Water Relations of Chelonian Eggs and Embryos: Is Wetter Better? 1

Transcription

1 AMER. ZOOL., 39: (1999) Water Relatins f Chelnian Eggs and Embrys: Is Wetter Better? 1 GARY C. PACKARD 2 Department f Bilgy, Clrad State University, Frt Cllins, Clrad SYNOPSIS. The exchange f water between a chelnian egg and its subterranean envirnment is influenced by numerus factrs, the mst imprtant f which are (1) structure f the calcareus layer f the eggshell, (2) water ptential and temperature in the nest, and (3) fractin f the eggshell that actually cntacts sil in the nest cavity. Eggs with relatively prus shells tend t absrb large quantities f water frm cl, mist envirnments and t lse large amunts f water t warm, dry nes. Net water-exchange by such eggs als tends t be mre favrable (= psitive) when sil cntacts the entire eggshell than when a large fractin f the shell is expsed t air trapped inside the nest cavity. In cntrast, eggs with relatively impermeable shells usually exchange nly small amunts f water with their envirnment, regardless f the physical cnditins that prevail inside the nest. The pattern f net water-exchange, tgether with size (and water cntent) f the freshly laid egg, determines the amunt f water that is available t sustain the embry. An embry having access t a relatively large reserve f water will cnsume mre f its ylk and grw t larger size befre hatching than will an embry having access t a smaller reserve f water. Large, well-hydrated hatchlings may survive better than small, dehydrated animals during the trek verland frm nest t water. If s, a cler, wetter nest will als be a better nest. INTRODUCTION Female turtles usually emerge frm water nt land in late spring r early summer t lay their eggs (Ernst et al, 1994). Each female first lcates an apprpriate site t nest and then uses her hind feet t dig a flaskshaped hle in the grund. She subsequently depsits her clutch f eggs in the subterranean chamber, clses the hle with spil, and abandns the nest (Fig. 1), leaving the eggs t develp under physical cnditins dictated largely by prevailing weather, depth f the nest, type f sil, and slpe, aspect and expsure f the site (Packard and Packard, 1988a). Chelnian eggs cntain sufficient water at vipsitin t meet the needs f develping embrys. Hwever, the reservir f water in an egg may be augmented r depleted during incubatin, depending n the mrphlgy f the eggshell and the site where incubatin ccurs. This paper ad- 1 Frm the Sympsium Aquatic Organisms, Terrestrial Eggs: Early Develpment at the Water's Edge presented at the annual meeting f the Sciety fr Integrative and Cmparative Bilgy, 3-7 January 1998, at Bstn, Massachusetts. 2 packard@lamar.clstate.edu dresses the cnsequences t chelnian embrys f varying the size f the reserve f water t which they have access. I first cnsider mrphlgy f eggs and eggshells t prvide necessary perspective, and then fllw with a summary f labratry and field experiments that pint t manifld effects f water exchanges by eggs n physilgy f develping embrys. Finally, I end by cnsidering the imprtance t nenatal animals f envirnmentally-induced variatin in their size and cnditin. EGGS AND EGGSHELLS The amnitic eggs f turtles are similar in many respects t thse f birds (Ewert, 1985). An embry (which is a gastrula at the time f vipsitin) rests atp a large, spherical ylk that is lcated centrally within a large mass f albumen. The ylk cnsists primarily f lipid, prtein, and water (Janzen et al, 1990; Packard et al, 1981a; Rwe et al, 1995; Wilhft, 1986), and is the main surce f nutrients t sustain the develping embry. The albumen is primarily a reservir fr water, albeit mst f the liquid in this cmpartment is translcated int the vitelline sac during the first 289

2 290 GARY C. PACKARD Lepidchelys kempii Gaps between shell units Shell unit / \ Fibrus shell membrane Gechelne elephantpus FIG. 1. Schematic illustratin shwing the arrangement f eggs inside the subterranean nest that is characteristic f mst turtles. Eggs f sme species are nearly spherical, but thse f ther species are ellipsidal (Iversn and Ewert, 1991). Air spaces persist between eggs fr mst f incubatin in sme nests but nt in thers (Packard el al, 1998; Ratterman and Ackerman, 1989). Frm Packard et al. (1981b). week r s f incubatin (Mrris et al, 1983; Packard et al., 1981a, 1983; Webb et al, 1986). This mvement f water causes the vitelline sac t enlarge appreciably while the albumen is simultaneusly diminished in size (Packard et al., 1981a, 1983; Webb et al, 1986). After the first 7-10 days f incubatin, the albumen is represented nly as a thin, rubbery layer lying beneath the shell membrane. The ylk and albumen f a fresh egg are surrunded by a shell membrane (r tw) t which an uter, calcareus layer is attached (Ewert, 1985; Packard and Packard, 1988a). The calcareus layer is cmprised f individual shell units that are bnded with the underlying membrane (Fig. 2). The shell units in many species {e.g., mst chelniids, chelydrids, dermchelyids, emydids, and pelmedusids; many batagurids) are relatively shrt and squat, and are separated frm ther shell units by large gaps that expse the underlying membrane (Hirsch, 1983; M. J. Packard and Hirsch, 1986; M. J. Packard et al, 1982). Such eggs can be defrmed withut cracking the shell, and are said t be "flexible" r "pliable." Hwever, the shell units in ther species (e.g., carettchelyids, chelids, dermatemydids, kinsternids, testudinids, trinychids) Calcareus layer Fibrus shell membrane FIG. 2. Schematic illustratin f a transverse sectin thrugh the flexible eggshell f a sea turtle (Lepidchelys kempii, upper panel) and the rigid eggshell f a trtise (Gechelne elephantpus, lwer panel). Bth schematics are drawn t the same scale, s the greater thickness f the trtise eggshell is apparent. Mdified frm Packard and Packard (1988a). are relatively tall and they cmmnly abut ne anther, thereby cnferring rigidity t the eggshell (Hirsch, 1983; M. J. Packard and Hirsch, 1986; M. J. Packard et al, 1982). These eggshells are penetrated by relatively infrequent pres that pass between adjacent shell units (Fig. 2), and the shells cmmnly are rigid and brittle. Eggs with flexible shells are highly cnductive t water whereas eggs with rigid shells are nt (Feder et al., 1982; Packard et al, 1979; M. J. Packard et al, 1982; Wdall, 1984). The frmer cnsequently exchange water with their envirnment mre freely than d the latter. Fr example, flexible-shelled eggs f cmmn snapping turtles (Chelydra serpentina) absrb large amunts f water and increase in mass when they are incubated in wet envirnments (i.e., in sils having high water ptentials) but lse large quantities f water and decline in mass when they are incubat-

