Southeastern Naturalist 141 C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 22001188 SOUTHEASTERN NATURALIST 1V7o(1l.) :1174,1 N–1o5. 41 Comparison of Relative Abundance and Microhabitat of Desmognathus organi (Northern Pygmy Salamander) and Desmognathus wrighti (Southern Pygmy Salamander) in North Carolina C. Reed Rossell Jr.1,*, Ivy C. Haas1, Lori A. Williams2, and Steven C. Patch3 Abstract - There are currently no quantitative studies describing the habitat of Desmognathus organi (Northern Pygmy Salamander) and D. wrighti (Southern Pygmy Salamander). We investigated the relative abundance and microhabitat selection of Northern Pygmy Salamander and Southern Pygmy Salamander in 3 forest types of different elevations across the mountains of North Carolina during the summer of 2015. We conducted 2-h time-constrained searches at 4 sites in Picea rubens (Red Spruce)–Abies fraseri (Fraser Fir) forests, northern hardwood forests, and mountain cove forests for each species. We quantified microhabitat characteristics at each pygmy salamander location and at a corresponding random location 2–30 m away. We captured a total of 98 pygmy salamanders (D. organi = 41, D. wrighti = 57): 52 in spruce–fir forests, 26 in northern hardwood forests, and 20 in cove forests, and recorded 655 other salamanders representing 15 species. Relative abundance of pygmy salamanders was greater in spruce–fir forests than in cove forests, but was not significantly different between spruce–fir and northern hardwood forests. Microhabitat did not differ between Northern Pygmy Salamander and Southern Pygmy Salamander for any of the variables examined, except for soil moisture, which was greater at Northern Pygmy Salamander locations and may have been a spurious result. We observed pygmy salamanders almost exclusively beneath wood cover-objects, and size of cover objects did not differ from the size available in the surrounding environments. Total area of small-sized down woody debris (DWD) and total area of large-sized DWD were the only variables associated with the presence of pygmy salamanders, suggesting that pygmy salamanders avoid predation and interspecific competition by selecting sites that minimize encounters with larger salamanders. Introduction Desmognathus organi Crespi, Brown, and Rissler (Northern Pygmy Salamander) and D. wrighti King (Southern Pygmy Salamander) are rare endemic species generally associated with high-elevation forests of the southern Appalachian Mountains (Crespi et al. 2010, Petranka 1998). In 2010, D. wrighti was split into 2 species based on genetic, morphometric, and ecological measurements (Crespi et al. 2010). The range of the 2 species is divided by the French Broad River, with Northern Pygmy Salamander occurring north and east of the river, and Southern Pygmy 1Department of Environmental Studies, University of North Carolina at Asheville, Asheville, NC 28804. 2North Carolina Wildlife Resources Commission, 177 Mountain Laurel Lane, Fletcher, NC 28732. 3Department of Mathematics, University of North Carolina at Asheville, Asheville, NC 28804. *Corresponding author - crrossell@aol.com. Manuscript Editor: Max Nickerson Southeastern Naturalist C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 142 Salamander occurring south and west of the river (Crespi et al. 2010). Both species occur in highly localized populations restricted to the most prominent mountains of western North Carolina, eastern Tennessee, and southwestern Virginia (Crespi et al. 2003, 2010; Petranka 1998). Both Northern Pygmy Salamander and Southern Pygmy Salamander are listed as Federal Species of Concern, with global rankings of G3 (vulnerable; LeGrand et al. 2014). Populations in North Carolina, Virginia, and Tennessee are considered vulnerable to extinction, with fewer than 100 occurrences documented in each state (LeGrand et al. 2014, Roble 2013, Withers 2009). In North Carolina, both species are listed as State Rare (LeGrand et al. 2014) and are considered Species of Greatest Conservation Need in the North Carolina Wildlife Action Plan (NCWRC 2015). Threats to the species include loss and fragmentation of habitat, environmental perturbations that reduce canopy cover such as acid rain deposition, outbreaks of insects including Choristoneura fumiferana Clemens (Spruce Budworm) and Adelges piceae Ratzeburg (Balsam Woolly Adelgid); and over-collecting for the pet industry (Hammerson 2004). Pygmy salamanders are one of the smallest and most terrestrial of all Desmognathus species (Petranka 1998). They are unique in that embryos completely absorb their gills before hatching (Organ 1961a, Petranka 1998). Pygmy salamanders are thought to have evolved in high-elevation Picea rubens Sarg. (Red Spruce)–Abies fraseri (Pursh) Poir (Fraser Fir) forests (Crespi et al. 2003, 2010), where they are most commonly found (Hairston 1949, King 1936, Organ 1961b), but they also are known to occur in deciduous forests as low as 762 m elevation (Bruce 1977, Rubin 1971, Tilley and Harrison 1969). However, reports of low-elevation occurrences are based on only a few observations (Bruce 1977, Hairston 1949, Rubin 1971, Tilley and Harrison 1969). Several authors have speculated that these low-elevation populations may be warm-tolerant, postglacial relicts that have persisted since the Pleistocene Era (Bruce 1977, Rubin 1971, Tilley and Harrison 