Reference: Biol Bull 187: 373-384. (December. 1994) Nest-Site Selection in the Horseshoe Crab, Limulus polyphemus DUSTIN PENN AND H. JANE BROCKMANN Department ofZoology, University ofFlorida, Gainesville, Florida 32611 Abstract. Like a number of other species, horseshoe Brockmann, 1983) and Massachusetts (Cavanaugh, 1975; crabs lay theireggs on beaches in the intertidal zone. The Barlow el a/., 1986; Badgerow and Sydlik, 1989) nest in elevation ofthe beach on which they nest differs among a narrow band in the upper middle quarter ofthe beach, populations. We examined two factors that potentially whereas crabs in the Delaware Bay nest in a wide band affect egg survival at different beach elevations: erosion overmost ofthewave-swept section ofthebeach (Shuster, and rate ofdevelopment. We found noevidencethateggs 1982; Shusterand Botton, 1985; Botton etal.. 1992). The buried at different elevations incur different risks ofero- adaptive significance ofnest-site preferencesofhorseshoe sionbywaveaction. However,theoptimalbeachelevation crabs is unclear, but it has been suggested that they nest foreggdevelopment differed between ourtwostudy sites, at beach elevations where environmental conditions are Florida and Delaware, and the difference was related to most conducive to egg development (Lockwood, 1870; beach morphology. Rate ofdevelopment increased with Shuster, 1982; Badgerowand Sydlik, 1989). Anotherpos- oxygen concentration, redox potential, and temperature, sibility is that horseshoe crabs nest at elevations that and all three ofthese factorschanged with elevation. The minimize the loss ofeggs due to beach erosion. In this nests in the lower beach failed to develop, especially in paper we examine these hypotheses and the costs and Florida, where the beach was fine-grained and poorly benefits ofthe nest-site preferences ofhorseshoe crabs. drained. The nests in the upper beach were prone to des- Horseshoe crabs spawn during the spring and early sication. especially in Delaware, where the beach was summer on certain beaches along the Atlantic and Gulf course-grained and well-drained. This means that differ- coasts ofthe United States, and in the Yucatan, Mexico ences between sites in the optimal location for egg devel- (Shuster. 1982). At high tide females bury themselves in opmentcoincidedwith horseshoecrabpreferences in nest- the sediments near the water's edge and lay a series of siteselection. Wesuggest that horseshoecrabssynchronize discrete egg clusters, each containing thousands ofeggs their nesting with the tides that reach the aerobic sedi- (Brockmann. 1990). These eggs are fertilized by sperm mentson thebeach, resultingin nestingpatternsthatdiffer released byan attached male and by oneormore satellite with differences in tidal regimes and beach morphology. malesthattypicallycongregatearoundthe nestingcouple (Rudloe, 1980; Brockmann. 1990; Brockmann and Penn, Introduction 1992; Brockmann et al., 1994). The eggs develop in sed- Horseshoe crabs (Limulus polyphemus) synchronize ibmeeancthssu5rftaoce20(Rcumdlo(em,ea1n9791;1.B5roc2k.m8anSnD,c1m9)90b).elAofwtetrh2e their spawning with flood tides, especially the highest to 4 weeksand fourembryonic molts, the embryos hatch (spring) tides associated with lunar syzygies (Lockwood. into"trilobite" larvae(Kingsley, 1892, 1893; Patten, 1896; 1870; Rudloe, 1980; Cohen and Brockmann, 1983; Bar- Sekiguchi et al.. 1982; Sekiguchi, 1988). The nonfeeding low el a/.. 1986). Although they deposit theireggs in sedi- larvae remain in the sand in distinct aggregations forsev- mentsfoundonthe midtoupperbeach, nestingelevation eraladditional weeks, untiltheyentertheocean bymoving varies among populations (Shuster. 