The New Mexico meadow jumping mouse Zapus hudsonius luteus was listed as endangered under the U.S. Endangered Species Act in 2014, with critical habitat designated in 2016. Despite these recent conservation actions, there is a paucity of published information regarding its habitat associations. The taxon is a riparian obligate that occurs along both low-elevation rivers and high-elevation headwater streams in several disjunct areas of the American Southwest. Habitat information from one region might not apply to others. The distribution and habitat preferences of the New Mexico meadow jumping mouse in the White Mountains in eastern Arizona are poorly known. Objectives of this study were to 1) identify and resurvey historical locations in the White Mountains, 2) survey for new populations in areas with potentially suitable habitat in the White Mountains, and 3) use quantitative data to evaluate habitat associations at the landscape and microhabitat scales and to compare habitat at sites where I captured or did not capture the New Mexico meadow jumping mouse. I found 123 historical records of the New Mexico meadow jumping mouse from 21 locations in the White Mountains, indicating a formerly broad distribution. I conducted field surveys and collected habitat data at 35 sites (14 historical, 21 new) and caught 37 (39 total captures) New Mexico meadow jumping mice at 12 sites, including 6 of 12 historical locations surveyed. The overall capture rate was 0.36%, with an average capture rate at sites where it was present of 1.28% (range = 0.25–2.5%). All historical sites where I caught the New Mexico meadow jumping mouse were in the drainage of the Black River. The six new sites included the first records for Nutrioso Creek and Corduroy Creek and confirmed persistence of the taxon in the East Fork Little Colorado River, San Francisco River, and Blue River watersheds. Habitat used by the New Mexico meadow jumping mouse in the White Mountains was similar to that reported for other montane populations, characterized by tall, dense herbaceous vegetation composed primarily of forbs and sedges on saturated soil in close proximity to flowing water. However, there was significantly more cover provided by alders Alnus spp. at capture sites at both the stream reach and microhabitat scales. All sites where I captured the New Mexico meadow jumping mouse had no authorized livestock grazing, and the taxon was more likely to occur at sites where there were no signs of unauthorized livestock grazing. Further, there was a significant positive relationship between alder cover and time since an area was excluded from livestock grazing. The widespread exclusion of livestock from riparian areas in the White Mountains may have contributed to the higher rate of population persistence of the New Mexico meadow jumping mice in the White Mountains compared with the Jemez and Sacramento mountains, New Mexico. Although the overall persistence rate in the White Mountains (47%) was higher than other populations, the population is at risk of further losses due to small, isolated occupied areas and ongoing threats.
The New Mexico meadow jumping mouse Zapus hudsonius luteus (some earlier papers referred to it as Z. luteus or Z. princeps luteus; Figure 1) is a well-differentiated monophyletic taxon that is endemic to the American Southwest (Miller 1911; Hafner et al. 1981; King et al. 2006; Malaney et al. 2012; Malaney and Cook 2013). It is a riparian obligate that is known to inhabit six regions, including tributaries to the San Juan River in southwestern Colorado; the upper Canadian River watershed in southeastern Colorado and northeastern New Mexico; the Rio Grande watershed in New Mexico (including the Jemez Mountains and portions of the Sangre de Cristo Mountains); watersheds of the Sacramento Mountains in southern New Mexico; the Verde River watershed in central Arizona; and watersheds of the White Mountains in eastern Arizona (Morrison 1992; Frey 2012; Malaney et al. 2012; USFWS 2014b). The New Mexico meadow jumping mouse is considered a habitat specialist associated with saturated soils along perennial flowing water that supports tall, dense herbaceous vegetation (Morrison 1990; Frey and Malaney 2009; USFWS 2014a, b). As a result of population declines due to habitat loss, the New Mexico meadow jumping mouse was listed as endangered under the U.S. Endangered Species Act (ESA 1973, as amended; USFWS 2014a) with critical habitat designated in 2016 (USFWS 2016).
Despite the recent conservation attention on this taxon due to the U.S. Endangered Species Act listing and critical habitat designation, published information about habitat relations of the New Mexico meadow jumping mouse is sparse and mostly based on anecdotal observations (e.g., Bailey 1931; Hall and Davis 1934; Harris 1963; Findley et al. 1975; Hoffmeister 1986; Hafner et al. 1981; Jones 1999; Morrison 1992). Only three published papers evaluated habitat of the New Mexico meadow jumping mouse. Morrison (1990) described habitat at sites where she caught the taxon in New Mexico (data were combined from montane and low-elevation sites) and compared habitat at capture versus noncapture sites in the Sacramento Mountains, although details about data collection and survey locations were not provided. Elsewhere (Morrison 1989:5), however, she explained that data were collected on 10 plots “along the waterway” and on 10 plots “on the slope just above the zone of riparian habitat, usually about 50 feet from the waterway.” Thus, her results described the general conditions of survey sites, but not necessarily specific habitats selected by New Mexico meadow jumping mice. Occupied sites were in riparian zones (including irrigation canals) or wet meadows, and were composed of diverse plant communities dominated by grasses and forbs. Based on qualitative observations, occupied sites had moist soil and ground covered by dense (≥0.5 m high) vegetation. Traps that captured New Mexico meadow jumping mice were near water. In the Sacramento Mountains, New Mexico meadow jumping mice were not captured at sites with greater cover of sedges and rushes in comparison with grasses and forbs. Although Morrison (1990) captured New Mexico meadow jumping mice at some sites where livestock grazing was occurring, she concluded that livestock grazing was the greatest threat to the taxon's habitat.
Frey and Malaney (2009) surveyed for the New Mexico meadow jumping mouse at historical locations and new sites with potentially suitable habitat in the Jemez and Sacramento mountains, New Mexico. They compared quantitative microhabitat features on 4-m-radius plots at sites where the taxon was captured versus sites where it was not captured. They found that capture locations were typified by saturated soil dominated by sedges and forbs with significantly higher vertical cover and stubble height in comparison with noncapture locations. Frey and Malaney (2009) concluded that livestock grazing was the primary cause of habitat and population loss. Results of that study have been considered representative of patterns of habitat use by the taxon in all montane areas. However, such inferences may not be appropriate given the small sample size in that study (14 capture locations at 7 sites) and variation in physical aspects, biogeographic histories, and land management approaches of different mountain ranges.
More recently, Wright and Frey (2015) analyzed habitat selection (i.e., use vs. availability) at three spatial scales by radiocollared New Mexico meadow jumping mice at Bosque del Apache National Wildlife Refuge along the Rio Grande in New Mexico. New Mexico meadow jumping mice used different habitats for different aspects of their life history: they selected canal banks and certain early seral herbaceous riparian vegetation types at the landscape and macrohabitat scale; and active jumping mice (i.e., foraging and traveling) selected microhabitat that contained certain wetland plants and was near water, had high soil moisture, high herbaceous cover, and lacked shrubs and trees. However, the study area is somewhat unique in that it is the lowest elevation site for the New Mexico meadow jumping mouse and it is on an intensively managed irrigated floodplain of a high-order river (Wright and Frey 2015). In contrast, most other persisting populations are in higher elevation montane areas where they occur on smaller order headwater streams with unmanaged flows (USFWS 2014a, b; 2016).
