Factors Influencing the Distribution of the Endemic Shorthead Garter Snake (Thamnophis Brachystoma) in Northwestern Pennsylvania

Transcription

1 Indiana University of Pennsylvania Knowledge IUP Theses and Dissertations (All) Factors Influencing the Distribution of the Endemic Shorthead Garter Snake (Thamnophis Brachystoma) in Northwestern Pennsylvania Julie Ellen Mibroda Indiana University of Pennsylvania Follow this and additional works at: Recommended Citation Mibroda, Julie Ellen, "Factors Influencing the Distribution of the Endemic Shorthead Garter Snake (Thamnophis Brachystoma) in Northwestern Pennsylvania" (2014). Theses and Dissertations (All) This Thesis is brought to you for free and open access by Knowledge IUP. It has been accepted for inclusion in Theses and Dissertations (All) by an authorized administrator of Knowledge IUP. For more information, please contact cclouser@iup.edu, sara.parme@iup.edu.

2 FACTORS INFLUENCING THE DISTRIBUTION OF THE ENDEMIC SHORTHEAD GARTER SNAKE (THAMNOPHIS BRACHYSTOMA) IN NORTHWESTERN PENNSYLVANIA A Thesis Submitted to the School of Graduate Studies and Research in Partial Fulfillment of the Requirements for the Degree Master of Science Julie Ellen Mibroda Indiana University of Pennsylvania December 2014

3 2014 Julie Ellen Mibroda All Rights Reserved ii

4 Indiana University of Pennsylvania School of Graduate Studies and Research Department of Biology We hereby approve the thesis of Julie Ellen Mibroda Candidate for the degree of Master of Science 10/22/14 Signature on file Jeffery L. Larkin, Ph.D. Professor of Biology, Advisor 10/22/14 Signature on file Joseph Duchamp, Ph.D. Associate Professor of Biology 10/22/14 Signature on file Josiah Townsend, Ph.D. Assistant Professor of Biology ACCEPTED Signature on file Timothy P. Mack, Ph.D. Dean School of Graduate Studies and Research iii

5 Title: Factors Influencing the Distribution of the Endemic Shorthead Garter Snake (Thamnophis brachystoma) in Northwestern Pennsylvania Author: Julie Ellen Mibroda Thesis Chair: Dr. Jeffery L. Larkin Thesis Committee Members: Dr. Joseph Duchamp Dr. Josiah Townsend The shorthead garter snake (Thamnophis brachystoma) has one of the most restricted ranges of any snake species in the United States, with approximately 90% residing within northwestern Pennsylvania. Although recent surveys indicate that the shorthead garter snake is relatively abundant in certain populations, there is at least some evidence of overall population decline. Furthermore, the specific requirements that may contribute to the restricted range of this species are currently unknown. A better understanding of shorthead garter snake habitat requirements and areas of occupancy is a necessary first step toward the development of an appropriate species-specific conservation strategy. In 2010, I conducted shorthead garter snake occupancy surveys and associated habitat sampling at 40 sites in northwestern Pennsylvania. I used logistic regression to model the relationship between shorthead garter snake occupancy and habitat covariates. I detected shorthead garter snakes at 18 of 40 sites surveyed. Based on the regression analysis, canopy cover (p=0.0242) and distance to water (p=0.0372) were found to be the best predicting factors for shorthead garter snake occupancy (AUC=0.809). Canopy cover was lower at occupied sites (avg=1.4%) than unoccupied sites (avg=10.9%), and distance to water was shorter from occupied sites (avg=397m) than unoccupied sites (avg=1598m). My results suggest that extensive canopy cover at sites with otherwise appropriate conditions may influence shorthead garter snake iv

6 presence, and that the decline of this species may be linked to patterns of landscape-scale regrowth of forests that have occurred throughout much of the species restricted range. v

7 ACKNOWLEDGEMENTS I would like to show my greatest appreciation to my advisor Dr. Jeffery Larkin for the useful comments, remarks, and suggestions through the learning process of this Master s thesis. Without his support and belief in me, this project would not have been possible. I would also like to thank Dr. Art Hulse and Dr. Tim Maret for introducing me to the shorthead garter snake as well for allowing me to increase my knowledge and passion for this species through three years of surveys (what seems like) so long ago. I would like to thank my committee members, Dr. Joseph Duchamp for his statistical genius, willingness to help, and generosity with encouragement, and Dr. Josiah Townsend for his contagious enthusiasm for all things herp, unlimited offering of help and support, and for making me believe that I really can make a living as a herpetologist. Special thanks to Wendy Leuenberger for being my one and only field technician. I would also like to thank each of the faculty members and staff in the biology department who were always interested in my progress and offered countless words of encouragement. I will be forever indebted to my fellow graduate students who always helped when they could, and never let me quit. Very special thanks to DJ McNeil, who sacrificed time, energy, and intellectual elbow grease helping me through the final stages. Funding for this project was provided in part by Indiana University of Pennsylvania. Last but not least, I would like to thank my family and friends who have supported me throughout this entire process. Words cannot express how grateful I am for your love and prayers- thank you all for never giving up on me. vi

