Final Report

Examining the Role of Cave Crickets (Rhaphidophoridae) in Central Texas Cave Ecosystems: Isotope Ratios (δ13C, δ15N) and Radio Tracking

Steven J. Taylor1, Keith Hackley2, Jean K. Krejca3, Michael J. Dreslik 1, Sallie E. Greenberg2, and Erin L. Raboin1

1Center for Biodiversity Illinois Natural History Survey 607 East Peabody Drive Champaign, Illinois 61820 (217) 333-5702 [email protected]

2 Isotope Geochemistry Laboratory Illinois State Geological Survey 615 East Peabody Drive Champaign, Illinois 61820

3Zara Environmental LLC 118 West Goforth Road Buda, Texas 78610

Illinois Natural History Survey Center for Biodiversity Technical Report 2004 (9)

Prepared for: U.S. Army Engineer Research and Development Center ERDC-CTC, ATTN: Michael L. Denight 2902 Newmark Drive Champaign, IL 61822-1076 27 September 2004

Cover: A cave cricket (Ceuthophilus The Red Imported Fire Ant (Solenopsis secretus) shedding its exuvium on a shrub (False Indigo, Amorpha fruticosa L.) outside invicta Buren, RIFA) has been shown to enter and of Big Red Cave. Photo by Jean K. Krejca. forage in caves in central Texas (Elliott 1992, 1994; Reddell 2001; Reddell and Cokendolpher 2001b). Many of these caves are home to federally endangered invertebrates (USFWS 1988, 1993, 2000) or closely related, often rare taxa (Reddell 2001, Reddell and Cokendolpher 2001a). The majority of these caves are small – at Fort Hood (Bell and Coryell counties), the mean length1 of the caves is 51.7 m (range 2.1 - 2571.6 m, n=105 caves). Few of the caves harbor large numbers of bats, perhaps because low ceiling heights increase their vulnerability to depredation by other vertebrate predators (e.g., raccoons, Procyon lotor). Lacking bats, these energy poor caves primarily receive energy from detritus and surface animals that fall or are washed into the caves and from energy brought into the caves by cave crickets (Ceuthophilus secretus, and perhaps Ceuthophilus cunicularis) and harvestmen (Lieobunum townsendii).

We know a great deal about the ecology of caves in general (Culver 1982, Howarth 1983, Poulson and White 1969, Vandel 1965). In many caves, nutrients appear to be concentrated near the cave entrance, arriving in the form of falling organic debris and presumably the feces and bodies of various organisms (Peck 1976). Deeper in the caves of central Texas, the feces, eggs, and bodies of Ceuthophilus spp. appear to comprise a more important energy source, a pattern similar to that observed for Hadenoecus sp. crickets in the eastern United States (e.g., Poulson et al. 1995). The above generalizations vary widely from cave to cave, but serve as a conceptual starting point for understanding the cave communities at Fort Hood, Texas (Bell and Coryell counties).

Land managers with an interest in protecting the rare and endangered karst invertebrates have expended considerable financial resources in an attempt to control RIFA activity around cave entrances. Effective control is accomplished through the killing of individual mounds with boiling water or steam applications which must be repeated on a regular basis. The area to be treated includes susceptible area around the cave entrance or cave footprint, with the intention of excluding the ants from the cave. As the foraging range of the red imported fire ant is about 25 meters, the area to be treated is at least 0.19 hectares, more if the footprint of the cave determines treatment area.

1 Data source (2001): Fort Hood Natural Resources Branch 2 These RIFA control efforts are further complicated by the foraging range of cave crickets. Ceuthophilus secretus forages at night on the surface and roosts in the caves during the day. Elliott (1992), working with C. secretus and a closely related, undescribed species (Ceuthophilus “species B,” which does not occur at Fort Hood), noted that “Cave crickets mostly feed within 5 or 10 m of the cave entrance, but large adults may travel 50 m or more.” Based on Elliott’s (1992, 1994) work, it is thought that most cave crickets forage within 30 m of the entrances of caves (Reddell and Cokendolpher 2001b). Because of the presumed interactions (competition and/or predation) of red imported fire ants and cave crickets on the surface, land managers have used this figure to enlarge the RIFA treatment area around cave entrances. More recently, Taylor et al. (USFWS 2003, pg 17159; Taylor et al. 2003, Taylor et al. submitted) found that C. secretus can travel up to 105 m away from a cave entrance. This figure greatly increases the area that would need to be treated to avoid fire ant/cave cricket interactions above ground,potentially increasing land management costs and other logistics associated with treating these larger areas.

Two species of cave crickets, Ceuthophilus secretus and Ceuthophilus cunicularis, occur in caves at Fort Hood. It is generally thought that cave crickets are scavengers or omnivores, and C. secretus is known to forage as far as 105 m from cave entrances (Taylor et al. 2003, Taylor et al. submitted). We suspect, but do not know, that C. secretus competes for food resources with RIFA. If land managers are attempting to manage landscapes around cave entrances to protect rare and endemic troglobites, it follows that they should have an understanding of what components of the epigean2 flora and fauna comprise major constituents of the energy brought into caves by C. secretus. That is, it seems reasonable to presume that protection of the cave fauna would be facilitated by encouraging populations or communities of epigean elements that are major contributors to the diet of C. secretus. Furthermore, we are nearly completely ignorant of the relationships among the various troglophiles and (often rare and/or endemic) troglobites that live in the caves at Fort Hood. Further, we know very little about trophic relationships of Fort Hood cavernicoles beyond occasional anecdotal observations of species interactions (e.g., predation events). Enhancing our understanding of food web relationships within the caves could prove useful in guiding management decisions.

Here we conduct two studies of cave crickets that attempt to: 1) gain an understanding of their trophic position in the food web of the cave ecosystem through analysis of the stable isotope ratios of nitrogen and carbon, and 2) gain further understanding of the epigean movements of

2 epigean – “Pertaining to, or living on, the surface of the earth” (Field 2002). 3 cave crickets during their foraging excursions through the use of very small radio transmitters. These two studies are presented in separate sections.

4

ISOTOPES

Analysis of Carbon and Nitrogen isotope ratios

Stable isotopes are popular tools for investigating ecosystems (Griffiths 1998, Lajtha and Michener 1994, Peterson and Fry 1987, Rounick and Winterbourn 1986, Rundel et al. 1988). The stable isotopes of carbon and nitrogen occur in virtually all animal tissues (Peterson and Fry 1987), and their ratios (δ13C, δ15N) have been used to track the movement of energy through a food web (Fry and Sherr 1984, Cabana and Rasmussen 1994, Ostrom et al. 1997, Ponsard and Arditi 2000, McNabb et al. 2001, Hobson et al. 2002, Blüthgen et al. 2003, Quinn et al. 2003) and to help identify the food sources of animals which are difficult to observe in the wild (e.g., Fry et al. 1978, Rico-Gray and Sternberg 1991, Markow et al. 2000, Hocking and Reimchen 2002, Carmichael et al. 2004). These isotope signatures essentially represent a running average3 of the feeding history of an organism (O’Reilly et al. 2002), and thus are not as biased as individual observations of instances of food resource utilization. Rather, this signature depends on the turnover rate of the isotopes in the tissue of the animals being examined and is closely tied to the isotope ratios in their diet (DeNiro and Epstein 1978, 1981). Species at higher trophic levels typically have enriched δ15N relative to their prey (e.g., Oelbermann and Scheu 2002), but often utilize a variety of food sources (Pain 1988, Persson 1999, Post 2002). Each trophic level is thought to be enriched by a factor of about 3 to 4 δ units (Michener and Schell 1994), typically 3.4 δ units (Wada et al. 1991), though this value for stepwise enrichment is not so clearly applicable to invertebrates (Scrimgeour et al. 1995, Scheu 2002) and potentially could be influenced even by microbial nitrogen fixation (e.g., Nardi et al. 2002) or the presence of mycoflora (e.g., Benoit et al. 2004). Quinn et al. (2003) reported enrichment between trophic levels varying from 2.1 to 4.9 ‰ (‘per mil’, parts per thousand), but the average value (3.5±0.8 ‰) is very close to that proposed by Wada et al. (1991), 3.4 ‰. The isotope ratios are, in actuality, partitioned among the various sources (e.g., Koch and Phillips 2002, Phillips and Koch 2002, Phillips and Gregg 2003, Robbins

3 Carbon isotope turnover in the chiton of locusts (Locusta migratoria Linnaeus 1758; Orthoptera: Acrididae) can occur in as little as eight days (Webb et al. 1998). 5 et al. 2002). In many cases, isotopic values form a continuum of δ15N values (e.g., Ponsard and Arditi 2000, Scheu and Falca 2000, Blüthgen et al. 2003, Quinn et al. 2003) and may vary seasonally (Neilson et al. 1998) and across relatively short distances (e.g., Hocking and Reimchen 2002). Thus, while isotope studies often result in new insights into trophic relationships, they rarely give completely decisive explanations of ecosystem functioning.

Stable isotopes of nitrogen and carbon have been used with some success to characterize food webs and trophic levels of a cave in Arkansas (Graening 2000; Graening and Brown 2000, 2003), with sea cave-inhabiting fruit bats in Mexico (Ceballos 1997), and in anchialine aquatic cave communities in Mexico (Pohlman et al. 1997, 2000) and northwestern Australia (Humphreys 1999). In spite of a fairly substantive body of research on the biology of North American cave crickets (cited earlier), we are not aware of any studies of cave-inhabiting Rhaphidophoridae (Ceuthophilus spp., Haedonoecus spp., etc.) that have utilized stable isotope analyses.

Methods

Our objective was to obtain δ13C and δ15N data at Fort Hood caves for a variety of possible constituents of a cave ecosystem. We obtained samples of the dominant cave crickets (C. secretus), RIFA, a selection of the more common plant taxa near the cave entrance, and other selected cavernicoles (e.g., C. cunnicularis, Cambala speobia millipeds, etc.). These collections were made at Mixmaster Cave and Streak Cave on 21 and 22 November 2003, respectively, and at Streak Cave and Big Red Cave on 23 May 2004.

Samples of cave crickets (C. secretus and C. cunicularis), RIFA, and other invertebrates were hand collected and kept on ice in the field, then were put on dry ice and shipped to the laboratory, where they remained frozen until processing could begin. For invertebrate samples, multiple individuals were sometimes pooled to obtain sufficient material for analysis when individual biomass was low. For rarer cave taxa this was not always possible or prudent. Vegetative samples of abundant plant species were collected and cleaned of arthropods. Selected individuals of each taxon (plants and animals) were preserved as vouchers, and identifications confirmed by an appropriate expert (SJT for most animals, Geoff Levin [INHS] for plants). Plant identifications follow Diggs et al. (1999), Correll and Johnston (1970), Hatch et al. (2001) and ITIS (2004). Vouchers specimens have been deposited in the Illinois Natural History Survey herbarium (plants) and insect (arthropods) collections. 6

Nitrogen and Carbon isotopes were measured using a Finnigan Mat 252 isotope ratio mass spectrometer (IRMS) with an attached Carlo Erba NC 2500 Elemental Analyzer (EA) with ConFlo II Split Interface (Figure 1,2). Insect and plant samples were dried in an oven at 80 °C for at least 72 hours and ground using a mortar and pestle (Figure 4a) to homogenize the samples. Dried samples were stored in a dessicator (Figure 4b) until they could be weighed (Figure 5) and analyzed for their carbon and nitrogen isotopic composition. Approximately 400-700 µg of sample was used for insect analyses and was analyzed for nitrogen and carbon at the same time. Plant samples were analyzed separately for nitrogen (~1500 µg sample) and carbon (~500 µg sample). Samples were wrapped in tin capsules (Figure 5) and combusted at 1020°C in the EA

(Figures 1b, 3). The N2 and CO2 gases were separated by a gas chromatograph column and introduced into the IRMS for analysis through a capillary tube.

The isotopic composition is reported in the delta (δ) notation which compares the ratio of two isotopes of the same element in a sample to the same ratio in an internationally accepted standard. The delta notation is defined as:

δ X = [(Rsmpl - Rstd)/ Rstd] C 1000

Where X is the isotope of interest such as, 15N or 13C, and R is the ratio of the isotopes being analyzed; for N the ratio is 15N/14N and for C the ratio is 13C/12C. Thus, the isotopic ratio in a sample is compared to the same ratio in the standard. The results are reported in parts per thousand or per mil (‰). If the sample contains a greater amount of the heavier isotopes compared to the standard, the delta value is positive, if there are less heavy isotopes in the sample compared to the standard, the δ value is negative.