Accurate information about habitat associations is essential to the conservation of a species, especially those regarded as habitat specialists or affected by habitat loss. Without this information it is difficult to target surveys to locate populations, interpret survey results, develop and implement management strategies for expanding or improving habitat for the species, evaluate the impacts of disturbances, develop monitoring strategies for determining trends in habitat, or construct hypotheses and implement rigorous experimental study designs to test important aspects of the species' biology. Published information on habitat relations for montane populations of the New Mexico meadow jumping mouse are scant overall, and nonexistent for the White Mountains, Arizona, which is the region that contains the greatest number of persisting populations (USFWS 2014a, b; 2016). Habitat relations for the New Mexico meadow jumping mouse could be different in the White Mountains because this region has a complex biogeographic history as a result of its high elevations, isolation, geology, and location at the intersection of several floristic areas (McLaughlin 1986; Long et al. 2006). In addition, information about the distribution of the New Mexico meadow jumping mouse in the White Mountains is fragmentary and poorly documented. The vast majority of existing information is contained in unpublished agency reports and historical museum specimens; the most recent published synopsis of the distribution of the New Mexico meadow jumping mouse in the White Mountains included just eight general locations (Hoffmeister 1986). Consequently, the purpose of this study was to better understand the distribution of the New Mexico meadow jumping mouse in the White Mountains and to evaluate habitat use of this montane population based on quantitative data collected in conjunction with that survey. Specific objectives were to 1) identify all historical locations for the New Mexico meadow jumping mouse in the White Mountains to better understand its historical distribution, 2) quantify the distribution and status of the New Mexico meadow jumping mouse at historical and new locations, and 3) evaluate habitat at two different scales (stream reach and microhabitat) at capture and noncapture sites to describe habitats used by the New Mexico meadow jumping mouse in the White Mountains.
The White Mountains are located on the southern edge of the Colorado Plateau in Apache, Navaho, and Greenlee counties in east-central Arizona. The headwaters of two major watersheds of the Colorado River converge in the White Mountains: the Little Colorado River on the north side of the mountains, and the Gila River on the east (San Francisco River), south (Blue River), and west (White River) sides of the mountains. Elevations range from 1,008 to 3,476 m, resulting in a complex admixture of vegetation types (summarized by Frey 2010). In brief, lower elevations consist of desertscrub and grassland biotic communities, midelevations consist of conifer woodlands, and high elevations consist of several montane and subalpine grassland and conifer forest biotic communities (Frey 2010). Thus, riparian zones traverse a diversity of biotic communities.
I compiled historical records of the New Mexico meadow jumping mouse in the White Mountains from the published literature, unpublished agency reports, records maintained by the Arizona Department of Game and Fish, Arizona Heritage Data Management System, and museum collections. I searched for specimens in 39 museum collections represented by the Mammal Networked Information System (http://manisnet.org/), as well as conducted separate searches of the following collections: University of Arizona, Arizona State University, Museum of Northern Arizona, U.S. National Museum, Denver Museum of Nature and Science, Texas Tech University Collection, New Mexico State University collections; New Mexico Museum of Natural History and Science, Gila Center for Natural History, Academy of Natural Science of Philadelphia, Museum of Southwestern Biology, University of Illinois Museum of Natural History, University of California Museum of Vertebrate Zoology, and San Diego Natural History Museum. In many instances records were reported in multiple sources, sometimes presenting slightly different data. Where possible, I cross-referenced and rectified variation in data with the original source (e.g., specimen tag) and I examined field notes or other records to more precisely determine capture locations.
Identification of nightly foraging habitat
The selection of new survey sites, transect locations, and microhabitat plot locations required identification of the specialized riparian habitat used by the New Mexico meadow jumping mouse for its nightly activities. The New Mexico meadow jumping mouse uses three habitat types for key aspects of its life history: herbaceous riparian wetland vegetation for its nightly activity, which primarily consists of foraging (but may also include other activities such as breeding and exploration; hereafter, termed “foraging habitat”); open grassy areas for above-ground nests where it sleeps during the day; and underground burrows in drier areas with more woody cover for maternal activities and hibernation (Wright and Frey 2015). Availability of foraging habitat is considered a primary limiting factor that is necessary for occurrence of the New Mexico meadow jumping mouse (Frey and Malaney 2009; Wright and Frey 2015). In addition, nightly foraging habitat is where New Mexico meadow jumping mice are most likely to be captured because this is the habitat where they are awake, moving above ground, and searching for food; hence, surveys for the New Mexico meadow jumping mouse target this habitat (USFWS 2015). Prior research on the New Mexico meadow jumping mouse and Preble's meadow jumping mouse Z. h. preblei demonstrated that they exhibit high site fidelity and use small patches of specific habitat for nightly foraging (Trainor et al. 2007; Wright and Frey 2015). Based on information from prior studies, characteristics I used to identify foraging habitat for the New Mexico meadow jumping mouse included near water; saturated soil; vegetation composed of tall, dense herbaceous vegetation; and lacking trees, rocks, coarse woody debris, or bare ground (Frey and Malaney 2009; Wright and Frey 2015).
Surveys for the New Mexico meadow jumping mouse
My primary objective was to identify all known historical locations for the New Mexico meadow jumping mouse in the White Mountains and to determine the status of the taxon and habitat characteristics at those sites. My second objective was to determine status of the New Mexico meadow jumping mouse and habitat at additional sites with potentially occupied habitat to better define the geographic range and habitat associations of the taxon. Consequently, I selected survey sites based on presence of a historical record for the New Mexico meadow jumping mouse regardless of current habitat conditions (i.e., historical site) and presence of foraging habitat for the New Mexico meadow jumping mouse (i.e., new site). Field surveys occurred on lands managed by the Arizona Game and Fish Department and Apache–Sitgreaves National Forest (Table 1). I did not survey three historical locations on the Fort Apache Indian Reservation because of lack of access. I could not determine locations of three historical records (sites 9, 15, 20) and did not survey them. I could not safely trap at two historical locations on the West Fork of the Little Colorado River (sites 3 and 4) because of intensive human recreational use, but I collected habitat data at these two sites and included them in statistical models. Persistence of the New Mexico meadow jumping mouse at one historical location (site 12) was confirmed in 2007 (J. Underwood, Arizona Department of Game and Fish, personal communication), and hence, I did not sample it during this study.
Survey methods were consistent with the U.S. Fish and Wildlife Service guidelines for surveys of the New Mexico meadow jumping mouse (USFWS 2015). I captured animals in accordance with a scientific collecting permit issued by the state of Arizona (SP627666, SP747593) and all capture and handling techniques were approved by the New Mexico State University Institutional Animal Care and Use committee. Because the New Mexico meadow jumping mouse hibernates, surveys occurred during 18 July to 13 September 2008 and 25 June to 6 September 2009, which insured that animals were active above ground (Frey 2015).