1982). For example, toward the beach surface when tidal inundation occurs horseshoe crabs in Florida (Rudloe, 1980; Cohen and again (Rudloe, 1979). Within 2 weeks, thefree-swimming trilobite larvae molt into juveniles that live in the near- Received 17 September 1993;accepted 28 September 1994. shore sand flats (Sekiguchi et al.. 1982). 373 374 D. PENN AND H. J. BROCKMANN The preference that horseshoe crabs show for nesting ofmixed sand and gravel locatedjust inside the mouth in the upper elevations ofthe beach is puzzling because ofthe Delaware Bay at the Cape Henlopen State Park it is associated with several costs. (1) Individuals risk (3847'N. 7506' W). strandingon nestingbeaches, which resultsin physiologi- cal stress ordeath from desiccation or predation (Botton Beach measurements and Loveland, 1989; Penn and Brockmann, in press). To quantify beach elevation, we measured the oblique Strandingisriskierat higherbeach elevationsfortworea- distancefrom azeropointonthebeachtolocationshigher sons: subsequent tidesare lesslikely to "rescue" stranded on the beach ("Beach Distance"). The zero point wasthe crabs; and beach slope, which is necessary for orienting bottom ofthe beach where the mud flat ended and the to the sea (Botton and Loveland, 1987; pers. obs.), di- beach began; i.e.. the slope and sediment composition minishes in the upper beach. (2) Eggs located higher on changed abruptly. Beach distance was more practical to thebeach are likely to incurgreater extremesoftempera- measure than elevation and the variationsin beach slope, ture and moisture and therefore may risk desiccation 2-5, were insufficient to affect our measurements. We (Middaugh ct al., 1983). (3)Nestingonthebeach requires quantified the location at which eggswere laid by placing thathorseshoecrabssynchronizespawningwith hightides. wire flagging stakes on either side of the female during but this synchronization leaves individuals fewer oppor- nesting. We marked each egg cluster that the female laid tunitiestoreproduceandincreasescompetition formates by putting these flags across the female's hinge each time and nesting sites (Ims, 1990). shemovedforwardinthesand;thisprocedureminimized These costs suggest that there is some compensating the possibility ofeggs from other females contaminating benefittothechoiceofspawningsitesby horseshoecrabs. the eggclusters (Brockmann. 1990). At low tide we mea- One possible explanation is that horseshoe crabs nest at sured the distance from the bottom ofthe beach to the elevations that optimize egg development (Lockwood, center ofeach set offlags ("Nesting Distance"). We also 1870; Badgerow and Sydlik, 1989). Shuster (1982) sug- measured the water depth in which females nested and gested that the development ofLimulus eggs depends on theirdistancestothewater'sedge. Wequantified the tim- acombination oftemperature, moisture, and oxygen gra- ing ofnestingby countingthe number ofnestingcouples dients on the beach, which vary according to local tidal on the beach on three tides for 3 hours before and after amplitude. Sea turtles nest high on the beach because the MHT in FL (1989) and DE (1991). "Tidal Distance" hatching success is increased by burying eggs in well-aer- wasquantified bymeasuringtheextentoftidal inundation atedsediments(Haysand Speakman, 1993; Horrocksand on the beach, using wooden stakes that were driven into Scott, 1991). A second possible advantage is that wave the beach at 2-m intervals(measured asbeach distances). action maybelesslikelytoexposeorwashawayhorseshoe We used tidaldata recorded bythe National Oceanic and crab eggs at