8 TABLE OF CONTENTS Chapter Page 1: INTRODUCTION. 1 Literature Review 3 General Description 3 Activity.. 4 Diet. 5 Reproduction... 5 Distribution 6 Restricted Range of an Endemic Species... 7 Habitat Selection 9 Bias of Roadside Surveys : METHODS Study Area Shorthead Garter Snake Surveys. 14 Habitat Surveys Data Analysis : RESULTS Shorthead Garter Snake Surveys. 19 Cover Objects.. 19 Logistic Regression. 21 4: DISCUSSION REFERENCES 31 APPENDICES.39 Appendix A- Snake Species and Total Numbers Detected During Surveys Conducted in Northwestern Pennsylvania in Species in Bold are Listed as Species of Conservation Concern in Pennsylvania. 39 vii

9 LIST OF TABLES Table Page 1 Occupied Shorthead Garter Snake Sites; Total Number Detected Across 3 Sampling Periods; Number of Visits With Detections; and Mean, Standard Deviation (s), Minimum, and Maximum Weight, Snout-Vent length (SVL) and Total Length (ToL) of Individuals Processed During Surveys Conducted in Northwestern Pennsylvania from April-October Shorthead Garter Snake Habitat Variables Quantified at 40 Sites in Northwestern Pennsylvania in Model Selection Results Generated Using Program R to Model Shorthead Garter Snake Occurrence Using Habitat Data Collected From 40 Sites Surveyed in viii

10 LIST OF FIGURES Figure Page 1 Two shorthead garter snakes showing the two extremes in color pattern variations. These snakes were collected in Erie County, PA by Julie Mibroda and Mark Lethaby in Shorthead garter snake site in Dubois, Clearfield County, Pennsylvania 10 3 Shorthead garter snake site along SR 666 in Forest County, Pennsylvania 10 4 Map showing sites that were occupied (red dots) and unoccupied (green dots) by shorthead garter snakes during surveys conducted in northwestern Pennsylvania during April-October Layout of habitat sampling plots located along randomly placed transects at sites monitored for shorthead garter snakes in northwestern Pennsylvania in Habitat features were recorded in 1m 2 plots (black dots; n=30) and 5m radius plots (large gray dots; n=5) ix

11 CHAPTER 1 Introduction Squamate reptiles represent a significant component of most ecosystems (McDiarmid, Guyer, Gibbons, & Chernoff, 2011). The small size and secretive habits of many squamates, particularly snakes, make them some of the least understood vertebrates in terms of their ecological requirements. The lack of information about many snake taxa has also led them to be overlooked in terms of conservation assessment and action. However recent efforts have seen snakes rapidly becoming recognized as priority species for conservation (Böhm et al., 2013; Gibbons et al., 2000; Reading et al., 2010). The shorthead garter snake (Thamnophis brachystoma) is considered a species of conservation concern in Pennsylvania and presently has a state ranking of vulnerable (S3). This species is endemic to northwestern Pennsylvania and adjacent southwestern New York, occurring primarily on the Allegheny High Plateau (Price, 1978). The shorthead garter snake has one of the most restricted ranges of any snake species in the United States, with approximately 90% residing within 14 Pennsylvania counties (Maret, 2008). Although recent surveys indicate that the shorthead garter snake is locally abundant in parts of northwestern Pennsylvania (Maret, 2008), absence from historical locations suggests that there is at least some evidence of overall population decline, apparently due to habitat loss resulting from anthropogenic factors and forest succession (Bothner, 1986). With the majority of the shorthead garter snake s global population residing in Pennsylvania, the Commonwealth has a responsibility to ensure the continued viability of populations and protect key habitats to the point that this vulnerable species is secure and population risks are minimized to the extent feasible. 1

12 The shorthead garter snake is almost always found in close proximity to water, and deep woodlands are anecdotally thought to be avoided (Ernst & Ernst, 2003). A general description of shorthead garter snake habitat includes open areas with low vegetation (e.g., old fields, wetland edges, roadside banks) that provide cover such as rocks, logs, or human litter (e.g., corrugated tin, plywood, tarps) (Ernst & Ernst, 2003; Ernst & Gotte, 1986; Hulse, McCoy, & Censky, 2001; Klingener, 1957). Nonetheless, no species-specific surveys designed to better understand how various cover types and microhabitat characteristics influence shorthead garter snake distribution and abundance have been conducted to date. As such, the specific factors that may contribute to the restricted range of this species are currently unknown. A better understanding of the factors that characterize shorthead garter snake habitat requirements at the landscape and microhabitat scales is a necessary first step toward the development of an appropriate species-specific conservation strategy. I conducted a study to quantify the characteristics of shorthead garter snake microhabitat. Specifically, I used shorthead garter snake presence/absence and habitat surveys to create an occurrence model for the species in northwestern Pennsylvania. I hypothesized that there were habitat variables that discriminated between sites that were occupied and unoccupied by the shorthead garter snake. Based on anecdotal accounts (Ernst & Ernst, 2003; Ernst & Gotte, 1986; Hulse et al., 2001; Klingener, 1957), I predicted that the discriminating variables would include type and availability of cover objects, tree canopy cover, and proximity to water. Information obtained from this study provides a valuable first step toward developing a science-based conservation strategy for this priority species. 2