3 WATER RELATIONS OF TURTLE EGGS 291 ed in dry surrundings (Mrris et al, 1983; Packard and Packard, 19886, 1989). Thus, the water reserve in a flexible egg can be augmented substantially in sme settings and depleted significantly in thers. In cntrast, rigid-shelled eggs like thse f sftshell turtles (Apalne spinifera) are largely unaffected by the wetness f their envirnment, and their mass at the time f hatching is nly slightly lwer in dry envirnments than in wet nes (Packard et al., 19816). Thus, the water reserve in a rigid egg is nt augmented appreciably in a wet envirnment nr is it depleted very much in a dry ne, and embrys in these eggs rely largely (if nt exclusively) n water supplied by the female parent at vipsitin. Indeed, the water reserves in rigid-shelled eggs are insulated s well frm the envirnment that tempral and spatial variatin in the nest envirnment is unlikely t have an appreciable influence n the availability f water t develping embrys except under the mst extreme f cnditins (Leshem and Dmi'el, 1986). As just nted, water exchanges by eggs with flexible shells are affected prfundly by water ptential f the sil n which the eggs are incubated (Fig. 3). Hwever, water exchanges by such eggs als are influenced by temperature, with eggs that incubate at lw temperatures experiencing mre favrable (r psitive) patterns f net water-exchange than eggs that incubate at high temperature (Fig. 3). The underlying cause fr the influence f temperature n patterns f water exchange is unreslved (Ackerman et al., 1985; Packard et al., 1981a), but available evidence nnetheless supprts the claim that water exchanges by eggs are a functin f temperature as well as water ptential (Gutzke et al., 1987; Packard et al., 1987). The pattern f net water-exchange by a flexible egg is affected als by the prprtin f the eggshell making cntact with air trapped inside the chamber (Packard, 1991). Fr example, an egg that is buried fully in a wet substrate at a mderate temperature gains water thrughut incubatin whereas an egg that rests n the surface f the same substrate gains water fr apprximately tw-thirds f incubatin and lses water <3> <U 03 C as 12 -i 10 " , Wet 28.5 Wet 31.0 Wet 26.0 Dry 28.5 Dry 31.0 Dry Percentage f incubatin Wet / buried Wet / half Wet / surface 13 - Dry/buried 12 H "& Dry/surface Dry/half 11 H Day f incubatin FIG. 3. (Tp) Eggs frm cmmn snapping turtles (Chelydra serpentind) were half-buried in wet r dry vermiculite and then incubated at 26.0, 28.5, r 31.0 C. An analysis f cvariance fr repeated measurements indicated that the pattern f change in mass by eggs was affected by bth temperature and misture regime. See Packard et al. (1987) fr details. (Bttm) Eggs frm snapping turtles were buried t varying degrees in wet r dry vermiculite and incubated at 26.0 C. An analysis f cvariance fr repeated measurements indicated that the degree f cntact between eggs and the substrate affected the pattern f change in mass fr eggs n the wet substrate but nt the dry ne (F 14J89 = 14.79,, fr the interactin between time X water ptential X degree f cntact). Frm Packard (1991). thereafter (Fig. 3). The water reserve in the buried egg is increased appreciably but that f the ther egg is nly slightly larger near t hatching than it was at vipsitin (Fig. 3). The different patterns f respnse by these eggs may reflect differences in the relative imprtance f water being exchanged as liquid r vapr (Packard and Packard, 1988a), but this explanatin is disputed (Ackerman, 1991; Ackerman et al., 1985).

4 292 GARY C. PACKARD Regardless f the explanatin, hwever, the degree f cntact between an egg and the substrate affects the verall pattern f net water-exchange, at least by eggs n wet media (Fig. 3). LABORATORY STUDIES OF WATER RELATIONS Flexible-shelled eggs f snapping turtles and painted turtles (Chtysemys picta) have been studied frequently in the labratry, because bth species are abundant in temperate regins f Nrth America where numerus bilgists live and wrk. Indeed, eggs f these tw species have becme "mdel rganisms" in several labratries fr nging research n the water relatins f chelnian eggs and embrys. Hwever, eggs f enugh ther species have been studied t prvide firm supprt fr generalizatins emerging frm research n the "mdel rganisms" (Janzen et al., 1995; Packard et al., 1982, 1985a; Tucker et al, 1998). Embrynic turtles in flexible eggs mbilize nutrients frm the ylk quicker, grw faster, and incubate lnger in wet envirnments than in dry nes (Fig. 4; als Janzen et al, 1990; Miller and Packard, 1992; M. J. and G. C. Packard, 1986, 1989). Nt surprisingly, the metablism f embrys als is higher in wet cnditins than in dry nes (Gettinger et al, 1984; Miller and Packard, 1992). Cnsequently, turtles develping in wet envirnments attain larger size befre hatching than animals develping in dry sites, but the relatively large animals als have smaller masses f unused (r residual) ylk t sustain them in the pstnatal perid (Fig. 5; als Gutzke et al, 1987; Packard et al., 1988). Incubating at a relatively high temperature has an effect n embrys and hatchlings similar t that elicited by a dry envirnment (Fig. 5; als Gutzke et al., 1987; Packard et al., 1988). These several effects f the envirnment apparently are mediated via its influence n patterns f net water-exchange and cnsequent effects n the availability f water t develping embrys. Supprt fr this cntentin cmes frm investigatins in which flexible-shelled eggs were shifted between wet and dry substrates at different times Q 2.1 -] i Wet substrate Dry substrate «0.6 I" a Day f incubatin FIG. 4. Means fr dry mass f unused ylk (Tp) and dry mass f the carcass (Bttm) frm embrynic snapping turtles (Chelydra serpenlina) at different times during incubatin. Eggs were half-buried in wet r dry vermiculite and incubated at 29 C. Arrws identify means fr hatchlings. Errr bars represent ± Vi the Least Significant Difference fr distinguishing means at P = Data fr embrys and hatchlings were examined in separate analyses, s different errr bars apply t the tw grups. Errr bars fr embrys are limited t the final tw sampling dates t preserve clarity. Mdified frm Mrris et al. (1983). during incubatin in rder t induce different patterns f net water-exchange between eggs and their envirnment (Gutzke and Packard, 1986; Packard and Packard, 1988fc, 1989). I used stepwise, linear regressin (Dixn and Jennrich, 1988) t re-examine data fr live mass, dry mass f the carcass, and dry mass f the unused ylk frm hatchling snapping turtles in ne f these investigatins (Packard and Packard, 19886). Change in mass f eggs was ne independent variable in the new analyses, and a matrix f dummy variables identifying clutch-