I captured animals using Sherman live traps (model LFATDG; H.B. Sherman, Tallahassee, FL) baited with commercial four-way horse sweet feed (i.e., a mixture of grains and molasses). At each site, I established trapping transects generally parallel to the stream in areas containing potential foraging habitat. I spaced traps closely (approx. 3–5 m apart) to saturate available habitat and insure trap placement in small patches of foraging habitat. Trap-night is a measure of survey effort wherein 1 trap-night is equivalent to one trap set for one night (Wilson et al. 1996). I attempted to accumulate an effort of 400 trap-nights over three consecutive nights at each site. This effort exceeded the mean number of trap-nights and encompassed the maximum range of variation required to capture the New Mexico meadow jumping mouse during my other surveys for this species (J.K. Frey, unpublished data). If I captured a New Mexico meadow jumping mouse prior to achieving 400 trap-nights, I usually ceased trapping except at four sites with complex habitat configurations to more thoroughly sample available habitats. I set traps in the evening and checked them at sunrise. I periodically cleaned and disinfected traps with Lysol or 10% bleach as a precaution against hantavirus, to control aquatic pathogens, and to remove odors of other species. For each New Mexico meadow jumping mouse captured, I took photographs of the animal and the capture location, recorded location of the trap with global positioning system, determined sex and reproductive condition, took standard measurements (hindfoot length, ear length, mass), and collected a tissue sample from the external ear pinna. I released animals at the capture location as quickly as possible. I rehabilitated torpid animals prior to release. If more than one New Mexico meadow jumping mouse was captured at a location, I collected a voucher specimen consisting of one male or one female that was not pregnant or lactating and deposited it in the Wildlife Museum, New Mexico State University. In compliance with a confidentiality agreement, I do not present detailed location data, but data are archived by the Arizona Game and Fish Department. I denoted historical sites by numerals and new survey sites by letter.
I calculated capture rate as the number of New Mexico meadow jumping mice captured per 100 trap-nights. Capture rate is interpreted as an index of relative abundance and can be used to make general comparisons across studies. However, capture rate must be interpreted cautiously because it does not correct for imperfect detection and might be influenced by other factors such as captures of other species.
Habitat data collection
The study design for the habitat analyses was mensurative sensu Morrison et al. (2008) and involved comparison of habitat attributes at the site level (i.e., I compared habitat at sites where New Mexico meadow jumping mice were captured vs. sites where they were not captured). Further, I collected habitat data at two scales: landscape (i.e., stream reach) and microhabitat. With respect to the four scales of habitat selection defined by Johnson (1980), my landscape scale corresponded to Johnson's first-order selection, which is defined as selection of the organism's physical or geographical range, while my microhabitat scale corresponded to Johnson's third-order selection, which is defined as selection of habitat components within an individual's home range. Thus, this study design was aimed at describing the landscape-scale habitat features that determine the local distribution of the New Mexico meadow jumping mouse and describing any differences in available foraging habitat (microhabitat scale) at sites where I captured or did not capture the New Mexico meadow jumping mouse. A mensurative study design does not control for confounding variables, manipulate the system, or involve replication as might occur in an experimental study design; therefore, the scope of inference for the habitat analyses is limited to the study sites (Morrison et al. 2008). However, given that the results are based on all known populations of the New Mexico meadow jumping mouse that persist in the White Mountains, they provide an important baseline for understanding the species' natural history, which can be used to evaluate potential threats, develop conservation and management plans, and inform future experimental studies.
Variables that I recorded that were relevant to both the landscape and microhabitat scale analyses were location, elevation, and presence or sign of livestock. To prevent bias in survey site selection, I did not obtain information on livestock grazing management until field work was completed (Table S1).
Stream reach cover
Previous studies identified vertical cover provided by herbaceous riparian vegetation as an important component of habitat used by the New Mexico meadow jumping mouse (Frey and Malaney 2009; Wright and Frey 2015). Thus, the stream reach cover method provided a landscape scale evaluation of components of vegetation cover along the stream reach at survey sites. At each survey site I established paired transects paralleling the stream course and located 0.5 m (i.e., “water-edge transect”) and 4.5 m (i.e., “inland transect”) from the edge of the green-line (i.e., vegetation closest to the water). Foraging habitat often occurs as narrow stringers along water edge; hence, the placement of the water-edge transect 0.5 m from the edge of the water helped to insure that patches of foraging habitat were recorded. I based the location of the inland transect on the standard 4-m reading distance of a Robel pole (Robel et al. 1970). However, this spacing also provided information on relative extent of the riparian vegetation at a site because on some small-order streams riparian vegetation transitions to upland vegetation within a few meters. I located the paired transects on the same side of the stream that trapping occurred, or I randomly determined the side if trapping occurred on both sides of the stream. I established sample stations each 20 m along transects, except at two sites (sites B and I) where the survey area was small and the interval was reduced to 10 m to increase sample size. The goal for transect length was 1 km, but site characteristics or lack of access reduced the length at some sites. At each station, I (or a field assistant) read the Robel pole (Robel et al. 1970) from the opposing transect (i.e., 4 m away) at 1-m eye level and I recorded the lowest 1-inch (25.4-mm) segment that was not obstructed by cover. In addition, I recorded the dominant plant or other structure that provided the cover: conifer, rush Juncaceae, sedge Carex spp., grass Poaceae, forb, willow Salix spp., alder Alnus spp., shrub cinquefoil Dasiphora fruticosa, Wood's rose Rosa woodsii, other shrub, other plant, dead standing plant, coarse woody debris, rock, and bank. I did not use this method at sites with complex wetlands spread across broad valleys because there was no defined stream edge.
I collected microhabitat data in the same manner as during other surveys of the New Mexico meadow jumping mouse, which allowed for comparison across different regions (e.g., Frey and Malaney 2009; Wright and Frey 2015). I collected microhabitat data on 4-m-radius plots that measured features of the specialized nightly foraging habitat where jumping mice are most likely to be captured. Consequently, at sites where I captured the New Mexico meadow jumping mouse, I established microhabitat plots at locations where I captured the species. At sites where I did not capture the New Mexico meadow jumping mouse, I selected locations for microhabitat plots haphazardly in representative foraging habitat. I did not use probabilistic selection of locations for microhabitat plots along trapping transects because it was not feasible given the complexity of habitat types in relation to survey trapping transects and because it would have biased results toward nonforaging habitat, which occupied the majority of area at most survey sites. This selection of plot locations insured that comparisons between sites where the New Mexico meadow jumping mouse was captured versus not captured were of foraging habitat. Thus, even though habitat on sampling plots were generally similar (i.e., representing foraging habitat), this analysis was aimed at revealing microhabitat variables that may not be immediately obvious that could vary between sites where the taxon is captured or not captured. However, as with the overall study design, inferences are restricted to study plots (Morrison et al. 2008).
At the center of the microhabitat plot, I visually estimated slope and estimated aspect with the aid of a compass. I measured canopy cover with a convex densitometer in the four cardinal directions. I obtained an index of soil moisture ranging from 1 (dry) to 10 (saturated) using a probe (Lincoln Irrigation, Lincoln, NE) inserted into the ground approximately 4 cm. I assessed vertical cover with a Robel pole (read in inches) from a 4-m distance at a 1-m eye level. I (or a field assistant) read the Robel pole at the plot center from three random azimuths as well as away from the plot center from three random azimuths. I established four 4-m perpendicular transects at a random azimuth from the plot center. At each 1-m interval along a transect, I used a Daubenmire (1959) frame to assess the percent ground cover of moss, field horsetail Equisetum arvense, rush, sedge, cattail Typha spp., grass, forb, willow, alder, redosier dogwood Cornus sericea, shrub cinquefoil, Wood's rose, other plants, coarse woody debris, litter, rock, gravel, bare ground, and open water. Cover classes were 1 for 0–5% cover, 2 for 5–25% cover, 3 for 25–50% cover, 4 for 50–75% cover, 5 for 75–95% cover, and 6 for 95–100% cover. In addition, I recorded soil moisture, litter depth, and stubble height in each frame. I measured stubble height with a ruler (in mm) and recorded it in two ways. I measured laid-over stubble height as the representative height of the vegetation as it naturally lay. I measured vertical stubble height by measuring the height of a representative blade of the dominant herbaceous vegetation that was fully extended vertically from the ground (Meehan et al. 2015). Finally, I recorded the number and identity of each tree and shrub within 1 m of transects.
