Abstract

In anticipation of U.S. federal status classification (warranted, warranted but precluded, not warranted), scheduled for 2023, we provide population viability analysis of the Blanding's turtle Emydoidea blandingii, a long-lived, late-maturing, semi-aquatic species of conservation concern throughout its range. We present demographic data from long-term study of a population in northeastern Illinois and use these data as the basis for viability and sensitivity analyses focused on parameter uncertainty and geographic parameter variation. We use population viability analysis to identify population sizes necessary to provide population resiliency to stochastic disturbance events and catastrophes, and demonstrate how alternative definitions of ‘foreseeable future' might affect status decisions. Demographic parameters within our focal population resulted in optimistic population projections (probability of extinction = 0% over 100 y) but results were less optimistic when catastrophes or uncertainty in parameter estimates were incorporated (probability of extinction = 3% and 16%, respectively). Uncertainty in estimates of age-specific mortality had the biggest impact on population viability analysis outcomes but uncertainty in other parameters (age of first reproduction, environmental variation in age-specific mortality, percent of females reproducing, clutch size) also contributed. Blanding's turtle demography varies geographically and incorporating this variation resulted in both mortality- and fecundity-related parameters affecting population viability analysis outcomes. Possibly, compensatory variation among demographic parameters allows for persistence across a wide range of parameter values. We found that extinction risk decreased and retention of genetic diversity increased rapidly with increasing initial population size. In the absence of catastrophes, demographic conservation goals could be met with a smaller initial population size than could genetic conservation goals; ≥20–50 adults were necessary for extinction risk <5%, whereas ≥50–110 adults were necessary to retain >95% of existing genetic diversity over 100 y. These thresholds shifted upward when catastrophes were included; ≥50–200 adults were necessary for extinction risk <5% and ≥110 to >200 adults were necessary to retain >95% of existing genetic diversity over 100 y. Impediments to Blanding's turtle conservation include an incomplete understanding of geographic covariation among demographic parameters, the large amount of effort necessary to estimate and monitor abundance, and uncertainty regarding the impacts of increasingly frequent extreme weather events.

Introduction

Assessments of population viability have become increasingly important in identifying and making management decisions for species of conservation concern (Akçakaya et al. 1999; Morris and Doak 2002; IUCN 2012; USFWS 2016; Lacy 2019). Unfortunately, as the number of species under threat increases worldwide (Gibbons et al. 2000; Houlahan et al. 2000; Schipper et al. 2008; Butchart et al. 2010), practitioners are often faced with an unsolvable dilemma of making decisions based on projections from incomplete data over uncertain time periods and under changing environmental conditions (Reed et al. 1998; Morrison et al. 2016). As a consequence, it is important to have a clear understanding of the uncertainties associated with population viability assessments and the decisions they engender. This is especially true for species that are long-lived and rare because these traits increase extinction risk and make precise estimates of key demographic parameters difficult to obtain (Mace and Kershaw 1997; Burnham and Anderson 2002).

The Blanding's turtle Emydoidea blandingii (Figure 1) is a long-lived, late-maturing, semi-aquatic species that often occurs at low density across large expanses of wetland and adjacent upland habitat (Congdon et al. 2008, 2011; Reid et al. 2016). Its distribution is centered on the North American Great Lakes, extends westward to the Sandhills of Nebraska, and includes disjunct populations in northeastern North America. Blanding's turtles are threatened by habitat loss, elevated rates of nest predation due to subsidized predators, and road mortality (Congdon et al. 2008). As a consequence of these threats, populations are frequently small and isolated (Congdon et al. 2008). The Blanding's turtle is ranked as endangered by the International Union for Conservation of Nature and is recognized as being in need of conservation or listed as threatened or endangered in each state and province in which it occurs (IDNR 2009; IUCN 2012; MNDNR 2013; MDIFW 2015; PWAP 2015; COSEWIC 2016; NHFG 2017; NGPC 2018; NYSDEC 2019; IDNR 2020; IESPB 2020; MDC 2020; MNHESP 2020; ODNR 2020; MSU 2021; NHI 2021; SDGFP 2021). A petition to list the Blanding's turtle under the U.S. Endangered Species Act (ESA 1973, as amended) was found to present “substantial scientific or commercial information indicating that the petitioned actions may be warranted,” thus triggering status review with a 12-mo finding that listing is warranted, warranted but precluded, or not warranted anticipated in 2023 (USFWS 2015, 2021).

Figure 1.

Adult Blanding's turtle Emydoidea blandingii, Ogle County, Illinois, 2019. Photo by Dee Hudson.

Figure 1.

Adult Blanding's turtle Emydoidea blandingii, Ogle County, Illinois, 2019. Photo by Dee Hudson.

Much of our knowledge of Blanding's turtle demography comes from long-term study at the E. S. George Reserve (University of Michigan) in southern Michigan that demonstrates low nest survival, delayed reproductive maturity, and high levels of adult survival (Congdon et al. 1993, 2000). Other long term studies corroborate these characteristics of Blanding's turtle demography (e.g., Standing et al. 1999; Reid et al. 2016). Associated life-table analyses indicate that population stability is most sensitive to juvenile and adult survival and less sensitive to nest survival, age at first reproduction, and fecundity (Congdon et al. 1993, 2000). But even for the Michigan study, where field research spans nearly 4 decades (Congdon et al. 2000), precise estimates for some demographic parameters are lacking. For example, age 0 mortality is equated to nest failure or nest failure combined with hatch failure and juvenile mortality is inferred from other demographic parameters by assuming a constant population size (Congdon et al. 1993, 2000).

Available information on Blanding's turtle demography has been used in population viability analyses (PVA) to assess impacts of road mortality (Beaudry et al. 2008), alternative management and urban development scenarios (Dillon Consulting Limited 2013), and head-starting (in which eggs are collected and hatchlings are reared in captivity to minimize mortality during vulnerable early life stages) as a management strategy (Buhlmann et al. 2015; Thompson et al. 2020). In general, PVA makes use of mathematical models to generate future projections of population dynamics (Morris and Doak 2002). Often, PVA incorporates one or more sources of stochasticity in vital rates and may include sensitivity analyses to assess how variation in parameter values affects PVA outcomes (McCarthy et al. 1995; Cross and Beissinger 2001; Prowse et al. 2016; Manlik et al. 2017). This variation might be the result of parameter uncertainty (e.g., the 95% confidence limits of a parameter estimate), parameter variability (e.g., from population to population), or parameter manipulation (e.g., observed or hypothetical changes in a parameter as the result of management or other human activities). Ideally, PVA is based on knowledge of demographic parameters estimated from long-term study of a single population (Morris and Doak 2002). Instead, existing Blanding's turtle PVAs have used a combination of estimates obtained from short-term studies and ‘borrowed' from the long-term Michigan study (Beaudry et al. 2008; Dillon Consulting Limited 2013; Buhlmann et al. 2015). Blanding's turtle sensitivity analyses have included only a subset of demographic parameters using life-table or matrix methods without stochasticity (Heppell 1998; Congdon et al. 1993, 2000). A comprehensive Blanding's turtle PVA based on locally derived parameter estimates and incorporating stochasticity and global sensitivity analysis is lacking.

To build on our understanding of Blanding's turtle demography and population viability, we 1) characterize Blanding's turtle demography from long-term study of a centrally located population, 2) model population viability and assess sensitivity, and 3) use population viability analysis to explore the possibility of setting population size thresholds for conservation. Importantly, we are able to estimate nearly all parameters used in PVA from a single Blanding's turtle population. This includes estimates of environmental (= process) variance, the component of the total variance in a demographic parameter attributable to year-to-year environmental variation separate from the error variance attributable to sampling (Franklin et al. 2002). These estimates of demographic parameters and environmental variances allow us to conduct sensitivity analyses designed specifically to address uncertainty in parameter estimates in our focal population, thus guiding future demographic study. We also conduct sensitivity analyses that address geographic variation in demography among Blanding's populations, thus potentially identifying key variables affecting persistence range-wide.

Population size criteria are often included among the goals of policy makers and managers tasked with endangered species protection and recovery despite active debate regarding their utility and generality (Jamieson and Allendorf 2012; Frankham et al. 2013, 2014; Franklin et al. 2014). Blanding's turtles exhibit remarkable variation in population size. At the upper extreme is the Valentine National Wildlife Refuge population in the Sandhills of north-central Nebraska, which is thought to exceed 100,000 animals (Lang 2004). Perhaps next largest is the Weaver Dunes population in southeastern Minnesota, numbering approximately 5,000 animals (Pappas et al. 2000; Lang 2003). More typically, Blanding's turtle populations number from tens to hundreds of individuals (Graham and Doyle 1977; Herman et al. 1995; Joyal et al. 2000; McNeil 2002; Kiviat et al. 2004; Rubin et al. 2004; Compton 2007; Congdon et al. 2008; Ruane et al. 2008; MWPARC 2010; COSEWIC 2016), a size where demographic and environmental stochasticity are likely to accelerate loss of genetic variability and magnify extinction risk (Ovaskainen and Meerson 2010). To address this variation in population size, we use PVA to identify population sizes for which extinction risk is projected to remain below and genetic diversity is projected to remain above threshold values, repeating our analyses with varying degrees of environmental stochasticity with and without catastrophes (rare events with impacts exceeding year-to-year variation in demography), and running simulations for differing durations to demonstrate how alternative definitions of ‘foreseeable future' might affect conclusions (USDOI 2009; Almy 2017; Lake and Petersen 2017; USFWS 2019).

We use our results to identify demographic characteristics associated with Blanding's turtle population resiliency, the ability of a population to withstand stochastic disturbance events and local catastrophes (Shaffer and Stein 2000; Wolf et al. 2015). Resiliency, redundancy (ability to withstand regional catastrophic events), and representation (ability to adapt to environmental change) constitute the three Rs used by the U.S. Fish and Wildlife Service (USFWS) to inform Endangered Species Act decisions (Shaffer and Stein 2000; Wolf et al. 2015; USFWS 2016), which, for Blanding's turtles, are scheduled to occur in 2023 (USFWS 2021). Blanding's turtles share life history traits and conservation threats with a number of other North American freshwater turtles, including snapping turtles, wood turtles, bog turtles, and spotted turtles (Congdon et al. 1994; Enneson and Litzgus 2008; Shoemaker et al. 2013; Feng et al. 2019), making our work relevant to turtle conservation more generally.

Study Site

Our data on Blanding's turtle demography mostly comes from long-term (2004–2018) monitoring at the Spring Bluff–Chiwaukee Prairie (SBCP) complex in Lake County, Illinois, and Kenosha County, Wisconsin. This site consists of 215 ha of high-quality coastal wetland habitat along Lake Michigan and is part of the Chiwaukee Illinois Beach Lake Plain, recognized among Wetlands of International Significance (https://rsis.ramsar.org/, January 2021). It is managed by the Lake County Forest Preserve District, Wisconsin Department of Natural Resources, and The Nature Conservancy, and has been the focus of efforts aimed at promoting Blanding's turtle recruitment, survival, and habitat quality, including prescribed fire, mechanical and chemical treatment of invasive plants, turtle head-starting, and mesopredator removal (Urbanek et al. 2016; Thompson et al. 2020). Supplemental information comes from shorter term (2017–2019) Blanding's turtle monitoring at Illinois Beach State Park (IBSP), an adjacent 1,680-ha coastal wetland complex immediately south of SBCP, and from the work of researchers elsewhere.

Materials and Methods

Field methods

From 2004 to 2018, researchers collected capture–mark–recapture data by capturing turtles in baited collapsible minnow traps (Promar, 30 × 30 × 60 cm, 0.6-cm mesh), nylon hoop traps (Memphis Net and Twine, 76-cm diameter with 2.5-cm mesh), and by hand during the active season (April–August). Researchers marked turtles with passive integrated transponder tags and notching of marginal scutes and photographed turtles to aid in future recognition (Cagle 1939; Buhlmann and Tuberville 1998). Researchers classified turtles weighing <750 g as juveniles and assigned ages by counting annuli from plastron photos (Germano and Bury 1998; Wilson et al. 2003). Photos that could not be scored consistently by two independent observers were excluded (n = 40 older juveniles with indistinct annuli). Researchers determined sex of adults by observing the concavity of the plastron (Graham and Doyle 1979). Researchers equipped a subset of adult Blanding's turtles, mostly females, with radiotransmitters, facilitating collection of reproductive data (Thompson et al. 2020).

Demography

We palpated inguinal pockets of known-age females during the weeks prior to nesting to determine reproductive status (nongravid, gravid), providing us with information on age at first reproduction. We used logistic regression to characterize the relationship between age and reproductive status of these females, providing us with an estimate of reproductive frequency early in life. We also palpated transmitter-equipped females of unknown age to obtain reproductive frequency among older females. We computed environmental variance in the proportion of females reproducing by subtracting mean annual binomial variance from the among-year variance following Akçakaya (2002). We obtained data on number of eggs per clutch from 120 clutches included in the Lake County head-starting program from 2008 to 2018 (Thompson et al. 2020). We estimated environmental variance in clutch size (= among-year variance) from variance components analysis computed using the restricted maximum likelihood method in IBM SPSS 25 with year and female ID as random factors.

To facilitate PVA, we defined age 0 to encompass the period from oviposition (late June –early July) through the resumption of activity following a turtle's first winter (typically in April or May). Mortality during this period includes 1) clutch failure mostly due to nest predation, 2) hatch failure of eggs within intact clutches, 3) posthatch mortality, including mortality prior to the cessation of activity in fall and over winter. Rates of clutch failure were obtained by using telemetry to observe nesting at SBCP and IBSP coupled with follow-up monitoring to determine nest outcome (depredated, destruction by other causes, undisturbed; Urbanek et al. 2016). Hatch failure was estimated from the Lake County Forest Preserve District head-starting program (Thompson et al. 2020) based on eggs incubated from 2008 to 2018. Rates of posthatch mortality were obtained from Kastle et al. (in press), who used telemetry to monitor 82 hatchlings for up to 88 d at sites in northern Illinois and southern Wisconsin, including SBCP, and then extrapolated mortality from hatching through resumption of spring activity. We combined these components to estimate Age 0 mortality as 1 − (clutch survival × hatch success × posthatch and overwinter survival).

We have data from too few years to estimate environmental variance of clutch failure at SBCP. Instead, we used data from a Michigan study where the fates of 238 nests were monitored from 1976 to 1998 (table 1 in Congdon et al. 2000) and estimated environmental variance following Akçakaya (2002). Environmental variance in hatch failure was estimated from the Lake County Forest Preserve District head-starting program following Akçakaya (2002). Environmental variance in posthatch and overwinter mortality is unknown but was assumed to be no greater than the environmental variance of clutch failure, allowing us to estimate lower and upper limits to environmental variance in age 0 mortality by combining these components following Goodman (1960).

We estimated age-specific juvenile survival from 265 captures of 127 wild-born turtles (recaptures = 52% of total captures) first captured as juveniles using Cormack–Jolly–Seber model selection in Program MARK (Golba 2019). We estimated adult mortality from 531 captures of 148 adults (recaptures = 72% of total captures) using Cormack–Jolly–Seber model selection. We estimated environmental variance in adult mortality using the variance component option in Program MARK (Golba 2019). We estimated adult population size from 531 captures of 148 adults using the Jolly–Seber model (Jolly 1965; Seber 1965) as implemented in Program JOLLY (https://www.mbr-pwrc.usgs.gov/software/jolly.shtml, June 2021). We computed confidence intervals using Manly's method (Manly 1984; Krebs 1998).

Population viability and sensitivity

Spring Bluff Chiwaukee Prairie model.