5 WATER RELATIONS OF TURTLE EGGS 293 ISS { 2 8 " mass Q 1.6 -, i C "5 </> OS E e f T fx D - <k Wet Intermediate Dry Incubatin temperature ( C) FIG. 5. Means ± Yi the Least Significant Difference (P = 0.05) fr dry mass f the carcass (Tp) and unused ylk (Bttm) frm hatchling snapping turtles (Chelydra serpentma) that cmpleted incubatin at different temperatures n wet, intermediate, r dry vermiculite. Mdified frm Packard el al. (1987). es was the ther (Draper and Smith, 1966). I first frced "clutch" t enter as a set int the regressin mdels in rder t accunt fr variatin in hatchlings resulting frm differences in parentage (Chen, 1991), and then studied the residuals as a functin f net change in mass f eggs. A residual is merely the difference between an bserved value and the crrespnding value predicted frm the regressin mdel, and therefre is a measure f the variatin that has nt been explained by the mdel (Draper and Smith, 1966). I als studied water cntent f carcasses frm the hatchlings but frced dry mass f carcasses t enter first int that regressin mdel in rder t cmpensate fr variatin in bdy size f turtles in the sampie. Size (judged bth as live mass and as dry mass f the carcass) and water cntent f hatchlings were psitively crrelated with net change in mass f incubating eggs, whereas the mass f unused ylk was negatively crrelated with water exchange by eggs (Fig. 6). Thus, embrys in eggs that were in psitive water balance cnsumed mre f their ylk and grew t larger size befre hatching than did embrys in eggs that were in negative water balance, and hatchlings in the frmer grup were mre fully hydrated than animals in the latter grup. The state f desiccatin in turtles hatching in dry envirnments is evident als in data fr smtic cncentratin f bld plasma, which varies frm 290 mosm/kg in hatchlings frm wet envirnments t nearly 375 mosm/kg in nenates frm dry cnditins (Packard and Packard, 1989). Dehydrated turtles are unable t perfrm as well in tests f perfrmance as turtles with higher levels f tissue hydratin. Fr example, nenatal snapping turtles that cmpleted incubatin n wet r dry substrates were frced t run at maximum speed in a labratry test f lcmtr capacity (Miller et al, 1987). Turtles hatching in wet envirnments were able t run much faster than thse frm dry envirnments (Fig. 7), apparently because the frmer were larger than the latter (Miller et al, 1987). The animals next were placed in water fr 1 week and allwed t becme fully hydrated, after which tests f lcmtr capacity were repeated. Turtles in bth grups were able t run faster after they had becme fully hydrated, but the difference between animals hatching n wet and dry substrates persisted (Fig. 7). Thus, lcmtr capacity f nenates is influenced by bth their size and their physilgical cnditin at the time f hatching. A further effect f the envirnment n develping embrys ften is detected when eggs are incubated in cnditins leading t substantial lsses f water: embrys cmmnly succumb when their eggs becme severely desiccated (Fig. 8). This influence f the envirnment n survival by embrys almst always is detected in studies f the relatively small eggs f painted turtles (Gutzke and Packard, 1985; Packard et al, 1981a, 1989, 1991), but nly ccasinally in investigatins f the larger eggs f snap-

6 294 GARY C. PACKARD Change in mass f egg (g) Change in mass f egg (g) FIG. 6. Residuals fr (A) live mass f hatchling snapping turtles (Chelydra serpentina), (B) dry mass f their carcasses, (C) water cntent f their carcasses, and (D) dry mass f their unused ylk in relatin t net change in mass f eggs during incubatin at 29 C in the labratry. Data fr live mass and fr dry mass f carcass and ylk were adjusted by stepwise regressin t remve effects f clutch, and thse fr carcass water were adjusted by stepwise regressin t remve effects f bdy size. The partial crrelatin = 0.93 () fr live mass, 0.88 () fr dry mass f the carcass, 0.83 (/> < 0.001) fr water cntent f the carcass, and (P < 0.001) fr dry mass f unused ylk. Data frm Packard and Packard (1988b). 5 -i H Q. CO E Dry incubatin Wet incubatin m FIG. 7. Mean running speed (+2 SEM) fr hatchling snapping turtles (Chelydra serpentina) emerging frm eggs that were incubated n wet r dry vermiculite at 29 C. Turtles were tested shrtly after hatching (befre) and after spending 1 week in cntainers f water (after). Data frm Miller el al. (1987). ping turtles (Bbyn and Brks, 1994; Packard et al, 1987). The greater sensitivity f embrynic painted turtles prbably reflects the smaller amunt f water available inside fresh eggs f this species cmpared t fresh eggs f the snapping turtles. Eggs f painted turtles cmmnly weigh 5-6 g whereas thse f snapping turtles ften weigh g; fresh eggs f bth species cntain apprximately 70% water (Packard and Packard, 1980). Embrynic snapping turtles cnsequently have access t a larger reservir f water than d embrynic painted turtles, and this larger reserve f water presumably sustains the frmer under cnditins that prve lethal t the latter (Gutzke and Packard, 1985). Mrtality t embrys increases als with temperature (Fig. 8). This mrtality culd result frm increased lss f water t the envirnment, r it culd mean simply that the upper limit f thermal tlerance has been exceeded fr sme individuals. It is wrth nting in this regard that hatching