I calculated statistics using SPSS 10.0 for Windows (SPSS 1999). I tested variables for normality using 1-sample Kolmogorov–Smirnov tests. I used Pearson and Spearman correlations to assess relationships among variables for normal and nonnormal variables, respectively. I tested for differences in the stream reach cover (Table S2) and microhabitat (Table S3) variables at survey sites where New Mexico meadow jumping mice were captured or not captured using 2-tailed t-tests (corrected using Levene's Test for equality of variances) and 2-sample Kolmogorov–Smirnov tests for parametric and nonparametric data, respectively.
Principal components analysis
I used principal components analysis (PCA) to describe variation in steam reach cover and microhabitat variables among the survey sites. For the stream reach data, I excluded five variables (percent cover of shrubby cinquefoil, other shrubs, coarse woody debris, rock, and other) to maintain a ratio of ≥2:1 in the number of samples to the number of variables, which is considered suitable for descriptive purposes (McGarigal et al. 2000). Thus, the analysis of stream reach cover included 14 variables: mean vertical cover height on both transects; variance of vertical cover height on both transects; and percent of stations covered by conifer, rush, sedge, grass, forb, willow, alder, rose, dead standing plant, and bank. For the PCA of microhabitat, I used elevation and all microhabitat variables except slope, which had missing values. The ratio of the number of samples to the number of variables (2.5:1) was considered suitable for descriptive purposes (McGarigal et al. 2000). I did not rotate variables and I only extracted components with eigenvalues ≥1.0 because these are usually sufficient to describe the variance within the variables (McGarigal et al. 2000). I retained components for interpretation based on the scree plot criterion (McGarigal et al. 2000, McCune and Grace 2002). I followed McGarigal et al. (2000) in considering variables with principal component loadings >|0.4| as important.
Discriminant function analysis
I used discriminant function analysis (DFA) to evaluate multivariate differences in stream reach cover and microhabitat at sites where New Mexico meadow jumping mice were captured versus not captured. To reduce the chance for multicollinearity problems in the data sets, I excluded from analysis variables that exhibited high (i.e., r > 0.6) correlations with other variables (McGarigal et al. 2000). For the stream-reach cover data, excluded variables included vertical cover on the inland transect, variance in vertical cover, and variance in coarse woody debris cover. For the microhabitat data, excluded variables included vertical cover at the plot center, vertical cover 4 m from the plot center, laid-over stubble height, variance in soil moisture, and slope (on account of missing data). I used a step-wise selection procedure and used Wilks' lambda to rank the variables in ability to discriminate by passing the tolerance tests (0.05 to enter; 0.10 to remove). I used a chi-square transformation of the overall Wilks' lambda to test for multivariate differences in habitat between sites where New Mexico meadow jumping mice were captured or not captured. I used a classification routine in the DFA to predict whether the New Mexico meadow jumping mouse would be captured or not captured at historical sites that were not trapped, to predict presence of New Mexico meadow jumping mice at sites where it was not captured, and to evaluate whether there were too many variables in the model. Because original classification results can provide overly optimistic estimates, I used a cross-validation procedure whereby each case in the analysis was classified by the functions derived from all cases other than that case (SPSS 1999). I compared the percentage of correct classifications between the original and cross-validated cases to assess whether there were too many predictors in the model. An excess of predictors was indicated by a substantially lower percentage of correct classification for the cross-validated cases.
I found 123 records of the New Mexico meadow jumping mouse from 21 locations in the White Mountains (Figure 2; Text S1). Three historical records (site 3, 8, 15) lacked voucher specimens. Prior researchers reported three specimens collected from two locations on Hannagan Creek—one from 8,600 ft ( = 2,621 m) elevation collected in 1932 and two from 8,200 ft elevation ( = 2,500 m) collected in 1933 (e.g., Hall and Davis 1934; Morrison 1991). However, based on the original field notes it is likely that the 8,600-ft elevation record was an error and that all three specimens were collected from the same location (Hannagan Creek, 8,200 ft; see Frey 2011 for more details). Morrison's (1991) report of a historical location approximately 0.6 mi downstream from Hannagan Lodge is an error.
Jumping mouse survey
I deployed 10,706 trap-nights at 33 survey sites, which included 12 historical locations. In total there were 1,413 captures of 19 species of mammals. This included capture of 37 (39 total captures) jumping mice at 12 sites (36% of all sites surveyed), including 6 of 12 historical sites surveyed and 6 new sites (Table 1). The New Mexico meadow jumping mouse was the sixth most abundant species captured (2.8% of captures) but it had a low overall capture rate of 0.36%, whereas capture rates at sites where it was present averaged 1.28% and ranged from 0.25 to 2.5% (Table 1). Other species that I captured were mountain cottontail Sylvilagus nuttallii, dusky shrew Sorex monticolus, western water shrew S. navigator, long-tailed weasel Mustela frenata, Mogollon vole Microtus mogollonensis, montane vole Microtus montanus, long-tailed vole Microtus longicaudus, southern red-backed vole Myodes gapperi, Mexican woodrat Neotoma mexicana, brush deermouse Peromyscus boylii, white-footed deermouse P. leucopus, North American deermouse P. maniculatus, northern rock deermouse P. nasutus, western harvest mouse Reithrodontomys megalotis, house mouse Mus musculus, golden-mantled ground squirrel Callospermophilus lateralis, gray-collared chipmunk Tamias cinereicollis, and least chipmunk Tamias minimus.
On stream reaches where the New Mexico meadow jumping mouse was captured, 70.5% of cover was provided by herbaceous vegetation (forbs, grasses, and sedges), while alders and willows together accounted for 13.4% of cover (Table 2; Figure 3). All other sources of cover were incidental (<5%). There was significantly more cover provided by alders and forbs on stream reaches where New Mexico meadow jumping mice were captured versus not captured (Table 2).
Five principal components were extracted, which together accounted for 79.9% of the variation in stream reach variables. The first two components were sufficient to describe stream reach habitat, which accounted for 31.8% and 20.5% of the variation, respectively (cumulative variation explained = 52.2%). On component 1, important variables with positive loadings included the four vertical cover measures and percent cover of willow, rose, and alder; the only important variable with a negative loading was cover of sedge. Thus, I interpreted component 1 as a “Riparian Community Gradient” extending from sites dominated by sedges (negative values) to sites dominated by shrubs (positive values). On component 2, there were no variables with positive loadings that ranked as important (variables with the highest positive loadings were percent cover of forbs, dead standing plant, and conifer), while important variables with negative loadings were percent cover of sedge, rush, and grass. Thus, I interpreted component 2 as a “Meadow-Forest Gradient” extending from open sites dominated by graminoid plants (negative values) to sites with woody plants, especially conifers, and an understory of forbs (positive values). Based on a scatter plot of surveyed stream reaches on components 1 and 2, stream reaches where New Mexico meadow jumping mice were captured overlapped those where they were not captured (Figure 4).