13 Literature Review General Description The shorthead garter snake is a small to medium-sized species, the maximum reported size for the species is 559 mm total length (Conant & Collins, 1991). The dorsal background coloration is usually olive to olive brown, and the dorsal pattern generally consists of one vertebral and two lateral stripes. The vertebral stripe varies from beige/yellowish to nearly the same olive brown as the background coloration (Figure 1). The lateral stripes are dull buff and not as distinctly set off from the background as the beige/yellowish vertebral stripe. All dorsal stripes are bordered by a series of small black dashes. The venter of all specimens is immaculate and is usually a shade of tan. The head is generally the same color as the body, and the chin and throat vary from creamcolored to yellow and are immaculate. The head of shorthead garter snakes is not distinct from the body and appears to be too small for the body. The dorsal scales are keeled and occur in 17 rows (occasionally 19) at midbody (Hulse et al., 2001). The anal plate is entire, and the subcaudals are in two rows. Shorthead garter snakes exhibit significant sexual dimorphism. Hulse et al. (2001) reports that the average snout-vent length (SVL) and total length (ToL) of adult females is greater than those of adult males that occur in Pennsylvania, with adult males averaging 273.1mm SVL and 366.8mm ToL, and mature females averaging 325.3mm SVL and 418mm ToL. Hulse et al. (2001) also found that males have significantly longer tails than females, reporting that the mean tail length for males is 34.8% of SVL and 25.6% of ToL, and the mean tail length for females is 29.7% of SVL and 23% of ToL. 3

14 Figure 1. Two shorthead garter snakes showing the two extremes in color pattern variations. These snakes were collected in Erie County, PA by Julie Mibroda and Mark Lethaby in Activity Seasonal activity of the shorthead garter snake starts as early as the beginning of April in Pennsylvania, but snakes do not become routinely active until the second half of the month (Hulse et al., 2001). Activity levels remain fairly constant through May, and then reach a peak in mid-june. Activity declines dramatically in July, exhibits a second peak in August, probably as a function of parturition, and then gradually decreases through September and October (Hulse et al., 2001). All shorthead garter snakes enter hibernation by the beginning of November. Bothner (1963) reported a hibernaculum in Cattaraugus County, NY, that was located on a west-facing shale bank located approximately 60 m from a stream. The opening to the hibernaculum was about one inch 4

15 wide and one inch high, and snakes were found wedged into cracks and crevices of the rocks, as opposed to a den, at depths between 50cm and 115cm (Bothner, 1963). Bothner (1963) also found red-spotted newts (Notophthalmus viridescens), spotted salamanders (Ambystoma maculatum), and northern red-bellied snakes (Storeria occipitomaculata) using the same hibernaculum. Diet It has been shown previously that in the wild, shorthead garter snakes feed exclusively on earthworms (Asplund, 1963; Ernst & Ernst, 2003). Several authors have suggested other potential food items, including fish, insects, leeches, frogs, and salamanders (Rossi, 1993; Sweeny, 1992; Tennant, 2003). Gray (2008) offered these items to two captive shorthead garter snakes and found that they attempted to eat and/or consumed leeches, fish (both live guppies and yellow perch fillet), slugs, salamanders, and frogs. His findings suggest that shorthead garter snakes may feed more opportunistically than previously reported, and that their diet may more closely resemble that of the closely related species Thamnophis butleri, which are known to consume amphibians and fish, in addition to earthworms (Gray, 2008). Reproduction Shorthead garter snake courtship and mating occur in the spring, shortly after emergence from hibernation, while males and females are in close proximity (Ernst & Gotte, 1986; Hulse et al., 2001; Pisani, 1967). In Pennsylvania, females usually emerge from hibernation with moderately well-developed ovarian follicles (Hulse et al., 2001). Sperm is apparently stored by females in seminal receptacles in the oviduct from mating in April until the eggs are ovulated from the end of May through early June (Hulse et al., 5

16 2001; Pisani & Bothner, 1970). Shorthead garter snakes are viviparous, and by late June, well-formed embryos are present in the oviducts (Hulse et al., 2001). Parturition occurs in August; litter size ranges from 5 to 14 young and is positively correlated with female body size (Hulse et al., 2001). Neonates average 111.3mm SVL and 148.5mm ToL (Hulse et al., 2001). Males emerge from hibernation with small testes. Testicular growth is rapid during the spring and early summer, and by mid-july the testes are at a maximum size (Hulse et al., 2001). At this time, the testes begin to regress and are very small when the males enter hibernation in November. While the testes are regressing, mature sperm are evacuated into the vas deferens, where they are stored through the winter until the spring mating season (Hulse et al., 2001). Both males and females mature in their second or third season. Males mature at about 200mm SVL and 261mm ToL, and females mature at about 273mm SVL and 350mm ToL (Hulse et al., 2001). Distribution The shorthead garter snake occurs primarily in the unglaciated portion of the Allegheny Plateau (Figure 2). Its natural range extends from the southern-tier counties of New York southward into the unglaciated area of northwestern Pennsylvania. Its occurrence in the urban areas of Allegheny, Butler, and Erie Counties, Pennsylvania, is the result of either intentional or unintentional human introductions (Hulse et al., 2001). 6

