Eastern Red-Backed Salamanders: A Comprehensive Review of an Undervalued Model in Evolution, Ecology, and Behavior

What defines a model organism for a particular biological field? Although this idea has captured much discussion in biomolecular literature (e.g., Müller and Grossniklaus 2010; Ankeny and Leonelli 2011; Leonelli and Ankeny 2013; Russell et al. 2017), the fields of evolution, ecology, and behavior (hereafter, EEB) have historically focused more on studying the systems suited to answer the scientific question at hand. Yet, even without a broader community discussion, particular research systems in EEB stand out for their body of knowledge and contributions to unifying principles, and researchers often invoke their study species as being excellent models for “robust inference” (sensu Travis 2006). Duffy et al. (2021) laid out a broad framework for defining, developing, and nurturing model organisms and model systems in EEB, including describing important contrasts with the biomolecular sciences, which we briefly summarize in the following section. Similarly, Edwards (2021) discussed the value of a model clade to EEB and systematic biology in particular. In this monograph, we argue that one of the most common vertebrates in northeastern North American forests, the Eastern Red-backed Salamander (Plethodon cinereus), deserves recognition and further development as an EEB model organism.

We make this argument based on several factors, including (1) the depth of knowledge published on this species that highlights the tractability and potential of P. cinereus both in the laboratory and field for answering EEB questions (Petranka 1998; Anthony and Pfingsten 2013; Jaeger et al. 2016; studies reviewed herein), (2) the critical niche position of woodland salamanders in forest ecosystems (Davic and Welsh 2004), (3) the accessibility of this species as easy to find, human-tolerant, and locally abundant across a large geographic and climatic range (Fig. 1; Petranka 1998), and (4) the potential for transferable methods and knowledge to a wide range of salamander species given the extensive diversity, abundance, and well-studied systematics of family Plethodontidae (e.g., Adams et al. 2009; Kozak and Wiens 2010; Pyron and Wiens 2011; Jaramillo et al. 2020). In making our argument, we review recent gains in knowledge of this species and discuss the next steps in investment and infrastructure needed for model organisms like P. cinereus to continue furthering our scientific understanding (Ankeny and Leonelli 2011; Leonelli and Ankeny 2013; Duffy et al. 2021; Wale and Duffy 2021).

Our monograph has four overarching goals and is organized as follows. First, we briefly discuss the definition and use of model organisms in EEB, highlighting salamanders more generally as good candidates for model clade status. Second, we determine the current level of EEB knowledge on P. cinereus compared to other well-studied salamander species. Third, we share the results of a comprehensive literature review of the EEB research on P. cinereus which focuses on literature published since Petranka (1998) and is written in the context of two more recent reviews (Anthony and Pfingsten 2013; Jaeger et al. 2016). Finally, we discuss the challenges of advancing P. cinereus and similarly well-studied species into the next stage of model organism status. We suggest that a collaborative ecological research network structure offers an opportunity for the continued development of P. cinereus as a model system in EEB.

As part of defining what EEB model organisms are, Duffy et al. (2021) summarized the considerations for using model organisms in cellular, molecular, and developmental biology (hereafter collectively referred to as the biomolecular sciences, e.g., Müller and Grossniklaus 2010; Ankeny and Leonelli 2011; Leonelli and Ankeny 2013; Russell et al. 2017), aptly noting that these fields tend to neglect the ecology and evolution of their models. For example, Russell et al. (2017) profile 14 “non-model model organisms” and focus on fundamental phenomena (i.e., the key processes and general principles of interest to a discipline; Ankeny and Leonelli 2011; Leonelli and Ankeny 2013; Wale and Duffy 2021) at the scale of molecules, genes, or cells with little regard for fundamental phenomena at the scale of populations, communities, or ecosystems. There is some discussion of how non-model model organisms have been used to study major evolutionary events (e.g., the origins of multicellularity with Volvox; Russell et al. 2017), but the fields of EEB are not well represented in the model organism literature. However, the EEB scientific community values many of these same organisms (e.g., Escherichia coli, Drosophila melanogaster, Arabidopsis thaliana), often for many of the same reasons (e.g., short generation time, lab tractability). Despite these overlaps in utility for both fields, Duffy et al. (2021: 55) argue that there are several challenges in applying the biomolecular definition of a model organism to EEB research. Instead, they propose the following definition for a model system in EEB:

In thinking about qualities that model systems in EEB share, the authors identify three axes which define the use of a given model system for a particular research question: lab tractability, field tractability, and the range of biological scales which can be addressed (see Duffy et al. 2021: fig. 1). It is important to highlight that their approach defines model systems—not always just a single species—which more accurately reflects how EEB scientists generally approach science from the perspective of larger biological scales compared to the biomolecular sciences. We add to Duffy et al. (2021) that model systems in EEB often benefit from encompassing taxa that are locally abundant and/or span relatively large geographic ranges. Local abundance and large geographic ranges both contribute to the accessibility of a model system and the potential for lessons learned from the model to have broader applicability to other species.

The exemplary taxa in EEB textbooks and research programs often take the form of model systems or clades: groups of closely related, ecologically diverse species, found in relatively well-defined geographic regions (e.g., Anolis lizards, Darwin’s Finches, Heliconius butterflies), which lend themselves to addressing a variety of broader EEB research questions. Instead of using common ancestry and evolutionarily conserved traits to declare EEB model systems representative of other species, as is done in the biomolecular sciences, EEB tries to conceptually unify the natural world. As Duffy et al. (2021: 55) state, EEB scientists “aim to see the forest and the trees” (their emphasis).

Important to the goals of this monograph, model systems in EEB must contribute to robust inference across the fundamental phenomena of the field (Travis 2006). Both Travis (2006) and Duffy et al. (2021) argue that an organism or system must generate a great body of knowledge to support continued study of EEB fundamental phenomena. In this monograph, we demonstrate that P. cinereus has well surpassed the bar of knowledge, but requires further development of infrastructure, community ethos, and funding investment to facilitate continued robust inference on woodland salamanders within their ecological communities.

Amphibians are represented in biomolecular model species by the Axolotl (Ambystoma mexicanum; Russell et al. 2017) and the African Clawed Frog (Xenopus laevis; Müller and Grossniklaus 2010). In EEB, pond-breeding amphibians have long played a major role in studies of metamorphosis and complex life cycles (Wilbur 1980), and salamanders have been the focus of a great number of studies on broad questions within ecology and evolutionary biology (e.g., Wake and Larson 1987; Bruce 2008; Wake 2009). Despite pond-breeding salamanders being more commonly recognized, the salamander family Plethodontidae represents nearly two-thirds of all salamander species (516 of 817; AmphibiaWeb 2024), and its members occupy a wide range of aquatic and terrestrial microhabitats, often with distinctly different life histories from pond-breeding amphibians (Wake and Lynch 1976; Petranka 1998; Blankers et al. 2012). Plethodontidae is the only salamander family to have radiated extensively in the tropics (tropical bolitoglossines; Wake 1970; Wake and Lynch 1976; Wiens 2007; Rovito et al. 2012, 2015; Jaramillo et al. 2020), and the only family of salamanders with species that have direct development (i.e., no aquatic larval stage; Wake and Hanken 2004). In fact, because Plethodontidae is the largest salamander family, direct development is the most common mode of reproduction in salamanders as a whole and was likely a key innovation involved in the diversification of Plethodontidae (Schluter 2000; Wake and Hanken 2004; Bonett et al. 2022). In terms of conservation status, species of Plethodontidae run the gamut from least concern to endangered or data deficient (see Supplemental Material S1; Supplemental Table S1, available online; IUCN 2020).

As amphibians and ectotherms, salamanders are generally sensitive to their environment (Duellman and Trueb 1986). Ecologically, salamanders are often mid-level consumers who play a critical role in nutrient cycling and other ecosystem processes throughout the aquatic and/or terrestrial ecosystems where they live (Davic and Welsh 2004). Plethodontid salamanders are particularly sensitive to climate, as these species are all lungless: all respiration is cutaneous, which requires moist skin. Climates and seasons that are too hot, too cold, or too dry are inhospitable to plethodontids. In part because of this environmental sensitivity, several studies have already made a case for the use of plethodontid salamanders as indicator species in biodiversity and ecosystem monitoring (Welsh and Ollivier 1998; Welsh and Droege 2001; Best and Welsh 2014; Farallo and Miles 2016; Fleming et al. 2020); however, others have argued for caution when using amphibians as “canaries in a coal mine” for anthropogenic change (e.g., Kerby et al. 2010).

The genus Plethodon includes 58 terrestrial woodland species found in North America; most of these species (49) form a clade of eastern Plethodon, and the remaining 9 species form a clade of western Plethodon, which are primarily found in the Pacific Northwest (Wiens et al. 2006; Kozak et al. 2009; Fisher-Reid and Wiens 2015; AmphibiaWeb 2024). Eastern Plethodon are further broken down into three well-supported clades: cinereus group, wehrlei-welleri group, and glutinosus group (Wiens et al. 2006; Kozak et al. 2009; Fisher-Reid and Wiens 2015). Phylogenetically, P. cinereus is a member of cinereus group, which is sister to the rest of eastern Plethodon (Wiens et al. 2006; Fisher-Reid and Wiens 2015). Plethodon cinereus is the sister species to the vulnerable P. shenandoah (IUCN 2020), which, although locally abundant where found, is geographically restricted to three mountaintops in Virginia, USA (Petranka 1998; Sites et al. 2004). Plethodon cinereus is a wide-ranging species found in temperate forest habitats along the eastern seaboard of North America from southern Canada to North Carolina, USA, and west to Minnesota, USA, in the north, and to the Appalachians in the south (Fig. 1). Within its range, P. cinereus is often locally abundant, with surface density estimates regularly ranging from 0.05 to 3.4 salamanders/m² (Petranka 1998; Anthony and Pfingsten 2013; Semlitsch et al. 2014; Grant et al. 2024), and some locations reporting even larger outlier surface densities. For example, Jung et al. (2000a) reported densities as high as 18.5 salamanders/m² in Shenandoah National Park, Virginia, and Hernández-Pacheco et al. (2019) reported densities of 2.5–6.3 salamanders/m² in urban Richmond, Virginia.

As the most common and most widely distributed species in the genus Plethodon, we argue that P. cinereus is an ideal candidate for an EEB model species and endeavor to provide an updated, range-wide review of knowledge on this salamander. There is a wealth of natural history, evolutionary, ecological, and behavioral research that has been done on P. cinereus over the decades; however, the most accessible and cited range-wide review was conducted over 20 yr ago as part of a book detailing the natural history of all North American salamanders (Petranka 1998). In the early 2000s, some authors wrote additional broad overviews of amphibian biology which included species-specific knowledge (Casper 2005; Wells 2007), but these compendiums largely rely on the details found in Petranka (1998), with minimal updates. In 2013, Anthony and Pfingsten (2013) wrote a detailed species account of P. cinereus for the book “Amphibians of Ohio” (Pfingsten et al. 2013). This review more than doubled the number of citations provided by Petranka (1998) and included Ohio-based updates, including several tables and figures of new data from Ohio populations and a detailed discussion as to how these data fit with the broader literature covering the rest of the geographic range of P. cinereus. We have therefore incorporated information first summarized by Anthony and Pfingsten (2013) in relevant sections. Finally, 50 yr of behavioral research on P. cinereus was recently reviewed in Jaeger et al. (2016), in which the authors highlight both Petranka (1998) and Anthony and Pfingsten (2013) as the authoritative accounts for P. cinereus life history and ecology (Jaeger et al. 2016: 3). Our review aims to synthesize these prior reviews with the entire EEB literature that has been published on P. cinereus since 1998.

In order to quantify the body of knowledge on P. cinereus for EEB fields in comparison to other well-studied salamanders, we conducted several rigorous literature searches. From these searches, we used the resulting studies on P. cinereus to comprehensively review the post-Petranka (1998) literature, and then make the argument for P. cinereus to be elevated to an EEB model organism.

As a first step for establishing a species of woodland salamander (P. cinereus) as a model organism in EEB, we investigated its prevalence in the literature relative to other well-studied salamanders. We conducted three sets of taxonomically progressive Web of Science topic searches using the Web of Science All Databases for All Years (1864–present). All searches were conducted on 18 June 2020. The goal of our search strategy was to restrict our focused efforts to the most well-studied salamanders, first by identifying the most well-studied families, then genera within those families, and finally species within those genera. Full details of our search methods, including search terms and database settings, as well as summary tables of the results (Supplemental Material S1; Supplemental Tables S1–S4, available online) can be found in Supplemental Material S2 (Supplemental Data S1), although raw results can be found in Supplemental Materials S2-S15 (Supplemental Data S1–S14, available online). Supplemental Material S2 (Supplemental Data S1) also contains detailed methods for our estimation of geographic range size and inclusion of sources in the comprehensive literature review which were not recovered by our Web of Science searches.

Results for the family-level and genus-level searches are shown in Supplemental Material S1 (Tables S2 and S3), respectively, and the full results for all 50 genera are in Supplemental Material S2 (Supplemental Data S1), and for all 179 species in Supplemental Material S3 (Supplemental Data S2). In the species-level search (Supplemental Material S1; Table S4), in terms of raw numbers of articles in EEB-related fields compared to non-EEB related fields, P. cinereus ranks fifth, behind two salamandrids from Europe (Triturus cristatus and Salamandra salamandra) and two ambystomatids from North America (Ambystoma tigrinum and A. maculatum). Like P. cinereus, all four of these species have large ranges and are locally abundant (Lannoo 2005; Wells 2007; IUCN 2020). Unlike P. cinereus, they are all largely pond-breeding species with an aquatic larval stage, the exception being Salamandra salamandra, for which some populations have a fully terrestrial, viviparous reproductive mode (e.g., Velo-Antón et al. 2012). Plethodon cinereus is the only representative of Plethodontidae among the top eight most well-studied salamander species, which is surprising, given that Plethodontidae is the largest family of salamanders (AmphibiaWeb 2024), and the only salamander family with a direct developing reproductive mode (Wake and Hanken 2004). However, the vast majority of salamander reproductive and developmental biology has been studied in species with an aquatic larval stage (e.g., see non-EEB sources for A. mexicanum, Supplemental Material S5; Supplemental Data S4).

Among all salamanders for which geographic range data were available (n = 607; IUCN 2020), P. cinereus has the 11th largest geographic range (Supplemental Material S13; Supplemental Data S12). Among the top eight most well-studied salamander species (Supplemental Material S1; Table S4), five have geographic range sizes larger than P. cinereus, supporting the idea that well-studied species often have large ranges. Among plethodontids, P. cinereus has the second largest range overall (1.89 million km²; Supplemental Material S13; Supplemental Data S12; IUCN 2020), and the largest range of North American plethodontids. The tropical South American plethodontid Bolitoglossa altamazonica is reported to have the largest plethodontid range at 1.97 million km² (Supplemental Material S13; Supplemental Data S12; IUCN 2020). However, recent molecular and morphological analyses of B. altamazonica and B. peruviana (Cusi et al. 2020; Jaramillo et al. 2020) suggest B. altamazonica has previously been used as a catchall taxon for any specimens from Venezuela south to Bolivia and from Ecuador east to Brazil, and that instead B. altamazonica is a microendemic species restricted to forests in northern Peru, between the rivers Nanay in the north, Tigre and Marañón in the south, and Amazon in the west. Therefore, it is likely that P. cinereus has the largest range of the plethodontids.

The following sections summarize the results of our comprehensive literature review of P. cinereus research since Petranka (1998) and published prior to our search date in June 2020. Additional sources published after our search date but suggested by anonymous reviewers during peer review have also been incorporated (see Supplemental Material S2; Supplemental Data S1).

We use the same section headers as Petranka (1998), but we have added subheaders for larger topics. We follow Petranka’s (1998) organization of topics closely. We have taken special care to note when more recent research 1) supports what is written in Petranka (1998), 2) fills in a gap in understanding which Petranka (1998) identified, or 3) contradicts what is written in Petranka (1998). In Supplemental Material S3 (Supplemental Data S2), we include our categorization and key results summaries of these sources, which we used to organize this review (Supplemental Material S12; Supplemental Data S11). We also include the full list of sources returned by Web of Science on P. cinereus (Supplemental Material S9; Supplemental Data S8) and as well as the full list of sources cited by Petranka (1998) and Anthony and Pfingsten (2013) for completeness (Supplemental Material S14; Supplemental Data S13).

Petranka (1998) describes the two predominant color phenotypes (striped morph or red-back and unstriped morph or lead-back) as well as the erythristic or all-red morph (Fig. 1). Since Petranka (1998), documentation of other color phenotype anomalies (e.g., leucistic) from throughout the geographic range have been published regularly, often in the Herpetological Review Natural History Notes section (e.g., Mendyk et al. 2010; Sawyers and Novick 2011; Jongsma 2012a,b). Recently, Moore and Ouellet (2014) reviewed color phenotypes in P. cinereus, summarizing eight diagnosable color phenotypes with photographs. Moore and Ouellet (2014) provided a dichotomous key for identifying the color phenotypes and established and maintain an open-access database of photographs, providing an email address for photograph submissions. The authors also provide a table documenting each state or province in which different morphs have been found, noting that the unstriped morph is absent at range edge populations in Minnesota, Kentucky, and Tennessee. Moore and Ouellet (2014) argue that the striped, unstriped, and erythristic morphs (Fig. 1) should be considered the primary color phenotypes of P. cinereus, as they are the most abundant and occur most widely through the geographic range. The remaining five color phenotypes (iridistic, albino, leucistic, amelanistic, melanistic) are somewhat rare anomalies, possibly simply teratological mutations, with patchy observations throughout the range. It is not known if any of these anomalous color phenotypes persist in any population long-term.

Additional work on color phenotypes since Petranka (1998) includes studies on the expression of different color phenotypes over time and during development. Over time, Kraemer et al. (2012) observed decreasing brightness of dorsal and lateral pigmentation and increasing ventral reflectance in both striped and erythristic morphs kept in a laboratory. The authors hypothesized that the changes in brightness were due to upregulation of dermal pigments (Kraemer et al. 2012) but noted that little is known about the cellular mechanisms of dermal pigmentation in this species. Kraemer et al. (2012) also caution that their study suggests laboratory quantification of coloration may not be representative of similar measurements recorded in natural settings.

Over the course of development, Evans et al. (2020) found that temperature may play a role in the expression of the color phenotype. The authors reared eggs at different temperatures and found a higher proportion of unstriped individuals hatch at higher temperatures. The authors note, however, that it is difficult to know if this is true phenotypic plasticity in color phenotypes (i.e., high temperatures induce expression of unstriped morph) or simply differential mortality based on temperature (i.e., eggs containing unstriped individuals are more likely to survive rearing at high temperatures compared to eggs containing striped individuals; Evans et al. 2020). Further work is needed to tease apart these hypotheses.

Recent research has revealed an additional interesting trait related to skin phenotypes, one that is likely sexually dimorphic (see Anthony et al. 2023). Muñoz (2018) described natural green fluorescence under a blacklight on the ventral side of tails for 45 salamanders (33 males and 12 females) in Pennsylvania during the autumn (September to November) 2014 season. In Massachusetts, one of us documented the same ventral fluorescence in 60 individual salamanders using a 395-nm blacklight during the autumn 2019 season (M. C. Fisher-Reid, personal observation), as well as in a handful of individuals in the autumn seasons of 2017 and 2018 (prior to intentionally recording these observations; Supplemental Material S16; Supplemental Material Fig. S1, available online). In 2019, we observed a strong male bias in fluorescence (n = 53/87 individual males (60.9%), 3/83 individual females (3.6%), 4/93 individual unknowns (4.3%); M. C. Fisher-Reid [personal observation]). We further observed that fluorescence persists for several weeks, as 15 individual males were recaptured 2–4 times (mean = 2.6) during the same season and exhibited fluorescence each time they were caught (sample dates are generally ≥7 d apart; M. C. Fisher-Reid [personal observation]). Muñoz (2018) speculates that the fluorescence may aid in night vision but acknowledges that our lack of knowledge on how the fluorescent compounds are generated and on the natural light conditions under which they fluoresce makes it difficult to hypothesize as to their ecological or evolutionary importance.

Anthony et al. (2023) was the first study to characterize tail fluorescence formally. The authors sampled more than 300 salamanders over 12 mo from a single population in Ohio plus more than 100 salamanders from additional populations in Indiana, Pennsylvania, northern Ohio, and southern Ohio, all during the month of October. Anthony et al. (2023) confirmed the fluorescence occurs in Serous 1 (S1) glands (Simons et al. 1999; Hecker et al. 2003; see Terrestrial Ecology: Territories and Home Range Size), the trait is more commonly observed in males than females, more commonly observed in autumn than spring, and only observed in adults. These patterns hold across the different geographic locations, which include three of the six phylogeographic clades of P. cinereus (Radomski et al. 2020). In their discussion, Anthony et al. (2023) provided an excellent starting point for further characterization of the trait and future tests of its hypothesized function.

More broadly, Petranka (1998) summarizes few sexually dimorphic traits in P. cinereus, with sexually active males showing swollen nasolabial glands, hedonic glands on the tail, and a mental gland. Two additional studies describe noninvasive methods for sexing adult salamanders. Quinn and Graves (1999a) report on sexual dimorphism in the internares distance (distance between nares and snout tip). They show that males had significantly larger internares distances than females, for salamanders whose sex was confirmed by dissection. Although this trait is extremely useful for sexing preserved specimens without dissection, it is more difficult to use with live animals. Measuring internares distance on a live animal requires that the animal be cooled on ice prior to measuring; thus this method is unlikely to be practical in the field (Quinn and Graves 1999a). Gillette and Peterson (2001) describe a candling method for determining sex non-invasively, which is used commonly today in both laboratory and field settings (e.g., Cabe et al. 2007; Novarro et al. 2018). One can determine sex by shining a light through the dorsal side of the salamander and looking at the ventral body wall for shadows of testes and vas deferens (if male) or eggs (if female). Gillette and Peterson (2001) also note that sexually active males have a whitish, enlarged area lateral to the cloaca, and gravid females have easily visible eggs once eggs are >2 mm. Gravid females can have their eggs counted by pressing gently on the body wall to make the individual eggs visible through the ventral skin. Distinguishing between subadult females with no eggs and subadult males with un- or underdeveloped testes is difficult with candling; however, in our experience, candling works well on adult salamanders (i.e., salamanders that are approximately 35 mm snout–vent length [SVL]; see Petranka [1998] for a summary of the geographic variation and sexual dimorphism in SVL at sexual maturity).

Genetic and bioinformatic tools for understanding the systematics and geographic variation of P. cinereus were only just becoming inexpensive and commonplace in 1998. The studies Petranka (1998) cites on this topic rely exclusively on allozyme data to distinguish species-level and population-level genetic variation (e.g., Highton and Webster 1976; Highton and Larson 1979). However, the early inferences made from these data have largely held up under the scrutiny of both Sanger and next-generation sequencing methods combined with more robust modern computational phylogenetic methods (e.g., Wiens et al. 2006; Fisher-Reid and Wiens 2011; Radomski et al. 2020). Today, we are confident that P. cinereus is genetically distinct from both its sister species P. shenandoah (Highton 1999; Sites et al. 2004; Wiens et al. 2006; Highton et al. 2012) and the phenotypically similar Southern Red-backed Salamander (P. serratus, Wiens et al. 2006; Highton et al. 2012). There is limited evidence of hybridization between P. cinereus and P. shenandoah (see summary in the discussion of Carpenter et al. 2001), no evidence for hybridization between P. cinereus and either P. sherando or P. serratus (Bayer et al. 2012), and limited evidence of hybridization between P. cinereus and P. electromorphus in Ohio (Kuchta et al. 2022; which reexamined data from Lehtinen et al. 2016).

The systematics of plethodontid salamanders has been one of the most active areas of research, both prior to Petranka (1998), and since. Because Plethodontidae is the largest family of salamanders, and the only group of salamanders to radiate in the tropics (Wake and Lynch 1976; Wiens 2007), there has been much interest using them to study evolutionary questions in a phylogenetic context. From a systematics perspective, recent research supports the earlier conclusions of Highton and colleagues that P. cinereus appears to be a single evolutionary species, that is, individuals from across the range form a single phylogenetic lineage (Highton and Webster 1976; Highton 1999; Wiens et al. 2006; Fisher-Reid and Wiens 2011). A recent study has identified six phylogeographic clades within P. cinereus, based on both mitochondrial and nuclear genes (Radomski et al. 2020). Waldron et al. (2019) found a clinal transition between two of these clades in Ohio, detectable in the mitochondrial cytochrome b gene, but not in the 10 nuclear microsatellite markers used. Waldron et al. (2019) suggest two scenarios to explain the discordance between mitochondrial and nuclear data: first the cline could be an artifact of genetic drift or lineage merger, and second, that there may yet be a nuclear cline that is outside their sampling transect. The authors note that a genomic study with wider sampling is needed to fully understand the discordance between mitochondrial and nuclear genomes in P. cinereus. For additional work in this particular geographic region see our discussion of Hantak et al. (2019).

At the population level, Petranka (1998) notes that P. cinereus populations in formerly glaciated regions are less genetically variable than southern populations that were never glaciated (see Fig. 1 for glacial extent). This observation supports a hypothesis of postglaciation range expansion from southern refugia, which has been further supported by more recent studies using either microsatellites (e.g., Cameron et al. 2017) or a combination of mitochondrial and nuclear genes (Radomski et al. 2020). Molecular ecology studies using microsatellites have been an area of active research since Petranka (1998). Connors and Cabe (2003) published the first set of microsatellite markers for P. cinereus using individuals from Mountain Lake Biological Station (MLBS). These markers have since been widely used by researchers to test hypotheses of: isolation by distance (Cabe et al. 2007); the strength of natural and human-made barriers to dispersal (Marsh et al. 2007, 2008); impacts of habitat fragmentation on urban populations (Noël et al. 2007; Noël and Lapointe 2010; Wilk et al. 2020); multiple paternity of clutches (Liebgold et al. 2006); kin discrimination (Liebgold and Cabe 2008); influence of historical landscape change on population structure (Jordan et al. 2008); and color-biased dispersal (Grant and Liebgold 2017). The results of these studies are discussed in detail in the appropriate sections. Here, we note that Cabe et al. (2007) found low-level but detectable population structure over 2 km of continuous habitat in Virginia (but see Cameron et al. 2019 for conflicting results in Ohio), consistent with an isolation-by-distance model, and no evidence for sex-biased dispersal. In Maryland, Grant and Liebgold (2017) also found no evidence for sex-biased dispersal but did find evidence for color-biased dispersal: striped salamanders showed evidence of philopatry, whereas unstriped salamanders showed evidence of dispersal. It remains to be seen if this pattern exists in other populations. For further discussion on sex-biased dispersal, and dispersal more generally, see the section on Terrestrial Ecology: Seasonal Activity and Dispersal.