7 WATER RELATIONS OF TURTLE EGGS i E Incubatin temperature ( C) FIG. 8. Hatching success by snapping turtles (Chelydra serpentina) incubating in the labratry. Eggs were half-buried in wet, intermediate, r dry vermiculite and incubated at cnstant temperature. Analysis f the data using the Genmd Prcedure in SAS versin 6.12 (SAS Institute, 1996) indicates that hatching success was affected by water ptential (x 2 = 14.3, df = 2, ) and temperature (x 2 = 35.1, df = 2, ) and pssibly by their interactin (x 2 = 9.9, df = 4, P = 0.042). Frm Packard et al. (1987). success by snapping turtles in dry envirnments and at mderate temperatures is higher than that f animals develping in wet envirnments at high temperatures (Fig. 8). Eggs in the frmer setting actually lse mre water during incubatin than eggs in the latter envirnment, yet survival is greater (Fig. 3). Althugh embrynic turtles develping in flexible eggs are affected prfundly by water ptential and temperature f the envirnment (Gutzke and Packard, 1986; Gutzke et al, 1987; Packard and Packard, 1988ft, 1989; Packard et al, 1987), thse in rigid eggs usually are nt (Gettinger et al, 1984; Packard and Packard, 1990; Packard et al, 1981*; M. J. and G. C. Packard, 1991; Thmpsn, 1983). This des nt necessarily mean that the physilgy f embrys in rigid eggs differs in sme fundamental way frm that f embrys in flexible eggs. Mre than likely it means simply that the relatively impervius shell f rigid eggs cnstrains water exchanges s much that induced variatin in physilgy f embrys is minimal r absent. FIELD STUDIES OF WATER RELATIONS Results frm experiments perfrmed in the labratry lead t predictins (1) that survival by embrys in the field is affected by water ptential and temperature in the nest and (2) that an imprtant fractin f phentypic variatin amng hatchlings in the field is envirnmentally induced (Packard, 1991). Axe these predictins accurate, r were cnditins used t incubate eggs in the labratry experiments s unlike thse that eggs encunter in the field that results f the labratry research have n parallel in nature (Ackerman, 1991; Brks et al, 1991; Htaling et al, 1985; Ratterman and Ackerman, 1989)? Three studies have addressed this issue directly. Eggs f painted turtles were examined in ne study (Cagle et al., 1993), and thse f snapping turtles in the ther tw (Packard et al, 1993, 1998). In all three f these investigatins, the cntents f fresh nests were manipulated by reciprcal transplant. Thus, each nest received a cmplement f eggs frm each f several females, and the eggs f each female were distributed amng each f several nests. Design f the experiments therefre was basically factrial, s effects f the nest envirnment culd be distinguished clearly frm effects f maternity {i.e., genetics plus effects related t the nutritinal state and/r age f the mther). The eggs were recvered befre embrys began t pip their eggshells, and then were held under ptimal cnditins in the labratry until hatching. Phentypic differences amng clutchmates that cmpleted mst f their incubatin in different nests therefre prvide cnservative estimates fr effects f the envirnment n develping embrys. Survival by embrynic painted turtles varied significantly amng nests (Cagle et al, 1993), but that f embrynic snapping turtles did nt (Packard et al, 1998). Mre painted turtles hatched frm eggs incubated in cl, mist nests than frm eggs in warm, dry nests (Fig. 9). This utcme is in accrd with results f labratry experiments (see Fig. 8) and implicates water as the factr that is limiting t survival by embrys. The absence f an apparent effect f

8 296 GARY C. PACKARD 99 -] S- 90- CO CO I 50- > 30 - ij 1 OO Mean water ptential (MPa) 99 -i O O s 10 H x, Mean temperature ( C) FIG. 9. Survival by embrynic painted turtles (Chrysemys picta) in relatin t mean water ptential (Tp) and mean temperature (Bttm) at the bttm f nests in the field. Water ptential was better than temperature at explaining survival by embrys and cnsequently entered a lgistic regressin mdel at the first step (x 2 = 9.58, df = I, P < 0.01, fr water ptential; X 2 = 7.85, df = 1, P < 0.01, fr temperature). Mdified frm Cagle et al. (1993). the envirnment n survival by embrynic snapping turtles is nt entirely unexpected, and prbably reflects n the larger eggs and deeper nests f this species. The relatively large eggs f snapping turtles cntain a larger quantity f water than thse f painted turtles, thereby buffering embrynic snapping turtles better against lethal effects f desiccatin. Additinally, the deeper nests f the snapping turtles prbably prtect eggs better frm extremes f water ptential and temperature than the shallw nests f painted turtles. Thus, eggs f snapping turtles may nt be expsed in the field t cnditins s extreme as t lead t detectable mrtality. Sublethal effects f the envirnment are apparent in bth painted turtles and snap- TABLE 1. Cmpnents f variance attributable t clutch (i.e., maternity) and nest (i.e., envirnment) in analyses f variance fr live masses f hatchling snapping turtles and painted turtles. variatin Nest Clutch Residual Variance Snapping turtle 1 p <0.001 <0.001 Variance Painted turtle 1 ' P < "Frm Packard et al. (1998). b Data frm Cagle et al. (1993) were re-analyzed using the Mixed Prcedure in SAS versin 6.12; levels f significance fr cmpnents f variance were btained by using the "test" ptin fr randm variables in the GLM Prcedure f SAS (SAS Institute, 1996). ping turtles, hwever. Fr example, mre than 40% f the variatin in live mass (which is used here as a measure f size) f hatchling painted turtles resulted frm differences in the envirnment in different nests and little (if any) resulted frm differences in maternity (Table 1). In cntrast, nly abut 10% f the variatin in live mass f hatchling snapping turtles resulted frm differences amng nests, and the cntributin f maternity was substantial (Table 1). The difference in apparent imprtance f the envirnment in these studies prbably reflects again n differences in size f eggs and depth f nests in painted turtles and snapping turtles. In any event, an imprtant fractin f the variatin in bdy size detected amng hatchlings in the field results frm variatin in the nest envirnment (Table 1). The study f painted turtles and ne f the investigatins f snapping turtles als reprted measurements f temperature and water ptential at the bttm f the nest cavities. I re-examined data fr size and cnditin f hatchlings in these investigatins using the stepwise, linear regressin prcedure described earlier (Dixn and Jennrich, 1988), and cnfirmed that live mass f hatchling painted turtles and snapping turtles in these studies was crrelated psitively with water ptential in nests and negatively with temperature (Table 2). Thus, wet nests (i.e., thse at high water ptential) tended t prduce larger yung than dry nests did, and cl nests yielded