17 Restricted Range of an Endemic Species The limited range of the shorthead garter snake may be characterized as a relict distribution that is shaped by the evolutionary and biogeographic history of the species and is also evident in the distributions of its two closest relatives (de Queiroz, Lawson, and Lemos-Espinal, 2002): the plains garter snake (Thamnophis radix) and Butler s garter snake (T. butleri). Thamnophis radix is listed as endangered in Ohio, occurring only in disjunct populations in relict prairies in Ohio, Missouri, Indiana, and Arkansas (Dalrymple & Reichenbach, 1981; Placyk et al., 2012). Although the range of T. butleri covers most of the glaciated Great Lakes Region in Michigan, Ohio, and Indiana, it occurs only in isolated colonies (Carpenter, 1952; Placyk et al. 2012). Butler s garter snakes in particular are relatively similar to shorthead garter snakes ecologically; having limited distributions, similar size and physical appearance, similar diet, and very similar natural histories (Ernst & Ernst, 2003). In addition to evolutionary and historical biogeographic influences, several factors may influence the distribution of the shorthead garter snake. Asplund (1963) investigated the feeding habits, marked individuals, and body temperatures of shorthead garter snakes, in order to identify possible factors that limited the species distribution. Asplund (1963) found that earthworms comprised 100% of the shorthead garter snake diet, noting that prey species such as slugs or small salamanders may have been digested beyond recognition. He argued that, although a strict earthworm diet may locally limit the snake s distribution, it was unlikely to restrict the range of the shorthead garter snake (Asplund, 1963). Contrary to this idea, Gray (2008) has more recently argued that the range overlap between native earthworms and the shorthead garter snake warrants 7

18 examination of this possible relationship. Through his investigation into the diet of T. brachystoma, Gray (2008) found that captive shorthead garter snakes attempted to eat other prey items, such as leeches, fish, slugs, salamanders, and frogs. He noted, however, that in the wild, non-native earthworms were likely to be abundant, widespread, and more easily obtained than leeches, small amphibians, and fish, perhaps resulting in a shift of the diet of T. brachystoma from more varied to almost exclusively earthworms (Gray, 2008). Considering the expansive range of non-native earthworms, Gray (2008) speculates that the almost exclusive earthworm diet of the shorthead garter snake should expand the snake s range into glaciated regions, and therefore does not explain the restricted range of the species. When considering the role of diet in the restricted range of shorthead garter snakes, authors have also studied potential competition for prey. Asplund (1963) suggested that competition from other worm-eating snakes may limit the range of the shorthead garter snake, reporting that species like common garter snake (Thamnophis sirtalis) and northern brown snake (Storeria dekayi) were rarely found to coexist with shorthead garter snakes. Supporting this hypothesis, Gray (2008) reported that in Erie County, Pennsylvania, T. brachystoma appeared to be more abundant in areas where those two other species were less abundant. Finally, Asplund (1963) hypothesized those ambient temperatures which are suitable for shorthead garter snake activity, may have the greatest influence on their distribution. He used body temperature data collected from 128 individuals (mean = 30 C) to imply that the species seems adapted to relatively cooler temperatures, thus limiting its range. Asplund s (1963) study provides relevant information to 8

19 understanding the ecology of the shorthead garter snake, and encourages more questions concerning its distribution and habitat use. Habitat Selection Previous studies have described the general habitat characteristics of shorthead garter snakes (Ernst & Ernst, 2003; Ernst & Gotte, 1986; Hulse et al., 2001; Klingener, 1957). It has been suggested that the shorthead garter snake prefers low herbaceous cover, such as that found in old fields, and is thought to avoid interior forest (Ernst & Gotte, 1986; Wozniak & Bothner, 1966). Shorthead garter snakes are commonly found in open areas where road or building construction has resulted in disturbed habitat, or along the banks of waterways, when those locations also provide suitable cover (i.e., rocks) (Figure 2). When evaluating shorthead garter snake habitat selection and use, it is important to note the findings of Klingener (1957) from his marking study of shorthead garter snakes in Crawford County, PA. He found that a high percentage (35.7%) of marked individuals were recovered during his sampling periods, concluding that the members of that local population were regularly using the available cover (a 2 x 7 sheet of corrugated galvanize iron) as a home site or refuge (Klingener, 1957). Klingener (1957) reported that the high recovery rate also indicated that the snakes were using the iron sheet to escape the heat, and that moisture and food was available in quantity, even in July/August when the study occurred. He also noted that an increase in movement and distance traveled that might exist at mating time, when individuals are searching for mates, would be absent during the mid-summer study period (Klingener, 1957). 9

20 Wozniak and Bothner (1966) found that shorthead garter snake distribution did not correlate with altitude, slope or soil type. Cover, however, appeared to be a significant part of shorthead garter snake habitat, as all specimens collected during their study were discovered under some sort of cover object, including stones, logs, boards, and tin. (Wozniak & Bothner, 1966). Figure 2. Shorthead garter snake site in Dubois, Clearfield County, Pennsylvania. Figure 3. Shorthead garter snake site along SR 666 in Forest County, Pennsylvania. 10

21 Bias of Roadside Surveys One major objective to understanding specific habitat characteristics that may predict limits of shorthead garter snake occupancy is to identify the species current distribution. There are several factors that limit the accuracy of species distribution models (Guisan & Thuiller, 2005), including the spatial bias of data frequently used to develop models (McCarthy, Fletcher, Rota, & Hutto, 2011). Possibly the most common reason for this prejudice is that surveys are often conducted near roads for ease of access (Kadmon, Farber, & Danin, 2004; Weir & Mossman, 2005). McCarthy et al. (2011) predicted species distributions of 15 bird species on the basis of a) roadside, and b) offroad point-count data. When the occupancy models generated from the data of each were compared, the results suggested that sampling along unpaved roads can be an impartial source of information for bird species distribution models (McCarthy et al., 2011). Thirty-six of 40 sites (90%) included in this study were located along paved roads, justifying further investigation into the accuracy of shorthead garter snake distribution using roadside surveys. From a conservation perspective, the use of rocky roadside habitat patches by this species may prove to be logistically appealing for both managing habitat and monitoring populations. 11