certain beach elevations. Some intertidally Atmospheric Administration (at Cedar Key, FL, and spawning fishes, such asgrunion, deposit theireggsin the Lewes, DE) to calculate mean high tide line (MHTL) for upperintertidal toavoidbeacherosion in thelowerbeach each. NOAA tide heights and ourtidaldistance measure- (Thompson, 1919; Taylor, 1984). To test hypotheses, we mentswere highly correlated (FL r = 0.92; DE r = 0.93), quantified the nesting locations of two populations of which enabled us to interpolate the MHTL (95% CD in horseshoe crabs, Florida and Delaware, and conducted tidal distance for April in FL and May in DE, 1991. experiments that evaluated the effect of environmental variables and erosion on egg development and survival. Eggdevelopment The results oftheseexperiments provide evidence forthe adaptiveness of nest-site selection and spawning syn- Totestthehypothesisthateggdevelopmentwasaffected chrony in horseshoe crabs. by beach elevation, we reburied newly laid eggs at three different beach distances and examined the eggs after 10 Materials and Methods days (1990). After marking nests, we excavated and col- lected the freshly laid eggs at low tide from 25 (FL) and Studv sites 22 (DE) different nests, broke up the clusters, separated We studied horseshoe crabs at two locations in the the eggs from the sediments, and measured their volume United States: Florida (25 May to 5 June 1990 and 27 (Brockmann, 1990). To separate the eggswe used a sieve March to 31 April 1991 ) and Delaware ( 13 to 23 July ( 1 mm mesh) in FL, but the mixed sand-gravel beach in 1990 and May to 15 June 1991). The Florida (FL) DE required that we elutriatethe eggs from thesediments. 1 1 site was a low-energy sandy beach on the south shore We reburied 60 (FL) and 90 (DE) egg clusters ("Egg ofSeahorse Key, a small island in the GulfofMexico, Batches") 8 cm below the sand surface at three different 4 km from Cedar Key (2906' N, 8304' W), Levy beach distances (3, 5, and 8 m) at 20 (FL) and 30 (DE) County. The Delaware site(DE) wasa low-energy beach locationsalong 1 km ofbeach, markingtheirlocation with NEST-SITE SELECTION IN UML'LL'S 375 flaggingstakes. Topreventegglossorcontamination from situ with a thermistor probe inserted 10cm below the other nests, half of the batches were placed in "cages" surface, and measurements were taken three (DE) or five mm (2 nylon mesh bagssealed with Velcro). After 10days (FL) times for each egg batch (means used in analyses). we excavated the egg batches and quantified their devel- (2) Moisture content (%H2O)was measured by collecting opment in two ways. First, we estimated the percent of sediment samples near each egg batch (in sealed plastic developed eggs by visually inspecting the developmental containers), weighing (within 2 h), drying(oven baked at stages ofa 1-ml sample ofeggs under a dissecting micro- 105C), and reweighing (%H2O = [(wet weight - dry scope, separating undeveloped eggs from embryos (i.e., weight)/wet weight] X 100). (3) Salinity ofthe interstitial those with limb buds which occur during stage 13; Seki- water (10 cm from surface) was measured in FL with a guchi el al.. 1982), and counting the number to obtain a refractometer. (4) Interstitial oxygen concentration ([O2]) "Percent Eggs Developed" within each batch. Second, we could not be measured directly because %H2O varies at measured the increase in the volumeofeggbatches("Egg different beach distances. Instead we used a portable YSI Batch Volume") due to developmental swelling (i.e.. the oxygen meter to measure the dissolved [O2] in 120 ml of 1.7-mm Linnilus eggs become 3.6-mm embryos before seawater inside a plastic bag that we buried in the sedi- hatching). For example, volume of egg batches (due to ments near the egg batches. Dissolved [O2] within the developmental swelling) increased significantly with de- bagsequilibrateswith thesurrounding [O2] sedimentsbe- v=el1o0.p6m,enPta=l0.r0a1t)e.