We used the PVA software program Vortex because of its flexibility and repeatability (Lacy and Pollak 2018; Lacy et al. 2018). We modeled a single Blanding's turtle population for 1,000 iterations over a time-frame of 100 y using the default order of simulation events. We defined extinction as occurring when only one sex remained. We assumed no inbreeding depression; evidence of inbreeding is equivocal for Blanding's turtles (Sethuraman et al. 2014; Anthonysamy et al. 2017) and for turtles generally (e.g., Alacs et al. 2007; Davy and Murphy 2014; Gallego-García et al. 2018; Buchanan et al. 2019) and evidence of inbreeding depression is scarce (Velo-Antón et al. 2011). We assumed no correlation between environmental variance in reproduction and survival based on the absence of a correlation between year-to-year variation in residual clutch size and survival at SBCP (r = 0.20, n = 8, P = 0.628). We specified a population state variable to track adult population size along with default tracking of total population size.

We set demographic parameters to values observed at SBCP except as described here and in the Results. We used environmental variances to compute environmental standard deviations (SD) by which Vortex specifies binomial distributions to simulate environmental variation in reproductive and mortality rates (Lacy et al. 2018). We assumed a polygynous mating system from marker-based analyses elsewhere (Refsnider 2009; Anthonysamy 2012; McGuire et al. 2015). Maximum lifespan and maximum age of reproduction are unknown at SBCP but were set to 83 y based on a Michigan population (https://news.umich.edu/oldest-well-documented-blanding-s-turtle-recaptured-at-u-m-reserve-at-age-83/, January 2021). Females at SBCP and other study sites produce at most one clutch per year (personal observation, Congdon et al. 1983; Standing et al. 1999). We set the sex ratio at birth to parity. Many turtles, including Blanding's turtles (Gutzke and Packard 1987), have temperature-dependent sex determination but lack secondary sex characteristics until maturity (Graham and Doyle 1979) so field data on hatchling sex ratio are mostly lacking (for an exception, see Schwanz et al. 2010). We allowed reproduction to be density-independent because information regarding density dependence in Blanding's turtles is lacking and because outcomes such as extinction vs. persistence are determined at densities below which density-dependence is expected to occur. We assumed mortality schedules were the same for males and females; adult mortality does not differ between sexes at SBCP but tests for differences in mortality for other age classes are lacking (Golba 2019).

The frequency and severity of catastrophes affecting Blanding's turtle populations are poorly known. Catastrophes have not been observed over 15 y of study at SBCP nor over nearly 4 decades at a Michigan study site (Congdon et al. 2000) but otter Lontra canadensis predation resulted in 53 deaths among 100 Blanding's turtles at an Ontario site (Gassbarini 2016) and 49 deaths of unknown cause were documented at another site (Sheppard 2014). Mass mortality events attributable to predation, disease, and severe weather have been documented in other freshwater and terrestrial turtles (Table S1, Supplemental Material; see Fey et al. 2015 for a review of mass mortality events in animals more generally). To assess how catastrophes might affect PVA outcomes, we contrasted no-catastrophe and catastrophe scenarios. Catastrophes occurred with a 4% probability/y (ca. once per generation) and resulted in a reduction in age-specific survival to 75% that of the no catastrophe scenario, which were plausible values based on mass mortality events in freshwater and terrestrial turtles generally (Table S1, Supplemental Material). We assumed that all adult males were potential breeders. Mortality in turtles typically decreases with size (e.g., Feng et al. 2019), so we restricted juvenile mortality to be greater than or equal to adult mortality by setting mortality to 5.3% for age 4 and older (confidence intervals [CI] for mortality of age 4 and older juveniles broadly overlapped those of adults). We used the stable age distribution option in Vortex to generate the number of individuals in each year class for a very large population (100,000) and then treated these as proportional values for a population in which the adult year classes summed to our observed adult population (Lacy et al. 2018:53). The carrying capacity of SBCP for Blanding's turtles is unknown. To allow for realistic amounts of population growth, we set carrying capacity equal to twice our initial adult population size. We specified that carrying capacity truncation be based on adult (rather than total) population size to reduce the impact that years with high reproductive output had on population limitation. Neither harvest nor supplementation were included. We used default settings, in which each individual in the initial population is assigned a unique heterozygous genotype at a single neutral locus and descendent genotypes are determined according to Mendel's principles, to track the loss of genetic variability over time (Lacy et al. 2018). Additional details are provided in Text S1 and S2 (Supplemental Material).

Head-started Blanding's turtles have been released annually since 2007 and predators (primarily raccoons Procyon lotor) have been removed annually since 2013 at SBCP (Urbanek et al. 2016; Thompson et al. 2020). However, head-starts had not yet reached reproductive maturity as of the completion of data collection in 2018 (Thompson et al. 2020) and thus do not affect estimates of adult mortality and population size. In addition, for PVA, we excluded head-starts from estimates of juvenile mortality (Golba 2019) and we used nest survival rates observed prior to predator removal in estimating age 0 mortality. Thus, while parameter estimates in our SBCP model may reflect the effects of habitat management (prescribed fire, chemical and mechanical methods), we sought to minimize effects of head-starting and predator removal.

Zero-growth model.

Our SBCP model resulted in positive population growth and low extinction risk (Results), making extinction risk of limited use in assessing sensitivity and contrasting with separate analyses indicating that adult SBCP Blanding's turtle population size is stable or only slightly increasing (personal observation). For these reasons we created a zero-growth model by increasing age-specific mortality rates and reducing clutch size to achieve a deterministic population growth rate close to zero. Catastrophes were implemented as in the Spring Bluff Chiwaukee Prairie model.

Sensitivity to parameter uncertainty.

We assessed the effect of parameter uncertainty on PVA outcomes using our zero-growth model and a combination of single-variable and multivariable tests in Vortex. Single-variable tests consisted of 1,000 iterations at each of 10 uniformly distributed parameter values while holding other parameters constant. We included 17 parameters in single parameter tests (Table 1) and evaluated them for the magnitude of their effects on population growth rate and probability of extinction. We based ranges of parameter values used in sensitivity tests on 95% confidence limits for parameter estimates observed at SBCP to the extent possible (we shifted clutch size and mortality CIs to match shifts in parameter estimates necessary to achieve zero growth). For parameters for which confidence limits were lacking, we derived plausible values from knowledge of Blanding's turtle biology. Specifically, we modeled inbreeding depression by assuming 0–7 lethal equivalents with 50% due to lethal recessive alleles, thus somewhat exceeding the mean number of lethal equivalents (6.29) affecting fecundity and first-year survival in a meta-analysis of wild birds and mammals (O'Grady et al. 2006). We allowed age of first reproduction to vary between 12 and 14, allowing for 1 y uncertainty in our assessment of the age of primiparous females (we rounded values selected for age of first reproduction to integer values within Vortex). We allowed maximum lifespan and maximum age of reproduction to vary jointly from 55 to 85 y. We allowed mean offspring sex ratio to vary from 40 to 60% males. We allowed carrying capacity to vary from 104 to 312 and the environmental SD in carrying capacity to vary from 0 to 50 adults. We allowed the environmental SD in juvenile and adult mortality to vary jointly from 0 to 4%.

Table 1.

Demographic parameter values used in Blanding's turtle Emydoidea blandingii population viability analyses and sensitivity tests. Parameter values for the Spring Bluff Chiwaukee Prairie (SBCP) model and in sensitivity tests of parameter uncertainty were estimated from data collected at the SBCP study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text. Clutch size and mortality rates were adjusted to achieve deterministic r = 0 for the zero-growth model. Parameter values used in sensitivity tests of range-wide parameter uncertainty are from throughout the species' range in North America (Table S11, Supplemental Material). Parameters are ordered to match their order in Vortex. SD refers to environmental standard deviation.

Demographic parameter values used in Blanding's turtle Emydoidea blandingii population viability analyses and sensitivity tests. Parameter values for the Spring Bluff Chiwaukee Prairie (SBCP) model and in sensitivity tests of parameter uncertainty were estimated from data collected at the SBCP study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text. Clutch size and mortality rates were adjusted to achieve deterministic r = 0 for the zero-growth model. Parameter values used in sensitivity tests of range-wide parameter uncertainty are from throughout the species' range in North America (Table S11, Supplemental Material). Parameters are ordered to match their order in Vortex. SD refers to environmental standard deviation.
Demographic parameter values used in Blanding's turtle Emydoidea blandingii population viability analyses and sensitivity tests. Parameter values for the Spring Bluff Chiwaukee Prairie (SBCP) model and in sensitivity tests of parameter uncertainty were estimated from data collected at the SBCP study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text. Clutch size and mortality rates were adjusted to achieve deterministic r = 0 for the zero-growth model. Parameter values used in sensitivity tests of range-wide parameter uncertainty are from throughout the species' range in North America (Table S11, Supplemental Material). Parameters are ordered to match their order in Vortex. SD refers to environmental standard deviation.

Multivariable tests included a subset of nine parameters identified as having moderate to large effects in single-variable tests or of intrinsic interest. We used Latin hypercube sampling to select 30,000 unique uniformly distributed combinations of parameter values and ran a single iteration of each (Prowse et al. 2016). We evaluated the sensitivity of PVA outcomes to variation in these nine parameters using logistic regression (McCarthy et al. 1995; Cross and Beissinger 2001). Analyses focused on four response variables: stochastic population growth rate (observed per capita growth rate averaged across years and iterations = stochastic-r), probability of extinction, gene diversity (proportion of initial expected heterozygosity remaining), and adult population size. To facilitate analysis via logistic regression, we transformed stochastic-r, gene diversity, and adult population size into binomial variables by equating values below and above the median equal to 0 and 1, respectively (with single iterations, probability of extinction is necessarily binomial). We opted to use logistic regression (vs. e.g., linear regression) to analyze stochastic-r, gene diversity, and adult population size because of nonnormality (gene diversity and adult population size equal 0 for simulations resulting in extinction, giving strongly bimodal distributions) and because relationships to independent variables are potentially nonlinear. We conducted multiple logistic regression analyses using IBM SPSS 25 with forced entry of the nine parameters as independent variables and stochastic-r, probability of extinction, gene diversity, or adult population size (analyzed separately) as dependent variables. We assessed sufficiency of sampling by selecting three sets of 10,000 random samples each and repeating analyses on each subsample (Prowse et al. 2016). We standardized partial logistic regression coefficients by dividing each by its standard error, providing a relative measure of the influence of each parameter (Cross and Beissinger 2001). To facilitate comparisons among dependent variables and among data subsamples, we scaled standardized partial logistic regression coefficients so that absolute values summed to 100.

Sensitivity to geographic parameter variation.

We reviewed published and unpublished sources for information on geographic variation in demographic parameters across the Blanding's turtle distribution. We excluded estimates of juvenile survival based on studies of head-started turtles (Arsenault 2011; d'Entremont 2014; Ritchie 2017; Starking-Szymanski et al. 2018; Golba 2019) or inferred from survival of other age classes by assuming constant population size (Congdon et al. 1993; Hawkins 2016). For sites where long-term study has produced multiple estimates, we used the estimate derived from the longest time interval. We assessed the effect of geographic parameter variation on PVA outcomes by repeating our multivariable tests in Vortex using an expanded range of parameter values. For parameters for which only our SBCP estimate was available (SD of age 0 mortality, juvenile mortality, SD of juvenile and adult mortality), we used parameter uncertainty to set the range of parameter values.

Sensitivity to population size.

We used our zero-growth model to investigate the impact of population size on PVA outcomes with and without catastrophes. We included a low, intermediate, and high levels of environmental stochasticity scenarios by setting the environmental SD in percent of females reproducing (hereafter, “% females”), clutch size, and percent mortality to the lower limits used in sensitivity testing, the base values used in the zero-growth model, and the upper limits used in sensitivity testing (Table 1). We implemented catastrophes as in the Spring Bluff Chiwaukee Prairie and zero-growth models. For each scenario, we ran 1,000 iterations with initial population sizes ranging from 5 to 200 adults (total population = 15 to 585 with a stable age distribution) and durations of 50 and 100 y.

Results

Demography

Nine females, first captured as juveniles, initiated reproduction during our study. The youngest of these primiparous females was 13 y old when she first reproduced. Estimates of reproductive frequency among females from 13 to 18 y of age ranged from 30 to 89% (Table S2, Figure S1, Supplemental Material). Among adult females of unknown age (n = 11–48/y from 2012 to 2018), we detected that females were gravid in 185 of 200 cases (92.5%, 95% CI = 87.9–95.7%). Environmental variance in reproductive frequency (and the environmental SD as used in Vortex) was estimated to equal 0.00 (Table S3, Supplemental Material). From 2008 to 2018, we obtained 120 clutches from 41 females (1–7/female; Table S4, Supplemental Material). Number of eggs per clutch averaged 13.0 (range = 7–24; Table S4, Supplemental Material). The estimated variance component attributable to year was 0.74 (environmental SD = 0.86).

We monitored 13 nests at SBCP prior to implementation of mesopredator control and 15 nests at IBSP, a site without mesopredator control. At SBCP, 12 nests were depredated; at IBSP, 6 nests were depredated and 3 others failed because of shoreline erosion, resulting in a clutch failure rate of 75%. We monitored 60 nests at SBCP during years with mesopredator control (n = 6–14 nests/y from 2013 to 2018 and observed a clutch failure rate of 33.3% (0–64.2%/y). Hatch failure among 1,380 eggs incubated for the Lake County Forest Preserve District head-starting program (n = 117–222 eggs/y from 2010 to 2018) averaged 19% (Table S5, Supplemental Material). Posthatching survival of 82 hatchlings tracked for up to 88 d using telemetry was 79.6% (standard error = 0.06; Kastle et al. in press). Extrapolating to the entire posthatch and overwinter period (ca. 240 d) using a Weibull survival function gave an anticipated mortality of 35% (Kastle et al. in press). Combining these components, age 0 mortality = 1 − [(1 − 0.750) × (1 − 0.187) × (1 − 0.350)] = 0.868 (86.8%) with uncertainty (sampling variance) = 0.0020 and 95% CI = 78.0–95.6% (Table S6.A, Supplemental Material).

Based on 238 nests monitored at a Michigan study site (n = 4–16 nests/y from 1976 to 1998; Congdon et al. 2000), environmental variance in clutch failure = 0.034 (Table S7, Supplemental Material). At SBCP, environmental variance in hatch failure was 0.0043 (Table S5, Supplemental Material). An estimate of environmental variance in posthatch mortality is lacking, but assuming it falls between 0.000 and 0.034 (the variance in clutch failure) results in a combined estimate of the environmental variance in age 0 mortality of 0.0096–0.0110 (environmental SD = 0.0980–0.1050 or 9.8–10.5%; Table S6.B, Supplemental Material).

Age-specific juvenile mortality decreased from 28.9% at age 1 to 14.6%, 6.9%, 3.6%, 2.5%, and 2.2% at age 2, 3, 4, 5, and 6 and older (converted from survival estimates in Golba 2019). Confidence intervals were broadest at age 1 (10.9–48.4%) and generally decreased with age (to 0.8–5.7% for age 6 and older; Golba 2019). Adult mortality was estimated to be 5.3% (CI = 3.8–7.3%) with environmental variance = 0.0011 (environmental SD = 0.0332 or 3.3%; Golba 2019). Adult population size estimates ranged from 62 to 111 among years (Table S8, Supplemental Material). For population viability analysis, we used the maximum 3-y running average ( = 104 over the years 2015–2017, CI = 87–138) as our initial adult population size and assumed a stable age distribution to determine the total population size.

Population viability and sensitivity

Spring Bluff Chiwaukee Prairie model.