22 CHAPTER 2 Methods Study Area This study was conducted in eight counties of northwestern Pennsylvania: Erie, Warren, McKean, Forest, Elk, Clearfield, Venango, and Jefferson (Figure 3). The shorthead garter snake is endemic primarily to the Northern Unglaciated Allegheny Plateau, with populations also occurring within the Southern Unglaciated Allegheny Plateau, Western Glaciated Allegheny Plateau, and Erie and Ontario Lake Plain (USFS, 1993). The landscapes of these ecoregions include extensively forested rounded hills, low mountains, and narrow valleys. I selected 40 sites for inclusion in this study. Study sites ranged in size from ha and all were within the historic range of the shorthead garter snake (Hulse et al., 2001). Sites were chosen based on results from previous herpetological surveys (Maret, 2008) or because of their potential to support shorthead garter snakes (e.g., appearance of appropriate habitat). A site was considered suitable for inclusion in this study when at least one rock measuring at least approximately 8 x 12 was clearly visible, and when turned, was not buried deep in the ground (allowed sufficient space for snakes to use as cover). Most (n = 35) of the study sites were located on public land (i.e., roadside right of ways, Allegheny National Forest, state parks) with the five remaining sites on private land. Thirty-six of the study sites were located along paved roadsides and were associated with disturbances created by road construction/maintenance. Surveyed roadside banks had rock cover that varied from covering the entire site to one or two 12

23 Figure 4. Map showing sites that were occupied (red dots) and unoccupied (green dots) by shorthead garter snakes during surveys conducted in northwestern Pennsylvania during April-October

24 single rocks scattered across the site. Two non-roadside sites were located along creek banks. Another non-roadside site was a rocky bank along an unpaved gas well road, and the final non-roadside site was along an old fence row in SB Elliott State Park. Shorthead garter snakes were previously detected during surveys conducted in , at 16 of the 40 sites surveyed in this study (Maret, 2008). Mean annual temperatures in 2010 during May, June, and September/October in northwestern Pennsylvania were 12.8 C, 17.2 C, and 13.9/7.8 C respectively ( accessed 12 June 2014). Mean annual precipitation for May, June, and September/October 2010 in northwestern Pennsylvania were 0.41 cm, 0.36 cm, and 0.30/0.23 cm respectively ( accessed 12 June 2014). Shorthead Garter Snake Surveys All methods used in this study followed the Guidelines for Use of Live Amphibians and Reptiles in Field and Laboratory Research comprised by the Herpetological Animal Care and Use Committee (Beaupre et al. 2004). This study was approved by the Institutional Animal Care and Use Committee (IACUC Permit # ), and the Pennsylvania Fish & Boat Commission (Scientific Collector Permit Number 96, Type 1). Each site was surveyed three times, once each during three different sampling periods: (a) May, (b) June, and (c) September/October. July surveys were avoided due to the potential of low detection probability while shorthead garter snake activity levels are reported to dramatically decline (Hulse et al., 2001). During each visit, ambient air temperature and general weather condition (cloudy, sunny, etc.) was recorded. Cover objects within each site were turned, and then replaced to their original position whenever possible. When a shorthead garter snake was found, it was captured, weighed, snout-vent 15

25 and total length was measured, observed for overall condition, photographed, and all were released immediately after processing. I measured the length, width, and thickness of rocks under which shorthead garter snakes were found. All other reptile and amphibian species found were recorded, and the occurrence and identity of multiple snakes under one rock was noted. Surveys were time constrained to ensure equal effort across sites. Habitat Surveys I conducted habitat surveys at all 40 sites where shorthead garter snakes were surveyed. Habitat surveys were conducted in September and early October I quantified 10 habitat features for each site: percent rock, percent bare ground, percent herbaceous cover, percent course woody debris (CWD), number of cover rocks, number of shrubs, number of saplings, number of trees, percent tree canopy cover; and distance to permanent water source. I estimated percent ground covered by rocks, bare ground, herbaceous (i.e., ferns, grass, forbs), CWD, and frequency of cover rocks within >30 plots measuring 1m 2 and spaced every 4 m along randomly placed transects at each study site (Figure 4). For number of cover rocks, I only considered rocks that measured at least the minimum length and width, and no greater than the maximum thickness of those under which I observed shorthead garter snakes ( 11 cm long x 11 cm wide x 9 cm thick). For CWD, I only considered objects that were > 2.5 cm diameter (Harmon et al., 1986). I also estimated percent tree canopy cover directly above each 1m² plot using a densitometer. I quantified number of shrubs, number of saplings (<10 cm dbh), and trees ( 10 cm dbh) within five 5m radius plots that were randomly distributed across each site. I used ArcGIS 10.1 and land cover data (National Agriculture Imagery Program [NAIP]) 16