(Cdheavnelgoepsmienntvaolluimndeewx)er(er2ca=lc0u.l6a5t.edF,by-8 cEavuasnes,po1l9y6e3t)h.yTlewnoe i(sFpLe)romreatbhrleeet(oDoEx)ygmeenas(uFrreemmleinntgsapnedr dividing the volume at 10 days by the initial volume of bagweretaken(mean used inanalyses). Wealsocollected abantecghgesbavtacrhie(dI(22/01-',5)0bemcla)u.sVeatrhieatiiniotniailnvioniltuiamlevoofltuhmeeegogf s(5e0dimmle)nttso (e5a0chmls)amnpelaer, eaancdhtehgegnbdaitrcehc,tlyadmdeeadsusreeawdattehre eggbatchesdid notsignificantlyaffectdevelopmental rate [O2] 1 h later at 30C ("Sediment Analysis"). (5) Redox (0 = 0.25. T= 1.9, P= 0.06). potential (Eh) was also measured inside the bags using a Totestthehypothesisthateggdevelopmentwasaffected Fisher955 pH/mV meterwith anaked platinum electrode by environmental variables that changed with beach ele- (Corning 476080 Redox Combination Electrode) because (vaDtEi)onn,ewstescaonlldecrteebdurfireedshltyhelaciadgeegdgsegfgrobmat1c3he(sFLa)ta8n(dF2L3) t[hEha]n i[sOa2]mmoeraesurreelimaebnletsin(dMiccaLtaocrhloafno,xy1g9e78n).ava[iElha]biallistoy and 7 (DE) different predetermined beach distances for indicates ion imbalances in sediments due to microbial s1t2akdeasysso(t1h9e9y1w).oIunlDdEnowtebeanecxhpoorseeddtbhyeoctahgeersnweistthinfglcargagbisn.g Sacetdiivimteynt(ZodBaetlal,we1r9e46t;akBeanngbeercaaunsde Ntiheemgirsation,s1i9ze78d)e.te(r6-) To quantify the average development ofeach egg batch mines the drainage and interstitial oxygen content of aTfhtiesr i1n2dedxaysp,rowveiduesdedmoare"DreevseolloutpimoennttahlanIn"pdeerxc"en(Dt.Id.e)-. abenadcthheesm(eEaagnleg,ra1i9n83si)z.eS(eMdzi)maenndtsthweevraeriactoilolne,ctoerdsoinrtiFnLg bcvoyenldtoihpteeidot"nism.ueseSrdaeqmuipnilree1sd99tf0orobrmeeacecaahucseheaceithggwsetbiaaggthecthusnwddeeerrveesltdoaipnvmdieadnretdd wci1noh9e8etf7rhf)eie.cri<tTae>nnh=tgee(-ar-,le02os)gu,lt2wtoSsa5aswn</em>d)re.eaSTshciuesormteuhpndeiatbrgsyreaadsirineewvoisiitnnzhgethd(ieaentspcmh1ri0mip(i<t/n(>i)tBoesonrcgsvagalsole,,sf into four categories based on developmental stages: eggs sediments from DE (Maurmeyer, 1978). (stages 1-17). embryos(stages 18-19), lateembryos(stages Because Linuilus embryos can postpone development 20-21). and larvae (Sekiguchi et al.. 1982). Aftercounting underanaerobicconditions(Palumbi andJohnson, 1982). thenumberofindividualsineach category,we multiplied wedetermined whetherundevelopedeggswerestillviable this numberby the numberofdaysrequired to reach that by transferring them to aerobic conditions in the labo- s1t9a8g1e).inWthee lcaablocrualtaotreyd (asntagoeveXradlalysD).I(.BrfoorwneaacnhdbCaltacphpebry, (ra1t0o0ry.egEggsgspefrrodmiseha)chwietghgbfaitlctehrewderseearewaarteedri(nppeertiroiddiicsahlelys taking the sum ofthe stage X days for all developmental changed) in a FormaScientific incubator (under a 14 h stages divided by the total number of individuals. For daylight schedule at 30C day, 26C nighttime temper- example, the D.I. for one batch that had 5 eggs, 14 early ature; Brown and Clapper, 1981). In DE we noted the e=mb[r5y(o0s).