The δ15N and δ13C values are reported relative to air and the Vienna Peedee belemnite (V- PDB) reference standard, respectively. The V-PDB reference standard is calibrated through analysis of NBS-19 with a carbon value -2.2 ‰ (Coplen et al. 1983). Reproducibility for both δ15N and δ13C is ± 0.15 ‰ based on multiple analyses of the homogeneous laboratory internal standard. Weight percents were calculated by comparing the measured peak area, sample weight, and a set peak area defined using a hydroxproline standard with 10.60 % nitrogen and 45.45 % carbon. The reproducibility of weight percents is ± 2% weight percent for samples containing 10 % or greater carbon and/or nitrogen. Each sequence of samples was run with the 7 following standards: an internal standard at the beginning and end of the run, a hydroxyproline to set weight percent, another hydroxyproline to verify weight percents, and every tenth sample was run as a duplicate.

Nitrogen and carbon isotope compositions derived from the above procedures were then plotted on graphs, and the distribution of the data will be used to assess potential trophic relationships among the taxa examined.

Results & Discussion

In most cases, we were able to identify taxa to species level in the field for animals, or by collection of vouchers for later identification by an appropriate expert for plants, facilitated by field photographs of the plants (Appendix 1). However, some taxa were pooled at the generic (e.g., Cicurina spp.[=eyeless troglobitic spiders]) or familial (e.g., Staphylinidae [rove beetles]) levels due to difficulties in obtaining species-level identifications in these groups prior to analysis. Some taxa are represented by single specimens (e.g., centipedes), and were only identified to the ordinal level. All animal taxa were collected in-cave, with the exception of Solenopsis invicta, which was collected from mounds within 10 m of the cave entrance, except at Streak Cave where we had go 30+ m from the cave entrance to locate a fire ant mound. Plants and the blue-green bacteria were collected within 30 m of the cave entrances.

Results for δ13C and δ15N analysis of 121 plant and animal samples are given in Table 1 and summarized by taxon in Table 2. Additional samples are still being processed in the isotope laboratory. For C4 plants (Sporobolus ozarkanus Fernald, Tridens muticus (Torr.) Nash var. muticus, and an unidentified grass [Poaceae]), δ13C ranged from -12.96 to -11.27 ‰ and δ15N ranged from -3.30 to 4.46 ‰, while for C3 plants (the majority of plant taxa sampled) , δ13C ranged from -31.17 to -22.97 ‰ and δ15N ranged from -3.79 to 3.27 ‰ (Figure 15, Tables 2, 3). Opuntia phaecantha Engelm. var. major Engelm. is a Crassulacean Acid Metabolism (CAM) plant, and δ13C ranged from -13.14 to -12.20 ‰ and δ15N ranged from -0.06 to 2.65 ‰. For animals, δ13C ranged from -26.17 to -16.65 ‰ and δ15N ranged from -0.76 ‰ for Mesodon roemeri to 13.25 ‰ for eyeless Cicurina sp. (Figure 7, Tables 1, 2). A single sample of blue-green bacteria was analyzed, and δ13C and δ15N were -19.5 and -1.23 ‰, respectively.

8 The relationships among various taxa show intriguing trends (Figure 7). Those animals at the lowest trophic levels (based on δ15N values: Mesodon roemeri, Ceuthophilus secretus, Leiobunum townsendii, and Speodesmus castellanus) were most similar in δ13C values to the plants Diospyros texana Scheele, Juniperus ashei J. Buchholz (both samples of berries and samples of leaves), and a liverwort (Figure 7, Table 1). The δ13C values for Helicodiscus eigenmanni are much more highly enriched than for other cave animals (Figures 7,8), especially given their intermediate trophic position (based on δ15N). Solenopsis invicta, collected outside of caves and known to be an omnivore, also displayed markedly enriched δ13C relative to the other cave animals (Figures 7, 8). It would appear, then, that the terrestrial cave snail, Helicodiscus eigenmanni, the Red Imported Fire Ant (Solenopsis invicta), and, to a lesser extent, the recently described (Elliott 2004) Fort Hood Cave Millipede (Speodesmus castellanus), consume foods which differ from the remaining cave animals examined in our study. Based on the values for δ13C, it may be that these taxa (Solenopsis invicata, Helicodiscus eigenmanni, and Speodesmus castellanus) have diets derived from differing sources, including C4 and/or Crassulacean Acid Metabolism (CAM) plants, relative to the diets of the cave crickets and other cavernicolous taxa. With these exceptions, the cave animals tend to show a remarkably tight relationship: on the average, with each increase in δ15N there was only small change in carbon isotope fractionation, suggestive of a single integrated pathway, or food chain, through which nutrients are passed up the trophic levels, as indicated by the gray region in Figure 8. The sequence of animal taxa along this linear trend from lowest to highest trophic level, then, is Mesodon roemeri, Ceuthophilus secretus, Lieobunum townsendii, Scutigeridae, Siphonophora sp., Batrisodes sp., Ceuthophilus cunnicularis, Rhadine reyesi, Ceuthophilus sp. (likely and immature Ceuthophilus cunnicularis), Haplotaxida, Cambala speobia, Lithobiomorpha, Staphylinidae, Cicurina varians, and eyeless Cicurina spp. Somewhat set off (in terms of δ13C values) from this trend are Speodesmus castellanus and Helicodiscus eigenmanni. The generally linear relationship of this seemingly tightly integrated food web is particularly important in that many of the taxa of concern at Fort Hood, and many of the federally endangered troglobites in the Austin and San Antonio areas, function at the higher trophic levels within the cave community (i.e., many are largely carnivorous). Therefore, the available data presented here, especially when combined with spatial and abundance data from Taylor et al. (2003), indicates that it is important to manage entire cave communities and the epigean community within which C. secretus forages if the species of concern (and the federally endangered cave species further south in Texas) are to be protected.

9 While the above observations are quite important in that they indicate a close dependence of taxa at higher trophic levels upon those at lower trophic levels within the cave ecosystem, assigning individual taxa to particular tropic levels quickly becomes problematic (Figure 10). The snail Mesodon roemeri is commonly encountered in the entrances of Fort Hood caves, appears to be herbivorous or possibly a detritovore. While gastropods are known to forage at a variety of trophic levels (Speiser 2001), Blinn (1963) recorded Mesodon species ingesting rotting wood and plant leaves, Pilsbry (1940) suggested that some Mesodon species are fungivores, Mesodon clarki has been found - perhaps foraging - on herbaceous vegetation (including stonecrop and Japanese honeysuckle) in the springtime (USFWS 1984), and Mesodon ferrissi feeds readily on Indian mustard (Brassica juncea) in a laboratory setting (Hanson et al. 2003). The two Cicurina taxa are clearly predators (as are nearly all spiders), and there is evidence from other work on isotopic enrichment in spiders that the high values for δ13C and δ15N could reflect either use of prey with high δ13C and δ15N values or that enrichment has resulted from starvation (Oelbermann and Scheu 2002). Because caves are generally considered low energy systems, where we might expect top predators (i.e., Cicurina spiders) to go for relatively long periods without prey, the latter option (enrichment associated with limited food resources) certainly should not be discounted. But beginning with Ceuthophilus secretus, omnivory clouds the picture. Based on their position (at a relatively low trophic level), C. secretus appears to consume plant material, consistent with the observations of Elliott (1992, 1994) who found surface-foraging Ceuthophilus feeding on fungi and ripe native persimmons. Elliott (1992, 1994) also observed that they feed on dead insects (including fire ants), an observation corroborated by our research (Taylor et al. submitted, Figure 10). Further, this species is readily attracted to baits such as tuna (Taylor et al. 2003). Though we know little about the feeding habits of Leiobunum townsendii, there is no reason to believe it is a strict herbivore either (given its’ similarity to C. secretus in isotopic composition), and centipedes are generally thought of as carnivores, but the Scutigeromorph centipede we analyzed appears to function at a fairly low trophic level. Taxa at higher trophic levels appear to include the carabid beetle Rhadine reyesi, known to be a predator of Ceuthophilus eggs (see photo of Rhadine reyesi with egg in Taylor [2003]). Ceuthophilus cunnicularis is very similar in isotopic composition to Rhadine reyesi (Figures 7-9), and thus appears to be more carnivorous than its congener, perhaps even preying upon C. secretus. The small, troglobitic beetle of the genus Batrisodes also falls out near this group. The relatively small numbers of C. cunnicularis relative to C. secretus, as reported by Taylor et al. (2003), is consistent with C. cunnicularis being largely predatory. We (Taylor et al. 2003) have also regularly attracted this species at meat baits. Further confusing this picture is an earthworm (Haplotaxida) whose isotopic composition is also 10 similar to C. cunnicularis and R. reyesi. Most earthworms are typically thought to be geophagous, though some typically feed on detritus which has been processed by microbes (Briones et al. 1999) and lumbricids are known to be omnivores (Schmidt 1999).

Effects of site and season

To look for possible differences between sites (caves) we used the Kruskal-Wallis Test to compare δ13C or δ15N values for each cricket species, the only taxa for which we had several individuals from each site. For C. cunnicularis, we did not detect a difference in the mean values for δ13C among caves (Table 3), though Big Red Cave appeared somewhat higher and the p value was not far from significance, thus a larger sample size may have yielded significant differences. For C. secretus, no difference in δ13C among caves was detected (Table 3). For both C. cunnicularis and C. secretus, there were significant differences among caves in δ15N values, with higher δ15N values for Mixmaster Cave (Table 3). Tayor et al. (2003) report that Mixmaster Cave, and the area around it, had the highest numbers of fire ants, and was one of the most highly impacted of the six caves they examined (including Streak and Big Red Caves) – it is unclear whether or not such conditions might contribute to the higher δ15N values observed for this cave in the present study.

To look for possible differences between seasons (May, November), we used the Mann- Whitney Test to compare δ13C or δ15N values for each cricket species, the only taxa for which we had several individuals from each site. We did not detect a difference in the mean values for δ13C for C. cunnicularis, but seasonal differences were significant for C. secretus (Table 3). For δ15N, we detected seasonal differences for both C. cunnicularis and C. secretus, with higher values for November samples than for May samples (Table 3). Although we have found some differences between sites and seasons, our data set is too small to effectively analyze results separately by site and season for all taxa. Therefore, some caution should be used in interpretation of pooled results presented herein.

Fagen et al. (2002) found general evidence of higher nitrogen content in predators versus herbivores, but when we looked for possible differences in percent nitrogen in C. cunnicularis (mean=10.57%) versus C. secretus (mean=10.20%) (Table 2), we were unable to detect a difference (two tailed t-test, unequal variances: df=21, t=0.8151, p=0.42418) – thus our data do not seem to fit the generalization of Fagen et al. (2002). Nonetheless, the differences in δ15N

11 between the two Ceuthophilus species, especially in light of the general consistency of our data with the theory of the isotopic trophic ecology, provides strong evidence that Ceuthophilus secretus and Ceuthophilus cunnicularis function at differing trophic levels.

Numerous authors have recorded isotopic values for plants which have shown similar results for δ13C among studies (e.g., Smith and Epstein 1971, Tieszen et al. 1979, Boutton et al. 1983, LePage et al. 1993), and these studies generally show marked differences between plant taxa following C3 versus C4 photosynthetic pathways. Values of δ13C for C3 plants generally fall in the range of -23 to -32 ‰, while the range for C4 plants is typically -11 to -15 ‰ (Smith and Epstein 1971, Smith and Brown 1973, Troughton and Card 1975, LePage et al. 1993). Our data are consistent with these earlier studies (Tables 1, 2, Figures 7-9).