26 to measure the distance between the center of each site and a permanent water source (i.e., stream, pond, lake). Figure 5. Layout of habitat sampling plots located along randomly placed transects at sites monitored for shorthead garter snakes in northwestern Pennsylvania in Habitat features were recorded in 1m 2 plots (black dots; n 30) and 5 m radius plots (large gray dots; n = 5). Data Analysis I developed logistic regression models to identify which habitat characteristics best predicted shorthead garter snake occupancy. I modeled species occurrence with the general linear model (glm) function in Program R v (R Development Core Team, Vienna, Austria; 2011) to generate logistic regression models. I analyzed species occurrence among study sites using shorthead garter snake presence/absence data for each site along with 10 habitat variables thought to potentially predict occupancy: rock cover (%), bare ground (%), herbaceous vegetation (%), coarse woody debris (%), tree canopy cover (%), frequency of cover rock, number of shrubs, number of saplings, number of trees, and the distance from each site to water. To prevent over-parameterized results, models were limited to a maximum of 2 variables. Multicollinearity was avoided by removing highly correlated variables, preventing a violation of the assumption that the variables included are reasonably 17

27 independent. To compare models while accounting for small sample size bias, Akaike s Information Criterion corrected for small sample size (AICc) was used. All models were ranked according to AICc values; the lowest AICc value (restricted to models containing 1-2 variables for this study) was considered the best combination of variables (Burnham & Anderson, 2002). Models within the range of 2 AICc of the best model were considered competing models (Burnham & Anderson, 2002). I evaluated the discriminatory power of each competing model containing 2 variables using a receiver operating characteristic (ROC) curve and tested for overdispersion. I tested for overdispersion by calculating the variance inflation factor (residual deviance over the residual degrees of freedom) (ĉ) on the most parameterized model in the competing set. Overdispersion occurs when the variance exceeds what is expected with the assumption of a binomial distribution and if the value ĉ produced is greater than 1 (MacKenzie et al., 2006). ROC curves were calculated for all competing models containing 2 variables. The area under the curve (AUC) of 1 would indicate a model perfectly predicted sites used by shorthead garter snakes, whereas a value of 0.5 for AUC indicated the model was equivalent to random guessing (Fielding & Bell, 1997). 18

28 CHAPTER 3 Results Shorthead Garter Snake Surveys In 2010, I detected a total of 99 individual shorthead garter snakes across 18 of 40 sites surveyed during three sampling periods. Sixteen of 18 occupied sites were confirmed to have retained occupancy of the species in 2010 from surveys conducted by the PABS in (Maret, 2008). The remaining two occupied sites were new local records for the species. Shorthead garter snakes were detected during one of three sampling periods (33% detection rate) at one site, two of three (67% detection rate) at 12 sites, and three of three (100% detection rate) at five sites (Table 1). The average mass of all captured individuals that were weighed was 17.5 g (3 32 g, SD = 4.5), the average SVL of all measured individuals was 28.9 cm (16 37 cm, SD = 2.3), and the average total length of all measured individuals was 36.7 cm (19 45 cm, SD = 3.1; Table 1). In addition to shorthead garter snakes, I detected at least one individual of seven additional snake species while conducting surveys (see Appendix A). Cover Objects A total of 92 rocks, under which shorthead garter snakes were found, were measured during the 3 sampling periods. The average length was 33.2 cm (min = 11 cm, max = 65 cm, s = 38.2), average width was 23.9 cm (min = 11 cm, max = 55 cm, s = 31.1), and average thickness was 4.0 cm (min = 1 cm, max = 9 cm, s = 5.7). On three occasions, shorthead garter snakes were found under drift fence that had been placed along ditches and near construction sites to prevent erosion. 19

29 Table 1. Occupied Shorthead Garter Snake Sites; Total Number Detected Across 3 Sampling Periods; Number of Visits with Detections; and Mean, Standard Deviation (s), Minimum, and Maximum Weight, Snout-Vent Length (SVL) and Total Length (ToL) of Individuals Processed During Surveys Conducted in Northwestern Pennsylvania from April-October Site # Detections over Visits w/ Weight (g) SVL (cm) ToL (cm) ID 3 periods detections Mean (SD) Min Max Mean (SD) Min Max Mean (SD) Min Max ª Avg ª - Only one of the two snakes detected at this site was weighed and measured 20

30 Logistic Regression I included habitat data from 40 sites in the habitat occupancy analysis. Of the 10 variables included in the analyses (Table 2), one variable (percent herbaceous vegetation cover) was highly correlated with the remaining 9 variables, and was therefore not included in the final analysis. The regression analysis of habitat characteristics resulted in ten models consisting of combinations of six variables (Table 3). Additionally, substantial model overdispersion did not occur because the ĉ was not greater much greater than one (ĉ = 1.138). The best model had no closely competing models and showed a substantial ability to distinguish occupied sites from unoccupied sites when compared to the data (AUC = 0.809). Model results demonstrated that sites occupied by shorthead garter snakes had less percent canopy cover (r = -1.04, SE = 0.46, p = 0.024) and were closer to water (r = -3.19, SE = 1.53, p = 0.037) than unoccupied sites (Table 2). Frequency of cover rock was also included in 3 of 4 top models (Table 3). Frequency of cover rock was higher at occupied sites (mean = 2.56 ± 0.81, 95CI = ) than unoccupied sites (mean = 1.82 ± 0.59, 95CI = ). Within the top 10 ranking models, the relative importance value (RI) of frequency of cover rock was 0.216, suggesting that frequency of cover rock could be an important feature of shorthead garter snake habitat. 21

31 Table 2. Shorthead Garter Snake Habitat Variables Collected at 40 Sites in Northwestern Pennsylvania in Occupied Unoccupied Habitat Characteristic Mean (SD) Min Max Mean (SD) Min Max Rock (%)ª Bare ground (%)ª Vegetation (%)ª Coarse woody debris (%)ª Canopy cover (%)ª Frequency of cover rockª No. of shrubs* No. of saplings* No. of trees* Distance to water (m) ª - Characteristic collected within 1m² plots *- Characteristic collected within 5m radius plots 22