+19184l(a1te5e)m+br1y9o8s,(2a1n)d+ 0l(a2rv6a)e]/is21I7[s=tag2e0.Xwdhauyrsh] nviuomubsleyrdoefadin(ddievciadyuaeldsawnidthcirnumebglgyb)aatfctheers1t2hadtaywseroen otbh-e means that the batch contains a large proportion oflate beach and compared the relative mortality ofeggs at dif- embryos. We used stepwise multiple regressions to deter- ferent beach distances. mine the environmental factors that best predicted de- velopmental rate. Egg loss due to erosion We measured six environmental variables near thede- To test the hypothesis that beach elevation affects ero- veloping egg batches. (1) Temperature was measured in sion ofeggs from the sand, we measured the net amount 376 D. PENN AND H. J. BROCKMANN ofsand lost (erosion) or deposited at different beach dis- FLORIDA Nestinglocation tances (1991). We hammered forty 1-m wooden stakes I 1 1- intothesand(thesamestakeswe used formeasuringtidal distance). After the tide receded, we recorded the change in beach level ("erosion/deposition") that occurred since the previous tide. In FL we recorded the change in sand DELAWARE = erosion/deposition at four beach distances (/; 7) using Nestinglocation the mean from 1 1 high tides. In DE, we recorded the i- mean erosion/deposition at five beach distances (n = 5) from five tides. We looked for evidence of egg loss by comparing the change in volume between matched pairs Beachdistance(m) omfenctageexdpearnidmeunntca(g1e99d0)e.ggRebdatuccheedsvforloummethieneugngcdaegveedloepg-g shoFeigcurraebs1n.estAicnoFmlpoarirdiasoanndofDethleawlaorcea.ti"oNneosntitnhgelboecaacthieosn"wihsetrheehboerasceh- batchesrelative tocaged oneswould provideevidence for distance (mean SD and range) where nesting occurs. MHTL is the egg loss due to erosion. We also used egg-sized, colored meanhightidelineandEHTListheextremehightideline.Themoisture glassbeadsto minimize theeffectsofpredation (FL 1991). gradientsonthebeachesareshownasthesaturatedzone(),moistzone We made 30 "bead batches" (3-g batches), and buried (D).anddryingzone(D). them at three different beach distances, 8 cm below the surface ( = 10). To monitor the erosion during the ex- tidal distance (FL: r2 = 0.25, FU4: = 433, P = 0.0001). periment, weburied each batch nearanerosion stake(us- The crabs nested in or just below the swash zone and ing the mean erosion/deposition for analyses). After 17 neither the distance from the water's edge nor the water days, thedepth ofthebead batches(thedistance fromthe depth (/-test, Tlig = 1.1, P= 0.29)ofnestingcrabsdiffered surface)wasremeasured, and, afterallowingthem todry, significantly between FL and DE populations (Table I). each batch was reweighed to estimate bead loss. The DEand FLpopulationsexhibiteddifferent patterns oflunarand tidal synchrony(Table I). (1) In FLthecrabs DE Data analyses spawned onlyduringspringtides. In thelargestpopu- lations emerged during spring tides, but they continued Before performingstatistical tests, weexamined there- tospawn during neaptides. (2)Spawningcrabswere more sidual plots to validate the assumptions ofthe tests and abundant during the higher ofthe two daily tides in DE used arcsine transformations on percentage data (to nor- and FL. We noted one exception in DE when tidal in- malize variances). We used higher-level regression models equality was reversing; i.e.. during lunar quadrature, the (such as quadratic versus linear) only ifthere was a sig- crabs were more abundant on the lower tide of the day nificant increase in the variation explained by adding an in DE (see also Barlow el ai. 1986). In FL the highest additionaltermtothemodel( = 0.05). Weused multiple tides were usually during the day, whereas in DE they regressions to evaluate the significance ofvarious factors were usually at night (1990; Table I). (3) In FL the largest and stepwise regression procedures to select the model numbers of horseshoe crabs nested around the time of when colinearity occurred. We used "Statview" (Abacus the MHT (see also Cohen and Brockmann, 