Herpetologíca, 40(1), 1984, 82-88 © 1984 by The Herpetologists' League, Inc.

ESTIMATING PREY SIZE AND NUMBER IN CRAYFISH-EATING , REGINA

J. S. GODLEY, R. W. McDiARMID, AND N. N. ROJAS

ABSTRACT: Snakes of the genus Regina feed almost exclusively on crayfish. The paired, sym- metrical gastroliths of crayfish are not digested and are detectable from x-rays of the . Gastrolith length is directly proportional to carapace length and can be obtained from x-rays. Carapace length can be converted to kcal of ingested energy. Using these relationships and repeated captures of radio-telemetered Regina, estimates of food consumption and energy intake by free- living snakes are feasible. New information on prey selectivity, feeding behavior, and predator- prey size relations in Regina grahami and R. septemvittata are presented and compared with similar data for other snakes. Key words: Reptilia; Serpentes; ; Regina; Food habits; Predator-prey size relation- ships; Crayfish

THE foraging ecology of snakes is less ing of specific events in the crayfish molt often studied than many aspects of snake cycle and the relationship of these events biology. Some are rare and others to prey selection in R. grahami and R. are secretive; consequently, obtaining ad- septemvittata. Briefly, individuals of var- equate samples is a common problem ious species of crayfish molt one to 12 times (Turner, 1977). In addition, most snakes per year depending upon sex, age and en- eat relatively large prey at infrequent in- vironment. Usually less than 15% of a tervals (e.g.. Beavers, 1976; Fitch, 1960, population is molting at any one time (Ca- 1965; Godley, 1980; Greene, 1983a,fe; pelli and Magnuson, 1975; Drach, 1939; Schoener, 1977; Seib, 1981), and their Prins, 1968; Stein, 1977; Stevenson, 1975). digestive tracts seldom contain food. Be- Prior to molt, calcium is extracted from cause of the prey's advanced state of the exoskeleton and stored in paired, sym- digestion in the hindgut (Brown, 1958; metrical gastroliths located in the cardiac Godley, 1980; Henderson, 1970; Skoczy- stomach (McWhinnie, 1962; Richards, las, 1970), positive identification of prey 1951; Stein and Murphy, 1976; Travis, species and important measures of prey 1960). The gastroliths are resorbed shortly size often are obtainable only while the after molt and contribute to the formation prey is in the snake's stomach. of a new exoskeleton. Gastroliths reappear Here we describe an accurate, non-in- upon initiation of the next molt cycle. vasive technique for estimating the num- Gastrolith growth and résorption are dis- ber and sizes of prey in the entire diges- tinct events and are tightly synchronized tive tracts of crayfish-eating snakes, genus with other identifiable stages in the molt Regina. The technique is designed specif- cycle (McWhinnie, 1962; Stevenson, 1975). ically for dietary studies of R. grahami As we shall show, these gastroliths provide and R. septemvittata, two species that feed a convenient, quantitative "marker" for almost exclusively on soft, freshly molted estimating rates of feeding and energy in- crayfish (Branson and Baker, 1974; Burg- take in R. grahami and R. septemvittata, hardt, 1968; Hall, 1969; Kofron, 1978; and perhaps in other species that feed ex- Mushinsky and Hebrard, 1977; Strecker, tensively on molting crayfish (Neill, 1951; 1926; Wood, 1949). We also provide new Penn, 1950). information on prey selectivity, feeding behavior, and predator-prey size relations MATERIALS AND METHODS in these species and summarize similar We estimated the number, size and data for other species of snakes. feeding sequence of crayfish ingested by The technique requires an understand- 36 R. grahami (USNM 12864-65, 12891, 82 March 1984] HERPETOLOGICA 83