32 Table 3. Model Selection Results Generated Using Program R to Model Shorthead Garter Snake Occurrence Using Habitat Data Collected From 40 Sites Surveyed in Models Were Considered Competing When AICc Model Model Variables AICc AICc Model Weight 1 Canopy Cover Distance to Water Freq of Cover Rock Distance to Water Freq of Cover Rock Number of Trees Canopy Cover Freq of Cover Rock Percent Rock Number of Trees Percent Rock Canopy Cover Number of Saplings Distance to Water Number of Shrubs Distance to Water Canopy Cover Number of Trees Percent Rock Distance to Water

33 CHAPTER 4 Discussion It is well-established that habitat loss and fragmentation are the most important drivers behind many wildlife population declines (Bender, Contreras, & Fahrig, 1998; Fahrig, 1997; Huxel & Hastings, 1999). This trend is repeated across an enormous breadth of taxa with mammals (Crooks, 2002; Maehr, 1990), birds (Conner & Rudolph, 1991; Lamberson, McKelvey, Noon, & Voss, 1992), amphibians (Dodd & Smith, 2002; Lehtinen, Galatowitsch, & Tester, 1999), insects (Bommarco et al., 2010; Saccheri, 1998), and plants (Helm, Hanski, & Pӓrtel, 2006). Indeed, the decline of many snake populations can be attributed to habitat loss and degradation, largely due to human development (Gibbons et al., 2000; Reading et al., 2010; Webb & Shine, 2000). An understanding of the habitat-requirements for snake species of conservation concern (e.g., shorthead garter snake) is critical to their conservation. My study is the first to empirically demonstrate the influence of canopy cover and distance to water on shorthead garter snake occurrence. Specifically, shorthead garter snakes were observed at sites with lower canopy cover and that were closer to water than sites where they were not detected. My results agree with reports that shorthead garter snakes typically inhabit open landscapes (Ernst & Gotte, 1986; Wozniak & Bothner, 1966), and are commonly found within a few hundred meters of water (Hulse et al., 2001). Shorthead garter snakes were surveyed in northwestern Pennsylvania in by the Pennsylvania Biological Survey (PABS) to examine the distribution and population dynamics of the species. Maret (2008) reported that shorthead garter snakes, while uncommon across most of the landscape, demonstrate the capacity to occur in high, localized abundances. My abundance estimates from 2010 support this idea since, although shorthead garter snakes 24

34 were not the most abundant (individuals/site) snake species in my study area, when present, shorthead garter snakes were frequently found at relatively high densities with respect to those of other snake species (see Appendix A). The average maximum number of shorthead garter snakes observed on a single survey (1.07/occupied site) exceeded all other common species, with northern red-bellied snakes (Storeria occipitomaculata) and eastern garter snakes (T. sirtalis) occurring at average densities of 0.61 and 0.56 snakes/occupied site, respectively. Moreover, all the sites where shorthhead garter snakes were detected during the PABS that I resurveyed in 2010 retained shorthead garter snakes. This finding suggests that shorthead garter snake populations are stable across these sites; however, a monitoring program that evaluates the long-term population trends at a set of sites across the species range would better assess this dynamic. Shorthead garter snake metapopulation persistence is likely reliant on attributes that affect habitat quality. Canopy cover was the habitat attribute that best predicted the presence of shorthead garter snakes at my study sites, with canopy cover being significantly lower (p=0.024) at sites where shorthead garter snakes were detected compared to sites where the species appeared to be absent. Canopy cover has been cited as an important factor in affecting habitat quality for several reptile species (Pike, Webb, & Shine, 2011; Pringle, Webb, & Shine, 2003; Webb, Shine, & Pringle, 2005). Specifically, decreased canopy cover has been shown to be positively correlated with reptile abundance and diversity (Nicoletto, 2013; Pike et al., 2011). Indeed, other conservation priority reptiles that occur in Pennsylvania, such as wood turtles (Glyptemys insculpta), Eastern massasauga rattlesnakes (Sistrurus catenatus), and Eastern hognose snakes (Heterodon platirhinos) are known to be heavily reliant on canopy openings (Compton, Rhymer, & McCullough, 2002; Harvey & Weatherhead, 2006; Lagory et al., 2009; 25

35 Moore & Gillingham, 2006). It is important to note that this study only examined diurnal habitat use of shorthead garter snake during portions of the active season. The species may utilize other habitat types (e.g., closed-canopy forest) during other parts of the day or phases of its lifecycle. A comparison of radio-tagged and non-radio tagged pine snakes (Pituophis melanoleucus) in New Jersey revealed that radio-tagged individuals spent equal amounts of time in forested and disturbed habitats, whereas non-radio tracked snakes were found over 90% of the time in disturbed habitats (Burger & Zappalorti, 1988). The relationship between shorthead garter snake occurrence and canopy cover may be driven by several aspects of the species biology including thermoregulation, insect prey availability, and concealment from predators (Dugay, Wood, & Miller, 2000; Pike et al., 2011; Pringle et al., 2003; Schowalter, Webb, & Crossley, 1981). Although this relationship was not the focus of my study, my findings contribute to a growing body of literature that identify the need to maintain adequate amounts of open/early-successional habitats as a component of forested landscapes to benefit many species of wildlife (Gilbart, 2011). As terrestrial ectotherms, snakes regulate body temperature by exploiting spatial and temporal variation in microclimates. Open canopies allow sunlight penetration, providing relatively warm microenvironments at ground level that shorthead garter snakes can utilize to avoid low critical body temperature. Open-canopy habitats also increase refuge availability in the form of dense herbaceous vegetation and coarse woody debris typically associated with canopy disturbance/removal. Additionally, macroarthropod abundance has been shown to increase in clearcut forest ecosystems (Dugay et al., 2000; Schowalter et al., 1981). Although captive shorthead garter snakes have shown no interest in consuming crickets (Gray, 2008), the species may 26