Additional microsatellite marker sets were developed from populations on Long Island, New York (Fisher-Reid et al. 2013), and from populations in Virginia, Ohio, and Pennsylvania (Cameron et al. 2017). Combined, these three marker sets cover five of the six phylogeographic clades identified by Radomski et al. (2020), all except the “North Carolina” clade at the southwestern range limit. Population-level comparisons between the phylogeographic clades have been hampered by low variation in the formerly glaciated regions (e.g., Fisher-Reid et al. 2013; Radomski et al. 2020) and difficulty amplifying markers outside of the populations in which they were developed in (e.g., Cameron et al. 2017).

Petranka (1998) focuses his discussion of geographic variation on ratios of color phenotypes across the range, hypotheses as to the persistence of color polymorphism, and why the color phenotype frequencies vary. Like the population genetic studies, this has been an extremely active area of research in the decades since Petranka (1998), and it has proven to be a complex puzzle. Generally speaking, we have ample evidence that the color phenotype frequencies observed prior to 1998 are stable 3–5 decades later (e.g., King et al. 1997; Fisher-Reid et al. 2013; Hantak et al. 2015; Evans et al. 2018; although, see our discussion of Grant et al. 2018). Less clear are the hypothesized relationships between color phenotypes and their physiology or climatic preferences. At the time Petranka (1998) wrote his chapter on P. cinereus, he summarized studies that provided evidence that striped morphs were more common in cooler climates (i.e., higher latitudes) and that unstriped morphs had a physiology consistent with being more tolerant of warmer conditions than striped morphs. However, more recent studies have identified exceptions to these general patterns and contradictory results that suggest high levels of local variation in both color phenotype frequency (e.g., Hantak et al. 2019) and color phenotype physiologies (e.g., Petruzzi et al. 2006). For example, Petruzzi et al. (2006) found that the patterns of physiological differences between color phenotypes first detected by Moreno (1989) in New Jersey were inconsistent and variable across several populations in Ohio. Additionally, Smith et al. (2015) found that, although both morphs were equally able to resist water loss, striped morphs recovered from dehydration more successfully than unstriped morphs, particularly in autumn.

Gibbs and Karraker (2006) show that frequencies of the unstriped morph have increased with sampling year alongside increasing global temperatures, supporting the patterns cited by Petranka (1998) and implying that the unstriped morph may become more common as global temperatures increase because of climate change. However, Moore and Ouellet (2015), Cosentino et al. (2017), and Evans et al. (2018) show no relationship between color phenotype frequency and sampling year. Cosentino et al. (2017) combined the Gibbs and Karraker (2006) and Moore and Ouellet (2015) data sets and found evidence that color phenotype frequency is related to an interaction between land use and temperature: The proportion of striped individuals was positively correlated with forest cover in warmer regions; however, this relationship weakened and disappeared as regional temperatures decreased with increasing latitude. This study suggests that forest cover may buffer natural selection on P. cinereus color phenotypes in warmer regions because of global climate change (Cosentino et al. 2017).

Recently, Hantak et al. (2021) looked at changes in morph frequency and body size (as measured by SVL) over time using specimens from Highton’s National Museum of Natural History (NMNH) collection. These specimens cover 37 geolocations (in New York, Maryland, and Virginia) and 43 yr (1956–1999). Hantak et al. (2021) found the frequency of striped morphs increased with increasing temperatures, increasing elevation, and through time, but decreased with increasing precipitation, contrary to previous studies (e.g., Lotter and Scott 1977; Gibbs and Karraker 2006; Fisher-Reid et al. 2013). Both morphs show decreases in body size with increases in temperature and increases in body size with increased elevation; however, these relationships show different strengths between morphs. Striped morph body size is more sensitive to temperature than unstriped morph body size, and unstriped morph body size is more sensitive to elevation than striped morph body size (Hantak et al. 2021). Over time, the body size of unstriped morphs is increasing, while the body size of striped morphs is decreasing. These complex patterns lead the authors to suggest that P. cinereus color morphs have a suite of co-adapted traits that are responding to multiple climate and geographic drivers over time (Hantak et al. 2021).

Cosentino et al. (2017) noted in their discussion that the many studies that examine trait differences between color phenotypes (e.g., Petruzzi et al. 2006; Anthony et al. 2008; Davis and Milanovich 2010; Fisher-Reid et al. 2013; Smith et al. 2015; Muñoz et al. 2016a; Cosentino and Droney 2016) suffer from being limited to a single population or geographic region. Related discussion of the pros and cons of single-population studies can be found in Hantak et al. (2019, 2020). In order to visualize the limitations in geographic scope, we recorded the geographic location (state) for 410 of the studies we reviewed in depth (Supplemental Material S15; Supplemental Data S14). Studies with data from multiple states were counted once for each state (see Supplemental Material S2; Supplemental Data S1 and Supplemental Material S3; Supplemental Data S2), resulting in 448 total state-level locations. Excluding range-wide studies and those with unknown locations, we found that just over half (52.2%, n = 234/448) of the locations where research has been conducted on P. cinereus since Petranka (1998) are in just three states in the center/southern part of the range: Virginia (29.5%, n = 132/448), Ohio (13.2%, n = 59/448), and Pennsylvania (9.6%, n = 43/448). Nearly one-third of the Virginia studies are from MLBS (9.2%, n = 41/448), a field station in southwestern Virginia, at elevation (∼1,100–1,200 m). MLBS has an atypical pattern of color phenotype frequencies for a high-elevation population (Angleberger and Chinnici 1975) and likely belongs to a geographically restricted clade (the Southern clade; sensu Radomski et al. 2020). Phylogeographically, these three states do include 5/6 clades (the North Carolina clade is not present); however, the most widespread Northern clade is not well represented by these three states (Radomski et al. 2020). This geographic bias in focal study populations indicates that, when trying to address geographic variation over large spatial scales, future work should be intentional in selection of new, unstudied site locations and take into consideration which of the six phylogeographic clades identified by Radomski et al. (2020) are under study.

Fisher-Reid and Wiens (2015) looked at the maintenance of color polymorphism throughout the genus Plethodon in relation to climate. Striped/unstriped polymorphism is observed in at least 8 other species of Plethodon, and another 14 species are fixed for either the striped or unstriped morph (7 species each). All of the polymorphic and monomorphic species are in the three most basal clades of Plethodon, suggesting color polymorphism existed within the ancestor of all Plethodon, and has been maintained in the genus for over 40 million years (Fisher-Reid and Wiens 2015). Interestingly, there is a strong relationship between color phenotype and climate among species which is similar, but not identical to that observed in P. cinereus, but there are only two significant within-species climate-morphology relationships among the eight polymorphic species (Fisher-Reid and Wiens 2015). Only P. cinereus and a western species, P. vehiculum, show significant climate-morphology relationships, and they are not identical (Fisher-Reid and Wiens 2015).

Several studies since Petranka (1998) have explored the variation of Plethodon color phenotypes for reasons other than climate, most using P. cinereus. Three of these focus on selection imposed by predators using clay models (Fitzpatrick et al. 2009; Kraemer et al. 2016; Grant et al. 2018), while two more focus on genetic structure and landscape heterogeneity (Fisher-Reid et al. 2013; Hantak et al. 2019). Fitzpatrick et al. (2009) used striped and unstriped clay models within the geographic range of polymorphic P. ventralis. The authors found that avian predators will attack the more common morph when presented with striped and unstriped clay models at varying frequencies, suggesting that negative frequency-dependent selection could be acting to maintain color polymorphism within Plethodon species. However, given the wide variation in morph frequencies in the literature (Petranka 1998; Cosentino et al. 2017), it is likely that other mechanisms are contributing to the proportion of different color morphs in a population.

Kraemer et al. (2016) used a clay model approach to test for predation selection on P. cinereus color phenotypes (striped, unstriped, and erythristic) imposed by mammals. In this study, the authors found complex results driving mammalian predator attacks on the clay models, including both directional selection favoring inconspicuous salamanders and, more strongly, negative frequency-dependent selection which favors novel (i.e., rare) salamander coloration. The latter pattern matches the findings of Fitzpatrick et al. (2009), which focused on avian predators. Kraemer et al. (2016) argue that the complexity of how mammals prey on salamanders is strongly influenced by local conditions and is only one piece of the complex puzzle of morph frequency variation throughout the range of P. cinereus.

Grant et al. (2018) combined the clay model approach with demographic data from 3 yr of P. cinereus mark–recapture at a Maryland population on the Delmarva Peninsula (i.e., the peninsula which includes portions of Deleware, Maryland, and Virginia) to explore morph-specific differences in survival. Delmarva is another location with atypical color phenotype frequencies (Highton 1977; Petranka 1998). Delmarva has a gradient in morph frequency, from majority striped in the north, to majority unstriped in the south (Highton 1977). Grant et al. (2018) found that there was a greater proportion of striped juveniles compared to the proportion of striped adults, suggesting differential survival of the two color phenotypes, with fewer striped individuals reaching adulthood. The authors also found that survival was only explained by the combination of color and age in spring, but not in autumn. Birds were more likely to attack the striped clay models, providing a potential mechanism for the differential survival, whereas mammals showed no differential predation by color phenotype (Grant et al. 2018). Interestingly, Grant et al. (2018) documented color phenotype frequencies that are opposite to what Highton (1977) found. In 1977, Highton recorded a color frequency of ∼70–80% striped at the sample locations closest to where Grant et al. (2018) conducted their study. Grant et al. (2018), on the other hand, recorded a color frequency of just 29% striped over their 3 yr of mark–recapture. This suggests that, unlike other locations which have been sampled repeatedly over time, the color phenotype frequencies on the Delmarva peninsula may not have been stable over 40 yr. It would be interesting to completely resurvey Highton’s (1977) localities and follow the modeling methods of Cosentino et al. (2017) to determine if the same covariates explain the change in color phenotype frequency over time.

Fisher-Reid et al. (2013) and Hantak et al. (2019) are two of several recent studies mentioned briefly above which have tried to leverage the technological advances in genetics since Petranka (1998) to understand variation in color phenotype frequencies in P. cinereus. Williams et al. (1968) were the first to note that the most common color morph ratio (70:30 striped:unstriped) is not observed on Long Island, New York: western Long Island is 100% striped, southeastern Long Island is 100% unstriped, and the center and northeastern parts of Long Island are roughly 50:50 striped:unstriped (Williams et al. 1968; Fisher-Reid et al. 2013). Using microsatellite markers, Fisher-Reid et al. (2013) found that the 100% unstriped populations are genetically isolated from the rest of the Long Island P. cinereus populations (as measured by F_ST), and show significant differences in morphology. Additionally, localities across Long Island show significant differences in microclimate and macroclimate when salamanders are active. Fisher-Reid et al. (2013) documented a soil moisture gradient on Long Island that correlated with the change in morph frequency, suggesting selection may have played a role in driving the less-fit morph extinct at the gradient extremes.

Anthony and Pfingsten (2013: Fig. 18-4) documented a similar correlation between a precipitation gradient and morph frequencies in Ohio. Much like on Long Island, their work shows that unstriped morph frequency increases in drier regions of Ohio. Further work in northern Ohio by Hantak et al. (2019) focused on the population genetics from 28 localities which span a range of color phenotype frequencies from 100% striped in eastern Ohio to nearly 100% unstriped in western Ohio. Using microsatellite markers, they found a complex interaction between gene flow, genetic drift, and natural selection working together to explain morph frequencies. Thus, both Hantak et al. (2019), at the genetic level, and Kraemer et al. (2016; see the foregoing), at the predator–prey interaction level, find evidence for complex evolutionary dynamics driving morph frequencies (see also Hantak et al. 2021). Using ecological niche modeling, Hantak et al. (2019) further found waterways were the most important factor in explaining morph frequency variation, likely, as the authors suggest, because waterways are strong dispersal barriers for a terrestrial salamander.

Our search recovered two additional studies that explored geographic variation not related to color phenotypes: one focused on skin microbial communities (Muletz-Wolz et al. 2017) and the other on phylogeographic patterns of pheromone profiles (Kunkel et al. 2019). Muletz-Wolz et al. (2017) is discussed in more detail in our section Additional Topics: Microbiota. In the context of geographic variation, Muletz-Wolz et al. (2017) sampled the skin microbial communities of 37 P. cinereus individuals across three localities in the central Appalachians, with one locality encompassing an elevation gradient. As other microbial studies have found (see Additional Topics: Microbiota), the authors report high levels of microbial diversity, and little overlap from one location to another, suggesting high taxonomic turnover of microbial communities across the landscape (Muletz-Wolz et al. 2017). Kunkel et al. (2019) compared the pheromone profiles of striped males from the Ohio clade to striped males from the Pennsylvania clade (Radomski et al. 2020) across a gradient in northern Ohio (overlapping the study area of Hantak et al. 2019). They found that, while pheromone profiles differed between populations, they did not differ between phylogeographic clades (Kunkel et al. 2019). These results suggest that the Ohio and Pennsylvania clades are not yet reproductively isolated from each other.

Taken together, the many studies focused on geographic variation in color morphology in P. cinereus make a strong case for complex interactions between macroclimate, microclimate, forest type, and predators on variation in color phenotype in P. cinereus. We refer the reader to the detailed discussion of Hantak et al. (2019) on the complex maintenance of color polymorphism in P. cinereus for additional thoughts on the complexity of color polymorphism in P. cinereus. Further, we discuss several studies that address the maintenance of the color polymorphism through other mechanisms (e.g., sexual selection, diet, territoriality) in greater detail in the Terrestrial Ecology section. We also note that there has been very little work on the genetic determination of color morphology in P. cinereus since Highton (1975) first documented that color is a polygenic trait. The genetics of color morphology in P. cinereus would benefit greatly from use of modern genetic tools to understand links between genes and color polymorphism better. Finally, our review of the literature clearly demonstrates a need for studies which focus on geographic variation in traits other than color morphology.

Petranka (1998) gives a brief description of the geographic range of P. cinereus, as well as the preferred habitat of adult salamanders. He noted that P. cinereus adults can be found in deciduous, northern conifer, and mixed deciduous–conifer habitats throughout their range, but he cited research that finds P. cinereus reach their greatest densities in well-drained, mature forests with deep soils. Petranka (1998) also cited evidence showing that populations are absent or at low densities in highly acidic soils, perennially wet soils, and shallow, rocky soils.

In compiling the more recent literature on the distribution and adult habitats of P. cinereus we noted more than 20 sources that simply document that P. cinereus was present at a given study site and categorized these as Regional Herpetofaunal Lists (Supplemental Material S12; Supplemental Data S11). In reviewing these sources, it is clear that the geographic distribution of P. cinereus has remained stable over the decades of intensive study, and if anything, it is underestimated by the most commonly cited range maps. That said, we found three sources explicitly documenting new records of P. cinereus. The first, Placyk et al. (2002), documented P. cinereus from two islands in Lake Michigan that had not been previously inventoried. Island populations of P. cinereus are common, particularly in areas that were formerly glaciated (e.g., Long Island, Martha’s Vineyard, Maine coastal islands, Canadian provincial islands; VertNet 2021). The second source recorded a new, high-latitude, high-elevation record for P. cinereus, found on Mount Katahdin at 1,173 m in central Maine, in what is considered subalpine habitat (Willey and Jones 2010). Finally, Baxter-Gilbert et al. (2022) documented a newly introduced population of P. cinereus on the island of Newfoundland, Canada, ∼530 km east of the nearest known native population on Nova Scotia.

Three studies noted fewer than expected (Silva et al. 2003; Meshaka et al. 2019) or no (Hartzell 2019) P. cinereus in suitable habitat; however, all of these studies focused their sampling during the summer months, when the heat drives the vast majority of P. cinereus underground (Fig. 2); thus we doubt these three studies represent true absences or declines in density at these locations (e.g., for evidence of true declines, see Highton 2005). Silva et al. (2003) were working at the northern edge of the geographic range of P. cinereus, where summer sampling is typically warranted (Fig. 2). They sampled from the start of May through the end of September and noted that prior work has shown P. cinereus is generally rare on Prince Edward Island (PEI), Canada. Their study suggests P. cinereus is even rarer than previously thought (Silva et al. 2003). That said, because island insulation likely means PEI forest climates are still suitable for P. cinereus in April and October, a survey that starts earlier in spring or continues later into autumn may provide a clearer picture of the actual abundance of P. cinereus on PEI. In general, the autumn season, particularly October, has the highest detection rates (e.g., Leclair et al. 2008; Anthony and Pfingsten 2013; McGhee 2013; Sutherland et al. 2016), and in the far southern part of the range, salamanders are surface active from autumn through the winter months into spring (Becker et al. 2013; Hernández-Pacheco et al. 2019; Fig. 2).

Much of the recent work on distribution and habitat has focused on the microhabitat preferences of both adult and juvenile salamanders, particularly with regards to refining sampling methods using natural or artificial cover objects (NCOs and ACOs, respectively). We categorized more than 50 studies as focusing on adult and/or juvenile habitat, as well as an additional 19 studies that test a variety of sampling methods based on presumed habitat preferences. Of those latter 19 studies, 18 are focused on ACO survey design, ACO survey performance compared to NCO and leaf litter surveys, and ACO best practices. Generally speaking, ACOs are a well-vetted method for standardized sampling of P. cinereus. Indeed, Hesed (2012) argues that ACO methodology has been optimized for P. cinereus over other woodland salamander species. Among our reviewed studies, at least one found ACOs are more efficient than leaf litter surveys (e.g., Otto and Roloff 2011) and several found ACO surveys are comparable to NCO surveys (e.g., Monti et al. 2000; Grover 2006; Moore 2009; Ciul et al. 2010).

In terms of best practices, wood boards (MacNeil and Williams 2013a; Hesed 2012), resting directly on the ground (Carfioli et al. 2000; Hickerson et al. 2012), and sampled no more frequently than once per week (Marsh and Goicochea 2003; Otto et al. 2013) have the highest encounter rates. Generally, ACOs are colonized by salamanders within 2 mo of installation (Bonin and Bachand 1997); however, other studies have documented faster colonization rates (e.g., 2 wk in Hickerson et al. 2012, which used ceramic floor tiles placed directly on the soil). There does not seem to be much impact on occupancy under ACOs based on the type of wood used (Bennett et al. 2003; Moore 2005) or the age of the ACO (Monti et al. 2000; Carlson and Szuch 2007). However, more recent work has shown ACO age may be important, with older, more weathered boards capturing more salamanders than younger, less weathered boards (Hedrick et al. 2021). In Virginia, one study found ACOs may be biased toward attracting adult salamanders and exclude juveniles compared to NCOs (Marsh and Goicochea 2003), but other studies from Maine and Québec found no age/size differences between ACO and NCOs (Monti et al. 2000; Moore 2009). When true density was estimated from removal, Siddig et al. (2015) found that ACO surveys overestimate the true density of salamanders, and NCO surveys underestimate the true density. However, the sampling in Siddig et al. (2015) was done in July in Massachusetts, when surface densities are at or approaching their lowest (see Terrestrial Ecology: Seasonal Activity and Dispersal and Fig. 2), and so should be repeated during the autumn or spring surface-active seasons for accuracy. Recent modeling using P. cinereus density estimates based on a mark–recapture data set has suggested that model-derived estimates of individual space use and detectability are both sensitive to the spacing of ACOs (0.5, 1, or 2 m apart) and the extent of an ACO array (total area occupied; Fleming et al. 2021). In contrast, model-derived estimates of population density are robust to variation in both ACO spacing and array extent (Fleming et al. 2021).

Beyond comparisons of ACOs to NCOs, several studies note that, although you can find P. cinereus in the leaf litter, they are more likely to be encountered under cover objects (Moore et al. 2001; Otto and Roloff 2011; McGhee 2013). Preference for woody NCOs vs rocky NCOs varies by study location. In one Ohio study, woody NCOs were preferred (Iverson and Smith 2010), whereas in Vermont (Richmond and Trombulak 2009) and Ohio more broadly (Anthony and Pfingsten 2013), rocky NCOs were preferred. Positive correlations between cover-object size (diameter or area) and salamander size (mass or snout–vent length) are inconsistent in those studies that have explored the relationship. Some studies found these correlations (e.g., Moore et al. 2001; Hickerson et al. 2004; Strojny and Hunter 2010a; Walton 2013), while others found no relationship between cover-object size and salamander size (e.g., Faragher and Jaeger 1997; Moore 2005, 2009). The inconsistencies may in part be because of different goals of these studies, and whether the study was designed to focus on adults, juveniles, or both age groups.

A study in Vermont by McKenny et al. (2006) suggests that structural complexity of the forest floor is important, finding P. cinereus to be more abundant when there was well-decayed coarse woody debris compared to less-well-decayed coarse woody debris. Strojny and Hunter (2010a) suggest that large coarse woody debris is important based on their study in Maine, which found salamanders to be more abundant under logs of larger diameter in canopy gaps after timber harvest. Using salamanders and substrate from the same population in Ohio, leaf litter preferences are inconsistent across a field-situated mesocosm experiment (Iverson and Smith 2010) and in a laboratory-based choice experiment (Renaldo et al. 2011). This result suggests that individual preferences are highly variable over time. Iverson and Smith (2010) found P. cinereus had no preference for deciduous vs. pine leaf litter, whereas Renaldo et al. (2011) found a preference for deciduous over pine, and the strongest preference for a combination of deciduous and pine litter.

Several studies directly contradict Petranka’s (1998) note that P. cinereus is uncommonly found on perennially wet or highly acidic soils: Chalmers and Loftin (2010) regularly observed P. cinereus in wetland habitats in Maine, and several studies have found P. cinereus in riparian environments (e.g., Moore et al. 2000, 2001 in Pennsylvania; Anderson et al. 2013 in West Virginia; McGhee 2013 in Virginia). In particular, Anderson et al. (2013) and McGhee (2013) both found equal encounter rates for P. cinereus between riparian and upland habitats. Moore and Wyman (2010) found a seemingly healthy population of P. cinereus in an extremely acidic forest in Québec, Canada. The pH under coverboards ranged from 3.1 to 5.2, and 82% of captures over 5 y were from coverboards with a soil pH ≤3.8 (Moore and Wyman 2010). This Québec population also had larger salamanders (both in SVL and mass; Moore and Wyman 2010) compared to other studies, directly contradicting an earlier laboratory experiment that found that P. cinereus living on low pH substrates experienced a significant loss of body mass and other potentially negative osmoregulatory effects (Frisbie and Wyman 1991). Similarly, Bondi et al. (2016) found P. cinereus to occupy a wider range of soil pH than earlier studies, 5 of their 34 sites in northern New York, Vermont, and New Hampshire had pH <3.0 and seemingly healthy P. cinereus populations. They found no evidence that P. cinereus was selecting microhabitats preferentially by pH (Bondi et al. 2016). Although these low pH values may be tolerable to P. cinereus, at least one recent study suggests they prefer more neutral soil pH. In Ontario, Gorgolewski et al. (2016) added wood ash to forest floors to raise the pH from ∼5 to ∼6.5 and found that the abundance of P. cinereus increased under coverboards with ash compared to those without ash. In contrast to Gorgolewski et al. (2016), Cameron et al. (2016) studied the effects of long-term soil liming (which increases soil pH) in an Ohio population and found no effect of liming on salamander body condition, population demographics, or surface density.

Three additional sources found P. cinereus in novel environments. Hughes et al. (1999) found one salamander dead inside a pitcher plant (Sarracenia purpurea) in two different locations: West Virginia and Pennsylvania. Both sites were the typical bog habitat of S. purpurea, which is an unusual place to find woodland salamanders. Hughes et al. (1999) hypothesize these two individuals were foraging when they died, given their stomach contents and the abundance of arthropod prey in the bog. Roble (1999) found an unstriped individual under driftwood on the sandy bank of the Potomac River in Virginia. Finally, LeGros (2018) found an individual hiding under the lid of a backcountry privy in Ontario, possibly feeding on the flies attracted to the privy. LeGros (2018) displaced the individual from the privy, and it returned within 9 h. Individual identification was confirmed using photographs of the spotting pattern on the head and tail (LeGros 2018). Such novel environment observations as those just cited are likely underreported in the literature. In our own field experience and those of our colleagues, P. cinereus is regularly found under anthropogenic objects (i.e., trash) within or near suitable forest habitats, including in suburban environments (e.g., Meshaka and Wright 2017). These observations are potentially important for understanding the dispersal and notably fast habitat colonization behavior of P. cinereus (e.g., see discussion in Williams et al. (1968) on colonization of formerly glaciated regions by P. cinereus).

Many studies attempt to characterize the preferred microhabitat of P. cinereus, either by measuring covariates in the field where salamanders are present or by doing behavioral choice trials of different habitats in a laboratory setting. In Virginia, Mitchell et al. (1997) found deep soils to be important in predicting P. cinereus presence in four out of five forest stands. Anthony and Pfingsten (2013) reviewed the maximum soil depths at which P. cinereus has been found, and noted that several pre-Petranka studies report finding salamanders at depths approaching 1 m. In the laboratory, individuals from an Ohio population preferred high pH, shaded habitat with medium-high moisture (Sugalski and Claussen 1997). The authors were able to show that pH was most influential in preferences, and that the combination of acceptable soil pH and moisture levels made a wider range of light intensities acceptable. In Pennsylvania, Brenner et al. (2005) found the presence of P. cinereus was correlated with soil moisture and temperature, and indirectly correlated with litter depth. Not surprisingly, given P. cinereus is fully terrestrial, they also found that salamander presence was not related to either distance or direction from vernal pools (Brenner et al. 2005).