In the absence of catastrophes, SBCP model settings and parameter values (Table 1; Text S1, S2, Supplemental Material) resulted in a generation time of 25 y, deterministic population growth rate of 5.8%/y, and stochastic population growth rate of 5.5%/y. No extinctions occurred among 1,000 iterations. Population size grew to carrying capacity within about 20 y (Figure S2, Supplemental Material). Extant populations averaged 1,031 individuals (253 adults) and retained 99% of initial genetic heterozygosity after 100 y. Individual iterations showed wide variation in population size over time (Figure 2). This variation was at least partly due to the large environmental SD of age 0 mortality (10.5%), which, when coupled with high levels of age 0 mortality (86.8%), resulted in runs of years with no surviving offspring and declining population size interspersed with occasional boom years of higher rates of offspring survival and rapid increases in population size (Figure 2; Figure S3, Supplemental Material). Including catastrophes had only small effects compared with the no-catastrophe scenario; deterministic population growth rate was 4.8%, stochastic population growth was 4.5%, no extinctions occurred among 1,000 iterations, and extant population size averaged 993 individuals (240 adults) and retained 99% of initial genetic heterogeneity after 100 y.

Figure 2.

Single iterations of the zero-growth Blanding's turtle Emydoidea blandingii population model showing the impact of the presence (blue) or absence (red) of environmental variation in age 0 survival on variation in population size. Parameter values were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1).

Figure 2.

Single iterations of the zero-growth Blanding's turtle Emydoidea blandingii population model showing the impact of the presence (blue) or absence (red) of environmental variation in age 0 survival on variation in population size. Parameter values were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1).

Zero-growth model.

Zero growth was achieved by increasing age-specific mortality by 4.5% for ages 0–4, 3.5% for ages 5–9, and 2.5% for ages 10 and above and reducing clutch size from 13.0 to 10.5 (Table 1), yielding a generation time of 25 y, deterministic population growth rate of 0.00%/y and stochastic population growth rate of –0.40%/y (Figure S2, Supplemental Material). Three extinctions occurred among 1,000 iterations (probability of extinction = 0.003). Extant populations averaged 248 individuals (88 adults) and retained 96.3% of initial genetic heterozygosity after 100 y. Including catastrophes had mostly modest effects compared with the no-catastrophe scenario; deterministic population growth rate was −1.0%, stochastic population growth was −1.6%, 29 extinctions occurred among 1,000 iterations (probability of extinction = 0.029), and 93% of initial genetic heterogeneity was retained after 100 y. Catastrophes had a larger effect on extant population size, which averaged 110 individuals (38 adults), about 44% that of the no-catastrophe scenario.

Sensitivity to parameter uncertainty.

In single-variable tests, uncertainty in age 0 mortality, juvenile mortality, and adult mortality had moderate to large effects on stochastic-r and probability of extinction (Table S9, Supplemental Material). Uncertainty in offspring sex ratio, age at first reproduction, % females reproducing, clutch size, the environmental SD in age 0 mortality, and the environmental SD in juvenile and adult mortality had more modest effects (Table S9, Supplemental Material). Uncertainty in other parameters had generally negligible effects (Table S9, Supplemental Material).

We selected nine variables for inclusion in multivariable sensitivity tests, age at first reproduction, % females reproducing, clutch size, age 0 mortality, the environmental SD in age 0 mortality, juvenile mortality, adult mortality, environmental SD in juvenile and adult mortality (combined), and initial population size. Although uncertainty in initial population size had a negligible effect in single-variable tests, its demographic significance led us to include it regardless. We did not include uncertainty in offspring sex ratio because, though it had a modest effect in single-variable tests, there is little information on offspring sex ratios in nature to guide selection of meaningful values for sensitivity testing.

Across 30,000 samples, stochastic-r ranged from −0.150 to 0.070 (median = −0.011, stochastic-r equaled or exceeded zero in 14,696 samples), gene diversity ranged from 0.00 to 0.991 (median = 0.961), and adult population size ranged from 0 to 363 (median = 43). Probability of extinction was 0.159 with median time to extinction of 78 y (range = 28–100 y; 5.5% of extinctions occurred within the first 50 y). Logistic regression revealed that uncertainty in age 0 mortality, followed by juvenile mortality and then adult mortality, had the largest effects on stochastic-r, probability of extinction, gene diversity, and adult population size with standardized relative influence of 19–33% (Table 2, Figure 3; Table S10, Figure S4, S5, S6, Supplemental Material). Uncertainty in age of first reproduction, environmental variation in age 0 mortality, % females reproducing, clutch size, and environmental variation in juvenile and adult mortality had smaller effects. In general, parameters related to mortality rates, together with initial population size, more strongly influenced extinction probability; whereas, parameters related to reproduction more strongly influenced stochastic-r, genetic diversity, and adult population size (Table 2, Figure 3; Table S10, Figure S4, S5, S6, Supplemental Material). Analysis of random subsamples of n = 10,000 gave highly concordant results (standardized relative influence varied by <1.8 from subsample to subsample; Table S10, Supplemental Material), confirming sufficiency of sampling.

Table 2.

Relative influence of parameter uncertainty and geographic parameter variation on Blanding's turtle Emydoidea blandingii population growth (Stochastic-r), probability of extinction, gene diversity, and adult population size. Font type indicates relative influence (from greatest to least: bold, italic, regular); sign indicates direction. SD refers to environmental standard deviation. Parameter values in sensitivity tests of parameter uncertainty were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1). Parameter values used in sensitivity tests of range-wide parameter uncertainty are from throughout the species' range in North America (Table 1; Table S11, Supplemental Material).

Relative influence of parameter uncertainty and geographic parameter variation on Blanding's turtle Emydoidea blandingii population growth (Stochastic-r), probability of extinction, gene diversity, and adult population size. Font type indicates relative influence (from greatest to least: bold, italic, regular); sign indicates direction. SD refers to environmental standard deviation. Parameter values in sensitivity tests of parameter uncertainty were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1). Parameter values used in sensitivity tests of range-wide parameter uncertainty are from throughout the species' range in North America (Table 1; Table S11, Supplemental Material).
Relative influence of parameter uncertainty and geographic parameter variation on Blanding's turtle Emydoidea blandingii population growth (Stochastic-r), probability of extinction, gene diversity, and adult population size. Font type indicates relative influence (from greatest to least: bold, italic, regular); sign indicates direction. SD refers to environmental standard deviation. Parameter values in sensitivity tests of parameter uncertainty were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1). Parameter values used in sensitivity tests of range-wide parameter uncertainty are from throughout the species' range in North America (Table 1; Table S11, Supplemental Material).
Figure 3.

Sensitivity of stochastic-r to parameter uncertainty in the zero-growth Blanding's turtle Emydoidea blandingii model. Parameter values were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and mortality adjusted to achieve deterministic r = 0 (Table 1). For each relationship, other parameters were held constant at their mean values. P(r > median) refers to the probability that the average population growth rate is above the median (−0.011). EV refers to environmental variation (standard deviation).

Figure 3.

Sensitivity of stochastic-r to parameter uncertainty in the zero-growth Blanding's turtle Emydoidea blandingii model. Parameter values were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and mortality adjusted to achieve deterministic r = 0 (Table 1). For each relationship, other parameters were held constant at their mean values. P(r > median) refers to the probability that the average population growth rate is above the median (−0.011). EV refers to environmental variation (standard deviation).

Sensitivity to geographic parameter variation.

We documented considerable geographic variation in Blanding's turtle demography, including variation in age at first reproduction (10–22 y), % females reproducing (40.5–94.0%), clutch size (7.4–17.7 eggs), and adult mortality (3.5–31.0%; Table S11, Supplemental Material). Components of age-0 mortality were also variable (Table S11, Supplemental Material), including nest failure (33–94%), hatch failure (13–53%), and hatch-to-hibernation mortality (20–80%). Combining variation in nest failure and hatch failure with interpolated hatch-to-spring emergence mortality gave lower and upper bounds on age 0 mortality of 72–99% (Table S11, Supplemental Material; Kastle et al. in press).

We repeated our multivariable sensitivity tests using observed geographic variation in age at first reproduction, % females reproducing, clutch size, age 0 mortality, and adult mortality. Ranges for juvenile mortality, environmental SD in age 0 mortality, and environmental SD in juvenile and adult mortality were identical to those used in sensitivity tests of parameter uncertainty. We varied initial population size from 15 to 403 (equivalent to a stable age distribution with ca. 5 to 138 adults), thus encompassing small to moderately sized Blanding's turtle populations. Larger Blanding's turtle populations occur at some sites but, given the results of our parameter uncertainty sensitivity tests, increasing initial population size above ca. 400 is unnecessary in evaluating population viability.

Across 30,000 samples, stochastic-r ranged from −0.558 to 0.131 (median = −0.077, stochastic-r equaled or exceeded zero in 4,705 samples), gene diversity ranged from 0.000 to 0.991 (median = 0.000 because gene diversity following extinction = 0), and adult population size ranged from 0 to 670 (median = 0). Probability of extinction was 0.691 with median time to extinction of 36 years (range = 4–100 y; 71.2% of extinctions occurred within the first 50 y). The median was zero for both gene diversity and adult population size, so logistic regression analysis was restricted to stochastic-r (transformed to a binomial variable as before) and probability of extinction. As in sensitivity tests of parameter uncertainty, geographic variation in age-specific mortality consistently had strong effects on both stochastic-r and probability of extinction (Table 2, Figure 4; Table S10, Figure S7, Supplemental Material). In addition, geographic variation in age of first reproduction, clutch size, and % females reproducing had greater effects on stochastic-r and probability of extinction than was the case for sensitivity tests of parameter uncertainty (Table 2, Figure 4; Table S10, Figure S7, Supplemental Material). Variation in initial population size had a noticeable impact on probability of extinction but little impact on stochastic-r (Table 2, Figure 4; Table S10, Figure S7, Supplemental Material). Effects of environmental variation in age 0 mortality and in juvenile and adult mortality were small (Table 2, Figure 4; Table S10, Figure S7, Supplemental Material).

Figure 4.

Sensitivity of stochastic-r to geographic parameter variation in the zero-growth Blanding's turtle Emydoidea blandingii population model. Parameter values are from throughout the species' range in North America from 1986 to 2018 (Table 1; Table S11, Supplemental Material). For each relationship, other parameters were held constant at their mean values. P(r > median) refers to the probability that the average population growth rate is above the median (−0.077). EV refers to environmental variation (standard deviation).

Figure 4.

Sensitivity of stochastic-r to geographic parameter variation in the zero-growth Blanding's turtle Emydoidea blandingii population model. Parameter values are from throughout the species' range in North America from 1986 to 2018 (Table 1; Table S11, Supplemental Material). For each relationship, other parameters were held constant at their mean values. P(r > median) refers to the probability that the average population growth rate is above the median (−0.077). EV refers to environmental variation (standard deviation).

Parameter values resulting in persistence (vs. extinction) and having stochastic-r ≥ 0 included the full range of values sampled except for age 0 mortality (only values <97.7% resulted in persistence with stochastic-r ≥ 0). However, combinations of parameter values resulting in persistence with stochastic-r ≥ 0 were more restrictive. For example, high rates of adult mortality (e.g., ≥15%) resulted in persistence with stochastic-r ≥ 0 only if age 0 mortality was low (e.g., <89%) and reproduction commenced at a young age (e.g., ≤16 y; Figure 5). Partial correlation analysis revealed that among samples resulting in persistence and stochastic-r ≥ 0, age of first reproduction was positively correlated with % females reproducing and with clutch size and negatively correlated with mortality rates (Table 3). That is, as age at first reproduction increased, % females reproducing and clutch size increased and mortality rates, especially adult mortality rate, decreased (Table 3). Similarly, as % females reproducing and clutch size increased, mortality rates also increased (Table 3). As % females reproducing increased, clutch size decreased; and as mortality in one age class increased, mortality in other age classes decreased (Table 3).

Figure 5.

Combinations of parameter values for annual adult mortality, annual age 0 mortality, and age at first reproduction resulting in persistence with stochastic-r ≥ 0 (green points), persistence but with stochastic-r < 0 (black points) or extinction (gray points) in sensitivity analyses of Blanding's turtle Emydoidea blandingii geographic parameter variation. Parameter values are from throughout the species' range in North America from 1986 to 2018 (Table 1; Table S11, Supplemental Material).

Figure 5.

Combinations of parameter values for annual adult mortality, annual age 0 mortality, and age at first reproduction resulting in persistence with stochastic-r ≥ 0 (green points), persistence but with stochastic-r < 0 (black points) or extinction (gray points) in sensitivity analyses of Blanding's turtle Emydoidea blandingii geographic parameter variation. Parameter values are from throughout the species' range in North America from 1986 to 2018 (Table 1; Table S11, Supplemental Material).

Table 3.

Bivariate correlations (above diagonal) and partial correlations (holding all other parameters constant, below diagonal) among Blanding's turtle Emydoidea blandingii demographic parameters resulting in population persistence and stochastic-r ≥ 0. SD refers to environmental standard deviation. Correlations with absolute values >0.20 are shown in bold; correlations with absolute values between 0.1 and 0.2 are italicized. Results are based on n = 4,704 iterations (out of a total of 30,000) generated in sensitivity analyses of geographic parameter variation using parameter values from throughout the species' range in North America from 1986 to 2018 (Table 1; Table S11, Supplemental Material).

Bivariate correlations (above diagonal) and partial correlations (holding all other parameters constant, below diagonal) among Blanding's turtle Emydoidea blandingii demographic parameters resulting in population persistence and stochastic-r ≥ 0. SD refers to environmental standard deviation. Correlations with absolute values >0.20 are shown in bold; correlations with absolute values between 0.1 and 0.2 are italicized. Results are based on n = 4,704 iterations (out of a total of 30,000) generated in sensitivity analyses of geographic parameter variation using parameter values from throughout the species' range in North America from 1986 to 2018 (Table 1; Table S11, Supplemental Material).
Bivariate correlations (above diagonal) and partial correlations (holding all other parameters constant, below diagonal) among Blanding's turtle Emydoidea blandingii demographic parameters resulting in population persistence and stochastic-r ≥ 0. SD refers to environmental standard deviation. Correlations with absolute values >0.20 are shown in bold; correlations with absolute values between 0.1 and 0.2 are italicized. Results are based on n = 4,704 iterations (out of a total of 30,000) generated in sensitivity analyses of geographic parameter variation using parameter values from throughout the species' range in North America from 1986 to 2018 (Table 1; Table S11, Supplemental Material).

Sensitivity to population size.

In the absence of catastrophes, extinction risk dropped rapidly with increasing initial population size for all three environmental stochasticity scenarios (Figure 6; Figure S8, Supplemental Material). The drop was most rapid when environmental variation was low and somewhat less rapid as environmental variation increased. At initial population sizes above ca. 100 adults, differences among environmental stochasticity scenarios were negligible. A projected extinction rate <5% over 100 y required an initial population size ≥20–50 adults, depending on scenario (Figure 6); a projected extinction rate <5% over 50 y required an initial population size ≥10–20 adults, depending on scenario (Figure S8, Supplemental Material). The proportion of genetic diversity retained increased rapidly with increasing initial population size for all three environmental stochasticity scenarios (Figure 6). The increase was most rapid when environmental variation was zero and was somewhat less rapid as environmental variation increased. At initial population sizes above 200 adults, differences among scenarios were negligible. To retain ≥95% of genetic diversity over 100 y required an initial population size ≥50–110 adults, depending on scenario (Figure 6); to retain ≥95% of genetic diversity over 50 y required an initial population size ≥25–40 adults, depending on scenario (Figure S8, Supplemental Material).

Figure 6.

Relationship of probability of extinction (upper panel) and genetic diversity (lower panel) to initial adult Blanding's turtle Emydoidea blandingii population size with (filled diamonds) and without (open circles) catastrophes in scenarios lasting 100 y. Parameter values were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1). Solid lines represent a scenario with intermediate environmental stochasticity and are bracketed by dashed (with catastrophes) and dotted (without catastrophes) lines that represent scenarios with low and high environmental stochasticity. Solid horizontal solid lines correspond to 5% probability of extinction (upper panel) and 95% retention of genetic diversity (lower panel).

Figure 6.

