TEXAS TECH UNIVERSITY Natural Science Research Laboratory Occasional Papers Museum of Texas Tech University Number 280 6 November 2008 Bat Diversity and Activity: A Comparison Among Texas Army National Guard Sites K. C. Loren Ammerman, RogelioM. Rodriguez, Robert Dowler, and Molly McDonough Abstract Five Texas Army National Guard training sites (Camp Maxey, Fort Wolters, Camp Swift, Camp Bowie, and Camp Mabry) were surveyed for bats using mist nets and Anabat units dur¬ ing spring, summer, and fall from October 2005 to November 2006. A total of seven species, Lasiurus borealis, L. cinereus, L. seminolus, Myotis \e lifer, Nycticeius humeralis, Perimyotis subflavus, and Tadarida brasiliensis, were documented across all five sites. Based on mist net captures, Camp Maxey had the highest species diversity (five species documented) whereas Camp Swift and Camp Mabry had the lowest (one species documented at each site). The capture of L. seminolus and L. cinereus represent county records for Lamar County (Camp Maxey) and the capture of T. brasiliensis was a county record for Parker County (Fort Wolters). Species occurrence was also recorded at each site using acoustic monitoring. Canonical correspondence analysis of acoustic data revealed no impact due to training on the bat communities. Key words: Anabat, bat diversity, Chiroptera, National Guard, survey, Texas Introduction Five Texas Army National Guard training sites (Travis Co.) is in the Edwards Plateau. Because of the (Camp Maxey, Fort Wolters, Camp Swift, Camp Bowie, diverse plant communities and habitats at each of these and Camp Mabry) were surveyed for bats using mist properties, different species of bats are expected to use nets and Anabat (Titley Electronics, Australia) units these sites for foraging and roosting activities. during spring, summer, and fall from October 2005 to November 2006. The training sites are located in Previous survey work using mist nets documented different vegetational areas of Texas. Camp Maxey three bat species at Camp Maxey (Edwards and John¬ (Lamar Co.) is in the Post Oak/Blackland Prairie region, son 2007), two at Fort Wolters (Thies 2004b), and Fort Wolters (Parker Co.) is in the Cross Timbers and three at Camp Bowie (Dowler et al. 2004). No bats Prairies, Camp Swift (Bastrop Co.) is located in the were captured by Thies (2004a) at Camp Swift or at Blackland Prairie region, but nearby is a small area Camp Mabry (McDonough et al. 2005). According to of relict pine forest (the “Lost Pines”), Camp Bowie Schmidly (2004), there are as many as 9-10 species (Brown Co.) is in the Rolling Plains, and Camp Mabry expected to occur at each of the sites. In this study, 2 Occasional Papers, Museum of Texas Tech University we employed traditional mist-netting techniques as technique is used most often when ample environmen¬ well as acoustic monitoring to assess bat diversity at tal data are available. Previous mammal studies that each site. have used this technique include the relationships of: bat assemblages and vegetation in Paraguay (Lopez- The five sites differ in overall size, amount of Gonzalez 2004); mammal abundance relative to veg¬ improved grounds, amount of water present, and biotic etation, soil, and slope (Brant and Dowler 2001); and communities. Additionally, the sites differ in frequency moth communities versus vegetation and geography and intensity of National Guard training activities that (Ober 2006). may affect bat foraging activity. To better understand the relationship between observed bat assemblages and The objective of this project was to determine environmental variables, a canonical correspondence the bat species richness and abundance at each train¬ analysis (CCA) was performed. A CCA is a multi¬ ing site and identify relationships between the relative variate statistical tool used in community ecology to abundance of species and the environmental variables examine relationships between environmental variables associated with each of the sites. and species data (ter Braak and Smilauer 1998). This Methods Sampling methods.