It appears that many of the cave taxa feed at more than one trophic level, and thus source partitioning of isotope fractionation (e.g., Phillips and Gregg 2003) appears to be complex for the Fort Hood cave communities, and far beyond the scope of the present study. It also appears that many of the taxa feed within a single food chain, and thus all are dependent on a single energy source – protection of species of concern, or (elsewhere in Texas) endangered species, then, depends on maintaining the entire cave ecosystem to ensure that the top predators are protected in perpetuity. These observations are consistent with findings in the eastern United States, where the guano of the rhaphidophorid Hadenoecus spp. is important as a food source for other cave inhabitants (Benoit 2004, Poulson et al. 1995). While it would be most ideal to use the information gleaned from stable isotope analyses and published literature to develop a detailed food web and assign taxa to particular trophic levels, we can, at this time, only make approximate assignments to trophic levels (Figure 9) and begin to sketch only the most basic outline of a food web (Figure 11). Clearly this avenue of investigation warrants further study to elucidate relationships among the various cave taxa.

12 Table 1. Full data for all samples used in isotope analysis.

Month Cave Site Sample identification δ15N δ13C %N %C Notes

Animals Nomember Streak Cave Batrisodes sp. 4.26 -23.52 8.6 - May Big Red Cave Cambala speobia 4.34 -24.03 4.1 30.5 May Big Red Cave Cambala speobia 5.55 -23.50 3.9 28.3 May Big Red Cave Cambala speobia 5.96 -23.15 3.9 25.6 November Mixmaster Cave Cambala speobia 8.65 -22.09 3.9 24.0 Reported as average November Streak Cave Cambala speobia 3.04 -21.56 3.5 20.0 November Streak Cave Cambala speobia 2.96 -24.11 4.7 30.9 May Streak Cave Cambala speobia 4.13 -22.99 4.3 27.8 May Streak Cave Cambala speobia 2.62 -23.39 - - May Big Red Cave Ceuthophilus cunnicularis 5.03 -22.38 10.7 46.7 May Big Red Cave Ceuthophilus cunnicularis 4.44 -23.28 11.4 49.3 May Big Red Cave Ceuthophilus cunnicularis 3.69 -22.40 10.8 46.8 May Big Red Cave Ceuthophilus cunnicularis 3.46 -22.32 10.9 47.0 May Big Red Cave Ceuthophilus cunnicularis 3.52 -22.54 11.2 47.8 November Mixmaster Cave Ceuthophilus cunnicularis 7.87 -22.71 11.3 50.1 November Mixmaster Cave Ceuthophilus cunnicularis 8.38 -22.52 12.5 52.2 Reported as average November Mixmaster Cave Ceuthophilus cunnicularis 8.27 -23.20 10.9 52.9 November Mixmaster Cave Ceuthophilus cunnicularis 8.14 -24.76 10.3 54.6 November Streak Cave Ceuthophilus cunnicularis 2.68 -22.98 10.1 49.3 November Streak Cave Ceuthophilus cunnicularis 3.93 -22.86 10.8 45.9 November Streak Cave Ceuthophilus cunnicularis 3.64 -23.16 10.0 48.3 May Streak Cave Ceuthophilus cunnicularis 2.60 -23.54 9.9 43.8 May Streak Cave Ceuthophilus cunnicularis 1.39 -23.37 8.7 51.3 May Streak Cave Ceuthophilus cunnicularis 2.70 -23.14 10.1 53.0 May Streak Cave Ceuthophilus cunnicularis 2.85 -23.57 9.7 50.3 May Big Red Cave Ceuthophilus secretus 0.58 -23.54 12.4 51.4 Reported as average May Big Red Cave Ceuthophilus secretus 0.82 -23.92 10.4 46.6 May Big Red Cave Ceuthophilus secretus 1.77 -24.84 12.5 56.0 May Big Red Cave Ceuthophilus secretus 0.81 -25.36 10.0 52.8 May Big Red Cave Ceuthophilus secretus 0.93 -24.76 9.9 55.9 November Mixmaster Cave Ceuthophilus secretus 3.31 -23.83 12.5 45.0 November Mixmaster Cave Ceuthophilus secretus 5.90 -24.48 8.1 53.5 November Mixmaster Cave Ceuthophilus secretus 4.26 -24.28 11.3 54.0

Continued on next page 13 Table 1. Continued.

Month Cave Site Sample identification δ15N δ13C %N %C Notes

November Streak Cave Ceuthophilus secretus 1.15 -19.72 6.9 52.7 November Streak Cave Ceuthophilus secretus 2.78 -20.46 10.3 48.9 N reported as average May Streak Cave Ceuthophilus secretus 1.52 -24.56 9.8 54.8 May Streak Cave Ceuthophilus secretus 0.02 -24.55 9.5 54.9 May Streak Cave Ceuthophilus secretus 0.82 -24.56 9.1 50.8 May Streak Cave Ceuthophilus secretus 0.43 -24.49 10.6 50.1 May Streak Cave Ceuthophilus secretus 0.82 -24.79 9.8 56.9 November Mixmaster Cave Ceuthophilus sp. 5.04 -23.99 7.6 42.3 May Big Red Cave Cicurina sp. 10.72 -22.29 12.0 49.4 November Mixmaster Cave Cicurina sp. 13.25 -19.86 8.4 30.5 November Streak Cave Cicurina sp. 8.50 -21.73 9.6 36.7 May Big Red Cave Cicurina varians 10.60 -22.40 11.9 44.6 November Mixmaster Cave Cicurina varians 11.48 -22.75 14.4 51.7 November Mixmaster Cave Cicurina varians 11.70 -22.42 12.8 49.3 November Streak Cave Cicurina varians 8.91 -22.09 11.7 42.4 May Streak Cave Cicurina varians 8.29 -21.82 11.1 45.3 May Streak Cave Haplotaxida 5.37 -23.39 6.4 38.7 May Big Red Cave Helicodiscus eigenmanni 3.84 -17.68 1.9 17.6 May Big Red Cave Helicodiscus eigenmanni 4.41 -16.65 1.7 16.8 May Big Red Cave Helicodiscus eigenmanni 5.18 -17.91 2.0 18.3 November Mixmaster Cave Helicodiscus eigenmanni 7.37 -18.07 1.8 17.9 May Streak Cave Helicodiscus eigenmanni 1.84 -17.39 3 21.3 May Big Red Cave Leiobunum townsendii 1.85 -24.68 8.7 50.6 May Big Red Cave Leiobunum townsendii 1.61 -24.92 10.9 49.0 May Big Red Cave Leiobunum townsendii 2.16 -24.56 9.5 47.6 May Big Red Cave Leiobunum townsendii 2.22 -24.80 9.7 54.1 May Big Red Cave Leiobunum townsendii 1.76 -25.18 9.6 51.5 November Mixmaster Cave Leiobunum townsendii 3.20 -23.99 11.3 49.3 May Mixmaster Cave Leiobunum townsendii 1.92 -25.66 8 53.6 May Mixmaster Cave Leiobunum townsendii 1.49 -24.71 10.6 49.6 Reported as average May Mixmaster Cave Leiobunum townsendii 0.64 -25.4 8.9 51.2 November Streak Cave Leiobunum townsendii 3.21 -24.47 11.5 44.8 November Streak Cave Leiobunum townsendii 1.98 -24.30 11.5 45.9 Reported as average November Streak Cave Leiobunum townsendii 6.32 -23.89 11.6 47.1

Continued on next page 14 Table 1. Continued.

Month Cave Site Sample identification δ15N δ13C %N %C Notes

May Streak Cave Leiobunum townsendii 0.51 -26.17 8.7 56.9 May Streak Cave Leiobunum townsendii 0.57 -24.67 9.6 47.5 May Streak Cave Leiobunum townsendii 2.22 -24.01 11.0 47.4 May Streak Cave Leiobunum townsendii 0.98 -25.50 9.2 50.4 May Streak Cave Leiobunum townsendii 0.49 -24.26 8.9 51.1 May Big Red Cave Lithobiomorpha 6.01 -25.00 11.7 48.7 May Streak Cave Mesodon roemeri -0.38 -23.82 6.0 45.7 May Streak Cave Mesodon roemeri 1.81 -22.47 7.4 51.9 Reported as average May Streak Cave Mesodon roemeri -0.76 -24.33 6.0 39.5 May Big Red Cave Rhadine reyesi 5.61 -22.98 9.3 55.7 November Streak Cave Rhadine reyesi 4.12 -24.37 8.3 57.0 Reported as average November Streak Cave Scutigeromorpha 2.44 -22.95 12.1 47.9 November Streak Cave Siphonophora sp. 3.63 -24.97 5.5 42.6 May Big Red Cave Solenopsis invicata 4.37 -18.80 9.6 53.7 November Mixmaster Cave Solenopsis invicata 8.66 -18.64 9.6 57.4 C reported as average November Mixmaster Cave Solenopsis invicata 7.90 -19.33 9.4 52.6 November Streak Cave Solenopsis invicata 4.50 -20.16 9.1 52.3 November Streak Cave Solenopsis invicata 3.96 -19.68 10.8 61.0 May Streak Cave Solenopsis invicata 3.99 -21.12 8.9 58.2 November Streak Cave Speodesmus castellanus 3.27 -21.28 4.8 23.3 November Streak Cave Speodesmus castellanus 1.42 -20.15 5.3 22.8 May Big Red Cave Staphylinidae 9.08 -23.23 10.1 51.3 May Big Red Cave Staphylinidae 8.78 -22.95 10.4 52.1 May Streak Cave Staphylinidae 5.99 -23.37 8.6 49.2 May Streak Cave Staphylinidae 4.98 -23.23 9.8 45.6 May Streak Cave Staphylinidae 7.85 -23.82 8.5 53.3 May Streak Cave Staphylinidae 4.82 -22.24 9.8 40.1

Bacteria November Streak Cave Blue-green bacteria -1.23 -19.15 3.2 38.9

Continued on next page

15 Table 1. Continued.

Month Cave Site Sample identification δ15N δ13C %N %C Notes

C3 Plants November Streak Cave Acalypha phleoides Cav. -2.02 -28.66 1.4 42.6 Reported as average November Streak Cave Coreopsis sp. 1.36 -31.17 3.7 42.1 November Mixmaster Cave Diospyros texana Scheele 0.96 -26.36 1.6 47.5 November Streak Cave Diospyros texana Scheele -1.11 -29.40 1.4 47.7 November Streak Cave Fraxinus texensis (A. Gray) Sarg. -3.17 -28.44 1.1 44.0 Reported as average November Mixmaster Cave Grindelia nuda A.W.Wood 2.88 -28.98 2.8 51.9 November Streak Cave Juniperus ashei J. Buchholz 0.98 -22.97 10.3 46.6 November Streak Cave Juniperus ashei J. Buchholz -0.67 -24.84 1.2 44.6 May Streak Cave Juniperus ashei J. Buchholz -3.79 -25.51 1.1 42.9 November Streak Cave Liverwort 0.23 -24.27 2.9 26.3 May Streak Cave Lupinus texensis Hook. -0.94 -27.39 2.3 35.9 May Streak Cave Panicum cf. depauperatum Muhl. -0.08 -26.53 1.2 35.3 November Mixmaster Cave Quercus fusiformis Small 2.20 -26.92 1.8 48.7 Reported as average November Streak Cave Quercus sinuata Walt. var. breviloba (Torr.) C.H. Müll. -2.73 -28.99 1.3 48.0 May Streak Cave Quercus sinuata Walt. var. breviloba (Torr.) C.H. Müll. -3.27 -28.8 2.6 41.5 November Streak Cave Rhus virens Lindh. ex A. Gray -2.09 -27.75 1.6 48.1 November Streak Cave Rhus virens Lindh. ex A. Gray -2.18 -28.15 0.9 60.9 November Mixmaster Cave Sideroxylon lanuginosum Michx. ssp. oblongifolium (Nutt.) T.D. Pennington 3.27 -27.74 1.7 50.4 Reported as average November Streak Cave Sideroxylon lanuginosum Michx. ssp. oblongifolium (Nutt.) T.D. Pennington -1.24 -30.10 1.4 51.8 November Streak Cave Sophora secundiflora (Ortega) Lag. ex DC. -3.10 -29.22 2.5 48.2 May Streak Cave Tetraneuris linearifolia (Hook.) Greene -3.36 -30.2 1.5 37.7 November Mixmaster Cave Ulmus sp. 0.68 -27.99 1.9 46.3 Reported as average

CAM1 plants November Streak Cave Opuntia phaecantha Engelm. var. major Engelm. -0.06 -13.14 0.7 36.9 November Streak Cave Opuntia phaecantha Engelm. var. major Engelm. 2.65 -12.20 1.1 37.8

Continued on next page

16 Table 1. Continued.

Month Cave Site Sample identification δ15N δ13C %N %C Notes

C4 plants May Streak Cave Poaceae -3.30 -12.53 3 37.4 November Streak Cave Sporobolus ozarkanus Fernald 4.46 -11.27 1.2 43.8 Reported as average November Streak Cave Tridens muticus (Torr.) Nash var. muticus -0.18 -12.96 0.7 45.8 Reported as average

1Crassulacean Acid Metabolism

17 Table 2. Classification, mean±standard deviation, and sample size for isotope samples, grouped data.

Taxon1 δ15N δ13C n

Gastropoda Sigmurethra Polygyridae Mesodon roemeri - TX 0.22±1.38 -23.54±0.96 3 Helicodiscidae Helicodiscus eigenmanni – TP/TB? 5.20±1.55 -17.58±0.64 4 Annelida Clitellata Haplotaxida (Earthworm) - ED 5.37 -23.39 1 Arthropoda Arachnida Opilionida Palpatores Sclerosomatidae Leiobunum townsendii - TP 2.08±1.50 -24.67±0.63 14 Araneae Dictynidae Cicurina spp. - TB 10.82±2.38 -21.29±1.62 3 Cicurina varians - TP 10.67±1.27 -22.42±0.27 4 Diplopoda Spirostreptida Cambalidae Cambala speobia - TB 5.78±2.33 -23.21±0.85 4 Polydesmida Fuhrmannodesmidae Speodesmus castellanus - TB 2.35±1.31 -20.72±0.80 2

Continued on next page.