12940-70, 13038-39) from New Orleans, were examined after dissection measure- Orleans Parish, Louisiana, and two ments were recorded. Because methods (197638) from Badger Lake, Monona varied between the snake species, results County, Iowa, by exposing identifiable are presented separately. prey with a single ventro-medial incision through the gut wall. The orientation RESULTS (head or abdomen first) and the ingestion position (dorsal, ventral or lateral) of the Regina grahami crayfish relative to the snake's skull were Only the remains of crayfish, repre- recorded prior to removal of prey from senting at least three species, Orconectes the snake's stomach. After removal, the palmeri croelanus (Creaser), Procamba- crayfish's carapace length (CL = maxi- rus clarkii (Girard) and P. vioscai (Penn), mum dorsal midline distance from tip of were recovered from the stomachs of 38 rostrum to posterior edge of carapace) and R. grahami. Of these prey, measurements gastrolith length (GL = maximum longi- of both GL and CL were obtainable from tudinal length of gastroliths) were mea- 17 crayfish (all were in molt stages Ai or sured with vernier calipers to the nearest early Aj), CL only from two crayfish (one 0.1 mm. Only GL was obtained from in- in stage B, the other in stage Aj or As but testinal contents; paired, symmetrical gas- gastroliths broken during processing), GL troliths were assumed to represent the di- only from six crayfish (all in stages Ai or gested remains of a single crayfish. The As but carapace partially digested), and molt stage of the crayfish (Drach, 1939; two crayfish were represented only by sin- McWhinnie, 1962; Stevenson, 1975) also gle, very large (57 and 63 mm total length) was recorded. Molt stages pertinent to this chaela, each in a different snake. The study are as follows. A;•Exoskeleton is largest chaela from an intact crayfish soft, gelatinous and easily torn; gastrolith (CL = 31.5 mm) in the sample was 36.0 is fully formed; duration is of a few hours. mm, suggesting that the latter two snakes Aj•Exoskeleton is leathery; gastrolith is encountered crayfish too large to consume reduced to about 50% of former size by whole and instead removed the only large end of stage; duration is ca. 24 h. B• portion (chaela) that could be disarticu- Rostrum and cephalic carapace are firmer lated. In addition, GL measurements from but flexible; gastrolith is very small to ab- 20 crayfish were obtained from intestinal sent at end of stage; duration is of one to contents. All of these crayfish appeared to three days. have been in molt stage Ai or early A2. Similar data were obtained from a se- Although most of the exoskeletons of ries of 36 R. septemvittata collected in crayfish were partially digested, the gas- Montgomery Co., Ohio (USNM 128974- troliths seemed normal and unaffected by 98, 129016-27). However, before dissec- passage through the snake's digestive tract. tion, each specimen was radiographed in In fact, captives of all species of Regina dorso-ventral view. The radiographs were defecate gastroliths which macroscopical- exposed for 30 s at instrument readings of ly are indistinguishable from those dis- 5 mamp and between 20 and 23 kV de- sected from live crayfish of a similar molt pending on snake size. Gastrolith mea- stage (J. S. Godley, personal observation). surements were taken from the radio- As is typical of freshly molted crayfish graphs with a dial caliper. Because (McWhinnie, 1962 for Orconectes virilis), gastroliths are slightly oblong in shape and GL is a linear function of CL (Fig. 1) and the x-rays provided only a single plane of provides an excellent estimate of crayfish view, the longest dimension of either gas- size. Most of the variance about this line trolith of the pair was chosen to represent probably is caused by three sources of ex- the GL. To avoid possible bias, GL mea- perimental error: (1) differences in GL: surements taken from the radiographs CL ratios among the three species of cray- 84 HERPETOLOGICA [Vol. 40, No. 1