Farallo and Miles (2016) compared the microhabitat characteristics of P. cinereus to two microendemic species, P. sherando and P. hubrichti in Virginia. They found evidence that all species seek their preferred habitats, as measured by temperature and moisture levels of the air and soil. There is additional evidence of microclimatic niche differentiation among all three species, suggesting character displacement when P. cinereus is sympatric with the microendemic species (Farallo and Miles 2016). A follow-up study on the same three species by Marsh et al. (2019) found that peak activity temperatures for P. cinereus were similar to P. sherando and cooler than P. hubrichti. There were no significant moisture patterns detected. As expected, the wide-ranging P. cinereus had a much larger range of active temperatures compared to P. hubrichti, but not when compared to P. sherando (Marsh et al. 2019). Yu et al. (1998) found that P. cinereus from Pennsylvania had high levels of constitutive heat shock protein 70; these levels did not change with acclimation to different temperatures in the laboratory. They compared P. cinereus to southern congener, P. jordani, and found that P. jordani was not as tolerant of heat shock as P. cinereus (Yu et al. 1998). In Ontario, Luymes and Chow-Fraser (2019) found populations to have stable abundances over an 18-y sample period with an average temperature of 12°C during the spring and autumn active seasons. The authors suggest that monitoring programs should plan on sampling for at least a decade to estimate occupancy more accurately (Luymes and Chow-Fraser 2019).

Kolozsvary and Swihart (1999) explored the relationship between patch size and P. cinereus occurrence in Indiana, finding a strong positive relationship as forest size increased from 1 ha to 1000 ha. McGhee and Killian (2012) found pH variability to be important in detecting P. cinereus using a regression model based on presence/absence data from a location in northern Virginia. They suggest salamanders actively select good microhabitat patches within less desirable patches. However, their sample size for their presence data was very low compared to their absence data (n = 42 salamanders over 2 y; presence detected on 29% of transects). McGhee and Killian (2012) argue that at their location, soil pH may be more important in determining which habitats salamanders avoid (low pH with low variability in pH). In Ontario, Brown et al. (2020) found that a suite of understory characteristics better predicted presence of P. cinereus than more macroscale variables like climate, disturbance, canopy cover, or landscape.

Finally, a handful of studies have explored how social hierarchies may be involved in habitat choices in P. cinereus. Forester and Anders (2000) found that hatchlings are more likely to associate with substrates marked by their mother than they are with substrates that are just wet or marked by a nonrelative conspecific female. Hurst and Smith (2006) found that females will avoid areas with cues from five or more conspecifics, whereas males do not discriminate. These two studies, as well as much of the work reviewed in Jaeger et al. (2016), imply that P. cinereus can not only recognize kin vs nonkin (but see Gibbons et al. 2003), and conceptualize the number of individuals in an area, but use this information to inform movement decisions. In West Virginia, Riedel et al. (2012) found that juveniles may actively avoid or be unable to compete for limited cover in nonforested habitats (e.g., meadows, pastures) that adults appear to live in without issue.

Future studies that try to understand the distribution and habitat of P. cinereus should first determine the phenology (Fig. 2) of their study sites, and then plan to sample during peak surface-active seasons. This will allow for more accurate presence, abundance, and density estimates, as well as increased sample sizes for characterizing microhabitat. Field methods for sampling P. cinereus during surface-active seasons are well developed and flexible, depending on the resources and personnel available. Well-designed choice experiments that offer P. cinereus different abiotic microhabitats or different cover objects are needed to characterize how salamanders perceive and select their habitat further. Once these abiotic requirements are met, there is ample evidence that social interactions are likely to shape habitat preferences further (Jaeger et al. 2016), and so these should be considered when designing experiments. Finally, and not surprisingly given the logistical difficulties of studying underground behavior, little is known about the underground activity of P. cinereus, and much of what is known about underground activity was largely documented pre-1998 (e.g., the depth that salamanders retreat to [Anthony and Pfingsten 2013]; see also the Terrestrial Ecology: Seasonal Activity and Dispersal section). Revisiting the methods of those earlier studies combined with creative solutions using modern radio-frequency tags (e.g., passive integrated transponders [PIT], tags; Sterrett et al. [2022]) provide an excellent starting point for further investigation of the underground activity of P. cinereus.

Petranka (1998) splits P. cinereus reproductive biology into two sections: Breeding and Courtship, which details all male and female behavior prior to insemination, and Reproductive Strategy, which focuses solely on female behavior after insemination. We follow a similar structure here. Petranka (1998) first describes the remarkably long mating season in P. cinereus based on studies that document the presence of sperm in female cloacae from autumn through winter to early spring. Mating is believed to occur primarily in the autumn. Breeding frequency is annual for males, whereas females breed biennially in most populations, particularly in more northern portions of the range, where the active season is shorter. That said, there is some evidence that larger females can breed annually even in more northern populations (>44 mm SVL in Connecticut; Lotter 1978). The courtship behaviors of P. cinereus are highly stereotyped, as they are for the entire plethodontid family, and therefore have been evolutionary stable for 50–100 million years (Houck and Arnold 2003). Detailed accounts of courtship behavior in P. cinereus have been well described elsewhere (Petranka 1998; Anthony and Pfingsten 2013; Jaeger et al. 2016). Although these descriptions are detailed, they are based on just two studies, one field study (Gergits and Jaeger 1990) and one lab study (Dyal 2006), both using salamanders from high-elevation Virginia populations. Additional observations of courtship throughout the range are needed. Briefly, P. cinereus courtship includes pheromone signaling, male abrasion of females with enlarged premaxillary teeth, nose tapping, and a coordinated tail-straddle walk, during which males deposit a spermatophore on the ground which females pick up by walking over it (Arnold 1977). Prior to 1998, sources cited by Petranka (1998) had observed that intruder males could disrupt courtship by biting the courting male. Lastly, Petranka (1998) summarizes two studies that support fecal pellet squashing behavior by gravid females as a mechanism for selecting mates with higher diet quality. Much of the related behavioral work on P. cinereus breeding and courtship has been thoroughly reviewed in Jaeger et al. (2016). Therefore, we focus this section of our review on recent studies that either (1) focus on nonbehavioral aspects of breeding and courtship (e.g., those focused on glandular physiology or life history timing) or (2) behavioral work independent of the R. G. Jaeger research program.

Since Petranka (1998), extensive work has been done to characterize the sexually dimorphic glands and the male pheromone proteins used in courtship by P. cinereus and other plethodontids. Sever (1997) and Sever and Siegel (2006, 2015) have characterized the sexually dimorphic glands using males and females from a single population of P. cinereus in Indiana. Sever (1997) observed that females usually store sperm for 3 to 4 mo, and up to 9 mo if mating occurred right before the winter hibernation period; however, contrary to other locations reviewed by Petranka (1998), autumn mating was rare in this population. A follow-up study described the complex spermatheca of females, where sperm clusters occurred in parallel arrays (Sever and Siegel 2006), and confirmed the predominance of spring mating in their population based on absence of sperm in females collected in October compared to those collected in April. Sever and Siegel (2015) found seasonal differences in the size and secretory activity of caudal courtship glands in males, with the greatest gland size and secretory activity occurring in April and October. Mental glands, found on the chins of mature males and used to produce and deliver pheromones to the female during courtship, show similar seasonal changes in size and secretory activity (Sever 2017). Related to these patterns, Church and Okazaki (2002) found that plasma testosterone levels and testes mass both vary seasonally in males from Virginia. Testosterone levels peak in the autumn in early October and reach their lowest point at the end of the spring in May. Peak testes mass is offset from testosterone, peaking in midsummer (July), and reaching its lowest point in the winter hibernation season (November–March). Additional details on the glands used for scent marking and territoriality can be found in the Terrestrial Ecology section.

Plethodon cinereus individuals from MLBS have been used to characterize both the evolution and function of courtship pheromones at the species, genus, and family level (e.g., Watts et al. 2004; Palmer et al. 2005, 2007, 2010; Wilburn et al. 2014). Here, we will focus on summarizing the pheromone protein research as it relates to courtship in P. cinereus, but we refer the reader to the extensive work of L. D. Houck and colleagues for greater detail on the evolution of plethodontid pheromones and courtship behaviors. To date, three pheromone proteins have been identified in Plethodontidae: the more recently evolved plethodontid receptivity factor (PRF), which is found only in eastern Plethodon (Palmer et al. 2005, 2007), and two ancestral proteins, plethodontid modulating factor (PMF) and sodefrin precursor-like factor (SPF), both of which were likely present in the ancestor of all plethodontids (Palmer et al. 2007, 2010). PRF and PMF are both expressed in the mental glands of sexually mature male P. cinereus (Wilburn et al. 2014), and there is evidence that rapidly evolving PRF has replaced the more ancestral SPF protein in P. cinereus (Palmer et al. 2007; Wilburn et al. 2014). PRF acts to increase female receptivity (Watts et al. 2004; Palmer et al. 2005), but much work remains to be done on the functional morphology and evolution of female receptors. As noted previously, Kunkel et al. (2019) did find population-level differences in pheromone chemistry, but these differences were not enough to prevent gene flow across two phylogeographic clades. Given the rapid evolution of these proteins (Palmer et al. 2005, 2007, 2010) and their importance in courtship success, further work on pheromone chemistry across the range of P. cinereus is warranted.

At the northern range limit, Leclair et al. (2006, 2008) document a much more delayed reproductive phenology in P. cinereus, particularly in females. In a Québec population, males become sexually mature between 3 and 5 y old and reproduce annually, whereas females were mostly unable to reproduce until 4–6 y old (Leclair et al. 2008), much later than results from studies conducted in more southern populations suggest (see sources cited by Petranka 1998). The decreased growth rate in northern populations is possibly due to colder temperatures and more limited opportunities for growth (Leclair et al. 2008). Both studies aged individuals using skeletochronology and found that these northern females typically only live 8 y, and it takes 3 y for a clutch of eggs to mature (Leclair et al. 2006, 2008). Therefore, most female P. cinereus in this population only reproduce once in their lifetime. Leclair et al. (2008) also found the juvenile sex ratio in this Canadian population was female-biased, likely because of the delayed female maturity and rare subsequent reproduction.

Male–female pairs are often observed sharing cover objects in P. cinereus populations, especially during the courtship season, whereas female–female pairs were uncommon and male–male pairs were very rare or nonexistent (Peterson et al. 2000; Anthony et al. 2008). Peterson et al. (2000) hypothesizes same-sex pairs occur coincidentally because of limited territory space. Meche and Jaeger (2002) argue that tail loss in females may impact the number of eggs that can be laid, as an individual female’s energy allocation for reproduction might shift towards tail regeneration. From this premise, the authors hypothesized that males would prefer tail-intact females over those who have recently lost their tails, with males using tail status of potential mates as a proxy for energy available for reproduction. However, males showed no preference for females based on tail status in laboratory experiments. This result is particularly interesting in the context of two studies discussed in detail in Reproductive Strategy: Madison et al. (1999a), which found tails were critically important for females to assume the posture needed to properly suspend a clutch of eggs, and Yurewicz and Wilbur (2004), who observed that tailless females did produce fewer ova.

In Ohio, Anthony et al. (2008) and Acord et al. (2013) both observed that male–female pairs found in the field are usually of the same color phenotype, suggesting positive assortative mating by color. However, a follow-up study by Jaworski et al. (2018) found that, in the laboratory, striped males cannot differentiate between female color phenotypes and thus showed no preference for either striped or unstriped females. These males did show a preference for larger females, presumably because female body size positively correlates with fecundity (Jaworski et al. 2018; see sources cited in Petranka 1998). There is some evidence that P. cinereus follow social monogamy for a mating strategy, and the male often aids the female in brooding eggs (e.g., Liebgold et al. 2006; see sources cited by Petranka 1998). Several articles from Jaeger and colleagues support the hypothesis of social monogamy (e.g., Guffey et al. 1998; Gillette et al. 2000; Jaeger et al. 2001; Jaeger et al. 2002a), noting that males and females exposed to unfamiliar individuals (and their scents) behave more aggressively than when exposed to familiar or partner individuals. Males and females were observed to be more aggressive towards polygamous partners than monogamous partners or strangers (Jaeger et al. 2016). Using this evidence, Jaeger et al. (2002a) hypothesized that P. cinereus uses sexual coercion against polygamous partners. For a thorough review of the behavioral work on P. cinereus, as it pertains to the courtship season and territoriality, we recommend Jaeger et al. (2016).

Atypical courtship behaviors have been observed in P. cinereus, although their relevance to the breeding and courtship of this species is unclear. In Virginia, Dyal (2006) observed a female-first tail straddle walk in P. cinereus that was not observed in two other plethodontid species studied. However, the data suggested that this behavior did not lead to successful spermatophore deposition in P. cinereus, and therefore Dyal’s (2006) hypothesis, that females have a larger role in courtship than previously thought, is difficult to evaluate. Courtship behaviors in plethodontids more generally were recently reviewed by Staub et al. (2020). They specifically describe a more active role for females in courtship than historical studies would suggest, a pattern of omission not unique to salamanders, as EEB fields strive to achieve greater gender parity in both scientists and their science (Staub et al. 2020).

Finally, a small number of recent studies have further investigated fecal pellet squashing behavior, which Petranka (1998) briefly summarizes at the end of his Breeding and Courtship section. To date, this behavior, a modification of nose tapping substrate to detect chemical cues, has only been observed in females and is thought to be involved in female choice of mates (e.g., Karuzas et al. 2004; Chouinard 2012; Jaeger et al. 2016). In laboratory settings, females have been observed squashing fecal pellets with their snouts, and experimental trials suggest they are detecting chemical cues about male diet and male health, which are then used to assess male territory quality and/or mate quality (Jaeger et al. 2016). Karuzas et al. (2004) found that while all females investigated male-marked substrates more than female ones, females on poor diets were especially likely to participate in the pellet-squashing behavior. Karuzas et al. (2004) agree with Petranka’s (1998) suggestion that the behavior is likely used to assess the diets of conspecifics; however, they suggest that the behavior is not limited to the context of mate selection. Instead, females could be using pellet-squashing to assess prey availability in the environment (Karuzas et al. 2004). Chouinard (2012) suggests that fecal pellets may not be the only method females have for assessing diet quality of potential mates. In a laboratory setting, Chouinard (2012) observed that males on high-quality diets had more protein in the mental gland compared to males on low quality diets. Females responded more quickly to and spent more time with males fed high quality diets (Chouinard 2012).

We still do not know what influences paternity, but females are known to be able to store multiple spermatophores and frequently lay clutches with multiple sires (Liebgold et al. 2006). We also do not know how the sperm from different males interact. There are additional gaps in our knowledge surrounding how or why males and females may occupy and co-defend (e.g., Mathis 1989) the same cover object, and if there is any geographic variation in mating behavior, as almost all of the research surrounding mating strategies has been done with individuals from MLBS. Morph-specific variation in mate quality has been documented at one population in Ohio: striped males cohabitate with larger, heavier females relative to those females cohabitating with unstriped males (Anthony et al. 2008). Given the variation in other morph-specific differences across populations (see above), this pattern should be explored at other sites. Related to this, more field research including capture–mark–recapture methods should be performed to determine if long-term pairing of males and females over multiple seasons occurs. Further research should also focus on mate selection in P. cinereus, specifically identifying the role of visual stimuli. Finally, we are unsure of the specific function of courtship pheromones, especially PRF and PMF, because previous research assumes they perform similar roles in P. cinereus as in other tested plethodontids.

In this section, Petranka (1998) summarizes what was known at the time about P. cinereus oviposition, parental care, and basic development. Generally, females oviposit in late spring or early summer, although observations of oviposition were reported as late as August in northern Michigan. Nests are made in cavities or crevices that females, and sometimes males, attend while brooding. Females aggressively defend eggs from conspecifics and predators, and Petranka (1998) cited evidence for reduced survivorship of abandoned clutches. During brooding, females have limited opportunities to feed, and grow significantly less, a pattern which is exacerbated in the presence of another female (Petranka 1998). However, females frequently cannibalize their own eggs when induced to oviposit in a lab and eat the eggs of abandoned clutches in both laboratory and field observations. Northern or high-elevation females tend to oviposit in cavities within decaying logs or beneath rocks and logs, whereas oviposition in southern populations was typically observed in subsurface retreats created by other animals. Petranka (1998) describes the gross anatomy of eggs and their arrangement in the clutch: Generally, clutches are suspended from the roof of the cavity in a grape-like cluster. Petranka (1998) further summarizes reports of clutch sizes and ovarian egg counts from several locations, with an average clutch size between six and nine eggs. In three populations where it was explored, the number of mature eggs was correlated with female SVL (Connecticut, Lotter 1978; Tennessee, Nagel 1977) or female body mass (New York, Fraser 1980). Egg incubation lasts for approximately 6 wk, and Petranka (1998) summarized studies of several populations where hatching occurs in August or September. In 1998, very little was known about the development of embryos, but mature embryos had been observed with three gills on either side of the head that are lost around the time of hatching.

Because P. cinereus has not yet been documented to oviposit naturally in the lab (published laboratory oviposition is induced via hormone injection, typically with agonists of luteinizing hormone-releasing hormone [LHRH], e.g., Evans et al. 2020), and their nests are often difficult to locate underground, studies on their reproductive strategy since Petranka (1998) have been limited. There have been several important studies, however, which document key oviposition and care behaviors and further characterize the development of P. cinereus eggs and embryos. Oviposition dates, when reported, generally match those reported in sources cited by Petranka (1998), occurring frequently around June, although, more recently, a nest in Virginia was found as early as April (Orr and Ernst 2005).

Madison et al. (1999a) described the exact ovipositing behavior of females induced to lay eggs in the lab. Females assume an upside-down S-shape to lay eggs, using their tail to brace themselves, and were in this position for over 6 h. From this position, females suspended their eggs from the base of a log and adjusted the positioning of the eggs after oviposition with her snout. Madison et al. (1999a) hypothesized that this behavior would place tailless females at a disadvantage for ovipositing, and more recent work has observed tailless females to produce fewer ova, indicating that reproductive success is inhibited by tail loss (Yurewicz and Wilbur 2004). Madison et al. (1999a) also hypothesized an evolutionary advantage for egg suspension, since unsuspended eggs were cannibalized 100% of the time, while suspended eggs were only cannibalized 50% of the time.

Controlled laboratory breeding of P. cinereus, for example, crosses of specific males and females as would be needed for selection experiments or quantitative trait loci (QTL) analysis, is not documented to date, and generally considered not possible in this species (Petranka 1998). Females do store sperm, and so can be induced to lay fertilized eggs in the lab (e.g., Madison et al. 1999a; Evans et al. 2020), thus cross-fostering experiments are possible once females produce clutches. Crespi and Lessig (2004) successfully cross-fostered 15 clutches, with 10 control clutches (eggs removed from mother’s care and then replaced). This study showed the presence of maternal effects, likely driven by differences in brooding behavior: There was a positive correlation between a foster mother’s SVL and hatchling’s body length. Behaviorally, larger mothers did not move their eggs as often and spent more time in direct contact with their eggs (Crespi and Lessig 2004).

Brooding females were more likely to abandon clutches in low-resource conditions (Yurewicz and Wilbur 2004). The authors found that smaller clutch sizes were not correlated with larger eggs, making clutch size an appropriate proxy for reproductive investment. Yurewicz and Wilbur (2004) also found that females were less likely to abandon or cannibalize larger clutches (about eight eggs), indicating that the high energetic cost of brooding outweighs the fitness value of smaller clutches (about five eggs). However, another study showed that brooding females defended large (10 eggs) and small (4 eggs) clutches with equal aggressiveness (Tornick 2010). These results seem contradictory, regarding the ability of females to differentiate between clutch sizes and make energetic investment decisions accordingly, but the discrepancy may be because Tornick (2010) did not investigate female behavior with clutches of fewer than four ova. Females were more aggressive when defending more mature clutches, indicating that the developmental stage of eggs influences investment decisions by brooding female P. cinereus (Tornick 2010). Ovarian egg counts were also used to determine the average clutch size of some populations, as these observations are typically more accessible and noninvasive. However, no studies have yet determined the relationship between ovarian egg counts and actual clutch size. Because P. cinereus are known to reabsorb eggs (Ng and Wilbur 1995), is it unclear how reliable this method is for predicting clutch size.

Kerney (2011) created a complete larval staging table for P. cinereus, showing the development of embryos over 72 d until hatching. Despite being a direct-developing species, P. cinereus embryos were found to have many characteristics indicative of a metamorphosing species, including hyobranchial skeleton development and the temporary presence of certain cartilages, gill clefts, and Leydig cells (Kerney 2011; Kerney et al. 2012). This is relevant to the loss and re-evolution of the larval stage by plethodontid species in the genus Desmognathus, as this shows that the development of key larval traits was not fully lost in direct-developing plethodontids (Kerney et al. 2012). Recent work by Lewis et al. (2022) has fully characterized the developmental basis of lung loss in Plethodontidae using P. cinereus, Desmognathus fuscus, and Hemidactylium scutatum. Interestingly, plethodontid embryos do begin to develop lungs, but these lung primordia regress via apoptosis prior to hatching (Lewis et al. 2022).

Future research on the reproductive strategy of P. cinereus females should be cognizant of potential behavioral changes due to laboratory settings (e.g., inducing oviposition increases oophagy; Petranka 1998; Madison et al. 1999a). More field-based studies and observations of reproduction and brooding are necessary for confidence in the results from laboratory studies. No study published since 1975 has used Highton’s (1975) method of burying jars with gravid females inside. Highton (1975) later collected the jars which contained both the brooding female and her clutch. This method provides an alternative to hormone injections for studies that require females and clutches and could possibly be refined for laboratory use. Future studies should also determine the accuracy of using ovarian egg counts to predict clutch size, as gravid females are much easier to capture and assess than nesting females. We still do not know what factors influence reabsorption of eggs, nor what factors induce females to cannibalize their own eggs. Finally, although reproductive timing and clutch size have been shown to be influenced by latitude, there have not been any recent studies investigating reproductive trends in populations at the southern range limit and our knowledge of patterns at the northern range limit is from a single population (Leclair et al. 2006, 2008). Those southern populations that have been studied are at elevation (i.e., at MLBS; Crespi and Lessig 2004; Yurewicz and Wilbur 2004) and have their own local phenology distinct from both southern and northern low elevation populations (Fig. 2), suggesting their results may not be generalizable to those regions.

The terrestrial ecology section is the lengthiest section of Petranka (1998). When categorizing studies, we found it useful to use a finer-scale breakdown of topics within terrestrial ecology, and these are represented by our subheaders in this section. There are some topics that overlap with other sections of this review, and are discussed in those other sections (e.g., juvenile habitat is summarized in Distribution and Adult Habitat), and some topics we have opted to combine (e.g., diet and feeding behaviors). Finally, we end this section with a catchall section on studies of general behavior and experimental methods not directly related to any of the preexisting categories in this or other sections.

Petranka (1998) described the diet of P. cinereus as any palatable metazoan prey an individual can capture. He also noted that diet is variable, but the literature Petranka (1998) reviews suggests there are no noticeable shifts in dietary preferences as individuals grow. Prior observations indicated that adults can be cannibalistic, consuming eggs and juveniles in both the lab, and on rare occasions, in the field. It was thought that P. cinereus fed primarily on the forest floor, but individuals have been observed to climb vegetation to forage. Petranka (1998) cited limited evidence for winter feeding in underground retreats from a population in Indiana, based on the stomach contents of specimens dug out of abandoned ant mounds. Adult P. cinereus are known to use both visual and olfactory cues to find prey. Individuals have been observed selecting higher-quality (i.e., higher caloric value) prey over low-quality prey when prey density is high, but eating all available prey when prey density was low (summarized in Petranka 1998).

At least 35 recent studies have focused on the diet and/or feeding behaviors of P. cinereus since Petranka (1998). Diet was assessed directly by examining stomach contents (via dissection or flushing) in 15 studies. As might be expected, prey items vary by location, but ants, mites, annelid worms, and springtails (Collembola) are the most common prey items across the range (Bellocq et al. 2000; Rooney et al. 2000; Adams and Rohlf 2000; Maerz and Karuzas 2003; Maerz et al. 2005, 2006; Arif et al. 2007; Anthony et al. 2008; Ivanov et al. 2011; Bondi et al. 2015, 2019; Paluh et al. 2015; Stuczka et al. 2016; Hantak et al. 2016, 2020). Interestingly, one study based in Delaware found an abundance of Collembola in the field, but none in salamander stomachs, although the sample size of individual stomach contents was small compared to other studies listed above (n = 12; Rooney et al. 2000). Three studies explored relationships between P. cinereus and nonnative prey, finding evidence that nonnative invertebrates can either serve as an added resource consumed alongside native prey (Maerz et al. 2005; Ivanov et al. 2011) or are ignored if deemed unpalatable (Ducey et al. 1999). Although by no means the first study to use the method, Bondi et al. (2015) tested the efficacy of the gastric lavage method (i.e., stomach flushing) as a technique for nonlethal sampling of salamander stomach contents. The authors find it to be reliable; 95% of salamanders regurgitated some or all of their stomach contents. Large prey (land snails, adult Coleoptera) were the least likely to be recovered through gastric lavage, and often these instances were singular items left behind while all smaller prey were recovered (Bondi et al. 2015). Finally, there were two studies that provide additional confirmation of cannibalism in the field (Maerz and Karuzas 2003; Gade et al. 2017).

Several studies have explored head morphology in P. cinereus and other congeners (e.g., Adams and Rohlf 2000; Maerz et al. 2006; Arif et al. 2007; Adams 2011), focusing on questions of character displacement in sympatry with congeners and relationships between head shape and prey size. These are all discussed in more detail in the Community Ecology: Other Salamanders and P. cinereus Interactions section. Here, we note that Adams (2011) explored the heritability and quantitative genetics of head shape variation in hatchling P. cinereus and P. nettingi. The heritability of head shape was generally high (h² > 0.65); however, Adams (2011) also found evidence that these two species would evolve differently in response to the same selection pressure on head shape, suggesting head shape as a potential trait axis by which diversification in salamanders occurs. Foraging-associated behaviors, such as territorial defense of prey-rich space, have also been studied in detail, and are summarized in the section Terrestrial Ecology: Territories and Home Range Size.