Relationship of probability of extinction (upper panel) and genetic diversity (lower panel) to initial adult Blanding's turtle Emydoidea blandingii population size with (filled diamonds) and without (open circles) catastrophes in scenarios lasting 100 y. Parameter values were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1). Solid lines represent a scenario with intermediate environmental stochasticity and are bracketed by dashed (with catastrophes) and dotted (without catastrophes) lines that represent scenarios with low and high environmental stochasticity. Solid horizontal solid lines correspond to 5% probability of extinction (upper panel) and 95% retention of genetic diversity (lower panel).

Including catastrophes resulted in a more gradual drop in extinction risk with increasing initial population size and greater population size thresholds to achieve conservation criteria (Figure 6; Figure S8, Supplemental Material). A projected extinction rate <5% over 100 y required an initial population size ≥50–200 adults depending on environmental stochasticity scenario (Figure 6). A projected extinction rate <5% over 50 y required an initial population size ≥20–30 adults depending on scenario (Figure S8, Supplemental Material). To retain ≥95% of genetic diversity over 100 y required an initial population ≥110 to >200 adults depending on scenario (Figure 6). To retain ≥95% of genetic diversity over 50 y required an initial population size ≥25–65 adults depending on scenario (Figure S8, Supplemental Material).

Discussion

Demography

Long-term study at SBCP provides an unusually complete assessment of Blanding's turtle demography. Such studies are necessary to understand temporal heterogeneity in demographic processes and develop predictive models (Reinke et al. 2019). As is the case elsewhere (Congdon et al. 1993, 2000; Reid et al. 2016), Blanding's turtles at SBCP experience low nest survival (25%), delayed reproductive maturity (13 y), and high rates of adult survival (94.7%). Our study goes further to fill knowledge gaps regarding Blanding's turtle demography. In particular, we confirm the existence of a positive relationship between the % of females reproducing and age among young adults, as hypothesized by Congdon et al. (1993). The SBCP study also provides more comprehensive estimates of juvenile survival than previously available. For example, prior studies have equated age 0 survival to nest success or nest success + hatch success (Congdon et al. 1993, 2000), whereas we combine these components with posthatch survival to estimate survival from egg deposition to resumption of activity in spring (Kastle et al. in press). Similarly, juvenile survival beyond age 0 has sometimes been represented by a single value inferred by computing the mean annual survival necessary to maintain a constant population size given observed rates of nest predation and adult survival (Congdon et al. 1993, 2000; Hawkins 2016). Capture–mark–recapture analysis of data from SBCP provides age-specific estimates of survival for many juvenile age classes (1, 2, 3, 4, 5, and 6+ juveniles; Golba 2019). These refinements provide a more complete understanding of Blanding's turtle survival and reproduction and facilitate population viability analysis.

We also provide the first estimates of environmental variation in Blanding's turtle demographic parameters, including environmental variation in % females reproducing, clutch size, age 0 mortality, and (from Golba 2019) adult mortality. We show that environmental variation in % females reproducing is surprisingly low (environmental SD = 0.00%) and although this estimate is based on just 7 y of data from SBCP, analysis of 18 y of data from the DuPage County, Illinois head-starting program (Thompson et al. 2020) gives identical results (% reproducing = 93%, environmental SD = 0.00%). Among other freshwater turtles, total variation in % females reproducing is greater in mud turtles (SD = 10.1%) and pond sliders (13.9%; Frazer et al. 1991) than we observed in Blanding's turtles, but the amount attributable to environmental vs. error variance is not known. We found that environmental variation in clutch size and adult survival were somewhat greater than for % females reproducing (environmental SD = 0.9 offspring and 3.3%, respectively), but what is most striking is the large amount of environmental variation in age 0 mortality (environmental SD = 10.5%). This is primarily due to high levels of year-to-year variance in nest predation rates (Congdon et al. 2000), a feature that may be common to fresh-water, estuarine, and marine turtle demographics (Engeman et al. 2016; Feinberg and Burke 2003; Schwanz et al. 2010; Munscher et al. 2012). Population viability analysis indicates that large amounts of environmental variation in age 0 mortality has the potential to generate dramatic changes in population size and age structure over time as a result of runs of years with 100% age 0 mortality (Figure 3; Figure S3, Supplemental Material). Predator removal began at SBCP in 2013 (Urbanek et al. 2016), so information on year-to-year variance in natural predation rates on Blanding's turtle nests is limited to Congdon's long-term study in Michigan (Congdon et al. 2000). Neither the generality of Congdon's result, nor its actual impact on population size and age structure are known. However, in other vertebrates, fluctuations in age structure affect population dynamics (Hoy et al. 2019). Comparable data on nest failure in painted turtles (table 3 in Schwanz et al. 2010) yield an estimated environmental variance of 0.064 as a result of nest predation and 0.054 if failure due to other causes (mostly parasite or fungus infestation) are included. These values are roughly similar to that computed for Michigan Blanding's turtles (0.034), suggesting that the environmental variance used in our analyses is not atypical. Hatch failure rates of intact nests of Michigan Blanding's turtles (19.5%, Congdon et al. 2000) are also similar to what we observed for SBCP (19%), despite our estimate being based on the results of artificial incubation.

Meaningful estimates of environmental variation in demographic parameters require long-term study, particularly if temporal trends or other variance sources are to be identified (Gould and Nichols 1998; Burnham and Anderson 2002; Anderson 2008; Reinke et al. 2019). As a consequence, the number of species for which estimates of environmental variation are available is limited (Franklin et al. 2002; King et al. 2018; Milligan et al. 2018; Cayuela et al. 2019). Among turtles, we are aware of just one other such estimate. In bog turtles, the environmental SD in survival is 7.2% for young juveniles and 1.9% for adults (computed from coefficients of variation in Shoemaker et al. 2013), values similar to what we report for age 0 and adult Blanding's turtles (10.5% and 3.3% respectively). In the absence of such estimates, values used when generating population projections are sometimes arbitrary and potentially inaccurate. For example, one might be tempted to use default Vortex settings for the environmental SD of % females reproducing (10%), age 0 mortality (10%), juvenile mortality (3%) and adult mortality (3%). And in the absence of long-term study with repeated observations of the same females, one might use the observed SD in clutch size (= 3.3 offspring at SBCP, a value that includes both environmental and sampling variation) in lieu of a variance component estimate (environmental SD = 0.9 offspring at SBCP). Population projections are likely to be biased as a result, potentially causing unwarranted optimism or pessimism about conservation status. Despite spanning 15 y, data from SBCP still represent <1 Blanding's turtle generation. Consequently, temporal trends in environmental variation and the magnitude of environmental variation over longer time frames remains unknown.

In turtle populations, an absence of juveniles has sometimes been interpreted to indicate unsustainably low recruitment (Browne and Hecnar 2007; Howell et al. 2019), leading managers to implement corrective measures aimed at increasing juvenile survival (nest caging, artificial hibernation, head-starting, predator control). Our results suggest that large amounts of variance in age 0 survival can also result in an absence of some juvenile classes (Figure S1, Supplemental Material). This observation, when coupled with possible differences in detection probability between juvenile and adult Blanding's turtles, suggests caution when interpreting observed population age structure. Possibly, years with high rates of recruitment may occur frequently enough to ensure population persistence. This is not to say that measures to increase juvenile survival are ineffective; our sensitivity analyses suggest that such efforts should fuel population growth and reduce extinction risk. Alternatively, surplus animals could be used in augmentation or reintroduction programs (Spencer et al. 2017).

Our review of Blanding's turtle studies demonstrates that as in other turtles (Miller and Dinkelacker 2007), demographic parameters vary geographically. For example, we report two-fold or greater differences in age at first reproduction, % females reproducing, clutch size, components of age 0 mortality, and adult mortality. Some of this variation may reflect geographic clines (e.g., age at first reproduction = 10–11 y in Nebraska, 12–14 y in the Midwest, and 17–22 y in Nova Scotia), suggesting a latitudinal gradient as seen in Wood turtles (Greaves and Litzgus 2009). Other parameters vary greatly even at a fine spatial scale. For example, mean clutch size varies from 8 to 14.5 among northern Illinois Blanding's turtle populations, presumably as a consequence of corresponding variation in mean female size (Ruane et al. 2008). Unfortunately, data are too few to characterize geographic clines and parameter covariation statistically as has been done for some reptiles (Moll 1973; Iverson et al. 1993; Ashton and Feldman 2003; Litzgus and Mousseau 2006; Greaves and Litzgus 2009; Jones et al. 2012; Hileman et al. 2017), leaving open the question of whether variation in one parameter (e.g., age at first reproduction) is offset by compensatory variation in other parameters (% females reproducing, clutch size, survival) as predicted from partial correlations among parameters resulting in population persistence and stochastic-r ≥ 0 in our sensitivity analysis of geographic parameter variation.

Population viability and sensitivity

Population viability analysis of the SBCP Blanding's turtle population suggests a low risk of extinction, even after mortality rates were increased and clutch size was reduced to achieve zero population growth. This result is consistent with the realized adult population growth parameter, which indicates that population size at SBCP is stable or only slightly increasing (95% confidence limits for λ = 0.99–1.03 for males and 1.01–1.07 for females; personal observation). Including catastrophes (at an average rate of one per generation and with a reduction in survival to 75% base-line levels) had only small to modest effects on PVA outcomes except for population size, which was markedly smaller (43%) in zero-growth models. Sensitivity tests designed to encompass uncertainty in parameter estimates at SBCP also result in a somewhat more cautionary interpretation; with parameter uncertainty, probability of extinction over 100 y is 16% compared to 0% when uncertainty is ignored. Uncertainty in age 0, juvenile, and adult mortality rates were the largest contributors to population growth rate and probability of extinction. The environmental SD in age 0 mortality had a modest effect on extinction risk and age at first reproduction had a modest effect on population growth rate. Uncertainty varies among parameters (e.g., uncertainty is greater for age 0 and juvenile mortality than for adult mortality), so our sensitivity tests of parameter uncertainty differ from what might be generated, for example, by using matrix methods (Mills et al. 1999; Caswell 2019). Thus, although life-table and matrix methods suggest that adult mortality may have the largest impact on Blanding's turtle population viability (Congdon et al. 1993, 2000; Heppell 1998), our results indicate that precise estimates of other demographic parameters (age 0 mortality, juvenile mortality) may be equally important in generating meaningful population projections. Our results also serve to identify demographic parameters that might effectively be targeted for research and management (Klein et al. 2017; Manlik et al. 2017). For example, viability analyses could be improved with more precise estimates of age-specific mortality. Given that adult mortality is already relatively low, management efforts might most effectively be focused on reducing age 0 and juvenile mortality and their environmental variation. Put simply, it may be easier to reduce age 0 mortality by 10% than to reduce adult morality by 1%. Similarly, although adult mortality was the most influential contributor to population growth in diamondback terrapins, reducing age 0 mortality is also necessary to ensure population growth (Crawford et al. 2014). This and other studies (e.g., Reed et al. 2009; Klein et al. 2017; Spencer et al. 2017; Mullin et al. 2020) argue for pluralistic approaches to turtle conservation management.

Sensitivity tests designed to encompass geographic variation in demography demonstrate that in addition to mortality rates, age at first reproduction, % females reproducing, clutch size, and initial population size all have the potential to influence Blanding's turtle population growth, extinction risk, or both. Furthermore, population growth and persistence is possible for nearly the full range of values sampled for any given parameter, but only when other parameters fall within certain bounds; of 30,000 parameter value combinations, just 16% resulted in population persistence with stochastic-r ≥ 0. Thus, although seven of eight estimates of annual adult mortality were <6.5%, the exception, a Nebraska population where adult mortality = 31% (Ruane et al. 2008) could represent a viable population if the values of other parameters compensated for high rates of adult mortality. First reproduction is estimated to occur at 10 y at this site (Germano et al. 2000), which is the youngest age of first reproduction of any Blanding's turtle population. High and female-biased (41% in adult females vs. 10% in adult males) mortality at this site has been attributed transportation infrastructure; a state highway lies immediately south and a two-track rail line lies immediately north of wetlands inhabited by Blanding's turtles (Ruane et al. 2008). However, given the early age of reproductive maturity at the site, Blanding's turtles may persist even with adult mortality rates that exceed those reported elsewhere. In contrast, in Nova Scotia, where age of reproductive maturity is ≥17 y (Standing 1997; McNeil 2002; The Blanding's Turtle Recovery Team 2003), low mortality rates, perhaps coupled with high reproductive rates (% female reproducing, clutch size), are likely necessary for persistence. Similar conclusions regarding expected patterns of parameter covariation emerge from life-table analyses but in the absence of demographic and environmental stochasticity (Congdon et al. 2000). Demographic data are too incomplete for most Blanding's turtle populations to meaningfully test for such patterns or to assess the degree to which observed variation is the result of sampling error. However, the direction and relative magnitude of parameter covariation can be predicted from partial correlations among parameters in PVA iterations resulting in population persistence and stochastic-r ≥ 0 in our sensitivity analysis (e.g., that there is strong negative covariation between age at first reproduction and adult survival across populations).

Sensitivity to Population Size

For populations with demographic characteristics similar to the SBCP population, our models demonstrate that when initial adult population size is small, risk of extinction is high and loss of genetic variation is rapid. In the absence of catastrophes, extinction risk drops and retention of genetic variation increases rapidly as initial adult population size increases such that even in our high environmental stochasticity scenario, extinction risk is low (<5% over 100 y) at initial population sizes >50 adults and retention of genetic variation is high (>95% over 100 y) at initial population sizes >120 adults. Initial population size thresholds to avoid extinction and retain genetic variation are even lower in 50 y scenarios (>20 and >50 adults in our high environmental stochasticity scenario, respectively). Including catastrophes shifts these population size thresholds upward (e.g., to >200 adults in 100 y scenarios and >65 adults in 50 y scenarios if catastrophes occur at an average rate of once per generation and reduce survival to 75% base-line levels). Given the geographic variation seen in Blanding's turtle demography, these results are likely to be most relevant for populations in the southern Great Lakes region and less so for populations to the southwest and northeast where age at first reproduction is lower and higher, respectively.

Our results also demonstrate the need for clear goals when setting population size criteria. As noted above, the initial adult population size necessary to retain some threshold proportion of genetic variation may differ from that necessary to exceed an extinction risk threshold. Furthermore, small populations necessarily retain less absolute genetic variation (number of alleles) than large populations even if the proportion of variation retained is high, potentially limiting adaptive responses within small isolated populations. Consequently, small Blanding's turtle populations may require genetic management (e.g., via facilitated gene flow [Frankham et al. 2017, 2019; Jordan et al. 2019]). Likewise, the initial population size necessary to meet genetic and extinction risk thresholds differs depending on the time period over which goals are to be met (e.g., 50 vs. 100 y in our analyses). Population projections generated using PVA assume that demographic parameters remain unchanged for the duration of simulations, which is an unlikely assumption over long time periods. But population dynamics play out on a time scale of generations, which for Blanding's turtles can be ≥25 y; consequently, simulations of 50 or 100 y represent just 2–4 generations. Although our thresholds for extinction risk and retention of genetic variability (5% extinction risk or 95% retention of genetic variation over 50 or 100 y) are widely used (IUCN 2012), more stringent thresholds and longer time frames are not uncommon (e.g., 150 y [Howell and Seigel 2019]; 1% extinction risk in 40 generations [Reed et al. 2003, Trail et al. 2007]; 1% extinction risk in 100 y [Wang et al. 2019]). This problem has practical implications for listing decisions for long-lived species as reflected in recent controversy surrounding the meaning of “foreseeable future” under the U.S. Endangered Species Act (ESA 1973, as amended; USDOI 2009; Almy 2017; Lake and Petersen 2017; USFWS 2019).