food, 14 contained one crayfish, six had two crayfish each, two contained three, Y.0.157X-0.5S2 • 5,0 and three had eaten four crayfish. SVL Ê r = 0.8 8 0 was positively correlated with the total E number of prey whether individuals with empty digestive tracts were included (Spearman rank correlation, r^ = 0.47; P = 0.0076, n = 38) or excluded (r, = 0.58, P = 0.0024, n = 25) from the analysis. Prey orientation could be determined for 26 of 27 crayfish recovered from the R. grahami stomachs; 12 were ingested head first and 14 abdomen first, suggest- 20 30 Carapace Length (mm) ing no orientation preference (x^ = 0.08, F > 0.75). Vertical position of prey could FIG. 1.•Relationship of gastrolith length and car- be determined for 22 of the 27 prey and apace length from 17 crayfish removed from the was not significantly different from ran- stomachs of Regina grahami. dom (x' = 4.0, 2 df, F > 0.10) with 12 crayfish ingested on their side with re- spect to the snake's skull, eight with ven- fish represented in the sample, (2) differ- ters up, and two with venters down. Anal- ences in crayfish molt stage, and (3) the ysis of covariance with GL as the difficulties of accurately measuring the CL dependent variable, SVL as the covariate, of soft-bodied crayfish taken from a and prey orientation and position as in- snake's stomach. In a sample of 18 Pro- dependent variables showed no significant cambarus fallax from Hillsborough Co., effects for the latter two variables. Florida, which were sacrificed within six h of molting, the correlation coefficient Regina septemvittata between GL and CL was 0.98 (J. S. God- In this species, specimens were x-rayed ley, unpublished data). The slope of this prior to dissection to determine if radio- line was not significantly different from graphs could provide reliable estimates of that shown in Fig. 1 (F^ai = 2.48, F = feeding activity. The two criteria we 0.1253). judged necessary for establishing reliabil- Predator-prey size relationships in R. ity were: (1) radiographs must detect all grahami can be examined by using GL as crayfish in the digestive tract, and (2) an estimate of crayfish size (GL estimated measurements of GL taken from x-rays for two crayfish from CL) and snake and those obtained from dissection must snout-vent length (SVL) as an estimate of show high correspondence. predator size (Fig. 2A). In our data set, Gastroliths representing 26 crayfish the regression of GL on CL (Fig. 1) al- were identified from the radiographs of lowed estimates of size for 27 digested the 36 R. septemvittata. The same num- crayfish and increased the usable crayfish ber of prey was found when the snakes sample size 158.8% (from 17-44). Snake were dissected. A representative radio- SVL explained a significant (F = 0.0029, graph is shown in Fig. 3. A strong corre- Y = 1.436 + 0.0046X) but small percent- lation exists between GL based on radio- age (r^ = 0.193) of the variation in the size graphs and GL based on dissections (Fig. of prey taken by R. grahami (Fig. 2A). 4). Although adult R. grahami consumed Only four of the 26 crayfish in the R. larger crayfish than juveniles, the mini- septemvittata were in the stomachs. These mum size of ingested prey did not change four prey (Cambarus diogenes Girard) during ontogeny. Of the 25 snakes with were partially digested and no CL mea- March 1984] HERPETOLOGICA 85

9 • . A B B •

7 U

"i , 1 6 • •

- c S " = • • • tJ Í t / Ô :: 4 • • tD • . • 3 • • • 2 • • • • •

r 400 500 600 300 Snout-Vent Lengtti (mm) Snout-Vent Lengtti (mm) Regina septemvrttota Regino gratiami

FIG. 2.•Size relationship between snakes and their respective crayfish prey. (A) Regina grahami (n •• 44). (B) Regina septemvittata (n = 26). surements were obtained; all were in molt clearance times and rates of digestion in stage Al and were ingested abdomen first. snakes of different size, sex and reproduc- Judging from the condition of the gas- tive condition (see Dandrifosse, 1974; troliths (representing 22 crayfish) found in Godley, 1980; Skoczylas, 1978 for re- the intestines, these crayfish also were in views). Here a major difficulty has been molt stage A; or A2 when eaten. The scat- the lack of a convenient, natural marker ter plot of crayfish GL on R. septemvit- to follow the course of prey digestion. tata SVL (Fig. 2B) was similar to that ob- X-rays made at appropriate intervals of served in jR. grahami (Fig. 2A), but the crayfish-eating snakes would provide such regression was not significant (r^ = 0.09, a system under controlled laboratory con- P = 0.139). ditions. Perhaps the greatest use of this technique lies in its potential field appli- DISCUSSION cation. By using the relationships devel- Our work has revealed an accurate, oped above, a portable x-ray unit, and re- nondestructive technique for estimating peated captures of Regina equipped with the number and size of prey in two species transmitters that measure heart rate or of Regina based on the following obser- body temperature, accurate estimates of vations. (1) Regina feed almost exclusive- the energy budget of free-living snakes are ly on crayfish. (2) The paired, symmetri- possible (the energy content of crayfish cal gastroliths of crayfish are not digested. is known, see Godley, 1980; Stein and (3) GL is directly proportional to CL. (4) Murphy, 1976). Concurrent x-rays of non- GL is obtainable from x-rays of Regina telemetered from the same pop- that have fed on crayfish. ulation could serve as a control for possi- Insight into several poorly known as- ble effects of experimental procedures. pects of snake foraging ecology and diges- Our results show that crayfish are sub- tive physiology could be gained using this ject to prédation by Regina grahami and technique. For example, little is known R. septemvittata only while they are in about the effects of different feeding, molt stages Ai or A2. Further, the low vari- temperature and activity regimes on ance in GL for crayfish of any size (Fig. 86 HERPETOLOGICA IVol. 40, No. 1