Anthony et al. (2008, 2017) explored the relationship between diet and color phenotype of a single population in Ohio. In autumn, Anthony et al. (2008) found that striped morphs eat more diverse and more profitable (i.e., easier to digest) prey than unstriped individuals, whereas the follow-up study found that striped morphs resided on territories with a greater abundance of available prey items (Anthony et al. 2017). At the same population, with regular sampling across 13 mo, Stuczka et al. (2016) found the opposite pattern to Anthony et al. (2008): The diet of unstriped morphs was more diverse than that of striped morphs; however, they also found that the diet differences observed between morphs were greatest during seasons with the greatest surface activity of salamanders (spring and autumn), that is, when competition among individuals is highest. Together with a related study from the same population on morph-specific aggression by Reiter et al. (2014), these four studies suggest that striped morphs may competitively exclude unstriped morphs from territories with larger quantities of more energetically profitable prey.

Two recent studies by Hantak et al. (2016, 2020) suggest that these documented diet differences between morphs are likely population specific and highly variable across both time and space. At one location in Ohio, Hantak et al. (2016) found no significant differences between the diets of sympatric striped (n = 29) and erythristic (n = 25) morphs. Hantak et al. (2020) looked at stomach contents for salamanders at six Ohio populations: two monomorphic striped, two polymorphic, and two nearly monomorphic (>99%) unstriped. One of the polymorphic populations was the same population sampled by Anthony et al. (2008, 2017), Reiter et al. (2014), and Stuczka et al. (2016). Hantak et al. (2020) also looked at temporal differences in diet, by sampling in both the spring and autumn seasons. Contrary to earlier studies discussed previously, they found high levels of diet overlap between morphs, both within a population, across populations, and across seasons. The polymorphic population sampled by both Hantak et al. (2020) and the studies cited above only showed morph-specific differences in diet in autumn. The other polymorphic population showed no significant morph-specific differences in diet (Hantak et al. 2020). The dietary breadth of both morphs did not change between monomorphic and polymorphic populations, and diet composition varied across seasons.

Hantak et al. (2020) noted in their discussion that the sample sizes in their study were much lower than the earlier studies: for the same polymorphic population, the sample size for Hantak et al. (2020) was n = 40, and for Stuczka et al. (2016) was n = 256. Thus, although Hantak et al. (2020) and other multipopulation studies (e.g., Petruzzi et al. 2006) suggest that many of previously documented differences between morphs may be highly localized, Hantak et al. (2020) particularly call attention to the logistical challenges of balancing sample size and effect size in multipopulation studies for a geographically wide-ranging species like P. cinereus. These challenges make detection of subtle differences between morphs difficult without large sample sizes; however, the challenges may be surmountable with a networked research approach by multiple research teams across the range of P. cinereus. Finally, Hantak et al. (2020) make a strong case for reanalysis of morph-specific trait complexes within a geographic context. As noted earlier, consideration of the phylogeographic clade to which a population belongs (Radomski et al. 2020) will help in this endeavor.

More generally, recent work confirms that P. cinereus forage at all times of day, although night foraging is likely the predominant mode (Placyk and Graves 2001a; Liebgold and Dibble 2011). Night-active salamanders had higher mass, and there was lower prey abundance in their field enclosures, compared to enclosures with fewer night-active salamanders (Liebgold and Dibble 2011). At night, olfactory cues are thought to be most important, as evidenced by increased nose tapping in the dark (Placyk and Graves 2001a), and decreased predation efficiency when the vomeronasal system is impaired (Placyk and Graves 2002). Telfer and Laberge (2013) found that salamanders were more likely to find and investigate soluble olfactory cues over volatile cues. The reliance on soluble olfactory cues is also supported by Maerz et al. (2001), which found that the foraging activity of individuals was reduced when exposed to water soiled by Common Garter Snakes (Thamnophis sirtalis). Social hierarchies also appear to influence foraging activity. Liebgold and Dibble (2011) found that both adults and juveniles foraged more actively when enclosed with familiar individuals of the opposite age class compared to when enclosed with unfamiliar individuals of the opposite age class.

Wise and Buchanan (2006) summarize three previously unpublished studies from their research group which explore how different levels of ambient light at night impact foraging. In all three studies, there is evidence that foraging is aided by increasing ambient light levels, especially if prey chemical cues are blocked. In the first two studies, when prey chemical cues are available, there is no change in foraging behavior under different light treatments (Wise and Buchanan 2006). However, when prey chemical cues are not available, salamanders are faster to orient to and attempt capture of prey in treatments that include either rapid shifts from total darkness (≤10⁻⁵ lux) to bright light (3.8 or 12 lux) or by increases in natural levels of ambient light (e.g., twilight, moonlight, starlight; treatments ranging from 10⁻¹–10⁻⁵ lux; Wise and Buchanan 2006). In the third study summarized by Wise and Buchanan (2006), salamanders were significantly more active in twilight (10⁻¹ lux) and day (photophase; 1 lux; equivalent to dim daylight under a forest canopy) treatments than under the night treatment (scotophase; 10⁻³ lux; equivalent to moonlight in the forest). However, the observed increases in activity did not change the amounts of prey consumed across treatments (Wise and Buchanan 2006).

Gibbons et al. (2005) found that although there is a genetic component to foraging efficiency, learning and exposure to prey can improve foraging efficiency over time. Sources cited by Petranka (1998) suggested that climbing behavior was likely a foraging tactic, and the results of Cote et al. (2016) support this. They found that salamanders exposed to dowels with prey scent trails climbed higher and for longer than on control (unscented) dowels. Individuals experimentally infected with the fungal pathogen Batrachochytrium dendrobatidis (Bd) appear to compensate for this infection by foraging more in a laboratory setting (Hess et al. 2015). The authors found minimal fitness effects of Bd infections compared to uninfected controls, suggesting that the extra foraging did help those individuals fight infection.

Physiologically, temperature seems to play a critical role in the efficiency of feeding. Diversity in the gut microbiome of P. cinereus begins to decline at 20°C, and digestion efficiency was the greatest at 15°C (Fontaine et al. 2018). Two studies have explored the use of stable isotope analysis to study diet and other factors (e.g., ecosystem-level nutrient cycling, Milanovich and Maerz 2012, 2013). These studies are discussed in more detail in the Additional Topics: Ecosystem Ecology section.

Gastric lavage has proven to be an invaluable method for nonlethal investigations into the diet of P. cinereus (e.g., Bondi et al. 2015). Future characterization of the diet of P. cinereus may be further improved by application of modern molecular methods, such as DNA metabarcoding (e.g., Harms-Tuohy et al. 2016), to stomach contents and fecal material. As with many topics reviewed here, and exemplified by the more spatially and temporally extensive work of Hantak et al. (2020), there is much to learn on the geographic variation and social context for both diet and foraging behaviors, including field observations of foraging behavior to complement lab studies.

Relative abundance and density were only briefly discussed by Petranka (1998). At that time, several studies had independently suggested that most of the P. cinereus population was below the soil surface on any given day (see also discussion of this topic in Anthony and Pfingsten 2013). Therefore, Petranka (1998) concluded, counts of salamanders on the surface and mark–recapture methods likely underestimate the true population size. Indeed, as part of a predator-removal experiment, Hickerson et al. (2017) removed all surface-active salamanders from under 144 ACOs during 98 visits across 4 y and estimated that this only depressed the salamander population density by 28%. Sources cited by Petranka (1998) estimated salamander density ranged from 0.21 salamanders/m² in Pennsylvania to 2.8 salamanders/m² in Virginia.

Both Semlitsch et al. (2014) and Grant et al. (2024) summarize previously published density estimates for P. cinereus. Both summaries report density estimates ranging from 0.05 to 3.3 salamanders/m² (Semlitsch et al. 2014; table 4; Grant et al. 2024: table 1). Grant et al. (2024) additionally present new data from a networked spatial capture–recapture project conducted in 18 study areas across the range of P. cinereus. Using these new data, Grant et al. (2024) estimated the range of average densities to be 0.195–3.43 salamanders/m², with a network-wide average density of 1.16 salamanders/m². In Virginia, additional outlier populations with even higher average densities have been recorded: first, in Shenandoah National Park, Virginia, Jung et al. (2000a) estimated density as high as 18.5 salamanders/m² and second, at the southern range edge in urban Richmond, Virginia, Hernández-Pacheco et al. (2019) estimated densities from 2.49 to 6.26 salamanders/m² per plot surveyed over 2 yr (Supplemental Material S16; Supplemental Fig. S2, available online; this population is also included in Grant et al. 2024).

More recent research focuses on how salamander abundance is impacted by human activities. These are discussed in more detail in the Conservation Biology and Additional Topics: Management sections to follow, but here we briefly describe the impact of roads, forest–farm edges, and prescribed fire on counts of surface-active salamanders. Plethodon cinereus has been observed to occupy forested locations along an urban to rural gradient (Gibbs 1998a). However, salamander counts are reduced near gravel roads, and this is likely due to a reduction in soil moisture near roads (Marsh and Beckman 2004). Dry conditions at forest-farm edges also lead to reduced counts of salamanders in these ecotones (Young and Yahner 2003). An experimental study corroborates these observations, with adults and juveniles found to be more abundant in plots with added moisture and more active in plots with more cover objects (Grover 1998). Increased abundance of cover objects near maintained recreational trails may be responsible for increased salamander counts near trails (Fleming et al. 2011). Prescribed fire did not impact capture rates of P. cinereus in Virginia (Keyser et al. 2004).

Seburn and Mallon (2017) highlight some of the challenges of using community science data to detect changes in abundance and occupancy. As a cryptic species, P. cinereus may not be well represented in community science data because of its low detection probability and therefore, may appear as if it is declining. When investigating potential declines of P. cinereus in Ontario, Canada over a 20-y period, Seburn and Mallon (2017) used natural cover searches and found that 84% of areas designated as unoccupied by community science data were actually occupied. From these findings, they suggest that at least 30 cover objects needed to be searched to attain 50% detection probability. The results of this study parallel other suggestions that accounting for detection is critical for determining the status (e.g., abundance or occupancy) of P. cinereus populations (e.g., Sutherland et al. 2016).

Petranka (1998) describes how little we know about the activity of salamanders underground and how these patterns vary geographically. Many of the studies cited by Petranka (1998) indicated that there is a seasonal pattern to vertical movement, with suitable soil moisture levels and temperatures driving movement to the surface during the spring and autumn in most locations (northern in Fig. 2). Petranka (1998) summarizes one study that found P. cinereus as deep as 1 m in winter and another study which demonstrated that P. cinereus not only relies on the burrows and tunnels made by other fossorial species to retreat underground, but is likely incapable of doing much more than enlarging existing burrows. Recent work by Cáceres-Charneco and Ransom (2010) confirms that P. cinereus uses earthworm burrows. Petranka (1998) did not mention any studies on natural dispersal or migration but he did summarize research documenting how P. cinereus can return to its original location after a 90-m displacement.

Based on Petranka’s (1998) summary, our own field experience with this species throughout its range, and the five recent supporting studies described here, we depict the surface activity phenology of P. cinereus in Fig. 2. There are at least three regional phenologies of surface activity that depend on latitude (north vs. south) and elevation (high vs. low). We hypothesize the latitudinal break between the northern and southern phenologies (Fig. 2) is at or near the extent of the Wisconsin ice sheet at the last glacial maximum (LGM; Fig. 1). In support of this hypothesis, Meshaka and Wright (2017) studied a southern Pennsylvania population of P. cinereus approximately 150–200 km south of the LGM ice edge and found individuals to have life history traits more similar to southern populations compared to northern ones (e.g., 75.9% of females at that site breed annually). Anthony and Pfingsten (2013: fig. 18-8) present data on the phenology of a single northeastern Ohio population using 6000+ observations collected over 4 yr. Their data closely match the northern phenology we suggest in Fig. 2 (Anthony and Pfingsten 2013). Leclair et al. (2008) examined a population near the northern range edge in Québec, showing a slight spring peak and a strong autumn peak in activity. This population also has a notably delayed life history (see foregoing discussion in Breeding and Courtship; Leclair et al. 2008). Finally, both Becker et al. (2013) and Hernández-Pacheco et al. (2019) have documented the southern phenology of continuous winter activity (Fig. 2) in Cary, North Carolina and Richmond, Virginia, respectively. Several sources and the references cited therein (e.g., Petranka 1998; Anthony and Pfingsten 2013; Stuczka et al. 2016) plus our own field experience suggest that P. cinereus is opportunistic about surface activity: If temperature and moisture conditions at the surface permit activity, some salamanders will be active, even in winter or summer. There is much to learn on this topic: We do not know how P. cinereus determines surface conditions from underground, what cues prompt vertical movement, or even how deep individuals can or will go during stressful seasons.

There is evidence that daily changes in ambient illumination play a role in cuing nightly emergence from leaf litter or cover object refugia. Wise and Buchanan (2006) summarize two of their previously unpublished field studies that conducted night censuses to determine the relationships between surface activity and ambient light. The first study found that surface activity of P. cinereus greatly increased 1–2 h after sunset, and was more strongly related to changes in ambient lighting than changes in temperature. In this study, the total duration of surface activity was positively related to humidity or moisture. The second study focuses more on the potential impacts of artificial night lighting on salamander activity. Wise and Buchanan (2006) describe an experiment that surveyed field transects with artificial light (10⁻² lux at leaf litter surface) and without artificial light (10⁻⁴ lux at leaf litter surface). Unlit transects had significantly more active salamanders during the first 1–2 h after dark, but later in the evening there were no significant differences between unlit and lit transects in the abundance of active salamanders. This suggests that increased artificial lighting delayed the onset of surface activity. Wise and Buchanan (2006) note that this delay in nightly emergence time could have several negative downstream effects, including reductions in: total foraging time, overall food intake, growth rates, reproduction, and survival during hibernation.

Additional studies that have examined natural and induced movements via displacement and homing to original locations have given some insight into movement patterns of P. cinereus. Seasonal movements of adults up a rocky hillside in autumn and down in spring were reported in New York (Woolbright and Martin 2014). The authors suggest that this movement is in response to increased availability of crevices in rocks uphill, allowing for more activity during winter months. A study using drift fences with pitfall traps found that the abundance of salamanders captured was not influenced by proximity to forest edges or streams, but roads likely acted as a barrier to dispersal (Gibbs 1998b). This is more rigorously supported by Marsh et al. (2005), who found that salamanders displaced across gravel and paved roads have reduced return rates to their original location in comparison to those returning across the same distance within continuous forest habitat. Similarly, small streams also reduce return rates of salamanders after displacement (Marsh et al. 2007; Snyder et al. 2022), and both roads and streams limit gene flow, suggesting these landscape features act as dispersal barriers for P. cinereus (Marsh et al. 2007, 2008; see also Hantak et al. 2019).

We still know relatively little about natural dispersal distances of P. cinereus. Petranka (1998), Williams et al. (1968), and Radomski et al. (2020) all speculate that P. cinereus is capable of moving great distances, since it presumably expanded from southern refugia to its current geographic extent in southern Canada (Fig. 1) during the last 10–15,000 yr after the Wisconsin glacier receded: a distance of at least 1000 km, suggesting dispersal rates are between 67 and 100 m (0.067–0.1 km) per year. Mark–recapture studies (e.g., Marsh et al. 2004, 2005, 2007) and studies of population genetic structure over relatively short distances in continuous forest or in habitat fragments separated by known dispersal barriers (Cabe et al. 2007; Marsh et al. 2007, 2008; Fisher-Reid et al. 2013; Cameron et al. 2019; Hantak et al. 2019) suggest that, although individuals generally do not disperse far, they are capable of longer movements across open fields, roads, or streams to return to familiar territory when displaced (25–55 m in Marsh et al. 2004, 2005, 2007; Snyder et al. 2022) or to colonize new habitat (5–25 m in Marsh et al. 2004). A recent observation increased the maximum distance that a single P. cinereus individual is known to have traveled, up to 143 m (Sterrett et al. 2015). Dispersing individuals are likely “floaters”—defined as adults without territories who readily colonize available habitat (Mathis 1991). Marsh et al. (2004) found that artificial habitat islands installed on an abandoned golf course (i.e., watered areas covered with shade cloth, supplemented with ACOs and leaf litter) were colonized within 1 yr of installation by young adult (based on SVL) P. cinereus in Virginia. Additionally, Snyder et al. (2022) found that displaced adults returned to their original location with greater frequency than did juvenile individuals.

The low lifetime dispersal distances of individuals likely helps P. cinereus persist in small forest fragments where other amphibians with greater dispersal distances have been extirpated (Gibbs 1998a). It is unclear if P. cinereus engages in sex-biased dispersal. Two studies, both conducted at MLBS in Virginia had conflicting results. Cabe et al. (2007) found no evidence of sex-biased dispersal using population genetic assignment indices. Liebgold et al. (2011) found evidence of females being more philopatric than males, confirmed through both lack of genetic structure at short distances in males and greater overall dispersal distances of males compared to females. A more recent study at a Maryland population using similar methods to Liebgold et al. (2011) did not find any evidence for sex-biased dispersal (Grant and Liebgold 2017). In an experiment exploring the movement behavior of individuals in unfamiliar environments, Cosentino and Droney (2016) found that individual movement behavior is highly repeatable and varies by forest age. In unfamiliar environments, individuals from older forest stands moved more than those from younger stands, and edge vs interior individuals within a stand did not show any movement differences. Individuals showed consistent movement patterns across time, suggesting movement personalities and heritable genetic variation for movement behavior (Cosentino and Droney 2016).

Our review of the literature on seasonal activity and dispersal shows that we now have a well-supported understanding of at least three surface-activity phenologies throughout the range of P. cinereus (Fig. 2). Awareness of the local phenology is important for appropriate scheduling of field sampling and experiments. Further studies show that P. cinereus can move substantial distances across the forest floor, but unless deliberately displaced, it seems that the majority of individuals do not move much within and between seasons (see also the Terrestrial Ecology: Territories and Home Range Size section). We do not know what abiotic cues drive movement—either vertically to/from underground and the soil surface, or horizontally during the surface-active seasons—although one study found changes in ambient light to be a potential cue for horizontal movement (Wise and Buchanan 2006). Methods discussed previously and in the following sections, such as tracking individuals with PIT tags (e.g., Sterrett et al. 2022) or fluorescent powder trails (e.g., Orlofske et al. 2009), may prove useful to investigate the abiotic cues and movement patterns more generally. Additional laboratory experiments following the methodological example of Cosentino and Droney (2016) could also further address these topics.

Petranka (1998) devoted several pages to summarizing research on territoriality in both sexes and age classes of P. cinereus, noting that P. cinereus is a model organism for studying the ecology and evolution of territorial behaviors. Much of the work he cites, and that which has been done since, has previously been thoroughly reviewed in Jaeger et al. (2016). To avoid duplication of effort, we focus this section of our review on recent studies that were either (1) independent of the R. G. Jaeger research program, (2) published after Jaeger et al. (2016), or (3) from non-MLBS locations. Reference information for all studies on this topic returned by our literature search are included in Supplemental Material S12 (Supplemental Data S11). We encourage those interested in territoriality, aggression, and related topics (e.g., chemical communication) in P. cinereus, to read Jaeger et al. (2016).

Briefly, there is ample evidence that both males and females defend territory, and that territorial behaviors (e.g., aggression, scent marking, avoidance of residents) vary both seasonally and by local abiotic (temperature, moisture) and biotic (food availability) conditions (Petranka 1998; Jaeger et al. 2016). At the time Petranka (1998) was published, estimates of home range sizes varied by study location. The average home range area in a Virginia population (MLBS) was estimated to be 0.16–0.33 m² with no significant differences between males, females, or juveniles (Mathis 1991), whereas the average home range area in a Michigan population was estimated to be 13 m² for males and juveniles and 24 m² for females (Kleeberger and Werner 1982). The latter study also reported daily movement of surface-active individuals (average = 0.43 m), with increases reported during and after precipitation events, and found that displaced individuals home at distances of 30–90 m (with declining success at longer distances; Kleeberger and Werner 1982). Petranka (1998) notes that the reason P. cinereus holds territories is unknown, but it is hypothesized that territories may be used to defend food, defend or attract mates, and/or defend suitable microhabitats. Petranka (1998) also acknowledges that a large proportion of individuals are floaters that do not hold territories (Mathis 1991; see also the Terrestrial Ecology: Seasonal Activity and Dispersal section). Larger individuals can successfully defend their territories from smaller individuals, and males who hold higher-quality territories are more likely to attract females. To mark their territories, P. cinereus use both fecal pellets and glandular secretions, and they are known to threaten and attack each other during territorial disputes (Petranka 1998; Jaeger et al. 2016). Interestingly, adult males are less aggressive to juveniles, and they are more tolerant of familiar intruders, which suggests the potential for kin recognition and/or a “dear enemy effect,” in which individuals tolerate near neighbors to reduce energy expended by acts of aggression (Petranka 1998; Jaeger et al. 2016).

Maerz and Madison (2000) offer a field-based study of territorial behavior from three non-Virginia populations, and specifically explore the relationship between territorial behavior and food availability. At a site in Pennsylvania with lower food density (as determined by total volume of prey in salamander stomachs), males and females were more territorial, exhibiting higher site fidelity, lower same-sex home area overlap, and body size and foraging success were positively correlated. At two sites in New York with higher food density there was lower site fidelity and more same-sex cohabitation of cover among females. Males in the high food environment still defended spaces, likely to attract mates (Maerz and Madison 2000).

Like many traits, there is geographic variation in territorial behavior; however, the scope and general patterns of that geographic variation are understudied (Jaeger et al. 2016). In the laboratory, Wise and Jaeger (2016) explored geographic variation in male aggression in P. cinereus using salamanders from eight populations from the center and southern parts of the range (Indiana, Maryland, Virginia, and West Virginia). They deliberately selected populations that were previously associated with two documented allozyme clades (Hass 1985), which now likely belong to at least three mitochondrial clades (see discussion in Radomski et al. 2020). Wise and Jaeger (2016) found significant variation in aggressive and submissive behaviors among males from different genetic groups and in individuals from different elevations. Individuals of one genetic group were more submissive than the other genetic group. The variation in territorial behaviors due to elevation was more complex, showing a significant negative correlation between submissive behavior and elevation, and a significant resident advantage (i.e., intruders are more submissive) that was independent of elevation. Anthony and Pfingsten (2013: table 18-3) summarize the variation in aggressive behavior in laboratory experiments across several Ohio populations as compared to one New York and one MLBS population. Interestingly, Ohio (central to the range) shows some of the longest average bouts of aggressive posture, whereas results from New York and MLBS are both shorter than Ohio and very similar to each other.

There is ample evidence from field studies that supports widespread expression of territorial behavior in P. cinereus. For example, field studies from across the geographic range show positive relationships between cover-object size and SVL (e.g., Moore et al. 2001; Hickerson et al. 2004); majority singleton observations under cover objects over opposite or same-sex pairs (e.g., Anthony et al. 2008); site fidelity (e.g., Gillette 2003; Reiter et al. 2014; see below); homing behavior to cover objects (e.g., Marsh et al. 2005; Ousterhout and Liebgold 2010; Martin and Hantak 2011); and invasion of experimentally vacated cover objects by smaller individuals (e.g., Marsh et al. 2004; Anthony and Pfingsten 2013). All of these studies are discussed in more detail in other sections of this review.

In addition to the comparisons of Ohio territorial behavior to New York and MLBS by Anthony and Pfingsten (2013), there are a handful of observations that suggest that expression of territorial behaviors in P. cinereus may be reduced in northern populations. One study (Quinn and Graves 1999b) and one observation (Placyk et al. 2000), both from the upper peninsula of Michigan, have reported large aggregations of P. cinereus under a single cover object throughout the active season (May through September). The observation by Placyk et al. (2000) is difficult to interpret in the context of territoriality, because sizes and sexes of the aggregating individuals are not reported. In Ontario, Rollinson and Hackett (2015), found male salamanders to be moderately aggressive in laboratory trials, but contrary to predictions from territoriality theory, resident salamanders were not more aggressive than intruders. However, intruding salamanders did exhibit more escape behavior than did residents, as expected from a territorial species. In the field, the authors did not observe any aggregations under cover objects and found a random spatial distribution of adults in 16 out of 17 visits between May and July, suggesting weak to no territoriality (Rollinson and Hackett 2015). Rollinson and Hackett (2015) hypothesize that the reduction in territoriality might be caused by less restricted daily foraging activity in northern populations compared to southern populations; however, this hypothesis has not yet been explicitly tested.

One of us (K.L. Grayson) has seen multiple large aggregations (n ≥ 7) of adult P. cinereus (>34 mm SVL) under a single coverboard at a low-elevation southern site (Richmond, Virginia; Supplemental Material S16; Supplemental Fig. S2). Aggregations are common throughout the active season (October to March; Fig. 2). This location also has some of the highest reported densities for this species (Hernández-Pacheco et al. 2019). Thus, it is not clear that latitude alone explains why Michigan and Ontario salamanders appear to be less territorial than Ohio and high-elevation Virginia salamanders. As Wise and Jaeger (2016) found, there is probably a genetic component to territoriality: the less territorial populations observed in Michigan, Ontario, and Richmond, Virginia, are likely all members of the Northern clade (Radomski et al. 2020). Additional work in both the field and laboratory is needed to both document and understand the causes of variation in expression of territorial behavior across the range of P. cinereus.