Ongoing management at SBCP and the protected status of Blanding's turtles in Illinois provides some assurance that the demographic parameter estimates we generated are unlikely to change dramatically in the near term. We sought to exclude the effects of predator removal on hatching success in our PVA, but beneficial effects of management on other parameters (e.g., juvenile survival) are unknown. Even at this site, the possibility exists that disease, climate change, invasive species, increased human population density, or catastrophic events will impact demographic parameters in the future. Climate change is expected to result in an increasing frequency of extreme weather, possibly resulting in mass mortality events that may increase extinction risk (IPCC 2014). This is of particular concern for turtles because of the magnitude of anticipated impacts and their limited ability to respond demographically (Ihlow et al. 2012; Keevil et al. 2018; Mullin et al. 2020). As linkages among extreme weather, mortality, and extinction risk are resolved for Blanding's turtles and other species, more sophisticated population projection models incorporating changing future conditions will be possible (e.g., Chan et al. 2005; Cardoso et al. 2008; Frederiksen et al. 2008) but even our fairly basic catastrophe scenario suggests that increasing extreme weather events and other catastrophes will increase extinction risk and the adult population size needed to withstand such events.

Regardless of conservation goals, our results demonstrate that small Blanding's turtle populations (fewer than ca. 50 adults) are unlikely to persist in the absence of active management. In theory, this population size cut-off could provide a criterion for prioritizing populations for management or selecting among management tactics. In practice, accurate estimates of population size, even for small populations, are challenging to generate. Effective Blanding's turtle population monitoring is a multiyear labor-intensive low-yield endeavor (Congdon et al. 2008, 2011; Reid et al. 2016). At SBCP, effort averaged >1,665 trap-nights/y and capture rate averaged just 0.02 adults/trap-night from 2004 to 2018. For many Blanding's turtle populations, systematic monitoring is lacking and information on abundance is limited to element occurrences of unmarked animals in state natural heritage databases without corresponding information on effort. Consequently, our ability to apply population size criteria for conservation and management decisions is currently limited.

Accurate estimates of population size are just one of several data gaps hindering effective Blanding's turtle conservation. Age-specific survival, particularly for younger age classes, has been estimated infrequently and with wide CIs, limiting our understanding of variation among populations and precision of population projections. Environmental drivers of demographic parameters are also poorly known. Our review suggests that age of first reproduction may vary clinally, possibly driven by climate. Habitat quality (e.g., resource availability) might also influence reproductive parameters (age at first reproduction, reproductive frequency, clutch size). If so, reproductive rates might be increased by targeting habitat characteristics for management (Tracy et al. 2006). Similarly, patterns of offspring sex ratio variation are unknown, limiting our ability to meaningfully model sex ratio effects on population projections or predict responses to climate change (Janzen 1994; Valenzuela et al. 2019). Models incorporating offspring sex ratio variation might also include differential male vs. female mortality, particularly that resulting from roads and railways that are thought to inflate female mortality in Blanding's and other freshwater turtles (Gibbs and Steen 2005; Steen et al. 2006; Ruane et al. 2008; Vanek and Glowacki 2019).

As with Blanding's turtles, detailed demographic data and associated population viability analyses are accumulating for other freshwater turtles, providing insights into threat impacts and alternative management strategies (Fordham et al. 2008; Famelli et al. 2012; Folt et al. 2016; Rachmansah et al. 2020). Some turtle PVAs result in quite pessimistic prognoses, such that even small changes in demographic parameters (e.g., slightly increased adult mortality) result in a high probability of extinction (Bulté et al. 2009; Midwood et al. 2015; Howell and Seigel 2019; Piczak et al. 2019; but see Shoemaker et al. 2013). More pessimistic results may arise from the use of a longer time frame (150–500 y), use of parameters borrowed from other locations or species, or inappropriate specification of initial population size (e.g., setting initial population size equal to estimated adult population size but then applying a stable age distribution, which, in Vortex, results in a smaller number of adults than intended; Bulté et al. 2009; Midwood et al. 2015; Howell and Seigel 2019). Regardless, freshwater turtle PVAs highlight the potential importance of metapopulation structure, inbreeding depression, catastrophes, road mortality, fisheries bycatch, invasive predators, and subsistence harvest (Fordham et al. 2008; Bulté et al. 2009; Enneson and Litzgus 2009,Famelli et al. 2012; Midwood et al. 2015; Howell and Seigel 2019; Piczak et al. 2019). Population viability analyses also demonstrates opportunities for management interventions, including head-starting, predator reduction, and roadside barriers (Spencer et al. 2017; Crawford et al. 2018; Mullin et al. 2020), suggesting future directions for Blanding's turtle PVA.

Population viability and sensitivity analyses indicate that Blanding's turtle populations with demographic characteristics similar to the SBCP population possess resiliency to withstand annual environmental stochasticity but are less resilient to catastrophes (Shaffer and Stein 2000; Wolf et al. 2015). Demographic characteristics contributing to resiliency include a population size sufficient to ensure persistence over a time frame of 50–100 y (20–30 adults for 50-y persistence and 50–150 adults for 100-y persistence with catastrophes as modeled here), annual adult survival that exceeds 90%, high reproductive rates (>90% of older adult females reproduce each year), and age 0 and juvenile survival sufficient to maintain a stable or growing population. Demographic characteristics of resilient populations may vary geographically, depending for example, on age of first reproduction. Resiliency also arises from habitat characteristics. Our SBCP study site encompasses 215 ha of high-quality coastal wetland that provides active-season, nesting, and overwinter habitat and is actively managed to promote Blanding's turtle recruitment and survival. Given the density of adult Blanding's turtles at SBCP (ca. 0.5/ha), sites <40–100 ha may be too small to support resilient populations. However, Blanding's turtle population density and home-range area can be quite variable (Piepgras and Lang 2000; Lang 2004; Schuler and Thiel 2008; Edge et al. 2010; Millar 2010; Congdon et al. 2011), making area a poor proxy for quantitative estimates of population size.

Future modeling efforts might profitably focus on regional patterns of Blanding's turtle persistence by including multiple populations and catastrophes of varying spatial extent, thus providing guidance on the level of redundancy necessary to meet conservation goals (Shaffer and Stein 2000; Wolf et al. 2015). Such models might also incorporate realistic rates of natural or facilitated gene flow, as well as the effects of inbreeding depression, to better guide genetic management (Frankham et al. 2017, 2019). Simultaneously, an expanded understanding of patterns of neutral and adaptive genetic variation, potentially coupled with species distribution modeling, would aid in determining the extent of range-wide representation necessary to ensure the potential for adaptation to environmental change (Shaffer and Stein 2000; Wolf et al. 2015; Hamilton et al. 2018; Jordan et al. 2019).

Supplemental Material

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Text S1. Comments on Blanding's turtle Emydoidea blandingii population viability analysis (PVA) implementation in Vortex.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S1 (20 KB DOCX).

Text S2. References cited in supplemental material.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S2 (35 KB DOCX).

Text S3. Vortex xml file.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S3 (177 KB XML).

Table S1. Examples of mass mortality events among freshwater and terrestrial turtles, including information on the number of deaths (“Deaths”), % population decline (“Decline”), and cause.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S4 (27 KB DOCX).

Table S2. Reproductive status (0 = nongravid, 1 = gravid) of known age female Blanding's turtles Emydoidea blandingii that reached reproductive maturity during the course of our study at Spring Bluff Chiwaukee Prairie in Lake County, Illinois, and Kenosha County, Wisconsin, 2004–2018. Proportion reproducing for turtles ages 13–18 is based on the results of logistic regression (Figure S1, Supplemental Material). Proportion reproducing for turtles ages 19 and older is based on the observed proportion among adult females of unknown age (Results).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S3. Annual proportion of adult female Blanding's turtles Emydoidea blandingii reproducing at Spring Bluff Chiwaukee Prairie in Lake County, Illinois, and Kenosha County, Wisconsin, 2012–2018, as determined by palpation. Total variance is the variance across years; error variance is the mean of year-specific binomial variances (= p(1 − p)/(n − 1); environmental variance = total variance − error variance (following Akçakaya 2002).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S4. (A) Size (number of eggs laid) of 120 Blanding's turtle Emydoidea blandingii clutches produced by 40 females from Spring Bluff Chiwaukee Prairie in Lake County, Illinois, and Kenosha County, Wisconsin, 2008–2018; and (B) associated variance components computed using the restricted maximum likelihood method in SPSS (13 females for which we have data for only a single clutch were excluded).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S5. Annual proportion of Blanding's turtle Emydoidea blandingii eggs that hatched from females in Spring Bluff Chiwaukee Prairie in Lake County, Illinois, and Kenosha County, Wisconsin, 2010–2018, as part of the Lake County Forest Preserve District head-starting program. Total variance is the variance across years; error variance is the mean of year-specific binomial variances (= p(1 − p)/(n − 1); environmental variance = total variance − error variance (following Akçakaya 2002).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S6. Worksheet for the calculation of Blanding's turtle Emydoidea blandingii age 0 survival uncertainty (A) and age 0 survival environmental variance (B). Nest success and hatch success are estimated from data collected at Spring Bluff Chiwaukee Prairie in Lake County, Illinois, and Kenosha County, Wisconsin, 2004–2018; posthatch survival is from Kastle et al. (in press). Calculation of the variance of a product (nest success × hatch success × posthatch survival) follows Goodman (1960) and assumes covariances among survival components are zero. Environmental variance in posthatch survival is unknown; therefore, we estimate lower and upper bounds by setting this variance to 0.0000 and 0.0340 (to equal environmental variance in nest success), respectively.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S7. Annual variation in the proportion of Blanding's turtle Emydoidea blandingii nests surviving at a Michigan study site 1976–1998 (from table 1 in Congdon et al. 2000) and its associated environmental variance. Total variance is the variance across years; error variance is the mean of year-specific binomial variances (= p(1 − p)/(n − 1); environmental variance = total variance − error variance (following Akçakaya 2002).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S8. Estimated Spring Bluff Chiwaukee Prairie (Lake County, Illinois, and Kenosha County, Wisconsin) adult Blanding's turtle Emydoidea blandingii population size, 2004–2018. Estimates (Ni) and confidence limits are highlighted in yellow, 3-y running average estimates and confidence limits are highlighted in blue. Symbolism and table organization follows Manly (1984).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S9. Results of single-variable sensitivity tests of parameter uncertainty from the Blanding's turtle Emydoidea blandingii population viability analysis (PVA). Parameter values and parameter uncertainty were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site in Lake County, Illinois, and Kenosha County, Wisconsin, from 2004 to 2018 except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1). Shown are the ranges of Stochastic-r and Probability of Extinction across parameter values and correlations between Stochastic-r and Probability of Extinction and parameter values.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S10. Results of logistic regression analysis of sensitivity test results and associated calculations of relative influence from the Blanding's turtle Emydoidea blandingii population viability analysis (PVA). Parts A–D represent analyses of sensitivity to parameter uncertainty; parts E–F represent analyses of sensitivity to geographic parameter variation. Included in A–D are analyses based on all 30,000 Latin hypercube sampling samples and 3 random subsamples of n = 10,000 each. Dependent variables include the binomial transformation of stochastic-r (A and E), probability of extinction (B and F), binomial transformation of gene diversity (C), and binomial transformation of adult population size (D). Parameter values in sensitivity tests of parameter uncertainty were estimated from data collected at the Spring Bluff Chiwaukee Prairie study site in Lake County, Illinois, and Kenosha County, Wisconsin, from 2004 to 2018, except as noted in the text and with clutch size and morality adjusted to achieve deterministic r = 0 (Table 1). Parameter values used in sensitivity tests of range-wide parameter uncertainty are from throughout the species' range in North America (Table 1; Table S11, Supplemental Material). Shown are standardized partial logistic regression coefficients (B), their standard errors (S.E.), Wald statistic (Wald) and associated degrees of freedom (df) and significance (Sig.), and odds ratio [Exp(B)]. Relative influence was calculated by computing B/S.E. and its absolute value [Abs(B/S.E.)], dividing the latter by the sum of Abs(B/S.E.) across parameters and multiplying by 100.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S5 (85 KB XLSX).

Table S11. Geographic variation in Blanding's turtle Emydoidea blandingii demographic parameters from throughout the species' range in North America from 1986 to 2018. Parameters are listed in the order that they are used in Vortex PVA software. Locations are ordered from northeast to southwest.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S6 (24 KB DOCX).

Figure S1. Logistic regression of reproductive status (gravid = 1, not gravid = 0) on age for nine Blanding's turtle Emydoidea blandingii females achieving reproductive maturity during our study at Spring Bluff Chiwaukee Prairie in Lake County, Illinois, and Kenosha County, Wisconsin, 2004–2018 (Table S1, Supplemental Material). Vertical dashed line at 13 y represents the youngest age of first reproduction in our study. The likelihood of reproducing increases with age as 1/{1 + [e(-1 × {−7.839 + [0.551 × Age)]}]} (χ2 = 34.26, P < 0.001), yielding probabilities of 0.34, 0.47, 0.73, 0.82, and 0.88 for 13, 14, 15, 16, 17, and 18-y-old turtles.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S7 (723 KB DOCX).

Figure S2. Blanding's turtle Emydoidea blandingii population viability analysis results showing mean (standard deviation) population size over 1,000 iterations of our Spring Bluff Chiwaukee Prairie (SBCP, blue) and zero-growth (red) models. Parameter values for the SBCP model and sensitivity tests of parameter uncertainty were mostly estimated from data collected at the SBCP study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018. Clutch size and mortality rates were adjusted to achieve deterministic r = 0 for the zero-growth model.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S7 (723 KB DOCX).

Figure S3. Example of annual variation in Blanding's turtle Emydoidea blandingii age structure in the zero-growth population model that result from years in which age 0 mortality = 100% (A) and the corresponding deterministic stable age distribution (B). Parameter values were mostly estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, with clutch size and mortality rates adjusted to achieve deterministic r = 0 (Table 1). Shown in A from top to bottom is a sequence of 20 y from one Vortex iteration. Histograms represent the number of individuals surviving to the end of a given age class. Adult age classes (≥13) are pooled.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S7 (723 KB DOCX).

Figure S4. Sensitivity of probability of extinction to parameter uncertainty in the zero-growth Blanding's turtle Emydoidea blandingii population model. Parameter values and parameter uncertainty were mostly estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, with clutch size and mortality rates adjusted to achieve deterministic r = 0 (Table 1). For each relationship, other parameters were held constant at their mean values. P(Extinction) refers to the probability of extinction. EV refers to environmental variation (standard deviation).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S7 (723 KB DOCX).

Figure S5. Sensitivity of gene diversity to parameter uncertainty in the zero-growth Blanding's turtle Emydoidea blandingii population model. Parameter values and parameter uncertainty were mostly estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, with clutch size and mortality rates adjusted to achieve deterministic r = 0 (Table 1). For each relationship, other parameters were held constant at their mean values. P(Gene Diversity > median) refers to the probability that the final genetic diversity is above the median (0.961). EV refers to environmental variation (standard deviation).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S7 (723 KB DOCX).

Figure S6. Sensitivity of adult population size to parameter uncertainty in the zero-growth Blanding's turtle Emydoidea blandingii population model. Parameter values and parameter uncertainty were mostly estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, with clutch size and mortality rates adjusted to achieve deterministic r = 0 (Table 1). For each relationship, other parameters were held constant at their mean values. P(Adults > median) refers to the probability that the final adult population size is above the median (43). EV refers to environmental variation (standard deviation).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S7 (723 KB DOCX).

Figure S7. Sensitivity of probability of extinction to geographic parameter variation in the zero-growth Blanding's turtle Emydoidea blandingii population model. Parameter values are from throughout the species' range in North America from 1986 to 2018 (Table 1; Table S11, Supplemental Material). For each relationship, other parameters were held constant at their mean values. (Extinction) refers to the probability of extinction. EV refers to environmental variation (standard deviation).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S7 (723 KB DOCX).