9 WATER RELATIONS OF TURTLE EGGS 297 TABLE 2. F-ralis, levels f significance, and partial crrelatin cefficients fr independent variables t enter stepwise regressin mdels fr live mass f hatchling snapping turtles and painted turtles. Independent variable (df) Clutches (16) A egg mass (1) Water ptential (1) Mean temperature (1) Clutches (14) A egg mass (1) Water ptential (1) Mean temperature (1) Step 1 Snapping turtles F = r = 0.87 F = r = 0.45 F = r = 0.30 F = 0.29 P = r = Painted turtles F = 5.40 r = 0.79 F = r = 0.62 F = P = r = 0.42 F = 6.66 P = r = Step 2 Entered F = r = 0.66 F = r = 0.30 F = r = Entered F = r = 0.70 F = r = 0.55 F = P = r = Step 3 Entered Entered F = 1.93 P = r = 0.09 F = 0.22 P = r = Entered Entered F = 0.02 P = 0.89 r = 0.02 F = 0.59 P = r = Data fr painted turtles frm Cagle et al. (1993); thse fr snapping turtles frm Packard et al. (1998). generally larger hatchlings than warm nests ptential) near t field capacity. Water cn- (Fig. 10). These findings were expected tent (and ptential) changes rapidly befrm results f research in the labratry, tween the dry layer and the wet layer. Turtle Hwever, neither water ptential nr tem- nests cmmnly are lcated in this transiperature was particularly gd at explaining tin zne, s eggs at the tp, middle, and variatin in live mass (Table 2) and this bttm f a nest may be expsed t very result cntrasts markedly with that frm ex- different misture regimes (Ackerman, periments in the labratry, where phen- 1991). If this is the case, measurements f typic variatin is tightly cupled t differ- water ptential at the bttm f nests f ences in the physical envirnment (Fig. 5). painted turtles and snapping turtles are un- This utcme f the field studies prba- likely t have yielded accurate infrmatin bly reflects the limitatins f existing data n the availability f water t all the eggs fr water ptential and temperature in nests, in a given nest. The limited explanatry Fr example, water drains dwnward value f water ptential is nt surprising thrugh the sil prfile in the days fllw- when viewed frm this perspective (Table ing a saturating rainfall, and evapratin si- 2). multaneusly cmmences drying f the sur- Data fr temperatures at the bttm f face layer. The upper 1-2 cm f the sil nests als are limited in value. Temperature cmmnly becmes extrardinarily dry at the tp f nests cmmnly fllws a diel while sil nly cm deeper in the pr- cycle that is distinctly different frm that file retains a quantity f water (and water ccurring at the bttm. Means are higher,

10 298 GARY C. PACKARD CO 2 -i en. 2-3 «1 r r = , -2 - Mean water ptential (MPa) r = Mean temperature ( C) FIG. 10. Residuals frm the regressin f live mass f hatchhng snapping turtles (Chelydra serpentina) against clutch (entered as a dummy variable) are pltted against means fr water ptential (Tp) and temperature (Bttm) in 10 nests in the field. The residuals crrect data fr live mass fr variatin amng ffspring frm different clutches. Frm Packard et al. (1998). and variances are greater, at the tp than at the bttm (Gerges, 1992; Packard et al, 1985 >; Ratterman and Ackerman, 1989; Thmpsn, 1988; Wilhft et al, 1983). Thus, mean values fr temperature at the bttm f nests in the present studies may apply t eggs lcated at the bttm, but they are unlikely t apply t eggs nearer t the surface. Additinally, eggs in natural nests cmmnly vary in the degree f cntact that they make with sil. Air spaces in sme nests becme bliterated by infiltrating sil, but thse in ther nests persist fr the duratin f incubatin (Cagle, 1950; Packard et al, 1998; Ratterman and Ackerman, 1989). Thus, eggs in sme nests may have a substantial fractin f their surface cntacting air whereas eggs in ther nests may be cmpletely encased in a matrix f sil. Indeed, similar variatin may ccur between the periphery and interir f a single nest. Such variatin prbably affects water exchanges by eggs prfundly (Fig. 3). Limitatins f data fr water ptential and temperature in nests, cupled with variatin amng and within nests in the fractin f each egg making cntact with the substratum, raise the pssibility that changes in mass f eggs are a better index t cnditins affecting eggs and embrys than means fr water ptential and temperature. Indeed, the stepwise regressins f data fr live mass f hatchlings supprt this view, because the partial crrelatins and F-ratis fr net change in mass f eggs are higher than thse fr either water ptential r temperature (Table 2). Further analyses cnfirmed that live mass f hatchling painted turtles and snapping turtles was psitively crrelated with water exchanges by their eggs (Fig. 11). These analyses als revealed that the dry mass and water cntent f carcasses f nenatal snapping turtles were crrelated psitively with water exchanges by eggs (Fig. 11), and that dry mass f unused ylk was negatively crrelated with change in mass f eggs (Fig. 11). Embrys develping in eggs with a net uptake f water used mre f the nutrients in the ylk t supprt metablism and grwth and grew t larger size befre hatching than did embrys in eggs that lst water t their surrundings. Cnsequently, the mass f unused ylk was smaller in animals hatching frm eggs in psitive water balance than in thse hatching frm eggs in negative water balance. The hatchlings als were mre fully hydrated when incubatin was cmpleted in wet envirnments than in dry nes. These findings are in full accrd with predictins emerging frm experiments perfrmed in the labratry (Fig. 6), and raise the pssibility that hatchlings in the field vary appreciably in their capacity fr mvement verland t water (Miller et al, 1987).