The DFA revealed a significant difference in stream reach cover variables between sites where New Mexico meadow jumping mice were captured or not captured (Wilks' lambda = 0.413, χ2 = 20.335, df = 4, P < 0.001). The best predictor for discriminating between these groups was percent cover by alder; other significant variables that contributed to the final model were percent cover by forbs, rushes, and bank, which was negative. The model correctly classified 85.2% and 77.8% of the original and cross-validated cases, respectively, indicating that there was no excess of predictor variables in the final model. Jumping mice were predicted to be not captured at two sites on the West Fork of the Little Colorado River (sites 3 and 4) that were not trapped because of heavy recreational use. There were more sites where New Mexico meadow jumping mice were captured but were predicted to be not captured, than vice versa, for both the original (i.e., 18.2% vs. 12.5%; sites 7 and 10) and cross-validated cases (27.3% vs. 18.8%; site M).
Microhabitat at sites where New Mexico meadow jumping mice were captured was dominated by tall, dense cover of predominantly forbs and sedges on saturated soil in immediate proximity to open water (Table 3; Figure 3). In comparison with sites where New Mexico meadow jumping mice were not captured, sites where they were captured had significantly greater and less variable soil moisture, greater variance in vertical cover, greater cover of alder and open water, less cover of litter and bare ground, and fewer shrub stems (Table 3).
Eleven principal components were extracted, which together accounted for 74.8% of the variation in microhabitat. Based on the scree plot criterion, a maximum of three components were required to describe habitat, which accounted for 15.8%, 12.0%, and 8.3% of the variation, respectively (36.1% total). On component 1, important variables with positive loadings included mean vertical cover, stubble height, willow cover, and number of shrub stems, while important variables with negative loadings were elevation and variance in soil moisture. I interpreted component 1 as a “Willow Wetland Gradient” of microhabitat at high elevation, variable soil moisture, and little willow (negative scores), to microhabitats at lower elevation, more uniform soil moisture, and with more willow cover (positive scores). On component 2, important variables with positive loadings included canopy cover, variance in vertical cover, vertical cover, coarse woody debris, alder cover, and forb cover; important variables with negative loadings were sedge cover and litter depth. I interpreted component 2 as a “Cover Variance Gradient” from microhabitat dominated by a uniform cover of sedges (negative scores) to sites with tall but variable cover due to presence of alder and forbs (positive scores). A scatter plot of microhabitat on the first two components revealed no separation between sites where New Mexico meadow jumping mice were captured and not captured, although microhabitat at sites where they were captured were never represented by extreme component scores (Figure 5). Evaluation of component 3 provided no additional explanation of patterns.
The DFA revealed a highly significant multivariate difference in microhabitats between sites where jumping mice were captured or not captured (Wilks' lambda = 0.601, χ2 = 34.612, df = 4, P < 0.001). The strongest predictor for differences in microhabitat between sites where New Mexico meadow jumping mice were captured versus not captured was alder cover, although other significant variables that contributed to the final model were cover by water, other plants, and bare ground, which was negative. The classification routine correctly classified 80.0% of both the original and cross-validated cases. The identity in percent correct classification between the original and cross-validated cases indicated that there were a reasonable number of predictor variables in the model. The proportion of misclassifications was equal between the groups. Misclassification of microhabitat at sites where New Mexico meadow jumping mice were not captured occurred at East Fork Little Colorado River Upper, Rudd Creek Upper, Hannagan Creek Lower, and KP Cienega. All but one of the misclassifications at sites where New Mexico meadow jumping mice were captured were at sites where there were multiple captures of jumping mice (sites 7, 8, 11, 13, E). One microhabitat plot established at West Fork Little Colorado River Upper (site 4), which was not trapped because of heavy recreational use, was predicted to be a noncapture location.
New Mexico meadow jumping mouse distribution
Surveys can suffer from two main types of error: false positive results (i.e., Type I error, which is due to misidentification) and false negative results (i.e., Type II error, which is due to methods that result in low detection rates such as inadequate effort or inappropriate targeting of habitat for sampling). The photographs and voucher specimens collected during this study provide high-quality physical evidence that negate false positive error. The high correct classification rates in the DFAs for both the stream reach and microhabitat data indicate that the survey methods were not prone to missing New Mexico meadow jumping mice at sites, if present. In addition, I captured New Mexico meadow jumping mice on the first night of trapping at 52% of sites, and on the second night of trapping at 42% of sites; in no instance did ≥3 nights of trapping produce a first capture. Finally, the overlap of microhabitat at capture and noncapture sites in the PCA scatter plot indicates that there was little, if any, bias in the selection of locations for microhabitat plots at noncapture sites, demonstrating that researchers with appropriate knowledge and experience can correctly identify key foraging microhabitat used by the New Mexico meadow jumping mouse. This provides strong support that trapping transects targeted habitats where New Mexico meadow jumping mice are most likely to be captured. Consequently, I conclude that the survey methods reliably detected New Mexico meadow jumping mice at sites where they were present. Similarly, at sites where I did not capture New Mexico meadow jumping mice I conclude that they were likely functionally absent at the time of the survey. These results allow comparisons of persistence rates at historical sites with results of other surveys using the same methods. However, it must be cautioned that failure to capture the New Mexico meadow jumping mouse during this survey does not necessarily mean that the population is permanently extinct. Although some occupied sites appeared isolated by unsuitable habitat, some sites (e.g., site 6) may have been temporarily unoccupied and habitat corridors to other occupied sites could allow for recolonization.
I did not capture jumping mice at half (6 of 12) of the historical sites trapped, and the DFA predicted they would not be captured at two additional historical sites that were not trapped. Together with an additional historical site where New Mexico meadow jumping mice were documented in 2007 (site 12), the overall rate of population persistence in the White Mountains was approximately 47%, which is more than that reported for the New Mexico meadow jumping mouse in the Jemez and Sacramento mountains, New Mexico (i.e., 27% and 6% respectively; Frey and Malaney 2009). All historical sites where I captured New Mexico meadow jumping mice were in the watershed of the Black River, which occupies the majority of the interior of the White Mountains, suggesting that threats to jumping mice may be more acute in peripheral watersheds such as the San Francisco and Little Colorado rivers. In addition, I did not capture New Mexico meadow jumping mice at 33% of nine historical sites where the species was known to occur as recently as the 1980s and1990s, indicating that there have been ongoing threats to population persistence in the White Mountains.
The six new sites fill important distributional gaps. These include the first records for the Nutrioso Creek watershed (site K), which is a major tributary to the Little Colorado River. Captures on Campbell Blue Creek (site O) verify occurrence in the Blue River watershed; prior, the only record from this watershed was a single specimen with obscure data. The capture of a New Mexico meadow jumping mouse on Corduroy Creek (site Q) is the first record from the Hannagan–KP peak massif since 1933. The two new sites in the upper San Francisco River drainage (sites L and M) represent the first verified (by physical evidence) records since 1933, although there also was an unverified report in 1981 (Text S1). Finally, the new site on the East Fork of the Little Colorado River (site E) was the only site where I captured New Mexico meadow jumping mice in the watershed of the upper Little Colorado River.
Overall, I caught New Mexico meadow jumping mice at 36% of 33 sites trapped, with an overall capture rate of 0.36%. In contrast, Morrison (1991) captured New Mexico meadow jumping mice at 26% of 19 sites in the White Mountains, with an overall capture rate of 0.22%. Difference in frequency and capture rate between the two studies may reflect differences in methods and areas surveyed. However, difference in capture rates also is likely influenced by annual variability that typifies small mammal communities.