FIG. 3.•Radiograph of Regina septemvittata (USNM 128998) from Ohio. This snake contains three 3 4 5 pairs of gastroHths and a pebble (arrow). Gastrolith Length (mm)-X-ray

FIG. 4.•Relationship between gastrolith length as measured from gastroliths removed from gut and 1) and the rapid temporal decline in GL those measured from x-rays of Regina septemvittata (n = 25); one gastrolith was damaged during pro- following molt (McWhinnie, 1962) sug- cessing and excluded from this comparison. gest that most crayfish are eaten by these snakes within 6 h of molt (stage A,). This extreme molt-stage selectivity may have important ecological consequences. In is independent of crayfish size (see also three species of crayfish found within the Hall, 1969). However, R. septemvittata, range of R. grahami or ñ. septemvittata, which feeds on crayfish of similar molt freshly molted individuals comprise less stages, apparently ingests crayfish only than 15% of the population averaged over abdomen first (see also Branson and Ba- the year (Prins, 1968; Stein, 1977). In ad- ker, 1974; Wood, 1949), suggesting that dition, all of these species seem to have a these two closely related species (Ross- "mass, synchronized molt" (Prins, 1968: man, 1963) have diverged behaviorally 678) which would further limit the bio- while maintaining similar molt-stage se- mass of snakes that could be supported on lectivity. R. alleni consumes crayfish ab- a given population of crayfish. In contrast, domen first and lateral with respect to the morphological and behavioral adaptations snake's skull regardless of crayfish size, of Regina alleni (Franz, 1977; Godley, molt-stage or snake feeding experience 1980) and R. rigida (Kofron, 1978) enable (Franz, 1977; Godley, 1980 and unpub- these species to exploit crayfish in all stages lished). The feeding behavior of R. rigida of their molt cycle. Available density es- is unknown but probably is similar to that timates for R. septemvittata (Branson and of R. alleni. Baker, 1974) and R. alleni (Godley, 1980) Snake SVL proved to be a relatively support this contention. poor predictor of the size of crayfish eaten Variation in molt-stage selectivity by Regina grahami (r^ = 0.19, P = 0.0029) among species of Regina also is reflected or R. septemvittata (r^ = 0.09, P = 0.139) in their feeding behavior. Covariance (Fig. 2). In reviewing similar data for 17 analysis suggests that in R. grahami, ori- other species of snakes distributed among entation and positioning of crayfish is ran- five families (Beavers, 1976; Godley, 1980; dom with respect to the snake's skull and Greene, 1983a,¿; Mushinsky et al, 1982; March 1984] HERPETOLOGICA 87