11 WATER RELATIONS OF TURTLE EGGS 299 CO "<6 d) 2 -i A <%* CO CO s esiidual fr c B c , C D " I CO "S** Q " M O sib I -. Wmm ?^l O) r i.o - g 't! (U Change in mass f egg (g) Change in mass f egg (g) FIG. 11. Residuals fr (A) live mass f hatchling snapping turtles (Chelydra serpentina), (B) dry mass f their carcasses, (C) water cntent f their carcasses, and (D) dry mass f their unused ylk in relatin t net change in mass f eggs during incubatin in natural nests. Data fr live mass and fr dry mass f carcass and ylk were adjusted by stepwise regressin t remve effects f clutch, and thse fr carcass water were adjusted by stepwise regressin t remve effects f bdy size. The partial crrelatin = 0.58 () fr live mass, 0.47 () fr dry mass f the carcass, 0.50 () fr water cntent f the carcass, and (P < 0.001) fr dry mass f unused ylk. Frm Packard et al. (1998). IMPORTANCE OF WATER EXCHANGES What is the eclgical imprtance f water exchanges by eggs incubating in the field? The answer t this questin is apparent when the eggs lse (r gain?) s much water as t affect survival by embrys, but the answer is nt s clear when effects f water exchange are sublethal. Des envirnmentally induced variatin in size, physilgical cnditin, r lcmtr capacity f nenatal turtles have any imprtance fr animals emerging frm nests in the field (Packard and Packard, 1988a)? The relatively large, well-hydrated animals hatching in wet nests might be better than the relatively small, dehydrated turtles emerging in dry nests at aviding predatrs, wing t the prtectin that ften is affrded t large animals by virtue f their size (Fig. 5) r that might attend the superir lcmtr ability f large turtles (Fig. 7). Alternatively, the relatively small animals hatching in dry envirnments might survive better than the large nes frm wet envirnments because the small turtles have larger masses f unused ylk t supprt them during the perid after hatching (Fig. 5), when fd may nt be as readily available as is generally suppsed (Brks et al., 1991). Unfrtunately, nly ne attempt has been made t assess this general questin empirically. This ne investigatin merits special mentin because the study prvides an extrardinarily useful mdel fr future research, yet flaws in design and analysis undermine cnclusins emerging frm the wrk. Janzen (1993) incubated eggs f cmmn snapping turtles n wet and dry substrates in the labratry in rder t prduce hatchlings that varied appreciably in size and physilgical cnditin. He marked the animals fr individual identificatin and then released them n a hillside abve

12 300 GARY C. PACKARD the Mississippi River, at a site knwn t be used fr nesting by gravid females. Janzen als placed a drift fence a few meters frm the river t intercept hatchlings as they mved dwn the hill tward the water (i.e., much like turtles wuld mve tward water after they had emerged frm a natural nest). Animals that were recaptured in pitfalls at the drift fence were assumed t have survived the trek t water and thse that were nt recaptured were assumed t have died during their verland jurney. The data then were examined fr sme indicatin that the prbability f recapturing a hatchling was affected by its size, the envirnment that it encuntered during incubatin, r its lcmtr capacity. Janzen cncluded that larger turtles were mre likely than smaller nes t survive the test, albeit the analysis did nt distinguish between effects f size related t maternity and thse related t the envirnment during incubatin. Nnetheless, the cnclusins are nt well funded. Janzen held the turtles in water between the time f hatching and the time that he released them in the field, s the hatchlings were able t ingest water and becme fully hydrated prir t being tested fr their ability t survive under natural cnditins. This prcedure fr handling animals prbably was inapprpriate, given that ne f his gals was t assess the eclgical imprtance f envirnmentally induced variatin. Hatchlings in the field seldm have access t water fr drinking until they actually enter the bdy f water t which they are immigrating, and the lcmtr ability f Janzen's animals prbably was altered significantly by this prcedure (Fig. 7). The utcme f the experiment might have been very different had the hatchlings been denied access t water prir t their being released in the field. In additin, the statistical analysis perfrmed by Janzen was inapprpriate, and may have led him t reach unwarranted cnclusins. First, he multiplied a dichtmus respnse (i.e., recapture = survive = 1, n recapture = succumb = 0) by a cnstant t "transfrm" the riginal variable int a pseud-cntinuus metric that he termed "relative fitness." Survivrs were assigned values f 1.7 fr relative fitness whereas nn-survivrs were assigned values f 0 (Janzen, 1993, persnal cmmunicatin; als see Brdie and Janzen, 1996), thereby magnifying the apparent differences in respnse. Janzen then studied data fr relative fitness by multiple linear regressin, and used the number f eggs in each clutch as ne f the ptential independent variables. Hwever, the cmputatin f "relative fitness" des nt alter the fact that the respnse variable is dichtmus, s multiple lgistic regressin shuld have been used instead f multiple linear regressin t examine the data (James and McCullch, 1990). Mrever, his use f clutch size as an independent variable is questinable, because the resultant analysis assessed survival by ffspring f different females nly in relatin t the number f eggs and nt in relatin t their quality. A better apprach wuld have been t cnstruct a matrix f dummy (r lcatin) variables t identify the different clutches and then t use the matrix as a set in lgistic regressin (Chen, 1991). Despite the fact that cnclusins frm Janzen's wrk are questinable, the imprtance f his study shuld nt be underestimated. He tk the critical step f trying t assess the imprtance f envirnmentallyinduced variatin under cnditins that turtles actually encunter in the field. He accrdingly has prvided imprtant directin fr future research. The fundamental questin that he addressed remains unanswered, hwever. ACKNOWLEDGMENTS I thank Karen Martin and Richard Strathmann fr inviting me t participate in this sympsium. Much f the wrk that I have summarized in this essay resulted frm cllabratins ver a perid f 20 years with numerus clleagues and students, and I am grateful t them all. Hwever, I want especially t acknwledge the majr cntributins made by G. F. Birchard, T. J. Bardman, L. L. McDaniel, K. Miller, and M. J. Packard. A draft f this essay was critiqued fr me by L. M. Hartley, M. J. Packard, P. Sims, and R. Willard. My research n rep-