Chemical communication is thought to be very important in territory establishment and retention, identification of individuals, and even determination of potential predation risk (see the Predators and Defense section). In Virginia, Simons et al. (1999) identified and described several mucous and serous glands in both the shoulder and postcloacal region of P. cinereus. Importantly, the authors compared the glands in the two regions, hypothesizing that the postcloacal region would show differences if those glands were important for scent marking and communication, as earlier studies had suggested. The shoulder and postcloacal glands had notable differences in both morphology and physiological status over time (full, empty, renewing) in response to individuals being placed on novel substrates. Simons et al. (1999) present evidence to support the hypothesis that the shoulder glands show morphology and physiological status consistent with the proposed ancestral function of serous glands in amphibians (predator defense), whereas the postcloacal glands have morphology and physiological status consistent with territorial scent marking. Simons et al. (1999) was followed by Hecker et al. (2003) in New York, who focused on the postcloacal mucous and two types of serous glands (S1 and S2). Hecker et al. (2003) hypothesized that the S1 serous glands are for scent marking, and the S2 serous glands are used in nutrient storage and defense.

Duhaime-Ross et al. (2013) tested both visual and chemical communication in adults and juveniles from Ontario, and found complex interactions between social context, cue type, and age class, suggesting multimodal communication is common. Wise and Buchanan (2006) summarize a previously unpublished study by Buchanan that measures territorial behavior under different lighting conditions: moonlight (10⁻² lux), starlight (10⁻⁴ lux), and total darkness (10⁻⁵ lux). Buchanan found that residents increased their use of visual displays with increasing ambient light (Wise and Buchanan 2006). Dawley et al. (2000, 2006) studied the cellular physiology of the olfactory and vomeronasal systems across seasons in a Pennsylvania population. Both studies found an increase of vomeronasal cellular proliferation in May and June, and Dawley et al. (2006) found a subsequent increase of vomeronasal apoptotic activity in July, suggesting the importance of these cells during the nonbreeding early summer months, when territorial behavior is believed to be strongest (Jaeger et al. 2016). Additional studies focused specifically on P. cinereus interpreting and responding to chemical cues from predators are summarized in the Predators and Defense: Defensive Behaviors section.

Site fidelity has been explored in Michigan, Ohio, and Virginia. In Michigan, Placyk and Graves (2001b) found the same individual under the same cover object 1 yr apart. In Ohio, Martin and Hantak (2011) document a low level of homing behavior (∼16% return, similar to reports in Jaeger et al. 2016) in both striped and erythristic individuals displaced 5 m from their original cover object. Larger individuals are more successful at homing in Virginia (Ousterhout and Liebgold 2010). This study also found low levels of between-year movements (0.85–1.22 m/yr), suggesting high site fidelity (Ousterhout and Liebgold 2010). Gillette (2003) found year-to-year site fidelity in Virginia, as did Reiter et al. (2014) in Ohio. Reiter et al. (2014) found individual residency (number of days between first and last capture under the same cover object) to range from 10 to 1281 d, with 46 individuals showing residencies longer than 700 d (four active seasons). To this we add that we have all seen some level of site fidelity across both seasons and years in our long-term research plots in Massachusetts, Virginia, Michigan, Pennsylvania, and New Jersey (M. C. Fisher-Reid, K. L. Grayson, L. S. Mead, T. H. J. Matlaga, and S. C. Sterrett, respectively, personal observations), but we have not yet quantified these patterns. Anecdotally, we expect formal analyses of our data will yield patterns similar to Reiter et al. (2014).

Recently, with the introduction of visual implant elastomer (VIE) marking methods (Heemeyer et al. 2007), studies have explored field-based space use in P. cinereus. In Virginia, Schieltz et al. (2010) found that adults often use more than one cover object, regardless of spacing between cover objects, with males never occupying adjacent cover objects. Muñoz et al. (2016b) used spatial capture–recapture (SCR) to understand space use, dispersal patterns, and density in Pennsylvania. This study found mean densities of 0.423–0.473 salamanders/m². Contrary to what is reported in Petranka (1998), Muñoz et al. (2016b) found males to use larger areas than females, regardless of season. Dispersal distances between seasons averaged <1 m, again suggesting high levels of site fidelity (Muñoz et al. 2016b). As described above, young adult floaters will quickly take over new habitat when it becomes available (Marsh et al. 2004; Anthony and Pfingsten 2013).

In Ohio, two studies explored territoriality differences between color morphs, and a third studied differences between sympatric P. cinereus and P. electromorphus.Reiter et al. (2014) found that striped morphs were more aggressive and held on to their territories for longer periods of time than unstriped morph, with both laboratory and field support for these patterns. A follow-up field study by Anthony et al. (2017) found that striped morph territories had 32% higher quantity of prey than unstriped morph territories. Other work in this population suggests that these differences in territories and territorial behavior between color morphs may play a role in mate choice (e.g., Anthony et al. 2008; Acord et al. 2013). Deitloff et al. (2009) explored the territorial behavior of sympatric and allopatric P. cinereus and P. electromorphus in response to heterospecific intruders in Ohio. Residents did not change behavior based on the source of their intruder; however, residents from sympatric populations of both species showed increased aggression toward heterospecific intruders than residents from allopatric populations.

In reviewing the territoriality literature on this species, it is clear that much of what we know stems from a small portion of the very large geographic range, 60% of the initial batch of studies we categorized into this topic are from MLBS or other Virginia locations (n = 21/35; Supplemental Material S12; Supplemental Data S11, Supplemental Material S15; Supplemental Data S14). Future work that tests previously documented patterns at other locations is imperative to first, fully understand the social behavior of P. cinereus across its range and second, tease apart abiotic and biotic factors that influence the expression of social behavior. We also have much to learn about the details of chemical communication, from the compounds and glands involved to the types of information P. cinereus regularly communicates about and with whom, and the geographic consistency of these patterns. Both Anthony and Pfingsten (2013) and Jaeger et al. (2016) provide excellent starting points for what we do know about chemical communication in P. cinereus, especially as it relates to territoriality, kinship, and mate choice.

In the course of our review, we found several studies that we initially classified as belonging to the primary category “other,” with further comments that these studies related to physiology, behavior, other methods, or sensory ecology. Those studies that remained difficult to place in other sections of our review are summarized here. Generally, these sources focus on experimental methods not directly related to field sampling salamanders and behavior that is not directly related to breeding and courtship, foraging, territoriality (all discussed above), or antipredator tactics (discussed in the next section).

Three studies documented specific methods or approaches that may be useful for those designing experiments with P. cinereus. First, using salamanders from a Michigan population, Placyk and Graves (2001c) tested the ability of P. cinereus to detect and respond to different wavelengths of light. Because the majority of P. cinereus behavior is nocturnal, the authors hoped to find a wavelength under which humans could observe salamanders behaving, but which could not be perceived by the salamander, to better simulate night conditions in experiments. When presented with a visual stimulus under broad spectrum fluorescent (400–700 nm) or black (4–400 nm) light, salamanders spent significantly less time looking at the stimulus under black light. Placyk and Graves (2001c) therefore suggest black light conditions will effectively simulate night conditions for future experiments.

Second, Wise and Buchanan (2006) reviewed the literature on salamander responses to artificial light. As part of this review, they criticize the experimental design of Placyk and Graves (2001c) and argue the results are difficult to interpret because (1) light treatment acclimation times that were too short, (2) bright light treatments had unrealistically high illuminations, and (3) there was missing information on the illumination and spectral properties of all treatments. The critiques of Placyk and Graves (2001c) by Wise and Buchanan (2006) apply to many of the studies the latter reviewed, and thus the authors make methodological suggestions for those wishing to conduct experiments on the impacts of light on salamander behavior: (1) use realistic, and thus ecologically relevant, artificial light levels that would be experienced by wild populations; (2) ensure full reporting of the intensity and spectral properties of their treatments. Wise and Buchanan (2006) conclude their chapter by noting that the physiological and behavioral responses of salamanders to artificial night lighting is understudied, particularly on longer time scales in both the field and the lab.

Third, using salamanders from MLBS, Novarro et al. (2017) expand upon earlier work by Lowe (2004) to determine what concentration of tricaine methanesulfonate (MS-222) works best for short-term anesthesia of P. cinereus.Lowe (2004) had found that, of five plethodontids tested, P. cinereus had the longest times to anesthetization, often 2–3× longer than the next longest, P. elongatus, and suggested higher concentrations might be needed for efficient anesthetization of P. cinereus. Novarro et al. (2017) determined the optimal MS-222 concentrations for rapid anesthetization (i.e., induction) time and a range of initial and full recovery times. In addition to publishing details of these concentrations and timings in their Table 1, Novarro et al. (2017) also explored the relationship between several morphological traits and induction time. Both tail length and mass significantly and positively affected induction time and should be factored into experiments requiring short-term anesthetization of P. cinereus.

Other studies that were difficult to categorize focused on the physiology of P. cinereus. Davis et al. (2009a) found that the roundness of red-blood cells is strongly positively correlated with body condition and body size. The authors suggest that taking blood samples might be a useful nondestructive method of tracking population health over time (Davis et al. 2009a). Fluorescent powder pigment has been used with success to track amphibian, reptile, and small mammal movement over short durations (e.g., Lemen and Freeman 1985; Graeter et al. 2008; Roberts and Liebgold 2008). Orlofske et al. (2009) was the first study to test explicitly how the fluorescent powders affect respiration and metabolism in a lungless amphibian using P. cinereus. They found that there was no impact on either oxygen consumption or standard metabolic rate (SMR). Instead, it appeared that the stress of being handled was greater than the stress of being temporarily covered in powder (Orlofske et al. 2009). Homyack et al. (2010a) found that SMR increased with both body mass and temperature. Novarro et al. (2018) complements this result by assaying corticosterone (CORT) of salamanders living at different temperatures. Salamanders housed at 25°C had higher CORT, levels indicating higher levels of stress, compared to those housed at 15°C. Additionally, salamanders housed at the warmer temperature ate more food than salamanders housed at the cooler temperature (Novarro et al. 2018). Takahashi and Pauley (2010) observed altitudinal differences in resource allocation. Salamanders from higher elevations (i.e., cooler climates) allocated more energy to storage than growth or reproduction compared to salamanders from lower elevations. Biddle et al. (2017) looked at factors that impact growth, using mark–recapture methods to track individuals over multiple years and seasons. Color morph had no effect on growth, but they did find an effect based on both initial size and environmental fluctuations. Small juvenile salamanders who experienced milder winters showed the greatest growth (Biddle et al. 2017).

From a purely behavioral perspective, at the end of Jaeger et al. (2016), the authors note the new work being done on salamander cognition and cognitive ecology. One of the first studies to explore this topic is Uller et al. (2003), which documents rudimentary numerical discrimination in P. cinereus. In choice trials containing one, two, three, four, or six food items, salamanders regularly opted for the larger quantity of food in one vs two and two vs three forced-choice tests, but could not distinguish between three vs four or four vs six. Kight et al. (2005) explored learning in P. cinereus using classical conditioning. Individuals from New Jersey were successfully conditioned to low-frequency sound and vibration, but not to visual or chemical stimuli. As Jaeger et al. (2016) note, cognition and cognitive ecology is an area of P. cinereus research that is particularly wide open for new questions and experiments.

The last two behavioral studies that were difficult to categorize both explored the social context of juvenile salamander behavior. Liebgold and Cabe (2008) explored if juvenile growth rates were affected by familiarity or relatedness of the adults nearby at MLBS. The results varied by year: in the first year, familiarity had a strong effect on mass and size, but not the following year. The authors hypothesize that temperatures may have played a role in the different familiarity results, as the first year was significantly cooler than the second year. Additionally, there was no evidence that familiarity or spatial proximity were indicative of relatedness, or that relatedness to nearby adults affected juvenile growth rates (Liebgold and Cabe 2008). Liebgold (2014) followed up on this by testing the effect of the presence of territorial adults on juvenile behavior using salamanders from Michigan (presumably nonterritorial) and Virginia (presumably territorial, see the Terrestrial Ecology: Territories and Home Range Size section). Michigan juveniles were more strongly affected by being housed with territorial adults. In juvenile–juvenile interactions, Michigan juveniles showed increased investigatory and escape behaviors compared to juveniles housed with nonterritorial adults.

In sum, much like in 1998, terrestrial ecology of P. cinereus remains a large and prominent focus of recent research. In nearly all instances, the topics summarized here are only just scratching the surface of our understanding of the ecology of this species, and offer many excellent starting points of background knowledge, useful methodology, and testable hypotheses for continued research to resolve some of the more challenging or elusive questions. Important to future study of terrestrial ecology topics will be expansion of research programs to investigate geographic and phylogeographic variation in terrestrial ecology, and experiments to determine the relative roles of phenotypic plasticity and local adaptation.

Petranka’s (1998) section on predators and defense in P. cinereus was brief, and there has been much work exploring these topics since. Petranka (1998) focused on describing general predators of P. cinereus (i.e., birds, snakes, spiders), with an emphasis on snakes, and summarized the experimental studies that have focused primarily on snake predators (Garter Snakes, Thamnophis spp., and Ringneck Snakes, Diadophis punctatus). Petranka (1998) describes the results of one study that suggests that Ringneck Snakes may be able to distinguish the scent of P. cinereus from nonedible prey and even distinguish particular parts of the body (i.e., tail vs. body; Lancaster and Wise 1996). In response to a predator, P. cinereus individuals generally change their behavior (i.e., staying motionless or coiling) or autotomize their tails (Petranka 1998). A measurable proportion of P. cinereus populations appear to have broken or regenerating tails, although it is not clear if this indicates predation rates, rates of intraspecific conflict, or, most likely, some combination of both (Jaeger et al. 2016). Anthony and Pfingsten (2013) concur in their review that crypsis in the form of immobility and tail autotomy are both primary defense strategies but add that noxious skin secretions (e.g., Brodie et al. 1979) may be another primary defensive trait for P. cinereus. In the following sections, we first discuss new knowledge on the predators and parasites of P. cinereus. Next, we discuss a handful of studies which have explored differential predation of color phenotypes of P. cinereus. Finally, we discuss the many studies which have explored P. cinereus defensive behaviors. There are many studies relating to skin and gut microbial communities, particularly how they are involved in defense against the amphibian fungal pathogen Bd. These have been summarized in their own subsection, Additional Topics: Microbiota.

Since Petranka (1998), a small number of studies have documented unusual predators for P. cinereus. Jung et al. (2000b) observed a Rove Beetle (Platydracus vidatus) preying upon live P. cinereus under a cover board in Maryland. The Creek Chub (Semotilus atromaculatus) was observed preying on P. cinereus, likely after the salamander was washed into the creek during a flash flood (Sharp et al. 2005). As noted earlier, Hughes et al. (1999) found two salamanders dead inside Pitcher Plants (Sarracenia purpurea) in bogs in West Virginia and Pennsylvania. These observations are incidental, but there are likely many species that find P. cinereus palatable, particularly woodland snakes and ground-foraging birds. Anthony and Pfingsten (2013) have a detailed summary of known instances of specific predators consuming P. cinereus. Most studies of antipredator responses use chemical cues from T. sirtalis fed a diet of P. cinereus, other salamander species, or earthworms. These are summarized in the next two subsections.

There are conflicting reports as to whether Northern Dusky Salamanders (D. fuscus) eat P. cinereus. Jaeger et al. (1998) found evidence to suggest that D. fuscus does eat P. cinereus, and that P. cinereus actively avoids encounters with D. fuscus. However, a follow up study by Ransom and Jaeger (2006) found no evidence of either predation or tail loss caused by D. fuscus, although P. cinereus still seemed to avoid D. fuscus in mesocosms. Other large salamanders likely to prey on P. cinereus where their ranges overlap include D. quadrimaculatus, Gyrinophilus porphyriticus, and Pseudotriton ruber (Petranka 1998).

Anthony and Pfingsten (2013) also reference and summarize Casper’s (2005) summary of parasites reported to infect P. cinereus, and indeed, there has been more work done on the parasites of P. cinereus since Petranka (1998). Muzzall et al. (1997) and Bolek and Coggins (1998) document several species of trematode, nematode, and protozoan parasites on P. cinereus. Most salamanders sampled in Muzzall et al. (1997) hosted at least one parasite, and a trematode (Brachycoelium spp.) was the most common parasite in P. cinereus from New Hampshire. In Wisconsin, Bolek and Coggins (1998) found lower overall rates of infection than Muzzall et al. (1997), but a nematode species (Rhabdias spp.) was the most common parasite on P. cinereus. In Michigan, Gillilland and Muzzall (2002) again found high rates of helminth infection, but no evidence that these parasites caused limb deformities. Most studies that have tested wild populations have found low prevalence of the amphibian pathogens Ranavirus or Bd on the skin of P. cinereus (e.g., Harris et al. 2009; Gratwicke et al. 2011; Wunder et al. 2012; Muletz et al. 2012, 2014; Augustine and Neff 2016), except Jongsma et al. (2019), which found much higher than typical (but still low overall) rates of Bd infection in P. cinereus from New Brunswick, Canada.

Rickettsial intracellular pathogens may also influence P. cinereus ecology and health. A rickettsial intraerythrocytic pathogen of the family Anaplasmatacea was found in 16.7% of 102 P. cinereus individuals tested from Pennsylvania, New York, and Virginia (Davis et al. 2009b). Infected salamanders had significantly larger snout–vent length and more circulating white blood cells than uninfected salamanders. Males in these populations were more likely to be infected than females. The authors hypothesize that the trombiculid mite, Hannemania dunni, a common amphibian ectoparasite, is the likely vector of the rickettsial pathogen, as these mites are the only ectoparasite which has been observed parasitizing plethodontid salamanders (Davis et al. 2009b). The male bias in infections also suggests a trombiculid mite, as they were previously documented parasitizing plethodontid salamanders differentially by sex (Anthony et al. 1994).

A follow-up study by Davis and Cecala (2010) focused on Desmognathus ocoee and found an obligate intracellular Rickettsia spp. at a much higher prevalence (35%) in D. ocoee compared to previous data on P. cinereus (8%–17%). Morphologically, the parasite infecting D. ocoee appeared to be a different type of Rickettsia spp. than the one found in P. cinereus by Davis et al. (2009b). The difference in average prevalence between Desmognathus spp. (30%) and Plethodon spp. (8%; see Table 2 in Davis and Cecala 2010) was significant, and the authors suggest that this difference is probably due to the difference in the habitat use of the two genera. Plethodon are completely terrestrial, whereas Desmognathus are semiaquatic; therefore these species are likely exposed to and contract different species of Rickettsia, which have different vectors and life cycles (Davis and Cecala 2010).

Predation risk for P. cinereus may be influenced by color phenotype, although patterns across the range of P. cinereus are complex and variable. There is some evidence that the color morphs respond differently to predators. In Ohio, Venesky and Anthony (2007) found that striped salamanders spend more time in the “all-trunk-raised” defensive posture in the presence of neonate T. sirtalis than unstriped salamanders, and unstriped salamanders were more mobile than striped salamanders. A concurrent look at museum specimens showed that unstriped morphs more frequently showed evidence of tail autotomy than striped morphs (Venesky and Anthony 2007). These differences in behavior and tail autotomy suggest that the two morphs experience different rates of predation, with the authors hypothesizing that unstriped morphs experience greater predation by T. sirtalis (Venesky and Anthony 2007). After simulated predation (i.e., a tail clip), Otaibi et al. (2017) found that the autotomized tails of striped morphs move longer and faster than those of unstriped morphs.

Although differential predation by color is well documented, which morph gets predated on more often seems to depend on the population and the predator (see additional discussion in the foregoing Systematics and Geographic Variation section). In the range of polymorphic P. ventralis in Tennessee, Fitzpatrick et al. (2009) found that the rare morph was attacked less than the common morph, arguing for negative frequency-dependent selection on color phenotype. This was further supported in a Massachusetts population of P. cinereus by Kraemer et al. (2016; see Systematics and Geographic Variation section). In Maryland, Grant et al. (2018) found positive frequency-dependent selection to favor the more common unstriped morph. In this population, the rarer striped morph individuals were attacked more often by avian predators and had lower survival rates over 3 yr compared to the more common unstriped morphs. However, there was no observed difference in survival between color phenotypes when mammalian predators were present (Grant et al. 2018). In Pennsylvania, Davis and Milanovich (2010) observed higher levels of hematological stress markers in unstriped individuals compared to striped, which may be explained by unstriped morphs being exposed to predator attacks more often. In Ohio, Hantak and Kuchta (2018) found that the striped morph had overall better camouflage, based on reflectance measurements taken in the field, but there was variation in camouflage by season, population, and body region.

Finally, there has been some work exploring the potential Batesian mimicry of the red eft stage of the Eastern Red-spotted Newt (Notophthalmus viridescens) by erythristic P. cinereus (Fig. 1C; Kraemer and Adams 2014; Kraemer et al. 2015a,b). In Pennsylvania, high frequencies of the erythristic morph (range = 21%–43.3%) have been found in populations that are syntopic with the eft stage of N. viridescens (Cassell and Jones 2005), supporting the possibility of selection for Batesian mimicry. Kraemer et al. (2015a) tested for four general patterns expected if a Batesian mimicry co-evolutionary dynamic existed between N. viridescens and erythristic P. cinereus. All four of these were supported, and additional studies by Kraemer and colleagues investigated specific details of the mimicry system. Kraemer and Adams (2014) suggest that mimicry by erythristic P. cinereus may be due to differences in color, not brightness. Using a visual model of predator spectral sensitivities and of the colors of newts, striped, and erythristic P. cinereus, Kraemer and Adams (2014) found that only birds could possibly discriminate between newts and striped P. cinereus based upon color, not snakes or mammals. Importantly, the visual model suggested that birds would not be able to differentiate between newts and erythristic morphs, suggesting bird predation would be the primary source of selection in a Batesian mimicry scenario. Kraemer et al. (2015b) found that there was no relationship between the toxicity of the local N. viridescens population and the frequency or presence of erythristic P. cinereus.

A number of predator-related cues, both simulated and natural, have been documented to alter behavior in P. cinereus, and presumably reduce the frequency of predation in nature. In Michigan, Mathis and Lancaster (1998) demonstrated that exposure to water that previously held stressed conspecifics led to P. cinereus spending more time trying to escape the study chamber relative to individuals exposed to control water. Another Michigan study followed up on this by showing that P. cinereus avoided the skin secretions of stressed conspecifics for up to 2 mins after the alarm pheromone had been collected (Graves and Quinn 2000). Sullivan et al. (2003) found that P. cinereus from New York avoid chemical cues from injured salamanders (both conspecific and heterospecific), but do not avoid chemical cues from noninjured salamanders. Earthworms share common vertebrate predators with salamanders (woodland snakes and birds), and thus could be a source of information on general presence of predators; however, salamanders did not respond to chemical cues from either injured or noninjured earthworms (Sullivan et al. 2003). More recently, also in New York, Sullivan and Jensen (2013) found that after exposure to different concentrations of damage-released conspecific cues, P. cinereus significantly decreased their activity compared to a control, but did not discriminate between the differing concentrations of the cue. After simulated predation by tail clips, Roberts and Liebgold (2008) found that individuals climbed higher than those without simulated predation in the field, especially male individuals.

Most defensive behaviors in P. cinereus have been induced using Garter Snake (Thamnophis spp.) chemical cues (Petranka 1998; Jaeger et al. 2016). Murray and Jenkins (1999) tested how P. cinereus responded to cues from T. sirtalis fed different diets in the lab. Using salamanders from Massachusetts, they showed that salamanders avoided the chemical cues of T. sirtalis compared to a control, regardless of what prey (P. cinereus or earthworms) the snakes had been fed prior to collecting the cue; however, they avoided the P. cinereus–fed snake cues more strongly than the earthworm-fed snake cues. Using salamanders from New York, Madison et al. (1999b) found the avoidance behavior of P. cinereus in response to chemical cues from T. sirtalis to be quite complex. Salamanders generally avoided substrates that smelled like T. sirtalis; however, their response varied by what the snake had eaten (conspecifics, earthworms, or nothing) and the time of day (night, day). During the day, when T. sirtalis is more active, salamanders avoided snake-scented substrates in all diet treatments. At night, when T. sirtalis are less active, salamanders only avoided the snake-scented substrates of those snakes who were fed conspecifics. The impact of time of day was not due to ambient lighting, suggesting circadian regulation is involved, in addition to detecting predator cues that are further modulated by predator diet (Madison et al. 1999b).

Studies that examine salamander responses to Common Garter Snake cues have mostly been done using P. cinereus individuals from upstate New York. Following up on Madison et al. (1999b), Madison et al. (2002), found similarly complex interactions between the time of day, and the combination of cues from snakes fed different diets and stressed or undisturbed conspecifics. Maerz et al. (2001) found that foraging activity decreased with exposure to chemical cues from T. sirtalis. Sullivan et al. (2002) found that P. cinereus hid out in refugia when exposed to chemical cues from T. sirtalis fed P. cinereus in both the lab and field; however, time spent in the refugia was significantly shorter in the field than in the lab. This study also speculated that presence of other individual salamanders under a cover object may provide additional protection by keeping salamanders under cover objects when predatory cues are present (Sullivan et al. 2002). A follow-up study supports this speculation by showing that P. cinereus avoided chemical cues from Common Garter Snakes fed Eurycea bislineata (Sullivan et al. 2004). The authors suggest that closely related prey species have similar antipredator behaviors and thus P. cinereus can avoid the cues of predators who have recently eaten either conspecifics or closely related species (Sullivan et al. 2004). Sullivan et al. (2005) further describe how the ability of P. cinereus to avoid chemical cues of predators is dependent upon both genetic and ecological relatedness of the prey which was consumed.

As noted by Petranka (1998) and in multiple studies already discussed, P. cinereus are known to autotomize their tail in the presence of a threat, and since then, researchers have learned more about the frequency, timing, and effects of this behavior. Wise and Jaeger (1998) found that in territorial contexts, a salamander’s personal tail loss did not have a large influence on either a resident or intruder’s behavior, but they are both more aggressive when facing an opponent without a tail. Meche and Jaeger (2002; see Breeding and Courtship section) found that males did not distinguish between tailed or tailless females in the lab; however, in the field, they found males associated with tailless females more often. Meche and Jaeger (2002) offer two hypotheses to explain this result (after rejecting male preference based on their lab results): that females without tails prefer to associate with males, or that male aggression from the courtship process itself results in lost tails for females.