Reynolds and Scott, 1982; Seib, 1981; LITERATURE CITED Shine, 1977; Voris and Moffett, 1981), we BEAVERS, R. A. 1976. Food habits of the western found four common, intraspecific trends. diamondback rattlesnake, Crotalus atrox, in Texas (1) Larger snakes can consume absolutely (Viperidae). Southwest. Nat. 10:503-515. larger prey than smaller individuals. (2) BRANSON, B. A., AND E. C. BAKER. 1974. An eco- logical study of the queen snake, Regina septem- The variance in prey size tends to increase vtttata (Say) in Kentucky. Tulane Stud. Zool. Bot. with increasing snake size. (3) A least 18:153-171. squares regression of prey size on snake BROWN, E. E. 1958. Feeding habits of the northern size usually yields a positive, significant watersnake, Natrix sipedon sipedon Linnaeus. slope but rarely accounts for more than Zoológica 43:55-71. BURGHARDT, G. M. 1968, Chemical preference 50% of the variation. (4) The slope of the studies on newborn snakes of three sympatric boundary line for minimum prey size is species of Natrix. Copeia 1968:732-737. much lower than that for maximum prey CAPELLI, G. M., AND J. J. MAGNUSON. 1975. Re- size such that the size of the smallest prey production, molting, and distribution of Orco- nectes propinquus (Girard) in relation to temper- is similar for adults and juveniles of the ature in a northern mesotrophic lake. Pp. 415-428, same species. As did Voris and Moffett In J, W, Avault, Jr, (Ed.), Second International (1981), we define prey boundary lines as Symposium on Freshwater Crayfish. Louisiana State the approximate lower and upper size University, Div, Continuing Education, Baton limits of prey eaten by a snake during its Rouge, CZAPLICKI, J, A., AND R, H. PORTER. 1974. Visual ontogeny. cues mediating the selection of goldfish (Carassius We are concerned with the increasing auratus) by two species of Natrix. J, Herpetol, 8: tendency among some workers to inter- 129-134, pret the shape of predator-prey size tra- DANDRIFOSSE, G, 1974. Digestion in , Pp, 249-275. In M. Florkin and B. T. Scheer (Eds.), jectories in snakes as a reflection of prey Chemical Zoology, Vol. 9. Academic Press, New size selection seemingly without con- York, sideration of alternative hypotheses. Con- DRACH, P, 1939, Mue et cycle d'intermue chez les straints imposed by head morphology crustacés décapodes, Ann, Inst, Oceanogr, Monaco alone would restrict the range of prey sizes 19:103-391, FITCH, H. S, 1960, Autecology of the copperhead, ingested by gape-limited snakes (Gans, Univ, Kansas Publ, Mus, Natur, Hist. 13:85-288, 1961; Greene, 1983a) and could produce , 1965. An ecological study of the garter each of the four intraspecific trends noted snake, Thamnophis strtalis. Univ. Kansas Publ. above. A morphological-constraint hy- Mus. Nat, Hist, 15:493-564, FRANZ, R, 1977, Observations on the food, feeding pothesis also could explain why, in our behavior and parasites of the striped swamp snake, opinion, experimental attempts to dem- Regina alleni. Herpetologica 33:91-94, onstrate prey selection based on size in GANS, C, 1961. The feeding mechanism of snakes snakes have failed (Czaplicki and Porter, and its possible evolution. Am. Zool. 1:217-227. 1974; Godley, 1980; Reynolds and Scott, GODLEY, J. S. 1980. Foraging ecology of the striped swamp snake, Regina alleni, in southern Florida, 1982; Smith and Watson, 1972). Although Ecol, Monogr. 50(4):411-436, we do not doubt that ecological factors GREENE, H. W, 1983íI. Dietary correlates of the influence prey choice in snakes, it remains origin and radiation of snakes. Am, Zool, 23:431- to be shown that any species of snake se- 441, lects a smaller range of prey sizes than the . 1983i>. Feeding behavior and diet of the eastern coral snake, Micrurus fulvius. Spec, Publ, limits imposed by its morphology. Mus. Nat, Hist. Univ. Kansas: In press. HALL, R. J. 1969. Ecological observations on Gra- Acknowledgments.•Margaret Daniel and Horton ham's water snake, Regina grahami (Baird and H. Hobbs, Jr. assisted with the crayfish identifica- Girard). Am. Midi. Nat, 18:156-163. tions; Claudia Angle prepared some of the illustra- HENDERSON, R. W. 1970. Feeding behavior, diges- tions; Susan Jewett assisted with the x-rays; and Ma- tion, and water requirements of Diadophis punc- rianne Scott typed the manuscript. Harry Greene tatus arnyi Kennicott, Herpetologica 16:520-526, and Henry Mushinsky reviewed an early draft of the KoFRON, C. P, 1978. Food and habitats of aquatic manuscript. To all of these people, we extend our snakes (Reptilia, Serpentes) in a Louisiana swamp. thanks. J. Herpetol. 12:543-554. 88 HERPETOLOGICA [Vol. 40, No. 1