1969). Currently, little is known about the relative abundance of these lower-elevation populations, particularly as they compare to populations at higher elevations. To date, there are no quantitative studies describing the habitat of Northern Pygmy Salamander and Southern Pygmy Salamander. Most of the available information on their habitat is general, and either comes from studies documenting the vertical and horizontal distribution of Desmognathus species in the mountains of North Carolina (Hairston 1949, Organ 1961b) or from observational reports (e.g., Bruce 1977, King 1936, Rubin 1971, Tilley and Harrison 1969). In general, pygmy salamander habitat is reported as mature forests in areas far from seeps and streams (Hairston 1949, Organ 1961b, Petranka 1998). However, some females reportedly nest in seeps and streambanks during the spring and early summer (Bruce 1977, Organ 1961a, Petranka 1998, Rossell et al. 2016), and aggregations have been found in streambanks in the fall and early winter (Bruce 1977, Organ 1961b). Pygmy Salamanders are known to utilize a variety of cover objects including logs, bark, rocks, and moss (Bruce 1977, King 1936, Hairston 1949, Tilley and Harrison 1969), and King (1936) indicated that they prefer small cover-objects versus larger ones. We Southeastern Naturalist 143 C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 have made similar field observations (C. Rossell Jr. and L. Williams; pers. observ.). However, preference for a habitat attribute cannot be determined with certainty unless the attribute is examined in terms of its use versus availability (Rossell et al. 2006, 2013). In the genus Desmognathus, microhabitat associations can be used to predict the macrohabitat of individual species (Krzysik 1979). Crespi et al. (2010) used climate models as supporting evidence for the taxonomic revision of D. wrighti. Their ecological-niche models indicated that localities of pygmy salamanders in areas north of the French Broad River were generally colder and drier than those south of the river. They hypothesized that unique environmental niche space was produced where divergent selective pressures, coupled with isolation, diversified the lineage of the species. However, this hypothesis remains untested because the habitat of Northern Pygmy Salamander and Southern Pygmy Salamander has not been quantified or compared. In this study we provide quantitative data on relative abundance and microhabitat of these taxa across 3 forest types of different elevations in the mountains of North Carolina. In addition, we compare microhabitat attributes used by both species and determine whether those attributes differ from those randomly available in the surrounding environment. Methods We conducted our study from 29 June to 25 August 2015. We selected our study sites from North Carolina Wildlife Resource Commission (NCWRC) historical occurrence records (Fig. 1). Occurrence records were divided into categories based on 3 major forest types at different elevations: spruce–fir forests at >1645 m, northern hardwood forests at 1310–1645 m, and mountain cove forests at less than 1310 m (Schafale 2012). Schafale (2012) provided complete floral descriptions of the forest types. In brief, spruce–fir forests are considered high-elevation forests dominated by Red Spruce and Fraser Fir, and also may include Betula alleghaniensis Britt. (Yellow Birch), Acer spicatum Lam. (Mountain Maple), and Sorbus americana Marshall (Mountain Ash). Shrubs are often sparse, but may be dense, and herbaceous layers vary from sparse to dense. Northern hardwood forests are mid-elevation forests dominated by mesophytic hardwoods including Yellow Birch, Acer saccharum Marshall (Sugar Maple), and Fagus grandifolia Ehrh. (American Beech). Shrub layers range from fairly sparse to moderately dense, and herbaceous layers are generally dense and moderately diverse. Mountain cove forests are low-elevation forests dominated by mesophytic species including Liriodendron tulipifera L. (Tulip Poplar), Prunus serotina Ehrh. (Black Cherry), Fraxinus americana L. (White Ash), and Tsuga canadensis (L.) Carr. (Eastern Hemlock). Shrub and herbaceous layers range from sparse to dense, and some herbaceous layers are highly diverse (Schafale 2012). We sampled 4 sites in each of the forest types for both Northern Pygmy Salamander and Southern Pygmy Salamander. We chose sites within each forest type that were as comparable in elevation as possible between the 2 species. We alternated sampling of each species within 1–4 d of each other to minimize effects of Southeastern Naturalist C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 144 changing weather patterns. We also sampled low-elevation sites first to increase our chances of finding animals active on the surface as the season p rogressed. We located pygmy salamander sites using GPS coordinates of occurrence records. We conducted a 2-h time-constrained search at each site (Corn and Bury 1990). A total of 4 individuals skilled at salamander surveys participated in searches, and usually 2–3 individuals worked a site at any given time. All search efforts were equal between sites. Researchers were careful to avoid overlap in areas searched, and search time was paused during all captures of pygmy salamanders. Searching for salamanders entailed looking under all potential cover-objects that were