—We employed mist nets, The Anabat Detection System, which consisted of harp traps, and acoustic monitoring to survey the bat a bat detector and CF (compact flash) storage ZCAIM community at all five Texas Army National Guard train¬ (Zero-Crossings Analysis Interface Module), was used ing sites. Bats were captured in mist nets or harp traps to acoustically record bat echolocation calls at selected placed in flyways (trails, paths, creeks, dirt roads) and sampling sites. The Anabat system was placed at an over water sources as recommended by Kunz (1988). approximate 45° angle from the ground facing over Water sources (small lakes) that were too large to use a water source or flyway. Two Anabat systems were capture devices were sampled acoustically. Sites were used in a night; one unit placed over the site where selected to represent the major habitat types present on bats were captured and another unit placed at another the area. Sampling sites were also selected to minimize location distant from the net. Call files were recorded interference of our sampling with National Guard ac¬ to the compact flash card contained within the CF tivities. Localities of all sampling sites were recorded storage ZCAIM. Call files were downloaded from the with a handheld global positioning system (GPS) using CF card using the CFCread ZCAIM interface software Universal Transverse Mercator (UTM) North American (Corben 2006, http://users.lmi.net/corben/anabat.htm# Datum of 1983 (NAD 83) coordinates. Searches for Technical Notes). Call files were labeled with date and potential roost sites were also conducted to completely locality information and analyzed to species using the assess bat diversity at a site. AnalookW software (Corben 2006, http://users.lmi.net/ corben/anabat.htm#Technical_Notes). Call files were Bats that were captured were identified, sexed, screened visually to remove files of non-bat calls, so aged, and measured. Some voucher specimens (Ap¬ that only suitable bat calls remained. Call files were pendix I) were collected from each site, but all other compared to libraries of known bat reference calls and bats were released at the site. Specimens (skin, skull, assigned to species. When a single species could not and tissues) were prepared and deposited in the Angelo be deciphered from the call, these calls were assigned State Natural History Collection. Relative abundance to species-group categories. This was possible only of bat species was assessed using capture data. Abun¬ when clear calls were recorded and only with certain dance based on captures was estimated for each site by species. Fragmented and unclear calls were assigned the number of individuals captured per net hour (one as “unknown.” Each group, whether single species “net hour” is equal to one net open for one hour). or species group, was considered a different phonic Ammerman et al.—Bat Diversity at Texas Army National Guard Training Sites 3 group for the calculation of species activity. Relative and environmental variables that might influence bat activity based on acoustic data was calculated based activity such as amount of water, insect diversity, pro¬ on the number of bat “passes” per unit time. These portion of improved grounds, and intensity of training data could not be used to estimate relative abundance activities. of species because individual bats might be detected multiple times. Training impact was measured as the number of man-days of training activities at each site from Sep¬ Each training center was surveyed 4-6 times tember 2005 to September 2006. The acoustic data over the approximate course of a year (October 2005- (number of passes for each phonic group) for each November 2006) to determine seasonal occurrence of training site was divided by the number of survey nights bat species. Camp Maxey was the only site sampled in to eliminate uneven sampling and then log-transformed the winter because of its high diversity and likelihood to account for the high values that could potentially of obtaining information on winter activity which is influence the ordination. For example, some sites lacking in many bat species (Boyles et al. 2006). Each had inflated call activity that might be due to multiple visit lasted two nights and sampling was conducted passes by a few bats. Training impact could not be from sunset to sunrise. estimated for Camp Mabry because this site serves as the headquarters and the primary activity is indoor of¬ Canonical Correspondence Analysis.—A total fice/classroom work and is not outdoor training activity. of 25 environmental variables (Appendix II) and nine Because Camp Mabry had no training impact or train¬ phonic groups were included in the CCA of the five ing acres estimated, we ran 2 separate CCA analyses National Guard training sites using CANOCO V4.0 - one that included Camp Mabry and one without - to (ter Braak and Smilauer 1998). CCA can help identify evaluate the effect. relationships between the observed bat assemblages Results Survey Results for Camp Maxey.