Yurewicz and Wilbur (2004) found that tail loss is energetically costly to females; those with lost or regenerating tails made fewer ova. Meshaka and Wright (2017) found that males and juveniles had a higher incidence of broken or regrown tails than females in a collection of museum specimens from south central Pennsylvania. Given that Madison et al. (1999a) documented the importance of having a tail for proper clutch suspension, it is likely that females are under stronger selection to retain their tails than either males or juveniles. This could mean that they are more behaviorally cautious, avoid risky situations, or are less likely to autotomize their tails than males or juveniles. As noted previously, Otaibi et al. (2017) found that the autotomized tails of striped morphs move longer and faster than those of unstriped morphs, suggesting striped morph tails may be more effective at distracting predators than unstriped morphs. Hessel et al. (2017) found no significant difference in jumping or mobility between individuals with and without tails.

Although the sources cited above indicate regular investigations of tail autotomy, there has been very little study on tail regeneration in P. cinereus (e.g., timing, physiological mechanisms, role of external factors on regrowth).Wise et al. (2022) tested the impacts of different levels of artificial light at night (ALAN) on tail regeneration, and found a complex, nonmonotonic pattern. Using four light treatments (10⁻⁴, 10⁻², 1, and 100 lux), they found that tail regeneration was significantly faster for salamanders exposed to 10⁻⁴ and 1 lux treatments compared to the 10⁻² and 100 lux treatments. These authors also explored the potential interaction of differences in prey consumption under different light treatments on tail regeneration; however, they did not find any significant differences in the amount of prey eaten between treatments, suggesting the impacts of ALAN on tail regeneration are independent of prey consumption effects on tail regeneration (Wise et al. 2022).

Finally, Liebgold (2019) documented one case of a female with a bifurcated tail at MLBS in Virginia. Both tail ends had regenerating tips, which indicated that autonomy and regrowth can occur in both the main and supernumerary tails. Two of us (T. J. H. Matlaga and K. N. Streeter) have also seen tail bifurcation during our own field work in Pennsylvania (Supplemental Material S16; Supplemental Fig. S3, available online). Additional work on the context of tail loss and regeneration across the range of P. cinereus is needed, specifically whether tail loss occurs primarily due to predation or intraspecific conflict. Research directions to fill the gaps in our comprehensive understanding of tail loss and regeneration in P. cinereus can be grouped into four main areas: (1) the selective pressures acting on tail loss and retention; (2) the behavior of autotomized tails; (3) the physiological and genetic mechanisms of tail regeneration; and (4) any morph, age, or sex-specific differences in tail autotomy or regeneration.

In reviewing the broader literature on predation of P. cinereus, it is clear that we have very few direct, nonincidental observations of predation on P. cinereus. The observations we do have are often single studies by scientists focused on the predator, not the prey, and most of these observations predate Petranka (1998), often by several decades (see summary in Anthony and Pfingsten 2013). The paucity of observations is likely, in part, because of the nocturnal nature of P. cinereus, but creative approaches should be able to address this challenge. For instance, collaboration with those who study the presumed vertebrate predators (birds, snakes, mammals, and larger salamanders) to complete targeted sampling of predator stomach contents during seasons of peak P. cinereus surface activity (Fig. 2) would help confirm the general palatability of P. cinereus, as well as contribute to our understanding of what proportion of predator diets are represented by P. cinereus. As we suggested earlier for refining our understanding of the diet of P. cinereus, the use of molecular methods, such as DNA metabarcoding (e.g., Harms-Tuohy et al. 2016), on predator stomach contents or predator fecal matter could greatly expand our ability to identify P. cinereus as a prey item as well as give us a relative idea of its prevalence in predator diets. We also need to expand our consideration of likely predators beyond vertebrates to include large arthropods (Anthony and Pfingsten 2013), and it is likely that DNA metabarcoding of arthropod gut contents and feces will also be an effective way to identify the invertebrates who are eating P. cinereus (e.g., Sint et al. 2015).

Our lack of knowledge of who specifically eats P. cinereus makes it difficult to draw conclusions about differential predation on and observed antipredator behaviors of color morphs. Many studies on this topic assume visual predators who can differentiate between the two color morphs in much the same way humans do. Kraemer and Adams (2014) show that only birds are likely to have this visual perception, not snakes or mammals. Because we have ample evidence that P. cinereus can detect chemical cues from predators, it would be interesting to test predators for chemical discrimination of salamander prey items more thoroughly, and for possible chemical discrimination of striped and unstriped morphs. Preliminary work in this area has been done with ringneck snakes (Lancaster and Wise 1996). Methods for these types of experiments with snakes are well established; for example, in the work of Brodie, Brodie, and colleagues on the predator–prey relationships of garter snakes (Thamnophis spp.) and newt species that produce tetrodotoxin (e.g., Avila et al. 2012). These studies often use the tongue-flick attack score method (reviewed by Cooper and Burghardt 1990).

It is also clear that our knowledge of P. cinereus parasites and pathogens is, at best, preliminary. Much of the work in this field has focused exclusively on testing for Bd or Ranaviruses. Surveillance testing for these pathogens has clear value for conservation of herpetofauna generally, but there are either no data on the impact or incidence of these pathogens on P. cinereus populations (e.g., Winzeler et al. 2014) or evidence of extremely low incidence of these pathogens within P. cinereus populations (e.g., Muletz et al. 2014; Augustine and Neff 2016; see discussion of Bd resistance in Additional topics: Microbiota). It will be important for future studies to follow up on the limited information we do have on the ecto- and endoparasites of P. cinereus to determine prevalence, transmission routes, natural susceptibility and symptomology, and natural resistance levels. Even from these few sources, we already have evidence suggesting a case of sex-biased vector transmission of a pathogen (Anthony et al. 1994; Davis et al. 2009b) and that there may be latitudinal (and thus likely climatic) differences in infection rates (Jongsma et al. 2019).

Petranka’s (1998) section on community ecology is limited to the interactions of P. cinereus with other salamanders, particularly congeners most closely related to it. The section primarily summarizes interactions (or lack thereof) with its sister species, Plethodon shenandoah, work largely done by Jaeger in the 1970s (Jaeger 1970, 1971a,b, 1972, 1974). According to Petranka’s (1998) summary of Jaeger’s work, P. shenandoah and P. cinereus are strongly segregated on the three Virginian mountain tops that P. shenandoah occupies, with P. shenandoah restricted to the talus habitat, and P. cinereus found in every other terrestrial habitat available on the mountains. It was thought that there was coexistence in the talus of Hawksbill Mountain; however, Petranka (1998) recounts Jaeger (1970), which suggests that this is only apparently true, as Jaeger (1970) found strong microspatial segregation between these two species. Jaeger (1970) found P. shenandoah existing in five talus isolates surrounded by continuous nontalus habitat occupied solely by P. cinereus. Jaeger (1970) notes that these are not actually cases of coexistence, as within the isolated talus, there is only P. shenandoah, and outside of the talus, there is only P. cinereus. Petranka (1998) goes on to review what we know about competitive behaviors between these species, and notes that the field and laboratory studies are often contradictory, making it difficult to determine what is happening. Petranka (1998) offers Jaeger’s (1972) hypothesis that perhaps adult P. cinereus exclude juvenile P. shenandoah. The remainder of this section of Petranka (1998) briefly discusses three additional interactions P. cinereus may (or may not) have with other salamander species: (1) potential for food competition between P. cinereus and either P. hoffmani or P. wehrlei, but lack of support when studied; (2) documented aggressive interactions between P. cinereus and both juvenile P. glutinosus and adult Desmognathus ochrophaeus; and (3) laboratory predation on P. cinereus by Ambystoma maculatum in 9% of staged encounters.

There has been much additional work on the community ecology of P. cinereus since Petranka (1998); importantly, these studies now regularly include interactions with nonsalamander species. In this section, we start by reviewing the more recent research on P. shenandoah and P. cinereus, continue by reviewing studies on interactions between P. cinereus and other salamander species, and finally, we review those studies focused on how P. cinereus interacts with nonsalamander species.

The same year that Petranka’s book was published, Griffis and Jaeger (1998) found further evidence supporting interspecific territoriality in P. cinereus but were unable to document any behavioral differences between these two species in biting, threat posture, or submissive posture. Evidence supporting the interspecific territoriality of P. cinereus came from a competitive release experiment, which removed P. cinereus residents from rocks close to a P. shenandoah isolate (experimental treatment) or left the P. cinereus residents in place (control treatment) and tracked how often and how quickly P. shenandoah were able to invade the experimental vs control rocks. Experimental rocks were invaded more quickly and more frequently by P. shenandoah than control rocks, suggesting active defense by P. cinereus against P. shenandoah (Griffis and Jaeger 1998).

Additional recent research further strengthens Jaeger’s (1972) argument that P. cinereus adults may exclude P. shenandoah juveniles. Myers and Adams (2008) observed phenotypic shifts between allopatric and sympatric populations of P. cinereus and P. shenandoah but no evidence of character displacement between sympatric populations. This observation suggests competitive exclusion is driving the segregation, restricting P. shenandoah to talus, an environment uninhabitable by P. cinereus because of its susceptibility to desiccation (Petranka 1998; Myers and Adams 2008). As noted earlier in Systematics and Geographic Variation, early morphological and genetics work on P. cinereus and P. shenandoah found limited evidence of introgression or hybridization (see summary in Carpenter et al. 2001).

Additional investigation into competitive interactions between the species found that, under current climate conditions, P. shenandoah lost more mass during competition than P. cinereus, indicating P. cinereus is competitively superior to P. shenandoah (Dallalio et al. 2017). Predicted future climate warming conditions suggest a more costly outcome for P. cinereus, though, as average mass loss was comparable for the two species under presumed future (warmer) temperatures. However, during intraspecific trials, P. cinereus experienced no difference in mass loss between current and future conditions. This suggests P. cinereus has a larger range of temperature tolerance than P. shenandoah (Dallalio et al. 2017). Finally, Chambert et al. (2018) found that although P. cinereus populations were relatively stable over 4 y, P. shenandoah experienced high turnover in occupancy.

Regarding competition with salamanders other than P. shenandoah, Petranka’s (1998) comments are limited. At that time, research summarized by Petranka (1998) indicated that interspecific competition for food between P. cinereus and P. hoffmani is unlikely and that dietary overlap between P. cinereus and P. wehrlei occurs but competition for food is unknown. Petranka (1998) also briefly summarizes interactions of P. cinereus with two more distantly related salamanders. Adult P. cinereus display similar levels of aggression towards and defend territories against juvenile P. glutinosus as they would towards adult conspecifics. Desmognathus ochrophaeus defends cover objects from P. cinereus and drives P. cinereus out of occupied territories (Petranka 1998).

Studies since Petranka (1998) have further described interactions between P. cinereus and other closely related salamander species. Multiple studies have examined one-on-one interactions between P. cinereus and other small-bodied Plethodon, looking for evidence for character displacement or competitive exclusion. Character displacement in both head morphology and diet is supported for P. cinereus and P. hoffmani (Adams 2000; Adams and Rohlf 2000), but not for teeth number or jaw muscle mass (Swart and Adams 2004). Although the microhabitats between these two species are not significantly different (Carlson et al. 2016), Jaeger et al. (2002b) found behavioral differences that may drive character displacement: P. cinereus is more aggressive and less submissive than both allopatric and sympatric P. hoffmani. They suggest that the boundary between the two species is either static, because of the aggressive superiority of P. cinereus and the exploitative superiority of P. hoffmani from eating larger prey, or that P. cinereus is slowly intruding on the range of P. hoffmani (Jaeger et al. 2002b).

Character displacement has also been supported for sympatric P. cinereus and P. nettingi (Adams et al. 2007; Adams 2011). Both species are smaller when in sympatry than when in allopatry, and sympatric populations show greater divergence in head shape between species relative to allopatric populations. These results suggest interspecific competition as a mechanism for phenotypic variation among Plethodon species (Adams et al. 2007). Over 32 yr, P. nettingi has decreased in abundance, which Kroschel et al. (2014) found may be correlated to P. cinereus expanding their range into the higher elevations that P. nettingi originally inhabited, but they could not rule out changes in environmental conditions affecting both species.

In Ohio and Kentucky, the ranges of P. cinereus and P. electromorphus overlap. In Ohio, sympatric localities within the broader counties of sympatry are not common, suggesting competitive interactions between these two species are important in determining species presence (Deitloff et al. 2008). In intraspecific trials, P. cinereus was more aggressive than P. electromorphus, and in interspecific trials, P. electromorphus was more submissive. These results suggest P. electromorphus may not be able to expand its range into that of P. cinereus, but P. cinereus may invade the range of P. electromorphus (Deitloff et al. 2008). In sympatric populations, both P. cinereus and P. electromorphus exhibited more aggressive behaviors and less submissive behaviors than allopatric populations (Deitloff et al. 2009). This study, however, only looked at one sympatric and one allopatric population, and the authors called for future research over the full extent of the overlapping ranges. Site-specific morphological differences also occur in sympatric P. cinereus and P. electromorphus populations (Deitloff et al. 2013). At some sympatric locations the two species experienced morphological divergence in head shape, while at other sympatric locations, morphological convergence occurred. These results suggest that abiotic and biotic disparities at different locations may be responsible for the morphological variation (Deitloff et al. 2013).

Hedeen (2000), working in both Kentucky and Ohio, found similar results to Deitloff et al. (2008): sympatry between P. cinereus and P. electromorphus was uncommon, and there seem to be climatic differences that correspond to differences in distribution. Plethodon electromorphus was found in drier upslope environments, whereas P. cinereus was found in wetter, lower, flatter areas (Hedeen 2000). Note that the populations of P. electromorphus sampled by Hedeen (2000) were previously classified as P. richmondi and are referred to as such by Hedeen (2000). Highton (1999) reclassified these populations as P. electromorphus based on allozyme data.

The ranges of P. cinereus and P. hubrichti are parapatric, with the range of P. cinereus completely surrounding that of P. hubrichti. Aasen and Reichenbach (2004) found no significant changes in relative proportions of P. cinereus and P. hubrichti over 10 yr at three populations in Virginia (one allopatric site for each species, and one sympatric site). Arif et al. (2007) found P. hubrichti to be more aggressive and less submissive than P. cinereus, but no differences in diet were observed between these two species. Head shape between these two species was different, but there was no evidence of ongoing character displacement in sympatric populations (Arif et al. 2007). The authors hypothesize P. hubrichti restricts the range of P. cinereus, even though it lives in suitable microclimates for P. cinereus. This hypothesis was further supported by Kniowski and Reichenbach (2009) which found that sympatric P. hubrichti and P. cinereus were similar in microhabitat preferences, movement patterns, growth rates, and adult and juvenile survival rates but differed in size. Sympatric P. hubrichti neonates and adults were larger than P. cinereus neonates and adults (Kniowski and Reichenbach 2009). A follow-up study observed non-fatal biting by a larger P. hubrichti of a smaller P. cinereus, suggesting interference competition contributes to the geographic distribution of these species (Kniowski and Pietsch 2015).

Several studies explored the position of P. cinereus within the larger woodland salamander community, focusing on its interactions with stream specialists (e.g., Desmognathus spp. and Eurycea spp.) and large-bodied Plethodon like P. glutinosus. As described previously in the Predators and Parasites section, it is unclear if D. fuscus preys on P. cinereus (Jaeger et al. 1998; Ransom and Jaeger 2006). Jaeger et al. (1998) found no evidence that larger P. glutinosus eat smaller P. cinereus, but a later study found that when sympatric, competition between P. glutinosus and P. cinereus led to reduced growth in juveniles of both species (Price and Secki Shields 2002). Two other studies support exclusion of P. cinereus by Desmognathus spp. Grover (2000) found little habitat overlap between Plethodon and Desmognathus, but some between Plethodon and Eurycea. Grover and Wilbur (2002) found that D. fuscus displaced P. cinereus closer to streams, and both abundance and mass of P. cinereus increased with distance from streams. When limited to a near-stream habitat, where stream specialist E. cirrigera likely outcompetes it, the body condition of P. cinereus individuals declined over a 3-wk period (Pasachnik and Ruthig 2004). A removal experiment at a stream/forest ecotone found that removal of P. cinereus had no effect on either P. cinereus abundance or Desmognathus sp. abundance; however, E. cirrigera moved closer to streams when P. cinereus was removed (Ransom and Jaeger 2008).

Plethodon salamanders are hypothesized to use behavioral mechanisms, such as climbing behaviors, that promote coexistence. A field study conducted by Mezebish et al. (2018) observed the climbing frequencies of three size guilds of Plethodon species. The authors found that the smallest species, P. cinereus, climbed more frequently than both the intermediate species, P. montanus, and the largest species, P. glutinosus. Plethodon cinereus also increased climbing frequency when exposed to intraspecific competitors but decreased climbing when exposed to interspecific competitors (Mezebish et al. 2018). The differences in climbing behaviors in different-sized Plethodon suggest that behavioral niche partitioning may be a mechanism that reduces competitive interactions and allows for the coexistence of species. In the lab, although adult P. glutinosus and P. cinereus did not directly interact when placed together, P. cinereus increased nose tapping behaviors when in the presence of P. glutinosus (Mezebish et al. 2018). This suggests that the individuals were chemically aware of the presence of the other. This study also found that more climbing behavior was observed in all species when there was more understory vegetation available to climb on.

Several studies since Petranka (1998) have described interactions of P. cinereus with nonsalamander species, focusing on (1) the potential roles of earthworms beyond simply being a prey item for P. cinereus, (2) the relationships between P. cinereus and intraguild predators common to the forest floor and soil (e.g., spiders, centipedes, beetles), (3) the role of P. cinereus in regulating soil and leaf-litter invertebrate communities, and (4) indirect interactions between P. cinereus and other woodland species.

Earthworms are likely to be important to P. cinereus community ecology in addition to being a prey item (e.g., Maerz et al. 2005, 2006; see the previous section on Terrestrial Ecology: Diet and Feeding Behavior). The work by Ransom and colleagues, conducted at MLBS in Virginia, focuses on the relationships between native and nonnative earthworms (Annelida) and P. cinereus. Cáceres-Charneco and Ransom (2010) found that P. cinereus uses earthworm burrows as refuge. In both lab and field, they found that P. cinereus was more likely to be under cover objects when earthworms were absent, and more likely to be in burrows when earthworms were present (Cáceres-Charneco and Ransom 2010). Ransom (2017) also observed that salamanders do not seem to distinguish between native vs nonnative earthworms, consuming and using the burrows of all species similarly. Ransom (2012) defines three distinct relationships between earthworms and P. cinereus: (1) earthworms are ecosystem engineers, building burrows that salamanders use regularly; (2) earthworms are potential prey for salamanders; and (3) earthworms compete with the other microinvertebrate prey species of P. cinereus leading to indirect effects on salamanders. Ransom (2012) found that the strength of these relationships varies by season for the non-native earthworm Lumbricus terrestris. Positive effects of ecosystem engineering were important to overwinter survival of adult P. cinereus. The positive effect of earthworms as prey was only greater than the negative effect of earthworms as competitors in the summer (Ransom 2012).

In Michigan, a survey by Brunges et al. (2020) investigated the effect of 11 different invasive earthworm species on terrestrial salamander abundance. They found a negative association between salamander counts and the presence of invasive, epigeic (i.e., found in the leaf litter) earthworms but no effect of invasive endogeic (i.e., found underground) earthworms. The authors suggest that this is because epigeic earthworms consume the O-horizon leaf litter layer, which P. cinereus depends on for moisture and protection (Brunges et al. 2020). Similarly, in Virginia, Ransom (2017) found that the presence of the invasive endogeic earthworm, Octolasion tyrtaeum, had an indirect, negative effect on nighttime P. cinereus counts. This earthworm species was also associated with a narrower O-horizon layer suggesting a similar mechanism to that proposed by Brunges et al. (2020). Finally, two studies in Ohio found that nonnative, epigeic Amynthas earthworms alter microhabitats and behavior of P. cinereus. The first found that earthworm presence decreased leaf litter mass over time, leading salamanders to increase movement, cover object use, and co-occurrence under cover objects with earthworms (Ziemba et al. 2015). A follow-up study found further evidence the salamanders were negatively impacted by Amynthas and Metaphire spp. (Ziemba et al. 2016). In the lab, salamanders opted for poorer-quality microhabitats and consumed fewer flies when earthworms were present. In the field, salamanders and earthworms shared cover objects much less than expected, and earthworm abundance was negatively associated with juvenile and male salamander abundance (Ziemba et al. 2016).

In Ohio, several studies have explored relationships between P. cinereus and spiders, centipedes, and carabid beetles. Anthony et al. (2007) found that P. cinereus exhibited more submissive behaviors when exposed to native centipedes as opposed to introduced centipedes, suggesting the presence of learned predator avoidance behaviors in P. cinereus. Burgett and Smith (2012) exposed P. cinereus to centipedes or conspecifics, and found salamanders showed increased aggression towards centipedes regardless of their residency status but showed no increase of aggression towards conspecifics in either setting. Hickerson et al. (2012) did a predator-removal experiment with the following treatments: P. cinereus removal, centipede removal, and a control with no removals. Interactions were complex: Removing either predator changed the abundance of other invertebrate predators (spiders and beetles) and had a mix of effects on invertebrate prey species. There was a negative relationship between spider or centipede abundance and P. cinereus abundance, suggesting interference competition between salamanders and the two invertebrate predators (Hickerson et al. 2012). However, there was a positive relationship between P. cinereus and carabid beetles. Hickerson et al. (2012) note that these relationships are strongest during the dry summer months, when these species all have fixed territories tied to suitable microhabitats. Interestingly, in the lab, Gall et al. (2003) found evidence to support intraguild predation by carabid beetles on P. cinereus. Residents of both species behaved more aggressively than intruders and attacked intruders of both species at similar levels. Plethodon cinereus bitten by carabid beetles often produced defensive, adhesive secretions which immobilized the beetles (Gall et al. 2003). Hickerson et al. (2004) found similar levels of interspecific aggression in laboratory mesocosms containing centipedes and P. cinereus. In this study, salamanders often escalated interactions to bite intruder centipedes, and did not exhibit biting behavior towards intruding conspecifics (Hickerson et al. 2004). Similarly, Hickerson et al. (2018) found that spiders in the genus Wadotes are also perceived by P. cinereus to be competitors, not as potential predators or prey. In laboratory trials, P. cinereus reacted to intruder Wadotes individuals in much the same way as they reacted to conspecific intruders, and there was no evidence of intraguild predation between spiders and adult or juvenile salamanders (Hickerson et al. 2018).

Plethodon cinereus is an important regulator of the composition of soil-litter invertebrate communities and thus, indirectly, the ecosystem processes these invertebrates perform. These interactions are thought to be crucially important for forest health, and often are used to justify the importance of woodland salamanders more generally (e.g., Welsh and Droege 2001; Davic and Welsh 2004). However, the community-level impacts of P. cinereus can depend on the habitat and climate context (and see Additional Topics: Ecosystem Ecology for broader impacts). In Ohio, Walton (2005, 2013) and Walton and Steckler (2005) all found that P. cinereus does have a top-down regulatory effect on the composition of soil-litter invertebrate communities, but the magnitude and direction of these effects varied both seasonally and annually (Walton 2013). An earlier, 2-yr study found that when leaf-litter mass and moisture were both high (typically in the spring), salamanders exerted stronger, negative top-down effects on mesofaunal invertebrates (Walton 2005). In complementary microcosm experiments, when P. cinereus was present, Walton and Steckler (2005) found more podomorphic Collembola, fewer enchytraeid worms, and fewer macroinvertebrate detritivores. Total invertebrate density, and densities of Collembola and Acari (mites) increased more quickly when salamanders were present. The authors argue that salamanders are either preying on larger detritivores that compete with these mesofaunal groups, releasing them from competition, or that salamander fecal matter promotes microflora growth, leading to bottom-up effects that promote mesofaunal population growth (Walton and Steckler 2005). A follow-up microcosm study found that salamander treatments again increased the density of Collembola, but this time only in urban leaf litter, not rural leaf litter (Walton et al. 2006). Other taxa, such as oribatid mites, pseudoscorpions, isopods, and millipedes (Walton et al. 2006), experienced decreased densities in both urban and rural leaf litters when salamanders were present.

Walton (2013) examined these patterns over a longer, 6-yr period and synthesized the complex relationships between P. cinereus and the mesofaunal invertebrates with which they share the leaf litter. During this longer study, Walton (2013) found positive, negative, and no effects of P. cinereus on invertebrate densities, with strong seasonal variation in strength and direction of the effects of salamanders on several invertebrate taxa. For example, the same negative relationships between P. cinereus and mesofaunal invertebrates observed in Walton (2005) were only observed in the spring and became positive relationships in autumn in the longer study (Walton 2013). Walton (2013) convincingly argues two complementary hypotheses to explain the change in dynamics: seasonal changes in prey-selection behavior by P. cinereus and seasonal changes in intraguild territoriality between P. cinereus and arthropod predators. Both hypotheses are mediated specifically by seasonal change in leaf-litter thickness, and the resulting interaction webs for spring and autumn are well summarized in Walton (2013: fig. 7). We refer the reader to Walton’s (2013) detailed discussion of the current support for these two hypotheses as well as starting points for additional experimental tests needed for further understanding.

In contrast, in a 2-yr study in Virginia, Homyack et al. (2010b) found no evidence that P. cinereus affected the number of invertebrate taxonomic groups or functional guilds. In New Hampshire, Hocking and Babbitt (2014) found over 4 yr that P. cinereus had no effect on intraguild predator abundance (carabid beetles, centipedes, and spiders). Walton (2013) offers testable hypotheses that may explain the lack of patterns found by Homyack et al. (2010b) and Hocking and Babbitt (2014). Additionally, Hickerson et al. (2017) provide a detailed argument that because of the differences both in methodology and duration of Homyack et al. (2010b) and Hocking and Babbitt (2014) compared to those studies discussed previously that do find strong patterns, the former are not directly comparable to the latter. We are sensitive to the Hocking and Babbitt (2014) argument that caution is warranted when using P. cinereus as an ecological indicator within a given forest until the local patterns at that site are sufficiently understood. Future studies that use standardized invertebrate sampling methods across sites and detailed, longitudinal data sets that explicitly incorporate seasonal and annual variability (e.g., like Walton 2013) will be important for generalizing the role of plethodontid salamanders in communities and ecosystems.