at least 3 cm wide x 5 cm long, the smallest size considered suitable to conceal an adult pygmy salamander (total length [TL] of adult pygmy salamander = 3.7–5.1 cm, Petranka 1998). During searches, we counted all salamanders and identified them to species when possible. When we found a pygmy salamander, we paused the search time to place the animal in a clean plastic bag with moss or leaf litter to prevent its desiccation and flagged the location before resuming the search. Once the search period expired, we determined the age and sex of each pygmy salamander selected for habitat analyses (see details below). We based age determinations on Figure 1. Location of study sites of Desmognathus organi (Northern Pygmy Salamander) and D. wrighti (Southern Pygmy Salamander) based on historic occurrence records in North Carolina, June–August 2015. Southeastern Naturalist 145 C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 the TL of animals (juveniles: TL < 3.7 cm, adults: TL ≥ 3.7 cm) and sex determination on the presence of a U-shaped mental gland in males and, in some cases, the observation of eggs in gravid females (Petranka 1998). We released each animal at its capture site immediately after processing. We collected habitat data immediately after each time-constrained search. At each pygmy salamander location (herein, referred to as occupied location), we recorded type of cover object (rock, round wood, bark, etc.), and measured cover-object dimensions (maximum length and width), and distance to water (up to 30 m). We measured soil pH and soil moisture under the cover object at a depth of 2 cm below the surface using a Kelway soil pH and moisture meter (Kel Instruments, Wyckoff, NJ). We also measured the soil surface temperature under the cover object using a Taylor digital thermometer (Taylor Precision Products, Las Cruces, NM). We estimated percent canopy cover of vegetation >0.5 m in height using a densitometer (Geographic Resource Solutions, Arcata, CA) held 0.5 m above the cover object. We used a 0.25-m2 quadrat frame centered over the cover object to estimate percent cover of plants <0.5 m in height; percent ground cover in each of 5 categories (bare soil, woody debris, leaf litter, rock, and water); and percent substrate composition in each of 5 size-classes (modified from Rosgen 1996): soil and sand (<2 mm), gravel (2–65 mm), cobble (66–250 mm), boulder (>250 mm), and bedrock. We used a 1-m–radius circular plot centered on the cover object to measure the density of down woody debris (DWD). We used a Biltmore stick to measure the length and width of each piece of DWD that met the minimum size criteria for cover objects (as described above); only portions of DWD that were touching the ground and within the plot were measured. If we captured more than 5 animals during a time-constrained search, we limited sampling to 5 randomly selected pygmy salamander locations per site. To determine whether habitat attributes at occupied locations differed from those randomly available in the surrounding environment, we identified a corresponding random location by finding a cover object 2–30 m from each occupied location. We employed a random numbers table to generate a compass bearing (0– 360°) and a distance (2–30 m) from each occupied location to locate cover objects. We considered a cover object to be suitable if it met the minimum size criteria (as described above) and had no pygmy salamander under it. In the event there was no suitable cover object at the random point, we followed the compass bearing until a suitable object was found. Once we selected a cover object, we collected habitat data using the same methods as described above. We collected data at each random location immediately after collecting data at the occupied location. We compared the total counts of pygmy salamanders between the 3 forest types and between Northern Pygmy Salamander and Southern Pygmy Salamander jointly using a negative binomial model for total counts at each site. We obtained the estimated means of pygmy salamanders (i.e., relative abundance) for each forest type and each species by reverse-transforming the estimates given by the model, which utilized a log-link function. We employed the Tukey–Kramer procedure applied to the negative binomial model to perform pairwise comparisons between forest Southeastern Naturalist C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 146 types. To test for differences in microhabitat variables between the 2 species, we combined data for the 3 forest types for each species because of limited sample size. Cover-object size may be an important habitat attribute to Desmognathus species (Keen 1982, Krzysik 1979); thus, we divided DWD into 2 categories: small DWD (width < 6 cm) and large DWD (width ≥ 6 cm). Cutoff between categories was based on the maximum TL of adult pygmy salamanders (5.1 cm; Petranka 1998). We then calculated means of habitat variables for occupied and random locations for each species and compared them using independent-sample t tests with the Satterthwaite approximation of degrees of freedom. We ran conditional logistic regression models with a stepwise procedure to compare habitat variables of occupied and random locations to determine if pygmy salamanders used any habitat attribute differently than was randomly available. We set an alpha of P < 0.05 to retain