—Camp Maxey subflavus (eastern pipistrelle) (n=4), Lasiurus cinereus is a 2,600-ha site in Lamar County of northeastern (hoary bat) (n=l), and Lasiurus seminolus (Seminole Texas located just south of Pat Mayse Lake. Habitat on bat) (n=l). The predominant species were L. borealis Camp Maxey falls in the Northern Post Oak Savannah and N. humeralis. The capture of L. seminolus and between the Northern Blackland Prairie and Red River L. cinereus were the first documented individuals in Bottomlands ecoregions of Texas. Plant communities Lamar County. In addition, the capture of L. semi¬ present include Post Oak-Black Hickory woodlands, nolus represented the most northern record in Texas. Shortleaf Pine forests and savanna, Little Bluestem- Nycticeius humeralis was the most abundant species Indiangrass grasslands, and Water Oak-Willow Oak during the spring sampling period and L. borealis was riparian forests (Farquharetal. 1996; Wolfe etal. 1996; most abundant during the summer (Fig. 2). Hunter 2005). Capture activity at Camp Maxey dropped off in This site had the highest bat species diversity. A June (n=8) as opposed to the preceding month (May, total of nine sites were sampled within training areas II, n=22) and the following two months (July, n=28; Au¬ IV, V, VI, and VII of the Camp Maxey training center gust n=24). Only two bats were captured in October (Fig. 1). Five sites were sampled with mist-netting and (L. borealis and N. humeralis). Higher capture rates acoustic monitoring and four sites were sampled only in May and July can be attributed to a single pond (15 by acoustic monitoring. Mist nets were monitored for S 262282 E 3742719 N) in training area IV that was a total of 667 net hours and resulted in 83 captures of surveyed during those months and not in June rather five species: Lasiurus borealis (eastern red bat) (n=45), than other environmental factors such as lunar illumi¬ Nycticeius humeralis (evening bat) (n=32), Perimyotis nation. This particular pond when surveyed produced 4 Occasional Papers, Museum of Texas Tech University 0 0.5 1 i Kilometers X 4 A OA JKy W X Of v. J yK Kj 1 Lyj, (V (cid:9734) (cid:9632)/ p 1 —3- k'M i i r Kaa a i a (cid:9734) Mist Net Sites r M1 h -/V o \y ANABAT Sites Roads %M_ 2- Streams A). J PondsAA/et lands Figure 1. Location of sampling sites at Camp Maxey (Lamar County). Stars represent sites sampled by mist net; circles represent sites that were sampled acoustically. TA=training area. higher captures and exhibited the majority of the overall numerous within the training center area compared to captures (n=58, 70%). Interestingly, all species found the few captures (n=4) that were made. Few calls were in Camp Maxey were captured at this pond. recognized for L. cinereus (n=42) and no calls were recognized for L. seminolus. A number of L. cinereus A total of 5,951 call files were recorded and and L. seminolus calls could be contained within the call designated into nine classes: LABO (L. borealis), classification LAsp (n=3 90). With the presence of three LABOPISU (L. borealis!P. subflavus), LACI (L. ci- species of Lasiurus, call identification can become dif¬ nereus), LAsp {Lasiurus sp.), NYHU (A. burneralis), ficult with much overlap among species. The case can NYHULABO (A. humeralis/L. Z>orea/A),NYHUPISU be more so for L. borealis and L. seminolus, which have (A. humeralis/P. subflavus), PISU (P. subflavus), and similar body sizes; total length =108 mm and forearm UNKNOWN (Fig. 3). LABO calls constituted the = 40 mm for L. borealis and total length =103 mm and majority of calls with NYHULABO calls being the forearm = 39 mm for L. seminolus (Schmidly 2004). second most recorded. Although A. humeralis repre¬ Congeneric bat species that are similar in size tend to sented 39% (n=32) of the captures, very few identifiable have a high degree of call similarities (Bogdanowicz NYHU calls were recorded (n=375, 6%). This can be et al. 1999). Lasiurus cinereus generally has a call the result of a large number of A. humeralis calls be¬ repertoire that is lower in frequency than L. borealis ing indistinguishable from L. borealis calls resulting and L. seminolus, but during times when L. cinereus in 1,372 calls (23%) being classified as NYHULABO. produces higher frequency calls, confusion can occur A large number of calls were recorded for P. subflavus with the two former species. (n=760), thus indicating that P. subflavus may be more Ammerman et al.—Bat Diversity at Texas Army National Guard Training Sites 5 ) r u o h t e n / s e r u t p a c ( e t a R e r u t p a C Figure 2. Capture rates of bats at Texas Army National Guard training sites in spring, summer, and fall (2005-2006). 