13 WATER RELATIONS OF TURTLE EGGS 301 tilian eggs and embrys was supprted mst recently by NSF grant IBN REFERENCES Ackerman, R. A Physical factrs affecting the water exchange f buried reptile eggs. In D. C. Deeming and M. W. J. Fergusn (eds.), Egg incubatin: Its effects n embrynic develpment in birds and reptiles, pp Cambridge University Press, Cambridge, UK. Ackerman, R. A., R. C. Seagrave, R. Dmi'el, and A. AT Water and heat exchange between parchment-shelled reptile eggs and their surrundings. Cpeia 1985: Bbyn, M. L. and R. J. Brks Incubatin cnditins as ptential factrs limiting the nrthern distributin f snapping turtles, Chelydra serpentina. Can. J. Zl. 72: Brdie, E. D. Ill and F. J. Janzen On the assignment f fitness values in statistical analyses f selectin. Evlutin 50: Brks, R. J., M. L. Bbyn, D. A. Galbraith, J. A. Layfield, and E. G. Nancekivell Maternal and envirnmental influences n grwth and survival f embrynic and hatchling snapping turtles {Chelydra serpentina). Can. J. Zl. 69: Cagle, F. R The life histry f the slider turtle, Pseudemys scripta trstii (Hlbrk). Ecl. Mngr. 20: Cagle, K. D., G. C. Packard, K. Miller, and M. J. Packard Effects f the micrclimate in natural nests n develpment f embrynic painted turtles, Chrysemys picta. Funct. Ecl. 7: Chen, A Dummy variables in stepwise regressin. Am. Statist. 45: Dixn, W. J. and R. Jennrich Stepwise regressin. In W. J. Dixn (ed.), BMDP statistical sftware manual, pp University f Califrnia Press, Berkeley. Draper, N. R. and H. Smith Applied regressin analysis. Jhn Wiley & Sns, New Yrk. Ernst, C. H., R. W. Barbur, and J. E. Lvich Turtles f the United States and Canada. Smithsnian Institutin Press, Washingtn, DC. Ewert, M. A Embrylgy f turtles. In C. Gans, F. Billett, and P. F. A. Madersn (eds.), Bilgy f the Reptilia, Vl. 14, pp Jhn Wiley & Sns, New Yrk. Feder, M. E., S. L. Satel, and A. G. Gibbs Resistance f the shell membrane and mineral layer t diffusin f xygen and water in flexibleshelled eggs f the snapping turtle (Chelydra serpentina). Respir. Physil. 49: Gerges, A Thermal characteristics and sex determinatin in field nests f the pig-nsed turtle, Carettchelys insculpta (Chelnia: Carettchelydidae), frm nrthern Australia. Aust. J. Zl. 40: Gettinger, R. D., G. L. Paukstis, and W. H. N. Gutzke Influence f hydric envirnment n xygen cnsumptin by embrynic turtles Chelydra serpentina and Trinvx spiniferus. Physil. Zl. 57: Gutzke, W. H. N. and G. C. Packard Hatching success in relatin t egg size in painted turtles (Chrysemys picta). Can. J. Zl. 63: Gutzke, W. H. N. and G. C. Packard Sensitive perids fr the influence f the hydric envirnment n eggs and hatchlings f painted turtles (Chrysemys picta). Physil. Zl. 59: Gutzke, W. H. N., G. C. Packard, M. J. Packard, and T. J. Bardman Influence f the hydric and thermal envirnments n eggs and hatchlings f painted turtles (Chrysemys picta). Herpetlgica 43: Hirsch, K. F Cntemprary and fssil chelnian eggshells. Cpeia 1983: Htaling, E. C, D. C. Wilhft, and S. B. McDwell Egg psitin and weight f hatchling snapping turtles, Chelydra serpentina, in natural nests. J. Herpetl. 19: Iversn, J. B. and M. A. Ewert Physical characteristics f reptilian eggs and a cmparisn with avian eggs. In D. C. Deeming and M. W. J. Fergusn (eds.), Egg incubatin: Its effects n embrynic develpment in birds and reptiles, pp Cambridge Univ. Press, Cambridge, UK. James, F. C. and C. E. McCullch Multivariate analysis in eclgy and systematics: Panacea r Pandra's bx? Annu. Rev. Ecl. Syst. 21: Janzen, F. J Persnal Cmmunicatin. Department f Zlgy and Genetics, Iwa State University, Ames, Iwa. Janzen, E J An experimental analysis f natural selectin n bdy size f hatchling turtles. Eclgy 74: Janzen, F. J., G. C. Packard, M. J. Packard, T. J. Bardman, and J. R. zumbrunnen Mbilizatin f lipid and prtein by embrynic snapping turtles in wet and dry envirnments. J. Exp. Zl. 255: Janzen, F. J., J. C. Ast, and G. L. Paukstis Influence f the hydric envirnment and clutch n eggs and embrys f tw sympatric map turtles. Funct. Ecl. 9: Leshem, A. and R. Dmi'el Water lss frm Trinyx triunguis eggs incubating in natural nests. Herpetl. J. 1: Miller, K. and G. C. Packard The influence f substrate water ptential during incubatin n the metablism f embrynic snapping turtles (Chelydra serpentina). Physil. Zl. 65: Miller, K., G. C. Packard, and M. J. Packard Hydric cnditins during incubatin influence lcmtr perfrmance f hatchling snapping turtles. J. Exp. Bil. 127: Mrris, K. A., G. C. Packard, T. J. Bardman, G. L. Paukstis, and M. J. Packard Effect f the hydric envirnment n grwth f embrynic snapping turtles (Chelydra serpentina). Herpetlgica 39: Packard, G. C Physilgical and eclgical imprtance f water t embrys f viparus reptiles. In D. C. Deeming and M. W. J. Fergusn