variables in the model. We used a separate conditional logistic regression model to determine if pygmy salamanders used smaller cover-objects than were randomly available in the surrounding habitat. We considered all statistical comparisons significant at alpha < 0.05. We considered an attribute selected if its use differed from its random availability. We performed all statistical analyses in Statistical Analysis System (SAS, version 9.2; SAS Institute, Cary, NC). Results We captured a total of 98 pygmy salamanders (41 Northern Pygmy Salamanders, 57 Southern Pygmy Salamanders) during 24, two-hour time-constrained searches. The estimated mean number of pygmy salamanders for all forest types combined did not differ between Northern Pygmy Salamander and Southern Pygmy Salamander (χ2 = 1.19, df = 1, P = 0.276). We found 52 pygmy salamanders in spruce–fir forests, 26 in northern hardwood forests, and 20 in cove forests. The estimated mean number of pygmy salamanders was significantly greater in spruce–fir forests (estimated mean = 6.38; 95% CI = 4.20, 9.69) than cove forests (estimated mean = 2.45; 95% CI = 1.43, 4.22; P = 0.017), but was not significantly different between spruce–fir and northern hardwood forests (estimated mean = 3.26; 95% CI = 1.98, 5.36; P = 0.107), or between northern hardwood and cove forests (P = 0.729). In addition to pygmy salamanders, we recorded 655 other salamanders representing 15 species: 173 in spruce–fir forests, 252 in northern hardwood forests, and 230 in cove forests (Appendix 1). Appendix 1 provides the relative abundance of salamanders at Northern Pygmy Salamander and Southern Pygmy Salamander sites in each forest type. We collected habitat data at 73 pygmy salamander locations (36 Northern Pygmy Salamander locations, 37 Southern Pygmy Salamander locations) in 32 spruce–fir forests, 24 northern hardwood forests, and 17 cove forests (elevation range of all sites = 966–1937 m). The sex and age composition of Northern Pygmy Salamander and Southern Pygmy Salamander representing the habitat data were similar between the 2 species (Table 1). The total sex ratio (males:females) of the sample population was 1.04 (24 males, 23 females), with adults comprising 66% (n = 47) of the population and juveniles 34% (n = 24) (Table 1). Summary statistics Southeastern Naturalist 147 C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 of the habitat variables at occupied and random Northern Pygmy Salamander and Southern Pygmy Salamander locations are provided in Tables 2 and 3, respectively. None of the habitat variables differed between occupied locations of Northern Pygmy Salamander and Southern Pygmy Salamander (all P > 0.05, Table 2), except for soil moisture which was significantly greater at Northern Pygmy Salamander locations (P = 0.005, Table 2). There were also no differences in any of the habitat variables between random locations (all P > 0.05, Table 3), except canopy cover, which was significantly less at Northern Pygmy Salamander locations (P = 0.039, Table 3). The stepwise logistic-regression models indicated that the only variables significantly associated with pygmy salamander locations were total area of small DWD (P = 0.0004) followed by total area of large DWD (P = 0.015); for all other variables, P > 0.05, (Table 4). The odds-ratio estimates for total area of small and large DWD were 1.398 and 1.052, respectively. This result indicates that the odds of a pygmy salamander being present at a location increases by factors of 1.398 and 1.052, respectively, as area of small DWD and area of large DWD increase by 100 cm2. Table 2. Summary statistics of habitat variables at occupied locations at historic occurrence sites of Northern Pygmy Salamander and Southern Pygmy Salamander in North Carolina, June–August 2015. Data presented are for occupied locations of Northern Pygmy Salamander (n = 36) and Southern Pygmy Salamander (n = 37) in 32 spruce–fir forests, 24 northern hardwood forests, and 17 cove forests. Historic occurrence sites of Northern Pygmy Southern Pygmy Salamander Salamander (n = 12) (n = 10) Variable Mean SD Mean SD t value df P-value Soil surface temperature (oC) 16.9 1.7 17.3 2.0 -0.51 17.8 0.618 Soil moisture (%) 59.9 12.2 41.9 13.6 3.18 18.2 0.005 Soil pH 6.5 0.4 6.6 0.2 -0.42 14.5 0.683 Distance to water (m) 26.8 7.7 25.5 8.2 0.36 18.7 0.721 Plant cover (%) 19.1 15.8 16.1 13.4 0.48 19.9 0.640 Canopy cover (%) 75.0 11.9 83.4 7.9 -1.96 19.1 0.065 Soil (%) 67.0 33.9 85.1 25.0 -1.43 19.8 0.167 Gravel/cobble (%) 0.0 0.0 2.5 5.6 -1.38 9.0 0.202 Boulder (%) 30.9 35.2 12.5 23.6 1.46 19.2 0.151 Other substrate (%) 2.1 5.8 0.0 0.0 1.24 11.0 0.241 Area of small DWD (cm2) 695.0 417.3 489.0 263.9 1.41 18.8 0.176 Area of large DWD (cm2) 2458.1 5539.7 1580.1 928.7 0.54 11.7 0.599 Table 1. Age and sex composition of Pygmy Salamanders represented in the habitat analysis. Data are missing for 1 Northern Pygmy Salamander and 1 Southern Pygmy Salamander. AttributeA Northern Pygmy Salamander Southern Pygmy Salamander Juvenile 12 12 Adult 23 24 Male 13 11 Female 10 13 Southeastern Naturalist C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 148 Cover objects used by pygmy salamanders consisted of either round wood (n = 59), bark (n = 13), or leaf litter (n = 1). We detected no pygmy salamanders under rocks or other types of cover objects. The size of cover objects used by pygmy