6 Occasional Papers, Museum of Texas Tech University 0 Bowie (cid:9632) Swift (cid:9633) Mabry (cid:9633) Maxey (cid:9633) Wolters Figure 3. Phonic groups identified using Anabat for bats at Texas Army National Guard training sites. LABO=Lasiurus borealis, LACl=Lasiurus cinereus, LAsp=Lasiurus species, MYsp=Myotis species, NYHU =Nycticeius humeralis, PISU=Perimyotis subflavus, and TABR=Tadarida brasiliensis. Call activity showed an increase from spring to Survey Results for Fort Wolters.—Fort Wolters is the summer months, yet call activity peaked in June a 1,614-ha site located in Parker and Palo Pinto coun¬ (Fig. 4). While more calls were recorded in June ties in north-central Texas near Lake Mineral Wells (n=1990) than in July (n=1478), more captures were State Park at the transition between the Oak Woods and made in July (n=28) versus June (n=8). The higher Prairies and Blackland Prairies natural regions of Texas number of captures in July are likely because sampling in an ecoregion called the Western Cross Timbers. at a pond (15 S 262282 E 3742719 N) in training area Plant communities present include Post Oak-Blackjack IV during this month resulted in the most captures of Oak Woodland, Ashe Juniper-Oak Woodland, Little all sites surveyed at Camp Maxey. Bats may be more Bluestem-Indiangrass Grassland and Sugar Hackberry- easily caught at this pond versus other sites. Also, the Elm Riparian Woodlands (Farquhar et al. 1996; Wolfe bat activity (based upon acoustic monitoring) at the et al. 1996; Hunter 2005). sites surveyed in June may have been higher, yet these sites did not facilitate the capture of bats. There was a A total of seven sites were sampled within train¬ single call recorded in December (NYHULABO). ing areas la, II, Ilia, Illb, IV, and VI of the Fort Wolters Ammerman et al.—Bat Diversity at Texas Army National Guard Training Sites 7 (cid:9633) Bowie (cid:9632) Swift (cid:9633) Mabry (cid:9633) Maxey (cid:9633) Wolters Figure 4. Monthly distribution of call files recorded at Texas Army National Guard training sites in 2005-2006. training center (Fig. 5). Five sites were sampled with ponds, thus demonstrating higher capture rates during mist-netting and acoustic monitoring and two sites were those months when it was surveyed. Overall, L. borea¬ sampled only by acoustic monitoring. Mist nets were lis and N. humeralis were captured in approximately monitored for a total of 534 net hours and resulted in equal frequency. However, more N humeralis were 52 captures of three species: Lasiurus borealis (n=23), captured in the summer and fall while more L. borealis Nycticeius humeralis (n=26), and Tadarida brasiliensis were captured in the spring (Fig. 2). (Brazilian free-tailed bat) (n=3). A total of 5,660 call files were recorded among The rate of captures varied throughout the year seven sites within the Fort Wolters training center des¬ with captures peaking in mid-summer (10 captures in ignated into seven classes: LABO (L. borealis), LAsp November 2005, 5 captures in May 2006, 14 captures CLasiurus sp.), NYHU (N. humeralis), NYHULABO in June 2006, 22 captures in July 2006, 1 capture in (N humeralis/L. borealis), TABR (T. brasiliensis), August 2006, and 0 captures in November 2006). This TABRLAsp (T. brasiliensis!Lasiurus sp.) and UN¬ fluctuation in activity is likely due to the particular KNOWN (Fig. 3). Although very few T. brasiliensis ponds surveyed during that time. For instance, a pond (n=3) were caught, they constituted the most calls re¬ (14 S 589651 E 3637414 N) in training area Ilia which corded. Tadarida brasiliensis, being a less maneuver- produced the most captures overall (n=27, 52%) was able but fast flying bat, requires large, open pools of wa¬ surveyed in November 2005, June 2006, and July 2006, ter in order to drink (Norberg and Rayner 1987). Thus, but not in May 2006 or August 2006. This specific a pond located within training area la (14 S 587901 E pond appears to facilitate the captures of bats over other 3635242 N) presented an adequate source in which to 8 Occasional Papers, Museum of Texas Tech University Figure 5. Location of sampling sites at Fort Wolters (Parker County). Stars represent sites sampled by mist net; circles represent sites that were sampled acoustically. TA=training area. Ammerman et al.