18 Table 2. Continued.

Taxon1 δ15N δ13C n

Siphonophorida Siphonophoridae Siphonophora? sp. – TB? 3.63 -24.97 1 Chilopoda Lithobiomorpha – TP? 6.01 -25.00 1 Scutigeromorpha Scutigeridae – TP? 2.44 -22.95 1 Insecta Orthoptera Rhaphidophoridae Ceuthophilus cunnicularis - TB/TP? 4.54±2.32 -23.05±0.64 16 Ceuthophilus secretus - TP 1.73±1.64 -23.88±1.61 15 Ceuthophilus sp. – TP? 5.04 -23.99 1 Coleoptera Carabidae Rhadine reyesi - TB 4.87±1.05 -23.68±0.98 2 Staphylinidae Pselaphinae Batrisodes sp. – TB 4.26 -23.52 1 Staphylinidae - TP 7.40±2.65 -23.40±0.56 3 Hymenoptera Formicidae Solenopsis invicata - AC 5.56±2.13 -19.62±0.92 6

C3 Plants - EP -0.78±2.13 -27.74±2.03 22 C4 Plants - EP 0.36±3.90 -12.25±0.88 3 CAM2 Plants - EP 1.30±1.92 -12.67±0.67 2 Blue-green bacteria - EP -1.23 -19.15 1

1AC=Accidental, ED=Edaphobite, EP=Epigean, TB= Troglobite, TP=Troglophile, TX=Trogloxene; 2Crassulacean Acid Metabolism

19

Table 3. Nonparametric tests for possible differences between sites (caves) and between seasons (months) for carbon and nitrogen isotopes (‰) in Ceuthophilus cunnicularis and Ceuthophilus sectretus.

Cave or Mean Mean Species Month δ13C δ15N n df Test Statistic1 P

C. cunnicularis Big Red -22.58 5 Mixmaster -23.30 4 Streak -23.23 7 2 5.1703 0.0754

C. secretus Big Red -24.48 5 Mixmaster -24.20 3 Streak -23.30 7 2 1.6582 0.4364

C. cunnicularis Big Red 4.03 5 Mixmaster 8.17 4 Streak 2.83 7 2 10.834 0.0044

C. secretus Big Red 0.98 5 Mixmaster 4.49 3 Streak 1.08 7 2 6.803 0.0333

C. cunnicularis May -22.95 9 November -23.17 7 57.0 0.8352

C. secretus May -24.54 10 November -22.55 5 60.0 0.0314

C. cunnicularis May 3.30 9 November 6.13 7 79.0 0.0443

C. secretus May 0.85 10 November 3.48 5 63.0 0.0152

1For comparisons among caves, the Kruskal-Wallis Test was used, and the test statistic is a χ2 value. For comparisons between months, the Mann-Whitney Test (=Wilcoxon Rank-Sum Test) was used, and the test statistic is U, with P values based on the two- tailed t approximation. In both cases, mean ranks are evaluated in the test, however we present mean isotopic values here to facilitate interpretation of the data.

20 A.

B.

Figure 1. A. Finnigan Mat 252 flow through isotope ratio mass spectrometer shown with peripherals, including the Elemental Analyzer on the right (see figure 3); B. Example of a partial chromatogram for nitrogen isotope results.

21

Figure 2. Close up of the isotope ratio mass spectrometer, showing inlet/source area, flight tube, and electromagnet.

22

Figure 3. Sallie Greenberg loading samples into automatic sample carousel of the Carlo Erba NC 2500 Elemental Analyzer used to combust organic matter to N2 and CO2 for isotope analysis.

23 A.

B.

Figure 4. A. Agate mortar and pestle used to grind up samples; B. Desiccator with Drierite ® absorbent used for storage of dried samples.

24

Figure 5. In preparation for elemental and isotopic measurement, samples were weighed into tin capsules using an electronic six-place Mettler AT balance, as shown above.

25

Figure 6. Distribution of N and C isotope results for various taxa sampled at Big Red, Streak, and Mixmaster caves in May and November at Fort Hood, Texas.

26

Figure 7. Mean and standard deviations for δ15N and δ13C values for all taxa examined. Sample size given in Table 2.

27

Figure 8. Gray area encompasses mean isotope values for cave animals, excluding H. eigenmanni, which appears to utilize a differing food source and S. invicta, which was collected from epigean habitats. The generally tight distribution of taxa suggests a relatively linear food chain, with dependence of taxa at higher trophic levels upon cavernicoles at lower trophic levels.

28

Figure 9. Dashed horizontal lines are visually fit based on apparent trophic level and knowledge of feeding habits of various taxa (see text for details).

29

Figure 10. Ceuthophilus secretus feeding on Triatoma sp. (Hemiptera: Reduviidae). Photo by Jean Krejca.

30

Figure 11. Generalized food web for caves at Fort Hood, Texas.

31

RADIO TELEMETRY

Radio tracking has been used with insects only in the last few years (Reynolds and Riley 2002) – since the radio telemetry technology has advanced to a point that transmitters under 1 g in weight can be produced. Perhaps the most successful implementation of this technology with insects is Lorch and Gwynne’s (2000) study of migration by the Mormon cricket (Anabrus simplex), a heavy bodied orthopteran in the family Tettigoniidae4 (Figure 12). They utilized 0.85 g transmitters to track the movement of the crickets. Other studies have used transmitters with dobsenfly larvae (Megaloptera: Corydalidae, an aquatic insect) (Hayashi and Nakane 1989), carabid beetles (Riecken and Raths 1996), and large, tropical scarab beetles (Beaudoin-Ollivier et al. 2003). Currently, even smaller transmitters (~0.35 g) are available, making it feasible to evaluate the use of radio telemetry with orthopterans somewhat less robust than A. simplex. For comparison, the relative sizes of A. simplex and C. secretus are shown in Figure 12.

Methods

Adult cave crickets emerging from Big Red Cave or Cicurina Sink at Fort Hood were captured using aquarium nets and were retained in a plastic bucket until a radio transmitter could be attached to their dorsum. We used 0.35 g transmitters (= “tags”) manufactured by Biotrack (Biotrack Ltd, 52 Furzebrook Road, Wareham, Dorset BH20 5AX, United Kingdom) and nominally spaced at about 10 kHz intervals (Table 4) within the 150.000 - 150.999 MHz range, so that multiple animals could be tracked

4 The family Tettigoniidae is fairly closely related to the family Rhaphidophoridae, to which the cave cricket belongs, and both Anabrus simplex and Ceuthophilus secretus are similar in that they are flightless ground dwellers.

32 simultaneously. These tags were the lightest commercially available radio transmitters at the time of our purchase. Prior to use, tags were stored in a refrigerator to conserve battery life. When crickets were captured and ready to receive a transmitter, the transmitter was unwrapped from the manufacturers’ package and the magnet was removed (thus activating the reed switch to initiate transmission). Transmission was tested using the receiver to ensure adequate signal. Of the 35 transmitters, one was found to be faulty and was not used. Tags were then glued to the dorsum of the crickets using either Loctite ® Quick Tite Super Glue Gel (n=13 individuals) or Elmers ® Super Glue (n=21 individuals) as a fixative (earlier trials with dummy transmitters demonstrated that these two glues were effective). We attempted to center transmitters over the legs (Figures 13, 14), but sometimes placement was less optimal (Figure 15). Transmitters were painted using UV-bright paint (Figures 14, 16), crickets were photographed with a ruler to determine length (Figure 17), and the crickets with transmitters were retained briefly for observation. They were released near the point of capture, typically within 3 m of the entrance to Big Red Cave or Cicurina Sink, always at the entrance where they had been initially captured. Total handling time from capture to release varied from 30 to 60 min.

The movements of individual crickets which left the cave to forage were tracked using flexible Yagi antennas (Biotrack) and standard radio tracking receivers, TR-5 Telemetry Scanning-Receivers (Telonics, 932 Impala Ave, Mesa, AZ) with headphones (Telonics RH-1 Headphones) (Figure 18). Crickets were located by following the signal to the area with the highest signal strength, then circling that area to verify the cricket location. Typically we could circle an area 3-6 m in diameter, so cricket locations have this level of accuracy. Specific locations where the crickets were found were marked using GPS receivers (Garmin ® Etrex Vista). These locations were recorded on a field data sheet along with other information such as time, description of habitat (Big Red Cave, Cicurina Sink, edge, grass, woodland), the receiver gain, highest signal strength detected, frequency, and comments (e.g., if the cricket was seen or not, if the cricket was moving or not). Crickets were located prior to sunset and then approximately one hour later. For crickets that did not seem to be moving that night, locations were made

33 approximately every two hours. For crickets that were moving, locations were made approximately every hour.

On many occasions the signal emitted by transmitters that were most likely inside Big Red Cave was actually strongest at an area on the ground near the entrance of the cave. For this reason our dataset contains many false readings of crickets just outside of the cave. There are several reasons why this phenomenon occurred. First, the signal strength reading on the receiver was greatly influenced by the orientation of the antenna to the ground or to the cave entrance (vertical, horizontal, parallel, perpendicular), and it is possible that the entrance pit and gate morphology affected the signal as it was coming out of the cave. This signal bounce, in combination with the variation in how different researchers held the antenna, could easily account for a higher signal strength coming from an area slightly away from the cave entrance. Another factor to be considered is that several of the false surface readings near the cave always came from the same spot on the ground. In several of these cases, that spot on the ground was entirely bare and it was easy to see that there was no cricket there. What was most likely occurring in these situations was that the cricket was located at a place in the cave with a particularly small amount of overburden, thus the signal was actually coming through the ground. At Big Red Cave (Figure 19) we have often observed many crickets roosting in domes located close to the surface (areas shaded gray on map). As we recognized these problems, we began to record signal strength and receiver gain for each reading in order to differentiate crickets that were in the cave versus on the surface.

There was insufficient time to fully integrate the above observations on signal strength into the analysis presented here, but we will use the following criteria in future analyses to determine which readings represented crickets above ground and which represented crickets that were below ground: First, on one occasion we observed an abrupt change in signal strength of a cricket, simultaneously, we observed the same cricket emerge from Big Red Cave. We can use these two signal strengths as one guide for determining whether readings were above or below ground. On another

34 occasion, we simultaneously recorded signal strengths of crickets from inside the cave and from the surface during the daytime. These differences in strengths can be another guide for differentiating cricket location. Finally, we have many observations of signal strengths of transmitters on newly released crickets that we knew were still outside of the cave, and these data can provide a range of values for crickets known to be above ground. Much of the analysis presented below in the results and discussion sections may be biased as a result of the problems in determining whether crickets were below or above ground – these problems will be rectified in future analyses of this dataset using the methods described above.