salamanders (mean = 384.3 cm2, SD = 476.1) did not differ from those randomly available in the surrounding habitat (mean = 496.0 cm2, SD = 646.9, P = 0.21). Table 4. Results of conditional logistic-regression models with a stepwise procedure comparing habitat variables of occupied and random locations of pygmy salamanders in North Carolina, June– August,2015. Habitat data of Northern Pygmy Salamander (n = 36) and Southern Pygmy Salamander (n = 37) locations were combined for analysis. P-value With area of With area of Variable 1-variable small DWD small and large DWD Soil surface temperature (oC) 0.871 0.348 0.779 Soil moisture (%) 0.011 0.047 0.054 Soil pH 0.209 0.383 0.182 Distance to water (m) 0.950 0.366 0.304 Plant cover (%) 0.021 0.177 0.266 Canopy cover (%) 0.306 0.228 0.574 Soil (%) 0.043 0.055 0.081 Gravel/cobble (%) 0.232 0.345 0.516 Boulder (%) 0.016 0.039 0.067 Other substrate (%) 0.891 0.907 0.476 Area of small DWD (cm2) 0.0004 - - Area of large DWD (cm2) 0.023 0.015 - Table 3. Summary statistics of habitat variables recorded at random locations at historic occurrence sites of Northern Pygmy Salamander and Southern Pygmy Salamander in North Carolina, June–August 2015. Data presented are for random locations of Northern Pygmy Salamander (n = 36) and Southern Pygmy Salamander (n = 37) that correspond to occupied locations in 32 spruce–fir forests, 24 northern hardwood forests, and 17 cove forests. Historic occurrence sites of Northern Pygmy Southern Pygmy Salamander Salamander (n = 12) (n = 10) Variable Mean SD Mean SD t value df P-value Soil surface temperature (oC) 16.9 1.3 17.5 2.1 -0.81 14.9 0.428 Soil moisture (%) 48.5 12.3 42.8 8.8 1.26 19.6 0.224 Soil pH 6.4 0.3 6.5 0.2 -1.04 19.8 0.309 Distance to water (m) 27.2 6.5 25.3 7.8 0.63 17.5 0.539 Plant cover (%) 30.7 19.3 23.5 19.6 0.86 19.2 0.402 Canopy cover (%) 72.1 21.0 86.9 7.2 -2.28 14.0 0.039 Soil (%) 86.1 24.0 87.3 18.2 -0.13 19.9 0.897 Gravel/cobble (%) 2.1 5.0 3.8 8.0 -0.59 14.5 0.565 Boulder (%) 13.5 26.5 6.5 11.6 0.83 15.7 0.417 Other substrate (%) 0.8 2.9 0.5 1.6 0.34 17.6 0.736 Area of small DWD (cm2) 255.7 175.5 291.2 197.0 -0.44 18.3 0.664 Area of large DWD (cm2) 797.0 716.0 1254.9 1084.6 -1.14 15.1 0.271 Southeastern Naturalist 149 C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 Discussion Searches of natural cover-objects have been shown to be a reliable index for absolute population size of terrestrial salamanders (Smith and Petranka 2000). The finding that the overall relative abundance of Northern Pygmy Salamander and Southern Pygmy Salamander did not differ, suggests that the size of the populations at our study sites was comparable, despite the fact that we were unable to find Southern Pygmy Salamander at 2 sites: a low-elevation cove forest and a mid-elevation northern hardwood forest. All sites were similar in that they were relatively intact stands of mature forests. The sex and age composition of Northern Pygmy Salamander and Southern Pygmy Salamander also were similar between the 2 species (Table 1). The total sex ratio (males:females) was 1.04, and the total age ratio (adults:juveniles) was 1.96, in contrast to the findings of Organ (1961b), who reported a total sex ratio of 1.62 and a total age ratio of 0.95 for populations of Northern Pygmy Salamander in the Balsam Mountains of southwestern Virginia. Pygmy salamander populations in our study appear to be relatively vigorous compared to those studied by Organ (1961b) in the late 1950s. We captured twice as many pygmy salamanders in spruce–fir forests than in northern hardwood forests and 2.6 times as many than in cove forests, which suggests that the most robust populations occur in spruce–fir forests. This result agrees with Hairston (1949) and Organ (1961b), who reported higher densities of pygmy salamanders in spruce–fir forests. However, we found that the estimated mean number of pygmy salamanders in spruce–fir forests was not significantly different from that in northern hardwood forests. This finding should be viewed with caution because of the low power of the statistical test due to the small sample size and high variability of the number of animals captured between sites within a forest type. The relative abundance of pygmy salamanders in cove forests was not significantly different from populations in northern hardwood forests; thus, population size of pygmy salamanders in some low-elevation cove forests may be comparable to those in northern hardwood forests. We also observed this pattern in a lowelevation cove site in the Great Smoky Mountains National Park that had the 5th highest capture rate (n = 7) of the 24 sites in our study. Low-elevation populations also may be relatively long-lived and persistent when the habitat remains intact, as exemplified by the successful capture of pygmy salamanders at 7 of the 8 historical low-elevation sites, including a 44-year-old record of Southern Pygmy Salamander in the Great Smoky Mountains National Park (same site as mentioned above). It also is worth noting that the low-elevation site that failed to yield any pygmy salamanders was a Southern Pygmy Salamander site impacted by a heavy infestation of Adelges tsugae Annand (Hemlock Woolly Adelgid). Although no cause and effect can be determined from 1 observation, we suspect that a reduced canopy from the loss of numerous large Eastern Hemlock trees