New Mexico meadow jumping mouse habitat in the White Mountains
Habitats used by the New Mexico meadow jumping mouse in the White Mountains were similar to those reported in other regions (e.g., Frey and Malaney 2009; Wright and Frey 2015). Microhabitat used by the New Mexico meadow jumping mouse in the White Mountains was characterized by 1) near flowing water (open water typically ranked as third or fourth most prevalent cover type); 2) on saturated soil (i.e., 9–10 on scale 0–10); 3) predominant plant cover consisted of forbs and sedges; 4) high vertical cover (mean = 24 inches); 5) low to moderate canopy cover (<50%); and 6) absence of trees. Thus, these results confirm that the New Mexico meadow jumping mouse is a riparian specialist that utilizes tall, dense herbaceous vegetation on saturated soil. This description pertains to habitats used by active jumping mice during their nightly foraging (and other) activities, which is where they are most vulnerable to trapping. The New Mexico meadow jumping mouse may use other habitats, including adjacent uplands, for traveling, seasonal foraging, day nesting, maternal nesting, and hibernating (Wright and Frey 2015). However, because herbaceous riparian habitat is limited in distribution and is particularly sensitive to disturbances, it is the availability of this foraging habitat that is a key limiting factor for the New Mexico meadow jumping mouse.
Sedges are considered an important habitat component for the New Mexico meadow jumping mouse (USFWS 2014a, b; 2016). For instance, Frey and Malaney (2009) found that microhabitat at capture locations in the Jemez and Sacramento mountains was composed primarily of sedges, followed by forbs and grasses. Results of this study confirm that sedges are an important constituent of the habitat used by New Mexico meadow jumping mice in the White Mountains, although not the dominant component: at sites where New Mexico meadow jumping mice were captured, sedges were the third most prevalent (behind forbs and grasses) source of cover at the landscape scale and were the second most prevalent (behind forbs) source of cover at the microhabitat scale. However, results indicated a trend of greater cover of sedges at sites where I did not capture New Mexico meadow jumping mice. Most sedges are associated with saturated soils and some species can tolerate waterlogged soils. Consequently, along streams sedges intermix with other plants or form small patches or stringers (usually along water's edge) on seasonally flooded soil within an admixture of other plant types. Such sites provide ideal foraging habitat for jumping mice. However, sedges also form nearly monotypic stands (sometimes codominant with willows) in areas with standing water. Prior studies concluded that New Mexico meadow jumping mice do not normally use areas of monotypic sedge on relatively deep, semiperennial standing water (Morrison 1990; Frey and Malaney 2009). Results of this study support that conclusion. For instance, I did not capture New Mexico meadow jumping mice at a site in a broad valley at the confluence of the East and West forks of the Little Colorado River (site D) that was dominated by sedges and willows on standing water, but I did capture New Mexico meadow jumping mice a short distance upstream on the East Fork of the Little Colorado River (site E) where the valley was narrower and sedges formed part of the admixture of diverse riparian plants. There are at least two reasons that sedge-dominated plant communities on relatively deep (i.e., from the perspective of a jumping mouse; >2 cm) standing water are not suited for the New Mexico meadow jumping mouse. First, although sedges can provide ideal cover, which is important for the New Mexico meadow jumping mouse, Carex constitutes a minor fraction of the known diet of Z. hudsonius (Quimby 1951). Second, although the New Mexico meadow jumping mouse swims well, it still requires drier areas for nesting. Thus, the New Mexico meadow jumping mouse primarily utilizes sedges as components of diverse herbaceous plant communities and does so primarily while engaged in nightly foraging activities. To summarize, presence of sedge is a good (but not only) indicator of New Mexico meadow jumping mouse habitat, although not all sedge-dominated communities are occupied.
Results of the multivariate analyses provide important insight into habitat relations of the New Mexico meadow jumping mouse. The DFA of stream reach cover suggested that there were sites where I did not capture the New Mexico meadow jumping mouse, but habitat was suitable. These cases may represent sites where New Mexico meadow jumping mice were extirpated. Morrison (1991) came to the same conclusion. Such extirpations could be temporary and recolonization may subsequently occur from nearby (but possibly unidentified) source populations. Alternatively, extirpation could have occurred historically but recolonization of subsequently restored habitat has not been possible because of intervening unsuitable environment. Connectivity of habitat is more likely to be an important issue for a small terrestrial mammal in comparison with many other endangered species such as birds, bats, and large mammals that have great vagility and ability to move through areas lacking suitable conditions. Sites with apparently suitable habitat but that are currently unoccupied may provide opportunity for restoration of New Mexico meadow jumping mice, either via translocation or natural recolonization through improvements in riparian habitat in intervening areas. However, it remains a possibility that there are other unmeasured aspects of the environment that could render an area unsuitable for the species. In contrast, I captured New Mexico meadow jumping mice at three sites (sites 7, 10, M) where they were predicted not to be captured based on stream reach cover data. This could be due to unmeasured factors or conditions at those sites might be marginal for New Mexico meadow jumping mice, and hence those populations might be at enhanced risk of extirpation.
Differences compared with other populations
High vertical cover provided by herbaceous plants is considered a key habitat requirement for the New Mexico meadow jumping mouse (USFWS 2014a, b; 2016). However, I found differences in the microhabitat results for vertical cover in the White Mountains in comparison with microhabitat results presented by Frey and Malaney (2009) for the Jemez and Sacramento mountains. First, mean vertical cover height at capture locations was lower in the White Mountains (24.9 inches [632 mm], SE = 1.75, n = 39) as compared with the Jemez and Sacramento mountains (32.6 inches [829 mm], SE = 3.58, n = 14, Frey and Malaney 2009). This difference is likely due to the small sample size in Frey and Malaney (2009) and differences in the natural growth potential of different riparian plant communities, rather than reflecting selection by jumping mice for different habitats in different regions. For instance, in the White Mountains the dominant plant cover at capture locations was forbs, whereas in the Jemez and Sacramento mountains it was sedges. Vertical cover provided by most riparian forbs is less than that provided by some species of sedge, such as beaked sedge Carex rostrata, which was common at capture locations in the Jemez and Sacramento mountains. Second, Frey and Malaney (2009) found that capture locations in the Jemez and Sacramento mountains had significantly greater vertical cover than microhabitat at sites where New Mexico meadow jumping mice were not captured, whereas no such difference was found during this study. Frey and Malaney (2009) attributed differences in microhabitat at sites where the New Mexico meadow jumping mouse was captured or not captured to livestock grazing; most of their capture sites were in grazing exclosures, whereas most of their noncapture sites had authorized livestock grazing. In contrast, during this study all but two sites had no authorized livestock grazing. Thus, it is not unexpected that vertical cover height did not differ between capture and noncapture sites during this study.