Accumulated data on individual crickets were later entered into a Filmaker Pro ® database to facilitate examination for possible patterns in movements and activity. Movement and home range analyses were performed in ARC-View® v3.2 using the Animal Movement Extension (Hooge and Eichlaub 1997). All other telemetry data were analyzed using the PopTools 2.5 Excel addin and SPSS® 10.1. All nominal alpha levels were set at 0.05 unless otherwise noted. For each radiolocated cricket, the minimum convex polygon (m2, essentially the home range for the cricket), maximum and mean distance moved per radiolocation, distance from original karst feature, and the total distance of the movement path (m) during tracking were calculated. Using linear regression, we tested if the movement parameters were biased by sample size. If movement parameters were biased we used the residuals of the relationship for all subsequent analyses. To determine if movement parameters or home range differed between sexes, we used t-tests. To determine if cricket size was related to movement parameters or home range, we performed linear regressions of cricket length versus movement parameters. We performed site fidelity tests using the Animal Movement extension in Arc-View. This process uses Monte Carlo simulated movement paths where the direction of travel is randomized and the distance between successive locations is the observed value. To test movement paths, mean squared distance from the center of activity was calculated and then an index of the path linearity was calculated using the observed distance between the origin and terminus of the path divided by the total distance traveled (Spencer, Cameron & Swihart, 1990). The

35 simulation results in dispersed (the distance between the first and last points are greater than that for simulated paths), random (the distance between the first and last points fall within the range of those for simulated paths), or constrained (the distance between the first and last points is shorter than those for simulated paths) paths. For this analysis we used 1000 Monte Carlo simulations. For raw habitat categories, we split the number of observation for each sex and averaged the proportion of use for each category for each sex. To determine if there were differences in between sexes and habitat categories we performed chi-square analyses on overall habitat use and terrestrial foraging habitat use.

Results

General Movement and Home Range Thirty-four crickets were tagged with radio transmitters (Table 4) over the course of the field work (Figure 20). Crickets averaged 18.08±0.22 mm (mean±std. err, n=30) in length, ranging from 15.53 to 20.13 mm. Females (mean =18.23 mm, range 16.11- 20.12, n=17) were generally larger than males (mean =17.87 mm, range 15.53-20.13, n=13) but this difference was not significant (t=0.8449, df =27, p=0.2028 [one-tail two- sample t-Test assuming unequal variances]). We made 1044 radiolocations of 32 crickets between 6 June and 28 June 2004 (Table 5, Figure 21). Overall, females and males had similar home range and movement statistics (Table 5). Home range area (r2 = 0.332, p < 0.001, df = 30), maximum distance (r2 = 0.240, p = 0.004, df = 30), mean distance (r2 = 0.240, p = 0.004, df =30) were biased by sample size. However, total distance moved was not biased (r2 = 0.084, p = 0.108, df = 30). Because the measures were biased we used the residuals of the respective relationships for the remainder of

the analyses. Males had larger home ranges when corrected for sample size (tstat = -

1.102, p = 0.004, df = 16.4) and a greater total distance moved (tstat = -0.852, p = 0.022,

df = 14.7) when compared to females. Neither mean distance moved (tstat = -0.491, p =

0.384, d.f. = 22.3), nor maximum distance moved (tstat = -1.022, p = 0.174, df = 21.2) differed between the sexes. Figure 22 shows there were no relationships between cricket size and the residual variance of home range area (r2 = 0.012, p = 0.574, df =

36 26), maximum distance moved (r2 = 0.059, p =0.212, df = 26), or mean distance moved (r2 = 0.013, p = 0.556, df = 26). Locations, home ranges, and movement paths for all crickets are given in Appendix 2.

Distance from Karst Features Male and female crickets tended to be, on average, only 11.1 and 13.3 m, respectively, from their home karst feature (Table 5). Three crickets switched between Circurina Sink and Big Red Cave and in all cases, these crickets were males (males 4, 17, and 32). Generally, crickets stayed in or within 30 m of their home karst feature.

Movement Paths Movement paths were predominantly constrained, whereby 14 of 19 females and 9 of 13 male crickets made constrained movements (Table 5). Average values for 23 constrained movement paths were 453.2±182.9 m2 for home range area, 29.5±5.6 m for maximum distance moved, 4.3±0.6 m for mean distance moved, and 160.7±38.8 m for total distance moved. Average values for nine random movement paths were 395.9±241.7 m2 for home range area, 21.2±8.6 m for maximum distance moved, 4.1±1.2 m for mean distance moved, and 102.4±45.8 m for total distance moved. Constrained and random movement paths did not differ significantly in their residual

variance for home range area (tstat = -1.775, p = 0.313, df = 21.1), maximum distance

moved (tstat = -0.488, p = 0.222, df = 19.3), and mean distance moved (tstat = -0.547, p = 0.949, df = 13.601).

Habitat Use and Foraging Habitat Of the 1044 radiolocations we recorded habitat for 899. Overall habitat use in order of most to least use was caves at 54.8%±5.7%, woodlands at 21.4%±4.1%, grasslands at 7.9%± 1.6%, sinks at 6.9%±3.5%, edges at 6.1%±2.1%, and cacti stands at 3.1%±0.9% (Table 6, Figure 23a). Habitat use was significantly different from equality of use (i.e., ~20% for each category) with caves and woodlands being used more, and all other categories being used less (chi-square = 945.6, p < 0.001, df = 5) When broken down by sex there were 513 observations for females which

37 corresponded to use of caves at 59.8%±7.6%, woodlands at 18.2%±5.6%, grasslands at 9.6%±2.5%, sinks at 5.2%±4.9%, edges at 3.9%±1.4%, and cacti stands at 3.2%±1.1% (Table 6, Figure 23a). For the 386 observations of males, use was categorized as caves at 47.9%±8.5%, woodlands at 25.7%±6.1%, grasslands at 5.5%±1.6%, sinks at 9.1%±4.9%, edges at 9.0%±4.5%, and cacti stands at 2.8%± 1.5%. By sex overall use was significantly different among habitat categories with females using caves, sinks, and woodlands less than expected and males using caves, sinks, and woodlands more than expected (chi-square = 31.5, p < 0.001, df = 5).

Because we are interested in surface foraging habitat we removed the karst habitat features and repeated the analysis. This resulted in 369 observations for which overall use was distributed as 54.3%±7.2% woodland, 22.4%±7.1% grassland, 13.8%±4.5% edge, and 9.4%±2.7% cacti stands. Overall foraging habitat use was significantly different from equality (chi-square = 278.7, p .001, df = 3) with woodlands being used more frequently than expected and grasslands, edges, and cacti stands being used less frequently (Table 6, Figure 23b). When partitioned by sex, foraging habitat use for 191 observations of females was distributed as; woodlands at 46.7%±8.2%, grasslands at 30.0%±7.7%, edges at 12.8%±5.9%, and cacti stands at 10.3%±3.6% (Table 6). For the 178 observations of males foraging habitat use was distributed as; woodlands at 63.4%±7.8%, edges at 14.9%±4.0%, grasslands at 13.2%±4.0%, and cacti stands at 8.4%±3.8% (Table 6). There were significant differences in terrestrial foraging habitat between the sexes (chi-square = 10.9, p = 0.012, df = 3). Woodlands were used more than predicted for males and less than predicted for females and grasslands were used more than predicted for females and less than predicted for males (Table 6, Figure 23b).

Discussion

The home ranges of cave crickets (defined as the minimum convex polygon) tracked during this study averaged 437 m2 and were as large as 4025 m2 (0.99 acres) for a single cricket over multiple nights. No previous studies have obtained home range

38 area data for cave crickets. However, we can compare the maximum distance moved by each individual with data obtained from an earlier mark recaptures study by Taylor et al (2003), in which crickets were painted as they left Big Red Cave, then relocated at various times throughout the night, and these relocations were plotted as distance from the cave entrance corrected for time and area searched (see figure 24a). That study found that 81.3% of the crickets were foraging less than 70 m from the cave, whereas we found that 81.25% of the crickets had a maximum foraging distance of 44 m or less (figure 24b). In general, telemetry data from our study indicate that individual cave crickets have high site fidelity, generally remaining within 12 m of their home karst feature (also see method caveat at end of discussion). The radio tracking method may have underrepresented the distance that crickets travel from the cave, and this may be related to sample size and/or the encumbrance of the transmitter. The earlier painting study (Taylor et al. 2003) marked many more crickets (over 1,000 individuals) than the radio telemetry study (34 individuals), resulting in generally less sampling of the total distribution of possible distances from the cave in the radio telemetry study. This would tend to lead to fewer observations of rare events such as distant traveling crickets. Additionally, the radio transmitter, while weighing very little (~0.35 g) for its capability, is a fairly large percentage of the body weight of the average cricket, and may affect the distance a cricket can efficiently travel. While we did not have appropriate scales for field weighing of crickets, a study by Studier et al. (1987) gives live weights of unfed Ceuthophilus stygius, a species roughly similar in size to Ceuthophilus secretus, as averaging 1.056 g for females and 1.304 g for males. One cricket that was relocated inside Big Red Cave had mud smeared on the top of the transmitter, in a position that suggests it was bumped against obstacles in the cave, which could effect the cricket’s behavior (figure 25). One similarity of the two studies worth noting is that one radio located cricket was found on the far side of a large gravel and bedrock road (see figure 20), and several painted crickets were also found on the far side of this same road by Taylor et al. (2003, figure 86). Thus, both of studies indicate that crickets will cross roads.

39 When we examined the effects of cricket size and sex on home range and movement statistics, it was found that cricket size was not related to the residual variation in home range area or to other movement statistics (mean, maximum and total distance moved). Cricket sex, however, when corrected for the bias in sample size, differed between sexes, with males having larger home ranges and greater total distance moved. Taylor et al. (2003, submitted) did not examine sex differences in foraging distance, but we reexamined those data and found that although mean distance for adult males (45.55 m, n=78) was greater than that for adult females (41.12 m, n=112), this difference was not significant (two tailed t-test, unequal variances: df=144, t=1.15, p=0.2538). That the three crickets which switched between karst features were all males further underscores the importance of sex as a determinant of behavior.

Our expectation based on available data on cavernicolous rhaphidophorid behavior was that crickets would leave the cave at dusk, forage outside of the cave, then return to the cave by morning (e.g., Barr 1963, Benoit et al. 2004, Campbell 1976, Park and Reichle 1963, Peck 1976, Reichle et al. 1965, Richards 1965), perhaps with a cycle of approximately 3-6 nights. As early as 1933, such behavior was observed by Husson and Remy for Dolichopoda sp. in Macedonia (Husson 1971), and such “nocturnal sorties” are typically thought to be for the purpose of feeding upon vegetation: “Orthoptères sortent régulièrement chaque nuit de leur retraite obscure pour s'approvisionner dans le domaine épigé (présence de végétaux dans le contenu intestinal).” (Husson 1971). The present study, however, reveals an exciting new observation – that three out of 34 crickets (9%) move from cave to cave. This observation suggests that this kind of dispersal may be much more common than previously reported, and that cave crickets from karst features within ~100 m of each other almost certainly represent a single population. These observations suggest that management of caves to protect invertebrate fauna should take into account nearby karst features, as these may play a significant role in the genetic diversity of cave crickets in target caves.

40 At least one study examining genetic structure of Rhaphidiphorid cave crickets in Italy showed that in general there was high population subdivision between sites, with

the mean FST value of 0.34, indicating very little mixing across populations (Allegrucci et al. 1996). That study, however, examined populations that were generally very distant and even occurring on separate islands (up to 125 km apart). Also in that study the number of migrants per generation (Nm) was estimated for sites as close as 100m apart, and was found to be 28. The highest number of migrants recorded was for two sites 20 km apart, but surrounded by mesic woods that are presumably hospitable habitat for dispersing, and this value was Nm = 54. The high values detected by Allegrucci et al. (1996) are in agreement with our observation that it is relatively common (3 out of 34 individuals) for individual crickets to move between features that are less than 100m apart.

Another unexpected result is that crickets that inhabited Big Red Cave (they were captured at the cave entrance as they exited the cave, after which a transmitter was attached) were found to occasionally spend the night on the surface, in areas with no obvious karst features. For example, crickets successfully stayed overnight in open grassy areas with small rocks as well as in dense wooded areas with thick leaf litter, woody debris and rocks. These observations indicate that it is important to preserve a variety of surface habitats around the cave that may be important in facilitating cricket movement between caves, thereby maintaining their genetic diversity and normal behavioral patterns.