1983), but in concepts, 1989a)and "SuperANOVA" (AbacusConcepts, DE they nested after MHT, i.e., during the receding tide 1989b) statistical software. Data are reported as mean (Fig. 2). standard deviation, unless otherwise stated. The FL and DE beaches differed in a number of im- portant respects (Table I). First, the extent oftidal inun- Results dation was greater in DE than in FL. Second, although both beachescan beclassifiedaslowenergy, theFLbeach Differences in nest-site location, tidalsynchrony, and was ofa much finer grain size than the DE beach. Third, beach characteristics associated with these two variables was the fact that the Horseshoe crabs in DE nested higher and overa wider saturated zone and the drying zones were higher on the tdr6iha1sna%tgnaenoiocnfefFtbahLenead(cbThteahabteclhheva,anIr)wi.thahIenonrcseeeDawiEsnerF9ien5Ls%F(iFLgoinfgi9.ft5ih1c%e)a;ncttorhlfaeybtgmshreeenaaehtsnoetrrensdeiesnsothDvioenEerg btoshfeeeadtichbmheeeaFincnLthsFbdLoeifastcDthhaEan.hnceeFlisndouawDrtaEttwh;eh,rii.Fceb.Lhe,ttttaehhnreedrtfehDianwEenarstbghereaaavcciahonieaelrsdasbdeslireefdfgoierxrmayeeidgnneteindsn crabs nested over 40% of the beach. The mean nesting (Table I). distances were not significantly different from the mean Egg development high tide line (MHTL) in FL or DE (/-test, P > 0.1) and FL and DE differed in the pattern ofegg development the nesting distances increased slightly with increasing with beachdistance. In FL,eggbatchesplacedatthehigh- NEST-SITE SELECTION IN LIMULUS 377 Table I Nesl-sileselectionbyhorseshoecrabs, thedevelopment oftheireggs, andcharacteristicsofthebeachat variouselevationsattwosites. Florida andDelaware Florida Delaware NestingBehavior Location ojNestingCrabs Mean nestingdistance 5.1 0.9 m (n = 434) 7.4 1.9 m(n = 1396) Rangeofnestingdistance 3 to9 m (n = 433) 1 to9m(n = 1395) Mean distance from water'sedge 68 68cm( = 37) 85 70cm (n = 145) Mean waterdepth 8 9cm (n = 37) 7 8 cm (n = 144) Nestdepth 11.5 2.8cm(n = 382) 9.3 3.9cm (n =112) SfawningSynchrony Lunar springtides springand neaptides Tidal bothdailytides(n =13) highertide(n = 13)which usuallyoccursat night Timingofmaximum crabdensity around high tide afterhigh tide EggDevelopment(Percent Developed, 1991) Beach Distance 11-12 m 9-10 m 7-8 m 5-6 m 3-4 m m 0-2 378 D. PENN AND H. J. BROCK.MANN (i.e.. < 5 cm) (Fig. 5). Egg batch volume (r2/F,) did not differbetween caged and uncaged treatments(F, 74 = 2.7, P= 0.12). Discussion Nest-site selection andegg development Horseshoe crabs nested at beach elevations where egg developmentwas maximized. They nestedjust abovethe mean high tide line, avoiding the lower and upper ele- vationswhereeggdevelopmentwasimpaired. Ourresults show that variation in egg development with elevation is Tidetime(hrs) due tochemical and thermal gradients on the beach. The Figure 2. Thetimeofmaximum horseshoecrabspawningcompared sedimentsin the upperbeach werewarmeranddrierthan withthetimeofmaximum hightideat(a)SeahorseKey.Florida(1989) the lower beach, which explains why eggs buried above and (b) Cape Henlopen, Delaware (1991). The bars indicate the mean the mean tide linewere more likely todesiccate than eggs number ofnestingcouplesoverthree daysand the points indicate the buried lower on the beach. The sediments in the lower mean tidal distance recorded during three Hood tides. "Tide time" in- dicatesthehoursbefore(-)andafter(+)themaximumhightide(adapted beach contained insufficient interstitial oxygen concen- from Burger el ill., 1977). Tidal synchrony was significantly different trations foreggdevelopment tooccur. The colorand odor between thetwopopulations(X: = 945.df= 4. P< 0.001). of the lower