Both the stream reach cover and microhabitat results revealed that although the dominant cover at sites where I captured New Mexico meadow jumping mice was provided by forbs, sedges, and grasses; capture sites in the White Mountains also had significantly more cover provided by alders as compared with noncapture sites. Cover by alder was the strongest predictor of differences between capture and noncapture sites in both the stream reach and microhabitat data. This pattern has not been reported previously. Alders are not known to be used by jumping mice for food, but they may be indicative of other factors that influence occurrence of New Mexico meadow jumping mice (e.g., restoration of riparian habitat following livestock exclusion) or may contribute to cover. Typically, alders are associated with streams at higher elevations and on narrower floodplains where they are restricted to soils that are temporarily flooded. In the White Mountains, I observed that alders usually occurred individually or in small clumps on the bank near water's edge. The herbaceous communities associated with alders in these narrow valleys tended to be dominated by forbs, with sedges occurring on saturated soil in sunny locations along water edge, although in some broader valleys sedges dominated saturated soils in the interspaces between alders. The growth form of alders usually consisted of several well-defined woody trunks with branches extending outward laterally from the trunks. I typically captured New Mexico meadow jumping mice in lush herbaceous vegetation in the interspaces between alders or at the “drip-line” of alder crowns. In these situations the tips of alder branches overlaid the herbaceous plant layer, contributing to the cover over the trap. However, the area around the trunks under the crown was devoid of herbaceous vegetation (i.e., it was bare or covered in litter) and appeared to be avoided by New Mexico meadow jumping mice. This explains why microhabitat at capture sites was associated with greater alder cover, but lower density of shrub stems. In addition, variance in vertical cover was significantly greater in microhabitat at capture sites, which reflects its significant positive correlation with alder cover (rs = 0.353; P < 0.01).
At Bosque del Apache National Wildlife Refuge, New Mexico meadow jumping mice selected a mesic graminoid–willow vegetation association at the landscape scale and were commonly associated with regenerating stands of willow that had an herbaceous understory (although jumping mice did not select willows more than available at the microhabitat scale; Wright and Frey 2015). However, in this study neither the stream reach cover nor microhabitat data revealed a difference in willow cover at capture versus noncapture sites. There are ≥11 species of willows occurring in the White Mountains (Dreesen et al. 2002), and they are natural components of most riparian plant communities, including in association with alders (Brown 1994). However, based on my observations, willows appeared to be less common than alders in the White Mountains. For example, willow cover was recorded in 13 (33%) microhabitat plots while alder cover was recorded in 27 (69%) microhabitat plots. There was a tendency for noncapture sites to have greater cover of willows and the PCA of microhabitat suggested that greater willow cover was associated with lower elevations and more uniform soil moisture. This may reflect the same pattern that was observed for sedges, in that areas with standing water and dominated by sedges and willows may represent inferior habitat for jumping mice. Thus, quantitative data provide little support for the statement by Hafner et al. (1981) that the White Mountain population of the New Mexico meadow jumping mouse is associated with dense willow thickets.
Livestock grazing is considered a key threat to the New Mexico meadow jumping mouse because it can alter the composition and structure of riparian plant communities (Morrison 1990, 1991; Frey and Malaney 2009; USFWS 2014a, b; 2016). In the Jemez and Sacramento mountains, all sites but one where the New Mexico meadow jumping mouse was captured were within fenced livestock exclosures; the exception occurred in a wetland created by beavers Castor canadensis that provided natural exclusion via a complex arrangement of water and mucky soils (Frey and Malaney 2009). Frey and Malaney (2009) concluded that livestock grazing was the primary proximate cause of population loss and attributed differences in vegetation composition and structure at capture and noncapture sites to livestock grazing. As in the current study, Frey and Malaney (2009) surveyed historical locations in addition to new areas that had potential habitat. However, although the majority of survey sites in the Frey and Malaney study were grazed by livestock (including most of the historical locations), during this study all of the survey sites except two were excluded from livestock grazing (including all historical locations). Thus, although it was not possible to compare sites that were excluded from livestock grazing versus those that were grazed as in Frey and Malaney (2009), results of this study support Frey and Malaney (2009) in that all sites where New Mexico meadow jumping mice were captured did not have authorized livestock grazing. However, this does not exclude the possibility that the New Mexico meadow jumping mouse can use some areas grazed by livestock; I did not design this study to test that hypothesis.
In the White Mountains there has been widespread exclusion of cattle from riparian zones over the past several decades (Table S1). This factor probably contributes to the higher persistence rate of the New Mexico meadow jumping mouse in the White Mountains as compared with the Jemez and Sacramento mountains. However, I observed evidence of unauthorized livestock use at several sites (Table S1). I captured jumping mice more often at sites where I did not observe sign of livestock grazing (i.e., 11 of 25 sites) as compared with sites where I observed livestock grazing (i.e., 1 of 8 sites). Thus, although livestock grazing does not guarantee absence of jumping mice and there may be tolerable levels of disturbance, livestock exclusion alone does not guarantee presence of jumping mice, even if habitat is restored. Recolonization of a site where habitat has been restored would require dispersal along riparian corridors with suitable environmental condition from persisting source populations. Based on my observations, many of the currently occupied sites were isolated by stream reaches with unsuitable conditions.
I did not find a significant relationship (P > 0.05) between the number of years since livestock were excluded at a site with presence or capture rate of jumping mice, but I did find that time since exclusion of livestock was positively correlated with two stream reach cover variables—percent alder cover (rs = 0.452, P = 0.023), and its strong correlate, variance in vertical cover at the stream-edge (rs = 0.408, P = 0.043). This suggests that the temporal pattern of livestock grazing may be a factor in the association of jumping mice with alders and the disproportionate abundance of alders observed relative to willows. In the White Mountains, most streamside riparian plant communities are alliances represented by various species of alders and/or willows (Brown 1994). However, in areas with heavy browsing alders are more likely to persist than willows because they are less palatable and reproduction occurs both by seeds and clonally via rhizomes. Studies have shown that once livestock grazing has been curtailed, alders can exhibit explosive growth resulting in significant increases in stem density and highly significant and dramatic increases in height; response of alders to livestock exclusion can be more dramatic than the response exhibited by willows (Green and Kauffman 1995; Case and Kauffman 1997). In contrast, willows are highly palatable to ungulates and overbrowsing can result in their differential loss from riparian plant communities (Belsky et al. 1999; Baker et al. 2005; Gage and Cooper 2005). Willows primarily reproduce via wind-dispersed seeds, which have limited dispersal ability, and asexual establishment of new plants via stems cut by beavers, which are dependent on downstream transport of stems from areas occupied by beavers (Gage and Cooper 2005). Thus, a combination of ungulate browsing and loss of beavers could reduce or eliminate occurrence of willows. Further, because reproduction by willows is primarily via wind-dispersed seeds, it may take much longer for willows to reestablish in an area, particularly if they have been altogether eliminated. Browsing by elk Cervus elaphus also could inhibit restoration of willows. Thus, the relatively recent (i.e., past two decades) widespread exclusion of cattle from riparian zones in the White Mountains may have had a differential impact on alders versus willows, such that alders are more likely to be recorded at sites where New Mexico meadow jumping mice persist.