During the night, crickets more often remained in the cave, as reflected in the comparison of habitat use. When the cave and sinkhole habitats were excluded from this analysis, it was found that male crickets foraged in the woodland habitat more than females, and females foraged in grassland areas more than males. We are unaware of any previous study that has demonstrated a differential use of habitat types for foraging Rhaphidophoridae. Our (Taylor et al. 2003, Taylor et al. submitted) early study using painted crickets was apparently biased against sampling in wooded areas, probably because these areas are very dense and difficult for researchers to travel through.

41

The movement paths for crickets were both constrained and random, depending on the individual, but when the residual variances in the spatial measures (such as maximum and mean distance moved) were compared, they were not significantly different. This indicates that all the crickets, whether they had random or constrained movement paths, had no detectable differences (between path types) in the distances they moved.

Finally, there are problems with the telemetry study that still need to be resolved. As discussed in the methods section, we appeared to locate many more crickets on the surface than were actually there (that is, a number of these data points should be classified as in-cave occurrences) because of problems differentiating signals that were below-ground from those that were actually above-ground. This created a bias in nearly all of the analyses towards finding a shorter mean distance moved and biased values for habitat use. In future analyses of this data set, these problems will be resolved by using the signal strength and other cues to determine whether each cricket location is actually below ground versus above ground (see earlier discussion).

Nonetheless, this telemetry study should be considered one of the most successful terrestrial insect radio tracking studies conducted to date, when compared with other such studies (e.g., Beaudoin-Ollivier et al. 2003, Lorch and Gwynne 2000, Riecken and Raths 1996), and has provided new insights that will facilitate better management of cave communities throughout central Texas as well as a better understanding of the biology of Ceuthophilus secretus.

42

Table 4. Lengths of cave crickets (Ceuthophilus secretus) to which radio transmitters were attached. Data are unavailable for four individuals.

Transmitter Frequency (MHz) Cricket Length (mm) Cricket Sex

150.010 18.59 male 150.037 - male 150.051 16.10 female 150.062 16.74 female 150.089 17.12 male 150.099 18.01 male 150.108 18.00 male 150.133 20.13 male 150.154 17.01 male 150.164 18.46 female 150.181 17.72 female 150.199 17.70 female 150.214 19.97 female 150.221 15.53 male 150.232 17.01 male 150.256 - female 150.266 17.75 female 150.318 19.32 female 150.359 16.26 female 150.397 17.57 male 150.417 18.68 female 150.442 17.76 female 150.553 17.31 male 150.574 18.19 male 150.587 20.12 female 150.631 - female 150.691 19.80 female 150.713 - male 150.747 18.49 male 150.792 18.58 female 150.851 19.01 female 150.868 18.60 female 150.874 19.29 male 150.910 17.40 female

43

Table 5. Cricket size (cm), minimum convex polygon (MCP) area (m2), maximum (MAXD), mean (MEAND), and total distance moved (TOTD) between radiolocations, mean distance from home karst feature (MEANKAR) number of radio locations, movement path type and proportion of simulated paths with greater distances than actual paths (Prop) for 32 Ceuthophilus secretus radiolocated from 6 June 2004 and 28 June 2004, in the vicinity of Big Red Cave, Fort Hood, Coryell County, Texas.

Females

Cricket Size MCP MAXD MEAND MEANKAR TOTD n Path Type Prop

1 68.0 10.4 1.6 1.5 40.7 27 Constrained 96.4 2 56.0 13.0 1.3 0.9 81.2 63 Constrained 99.9 5 19.0 98.5 22.6 3.6 3.9 155.5 44 Constrained 99.9 7 17.7 466.0 70.8 3.9 8.0 161.0 42 Constrained 95.5 9 17.8 347.5 28.2 3.9 4.5 176.7 46 Constrained 99.8 10 16.3 112.5 11.4 3.1 2.2 131.0 43 Constrained 99.9 13 20.1 705.5 39.1 3.7 11.5 149.9 41 Random 84.6 14 19.8 113.0 17.0 1.9 62.6 76.3 41 Constrained 97.8 18 19.3 1314.0 44.0 10.6 42.6 402.1 39 Constrained 99.2 19 18.6 478.5 83.0 7.2 3.6 239.1 34 Constrained 99.9 20 17.4 964.5 51.9 6.8 50.1 237.5 36 Constrained 99.5 21 18.7 314.5 20.6 5.5 8.1 115.9 22 Random 73.7 22 18.5 32.5 10.0 2.3 1.2 42.8 20 Constrained 98.9 23 16.1 6.0 3.6 0.9 0.6 14.6 18 Random 68.9 24 17.8 168.5 19.6 3.9 3.9 81.8 22 Constrained 98.7 27 18.6 2.0 2.2 0.3 0.7 4.2 15 Random 89.5 28 16.7 4.0 3.6 0.7 0.6 10.3 16 Random 38.6 29 17.7 6.0 12.5 2.2 1.1 30.5 15 Constrained 99.9 31 20.0 100.0 13.6 6.9 3.5 117.4 18 Constrained 99.9

Mean 18.2 282.0 25.1 3.7 11.1 119.4 StDev. 1.2 364.5 22.8 2.7 18.6 99.0

Continued on next page

44

Table 5. Continued.

Males

Cricket Size MCP MAXD MEAND MEANKAR TOTD n Path Type Prop

3 141.0 15.7 1.2 15.2 58.2 48 Constrained 96.8 4 4025.0 83.0 11.8 68.0 883.1 76 Constrained 99.9 6 17.1 31.5 10.8 1.2 0.8 48.7 42 Constrained 99.2 8 17.3 222.5 21.1 3.0 2.5 128.2 44 Constrained 99.9 11 20.1 53.0 10.8 2.9 2.4 64.4 23 Constrained 99.9 12 17.6 1435.5 89.7 9.2 6.0 349.6 39 Constrained 99.9 15 17.0 65.5 11.3 2.7 2.5 92.7 35 Constrained 99.9 17 18.6 2231.5 83.0 11.7 49.8 444.6 39 Random 82.6 25 18.0 9.0 5.4 1.2 1.0 22.4 20 Constrained 98.2 26 19.3 73.5 12.2 4.2 5.5 62.8 16 Random 87.7 30 17.0 115.0 23.0 5.4 3.5 75.8 15 Constrained 99.9 32 18.0 182.5 17.2 6.4 12.0 77.3 13 Random 61.0 34 15.5 43.5 9.2 3.2 3.6 41.7 14 Random 94.2

Mean 17.8 663.8 30.2 4.9 13.3 180.7 StDev. 1.2 1212.6 31.8 3.8 21.0 246.3

Grand Mean 18.1 437.1 27.2 4.2 12.0 144.3 StDev. 1.2 826.2 26.4 3.2 19.3 173.5

45

Table 6. Overall and epigean foraging habitat use of Ceuthophilus secretus radiolocated from 6 June 2004 to 28 June 2004 near Big Red Cave, Fort Hood, Coryell County, Texas. Each habitat category is subdivided by sex and includes the total number of observations, mean proportion of individual use and the standard error of individual use.

Overall Habitat Use

Cave Edge Grassland Sink Woodland Cacti Total

Females Observations 294 18 50 28 103 20 513 Mean Proportions 0.598 0.039 0.096 0.052 0.182 0.032 S.E. 0.076 0.014 0.025 0.049 0.056 0.011

Males Observations 167 17 26 41 123 12 386 Mean Proportions 0.479 0.090 0.055 0.091 0.257 0.028 S.E. 0.085 0.045 0.016 0.049 0.061 0.015

Overall Observations 461 35 76 69 226 32 899 Mean Proportions 0.548 0.061 0.079 0.069 0.214 0.031 S.E. 0.057 0.021 0.016 0.035 0.041 0.009

Epigean Foraging Habitat Use

Edge Grassland Woodland Cacti Total

Females Observations 18 50 103 20 191 Mean Proportions 0.129 0.300 0.468 0.103 S.E. 0.059 0.077 0.082 0.036

Males Observations 17 26 123 12 178 Mean Proportions 0.149 0.132 0.635 0.084 S.E. 0.073 0.040 0.078 0.038

Overall Observations 35 76 226 32 369 Mean Proportions 0.138 0.224 0.543 0.095 S.E. 0.045 0.071 0.072 0.027

46

Figure 12. Relative size of the Mormon cricket (Anabrus simplex), a large orthopteran used in a radio tracking study by Lorch and Gwynne (2000) and Ceuthophilus secretus, to which radio transmitters were attached in the present study.

47

A.

B.

Figure 13. A. Adult female of Ceuthophilus secretus with ~0.45 g radio transmitter attached during preliminary trials; B. Adult female Ceuthophilus secretus with 0.35 g radio transmitter attached (frequency: 150.442 MHz), during present study. Photos by Jean K. Krejca.

48

Figure 14. Two cave crickets (left, male; center, female) with radio transmitters painted with UV bright paint to facilitate relocation. An unpainted female is present in the upper right. Photo by Jean K. Krejca.

49

Figure 15. An adult female cave cricket with attached radio transmitter (frequency: 150.226 MHz). Placement of the transmitter on this individual is somewhat less than optimal, as the weight is not centered above the legs, and glue may interfere somewhat with flexion of the abdominal segments. Nonetheless, the cricket appeared to move about normally. Figures 13 and 14 show optimal placement. Photo by Jean K. Krejca.

50

Figure 16. Cricket with UV-bright painted transmitter moving through grass. Photo by Jean K. Krejca.

51

Figure 17. Sample of photograph used to measure cricket length. Small marks on ruler are millimeters. Cricket length is measured along dorsal midline from vertex of head to posterior end of abdomen excluding genital segments. This specimen is a male, with radio transmitter attached. Relative lengths of the two dark lines, drawn in Adobe Illustrator, are used to calculate cricket length. Original photo by Jean K. Krejca.

52

Figure 18. Erin Raboin with full tracking gear on a moonlit night. Gear includes flexible Yagi antenna (right), TR-5 Telemetry Scanning-Receiver (lower left) and Telonics RH-1 headphones. Photo by Jean K. Krejca.

53

Figure 19. Big Red Cave (Fort Hood, Coryell County, Texas), showing locations of domes where crickets were observed during radio tracking study.

54

Figure 20. Map of all cricket radiolocations made between 6 June 2004 and 28 June 2004 in the vicinity of Big Red Cave and Cicurina Sink, Coryell County, Fort Hood, Texas.

55

36

34

32

30

28

26

24

22

20

18

16

Cricket Number Cricket 14

12

10

8

6

4

2

0 6/5/2004 6/9/2004 6/13/2004 6/17/2004 6/21/2004 6/25/2004 6/29/2004 Date

Figure 21. Line plot of the duration (days) each Ceuthophilus secretus was radiolocated.

56

Figure 22. Scatter plots of home range and movement statistics versus size (body length) for 30 radio equipped Ceuthophilus secretus tracked from 6 June 2004 to 28 June 2004. Filled circles are females and open circles are males.

57 0.8 Females 0.7 Males

0.6

0.5

0.4

0.3 Proportion of Use

0.2

0.1

0 Cave Edge Grass Sink Woodland Cacti Habitat Type A.

0.8 Females 0.7 Males

0.6

0.5

0.4

0.3 ProportionUse of

0.2

0.1

0 Edge Grass Woodland Cacti Habitat Type B.

Figure 23 Relative habitat use data from radiolocations of 34 Ceuthophilus secretus. A. Grouped by sex; B. Grouped by sex with cave and sink data removed, allowing focus on the epigean habitat used for foraging. Bars are represented with ± one standard error of the mean proportion of use.

58

Figure 24. Comparison of cricket distance data from mark recapture study of Taylor et al. (2003) and the present radio tracking study. A. Cumulative proportion of the total crickets per minute adjusted for the total area at each distance interval from cave (modified from Taylor et al. 2003); B. Maximum distance from home cave for each cricket with telemetry data.

59

Figure 25. Adult female Ceuthophilus secretus (frequency: 150.164 MHz, cricket #22) located inside Big Red Cave, Coryell County, Texas on 21 June, 2004. Note soil on anterior surface of radio transmitter.

60

Acknowledgements

For assistance with radio tracking field work we thank S. Casey Funderburk, Chris Krejca, Vivian Loftin, and Ian Quigley. Plant identifications were made by Geoff Levin (Center for Biodiversity, Illinois Natural History Survey), and curation of plant vouchers was facilitated by Rick Phillippe (Center for Biodiversity, Illinois Natural History Survey).