may be the reason for the apparent demise of this population. Currently, the effects of Hemlock Woolly Adelgid on salamander populations are unknown. However, variables that could affect salamanders, including coarse woody debris, herbaceous cover, and light levels, have been shown to significantly increase in Hemlock stands with canopy decline caused Southeastern Naturalist C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 150 by the Hemlock Woolly Adelgid (Cleavitt et al. 2008). Therefore, further study is needed to investigate the impacts of Hemlock Woolly Adelgid on pygmy salamanders in low-elevation cove forests. Strong evidence suggests that predator avoidance and interspecific competition with other salamanders shape the habitat selection of Desmognathus species in the southern Appalachian Mountains (e.g., Bruce 1996, Hairston 1986, Krzysik 1979, Southerland 1986). We recorded 655 salamanders inhabiting the same areas as pygmy salamanders, with the majority either Desmognathus or Plethodon species (Appendix 1). The total number of salamanders observed in each of the 3 forest types far exceeded the total number of pygmy salamanders in each forest type (Appendix 1). These findings suggest that the general habitat of pygmy salamanders is suitable to a variety of salamander species, and that pygmy salamanders are exposed to potential predation by and interspecific competition from other salamanders. The findings that none of the habitat variables differed between Northern Pygmy Salamander and Southern Pygmy Salamander, except for soil moisture, suggests that these 2 species use similar types of microhabitats (Table 2). We are uncertain of the reason why soil moisture levels would be higher at Northern Pygmy Salamander locations compared to Southern Pygmy Salamander locations; perhaps it is a spurious result given the small sample size. Therefore, additional data are needed to validate this result. Nonetheless, our results do indicate that the microhabitat of pygmy salamanders may be characterized by: relatively high soil moisture, neutral soil pH, moderate plant cover, a closed canopy, moderate amounts of exposed soil, and large amounts of small- and large-sized DWD (Table 2). In addition, microhabitat was not associated with surface water; 78% (n = 57) of pygmy salamander locations were located >30 m from seeps or streams. Although analyses of microhabitat associations have not been previously reported for pygmy salamanders, our results generally agree with those of others who have reported that pygmy salamanders utilize relatively moist areas in undisturbed forests (Hairston 1949, King 1936, Tilley and Harrison 1969). In addition, Hairston (1949) and Organ (1961b) both noted that pygmy salamander habitat is often found far from water, and suggested that this is an adaptation to minimize potential predation from larger predatory salamanders that reside closer to water. Our results indicate that total amount of small DWD was the most important habitat variable related to the presence of pygmy salamanders followed by total amount of large DWD (Table 4). The highly significant association of small DWD in the final regression model also highlights the importance of this attribute to pygmy salamanders (Table 4). Cover objects protect salamanders from stressful environmental conditions because they provide a cool, moist refuge from predators and a possible source of invertebrate prey (Heatwole 1962, Jaeger 1980, Mathis 1990). Density of cover objects has been reported as an attractive habitat attribute for several salamander species. For example, Grover (1998) reported that the abundance of Plethodon glutinosus Green (Northern Slimy Salamander) and P. cinereus Green (Eastern Red-backed Salamander) was greater in plots with a high density of Southeastern Naturalist 151 C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 cover objects than in plots with a low density of cover objects. Numerous authors also have suggested that salamanders differentially use cover objects relative to their body length to avoid intra- and interspecific encounters (e.g., Colley et al. 1989, Keen 1982, Kryzysik 1979, Southerland 1986). Therefore, we suggest that pygmy salamanders prefer microhabitats with greater amounts of small DWD to minimize potential encounters with larger salamanders. It also should be noted that pygmy salamanders may select microhabitats with higher soil-moisture levels; soil moisture was significant in the first step of the regression model when not adjusting for other variables (P = 0.011), and was almost significant in the last step of the regression model after adjusting for small and large DWD (P = 0.054) (Table 4). However, additional research is needed to verify this assertion. Pygmy salamanders use a variety of cover objects including logs, rocks, moss, and leaf litter (Caruso 2016, Hairston 1949, King 1936, Organ 1961b, Tilley and Harrison 1969). We found pygmy salamanders almost exclusively under wood cover- objects, consisting of either round wood (81%) or bark (18%). These specimens included 3 gravid females under small round wood and 3 under bark, with 5 located >30 m from water and one 9.7 m from water. Caruso (2016) reported that the majority of cover objects used by Southern Pygmy Salamander in the Great Smoky Mountains National Park consisted of round wood (82%), followed by rocks (14%) and bark (4%). Caruso (2016) also reported that terrestrial Desmognathus species preferred wood cover-objects over rock cover-objects. This preference for wood is likely related to the higher moisture levels maintained under wood cover than rock cover (Grover 1998). The finding that size of cover objects used by pygmy salamanders did not differ from the size of cover objects randomly available suggests that pygmy salamanders showed no preference for small-sized cover objects. However, this finding should be viewed with caution, as observations during our field work suggest that individual pieces of small-sized DWD were much more prevalent across the forest floor than pieces of large-sized DWD. Therefore, any preference pygmy salamanders might have for smaller-sized cover objects may not be apparent in our analysis. However, data from Caruso (2016) indicate that the mean size of cover objects used by Southern Pygmy Salamander were consistently the smallest among 11 species of salamanders in the Great Smoky Mountains National Park. That finding is consistent with those of King (1936) that pygmy salamanders are found most often under small rather than large cover-objects. The results of our study indicate numerous similarities between populations of Northern Pygmy Salamander and Southern Pygmy Salamander among the 3 forest types investigated. Crespi et al. (2010) purported that the environmental niche space between the 2 species is unique because of different climatic conditions on either side of the French Broad River. Our findings that none of the microhabitat variables differed between the 2 species for either the occupied (with the exception of soil moisture, see discussion above; Table 2) or random locations (with the exception of percent canopy cover, both means >72%; Table 3) do not support this assertion, and suggest that the environmental niche space is commensurate between Southeastern Naturalist C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 152 the 2 species. Atypical weather did not likely affect results of our study because average temperatures and total precipitation for August 2015 were comparable to historic averages at several weather stations in the northern and southern study areas (US Climate Data 2017a,b). In view of these findings, we suggest further studies are needed to investigate the effects of incorporating microhabitat data into ecological-niche models that rely solely on climatic data. In addition, more research is needed across different seasons and years to validate our conclusion of similar environmental-niche space between the 2 species. Acknowledgments We thank Bill and Peggy Steiner for providing a scholarship to I. Haas to help fund her participation in this research, C. Lawson and K. Parker for field assistance, and I. Rossell for helpful comments on the manuscript. We also thank K. Weeks and A. Boynton, with the NCWRC’s Wildlife Diversity Program, for administrative and project support. Literature Cited Bruce, R.C. 1977. The Pygmy Salamander, Desmognathus wrighti (Amphibia, Urodela, Plethodontidae), in the Cowee Mountains, North Carolina. Journal of Herpetology 11:246–247. Bruce, R.C. 1996. Life-history perspective of adaptive radiation in desmognathine salamanders. Copeia 4:783–790. Caruso, N.M. 2016. Surface retreats used among 4 genera of terrestrial salamanders in the Great Smoky Mountains National Park. Journal of Herpetology 50: 87–93. Colley, S.A., W.H. Keen, and R.W. Reed. 1989. Effects of adult presence on behavior and microhabitat use of juveniles of a desmognathine salamander. Copeia 1:1–7. Corn, P.S., and R.B. Bury. 1990. Sampling Methods for Terrestrial Amphibians and Reptiles. General Technical Report PNW-GTR-256. US Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland, OR. Cleavitt, A.K., A.K. Eschtruth, J.T. Battles, and T.J. Fahey. 2008. Bryophyte response to Eastern Hemlock decline caused by Hemlock Woolly Adelgid infestation. Journal of the Torrey Botanical Society 135:12–25. Crespi, E.J., L.J. Rissler, and R.A. Browne. 2003. Testing Pleistocene refugia theory: Phylogeographical analysis of Desmognathus wrighti, a high-elevation salamander in the southern Appalachians. Molecular Ecology 12:969–984. Crespi, E.J., R.A. Browne, and L.J. Rissler. 2010. Taxonomic revision of Desmognathus wrighti (Caudata: Plethodontidae). Herpetologica 66:283–295. Grover, M.C. 1998. Influence of cover and moisture on abundance of the terrestrial salamanders Plethodon cinerus and Plethodon glutinosus. Journal of Herpetology 32:489–497. Hairston, N.G. 1949. The local distribution and ecology of the Plethodontid salamanders of the southern Appalachians. Ecological Monographs 19:47–73. Hairston, N.G. 1986. Species packing in Desmognathus salamanders: Experimental demonstration of predation and competition. American Naturalist 127:266–291. Hammerson, G. 2004. Desmognathus wrighti. The IUCN Red List of Threatened Species. Available online at http://dx.doi.org/10.2305/IUCN.UK. 2004. RLTS. T59259A11907504.en. Accessed 18 April 2016. Heatwole, H. 1962. Environmental factors influencing local distribution and activity of the salamander Plethodon cinerus. Ecology 43:460–472. Southeastern Naturalist 153 C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 Jaeger, R.G. 1980. Microhabitats of a terrestrial forest salamander . Copeia 2:265–268. Keen, W.H. 1982. Habitat selection and interspecific competition in 2 species of Plethodontid salamanders. Ecology 63:94–102. King, W. 1936. A new salamander (Desmognathus) from the southern Appalachians. Herpetologica 1:56–60. Krzysik, A.J. 1979. Resource allocation, coexistence, and the niche structure of a streambank salamander community. Ecological Monographs 49:173–194. LeGrand, H.E., Jr., J.A. Ratcliffe, and J.T. Finnegan. 