Finally, a small number of studies have looked at the indirect relationships between P. cinereus and other woodland species, including White-tailed Deer (Odocoileus virginianus), Garlic Mustard (Alliaria petiolata), Spongy Moths (Lymantria dispar dispar), and fungal communities. In Ohio, Greenwald et al. (2008) found a positive, indirect effect of deer presence on P. cinereus abundance. Salamanders were three times more abundant in deer-permitted control plots compared to deer-exclusion plots. Greenwald et al. (2008) suggest that deer herbivory increases invertebrate species richness through increased nutrient deposition and plant growth, which in turn attracts salamanders. In Massachusetts, Brooks (1999) found that the abundance of P. cinereus did not change across forests with higher or lower densities of White-tailed Deer. In Ohio, removal of invasive Garlic Mustard had no effect on P. cinereus abundance (Smith 2018). Mitchell (2015) explored how P. cinereus counts changed after invasion by Spongy Moths. Despite recovery of the canopy and increased amounts of coarse woody debris, Mitchell (2015) observed a >60% decline in P. cinereus counts in 1993 compared to 1988 (prior to the 1989 arrival of Spongy Moths). It is not clear if the decline in P. cinereus was related to the arrival of Spongy Moths, other factors, or some combination of these. For interactions between P. cinereus and the soil microbial community, please see the forthcoming section Additional Topics: Microbiota.

In the decades since Petranka (1998), the community ecology of P. cinereus is one of the fields which has grown the most, but for which there are still many unanswered questions. We encourage the reader interested in these topics to also read Bruce (2008) which thoroughly reviews interspecific competition and intraguild predation in Plethodontidae more generally. Bruce (2008) reviews several studies cited both here and by the three prior reviews (Petranka 1998; Anthony and Pfingsten 2013; Jaeger et al. 2016), offering conclusions which feature potential avenues for future work. Broadly, Bruce (2008) highlights the need for long-term, time-series studies that target the relationship between population growth rates and densities of Plethodon spp., as well as the continued development and use of manipulative experiments, emphasizing a need for those in unenclosed field plots. Specifically, regarding P. cinereus, Bruce (2008) points out that although we have a good basic understanding of the top-down impacts of P. cinereus on their prey, little is known about how those prey exert bottom-up control on P. cinereus populations.

In addition to better understanding interspecific competition and intraguild predation within Plethodon, continued study of the interactions between P. cinereus and its close relatives in Plethodon will hopefully help us understand not only how P. cinereus was able to spread so far and wide across northeastern North America, but also how reliable of a stand-in P. cinereus biology could be for its less abundant, less wide-ranging relatives. The crucial role of P. cinereus as an invertebrate mesopredator in northeastern woodlands suggests they interact with many other species, both directly and indirectly, both positively and negatively, and the research reviewed here only begins to reveal the complexity of these dynamics. Importantly, the variation in P. cinereus dynamics across both time and space makes generalizing their community-level impacts difficult. The lack of standardized methods for tracking such impacts as well as the short-term nature of most investigations further compounds these difficulties. These difficulties make the “ecological indicator” designation (e.g., Welsh and Droege 2001; Siddig et al. 2019) of P. cinereus harder to sell, but we argue that the variation and complexity in P. cinereus community dynamics offer countless opportunities for broader ecological research on these topics, lending support to our argument that P. cinereus has model organism potential, regardless of its use as or the specific requirements for an indicator species. A coordinated effort that (1) spans several years and surface-active seasons, (2) is replicated across the range of P. cinereus, and (3) monitors and measures salamander and invertebrate densities and relevant abiotic covariates (e.g., leaf litter biomass, soil moisture) will be critical in establishing baseline, range-wide understanding for future experimentation. Walton (2013) and Hickerson et al. (2017) both provide robust methodology and foundational knowledge on which to build this research program.

Petranka’s (1998) summary of the conservation biology of P. cinereus was very brief, focusing on sensitivity of P. cinereus to soil acidification (e.g., Wyman 1988) and intensive forest harvesting practices (e.g., deMaynadier and Hunter 1995). Historically, there has been very little concern for the impacts of human-mediated effects on P. cinereus populations, in part because of the then-growing understanding that P. cinereus are remarkably abundant in some forests, with biomass greater than other prominent vertebrate groups (e.g., New Hampshire, Burton and Likens 1975a). However, shortly after the publication of Petranka (1998), Lannoo (2005) was published, comprehensively reviewing contemporary amphibian declines, their hypothesized causes, potential conservation, survey and monitoring strategies, and the statuses of United States species. Highton’s (2005) chapter within this volume documents the declines of 44 Plethodon species across 205 populations, including P. cinereus. Highton (2005) reports that 19 out of 24 populations of P. cinereus across eight states (Indiana, Maryland, New York, North Carolina, Tennessee, Pennsylvania, Virginia, and West Virginia) show declines in mean number of individuals collected per person from visits pre-1990 (range = 1956–1986) to visits post-1990 (range = 1990–1999), some of which are quite dramatic (see Highton 2005: table 8-1). These patterns, and those of global amphibian declines more generally (e.g., Alford and Richards 1999; Lannoo 2005; Caruso and Lips 2013), are alarming and suggest that the historically high abundances of P. cinereus in northeastern North American forests should not be taken for granted.

Like many of the topics we have reviewed thus far, the conclusions of conservation-focused studies of P. cinereus are geographically variable, with many studies finding contradictory patterns, suggesting the importance of understanding both land use history and current anthropogenic threats to best determine the likelihood of persistence for any given population. We found many recent papers that have looked at how P. cinereus responds to specific kinds of anthropogenic change, as well as many studies that focus on how to manage P. cinereus populations or use P. cinereus to manage forest ecosystems. We have separated the more management-focused sources to be summarized under the Additional Topics: Management section, and here focus largely on concerns related to how P. cinereus responds to anthropogenic land use changes and climate change. We also limited our discussion of papers that have conservation as a secondary category (Supplemental Material S12; Supplemental Data S11), as these studies are nearly all discussed elsewhere in this review.

Since Petranka (1998), many studies have further explored the sensitivity of P. cinereus to anthropogenic influences, despite low prioritization of this species as a conservation concern. deMaynadier and Hunter (1998) found that P. cinereus in Maine are sensitive to clear cutting and edge effects, and have an affinity for mature forest. Further, it was suggested that modified, open canopy sites may represent sink habitat, meaning there are significantly more juveniles than adults and these juveniles may not have established territories. Despite this finding, at populations in Connecticut and Massachusetts, Brooks (2001) found that P. cinereus abundance only decreases temporarily, both when Eastern Hemlock (Tsuga canadensis) is partially harvested as well as when invasion of Hemlock Wooly Adelgid (Adelges tsugae) causes Eastern Hemlock defoliation and tree loss. Within 3 yr of initial tree loss, the abundance of P. cinereus had already begun to increase (Brooks 2001). Also working in Connecticut, Mossman et al. (2019) found that abundance of P. cinereus was the highest in 20–25-yr-old shelterwoods and significantly higher than in mature or unharvested forest sites, suggesting P. cinereus may benefit from some infrequent level of disturbance, particularly disturbance that results in increased availability of coarse woody debris, like timber harvesting. This contrasts with the findings of Rodewald and Yahner (1999) in Pennsylvania, who found that harvested areas had no salamanders, and the abundance of P. cinereus in mature forest was positively correlated with the number of trees. At Appalachian populations in southwestern Virginia and neighboring West Virginia, Knapp et al. (2003) compared declines in salamander abundance after clear-cutting to six other silvicultural methods, and found that if the canopy was disturbed, the other silvicultural methods were as bad as clear-cutting for salamander abundance. After a forest fire in Nova Scotia, only one P. cinereus individual was found alive about 300 m from the edge of the wildfire in an unburned refuge (Russell 2011). No salamanders were found in the burned area when searched over the next 4 mo. This suggests that P. cinereus may be negatively affected by wildfires in terms of habitat loss; however, given their ability to persist in subterranean retreats during a wildfire, they likely could move to areas nearby that were not affected by the wildfire once it is safe to do so (Russell 2011).

Various modeling approaches by Leroux et al. (2017) suggest that patch size is more important for persistence of P. cinereus populations than either patch isolation or environmental conditions (but see the following discussion of conservation genetics). A model by Venier et al. (2007) based on P. cinereus in Ontario concluded that P. cinereus are sensitive to changes in habitat availability, regardless of whether the disturbance is caused naturally or from intensive harvesting. As canopy cover and downed coarse woody debris increases, P. cinereus abundance increases (Venier et al. 2007). In Indiana, however, while Peele et al. (2017) also found these factors to be positively related to the abundance of P. cinereus, neither one had a significant effect on the survival rates of the P. cinereus. Otto et al. (2014) found similar importance of canopy cover and coarse woody debris availability in Michigan at the landscape scale, but not at the local scale. A model by Gordon et al. (2012) found that P. cinereus abundance decreases following forest harvesting in Ontario. Vanderwel et al. (2011) modeled the value of different habitats following deforestation, using P. cinereus data from Québec and Vermont. The habitat value for P. cinereus increased after most stages of deforestation, likely due to increases in coarse woody debris, suggesting that P. cinereus can recover after disturbances to their environment. A field study in West Virginia by Wood and Williams (2013) found that reclaimed mountaintop removal mines are not suitable habitat for P. cinereus, as these habitats favor succession towards a grassland or shrubland habitat than forest. They argue that the recovery of the forest habitat after mountaintop mining may take longer than areas that are clear-cut because of factors like soil compaction.

In Virginia, Marsh and colleagues have published several studies on the effects of roads on P. cinereus, many of which we have previously discussed in this monograph (Marsh and Beckman 2004; Marsh et al. 2005, 2007, 2008). From a conservation perspective, both Marsh and Beckman (2004) and Marsh (2007) examine the edge effects of roads. The first study found strong edge effects—fewer P. cinereus were found close to roads in all three populations studied. The authors suggest that the decline of P. cinereus as you approach a road edge is due to the parallel decline in both soil moisture and cover object area near roads (Marsh and Beckman 2004). Marsh (2007) focused on gated (closed to vehicles) vs. ungated roads and found no edge effects on P. cinereus from roads that are narrow and gated, while ungated roads show the same pattern as observed in Marsh and Beckman (2004). Marsh et al. (2005, 2008) both find that roads are dispersal barriers: Marsh et al. (2005) used a displacement and homing experiment and Marsh et al. (2008) measured population genetic structure on different sides of roads of different widths. In Connecticut, Gibbs (1998b) also found roads to be a dispersal barrier for P. cinereus.

As reviewed earlier, several studies have explored the relationship between habitat fragmentation and population genetics. Gibbs (1998c) compared continuous to fragmented populations in Connecticut using random amplified polymorphic DNA (RAPD) markers and found that the fragmented populations were more divergent, that is, showed greater genetic structure, compared to the continuous populations. Gibbs (1998c) also noted a difference in morph frequency: fragmented populations were close to 50:50 striped:unstriped, whereas continuous populations were nearly 100% striped. In Québec, Noël et al. (2007) found that urban P. cinereus have both lower allelic richness and lower heterozygosity. A follow-up study found that P. cinereus showed more genetic differentiation in areas where anthropogenic disturbances have persisted for a longer time (Noël and Lapointe 2010). The authors suggest that this pattern is because P. cinereus in these environments tend to have a low effective population size. Further implications from Noël and Lapointe (2010) are that it is important to preserve large, forested areas within cities to increase genetic variation and conserve or increase effective population size. Cosentino and Brubaker (2018) demonstrate the importance of connectivity for post-agricultural recovery. Salamanders were affected by historical land use and were more abundant in areas of historical forest cover. This suggests connectivity between relic source populations and regenerating forests on postagricultural lands is important for colonization of the new forest habitat (Cosentino and Brubaker 2018).

In Ohio and Pennsylvania, Cameron et al. (2019) compare the genetic differentiation and population structure between fragmented populations (Cleveland, Ohio) and continuous populations (Allegheny National Forest, Pennsylvania) to understand the relative roles of genetic drift and effective population size better. In the fragmented landscape, populations did show genetic differentiation and structuring; these patterns were best explained by spatial variation in effective population size and driven by one of the four sampled populations. The three non-differentiated fragmented populations showed levels of gene flow comparable to the four continuous populations. Interestingly, the genetically distinct population in the fragmented landscape had the lowest effective population size, but the largest allelic richness and highest observed heterozygosity within that landscape. The authors argue that this may be due to multiple paternity in P. cinereus (see Liebgold et al. 2006) and/or the presence of unsampled populations within the fragmented landscape. In the continuous landscape, populations were not genetically differentiated, lacked structure, and differences in allele frequency between continuous sites were best explained equally by matrix resistance and geographic distance, similar to Noël et al. (2007). Cameron et al. (2019) conclude that the ability of P. cinereus to persist in marginal habitats likely allows them to maintain levels of gene flow in highly modified habitats that are similar to those observed in larger forests.

Using a combination of contemporary field measurements of SVL and historical measurements from museum specimens, Caruso et al. (2014) found that the body size of P. cinereus has decreased by 7% in several Appalachian Mountain locations over 55 yr. The greatest decreases in body size were in the southern Appalachians, which have had the greatest increase in both drying and warming. The authors hypothesize that this correlation between location and body size likely comes from an increase in metabolic expenditure (Caruso et al. 2014). In contrast, using only museum specimens but including specimens from both high and low elevations, McCarthy et al. (2017) reported a similar precipitation pattern, with body size being negatively correlated precipitation in the driest month, but the opposite temperature pattern: larger P. cinereus were correlated with higher mean maximum July temperatures, and specimens on the coastal plain showed a temporal increase in body size correlated with a temporal increase in mean maximum July temperatures.

A third study using only museum specimens looked at morph-specific change in SVL at 37 locations over 43 yr, at high and low elevations (Hantak et al. 2021). This study was summarized in detail in the Systematics and Geographic Variation section; however the patterns documented by Hantak et al. (2021) are more similar to Caruso et al. (2014) than to McCarthy et al. (2017), with both morphs showing decreases in SVL with increases in temperature and increases in SVL with increasing elevation. The strength of these changes was different between the two morphs, with striped morphs responding more strongly to temperature, and unstriped morphs responding more strongly to elevation. Striped morphs also showed a decrease in SVL over time, while SVL in unstriped morphs is increasing over time (Hantak et al. 2021).

Specifically in response to Caruso et al. (2014), and prior to the publication of McCarthy et al. (2017) or Hantak et al. (2021), both Grant (2015) and Connette et al. (2015) offer alternative explanations for shrinking salamanders. They argue that the results of Caruso et al. (2014) could be explained by the sampling approaches used in the study and that both museum specimens and contemporary surface-active individuals likely only represent the surface-active portion of the inferred population. Indeed, in their discussion, McCarthy et al. (2017) argue that they designed their study specifically to address Grant’s (2015) criticisms of Caruso et al. (2014). The most recent study by Hantak et al. (2021) suggests complex patterns driving body size responses to abiotic factors, and an impact of color morph, which was not considered in Caruso et al. (2014) or McCarthy et al. (2017). Methodologically, the two more recent studies are difficult to compare. McCarthy et al. (2017) grouped specimens by county and into two time brackets (1950–1970 and 1980–2000), and used mean maximum July temperature for each county in the two time brackets to explore how body size changed in counties which experienced warming vs. counties which did not experience warming. Hantak et al. (2021) focused on 37 geolocations which had sampling over a minimum time series of 20 y, grouped geolocations separated less than 15 km into eight composite sites, and used mean annual temperature for each geolocation and each year of specimen collection. Further work on this topic, which tracks contemporary populations and includes metrics like body size, individual growth rates, and changes in mass across the range over time using spatial capture–recapture methods (e.g., Sutherland et al. 2016) may be able to detect these patterns more confidently.

Although P. cinereus is not itself considered a species of conservation concern, it is clearly affected by a variety of land use modifications and climate change, both of which are often anthropogenically caused. Declines in P. cinereus populations have been documented (Highton 2005), and population and conservation genetic approaches discussed in detail above suggest that the evolutionary resiliency (sensu Sgrò et al. 2011) of P. cinereus may be at risk from continued changing land use and climates. Plethodon cinereus also clearly has strong impacts on the ecological community (see the Community Ecology section) such that even slight changes in salamander abundance could have much larger ripple effects throughout the community and ecosystem (Gaston and Fuller 2008). Regular monitoring of common, abundant, not-at-risk species like P. cinereus will help conservation biologists detect declines earlier and have a better chance to mitigate their effects (e.g., Gaston and Fuller 2008; Gaston 2010, 2011; Lindenmayer et al. 2011, 2013; Sterrett et al. 2019).

In addition to the miscellaneous physiology and behavior sources we report on previously, we found four additional topics not covered by Petranka (1998) for which there has been enough literature to warrant their own sections in the present review. First, we summarize studies on the ecosystem ecology of P. cinereus. Next, we summarize studies that focus on the natural resources management of P. cinereus and the forests in which they live. Third, we describe the many studies exploring the skin and gut microbiota of P. cinereus and their relationship to disease resistance. Finally, we found several ecotoxicology studies that explore the tolerance of P. cinereus to various anthropogenic pollutants, particularly on current and former United States military installations.

Petranka (1998) mentions the Burton and Likens (1975a) study at Hubbard Brook Experimental Forest in New Hampshire for their estimates of P. cinereus density, but similarly important is Burton and Likens (1975b), a simultaneous study on the energy flow and nutrient cycling that must pass through P. cinereus, given its position in the trophic web and the high densities observed at Hubbard Brook. These studies (Burton and Likens 1975a,b) have since been cited by hundreds of subsequent studies, some of which argue for the use of P. cinereus as an ecological indicator species (e.g., Welsh and Droege 2001; Siddig et al. 2019), or for the overall importance of plethodontids to terrestrial forest ecosystems (e.g., Davic and Welsh 2004), and indeed have shaped the thinking of many researchers working on North American plethodontid salamanders. As we discuss in Additional Topics: Management, there are challenges to the acceptance and use of P. cinereus as an indicator species (e.g., Riedel et al. 2008; Moore and Ouellet 2015). In particular, the wide geographic range and the high tolerance of P. cinereus to disturbed habitats (e.g., Mossman et al. 2019) make P. cinereus an outlier within Plethodontidae, and a potentially poor proxy for more threatened Plethodon salamanders.

Since Petranka (1998), several studies have investigated how P. cinereus contribute to ecosystem processes like nutrient cycling or energy movement. Some of these have manipulated densities of P. cinereus or examined how natural density variation influences ecosystem processes, albeit with contradictory findings (Wyman 1998; Walton 2005; Sucre et al. 2010; Homyack et al. 2010b; Walker et al. 2014; Hocking and Babbitt 2014; Pelini et al. 2015; Hickerson et al. 2017). In New York, Wyman (1998) found that experimental cages with two or six salamanders (compared to controls with zero salamanders) significantly decreased forest floor prey communities, which had an indirect negative effect on the rate and magnitude of leaf litter decomposition, likely because of P. cinereus feeding on leaf shredders. In Ohio, Hickerson et al. (2017) also manipulated P. cinereus densities by removal using an open field design and found similar results to Wyman (1998). Repeated removal of salamanders over a 4-yr period increased the rate and magnitude of leaf litter decomposition compared to treatments with fewer or no salamanders removed (Hickerson et al. 2017). In contrast, other experimental studies found no effects of P. cinereus on ecosystem measures, such as leaf litter and wood decomposition or nitrogen mineralization rates, among others, even when manipulated densities were as high as 4 salamanders/m² (Homyack et al. 2010b in Virginia; Hocking and Babbitt 2014 in New Hampshire; but see discussion of these studies above in Community Ecology). Additionally, Sucre et al. (2010) found that salamander abundance was not a reliable indicator of nitrogen availability in Virginia.

Milanovich and Maerz (2012, 2013) describe the use of stable isotopes from P. cinereus tail clips collected from New York for tracking ecosystem nutrient and energy movement. They note that these isotopes are accurate even when the salamander has been fasted for up to 35 d, making their interpretation much easier compared to organisms for which fasting does change these values. Also working in New York, Gilbert et al. (2014) used carbon-13 to understand how carbon from sugar maple (Acer saccharum) tree roots moves through the ecosystem and found that P. cinereus individuals contained large quantities of root-derived carbon. This was true whether earthworms were part of the treatment conditions or not (Gilbert et al. 2014).

Another line of investigation has made connections between ecosystem changes or elemental gradients and P. cinereus (Homyack et al. 2011; Beier et al. 2012; Ochs and Siddig 2017). In recently harvested forests in West Virginia, P. cinereus expends more energy on body maintenance, although there were no consistent patterns related to body condition or invertebrate abundance (Homyack et al. 2011). In New York, Beier et al. (2012) found a positive relationship between P. cinereus abundance and a well-defined calcium gradient, and in Massachusetts, Ochs and Siddig (2017) found P. cinereus was more prevalent in forests with lower soil pH, lower carbon to nitrogen (C:N) ratios, and lower temperatures. The authors note that the first two characteristics, low soil pH and low C:N ratios, are characteristic of Eastern Hemlock stands, which used to be much more common within the range of P. cinereus prior to the arrival of Hemlock Wooly Adelgid (Siddig et al. 2019). The loss of Eastern Hemlocks and other ecosystem changes as a result of invasive species who affect fundamental environmental properties relied on by forest floor inhabitants (e.g., soil chemistry, temperature, humidity), may also influence P. cinereus (e.g., Cáceres-Charneco and Ransom 2010; Ochs and Siddig 2017). Continental invasions of earthworms from Europe and Asia undoubtedly influence P. cinereus in many ways, but especially by changing soil properties and P. cinereus behavior, which likely influence the role of salamanders in shaping North American forest ecosystems (Cáceres-Charneco and Ransom 2010; see also the Community Ecology section).

Lastly, as discussed in Community Ecology above, the role of P. cinereus on the forest floor as a predator (e.g., Walton 2005; Walker et al. 2014) may influence heterotrophic respiration rates, having measurable implications for climate change in forest soils, but these effects may vary latitudinally (Pelini et al. 2015). The inconsistent results of these studies (see our discussion above, in Community Ecology, as well as Hickerson et al. 2017) are yet another reason it is critical to design robust experiments to ensure comparability across populations and better understand variation in demography and community interactions across the range of P. cinereus.

In Petranka (1998) there was relatively little consideration of management for P. cinereus populations or their habitat, presumably because P. cinereus is considered a common species throughout its range, despite its proposed sensitivity to changing landscapes driven by human activity and our knowledge of these practices on other woodland salamanders. However, since Petranka (1998), there has been a great deal of research on the impacts of forest management on amphibian communities, including the ubiquitous P. cinereus. Here, we summarize those studies which focus on managing forests with explicit mention of impacts to P. cinereus.

In an experimental management context (i.e., either manipulated or observational), P. cinereus abundance or distribution may be influenced by factors as diverse as prescribed fire (Mitchell 2000), tree harvest management strategy or intensity (Messere and Ducey 1998; Rodewald and Yahner 1999; Harpole and Haas 1999; Yahner et al. 2001a; Duguay and Wood 2002; Perkins and Hunter 2006; Patrick et al. 2006; Homyack and Haas 2009; Strojny and Hunter 2010b; Hocking et al. 2013; MacNeil and Williams 2013b, 2014; Mossman et al. 2019), deer management (Greenwald et al. 2008; see discussion in Community Ecology), managed grazing (Riedel et al. 2008), or subsequently in forests with varying densities and qualities of coarse woody debris (McKenny et al. 2006; Strojny and Hunter 2010a). There is some evidence that plantation-style forests, common on lumber farms, are not suitable for P. cinereus (Waldick et al. 1999; Reams et al. 2008) despite the closed canopy, perhaps because of the lack of cover objects.

Studying how timber harvest practices influence P. cinereus energetic response, Homyack et al. (2011) reported on short-term effects of forest floor temperatures from overstory removal harvest, which may lead to an energetic cost for P. cinereus. The authors also suggest that retaining downed logs and other large woody debris in addition to promoting regrowth of understory may contribute to rapid recovery of forest floor conditions conducive to stable P. cinereus populations. Electric transmission right-of-way (ROW) areas are often heavily managed to remove vegetation and to provide continual access for maintenance. Yahner et al. (2001a) report that P. cinereus were among the most common amphibians found in these areas and that ROW areas housed a greater diversity of amphibians than adjacent forests; however, the authors do not discuss how detection influenced their results. In a follow-up study, P. cinereus were more commonly found in areas in Pennsylvania that are managed through timber harvesting than in areas that are undisturbed (Yahner et al. 2001b), which contradicts the deMaynadier and Hunter (1998) finding that P. cinereus prefer interior, mature forests. The work of Yahner et al. (2001b) does fit with the suggestion of Riedel et al. (2008) that P. cinereus may be less sensitive to anthropogenic change than previously thought, provided they have appropriate microhabitats (i.e., cool and moist cover objects; but see Wood and Williams 2013 for a counterexample).

Some studies demonstrated no effect of experimental management on P. cinereus individuals or populations. Mahoney et al. (2016) showed no effect of shelterwood-burn or fenced shelterwood-burn treatments on P. cinereus abundance, but stressed that leaf litter and fine woody debris cover objects were influential on P. cinereus abundance when considering best approximating single variable models. Similarly, Moore (2014) used a mesocosm experiment to demonstrate that the addition of lime to acidic, maple forests (liming), has no impact on survival of P. cinereus. Cameron et al. (2016) found similar results in the field comparing a long-term liming site to an unlimed control area in Ohio.

In the last two decades, it has been suggested that P. cinereus be considered as an indicator species for a variety of ecosystem properties (e.g., Welsh and Ollivier 1998; Welsh and Droege 2001). Siddig et al. (2019) agreed that P. cinereus may be an indicator species used to monitor changes in eastern hemlock forest ecosystems because of their sensitivity to key stressors and changing environmental conditions. Along with several studies hypothesizing and testing the connection between a changing climate and color phenotype ratios in a population (see the Systematics and Geographic Variation section), Moore and Ouellet (2015) concluded that climate variables do not influence color phenotype ratios broadly across the range of P. cinereus. The indicator species concept (also known as “surrogate species concept”) arose as a shortcut approach to evaluating the complex state of ecosystems by measuring a few, relatively inexpensive observable characteristics (Landres et al. 1988; Caro 2010; Lindenmayer et al. 2015). The degree to which P. cinereus can successfully be considered an indicator species in the future will depend upon making direct connections between P. cinereus and specific drivers of interest as well as understanding the mechanistic relationship between the indicator and ecosystem property of interest, which is often an unobservable ecological process (Fleming et al. 2020).