Literature Cited

Allegrucci, G., M. G. Minasi and V. Sbordoni. 1996. Patterns of gene flow and genetic structure in cave-dwelling crickets of the Tuscan endemic, Dolichopoda schiavazzi (Orthoptora, Rhaphidophoridae). Heredity 78:665-673.

Barr, T. C., Jr. 1963. The fauna of Mammoth Cave: A study in community composition and evolution. Bulletin of the Ecological Society of America 44(1):126.

Beaudoin-Ollivier, L., F. Bonaccorso, M. Aloysius, and M. Kasiki. 2003. Flight movement of Scapanes australis australis (Boisduval) (Coleoptera: Scarabaeidae: Dynastinae) in Papua New Guinea: a radiotelemetry study. Australian Journal of Entomology 42:367- 372.

Benoit, J. B., J. A. Yoder, L. W. Zettler, and H. H. Hobbs, III. 2004. Mycoflora of a trogloxenic cave cricket, Hadenoecus cumberlandicus (Orthoptera: Rhaphidophoridae), from two small caves in northeastern Kentucky. Annals of the Entomological Society of America 97(5):989-993.

Blinn, W.C. 1963. Ecology of the land snails, Mesodon thyroidus and Allogona profunda. Ecology 44(3):498-505.

Blüthgen, N., G. Gebauer, and K. Fiedler. 2003. Disentangling a rainforest food web using stable istopes: dietary diversity in a species-rich ant community. Oecologia 137:426-435.

Briones, M. J. I., R. Bol, D. Sleep, L. Sampedro, and D. Allen. 1999. A dynamic study of earthworm feeding ecology using stable isotopes. Rapid Communications in Mass Spectrometry 12:1300-304.

Cabana, G. and J. B. Rasmussen. 1994. Modeling food chain structure and contaminant bioaccumulation using stable nitrogen isotopes. Nature 372(6503):255- 257.

61

Campbell, G. D. 1976. Activity rhythm of the cave cricket, Ceuthophilus conicaudus Hubbell. American Midland Naturalist 96:350-366.

Carmichael, R. H., D. Rutecki, B. Annett, E. Gaines, and I. Valiela. 2004. Position of horseshoe crabs in estuarine food webs: N and C stable istopic study of foraging ranges and diet composition. Journal of Experimental Marine Biology and Ecology 299:231- 253.

Ceballos, G., T. H. Fleming, C. Chavez, and J. Nassar. 1997. Population dynamics of Leptonycteris curasoae (Chiroptera: Phyllostomidae) in Jalisco, Mexico. Journal of Mammalogy 78(4):1220-1230.

Coplen, T. B, C. Kendall and J. Hopple. 1983. Comparison of stable isotope reference samples. Nature 302:236-238.

Correll, D. S. and M. C. Johnston. 1970. Manual of the vascular plants of Texas. Texas Research Foundation, Renner, Texas. 188 pp.

Culver, D.C. 1982. Cave life-Evolution and ecology. Cambridge, Massachusetts, Harvard University Press, 189 p.

DeNiro, M. J. and S. Epstein. 1978. Influence of diet on the distribution of carbon isotopes in animals. Geochemica et Cosmochimica Acta 42:341-351.

DeNiro, M. J. and S. Epstein. 1981. Influence of diet on the distribution of nitrogen isotopes in animals. Geochimica et Cosmochimica Acta 45:341-351.

Diggs, George M., Jr., Barney L. Lipscomb, and Robert J. O'Kennon. 1999. Shinner & Mahler's illustrated flora of north central Texas. Sida, Botanical Miscellany (Botanical Research Institute of Texas, Fort Worth, Texas) 16:1-1626.

Elliott, W. R. 1992 (revised 1993). Fire ants and endangered cave invertebrates: a control and ecological study. Final report submitted to Texas Parks and Wildlife, Austin, Texas. 33 pp.

Elliott, W. R. 1994. Community ecology of three caves in Williamson County, Texas: A three-year summary. 1993 Annual Report for Simon Development Co., Inc., U. S. Fish and Wildlife Service, and Texas Parks & Wildlife Department. 46 pp.

Elliott, William R. 2004. Speodesmus cave millipedes. Four new species from Central Texas (Diplopoda: Polydesmida: Polydesmidae). Texas Memorial Museum, Speleological Monographs 6:163-174.

62 Fagen, W. F., E. Siemann, C. Mitter, R. F. Denno, A. F. Huberty, H. A. Woods, and J. J. Elser. 2002. Nitrogen in insects: Implications for trophic complexity and species diversification. The American Naturalist 160(6):784-802.

Field, M. S. 2002. A lexicon of cave and karst terminology with special reference to environmental karst hydrology. National Center for Environmental Assessment, U. S. Environmental Protection Agency. EPA/600/R-02/003. 214 pp.

Fry, B., A. Joern, and P. L. Parker. 1978. Grasshopper food web analysis: Use of carbon isotope ratios to examine feeding relationships among terrestrial herbivores. Ecology 59(3):498-506.

Fry, B. and E. B. Sherr. 1984. δ13C measurements as indicators of carbon flow in marine and freshwater ecosystems. Contrib. Mar. Sci. 27:13-47.

Graening, G. O. 2000. Ecosystem dynamics of an Ozark cave. Ph.D. dissertation, University of Arkansas, Fayetteville. vi + 99 pp.

Graening, G. O. and A. V. Brown. 2000. Trophic dynamics and pollution effects in Cave Springs Cave, Arkansas. Final report submitted to the Arkansas Natural Heritage Commission. iv + 44 pp.

Graening, G. O. and A. V. Brown. 2003. Ecosystem dynamics and pollution effects in an Ozark cave stream. Journal of the American Water Resources Association 39(6):1497-1507.

Griffiths, H. (ed.). 1998. Stable isotopes: Integration of biological, ecological and geochemical processes. Herndon, Virginia, BIOS Scientific, 438 pp.

Hanson, B., G. F. Farifullina, S. D. Lindblom, A. Wangeline, A. Ackley, K. Kramer, A. P. Norton, C. B. Lawrence, and E. A. H. Pilon-Smits. 2003. Selenium accumulation protects Brassica juncea from invertebrate herbivory and fungal infection. New Phytologist 159(2):461-469.

Hatch, Stephan L., Kancheepuram N. Gandhi, and Larry E. Brown. 2001. Checklist of the vascular plants of Texas. http://www.csdl.tamu.edu/FLORA/taes/tracy/regeco.html

Hayashi, F. and M. Nakane. 1989. Radio tracking and activity monitoring of the dobsonfly larva, Protohermes grandis (Megaloptera: Corydalidae). Oecologia 78:468- 472.

Hobson, K. A., A. Fisk, N. Karnovsky, M. Holst, J.-M. Gagnon, and M. Fortier. A stable isotope (δ13C, δ15N) model for the North Water food web: implications for evaluating trophodynamics and the flow of energy and contaminants. Deep-Sea Research Part II 49:5131-5150.

63 Hocking, M. D., and T. E. Reimchen. 2002. Salmon-derived nitrogen in terrestrial invertebrates from coniferous forests of the Pacific Northwest. BMC Ecology 2:4. [online at: http://www.biomedcentral.com/1472-6785/2/4]

Hooge, P. N., and B. Eichenlaub. 1997. Animal movement extension to Arcview ver 1.1. Alaska Bio. Sci. Cen., U. S. Geological Survey., Anchorage, Alaska, USA.

Howarth, F.G. 1983. Ecology of cave arthropods. Annual Review of Entomology 28:365-389.

Humphreys, W. F. 1999. Physico-chemical profile and energy fixation in Bundera Sinkhole, an anchialine remiped habitat in north-western Australia. Journal of the Royal Society of Western Australia 82(3):89-98.

Husson, R. 1971. Rythmes biologiques et vie cavernicole. Bulletin de la Société Zoologique de France 96:301-316.

Integrated Taxonomic Information System (ITIS). 2004. On-line database. http://www.itis.usda.gov

Koch, P. L. and D. L. Phillips. 2002. Incorporating concentration dependence in stable isotope mixing models: A reply to Robbins, Hilderbrand and Farley (2002). Oecologia 133:14-18.

Lajtha, K. and R. H. Michener (eds.). 1994. Stable isotopes in ecology and environmental science. Blackwell Scientific Publications, Oxford, 316 p.

LePage, M., L. Abbadie and A. Mariotti. 1993. Food habits of sympatric termite species (Isoptera, Macrotermitinae) as determined by stable carbond isotope analysis in a Guinean savanna (Lamto, Côte d’Ivoire). Journal of Tropical Ecology 9:303-311.

Lorch, P. D. and D. T. Gwynne. 2000. Radio-telemetric evidence of migration in the gregarious but not the solitary morph of the Mormon cricket (Anabrus simplex: Orthoptera: Tettigoniidae). Naturwissenschaften 87:370-372.

Markow, T. A., S. Anwar, and E. Pfeiler. 2000. Stable isotope ratios of carbon and nitrogen in natural populations of Drosophila species and their hosts. Functional Ecology 14:261-266.

McNabb, D. M., J. Halaj, and D. H. Wise. 2001. Inferring trophic positions of generalist predators and their linkage to the detrital food web in agroecosystems: A stable isotope analysis. Pedobiologia 45(4):289-297.

Michener, R. H. and D. M. Schell. 1994. Stable isotope ratios as tracers in marine and aquatic food webs. Pages 138-157 in Lajtha, K. and R. H. Michener (eds.). Stable

64 isotopes in ecology and environmental science. Blackwell Scientific Publications, Oxford.

Nardi, J. B., R. I. Mackie and J.O. Dawson. 2002. Could microbial symbionts of arthropod guts contribute significantly to nitrogen fixation in terrestrial ecosystems? Journal of Insect Physiology 48: 751–763

Neilson, R., D. Hamilton, J. Wishart, C. A. Marriott, B. Boag, L. L. Handley, C. M. Scrimgeour, J. W. McNicol and D. Robinson. 1998. Stable isotope natural abundances of soil, plants and soil invertebrates in an upland pasture. Soil Biology and Biochemistry 30:1773-1782.

Oelbermann, K. and S. Scheu. 2002. Stable istope enrichment (δ15N and δ13C) in a generalist predator (Pardosa lugubris, Araneae: Lycosidae): effects of prey quality. Oecologia 130:337-244.

O’Reilly, C. M., R. E. Hecky, A. S. Cohen, and P. D. Plisnier. 2002. Interpreting stable isotopes in food webs: Recognizing the role of time averaging at different trophic levels. Limnology and Oceanography 47(1):306-309.

Ostrom, P. H., M. Colunga-Garcia, and S. H. Gage. 1997. Establishing pathways of energy flow for insect predators using stable isotope ratios: Field and laboratory evidence. Oecologia 109:108-113.

Paine, R. T. 1988. Food webs: road maps of interactions or grist for theoretical development? Ecology 69:1648-1654.

Park, O. and D. Reichle. 1963. Observations on the ecology and behavior of the cave cricket, Hadenoecus subterraneus (Scudded[sic]). Bulletin of the Ecological Society of America 44(1):126-127.

Peck, S. B. 1976. The effect of cave entrances on the distribution of cave inhabiting terrestrial arthropods. International Journal of Speleology 8(4):309-321.

Persson, L. 1999. Trophic cascades: abiding heterogeneity and the trophic level concept at the end of the road. Oikos 85:385-397.

Peterson, B. J. and B. Fry. 1987. Stable isotopes in ecosystem studies. Annual Review of Ecology and Systematics 18:293-320.

Phillips, D. L. and J. W. Gregg. 2003. Source partitioning using stable isotopes: Coping with too many sources. Oecologia (preprint from Springer-Verlag website: http://link.springer-ny.com/link/service/journals/00442/contents/03/01218/).

Phillips, D. L. and P. L. Koch. 2002. Incorporating concentration dependence in stable isotope mixing models. Oecologia 130:114-125.

65

Pilsbry, H.A. 1940. Land Mollusca of North America (North of Mexico). Acad. Nat. Sci. Monogr. No.3. Vol.1. Part 2.

Pohlman, J. W., T. M. Iliffe, and L. A. Cifuentes. 1997. A stable isotope study of organic cycling and the ecology of an anchialine cave ecosystem. Marine Ecology Progress Series 155:17-27.