Figure S8. Relationship of probability of extinction (upper panel) and genetic diversity (lower panel) to initial adult Blanding's turtle Emydoidea blandingii population size with (filled diamonds) and without (open circles) catastrophes in scenarios lasting 50 y. Parameter values were mostly estimated from data collected at the Spring Bluff Chiwaukee Prairie study site (in Lake County, Illinois, and Kenosha County, Wisconsin) from 2004 to 2018, with clutch size and mortality rates adjusted to achieve deterministic r = 0 (Table 1). Solid lines represent a scenario with intermediate environmental stochasticity and are bracketed by dashed (with catastrophes) and dotted (without catastrophes) lines that represent scenarios with low and high environmental stochasticity. Solid horizontal solid lines correspond to 5% probability of extinction (upper panel) and 95% retention of genetic diversity (lower panel).

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S7 (723 KB DOCX).

Reference S1. Lang JW. 2004. Blanding's turtles on Valentine NWR, Nebraska: population status, estimate of population size, and road mortality. St. Paul, Minnesota: Final Report to Nebraska Department of Roads.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S8 (27.9 MB PDF).

Reference S2. McNeil JA. 2002. Distribution, movements, morphology and reproduction in a population of Blanding's turtle (Emydoidea blandingii) in an unprotected landscape in southwestern Nova Scotia. Master's thesis. Wolfville, Nova Scotia, Canada: Acadia University.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S9 (7.24 MB PDF).

Reference S3.The Blanding's Turtle Recovery Team. 2003. National recovery plan for the Blanding's turtle (Emydoidea blandingii) Nova Scotia population. Nova Scotia, Canada.

Found at DOI: https://doi.org/10.3996/JFWM-20-063.S10 (1.1 MB PDF).

Acknowledgments

Funding for this project was provided by the U.S. Fish and Wildlife Service – Great Lakes Fish and Wildlife Restoration Act (FWS Agreement Number F16AP01040), the Illinois Department of Natural Resources (State Wildlife Grant T-111-R-1), Lake County Forest Preserve District, and Northern Illinois University. Work was carried out under permits from the Illinois Department of Natural Resources (05-11s, 07-04s, 16-045). Institutional Animal Care and Use Committee approval was provided by Northern Illinois University (LA16-0015) and the University of Illinois (06129). We thank field technicians, Lake County Forest Preserve District staff, interns, and volunteers for their contributions to field surveys. We are grateful to D. Bradke, B. Crawford, R. Lacy, J. Maerz, P. Miller, J. Vanek, and three anonymous reviewers for their very constructive suggestions to improve the manuscript.

Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.