MCWHINNIE, M. A. 1962. Gastrolith growth and SMITH, G. C, AND D. WATSON. 1972. Selection calcium shifts in the freshwater crayfish, Orco- patterns of corn snakes, Elaphe guttata, of differ- nectes virilis. Comp. Biochem. Physiol. 7; 1-14. ent phenotypes of the house mouse. Mus muscu- MusHiNSKY, H. R., AND J. J. HEBRARD. 1977. Food lus. Copeia 1972:529-532. partitioning by five species of water snakes in Lou- STEIN, R. A. 1977. Selective prédation, optimal for- isiana. Herpetologica 33:162-166. aging, and the predator-prey interaction between MusHiNSKY, H. R., J. J. HEBRARD, AND D. S. fish and crayfish. Ecology 58:1237-1253. VODOPICH. 1982. Ontogeny of water snake for- STEIN, R. A., AND M. L. MURPHY. 1976. Changes aging ecology. Ecology 63:1624-1629. in proximate composition of the crayfish Orco- NEILL, W. T. 1951. Notes on the role of crawfishes nectes propinquus with size, sex, and life stage. J. in the ecology of reptiles, amphibians, and fishes. Fish. Res. Board Can. 33:2450-2458. Ecology 32:764-766. STEVENSON, J. R. 1975. The molting cycle in the PENN, G. H. 1950. Utilization of crawfishes by cold- crayfish: recognizing the molting stages, effects of blooded vertebrates in the eastern United States. ecdysone, and changes during the cycle. Pp. 255- Am. Midi. Nat. 44:643-658. 267. In J. W. Avault, Jr. (Ed.), Second Internation- PRINS, R. 1968. Comparative ecology of the cray- al Symposium on Freshwater Crayfish. Louisiana fishes Orconectes rusticus rusticus and Cambarus State Univ., Div. Continuing Education, Baton tenebrosus in Doe Run, Meade County, Kentucky. Rouge. Int. Rev. Ges. Hydrobiol. 53:667-714. STRECKER, J. K. 1926. On the habits of some south- REYNOLDS, R. P., AND N. J. SCOTT, JR. 1982. Use ern snakes. Baylor Univ. Mus. Contri. 4:3-11. of a mammalian resource by a Chihuahuan snake TRAVIS, D. F. 1960. Matrix and mineral deposition community. Pp. 99-118. In N. J. Scott, Jr. (Ed.), in skeletal structures of the decapod Crustacea. Pp. Herpetological Communities. U.S. Dept. Int. Fish 57-116. In R. F. Sognnaes (Ed.), Calcification in Wildlife Ser., Wildlife Research Report 13. Biological Systems. Publ. 64, Am. Assoc. Adv. Sei., RICHARDS, A. G. 1951. The integument of arthro- Washington, D.C. pods. University of Minnesota Press, Minneapolis. TURNER, F. B. 1977. The dynamics of populations ROSSMAN, D. A. 1963. Relationships and taxonomic of squamates, crocodilians and rhynchocephalians. status of North American natricine snake genera Pp. 157-264. In C. Gans and D. W. Tinkle (Eds), Liodytes, Regina, and Clonophis. Occ. Pap. Mus. Biology of the Reptilia. Vol. 7. Academic Press, Zool., Louisiana State Univ. 29:1-29. New York. SCHOENER, T. W. 1977. Competition and the niche. VORIS, H. K., AND M. W. MOFFETT. 1981. Size and Pp. 35-136. In C. Gans and D. W. Tinkle (Eds.), proportion relationship between the beaked sea Biology of the Reptilia. Vol. 7. Academic Press, snake and its prey. Biotropica 13:15-19. New York. WOOD, J. T. 1949. Observations on Natrix septem- SEIB, R. L. 1981. Size and shape in a neotropical vittata (Say) in southwestern Ohio. Am. Midi. Nat. burrowing colubrid snake, Geophis nasalis, and its 42:744-750. prey. Am. Zool. 21:933. SHINE, R. 1977. Habitats, diets, and sympatry in Accepted: 25 August 1983 snakes: a study from Australia. Can. J. Zool. 55: Associate Editor: Kentwood Wells 1118-1128. JSG and NNR: Department of Biology, SKOCZYLAS, R. 1970. Influence of temperature on gastric digestion in the grass snake, Natrix natrix University of South Florida, Tampa, FL L. Comp. Biochem. Physiol. 33:793-804. 33620, USA; JSG and RWM: U.S. Fish . 1978. Physiology of the digestive tract. Pp. and Wildlife Service, National Museum 589-717. In C. Gans and K. A. Gans (Eds.), Biol- of Natural History, Washington, DC ogy of the Reptilia. Vol. 8. Academic Press, New York. 20560, USA