Pohlman, J.W., L. Cifuentes and T.M. Iliffe 2000. Food web dynamics and biochemistry of anchialine caves: a stable isotope approach. Pages 345-357 in: Ecosystems of the World. 30. Subterranean Ecosystems, H. Wilkens, D.C. Culver, & W.F. Humphreys (eds.), Elsevier Science, Amsterdam.

Ponsard, S. and R. Arditi. 2000. What can stable isotopes (δ15N and δ13C) tell about the food web of soil macro-invertebrates? Ecology 81(3)852-864.

Poulson, T. L., K. Lavoi, and K. L. Helf. 1995. Long-term effects of weather on the cricket Hadenoecus subterraneus (Orthoptera, Rhaphidophoridae) guano community in Mammoth Cave National Park. American Midland Naturalist 134:226-236.

Poulson, T. L. and W. B. White. 1969. The cave environment: Science 165:971-981.

Quinn, M. R., X. Feng, C. L. Folt, and C. P. Chamberlain. 2003. Analyzing trophic transfer of metals in stream food webs using nitrogen isotopes. The Science of the Total Environment 317:73-89.

Reddell, J. R. 2001. Cave invertebrate research on Fort Hood, Bell and Coryell counties, Texas. Report to Texas Nature Conservancy, Fort Hood Field Office. 278 pp.

Reddell, J. R and J. C. Cokendolpher (eds). 2001a. Studies on the Cave and Endogean Fauna of North America III. Texas Memorial Museum, Speleological Monographs, 5. viii + 192 pp.

Reddell, J. R. and C. C. Cokendolpher. 2001b. Ants (Hymenoptera: Formicidae) from the caves of Belize, Mexico, and California and Texas (U.S.A.). Texas Memorial Museum, Speleological Monographs 5:129-154.

Reichle, D. E., J. D. Palmer and O. Park. 1965. Persistent rhythmic locomotor activity in the cave cricket, Hadenoecus subterraneus, and its ecological significance. American Midland Naturalist 74:57-66.

Reynolds, D. R. and J. R. Riley. 2002. Remote-sensing, telemetric and computer- based technologies for investigating insect movement: a survey of existing and potential techniques. Computers and Electronics in Agriculture 35:271-307.

66 Richards, A. M. 1965. The effect of weather on Rhaphidophoridae (Orthoptera) in New Zealand and Australia. Annales de Spéléologie 20(3):391-400.

Rico-Gray, V. and L. da S. L. Sternberg. 1991. Carbon isotopic evidence for seasonal change in feeding habits of Camponotus planatus Roger (Formicidae) in Yucatan, Mexico. Biotropica 23(1):93-95.

Riecken, U. and U. Raths. 1996. Use of radio telemetry for studying dispersal and habitat use in Carabus coriaceus L. Ann. Zool. Fenn. 33:109-116.

Robbins, C. T., G. V. Hilderbrand, and S. D. Farley. 2002. Incorporating concentration dependence in stable isotope mixing models: A response to Phillips and Koch (2002). Oecologia 133:10-13.

Rounick, J. S. and M. J. Winterbourn. 1986. Stable carbon isotopes and carbon flow in ecosystems. Bioscience 36(3):171-177.

Rundel, P. D., J. R. Ehleringer, and K. A. Nagy (eds.). 1988. Stable isotopes in ecological research. Ecological Studies (vol. 68). Springer-Verlag, New York.

Scheu, S. and M. Falca. 2000. The soil food web of two beeech forests (Fagus sylvatica) of contrasting humus type: stable isotope analysis of a macro- and mesofauana-dominated community. Oecologia 123:285-296.

Scheu, S. 2002. The soil food web: structure and perspectives. European Journal of Soil Biology 38:11-20.

Schmidt, O. 1999. Intrapopulation variation in carbon and nitrogen stable isotope ratios in the earthworm Aporrectaodea longa. Ecological Research 14:317-328.

Scrimgeour, C. M., S. C. Gordon, L. L. Handley and J. A. T. Woodford. 1995. Trophic levels and anomalous δ15N of insects on raspberry (Rubus idaeus L.). Isotopes and Environmental Health Studies 31:107-115.

Smith, B. and S. Epstein. 1971. Two categories of 13C/12C ratios for higher plants. Plant Physiology 47:380-384.

Smith, B. N. and Brown, W. B. 1973. The Kranz syndrome in the Gramineae as indicated by carbon isotopic ratios. American Journal of Botany 60:505-513.

Speiser, B. 2001. Food and feeding behavior. Pages 259-288 in: Barker, G. M. (ed.). The Biology of Terrestrial Molluscs. CAB International.

Spencer, S. R., G. N. Cameron and R. K. Swihart. 1990. Operationally defining home range: temporal dependence exhibited by hispid cotton rats. Ecology (Washington D C) 71:1817-1822.

67

Studier, E. H., K. H. Lavoie, W. D. Wares, II, and J. A. M. Linn. 1987. Bioenergetics of the camel cricket Ceuthophilus stygius. Comparative Biochemistry and Physiology 86A(2):289-293.

Taylor, S. J. 2003. America, North: Biospeleology. Pages 45-49 in Gunn, John (ed.). Encyclopedia of Cave and Karst Science. Fitzroy Dearborn, New York. xviii + 902 pp.

Taylor, S. J., J. K. Krejca, J. E. Smith, V. R. Block, and F. Hutto. 2003. Investigation of the potential for Red Imported Fire Ant (Solenopsis invicta) impacts on rare karst invertebrates at Fort Hood, Texas: a field study. Illinos Natural History Survey, Center for Biodiversity Technical Report 2003(28):1-153.

Taylor, S. J., J. K. Krejca, and M. L. Denight. ____. Foraging range and habitat use of Ceuthophilus secretus (Orthoptera: Rhaphidophoridae), a keystone trogloxene in central Texas cave communities. American Midland Naturalist, Submitted August 2004.

Troughton, J. H. and K. A.Card. 1975. Temperature effects on the carbond-isotope ratio of C3, C4 and crassulacean-acid-metabolism (CAM) plants. Planta (Berlin) 123:185- 190.

U.S. Fish and Wildlife Service (USFWS), Department of the Interior. 1984. Recovery Plan for the Noonday Snail. U.S. Fish and Wildlife Service. Atlanta, GA. 30 pp.

U.S. Fish and Wildlife Service (USFWS), Department of the Interior. 1988. Endangered and Threatened Wildlife and Plants; Final rule to determine five Texas cave invertebrates to be endangered species. Federal Register 53(180):36029-36033.

U.S. Fish and Wildlife Service (USFWS), Department of the Interior. 1993. Endangered and Threatened Wildlife and Plants; Coffin Cave Mold Beetle (Batrisodes texanus) and the Bone Cave Harvestman (Texella reyesi) determined to be endangered. Federal Register 56(158):43818-43820.

U.S. Fish and Wildlife Service (USFWS), Department of the Interior. 2000. Endangered and Threatened Wildlife and Plants; Final rule to list nine Bexar County, Texas invertebrate species as endangered. Federal Register 65(248):81419-81433.

U.S. Fish and Wildlife Service (USFWS), Department of the Interior. 2003. Endangered and Threatened Wildlife and Plants; Designation of critical habitat for seven Bexar County, TX, invertebrate species. (Final Rule). Federal Register 68(67):17156-17231.

Vandel, A. 1965. Biospeleology-The biology of cavernicolous animals. London, England, Pergamon Press. 524 p.

68 Wada, E. H. Mizutani, and M. Minagawa. 1991. The use of stable istopes for food web analysis. Crit. Rev. Food Sci. Nutr. 30:361-371.

Webb S. C., R. E. M. Hedges and S. J. Simpson. 1998. Diet quality influences the δ13C and δ15N of locusts and their biochemical components. Journal of Experimental Biology 201:2903–2911.

69 Appendix 1. Images of 25 selected plant taxa which were used in isotopic analysis. Photos by Steve Taylor.

i.

ii.

A. Prickly-pear, Opuntia phaecantha Engelm. var. major Engelm.: i. with fruit, 22 November 2003, Streak Cave; ii. with bloom, 23 May 2004, Streak Cave.

70

B. Sweet mountain grape, Vitus cf. monticola Buckley, 23 May 2004, Streak Cave.

71

C. Texas ash, Fraxinus texensis (A. Gray) Sarg., 23 May 2004, Streak Cave.

72

D. Golden tickseed or plains coreopsis, Coreopsis tinctoria Nutt., 23 May 2004, Streak Cave.

73

E. Mescal bean, Sophora secundiflora (Ortega) Lag. ex DC., 23 May 2004, Streak Cave.

74

F. Violet lespedeza, Lespedeza cf. violacea (L.) Pers., 23 May 2004, Streak Cave.

75 i.

ii.

G. Plateau oak, Quercus fusiformis Small: i. 23 May 2004, Streak Cave; ii. 21 November 2003, Mixmaster Cave.

76

H. Fineleaf fournerved daisy, Tetraneuris linearifolia (Hook.) Greene, 23 May 2004, Streak Cave.

77

I. Starved rosettegrass, Panicum cf. depauperatum Muhl., 23 May 2004, Streak Cave.

78

J. Texas bluebonnet, Lupinus texensis Hook., 23 May 2004, Streak Cave.

79

K. White shin oak, Quercus sinuata Walt. var. breviloba (Torr.) C.H. Müll., 23 May 2004, Streak Cave.

80

L. False Indigo, Amorpha fruticosa L., 23 May 2004, Big Red Cave.

81

M. Bouteloua curtipendula (Michx.) Torr., 23 May 2004, Big Red Cave.

82

N. Common greenbriar, Smilax rotundifolia L., 23 May 2004, Big Red Cave.

83

O. Skunk bush, Rhus trilobata Nutt., 23 May 2004, Big Red Cave.

84

P. Hairy tridens, Erioneuron pilosum (Buckley) Nash, 23 May 2004, Big Red Cave.

85

Q. Texas wintergrass, Nassella leucotricha (Trin. & Rupr.) Pohl, 23 May 2004, Big Red Cave.

86

i.

ii.

R. Texas persimmon, Diospyros texana Scheele: i. 23 May 2004, Big Red Cave; ii. 21 November 2003, Mixmaster Cave.

87

S. Velvet buttonflower, Desmanthus velutinus Scheele, 23 May 2004, Big Red Cave.

88

T. Firewheel, Gaillardia pulchella Foug., 23 May 2004, Big Red Cave.

89

U. Berlandier's sundrops, Calylophus berlandieri Spach ssp. pinifolous (A. Gray) Towner, 23 May 2004, Big Red Cave.

90

V. Netleaf hackberry, Celtis laevigata Willd. var. reticulata (Torr.) L. Benson, 23 May 2004, Big Red Cave.

91

W. Yellow stonecrop, Sedum nuttallianum Raf., 23 May 2004, Big Red Cave.

92

X. Ashe juniper, Juniperus ashei J. Buchholz, 21 November 2003, Mixmaster Cave.

93

Y. King Ranch bluestem, Bothriochloa ischaemum (L.) Keng var. songarica (Rupr. ex Fisch. & C.A. Mey.) Celarier & Harlan, 21 November 2003, Mixmaster Cave.

94

Z. Common broomweed, Gutierrezia dracunculoides (DC.) S.F. Blake, 21 November 2003, Mixmaster Cave.

95 Appendix 2. Locations, movement path, and home range maps for radiolocated Ceuthophilus secretus. Background is general vegetation types, derived from aerial photography, colors are: gray – bare ground, light green – grassland, dark green – forested/shrub (modified from Taylor et al. submitted).

Cricket 1

96 Cricket 2

97 Cricket 3

98 Cricket 4

99 Cricket 5

100 Cricket 6

101 Cricket 7

102 Cricket 8

103 Cricket 9

104 Cricket 10

105 Cricket 11

106 Cricket 12

107 Cricket 13

108 Cricket 14

109 Cricket 15

110 Cricket 17

111 Cricket 18

112 Cricket 19

113 Cricket 20

114 Cricket 21

115 Cricket 22

116 Cricket 23

117 Cricket 24

118 Cricket 25

119 Cricket 26

120 Cricket 27

121 Cricket 28

122 Cricket 29

123 Cricket 30

124 Cricket 31

125 Cricket 32

126 Cricket 33

127 Cricket 34

128