—Bat Diversity at Texas Army National Guard Training Sites 9 catch the three individuals of T. brasiliensis that rep¬ Savannah between the Northern Blackland Prairie resent a first record for Parker County. Few calls were and Bastrop Lost Pines ecoregions of Texas. Plant classified as T. brasiliensis/Lasiurus sp. (n=36), which communities present include Oak-Eastern Red Cedar is indicative of the presence of L. cinereus because T. Forest, Little Bluestem-Indiangrass Grassland, Green brasiliensis and L. cinereus calls often can be confused. Ash-American Elm Riparian Forest, and Loblolly Pine Also, a number of calls were identified as Lasiurus sp. Forest (Wolfe et al. 1996; Fischer and Senseman 2003; (n=148) leading to the possibility of other Lasiurus Williams 2003). species being present in addition to L. borealis. No calls were found to strongly resemble L. cinereus calls A total of nine sites were sampled at the Camp and a capture of L. cinereus was not made. Lasiurus Swift training center (Fig. 6). Four sites were sampled borealis and N. humeralis exhibited similar activity by with both mist nets and Anabat, four sites were sampled captures (n=23 and n=26, respectively), yet call activ¬ with Anabat only, and one site was sampled with mist ity varied (n=704 and n=425, respectively). Some of nets alone. Mist nets were monitored for a total of 431 the discrepancy observed might be accounted for by net hours and resulted in five captures of a single spe¬ the large number of calls that were classified as either cies, Lasiurus borealis (Fig. 2). Four individuals were N. humeralis or L. borealis (NYHULABO, n=1446). captured in April (two males, two pregnant females) The lack of Myotis calls at this site was unexpected and and one in July (juvenile). This site had the lowest net¬ may possibly be a result of placement of Anabat units ting success of all of the training sites. Most captures in more open areas. occurred at Long Skinny Pond. Call activity increased from fall 2005 (Novem¬ Camp Swift had the lowest call activity of all five ber) to spring and summer months of 2006 with peak sites. A total of 661 call files were recorded with most activity in August 2006 (Fig. 4). During the summer of the call activity in April (Fig. 3). The dominant call months, overall call activity in June (n= 1,063) and activity in April (59% of calls) was T. brasiliensis, but July (n=l,151) was approximately equal. However, none were captured in nets. The highest call activity considerably more captures were made in July (n=28) in summer was the NYHULABO phonic group (Fig. versus June (n=8). The higher number of captures in 4). Although it is possible that N. humeralis and P. July primarily are attributable to captures at a pond in subflavus are present at Camp Swift, no captures were training area la (14 S 587901 E 3635242 N), which was made and no undisputed call files were scored. the site with the most captures of any surveyed. One interesting result is the comparison of activity between Survey Results for Camp Bowie.—Camp Bowie is November 2005 and November 2006 in which surveys a 3,542-ha site located in Brown County in west-central were conducted around the same time of the month (11- Texas at the transition between the Western Cross 12 November 2005 and 10-11 November 2006). Higher Timbers and Limestone Plains ecoregions. Plant com¬ capture and call activity was found in November 2005 munities present include: Plateau Live Oak-Midgrass (n=10 and n=652, respectively) in contrast to Novem¬ Woodland, Post Oak-Blackjack Oak Woodlands, Texas ber 2006 (n=0 and n=60, respectively) (Fig. 4). These Oak Woodlands, American Elm-Cedar Elm Wood¬ vast differences in activity levels can be attributed to lands, Pecan-Sugarberry Woodlands, Ashe Juniper- a notable disparity in temperature for the two survey Oak Woodlands, Ashe Juniper Woodlands, Mesquite periods. When looking at archived temperature data Woodlands and Forests, and Sideoats Grama-Little (Miami Herald 2007) the two nights surveyed during Bluestem Grasslands (Wolfe et al. 1996; Fischer and November 2005 (20.11°C and 18.06°C, respectively) Senseman 2003). showed an average of 11 degrees higher than the two nights surveyed during November 2006 (9.83°C and A total of five different sites were sampled at 6.06°C, respectively). Camp Bowie training center (Fig. 7) with mist-netting and acoustic monitoring. Mist nets were monitored for Survey Results for Camp Swift.—Camp Swift is a total of 428 net hours and resulted in 18 captures of a 4,718-ha site located in Bastrop County in central three species. Myotis velifer (cave myotis) was cap¬ Texas and is located within the Southern Post Oak tured most often (n=ll), followed by five L. borealis 10 Occasional Papers, Museum of Texas Tech University Figure 6. Location of sampling sites at Camp Swift (Bastrop County). Stars represent sites sampled by mist net; circles represent sites that were sampled acoustically. TA=training area.