36 opportunistically feed on other types of invertebrates that are more readily available in opencanopy communities. Distance to water was the second habitat feature that my analysis revealed to be an important predictor of shorthead garter snake presence in the sites monitored during my study. Specifically, shorthead garter snakes were more likely to occur at habitat patches that were closer to a permanent water source. Sites occupied and unoccupied by shorthead garter snakes in my study were on average 397 m and 1598 m from a permanent water source (Table 2). A possible explanation of this observation may be related to differences in the availability of shorthead garter snake prey across a soil moisture gradient. Earthworm populations, a major prey item for shorthead garter snakes, are known to be effected by soil moisture (Hallatt et al. 1992). Soil moisture is generally considered to be an important environmental factor that influences earthworm growth and reproduction (Hallatt, Viljoen, & Reinecke, 1992). Future research should focus on quantifying shorthead garter snake home range size and movement in relation to distance to water sources and seasonal changes in soil moisture across a home range. Despite our gap in knowledge regarding the causal mechanism behind the positive relationship between shorthead garter snakes and distance to water, I suggest that conservation efforts intended to create or enhance shorthead garter snake habitat should target areas close (<400 m) to a permanent water source. Previous investigators have noted an apparent close association between shorthead garter snake presence and the availability of cover objects (Hulse et al., 2001). Numerous studies have demonstrated that habitat attributes (e.g., cover availability) strongly affect the survival of many snake species (Beck & Jennings, 2003; Huey, 1991; Shine, Webb, Fitzgerald, & Sumner, 1998). Cover objects may serve as shelter from predators/extreme temperatures, habitat for prey, or den 27

37 sites/hibernacula (Huey, Peterson, Arnold, & Porter, 1989). Huey et al. (1989) concluded that ectotherms can sometimes thermoregulate more effectively in retreats (e.g., under rocks and in burrows) than in the open. Although Ernst and Barbour (1989) reported that shorthead garter snakes frequently basked conspicuously during warm months, all individuals of this species that I observed were found beneath cover objects. Moreover, the recurrent use of cover objects by shorthead garter snakes has been documented by other biologists (Hulse et al., 2001; M. Lethaby, personal communication, March 2009). Despite the apparent importance of cover to shorthead garter snakes, my study did not find the frequency of cover rock to be the most important difference between occupied and unoccupied sites. However, this result is not surprising given the fact that all the sites included in my study were in part selected based on the presence of large cover rocks. As such, the frequency of cover rock varied only slightly among any sites regardless of occupancy by shorthead garter snakes. Thus, the models were unable to more strongly reflect the potential importance of cover rocks to shorthead garter snake presence. With that said, the variable frequency of cover rock was included in 3 of the 4 top models (Table 3), supporting the notion that the availability of cover objects could be an important characteristic in shorthead garter snake habitat. Research designed to specifically quantify the importance of cover rocks to shorthead garter snakes presence and abundance is warranted. My study empirically describes some of the important habitat features of the shorthead garter snake. Based on my model, open communities with limited canopy (<10%), close proximity to water (<400 m), and abundant rocky cover would constitute optimal habitat for this species. As such, these areas should be considered high-priorities for conservation within the limited range of the species. One of the most serious threats to shorthead garter snake habitat may actually be natural ecological succession. The manipulation of canopy cover through the 28

38 removal of shrubs, saplings, and tree at sites with sufficient rocky cover and in close proximity to water may be an effective strategy to promote the stability of shorthead garter snake populations (Webb et al., 2005). An alternative to manipulating canopy cover, albeit potentially more costly, would be to create new rocky habitat patches in existing open areas near water. The use of restored or human-created rocky habitats by reptiles has been reported elsewhere (Webb & Shine, 2000). Regardless of the management technique increase habitat availability, sites should be closely monitored to evaluate shorthead garter snake response and to modify management prescription should it be warranted. The results of my study represent an important first step to understanding factors that limit the distribution of the shorthead garter snake. The findings presented here also provide insight to the potential direction of future studies that are necessary to better understand the resource needs of the shorthead garter snake. Future studies should use repeat surveys to estimate detection probability, occupancy, and abundance for the shorthead garter snake in various habitats. Additionally, studies should also quantify home range size and movement patterns of the shorthead garter snake to more completely characterize its habitat requisites. A thorough evaluation of snake community assemblages in patches occupied and unoccupied by shorthead garter snakes is also a topic worthy of investigation. Such a study may reveal potential competitor-mediated factors that influence the species limited distribution. Finally, given the restricted range and apparent patchy distribution of the shorthead garter snake, genetic analyses will reveal the extent of genetic diversity within the species and among subpopulations, quantify gene flow, and identify key landscape features that encourage connectivity among habitat patches. The completion of the above mentioned studies and others will be necessary to develop 29