Studies that explore the community-level interactions between P. cinereus and microbial species generally focus on three distinct groups of microbiota: salamander skin microbiota, salamander gut microbiota, and environmental (typically soil) microbiota. In this section, we discuss all sources which address how P. cinereus interacts, directly or indirectly, with microbial taxa. Prior to 1998, information regarding the microbiota that P. cinereus interact with was scarce, in part because the technological developments which enabled economical microbial community profiling were only just becoming available. Petranka (1998) mentions that there has been no evidence for antibiotic properties in the skin secretions of brooding females, a claim that has since been refuted by several recent studies that document microbial taxa that produce both antibiotic and antifungal compounds in clinically relevant levels (e.g., Fredericks and Dankert 2000; Harris et al. 2006; Culp et al. 2007; Lauer et al. 2007; Brucker et al. 2008a,b; Becker et al. 2009). Most research to date regarding the skin microbiota of P. cinereus has focused on two overarching goals: (1) to identify microbial taxa found on salamander skin and in their guts, and (2) to understand the potential these microbial taxa have to produce antibacterial and antifungal compounds. The latter goal has been specifically focused on those microbiota which may be useful in targeting the deadly amphibian fungal pathogen Bd, to which P. cinereus is thought to be largely resistant. Compared to these two topics, relatively few studies have looked at the impact of P. cinereus on the soil microbial communities.

Fredericks and Dankert (2000) is one of the earliest studies to identify antimicrobial proteins on the skin of P. cinereus, shortly after Petranka (1998) was published. However, as this study focused on protein fractions instead of whole-community DNA, they are unable to determine if these compounds are salamander-made or microbe-made. Harris et al. (2006) document three genera of bacteria on the skin of P. cinereus that inhibit the growth of Bd in challenge assays (Bacillus, Lysobacter, Pseudomonas). Another early study of skin microbes by Culp et al. (2007) isolated five bacterial taxa (Pseudomonas fluorescens, Microbacterium testaceum, Flavobacterium johnsoniae, Bacillus cereus, and Corynebacterium spp.) and one yeast (Candida molishiana), as well as several unidentifiable isolates. Of these, Pseudomonas fluorescens was especially notable, because it is known for having antimicrobial and toxic (hydrogen cyanide) properties. Culp et al. (2007) also noted that many of the bacterial colonies were mucoid, and thus are likely involved in preventing salamander skin dehydration.

Shortly after these early studies were published, metagenomic methods became less expensive and more accessible, and most subsequent work using these methods suggests that microbes produce many of the beneficial proteins previously identified by Fredericks and Dankert (2000). Lauer et al. (2007) found that 89% of the salamanders in their sample had strong and/or weak antifungal bacteria on their skin. They used a fungal pathogen, Mariannaea sp., isolated from dead Hemidactylium scutatum eggs in their challenge assays. Lauer et al. (2007) highlight their discovery of a specific bacterial strain closely related to Janthinobacterium lividum, a bacterium which was previously known to produce antifungal compounds. In their study, J. lividum was found on every salamander (Lauer et al. 2007). This result has since been confirmed by several additional studies that focus on J. lividum and the ability of P. cinereus to either resist or quickly recover from Bd infection (e.g., Brucker et al. 2008a; Becker et al. 2009; Wiggins et al. 2011; Muletz et al. 2012; Hess et al. 2015). A follow-up study found similar microbial composition at the genus and family levels, but little overlap at the species level, on the skin of confamilial H. scutatum (Lauer et al. 2008).

Janthinobacterium lividum makes at least two known antifungal metabolites, indole-3-carboxaldehyde and violacein, that are lethal to Bd in naturally occurring concentrations found on the skin of P. cinereus (Brucker et al. 2008a). When Bd is experimentally introduced salamanders, 8 out of 10 salamanders who died due to Bd infection had no violacein-producing microbes and the remaining 2 had lower than average violacein levels (Becker et al. 2009). Other Bd inhibiting or mitigating microbial taxa found in the skin or in the guts of P. cinereus include Lysobacter gummosus (Brucker et al. 2008b), Pseudomonas reactans (Harris et al. 2006, 2009), and several additional taxa identified by Muletz-Wolz et al. (2017) and Fontaine et al. (2018). In a lab experiment, Loudon et al. (2014a) cultured bacterial isolates including Bacillus spp., J. lividum, Pseudomonas spp., and Chitinophaga arvensicola, from P. cinereus as cocultures and monocultures to determine what bacterial species or combinations most successfully inhibit the growth of Bd. Metabolites in the co-cultures were most inhibitory to Bd as they interact synergistically and additively (Loudon et al. 2014a).

There is some evidence that these protective bacterial taxa can be inoculated on salamanders who lack them, and that inoculation relieves symptoms of infection (Harris et al. 2009; Muletz et al. 2012; Loudon et al. 2014b). Harris et al. (2009) used Provosoli medium to expose individuals of P. cinereus to protective Pseudomonas reactans, with and without Bd present in the same treatment. Salamanders exposed to Bd alone lost more body mass compared to both those in the control groups, which were exposed to either sterile medium or Pseudomonas reactans alone, and those in the treatment group exposed to Bd and Pseudomonas reactans. Muletz et al. (2012) sought to use J. lividum to protect the globally declining amphibian populations as a result of Bd, and learned that through soil bioaugmentation, it is possible to experimentally introduce J. lividum into soil which is then environmentally transferred to amphibian skin. Within 5 d post-Bd exposure, the environmental transmission of J. lividum was able to inhibit the colonization of Bd on the skin of P. cinereus (Muletz et al. 2012).

Loudon et al. (2014b) identified a “core community” of eight operational taxonomic units (OTUs) which was found on at least 90% of the salamanders throughout the experiment. Their experiment revealed that the most abundant and prevalent OTU was a member of the phylum Verrucomicrobia, and that five out of the eight core OTUs were in the family Pseudomonadaceae which all have antifungal properties. Their results also indicate that the availability of different bacteria in the environment influences what bacteria are found on the salamander’s skin (Loudon et al. 2014b). Specifically, they suggest that the environment works as a reservoir for bacteria. Without access to this reservoir, the core community of skin microbiota dominates, and the diversity of the microbiome is lower (Loudon et al. 2014b). Loudon et al. (2016) supports these results: 9 out of 12 (75%) OTUs that were overrepresented on P. cinereus skin had at least a 97% match to antifungal isolates. They also noted that four of these overrepresented OTUs were part of the core community identified by Loudon et al. (2014b). Loudon et al. (2016) suggest that salamander skin, as an environment, may actively select for bacteria with antifungal activity. Future efforts to protect amphibians from mortality due to diseases such as Bd may include exposing captive animals to environmental reservoirs with defensive bacteria. This may improve captive-rearing programs for amphibians to prevent colonization of Bd (e.g., Passos et al. 2018).

Protective bacteria for P. cinereus against fungal diseases may also originate from the gut of the salamander, although we only found two studies that looked at gut microbes (Wiggins et al. 2011; Fontaine et al. 2018). Wiggins et al. (2011) suggest that the gastrointestinal tract of P. cinereus could serve as a reservoir for J. lividum. Fontaine et al. (2018) also find evidence of Janthinobacterium and other disease resistance related taxa in the guts of P. cinereus. This study found an interesting pattern with temperature. Microbial diversity, including those taxa involved in disease resistance, declined at 20°C compared to 10°C and 15°C, and digestive performance was greatest at 15°C. Additionally, the relative abundance of pathogenic taxa increased at 20°C (Fontaine et al. 2018).

In terms of how P. cinereus responds to Bd infection, a small number of studies have documented behavioral effects of Bd infection, loss of body mass, differential infection by color phenotype, and gene expression. As noted earlier in the Predators and Parasites section, natural incidence of Bd infection in P. cinereus is generally quite low (≤2%, with the exception of a New Brunswick population with infection rates around 13%; Jongsma et al. 2019). In a behavioral study, Bd-infected salamanders attacked prey significantly more often than noninfected individuals, and the prevalence of Bd among P. cinereus decreased over time (Hess et al. 2015). The latter result led Hess et al. (2015) to suggest that P. cinereus may have natural resistance to Bd. The former result makes biological sense in the context of Harris et al. (2009) and Becker and Harris (2010), which both show that Bd-infected P. cinereus without protective bacteria lose significantly more body mass than controls (in both studies) and individuals inoculated with protective bacteria (Harris et al. 2009). Additionally, Becker and Harris (2010) showed that a reduction of cutaneous bacteria prior to Bd infection led to increased limb-lifting behavior, which had not been observed in this species before, and was only observed in infected individuals.

Venesky et al. (2015) looked at differential effects of Bd infection by color phenotype in the laboratory. Unstriped P. cinereus had both a higher prevalence of infection and higher mortality due to Bd than the striped individuals (Venesky et al. 2015). The authors note that these results conflict with a previously noted connection between high levels of melanin and disease resistance, but that the behavioral differences between morphs (e.g., Venesky and Anthony 2007; Davis and Milanovich 2010; Reiter et al. 2014) may increase the stress experienced by unstriped morphs thus reducing their resistance to infection. Finally, Ellison et al. (2020) found changes in gene expression, mediated by temperature, in P. cinereus who were uninfected, recently infected, or naturally recovered from infection with Bd. At lower temperatures, inflammation transcripts were more commonly expressed, while at higher temperatures, salamanders shift to increased expression of adaptive immune genes, including MHC (major histocompatibility complex) genes (Ellison et al. 2020).

We found three studies that explore how presence or absence of P. cinereus may impact soil microbial communities, and the results are somewhat equivocal (Walker et al. 2014; Hickerson et al. 2017; Walker et al. 2018). Two studies, one based in Ohio (Walker et al. 2014) and the other in West Virginia (Walker et al. 2018), used a metagenomics approach to explore the impact of the removal of P. cinereus on the fungal detritivore food web. Fungal community diversity and composition were not affected by removal of P. cinereus, but there were significant shifts in relative abundance of the fungal groups (Walker et al. 2014, 2018). Hickerson et al. (2017) used phospholipid fatty acid (PLFA) analysis to explore the effect of P. cinereus on the microbial community in Ohio. Although removal of intraguild predators had little effect on soil and leaf litter microbial communities, the treatment in which salamanders were most abundant (centipedes removed) suppressed bacteria, which in turn decreased the rate of leaf litter decomposition (Hickerson et al. 2017).

Additionally, only a small number of studies explored change in microbial communities over space and time (Muletz-Wolz et al. 2017, 2018, 2019; Barnes et al. 2020). The work of Muletz-Wolz and colleagues has explored skin microbial communities across an elevational gradient in Maryland and Virginia. They found that location, rather than host characteristics, was the best predictor of prevalence of antifungal bacteria (Muletz-Wolz et al. 2017); higher elevations had greater alpha diversity of bacterial taxa than lower elevations (Muletz-Wolz et al. 2018); and increasing the pathogen load and increasing temperatures changed the microbial community structure, which suggests that P. cinereus may not be resistant to Bd in all ecological contexts (Muletz-Wolz et al. 2019). Finally, in New York, Barnes et al. (2020) found that the composition of bacterial communities, but not overall diversity, varied with a land use gradient (urban, suburban, exurban), but there were similar levels of Bd inhibition among each of these different land use bacterial communities.

Studying the interactions of P. cinereus with skin, gut, and environmental microbes is still a relatively new research focus. We also note that this topic is one of the most geographically restricted, with 65% (17/26) of studies discussed earlier using salamanders from Virginia, and the remaining nine studies using salamanders from adjacent central, mid-Atlantic states (Ohio, West Virginia, Pennsylvania, Maryland, and New York). There is evidence that microbial taxa change over time, space, and with access to soil reservoirs (e.g., Loudon et al. 2014b; Muletz-Wolz et al. 2018; Barnes et al. 2020), thus standardized microbial sampling, such as that described by Walker et al. (2015), is needed throughout the range of P. cinereus for comparable data that can then be used to understand the relationships of P. cinereus with microbial taxa fully. Other gaps in salamander microbiota research include exploring the distribution of different bacteria across the body regions of the salamander. Culp et al. (2007) suggests there may be a nonrandom distribution of different species in different body regions and that future work should look to swab the whole animal at multiple locations on the body to determine if/where different species may congregate. We also do not fully understand the relative roles and interactions of microbial metabolites versus salamander skin secretions. Lauer et al. (2007) suggest separating and analyzing the antibiotic chemicals produced by the animal from those produced by bacteria on the skin. They also suggest conducting experiments using antibiotics to determine the roles of different cutaneous bacteria.

The final additional topic that has grown considerably since Petranka (1998) explores the responses of P. cinereus to a variety of anthropogenic environmental contaminants. These ecotoxicology papers generally subject individual salamanders to multiple concentrations of compounds with known toxicity in other vertebrates. In most cases, the concentrations tested are well below field values at contaminated sites, and salamanders show some ability to detect and avoid some of these contaminants.

Several studies using salamanders from Massachusetts and Maryland have focused on contaminants commonly found on current and former United States military installations, including RDX (1,3,5-trinitrohexahydro-1,3,5-triazine; Johnson et al. 2004), 2,4-DNT (dinitrotoluene; Johnson et al. 2007), TNT (trinitrotoluene; Bazar et al. 2008), copper (Bazar et al. 2009), lead (Bazar et al. 2010), and HMX (high melting explosive or octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine; Johnson et al. 2010). All of these except HMX had negative effects on P. cinereus at the highest concentrations tested (Johnson et al. 2004, 2007, 2010; Bazar et al. 2008, 2009, 2010). The most common symptoms were weight loss, depressed white blood cell counts, and death. Additionally, in the copper treatments, the authors observed lethargy, unresponsiveness, and inhibition of the righting reflex (Bazar et al. 2009). HMX did not seem to affect P. cinereus, and the authors suggest that this could be because of differences in how HMX is absorbed through salamander skin versus through the gastrointestinal tract of the other vertebrates tested (Johnson et al. 2010).

Mercury (Hg) concentrations found in P. cinereus increase when individuals are exposed to contaminated soil compared to individuals from noncontaminated sites (Bergeron et al. 2010a). Bergeron et al. (2010a) reported higher Hg in P. cinereus than any previous literature, but also noted a low sample size of 24 individuals. It is not clear from this study if Hg accumulation has negative effects on the salamanders or if it should be more of a concern for the predators that eat salamanders. A follow-up study found that total Hg and (mono)methylmercury (MMHg) concentrations are positively correlated in P. cinereus from contaminated sites (Bergeron et al. 2010b). The same study did not find accumulation of selenium (Se) above what is expected for normal background. Selenium is thought to be antagonistic with Hg, offering some protection from Hg toxicity (Bergeron et al. 2010b). Because of the capacity of P. cinereus to take in Hg from the surrounding environment, it is suggested that P. cinereus be used as a bioindicator to monitor Hg accumulation in the landscape (Townsend and Driscoll 2013).

Two laboratory studies have demonstrated that P. cinereus is able to detect and avoid contaminated soils, suggesting that contaminated sites in nature would simply not be colonized by salamanders. First, Gertzog et al. (2011) found that P. cinereus can detect and avoid substrates which have been contaminated with herbicides such as Roundup, Bayer Brushkiller, and Spectracide Brush Killer. Second, Gaglione et al. (2011) found that P. cinereus can detect and avoid substrates contaminated with urea. Finally, a third study tested the effects of the microbial pesticide Bacillus thuringiensis var. kurstaki (Btk) on P. cinereus abundance or diet in West Virginia and found no adverse effects (Raimondo et al. 2003).

In completing the comprehensive review, it is clear that the Duffy et al. (2021: p. 55) requirement that a model system have been studied long enough to generate a “substantive body of knowledge” has been met by the research to date on P. cinereus. In defining EEB model systems, Duffy et al. (2021) highlight Travis’s (2006: p. 307) related stipulations that “robust inference requires horizontal comparisons and vertical integration.” These, too, have been clearly met by the research to date on P. cinereus, which has been used to better understand closely related, but more geographically restricted, taxa (horizontal comparisons) and has been studied at nearly every scale of biological organization (vertical integration; Fig. 3). Outside of those who work on plethodontid salamanders, however, we do not think the EEB community would necessarily designate P. cinereus as a model organism—not in the comprehensive way that others readily provide examples such as Anolis lizards, Mimulus plants, or Drosophila flies (e.g., see Duffy et al. 2021: Supplemental Material S1; table S1).

Thus, to date, P. cinereus falls short of model organism status. This is likely due in part to many of the broader challenges within EEB fields around developing new model systems. In particular, the broader appeal of P. cinereus research and plethodontids as model organisms more generally may have been hampered by a lack of methods standardization, barriers to participation in the field of herpetology (e.g., Perry et al. 2020), funding constraints (e.g., Clutton-Brock and Sheldon 2010), and other limitations imposed on both academic and nonacademic scientists in the United States and Canada (e.g., lack of data sharing ethos). However, with the vast body of knowledge on P. cinereus, we have the opportunity to build future work with a deliberate eye towards inclusive, coordinated collaboration with an emphasis on knowledge and data sharing. Specifically, the benefits of collaborative ecological research networks can be leveraged for collecting organismal data at larger temporal and spatial scales than individual researchers (Hughes et al. 2017; Clutton-Brock and Sheldon 2010).

Illuminating fundamental phenomena in EEB (Sutherland et al. 2013) not only benefits from the use of diverse model systems (Duffy et al. 2021; Wale and Duffy 2021), but also from long-term data sets (Clutton-Brock and Sheldon 2010). Access to high-speed internet connections and smart devices has rapidly increased, making it easier to manage large collaborative research networks that coordinate large scale experiments and surveys across time and space. One such network, the Salamander Population and Adaptation Research Collaboration Network (SPARCnet) was founded in 2013 by researchers at the United States Geological Survey Amphibian and Reptile Monitoring Initiative (USGS ARMI) and Pennsylvania State University (SPARCnet 2024). Since then, it has grown to include over 30 research collaborators and 13 education collaborators (e.g., nature centers) across 13 states and provinces within the geographic range of P. cinereus (Supplemental Material S16; Supplemental Fig. S4, available online).

SPARCnet methodology employs a simple and inexpensive, yet robust and spatially explicit mark-recapture design (Sutherland et al. 2016; Muñoz et al. 2016b; Fleming et al. 2020, 2021; Grant et al. 2024) using standardized ACO plots, arranged in pairs for potential experimental and control studies (e.g., snow removal, forest management). SPARCnet holds annual meetings and has developed research protocols and minimum core data collection agreements among researchers that allow this network to meet the growing needs of long-term studies focusing on changing demography of the focal species, P. cinereus (Clutton-Brock and Sheldon 2010). SPARCnet’s framework is flexible, allowing individual researchers to design side projects that are site-specific or network-wide. Additionally, the flexibility of SPARCnet structure allows researchers to exploit observations and research questions that emerge in real time (i.e., emergence of disease or toxicology) and work together to study these phenomena across the geographic range in tandem rapidly.

Collaborative research networks provide a solution for consistent data collection methods and help network participants address the many inconsistencies that come with studying a wide-ranging species like P. cinereus. It is well known that it is difficult to study patterns from past data based on differences in methodology and confounding factors from data collection (Fraser et al. 2013; Grant 2015), and that EEB studies typically suffer from geographic and taxonomic biases, as we have demonstrated throughout this review (Supplemental Material S15; Supplemental Data S14). A geographically and intellectually distributed network, such as SPARCnet, addresses these concerns to provide a robust spatial and temporal assessment of the current state and change over time of P. cinereus populations across its range. The large geographic range and high local abundances make P. cinereus a highly accessible model system for a great number of diverse institutions, researchers, and their students. We hope that our experience provides a blueprint for other researchers to more confidently develop a diversity of ecological model systems called for by Duffy et al. (2021).

As we have seen in nearly every section of our review, the great amount of geographic variation in P. cinereus biology often leads to puzzling contradictions of P. cinereus physiology, behavior, roles in ecosystems when studies are repeated in new locations. These contradictions and their impact on the potential role of P. cinereus as an indicator species (especially among color phenotypes; see Systematics and Geographic Variation) have resulted in confusion, valid criticisms, and necessary caution (e.g., Moore and Ouellet 2015). However, considering Duffy et al. (2021) and in the hindsight of this review, we think the inconsistencies—the natural variation of a single species over a wide geographic area—are exactly what makes P. cinereus an ideal EEB model organism.

Unlike the other seven of the top eight most well-studied salamander species, P. cinereus is the only member of the Plethodontidae (Supplemental Material S1; Table S4, Supplemental Material S3; Supplemental Data S2). As a direct-developing member of the largest family of salamanders, P. cinereus is more representative of the majority of ecologically and evolutionarily successful salamander taxa than any of the other top seven species. What we learn by studying P. cinereus is more likely to apply to vulnerable and endangered plethodontids than what we learn by studying A. maculatum or N. viridescens (Supplemental Material S1; Table S4). Combined with their large population sizes and position in the forest food web, we see endless potential for illuminating fundamental phenomena in EEB with P. cinereus. On top of the overwhelming volume of ecological and behavioral knowledge we have summarized above, the well-supported systematics and evolutionary knowledge of both Plethodontidae and Plethodon (e.g., Wake and Larson 1987; Wake 2009; and see Staub and Muller 2017 and Kuchta 2019 for overviews of the work of D. B. Wake and R. Highton, respectively) there is also a very healthy level of recent systematic interest in both the family and the genus (e.g., Wiens et al. 2006; Fisher-Reid and Wiens 2011; Fisher-Reid et al. 2012), making plethodontids truly a model clade, not just P. cinereus a model organism.

In terms of the three axes Duffy et al. (2021) include in their Fig. 1, we know, and demonstrate with our review, that P. cinereus is both highly lab and field tractable, and research can be and has been done at all biological scales except for global (Fig. 3). That said, there are some weaknesses, particularly when it comes to lab tractability (Fig. 3), which have yet to be satisfactorily addressed, and indeed may never be so. In the field, individuals spend much of the year underground, no matter the broader phenological pattern (Fig. 2), and when they are surface active, they are most active at night, which can be logistically difficult to observe. In the lab, rearing P. cinereus has proven difficult for two primary reasons: First, they are a relatively long-lived species. Direct measures of lifespan, based on a single population at the northern range limit, are between 8 and 9 y (Leclair et al. 2006, 2008). Second, controlled laboratory breeding has not yet been possible. Females can be induced to lay fertilized eggs (e.g., Evans et al. 2020), and courtship behaviors can be observed successfully (Dyal 2006; Jaeger et al. 2016), so perhaps one day we will crack the code for controlled breeding of P. cinereus, but, as we have argued here, this requires infrastructure, community ethos, and investment from funding agencies.

1. Model systems in evolution, ecology, and behavior (EEB) are similar, but also different from those in biomolecular and biomedical sciences. Variation is a feature, not a bug.

2. As lungless ectotherms, salamanders in the family Plethodontidae have great potential to serve as a model clade within EEB.

3. One plethodontid in particular, P. cinereus, stands out for the overwhelming depth of knowledge accumulated in the last several decades, its critical niche position in forest ecosystems, its accessibility and tractability in both lab and field, and its close, well-studied relationships with other members of its taxonomic family.

4. In just 22 yr, over 400 papers on all aspects of the evolution, ecology, and behavior of P. cinereus have been published in peer-reviewed journals. Although studies are often highly localized and demonstrate high levels of local variation across the large geographic range of P. cinereus, they also demonstrate the breadth of EEB fundamental phenomena that can be studied using P. cinereus.

5. Plethodon cinereus falls short of model system status because of lack of infrastructure, lack of community ethos, and lack of investment by funding agencies.

6. The collaborative research network SPARCnet follows in the footsteps of other collaborative research networks to study P. cinereus populations cohesively and repeatedly across the geographic range. Local variation is expected and embraced as we leverage our intellectually and geographically distributed network to illuminate fundamental phenomena in EEB through the eyes of a woodland salamander.

This monograph is the product of a Summer 2020 Virtual Lab Meeting, which met weekly on Zoom to remain engaged in research at the start of the COVID-19 pandemic. We thank members of SPARCnet for their support, discussion, and comments on earlier versions of the monograph, especially J.E. Fleming and E.H.C. Grant. We also thank Editor M. Harvey and four anonymous reviewers for extremely thoughtful and thorough reviews on what can only be described as a massive beast of a manuscript, which we feel greatly improved the finished product. We thank B. Waldron for additional helpful comments on the preprint. Photos of P. cinereus color morphs that are included in Fig. 1 were graciously provided by T. Pierson (striped and unstriped morphs) and T. Leenders (erythristic morph). M.C.F.-R., K.L.G., and L.S.M. were supported by a collaborative National Science Foundation Improving Undergraduate STEM Education grant (DUE Award Nos.: 1914798, 1914727, and 1914722, respectively). M.C.F.-R. was additionally supported by Bridgewater State University’s Center for the Advancement of Research and Scholarship Faculty and Librarian Research Grant and the Bridgewater State University Office of Undergraduate Research. S.R.G. and A.K.I. were supported by the Urban Coast Institute and the Summer Research Program at Monmouth University. M.A.H. was supported by Arts & Sciences Summer Research Fellowship from the University of Richmond. T.J.H.M. and K.N.S. were supported by Richard King Mellon Foundation Freshwater Research Institute funds for summer research at Susquehanna University. A.S.J. was supported by an Adrian Tinsley Program Summer Grant from the Bridgewater State University Office of Undergraduate Research. M.S. was supported by the BEACON Center for the Study of Evolution in Action at Michigan State University.

Supplemental material associated with this monograph can be found online at https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S1; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S2; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S3; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S4; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S5; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S6; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S7; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S8; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S9; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S10; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S11; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S12; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S13; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S14; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S15; https://doi.org/10.1655/HERPMONOGRAPHS-D-21-00003.S16.