14 302 GARY C. PACKARD (eds.), Egg incubatin: Its effects n embrynic develpment in birds and reptiles, pp Cambridge University Press, Cambridge, UK. Packard, G. C, K. Miller, and M. J. Packard Envirnmentally induced variatin in bdy size f turtles hatching in natural nests. Oeclgia 93: Packard, G. C, K. Miller, M. J. Packard, and G. F Birchard Envirnmentally induced variatin in bdy size and cnditin in hatchling snapping turtles {Chelydra serpentina). Can. J. Zl. (In press.) Packard, G. C. and M. J. Packard Evlutin f the cleidic egg amng reptilian antecedents f birds. Amer. Zl. 20: Packard, G. C. and M. J. Packard. 1988a. The physilgical eclgy f reptilian eggs and embrys. In C. Gans and R. B. Huey (eds.), Bilgy f the Reptilia, Vl. 16, pp Alan R. Liss, New Yrk. Packard, G. C. and M. J. Packard Water relatins f embrynic snapping turtles {Chelydra serpentina) expsed t wet r dry envirnments at different times in incubatin. Physil. Zl. 61: Packard, G. C. and M. J. Packard Cntrl f metablism and grwth in embrynic turtles: A test f the urea hypthesis. J. Exp. Bil. 147: Packard, G. C. and M. J. Packard Grwth f embrynic sftshell turtles is unaffected by uremia. Can. J. Zl. 68: Packard, G. C, M. J. Packard, and L. Benigan Sexual differentiatin, grwth, and hatching success by embrynic painted turtles incubated in wet and dry envirnments at fluctuating temperatures Herpetlgica 47: Packard, G. C, M. J. Packard, and G. E Birchard Sexual differentiatin and hatching success by painted turtles incubating in different thermal and hydric envirnments. Herpetlgica 45: Packard, G. C, M. J. Packard, and T. J. Bardman. 1981a. Patterns and pssible significance f water exchange by flexible-shelled eggs f painted turtles. Physil. Zl. 54: Packard, G. C, M. J. Packard, and T. J. Bardman An experimental analysis f the water relatins f eggs f Blanding's turtles (Emydidea blandingii). Zl. J. Linn. Sc. 75: Packard, G. C, M. J. Packard, T. J. Bardman, K. A. Mrris, and R. D. Shuman Influence f water exchanges by flexible-shelled eggs f painted turtles Chrysemys picta n metablism and grwth f embrys. Physil. Zl. 56: Packard, G. C, M. J. Packard, and W. H. N. Gutzke 1985a. Influence f hydratin f the envirnment n eggs and embrys f the terrestrial turtle Terrapene rnala. Physil. Zl. 58: Packard, G. C, M. J. Packard, K. Miller, and T. J Bardman Influence f misture, temperature, and substrate n snapping turtle eggs and embrys. Eclgy 68: Packard, G. C, M. J. Packard, K. Miller, and T. J. Bardman Effects f temperature and misture during incubatin n carcass cmpsitin f hatchling snapping turtles {Chelydra serpentina). J. Cmp. Physil. B 158: Packard, G. C, G. L. Paukstis, T. J. Bardman, and W. H. N. Gutzke Daily and seasnal variatin in hydric cnditins and temperature inside nests f cmmn snapping turtles {Chelydra serpentina). Can. J. Zl. 63: Packard, G. C, T. L. Taigen, M. J. Packard, and T. J. Bardman Changes in mass f eggs f sftshell turtles {Trinyx spiniferus) incubated under hydric cnditins simulating thse f natural nests. J. Zl. 193: Packard, G. C, T. L. Taigen, M. J. Packard, and R. D. Shuman Water-vapr cnductance f testudinian and crcdilian eggs (class Reptilia). Respir. Physil. 38:1-10. Packard, M. J. and K. E Hirsch Scanning electrn micrscpy f eggshells f cntemprary reptiles. Scanning Electrn Micrsc. 1986/IV: Packard, M. J. and G. C. Packard Effect f water balance n grwth and calcium mbilizatin f embrynic painted turtles {Chrysemys picta). Physil. Zl. 59: Packard, M. J. and G. C. Packard Envirnmental mdulatin f calcium and phsphrus metablism in embrynic snapping turtles {Chelydra serpentina). J. Cmp. Physil. B 159: Packard, M. J. and G. C. Packard Surces f calcium, magnesium, and phsphrus fr embrynic sftshell turtles {Trinyx spiniferus). J. Exp. Zl. 258: Packard, M. J., G. C. Packard, and T. J. Bardman Structure f eggshells and water relatins f reptilian eggs. Herpetlgica 38: Ratterman, R. J. and R. A. Ackerman The water exchange and hydric micrclimate f painted turtle (Chrysemys picta) eggs incubating in field nests. Physil. Zl. 62: Rwe, J. W, L. Hly, R. E. Ballinger, and D. Stanley- Samuelsn Lipid prvisining f turtle eggs and hatchlings: Ttal lipid, phsphlipid, triacylglycerl and triacylglycerl fatty acids. Cmp. Bichem. Physil. B 112: SAS Institute Inc SAS/STAT sftware: Changes and enhancements thrugh release SAS Institute, Cary, NC. Thmpsn, M. B The physilgy and eclgy f the eggs f the pleurdiran trtise Emydura macquarii (Gray), Ph.D. Thesis, University f Adelaide, Suth Australia. Thmpsn, M. B Nest temperatures in the pleurdiran turtle, Emydura macquarii. Cpeia 1988: Tucker, J. K., N. I. Filram, G. L. Paukstis, and F. J. Janzen Respnse f red-eared slider, Trachemys scripta elegans, eggs t slightly differing water ptentials. J. Herpetl. 32: Webb, G. J. W, D. Chquent, and P. J. Whitehead Nests, eggs, and embrynic develpment f Carettchelys insculpta (Chelnia: Carettcheli-

15 WATER RELATIONS OF TURTLE EGGS 303 dae) frm nrthern Australia. J. Zl. (B) 1:521- brys f the snapping turtle, Chelydra serpentina J. Herpetl. 17: Wilhft, D. C Eggs and hatchling cmpnents Wdall, P. E The structure and sme functinal f the snapping turtle (Chelydra serpentina). aspects f the eggshell f the brad-shelled river Cmp. Bichem. Physil. 84A: trtise Cheldina expansa (Testudinata: Cheli- Wilhft, D. C, E. Htaling, and P. Franks Efdae >- Aust - J - z 1-32:7-14. fects f temperature n sex determinatin in em- Crrespnding Editr: Paul Verrell