beach indicated the presence of hydrogen sulfide and explains why the eggs became black and crumbly. The inability ofeggs to survive in anoxic sedi- intercorrelated with each other (Table III). A stepwise mentssupportsthesuggestion that nestinghorseshoecrabs breesgtrepsrseidoinctaenadltyhsiesdefvoeulnodptmheantta[lO2i]nwdeaxsitnheFLvar(irab=le0.t5h3a,t aevmobirdyobneiacchdeesvweiltohpmaneonxtic(Bpoetattonbeedtsaib,ec1a9u8s8e).thHeoyrsiemsphaoier F,,22 = 25.2, P = 0.0001), whereas the best predictor of crabs may avoid anaerobicconditions(anoxic sediments, D.I. in DEwas%H:O(r2 = 0.67, F2.2, = 23.6. P= 0.0001). peat beds, and sewer outflows; Rudloe, 1971) by using incTrheaese[dOs:i]ginniftichaentsleydwiimtehntbesaacmhpdliessta(nsceed(irm2e=nt0.a4n5a,lyFsii,s2)2 oPxayggee,n-1s9e7n4s)itaivnedeplremoesnotmsain(Tthheoirmpbsoookngailnlsd(CPraagbet,re1e97a5n)d. = 18, P = 0.0001) in FL. Sand color below the surface It has also been suggested that horseshoe crabs have H2S w=as143l,igPhte=r0a.t00h0i1gh)earndbewaacshbldaicstkaonrcegsre(yAfNrOomVA,toF3,m,5.K recHeoprtosressh(oBeotctroanbsetitnil.D.e1l9a8w8a).re nest over a wider range indicating anaerobic conditions (Eagle, 1983). Develop- ofthe beach than in Florida, probably because the range mental rate (D.I.) ofeggs in the incubator increased sig- ofelevations conducive to development is wider in Del- nificantly with beach distance (r2 = 0.33, F\ j4 = 16.7, P aware. The differences in egg development between the = 0.0003) and eggs that had been buried in the lower two sites can be explained by differences in oxygen and beach (below 3 m) never hatched. Unlike FL. the sedi- moisture gradients on the beaches. In Florida the inter- mentsbeneath thesurface in DE were rarelygreyorblack, stitial oxygenconcentration increasedslowlywithdistance and most eggs in the lower beach showed some signs of (not reaching [O2] > 1 ppm and +Eh until 3m), whereas development. oxygen concentration increased sharply with beach dis- tance in Delaware ([O2] > 1 ppm and Eh > +80 above m Egg /o.v.v din' to erosion 1 section of the beach) (Figs. 4a, e). Also the redox potentials in the lower beach in Delaware (+140 to In FLsand iseroded from the lowerpartofthebeach -30 mV) were never as low as in Florida (-50 to and deposited ai higherbeach distances(Fig. 5). However, -250 mV) (Figs. 4b, f)- This means that the Delaware in the artificial experiment, bead loss occurred only crabs can nest lower on the beach without adversely af- at the highest beach distance (ANCOVA, F,.28 = 7.92. P fecting the development oftheir eggs. = 0.009). Nosignificant associationswere found between The differences in interstitial oxygen (and redox gra- bead loss (P = 0.75) or final bead depth (P = 0.70) and dients) and moisture gradients between the Florida and sand erosion. In the egg-loss experiment, volume (Fi/I',) Delaware sites can ultimately be attributed to differences did not differ between caged and uncaged treatments in beach morphology (Table I). Sediment grain size de- (Wilcoxon Signed-Ranks, r = 0.77. n= 13, P = 0.22). In terminesthedrainage ofabeach, whichgreatlyaffectsthe DE sand was eroded only at the highest beach distance interstitial oxygencontent(Gordon, 1960; Brafield, 1964; (/-: = 0.23, F,,22 = 6.9. P = 0.02) and this was negligible Eagle, 1983). The Floridasedimentswere fine to medium NEST-SITE SELECTION IN L1MULUS 379 FLORIDA (a) Egg Development (b) Egg-Batch Volume 30 - CD 20 - I -11357 -11357 10 Beachdistance(m) Beachdistance(m) DELAWARE (c) Egg Development (d) Egg-Batch Volume 60- 50- 40 - 30- 20- 10 024 24 6 8 10 12 -2 6 8 10 12 Beachdistance(m) Beachdistance(m) Figure3. Development