References

Akçakaya
HR.
2002
.
Estimating the variance of survival rates and fecundities
.
Animal Conservation
5
:
333
336
.
Akçakaya
HR,
Burgman
MA,
Ginzburg
LR.
1999
.
Applied population ecology: principles and computer exercises using RAMAS® EcoLab 2.0. 2nd edition
.
Setauket, New York
:
Applied Biomathematics
.
Alacs
EA,
Janzen
FJ,
Scribner
KT.
2007
.
Genetic issues in freshwater turtle and tortoise conservation
.
Chelonian Research Monographs
4
:
107
123
.
Almy
BW
2017
.
Alaska Oil & Gas Association v. Pritzker
.
Public Land and Resources Law Review 0, Article 18.
Anderson
DR.
2008
.
Model based inference in the life sciences: a primer on evidence
.
New York
:
Springer
.
Anthonysamy
WJB.
2012
.
Spatial ecology, habitat use, genetic diversity, and reproductive success: measures of connectivity of a sympatric freshwater turtle assemblage in a fragmented landscape. Doctoral dissertation
.
Urbana-Champaign
:
University of Illinois
.
Anthonysamy
WJB,
Dreslik
MJ,
Douglas
MR,
Thompson
D,
Klut
GM,
Kuhns
AR,
Mauger
D,
Kirk
D,
Glowacki
GA,
Douglas
ME,
Phillips
CA.
2017
.
Population genetic evaluations within a co-distributed taxonomic group: a multi-species approach to conservation planning
.
Animal Conservation
21
:
137
147
.
Arsenault
LM.
2011
.
Headstarting Blanding's turtles (Emydoidea blandingii) in Nova Scotia: an investigation of artificial incubation, captive-rearing, and release to natural habitats. Master's thesis
.
Wolfville, Nova Scotia, Canada
:
Acadia University
.
Ashton
KG,
Feldman
CR.
2003
.
Bergmann's rule in nonavian reptiles: turtles follow it, lizards and snakes reverse it
.
Evolution
57
:
1151
1163
.
Beaudry
F,
deMaynadier
PG,
Hunter,
ML
Jr.
2008
.
Identifying road mortality threat at multiple spatial scales for semi-aquatic turtles
.
Biological Conservation
141
:
2550
2563
.
Browne
CL,
Hecnar
SJ.
2007
.
Species loss and shifting population structure of freshwater turtles despite habitat protection
.
Biological Conservation
138
:
421
429
.
Buchanan
SW,
Kolbe
JJ,
Wegener
JE,
Atutubo
JR,
Karraker
NE.
2019
.
A comparison of the population genetic structure and diversity between a common (Chrysemys p. picta) and an endangered (Clemmys guttata) freshwater turtle
.
Diversity 11, 99.
Buhlmann
KA,
Koch
SL,
Butler
BO,
Tuberville
TD,
Palermo
VJ,
Bastarache
BA,
Cava
SA.
2015
.
Reintroduction and head-starting: tools for Blanding's turtle (Emydoidea blandingii) conservation
.
Herpetological Conservation and Biology
10
:
435
454
.
Buhlmann
KA,
Tuberville
TD.
1998
.
Use of passive integrated transponder (PIT) tags for marking small freshwater turtles
.
Chelonian Conservation and Biology
3
:
102
104
.
Bulté
G,
Carrière
MA,
Blouin-Demers
G.
2009
.
Impact of recreational power boating on two populations of northern map turtles (Graptemys geographica)
.
Aquatic Conservation: Marine and Freshwater Ecosystems
20
:
31
38
.
Burnham,
KP,
Anderson
DR.
2002
.
Model selection and multimodel inference: a practical information-theoretic approach
.
New York
:
Spring-Verlag
.
Butchart
SHM,
Walpole
M,
van Strien
A,
Scharlemann
JPW,
Almond
REA,
Baillie
JEM,
Bomhard
B,
Brown
C,
Bruno
J,
Carpenter
KE,
Carr
GM,
Chanson
J,
Chenery
AM,
Csirke
J,
Davidson
NC,
Dentener
F,
Foster
M,
Galli
A,
Galloway
JN,
Genovesi
P,
Gregory
RD,
Hockings
M,
Kapos
B,
Lamarque
J-F,
Leverington
F,
Loh
J,
McGeoch
MA,
McRae
L,
Minasyan
A,
Morcillo
MH,
Oldfield
TEE,
Pauly
D,
Quader
S,
Revenga
C,
Sauer
JR,
Skolnik
B,
Spear
D,
Stanwell-Smith
D,
Stuart
SN,
Symes
A,
Tierney
M,
Tyrell
TD,
Vié
J-C,
Watson
R.
2010
.
Global biodiversity: indicators of recent declines
.
Science
328
:
1164
1168
.
Cagle
FR.
1939
.
A system for marking turtles for future identification
.
Copeia
1939
:
170
173
.
Cardoso
PG,
Raffaelli
D,
Pardal
MA.
2008
.
The impact of extreme weather events on the seagrass Zostera noltii and related Hydrobia ulvae population
.
Marine Pollution Bulletin
56
:
483
492
.
Caswell
H.
2019
.
Sensitivity analysis: matrix methods in demography and ecology
.
Cham, Switzerland
:
Springer Open
.
Cayuela
H,
Cruickshank
SS,
Brandt
H,
Ozgul
A,
Schmidt
BR.
2019
.
Habitat-driven life history variation in an amphibian metapopulation
.
Oikos
128
:
1265
1276
.
Chan
K-S,
Myserud
A,
Øritsland
NA,
Severinsen
T,
Stenseth
NC.
2005
.
Continuous and discrete extreme climatic events affecting the dynamics of a high-arctic reindeer population
.
Oecologia
145
:
556
563
.
[COSEWIC] Committee on the Status of Endangered Wildlife in Canada.
2016
.
COSEWIC assessment and status report on the Blanding's turtle Emydoidea blandingii, Nova Scotia population and Great Lakes/St. Lawrence population, in Canada
.
Committee on the Status of Endangered Wildlife in Canada, Ottawa, Quebec, Canada.
Compton
BW.
2007
.
Status assessment for the Blanding's turtle (Emydoidea blandingii) in the Northeast
.
Congdon
JD,
Dunham
AE,
van Loben Sels
RC.
1993
.
Delayed sexual maturity and demographics of Blanding's turtles (Emydoidea blandingii): implications for conservation and management of long-lived organisms
.
Conservation Biology
7
:
826
833
.
Congdon
JD,
Dunham
AE,
van Loben Sels
RC.
1994
.
Demographics of common snapping turtles (Chelydra serpentina): implications for conservation and management of long-lived organisms
.
American Zoologist
34
:
397
408
.
Congdon
JD,
Graham
TE,
Herman
TB,
Lang
JW,
Pappas
MJ,
Brecke
BJ.
2008
.
Emydoidea blandingii (Holbrook 1838)—Blanding's turtle
.
Pages
1
12
in
Rhodin
AGJ,
Pritchard
PCH,
van Dijk
PP,
Saumure
RA,
Buhlmann
KA,
Iverson
JB,
editors.
Conservation biology of freshwater turtles and tortoises: a compilation project of the IUCN/SSC tortoise and freshwater turtle specialist group. Chelonian Research Foundation Monographs No. 5:-15
.
1–015.12.
Congdon
JD,
Kinney
OM,
Nagle
RD.
2011
.
Spatial ecology and core-area protection of Blanding's turtle (Emydoidea blandingii)
.
Canadian Journal of Zoology
89
:
1098
1106
.
Congdon
JD,
Nagle
RD
Kinney
OM,
Oseioski
M,
Avery
HW,
van Loben Sels
RC,
Tinkle
DW.
2000
.
Nesting ecology and embryo mortality: implications for hatchling success and demography of Blanding's turtles (Emydoidea blandingii)
.
Chelonian Conservation and Biology
3
:
569
579
.
Congdon
JD,
Tinkle
DW,
Breitenbach
GL,
van Loben Sels
RC.
1983
.
Nesting ecology and hatching success in the turtle Emydoidea blandingii
.
Herpetologica
39
:
417
429
.
Crawford
BA,
Maerz
JC,
Nibbelink
NB,
Buhlmann
KA,
Norton
TM.
2014
.
Estimating consequences of multiple threats and management strategies for semi-aquatic turtles
.
Journal of Applied Ecology
51
:
359
366
.
Crawford
BA,
Moore
CT,
Norton
TM,
Maerz
JC.
2018
.
Integrated analysis for population estimation, management impact evaluation, and decision-making for a declining species
.
Biological Conservation
222
:
33
43
.
Cross
PC,
Beissinger
SR.
2001
.
Using logistic regression to analyze the sensitivity of PVA models: a comparison of methods based on African wild dog models
.
Conservation Biology
15
:
1334
1346
.
Davy
CM,
Murphy
RW.
2014
.
Conservation genetics of the endangered spotted turtle (Clemmys guttata) illustrate the risks of “bottleneck tests”
.
Canadian Journal of Zoology
92
:
149
162
.
d'Entremont
ND.
2014
.
Comparative growth and movement analysis of headstarted Blanding's turtles (Emydoidea blandingii) at Kejimkujik National Park and National Historic Site of Canada. Master's thesis
.
Wolfville, Nova Scotia, Canada
:
Acadia University
.
Dillon Consulting Limited.
2013
.
City of Ottawa South March Highlands Blanding's turtle conservation needs assessment, Project No. 12-6060
.
Edge
CB,
Steinberg
BD,
Brooks
RJ,
Litzgus
JD.
2010
.
Habitat selection by Blanding's turtles (Emydoidea blandingii) in a relatively pristine landscape
.
Ecoscience
17
:
90
99
.
Engeman
RM,
Addison
D,
Griffin
JC.
2016
.
Defending against disparate marine turtle nest predators: nesting success benefits from eradicating invasive feral swine and caging nests from raccoons
.
Oryx
50
:
289
295
.
Enneson
JJ,
Litzgus
JD.
2008
.
Using long-term data and a stage-classified matrix to assess conservation strategies for an endangered turtle (Clemmys guttata)
.
Biological Conservation
141
:
1560
1568
.
Enneson
JJ,
Litzgus
JD.
2009
.
Stochastic and spatially explicit population viability analyses for an endangered freshwater turtle, Clemmys guttata
.
Canadian Journal of Zoology
87
:
1241
1254
.
Famelli
S,
Pinhiro
SCP,
Souza
FL,
Chiaravalloti
RM,
Bertoluci
J.
2012
.
Population viability analysis of a long-lived freshwater turtle, Hydromedusa maximiliani (Testudines: Chelidae)
.
Chelonian Conservation and Biology
11
:
162
169
.
Feinberg
JA,
Burke
RL.
2003
.
Nesting ecology and predation of diamondback terrapins, Malaclemys terrapin, at Gateway National Recreation Area, New York
.
Journal of Herpetology
37
:
517
526
.
Feng
CY,
Ross
JP,
Mauger
D,
Dreslik
MJ.
2019
.
A long-term demographic analysis of spotted turtles (Clemmys guttata) in Illinois using matrix models. Diversity 11, 226
.
Fey
SB,
Siepielski
AM,
Nusslé
S,
Cerantes-Yoshida
K,
Hwan
JL,
Huber
ER,
Fey
MJ,
Catenazzi
A,
Carlson
SM.
2015
.
Recent shifts in the occurrence, cause, and magnitude of animal mass mortality events
.
Proceedings of the National Academy of Sciences
112
:
1083
1088
.
Folt
B,
Jensen
JB,
Teare
A,
Rostral
D.
2016
.
Establishing reference demography for conservation: a case study of Macrochelys temminckii in Spring Creek, Georgia
.
Herpetological Monographs
30
:
21
33
.
Fordham
DA,
Georges
A,
Brook
BW.
2008
.
Indigenous harvest, exotic pig predation and local persistence of a long-lived vertebrate: managing a tropical freshwater turtle for sustainability and conservation
.
Journal of Applied Ecology
45
:
52
62
.
Frankham
R,
Ballou
JD,
Ralls
K,
Eldridge
M,
Dudash
MR,
Fenster
CB,
Lacy
RC,
Sunnocks
P.
2017
.
Genetic management of fragmented animal and plant populations
.
Oxford
:
Oxford University Press
.
Frankham
R,
Ballou
JD,
Ralls
K,
Eldridge
M,
Dudash
MR,
Fenster
CB,
Lacy
RC,
Sunnocks
P.
2019
.
A practical guide for genetic management of fragmented animal and plant Populations
.
Oxford
:
Oxford University Press
.
Frankham
R,
Bradshaw
CJA,
Brook
BW.
2014
.
Genetics in conservation management: revised recommendations for the 50/500 rules, Red List criteria, and population viability analyses
.
Biological Conservation
170
:
56
63
.
Frankham
R,
Brook
BW,
Bradshaw
CJA,
Trail
LW,
Spielman
D.
2013
.
50/500 rule and minimum viable populations: response to Jamieson and Allendorf
.
Trends in Ecology and Evolution
28
:
187
188
.
Franklin
AB,
Anderson
DR,
Burnham
KP.
2002
.
Estimation of long-term trends and variation in avian survival probabilities using random effects models
.
Journal of Applied Statistics
29
:
267
287
.
Franklin
IR,
Allendorf
FW,
Jamieson
IG.
2014
.
The 50/500 rule is still valid—reply to Frankham et al
.
Biological Conservation
176
:
284
286
.
Frazer
N,
Gibbons
J,
Greene
J.
1991
.
Growth, survivorship and longevity of painted turtles Chrysemys picta in a southwestern Michigan marsh
.
American Midland Naturalist
125
:
245
258
.
Frederiksen
M,
Daunt
F,
Harris
MP,
Wanless
S.
2008
.
The demographic impact of extreme events: stochastic weather drives survival and population dynamics in a long-lived seabird
.
Journal of Animal Ecology
77
:
1020
1029
.
Gallego-García
N,
Vargas-Ramírez
M,
Forero-Medina
G,
Caballero
S.
2018
.
Genetic evidence of fragmented populations and inbreeding in the Columbian endemic Dahl's toad-headed turtle (Mesoclemmys dahli)
.
Conservation Genetics
19
:
221
233
.
Gassbarini
DM.
2016
.
Investigation into the cause(s) of a mass mortality of a long-lived species in a Provincial Park and an evaluation of recovery strategies
.
Master's thesis
.
Sudbury, Ontario, Canada: Laurentian University Available: https://zone.biblio.laurentian.ca/handle/10219/2734 (May 2021).
Germano
DJ,
Bury
RB.
1998
.
Age determination in turtles: evidence of annual deposition of scute rings
.
Chelonian Conservation and Biology
3
:
123
132
.
Germano
DJ,
Bury
RB,
Jennings
M.
2000
.
Growth and population structure of Emydoidea blandingii from western Nebraska
.
Chelonian Conservation and Biology
3
:
618
625
.
Gibbons
JW,
Scott
DE,
Ryan
TJ,
Buhlmann
KA,
Tuberville
TD,
Metts
BS,
Greene
JL,
Mills
T,
Leiden
Y,
Poppy
S,
Winne
CT.
2000
.
The global decline of reptiles, déjà vu amphibians
.
Bioscience
50
:
653
666
.
Gibbs
JP,
Steen
DA.
2005
.
Trends in sex ratios of turtles in the United States: implications of road mortality. Conservation
.
Biology
19
:
552
556
.
Golba
CK.
2019
.
Growth and survival of wild and head-started Blanding's turtles (Emydoidea blandingii). Master's thesis
.
DeKalb
:
Northern Illinois University
.
Goodman
LA.
1960
.
On the exact variance of products
.
Journal of the American Statistical Association
55
:
708
713
.
Gould
WR,
Nichols
JD.
1998
.
Estimation of temporal variability of survival in animal populations
.
Ecology
79
:
2531
2538
.
Graham
T,
Doyle
T.
1977
.
Growth and population characteristics of Blanding's turtle, Emydoidea blandingii, in Massachusetts
.
Herpetologica
33
:
410
414
.
Graham
T,
Doyle
T.
1979
.
Dimorphism, courtship, eggs, and hatchlings of the Blanding's turtle, Emydoidea blandingii (Reptilia, Testudines, Emydidae) in Massachusetts
.
Journal of Herpetology
13
:
125
127
.
Greaves
WF,
Litzgus
JD.
2009
.
Variation in life-history characteristics among populations of North American wood turtles: a view from the north
.
Journal of Zoology
279
:
298
309
.
Gutzke
W,
Packard
G.
1987
.
The influence of temperature on eggs and hatchlings of Blanding's turtles, Emydoidea blandingii
.
Journal of Herpetology
21
:
161
163
.
Hamilton
CM,
Bateman
BL,
Gorzo
JM,
Reid
B,
Thogmartin
WE,
Peery
MZ,
Heglund
PJ,
Radeloff
VC,
Pidgeon
AM.
2018
.
Slow and steady wins the race? Future climate change and land use leaves the imperiled Blanding's turtle (Emydoidea blandingii) behind
.
Biological Conservation
222
:
75
85
.
Hawkins
E.
2016
.
Demography, movement patterns, and habitat selection of Blanding's turtles at Canadian nuclear laboratories in Chalk River, Ontario. Master's thesis
.
Ottawa
:
University of Ottawa
.
Heppell
SS.
1998
.
Application of life-history theory and population model analyses to turtle conservation
.
Copeia
1998
:
367
375
.
Herman
TB,
Power
TD,
Eaton
BR.
1995
.
Status of Blanding's turtles (Emydoidea blandingii) in Nova Scotia, Canada
.
Canadian Field-Naturalist
109
:
182
191
.
Hileman
ET,
King
RB,
Adamski
JM,
Anton
TG,
Bailey
RL,
Baker
SJ,
Bieser
ND,
Bell
TA
Jr,
Bissell
KM,
Bradke
DR,
Campa
H
III,
Casper
GS,
Cedar
K,
Cross,
MD,
DeGregorio
BA,
Dreslik
JM,
Faust
LJ,
Harvey
DS,
Hay
RW,
Jellen
BC,
Johnson
BE,
Johnson
G,
Kiel
BE,
Kingsbury
BA,
Kowalski
MJ,
Lee
YM,
Lentini
AM,
Marshall
JC,
Mauger
D,
Moore
JA,
Paloski
RA,
Phillips
CA,
Pratt
PD,
Preney
T,
Prior
KA,
Romaine
A,
Redmer
M,
Reinert
HK,
Rouse
JD,
Shoemaker
KT,
Sutton
S,
VanDeWalle
TJ,
Weatherhead
PJ,
Wynn
D,
Yagi
A.
2017
.
Climatic and geographic predictors of life history variation in Eastern massasauga (Sistrurus catenatus): a range-wide synthesis. PLoS One 12, e0172011
.
Houlahan
JE,
Findlay
CS,
Schmidt
BR,
Meyer
AH,
Kuzmin
SL.
2000
.
Quantitative evidence for global amphibian population declines
.
Nature
404
:
752
755
.
Howell
HJ,
Legere
RH
Jr,
Holland
DS,
Seigel
RA.
2019
.
Long-term turtle declines: protected is a verb, not an outcome
.
Copeia
107
:
493
501
.
Howell
HJ,
Seigel
RA.
2019
.
The effects of road mortality on small, isolated turtle populations
.
Journal of Herpetology
53
:
39
46
.
Hoy
SR,
MacNulty
DR,
Smith
DE,
Stahler
DR,
Lambin
X,
Peterson
RO,
Ruprecht
SJ,
Vutetich
JA.
2019
.
Fluctuations in age structure and their variable influence on population growth
.
Functional Ecology
34
:
203
216
.
Ihlow
F,
Damach
J,
Engler
JO,
Flecks
M,
Hartmann
T,
Nekum
S,
Rajaei
H,
Rödder
D.
2012
.
On the brink of extinction? How climate change may affect global chelonian species richness and distribution
.
Global Change Biology
18
:
1520
1530
.
[IESPB] Illinois Endangered Species Protection Board.
2020
.
Checklist of Illinois endangered and threatened animals and plants (May 2020 version)
.
[IDNR] Indiana Department of Natural Resources.
2020
.
Species of greatest conservation need (November 2020 version)
.
[IPCC] Intergovernmental Panel on Climate Change.
2014
:
climate change 2014: synthesis report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team,
Pachauri
RK
and
Meyer
LA,
editors].
Geneva, Switzerland: IPCC.
[IUCN] International Union for Conservation of Nature.
2012
.
IUCN Red List categories and criteria: version 3.1. 2nd edition
.
Gland, Switzerland
:
IUCN
.
[IDNR] Iowa Department of Natural Resources.
2009
.
Endangered, threatened, and special concern animals (October 2009 version)
.
Iverson
JB,
Balgooyen
CP,
Byrd
KK,
Lyddan
KK.
1993
.
Latitudinal variation in egg and clutch size in turtles
.
Canadian Journal of Zoology
71
:
448
2461
.
Jamieson
IG,
Allendorf
FW.
2012
.
How does the 50/500 rule apply to MVPs?
Trends in Ecology and Evolution
27
:
578
584
.
Janzen
FJ.
1994
.
Climate change and temperature-dependent sex determination in reptiles
.
Proceedings of the National Academy of Science
91
:
7487
7490
.
Jolly
GM.
1965
.
Explicit estimates from capture–recapture data with both death and immigration-stochastic model
.
Biometrica
52
:
225
247
.
Jones
PC,
King
RB,
Bailey
RL,
Bieser
ND,
Bissell
K,
Campa
H
III,
Crabill
T,
Cross
MD,
DeGregorio
BA,
Dreslik
MJ,
Durbian
FE,
Harvey
DS,
Hecht
SE,
Jellen
BC,
Johnson
G,
Kingsbury
BA,
Kowalski
MJ,
Lee
J,
Manning
JV,
Moore
JA,
Oakes
J,
Phillips
CA,
Refsnider
JM,
Rouse
JD,
Sage
JR,
Seigel
RA,
Shepard
DB,
Smith
CS,
VanDeWalle,
TJ,
Weatherhead
PJ,
Yagi
A.
2012
.
Range-wide analysis of eastern massasauga survivorship
.
Journal of Wildlife Management
76
:
1576
1586
.
Jordan
MA,
Mumaw
V,
Millspaw
N,
Mockford
SW,
Janzen
FJ.
2019
.
Range-wide phylogeography of Blanding's turtle [Emys (= Emydoidea) blandingii]
.
Conservation Genetics
20
:
419
430
.
Joyal
LA,
McCollough
M,
Hunter
ML.
2000
.
Population structure and reproductive ecology of Blanding's turtle (Emydoidea blandingii) in Maine, near the northeastern edge of its range