2014. Natural Heritage Program List of the Rare Animal Species of North Carolina. North Carolina Natural Heritage Program, North Carolina Department of Environment and Natural Resources, Raleigh, NC. Mathis, A. 1990. Territoriality in a terrestrial salamander: The influence of resource quality and body size. Behavior 112:162–175. North Carolina Wildlife Resources Commission (NCWRC). 2015. North Carolina Wildlife Action Plan, Raleigh, NC. Organ, J.A. 1961a. Life history of the Pygmy Salamander, Desmognathus wrighti, in Virginia. American Midland Naturalist 66:384–390. Organ, J.A. 1961b. Studies of the local distribution, life history, and population dynamics of the salamander genus Desmognathus in Virginia. Ecological Monographs 31:189–220. Petranka, J.W. 1998. Salamanders of the United States and Canada. Smithsonian Institution Press, Washington, DC. 592 pp. Roble, S.M. 2013. Natural Heritage Resources of Virginia: Rare Animal Species. Natural Heritage Technical Report 13-05. Virginia Department of Conservation and Recreation, Division of Natural Heritage, Richmond, VA. Rosgen, D. 1996. Applied River Morphology. Wildland Hydrology Books, Pagosa Springs, CO. 378 pp. Rossell, C.R., Jr., I.M. Rossell, and S. Patch. 2006. Microhabitat selection by Eastern Box Turtles (Terrapene c. carolina) in a North Carolina mountain wetland. Journal of Herpetology 40:280–284. Rossell, C.R., Jr., P. McNeal, D.P. Gillette, L.A. Williams, S.C. Patch, and A.G. Krebs. 2013. Attributes of shelters selected by Eastern Hellbenders (Cryptobranchus a. alleganiensis) in the French Broad River Basin of North Carolina. Journal of Herpetology 47:66–70. Rossell, C.R., Jr., E. Schwartzman, H.D. Clarke, M. Schultz, and S.C. Patch. 2016. Description of rich montane seeps and effects of Wild Pigs on the plant and salamander assemblages. American Midland Naturalist 175:139–154. Rubin, D. 1971. Desmognathus aenus and D. wrighti on Wayah Bald. Journal of Herpetology 5:66–67. Schafale, M.P. 2012. Guide to the Natural Communities of North Carolina, 4th Approximation. North Carolina Natural Heritage Program, Department of Environment and Natural Resources, Raleigh, NC. Smith, C.K., and J.W. Petranka. 2000. Monitoring terrestrial salamanders: Repeatability and validity of area-constrained cover-object searches. Journal of Herpetology 34:547–557. Southerland, M.T. 1986. Behavioral interactions among 4 species of the salamander genus Desmognathus. Ecology 67:175–181. Tilley, S.G., and J.R. Harrison. 1969. Notes on the distribution of the Pygmy Salamander, Desmognathus wrighti King. Herpetologica 25:178–180. US Climate Data. 2017a. Climate Boone - North Carolina. Available online at https://www. usclimatedata.com/climate/boone/north-carolina/united-states/usnc0072. Accessed 1 August 2017. Southeastern Naturalist C.R. Rossell Jr., Ivy C. Haas, L.A. Williams, and S.C. Patch 2018 Vol. 17, No. 1 154 US Climate Data. 2017b. Climate Brevard - North Carolina. Available online at https:// www.usclimatedata.com/climate/boone/north-carolina/united-states/usnc0076. Accessed 1 August 2017. Withers, D.I. 2009. Tennessee Natural Heritage Program Rare Animals List. Tennessee Department of Environment and Conservation, Division of Natural Areas, Nashville, TN. Appendix 1. Total number of salamanders recorded at historic occurrence sites of Northern Pygmy Salamander and Southern Pygmy Salamander in 3 forest types in North Carolina, June–August 2015. Totals are based on number of animals counted during four 2-h timeconstrained searches in each forest type at Northern Pygmy Salamander and Southern Pygmy Salamander sites. C = cove, NH = northern hardwood, and SF = spruce–fir. Forest Type Salamander species C NH SF Northern Pygmy Salamander Sites Desmognathus organi Crespi, Browne, and Rissler (Northern 7 14 20 Pygmy Salamander) Desmognathus carolinensis Dunn (Carolina Mountain Dusky 72 4 14 Salamander) Plethodon montanus Highton & Peabody (Gray-cheeked Salamander) 54 34 19 Plethodon cylindraceus (Harlan) (White-spotted Slimy Salamander) 3 1 1 Plethodon yonahlossee Dunn (Yonahlossee Salamander) 1 0 0 Notophthalmus v. viridescens (Rafinesque) (Red-spotted Newt) 2 0 0 Eurycea wilderae Dunn (Blue Ridge Two-lined Salamander) 25 4 8 Unknown spp. 0 3 0 Southern Pygmy Salamander Desmognathus wright King (Southern Pygmy Salamander) 13 12 32 Desmognathus fuscus (Rafinesque) (Dusky Salamander) 0 6 0 Desmognathus santeetlah Tilley (Santeetlah Dusky Salamander) 1 8 5 Desmognathus imitator Dunn (Imitator Salamander) 0 1 0 Desmognathus ocoee Nicholls (Ocoee Salamander) 7 8 42 Desmognathus spp. 2 0 18 Plethodon metcalfi Brimley (Southern Gray-cheeked Salamander) 30 156 32 Plethodon jordani Blatchley(Red-cheeked Salamander) 30 4 2 Plethodon teyahalee Hairston (Southern Appalachian Salamander) 1 1 0 Plethodon shermani Stejneger (Red-legged Salamander) 0 6 0 Plethodon serratus Grobman (Southern Redback Salamander) 0 1 0 Eurycea wilderae Dunn (Blue Ridge Two-lined Salamander) 2 2 12