ofLimuluseggs,embryosand larvae(mean SE)atdifferentbeach distances inFlorida(aandb)andDelaware(candd). Development, asmeasuredbythedevelopmentalindex, varied Fwi2t4h2b=ea1c6h,dnis=tan1,cePin=(a0).0F0l0o1r)i.daE(grg=ba0t.c3h5,vFolt.u2lme=d1u1e.3,tond=ev3e,lPop=me0n.t0a0l3)sawneldliinng(ca)lsDoelvaarwiaerdew(irth=b0e.a4c4h. distance in (b) Florida (r2 = 0.12, F,j, = 2.9, P = 0.10) and (d) Delaware (r = 0.32. F2.a = 9.7. P = 0.0003). grained and had poordrainage(e.g., the moisturecontent beach. Increased risk ofdesiccation on higherpartsofthe decreased gradually, not dropping to 5% until 7 m), beach mayexplain whycrabsin Massachusetts shift their whereasthe Delaware sedimentswerecoarsegrainedwith nesting sites to lowerbeach distances during the summer relatively high drainage (e.g., the interstitial waterdropped (Barlow et ai, 1986). sharply, falling below 5% at 4 m) (Figs. 4d, and h). Thus, eventhough thetidalamplitudesweregreaterin Delaware, Nesting synchrony the watercontent ofthe beach was lowerthan in Florida. Fine-grained sediments, such as in Florida, also have Nestingsynchronyvariedsignificantly between thetwo greater surface areas for microbial growth. This further populations. In the Delaware Bay, horseshoe crabs often depletes interstitial oxygen, increases hydrogen sulfide, spawn during neap tides, as reported for Massachusetts and lowers redox levels (Eagle, 1983; Boaden, 1985). (Cavanaugh, 1975; Barlow et a/., 1986; pers. obs.). How- Variation in beach geochemistry may also explain why ever, crabs in Florida almost never spawn during neap thecrabsin Floridanested uptotheEHTLbutwellbelow tides (Rudloe, 1980; Cohen and Brockmann, 1983; pers. this mark in Delaware (Fig. 1). In Delaware the eggs in obs.). Our results from the egg-development experiments the highest partofthe beach were usually desiccated. Wa- suggest an adaptive explanation fordifferences in spawn- terretention wasso poor in Delaware that water concen- ing synchrony between these populations. The aerobic tration was the best predictor ofegg development on the sediments occur at higher elevations in Florida than in 380 D PENN AND H. J. BROCKMANN Table II Thedevelopment ofLimuluseggsplacedaldifferent locationsonthe beach NEST-SITE SELECTION IN LIMULUS 381 FLORIDA (a) Oxygenconcentration (b) Redoxpotential 200- 3.0- 100- 1 o- 2.0- W_c -100- 1.0- -200- 1357 0.0 -300 -1 1 3 5 -1 Beachdistance(m) Beachdistance(m) (c)Temperature 25 (d) Moisturecontent 30- 20- 29- 28- 15 . 10. 27- 26- 5 . 5 25 -1 1 3 5 1 3 5 Beachdistance(m) Beachdistance(m) DELAWARE (e) Oxygen concentration (f) Redoxpotential 4.0- 3.0- Q. ^. 2.0- d 1.0- 0.0 024 246 6 8 10 12 -2 o 10 12 Beachdistance(m) Beachdistance(m) (g)Temperature 20- (h) Moisturecontent 28-1 15 27- 10- 2468 -2 10 12 -2 2 4 6 8 1012 Beachdistance(m) Beachdistance(m) Figure 4. The conditions ofthe sediments at different beach distances (1991; mean SE, ;; - 3) in Florida: (a) interstitial oxygen concentration (r2 = 0.78: Fl:22 = 78; P = 0.0001), (b) redox potential (r: (F==rI2U01,.=072=9;0;.3P8F50:u;=P,F0==1.20200700.=;10)0P,0817;)(=,0P0ar.ne0=dd00o0(1x.h)0).p0o0m(t1oce))in;stttieauamnrlpdee(rrcDa2oetnl=utarew0ena.tr3(er6(:;r22=(F=e2)0.3.0i,6.n48t=;3er;sF\t,Fli.;t2,i.22aP7l===o3x690y7;.g;0eP0Pn0=2=c)o,00n..c0(0eg0)0n00t1t1)re),am.tp(iedo)rnamt(orui2rset=u(rr0e2.8c=2o;n0tF.e62n1.t; 382 D. PENN AND H. J. BROCKMANN Table III Theeffectsoffourenvironmentalfactors in beachsedimentson thedevelopment <>/Limuluseggs(D.I.)attheFL andDEsites(1991) Florida Delaware Parameter Significancetest Significance test [0,] Note: A polynomial regression was usedon the Delawaredataand nonsignificanteffectswere not included.