.
Chelonian Conservation and Biology
3
:
580
588
.
Kastle
M,
Kapfer
J,
Kuhns
AR,
Graser
W,
Glowacki
G,
Ibach
A,
Mitchem,
L,
Mozuch
J,
Rudolph
N,
Rutzen
K,
King
R.
In press
.
Blanding's turtle hatchling survival and movements following natural vs. artificial hibernation
.
Journal of Herpetology
.
Keevil
MG,
Brooks
RJ,
Litzgus
JD.
2018
.
Post-catastrophe patterns of abundance and survival reveal no evidence of population recovery in a long-lived animal
.
Ecosphere 9:e02396. Available: https://doi.org/10.1002/ecs2.2396
King
RB,
Stanford
KM,
Jones.
PC
2018
.
Sunning themselves in heaps, knots, and snarls: the extraordinary abundance and demography of island watersnakes
.
Ecology and Evolution
8
:
7500
7521
.
Kiviat
E,
Stevens
G,
Munger
KL,
Heady
LT,
Hoeger
S,
Petokas
PJ,
Brauman
R.
2004
.
Blanding's turtle response to wetland and upland habitat construction
.
Pages
93
99
in
Swarth
CW,
Roosenburg
WM,
Kiviat
E,
editors.
Conservation and ecology of turtles of the Mid-Atlantic region: a symposium
.
Klein
CJ,
Beher
J,
Chaloupka
M,
Hamann
M,
Limpus
C,
Possingham
HP.
2017
.
Prioritization of marine turtle management projects: a protocol that accounts for threats to different life history stages
.
Conservation Letters
10
:
547
554
.
Krebs
CJ.
1998
.
Ecological methodology
.
San Francisco
:
Benjamin Cummings
.
Lacy
RC.
2019
.
Lessons from 30 years of population viability analysis of wildlife populations
.
Zoo Biology
38
:
67
77
.
Lacy
RC,
Miller
PS,
Traylor-Holzer
K.
2018
.
Vortex 10 user's manual, 1 June 2018 update. Apple Valley, Minnesota: International Union for Conservation of Nature Species Survival Commission Conservation Breeding Specialist Group and Chicago Zoological Society
Lacy
RC,
Pollak
JP.
2018
.
Vortex: a stochastic simulation of the extinction process, version 10.3.1. Brookfield, Illinois: Chicago Zoological Society
Available: https://scti.tools/vortex/ (May 2021).
Lake
EA,
Petersen
R.
2017
.
Projecting the future: ninth circuit upholds ESA listing for bearded seals
.
Environmental Law Reporter
47
:
10217
10219
.
Lang
JW.
2003
.
Blanding's turtle studies in southwestern Minnesota in 2002. New Ulm, Minnesota: Minnesota Department of Natural Resources: Final Report to Minnesota DNR Nongame, Region 4.
Lang
JW.
2004
.
Blanding's turtles on Valentine NWR, Nebraska: population status, estimate of population size, and road mortality
.
St. Paul, Minnesota Final Report to Nebraska Department of Roads
(see Supplemental Material, Reference S1).
Litzgus
JD,
Mousseau
TA.
2006
.
Geographic variation in reproduction in a freshwater turtle (Clemmys guttata)
.
Herpetologica
62
:
132
140
.
Mace
GM,
Kershaw
M.
1997
.
Extinction risk and rarity on an ecological timescale
.
Pages
130
149
in
Kunin
WE,
Gaston
KJ,
editors.
The biology of rarity: causes and consequences of rare-common differences
.
Dordrecht, Netherlands
:
Springer
.
Manlik
O,
Lacy
RC,
Sherwin
WB.
2017
.
Applicability and limitations of sensitivity analyses for wildlife management
.
Journal of Applied Ecology
55
:
1430
1440
.
Manly
BFJ.
1984
.
Obtaining confidence limits on parameters of the Jolly–Seber model for capture–recapture data
.
Biometrics
40
:
749
758
.
McCarthy
MA,
Burgman
MA,
Ferson
S.
1995
.
Sensitivity analysis for models of population viability
.
Biological Conservation
73
:
93
100
.
McGuire
JM,
Congdon
JD,
Kinney
OM,
Osentoski
M,
Scribner
KT.
2015
.
Influences on male reproductive success in long-lived Blanding's turtles (Emydoidea blandingii)
.
Canadian Journal of Zoology
93
:
487
497
.
McNeil
JA.
2002
.
Distribution, movements, morphology and reproduction in a population of Blanding's turtle (Emydoidea blandingii) in an unprotected landscape in southwestern Nova Scotia. Master's thesis
.
Wolfville, Nova Scotia, Canada
:
Acadia University (see Supplemental Material, Reference S2)
.
[MDIFW] Maine Department of Inland Fisheries & Wildlife.
2015
.
Maine's endangered and threatened wildlife (October 2015 version)
.
[MNHESP] Massachusetts Natural Heritage and Endangered Species Program.
2020
.
List of endangered, threatened, and special concern species (January 2020 version)
.
[MSU] Michigan State University.
2021
.
Michigan natural features inventory
.
[MWPARC] Midwest Partners for Amphiban and Reptile Conservation.
2010
.
Blanding's turtle (Emydoidea blandingii) conservation assessment survey
.
Available: http://www.mwparc.org/ (May 2021).
Midwood
J D,
Cairns
NA,
Stoot
LJ,
Cooke
SJ,
Blouin-Demers,
G.
2015
.
Bycatch mortality can cause extirpation in four freshwater turtle species
.
Aquatic Conservation: Marine and Freshwater Ecosystems
25
:
71
80
.
Millar
C.
2010
.
The spatial ecology of Blanding's turtles (Emydoidea blandingii): from local movement patterns, home ranges and microhabitat selection to Ontario-wide habitat suitability modelling. Master's thesis
.
Ottawa
:
University of Ottawa
.
Miller
JD,
Dinkelacker
SA.
2007
.
Reproductive structures and strategies of turtles
.
Pages
225
278
in
Wyneken
J,
MH,
Godfrey
Bels
V,
editors.
Biology of turtles: from structures to strategies of life
.
Boca Raton, Florida
:
CRC Press
.
Milligan
MC,
Wells
SL,
McNew
LB.
2018
.
A population viability analysis for sharp-tailed grouse to inform reintroductions
.
Journal of Fish and Wildlife Management
9
:
565
581
.
Mills
LS,
Doak
DF,
Wisdom
MJ.
1999
.
Reliability of conservation actions based on elasticity analysis of matrix models
.
Conservation Biology
13
:
815
829
.
[MNDNR] Minnesota Department of Natural Resources.
2013
.
Minnesota's list of endangered, threatened, and special concern species (August 2013 version)
.
[MDC] Missouri Department of Conservation.
2020
.
Missouri species and communities of conservation concern checklist (January 2020 version)
.
Moll
EO.
1973
.
Latitudinal and intersubspecific variation in reproduction of the painted turtle, Chrysemys picta
.
Herpetologica
29
:
307
318
.
Morris
WF,
Doak
DF.
2002
.
Quantitative conservation biology: theory and practice of population viability analysis
.
Sunderland, Massachusetts
:
Sinauer Associates
.
Morrison
C,
Wardle
C,
Castley
JG.
2016
.
Repeatability and reproducibility of population viability analysis (PVA) and the implications for threatened species management
.
Frontiers in Ecology and Evolution
4
:
98
.
Mullin
DI,
White
RC,
Lentini
AM,
Brooks
RJ,
Bériault
KR,
Litzgus
JD.
2020
.
Predation and disease limit population recovery following 15 years of headstarting an endangered freshwater turtle
.
Biological Conservation
245
.
Munscher
EC,
Kuhns
EH,
Cox
CA,
Butler
JA.
2012
.
Decreased mortality for the Carolina diamondback terrapin (Malaclemys terrapin centrata) following removal of raccoons (Procyon lotor) from a nesting beach in northeastern Florida
.
Herpetological Conservation and Biology
7
:
176
184
.
[NHI] Natural Heritage Inventory.
2021
.
Wisconsin natural heritage working list (January 2021 version)
.
[NGPC] Nebraska Game and Parks Commission.
2018
.
Revision of the tier 1 and 2 lists of species of greatest conservation need (2018 version)
.
[NHFG] New Hampshire Fish and Game.
2017
.
Endangered and threatened wildlife of New Hampshire (March 2017 version)
.
[NYSDEC] New York State Department of Environmental Conservation.
2019
.
Checklist of amphibians, reptiles, birds, and mammals of New York State. Effective February 2019
.
O'Grady
JJ,
Brook
BW,
Reed
DH,
Ballou
JD,
Tonkyn
DW,
Frankham
R.
2006
.
Realistic levels of inbreeding depression strongly affect extinction risk in wild populations
.
Biological Conservation
133
:
42
51
.
[ODNR] Ohio Department of Natural Resources.
2020
.
Ohio's listed species (July 2020 version)
.
Ovaskainen
O,
Meerson
B.
2010
.
Stochastic models of population extinction
.
Trends in Ecology and Evolution
25
:
643
652
.
Pappas
MJ,
Brecke
BJ,
Congdon
JD.
2000
.
The Blanding's turtles (Emydoidea blandingii) of Weaver Dunes, Minnesota
.
Chelonian Conservation and Biology
3
:
557
568
.
[PWAP] Pennsylvania Wildlife Action Plan.
2015
.
Species of greatest conservation need species accounts (September 2015 version)
.
Petokas
PJ.
1986
.
Patterns of reproduction and growth in the freshwater turtle Emydoidea blandingii. Dissertation
.
Binghamton, New York
:
University Center at Binghamton (see Supplemental Material, Reference S4)
.
Piczak
ML,
Markle
CE,
Chow-Fraser
P.
2019
.
Decades of road mortality cause severe decline in a common snapping turtle (Chelydra serpentina) population from an urbanized wetland
.
Chelonian Conservation and Biology
18
:
231
240
.
Piepgras
SA,
Lang
JW.
2000
.
Spatial ecology of Blanding's turtle in central Minnesota
.
Chelonian Conservation and Biology
3
:
589
601
.
Prowse
TAA,
Bradshaw
CJA,
Delean
S,
Cassey
P,
Lacy
RC,
Wells
K,
Aiello-Lammens
ME,
Akçakaya
HR,
Brook
BW.
2016
.
An efficient protocol for the global sensitivity analysis of stochastic ecological models. Ecosphere 7:e01238
.
Rachmansah
A,
Norris
D,
Gibbs
JP
2020
.
Population dynamics and biological feasibility of sustainable harvesting as a conservation strategy for tropical and temperate freshwater turtles. PLoS ONE 15:e0229689
.
Reed
DH,
O'Grady
J,
Brook
BW,
Ballou
JD,
Frankham
R.
2003
.
Estimates of minimum viable population size for vertebrates and factors influencing those estimates
.
Biological Conservation
113
:
23
34
.
Reed
JM,
Fefferman
N,
Averill-Murray
RC.
2009
.
Vital rate sensitivity analysis as a tool for assessing management actions for the desert tortoise
.
Biological Conservation
142
:
2710
2717
.
Reed
JM,
Murphy
DD,
Brussard
PF.
1998
.
Efficacy of population viability analysis
.
Wildlife Society Bulletin
26
:
244
251
.
Refsnider
JM.
2009
.
High frequency of multiple paternity in Blanding's turtle (Emys blandingii)
.
Journal of Herpetology
43
:
74
81
.
Reid
BN,
Thiel
RP,
Perry
MZ.
2016
.
Population dynamics of endangered Blanding's turtles in a restored area
.
Journal of Wildlife Management
80
:
553
562
.
Reinke
BA,
Miller
DAW,
Janzen
FJ.
2019
.
What have long-term field studies taught us about population dynamics?
Annual Review of Ecology, Evolution and Systematics
50
:
11.1
11.18
.
Ritchie
S.
2017
.
Overwintering ecology of head-started Blanding's turtles (Emydoidea blandingii) in an artificial wetland complex. Master's thesis. Toronto: University of Toronto
Ruane
S,
Dinkelacker
SA,
Iverson
JB.
2008
.
Demographic and reproductive traits of Blanding's turtles, Emydoidea blandingii, at the western edge of the species' range
.
Copeia
2008
:
771
779
.
Rubin
CS,
Warner
RE,
Ludwig
DR,
Thiel
R.
2004
.
Population structure of Blanding's turtles (Emydoidea blandingii) in two suburban Chicago forest preserves
.
Natural Areas Journal
24
:
44
48
.
Schipper
J,
Chanson
JS,
Chiozza
F,
Cox
NA,
Hoffmann
M,
Katariya
V,
Lamoreux
J,
Rodrigues
ASL,
Stuart
SN,
Temple
HJ,
Baillie
J,
Boitani
L,
Lacher
TE
Jr,
Mittermeier
RA,
Smith
AT,
Absolon
D,
Aguiar
JM,
Amori
G,
Bakkour
N,
Baldi
R,
Berridge
RJ,
Bielby
J,
Black
PA,
Blanc
JJ,
Brooks
TM,
Burton
JA,
Butynski
TM,
Catullo
G,
Chapman
R,
Cokeliss
Z,
Collen
B,
Conroy
J,
Cooke
JG,
da Fonseca
GAB,
Derocher
AE,
Dublin
HT,
Duckworth
JW,
Emmons
L,
Emslie
RH,
Festa-Bianchet
M,
Foster
M,
Foster
S,
Garshelis
DL,
Gates
C,
Gimenez-Dixon
M,
Gonzalez
S,
Gonzalez-Maya
JF,
Good
TC,
Hammerson
G,
Hammond
PS,
Happold
D,
Happold
M,
Hare
J,
Harris
RB,
Hawkins
CE,
Haywood
M,
Heaney
LR,
Hedges
S,
Helgen
KM,
Hilton-Taylor
C,
Hussain
SA,
Ishii
N,
Jefferson
TA,
Jenkins
RKB,
Johnston
CH,
Keith
M,
Kingdon
J,
Knox
DH,
Kovacs
KM,
Langhammer
P,
Leus
K,
Lewison
R,
Lichtenstein
G,
Lowry
LF,
Macavoy
Z,
Mace
GM,
Mallon
DP,
Masi
M,
McKnight
MW,
Medellín
RA,
Medici
P,
Mills
G,
Moehlman
PD,
Molur
S,
Mora
A,
Nowell
K,
Oates
JF,
Olech
W,
Oliver
WRL,
Oprea
M,
Patterson
BD,
Perrin
WF,
Polidoro
BA,
Pollock
C,
Powel
A,
Protas
Y,
Racey
P,
Ragle
J,
Ramani
P,
Rathbun
G,
Reeves
RR,
Reilly
SB,
Reynolds
JE,
Rondinini
C,
Rosell-Ambal
RG,
Rulli
M,
Rylands
AB,
Savini
S,
Schank
CJ,
Sechrest
W,
Self-Sullivan
C,
Shoemaker
A,
Sillero-Zubiri
C,
De Silva
N,
Smith
DE,
Srinivasulu
C,
Stephenson
PJ,
van Strien
N,
Talukdar
BK,
Taylor
BL,
Timmins
R,
Tirira
DG,
Tognelli
MF,
Tsytsulina
K,
Veiga
LM,
Vié
J-C,
Williamson
EA,
Wyatt
SA,
Xie
Y,
Young
BE.
2008
.
The status of the world's land and marine mammals: diversity, threat, and knowledge
.
Science
322
:
225
230
.
Schuler
M,
Thiel
RP.
2008
.
Annual vs. multiple-year home range sizes of individual Blanding's turtles, Emydoidea blandingii, in central Wisconsin
.
Canadian Field-Naturalist
122
:
61
64
.
Schwanz
LE,
Spencer
R-J,
Bowden
RM,
Janzen
FJ.
2010
.
Climate and predation dominate juvenile and adult recruitment in a turtle with temperature-dependent sex determination
.
Ecology
9
:
3016
3026
.
Seber
GAF.
1965
.
A note on the multiple recapture census
.
Biometrica
52
:
249
259
.
Sethuraman
A,
McGaugh
SE,
Becker
ML,
Chandler
CH,
Christiansen
JL,
Hayden
S,
LeClere
A,
Monson-Miller
J,
Myers
EM,
Paitz
RT,
Refsnider
JM,
VanDeWalle
TJ,
Janzen
FJ.
2014
.
Population genetics of Blanding's turtle (Emys blandingii) in the Midwestern United States
.
Conservation Genetics
15
:
61
773
.
Shaffer
ML,
Stein
B.
2000
.
Safeguarding our precious heritage
.
Pages
301
322
in
BA,
Stein
Kutner
LS,
Adams
JS,
editors.
Precious heritage: the status of biodiversity in the United States
.
Oxford
:
Oxford University Press
.
Sheppard
AC.
2014
.
Emydoidea blandingii (Blanding's turtle) mass mortality
.
Herpetological Review
45
:
312
313
.
Shoemaker
KT,
Breisch
AR,
Jaycox
JW,
Gibbs
JP.
2013
.
Reexamining the minimum viable population concept for long-lived species
.
Conservation Biology
27
:
542
551
.
[SDGFP] South Dakota Game, Fish, and Parks.
2021
.
Amphibians and reptiles of South Dakota
.
Spencer
RJ,
Van Dyke
JU,
Thompson
MB.
2017
.
Critically evaluating best management practices for preventing freshwater turtle extinctions
.
Conservation Biology
31
:
1340
1349
.
Standing
KL.
1997
.
Reproduction, nest site selection, and neonatal behavior in a northern peripheral populations of Blanding's turtle. Master's thesis
.
Wolfville, Nova Scotia
:
Acadia University
.
Standing
KL,
Herman
TB,
Morrison
IP.
1999
.
Nesting ecology of Blanding's turtle (Emydoidea blandingii) in Nova Scotia, the northeastern limit of the species' range
.
Canadian Journal of Zoology
77
:
1609
1614
.
Starking-Szymanski
M,
Yoder-Nowak
T,
Rybarczyk
G,
Dawson
HA.
2018
.
Movement and habitat use of headstarted Blanding's turtles in Michigan
.
Journal of Wildlife Management
82
:
1516
1527
.
Steen
DA,
Aresco
MJ,
Beilke
SG,
Compton
BW,
Condon
EP,
Dodd
CK,
Forrester
H,
Gibbons
JW,
Greene
JL,
Johnson
G,
Langen
TA,
Oldham
MJ,
Oxier
DN,
Saumure
RA,
Schueler
FW,
Sleeman
JM,
Smith
LL,
Tucker
JK,
Gibbs
JP.
2006
.
Relative vulnerability of female turtles to road mortality
.
Animal Conservation
9
:
269
273
.
The Blanding's Turtle Recovery Team.
2003
.
National recovery plan for the Blanding's turtle (Emydoidea blandingii) Nova Scotia population. Nova Scotia, Canada
(see Supplemental Material, Reference S3).
Thompson
D,
Glowacki
G,
Ludwig
D,
Reklau
R,
Kuhns
AR,
Golba
CK,
King
R.
2020
.
Benefits of head-starting for Blanding's turtle size distributions and recruitment
.
Wildlife Society Bulletin
44
:
57
67
.
Tracy
CR,
Nussear
KE,
Esque
TC,
Dean-Bradley
K,
Tracy
CR,
DeFalco
LA,
Castle
KT,
Zimmerman
LC,
Espinoza
RE,
Barber
AM.
2006
.
The importance of physiological ecology in conservation biology
.
Integrative and Comparative Biology
46
,
1191
1205
.
Trail
LW,
Bradshaw
CJA,
Brook
BW.
2007
.
Minimum viable population size: a meta-analysis of 30 years of published estimated
.
Biological Conservation
139
:
159
166
.
Urbanek
RE,
Glowacki
GA,
Nielsen
CK.
2016
.
Effect of raccoon (Procyon lotor) reduction on Blanding's turtle (Emydoidea blandingii) nest success
.
Journal of North American Herpetology
2016
:
39
44
.
[USDOI] U.S. Department of the Interior.
2009
.
The meaning of “Foreseeable Future” in Section 3(20) of the Endangered Species Act. Memorandum M-37021
.
[ESA] U.S. Endangered Species Act of 1973, as amended, Pub. L. No. 93-205, 87 Stat. 884 (Dec. 28, 1973)
.
[USFWS] U.S. Fish and Wildlife Service.
2015
.
Endangered and threatened wildlife and plants; 90-day findings on 31 petitions
.
Federal Register
80
:
37568
37579
.
[USFWS] U.S. Fish and Wildlife Service.
2016
.
USFWS species status assessment framework: an integrated analytical framework for conservation, version 3.4
.
[USFWS] U.S. Fish and Wildlife Service.
2019
.
Endangered and threatened wildlife and plants: regulations for listing species and designating critical habitat
.
Federal Register
84
:
45020
45053
.
[USFWS] U.S. Fish and Wildlife Service.
2021
.
National domestic listing workplan: fiscal years 21–25 5-year workplan (January 2021 version)
.
Valenzuela
N,
Literman
R,
Neuwald
JL,
Mizoguchi
B,
Iverson
JB,
Riley
JL,
Litzgus
JD.
2019
.
Extreme thermal fluctuations from climate change unexpectedly accelerate demographic collapse of vertebrates with temperature-dependent sex determination
.
Scientific Reports
9
:
4254
.
Vanek
JP,
Glowacki
GA.
2019
.
Assessing the impacts of urbanization on sex ratios of painted turtles (Chrysemys picta)
.
Diversity
11
:
72
.
Velo-Antón
G,
Becker
CG,
Cordero-Rivera
A.
2011
.
Turtle carapace anomalies: the roles of genetic diversity and environment. PLoS ONE 6:e18714
.
Wang
T,
Fujiwara
M,
Gao
X,
Liu
H.
2019
.
Minimum viable population size and population growth rate of freshwater fishes and their relationships with life history traits
.
Scientific Reports
9
:
3612
.
Wilson
DS,
Tracy
CR,
Tracy
R.
2003
.
Estimating age of turtles from growth rings: a critical evaluation of the technique
.
Herpetologica
59
:
178
194
.
Wolf
S,
Hartl
B,
Carroll
C,
Neel
MC,
Greenwald
DN.
2015
.
Beyond PVA: why recovery under the Endangered Species Act is more than population viability
.
Bioscience
65
:
200
207
.

The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.

Author notes

Citation: King RB, Golba CK, Glowacki GA, Kuhns AR. 2021. Blanding's turtle demography and population viability. Journal of Fish and Wildlife Management 12(1):112–138; e1944-687X. https://doi.org/10.3996/JFWM-20-063

Supplementary data