The 13 sites known to be occupied by the New Mexico meadow jumping mouse in the White Mountains since 2007 represents a majority of the sites where the taxon has been documented since 2005 range-wide (i.e., Jemez Mountains = 5; Sacramento Mountains = 4; Rio Grande = 1; Coyote Creek = 2; Sugarite Canyon = 3; San Juan River watershed = 2; USFWS 2014a, b; 2016). Consequently, the White Mountains currently represents the single greatest potential for the long-term persistence of the taxon. However, based on my survey results and my observations of habitat conditions, many of the populations appeared small and isolated, which can reduce potential for long-term population persistence (Frey 2011). The New Mexico meadow jumping mouse is particularly vulnerable to population extinction as a result of a variety of factors. First, it has a low reproductive rate. Only older females may be capable of breeding and only a single litter may be produced each year in montane populations (Frey 2015). Organisms with low intrinsic rate of population growth are at greater risk of extinction because they recover more slowly from reductions in population size and they also remain threatened longer because of demographic and genetic stochasticity (Beissinger 2000). Second, the New Mexico meadow jumping mouse is naturally rare within small mammal communities. Reasons for this rarity may include its low biotic potential resulting in a lack of wide population fluctuations (Kirkland and Kirkland 1979), predation by a spectrum of predators, and competition with more aggressive and abundant species that may suppress populations of jumping mice (e.g., Boonstra and Hoyle 1986). Third, the New Mexico meadow jumping mouse has limited dispersal capability, which restricts its ability to reoccupy habitats where it was extirpated. This problem is compounded by the fact that corridors of riparian habitat are linear; consequently, reaches with unsuitable habitat can cause isolation of populations. Fourth, the New Mexico meadow jumping mouse is an extreme habitat specialist at multiple scales: it is exclusively associated with certain riparian ecosystems at the landscape scale, and is associated with a narrowly defined microhabitat consisting of tall, dense herbaceous plants, usually with a strong sedge or forb component on saturated soil. Water is a rare commodity in the American Southwest and riparian zones are disproportionately used by wildlife and humans. The hydrology and vegetation of these systems can be altered by a wide array of factors, such as climate, livestock grazing, and human uses. Consequently, unless successful management actions are implemented, it is likely that there will be further reductions in the number of occupied sites in the White Mountains, with particular risk of extinction of the species in peripheral watersheds including the Little Colorado River, San Francisco River, and Blue River.
The New Mexico meadow jumping mouse is a riparian obligate that requires tall, dense herbaceous vegetation composed primarily of forbs and sedges on saturated soils in close proximity to flowing water for key life-history functions. Factors that degrade this habitat may pose a threat to the taxon. Consequently, management to conserve the New Mexico meadow jumping mouse should seek to reduce degradation and enhance restoration of this habitat. Additional studies on the New Mexico meadow jumping mouse are needed to evaluate habitat selection at multiple scales; evaluate influence of disturbances (especially livestock grazing) on habitat use and occupancy; and to critically evaluate habitats used for other key life-history functions including dispersal, day nesting, maternal nesting, hibernation, and seasonal shifts in foraging. Results indicate that the rate of population loss of the New Mexico meadow jumping mouse in the White Mountains has been less than in some other mountain ranges, likely because of the more widespread exclusion of livestock grazing in this range. However, although persistence of the New Mexico meadow jumping mouse has been confirmed at seven historical and six new sites in the White Mountains, this study was not an exhaustive survey of all possible sites within the mountain range. Further, no effort was made to critically evaluate the distributional limits of populations represented by capture locations. Some occupied sites might form single panmictic populations, especially in watersheds where there has been widespread reduction in livestock grazing and other disturbances to riparian habitats (e.g., West Fork of the Black River), whereas other sites might be isolated. Size and connectivity of populations are linked to extinction risk, so it is important to conduct more surveys to find additional new populations and to better define the limits of distribution of the taxon within each watershed. Results of the habitat analyses provide crucial baseline information about habitats used by the New Mexico meadow jumping mouse in the White Mountains that can inform site selection for future surveys and inform development of species distribution models.
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.
Table S1. Livestock grazing history at survey sites for the New Mexico meadow jumping mouse Zapus hudsonius luteus in the White Mountains, Apache and Greenlee counties, Arizona, 2008–2009.
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S1 (36KB PDF).
Table S2. Archive of stream-reach cover data collected at sites surveyed for the New Mexico meadow jumping mouse Zapus hudsonius luteus in the White Mountains, Apache and Greenlee counties, Arizona, 2008–2009.
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S2 (22KB XLSX).
Table S3. Archive of microhabitat data collected at sites surveyed for the New Mexico meadow jumping mouse Zapus hudsonius luteus in the White Mountains, Apache and Greenlee counties, Arizona, 2008–2009.
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S3 (28 KB XLSX).
Text S1. History of studies reporting records of Zapus hudsonius luteus in the White Mountains, Arizona.
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S4 (33 KB PDF).
Reference S1. Dreesen D, Harrington J, Subirge T, Stewart P, Fenchel G. 2002. Pages 253–272 in Dumroese RK, Riley LE, Landis TD, technical coordinators. National proceedings Forest and Conservation Nursery Associations—1999, 2000, and 2001. Ogden, Utah: U.S. Department of Agriculture Forest Service, Rocky Mountain Research Station. Proceedings RMRS-P-24.
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S5 (234 KB PDF); also available at http://www.nrcs.usda.gov/Internet/FSE_PLANTMATERIALS/publications/nmpmcsy03852.pdf (234 KB PDF).
Reference S2. Frey JK. 2011. Inventory of the meadow jumping mouse in Arizona. Report to Arizona Game and Fish, Phoenix.
Reference S3. [USFWS] U.S. Fish and Wildlife Service. 2014a. Final rule: determination of endangered status for the New Mexico meadow jumping mouse throughout its range. Federal Register 79(111):33119–33137; 10 June 2014.
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S7 (288 KB PDF); also available at https://www.gpo.gov/fdsys/pkg/FR-2014-06-10/pdf/2014-13094.pdf (288 KB PDF).
Reference S4. [USFWS] U.S. Fish and Wildlife Service, Listing Review Team. 2014b. Species status assessment report: New Mexico meadow jumping mouse (Zapus hudsonius luteus).
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S8 (27 MB PDF); also available at http://www.fws.gov/southwest/docs/NewMexicomeadowjumpingmousefinalSSA.pdf (27 MB PDF).
Reference S5. [USFWS] U.S. Fish and Wildlife Service. 2015. Interim survey guidelines for the New Mexico meadow jumping mouse.
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S9 (71 KB PDF); also available at https://www.fws.gov/southwest/es/NewMexico/documents/SP/New_Mexico_Meadow_Jumping_Mouse_survey_protocol.pdf (71 KB PDF).
Reference S6. [USFWS] U.S. Fish and Wildlife Service. 2016. Final rule: designation of critical habitat for the New Mexico meadow jumping mouse. Federal Register 81 (51):14264–14325; 16 March 2016.
Found at DOI: http://dx.doi.org/10.3996/062016-JFWM-043.S10 (60 KB PDF); also available at https://www.gpo.gov/fdsys/pkg/FR-2016-03-16/pdf/2016-05912.pdf (60 MB PDF).
I thank M. Calkins, J. Malaney, M. Moses, J. Redman, and G. Wright for assistance in the field. T. Frey assisted with logistics and other support. I thank the Apache–Sitgreaves National Forest, especially L. White-Trifaro, for assistance and information. I thank the Arizona Game and Fish Department for assistance and permits to access lands managed by the agency, especially to D. Cagle, J. Cooley, B. Crawford, M. Goodwin, and R. Sieg. I thank J. Underwood for information regarding field surveys by the Arizona Game and Fish Department and S. Schwartz for providing records in the Arizona Heritage Data Management System. I thank R. Goljani-Amirkhiz for preparing Figure 1. Special thanks are due to the various museums for help with data queries, loans of specimens, and arrangements to visit collections. Funding for this study was provided by Arizona Game and Fish Heritage Fund Grant I09004. I thank three anonymous reviewers and the editors for helpful comments on a previous version of this 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.
Citation: Frey JK. 2017. Landscape scale and microhabitat of the endangered New Mexico meadow jumping mouse in the White Mountains, Arizona. Journal of Fish and Wildlife Management 8(1):39-58; e1944-687X. doi:10.3996/062016-JFWM-043
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.