1993. TheJournal ofArachnology 21:107-119 CONSTRAINTS AND PLASTICITY IN THE DEVELOPMENT OF JUVENILE NEPHILA CLAVIPES IN MEXICO Linden Higgins': Centro de Ecologia, Universidad Nacional Autonoma de Mexico, Apartado Postal 70-275, Ciudad Universitaria, C. P. 01450, Mexico ABSTRACT: Thelarge,orb-weavingspiderNephilaclavipesisfoundinadiversityofhabitatswithinanarrow latitudinal range in Mexico. This allowed nearly simultaneous study ofpost-embryonic development ofsix disjunctpopulationsindissimilarenvironments.Acommon-gardenlaboratorystudyutilizingjuvenilescollected infoursitesreinforcedtheconclusionsfromthefield. Thedevelopmentalparametersinfluencinggrowthinsize at ecdysis did not vary within oramong populationsand may be genetically determined. Although very small juveniles exhibit variation in the growth per ecdysis, largerjuveniles exhibit very little variation. These data, compared to data from field and laboratory studies ofother tropical populations ofN. clavipes, indicate that growth per ecdysis is highly constrained. Thus, this developmental parameter establishes a developmental trajectory that may be geneticallydetermined and therefore subject to natural selection. RESUMEN: La arana Nephila clavipes, tejedora de telas orbiculares, se encueritra en habitats diversos en Mexicodentrodeunlimiteangustodelatitud. Esopermitioestudioscasisimultaneosdeontogeniadejuveniles enseispoblacionesdesunidosenambientesdistintos.Unestudiodellaboratorioutilizandojuvenilescolectados encuatrositiosfortalecelasconclusionesdelcampo.Losparametrosontogeneticosdeterminandoelcrecimiento por muda no vario dentro ni entre poblaciones, y posiblemente son geneticamente determinados. Aunque juvenilesmuypequenos mostraron variacion en elcrecimientopormuda,juvenilesmasgrandesno mostraron variacionenesteparametro.Estosdatos,combinadoscondatosdeestudiosdeotraspoblacionesenellaboratorio y en el campo, indican que el crecimiento por muda esta muy coiistrenido. Asi, este parametro establece una trayectoriaontogenetica que posiblemente sea geneticamente determinado y sujeto a seleccion natural. Determininghowenvironmentalfactorsinflu- weight gain and the number ofjuvenile molts ence life history requires determination of the varied with habitat within and among popula- developmental parameters at each life-history tions, thegrowthperecdysisdidnotvarywithin stageandexaminationofwhethertheparameters a population or between two populations in the are phenotypically plastic, respondingto the en- tropics (Higgins 1992a). The constraints im- vironment, or are genetically determined (Ca- posedupondevelopmentbytheconstantgrowth swell 1983, Via & Lande 1985, Pease & Bull per ecdysis were countered by phenotypic plas- 1988). The development ofthe large orb-weav- ticity in other developmental parameters, par- ingspiderNephila clavipes(Linnaeus) (Araneae: ticularly the number ofinstars, generating vari- Tetragnathidae) can be expressed as a group of ationinsizeandageatmaturity(Higgins 1992a; interdependentparameterswith varyingdegrees Higgins pers. obs.). ofphenotypicplasticity (Higgins 1992a). Exper- The previous studies utilized a range ofsites imental trials showed that weight gain and in- withwidelydifferingphysicalandbiologicalcon- termolt interval duration responded to shifts in ditionsincludingphotoperiod,makingitdifficult food availability, butgrowth perecdysisdid not to distinguish the relative influence ofdifferent (Higgins pers. obs.). The duration ofthe inter- ecological factors on post-embryonic develop- moltintervalwasfoundtobecorrelatedwiththe ment. Inordertobetterunderstandtheenviron- size of the spider and apparently reflected the mentalinfluenceon development, a secondfield lengthoftimetheindividualrequiredtoachieve Study was undertaken, utilizing the diversity of the minimum weight necessary to molt to the habitatsinwhichN. clavipesisfoundinMexico. next instar. In field studies, whereas the rate of Choosing sites within a U latitude range elimi- nated variation in photoperiod among sites and 'Currentaddress:Dept,ofZoology,UniversityofTex- allowed nearly simultaneous study of popula- as atAustin, Austin, Texas 78712 USA tions experiencing very different environments. 107 108 THEJOURNALOFARACHNOLOGY In orderto more fully explore the plasticity and terswith minimumtemperatures aslowas 0°C. constraints ofdevelopment, juveniles from the On the western, dry side of the Sierra Madre fourmostdistincthabitatswerebroughtintothe Oriental inthe valleyofTehuacan, I studiedthe laboratory in a common-garden experiment. spidersattheSecretariadeEcologiayDesarrollo Common-gardenexperiments, whereindivid- Urbano cactus garden near Zapotitlan Salinas, uals from different environments are held in a Puebla. Annual rainfall in Tehuacan is very low commonenvironment, allowpreliminarydiffer- and there is a relatively cold winter. Climato- entiation between those parameters that are ap- logical data are not available for Arroyo Frio, parently genetically determined and those that located near Perdenales in southwestern Mi- are phenotypically plastic (Wise 1987). Param- choacan. The area is seasonally dry and, due to eters that are genetically determined and vary the altitude, seasonally cool. However, there is amongpopulationswillexpresssimilarvariation a permanent stream through the site that main- in the laboratory among individuals from those tains high relative humidity within the arroyo populations. In contrast, parameters that are wherethespidersarefound. OnthePacificcoast phenotypically plastic and vary among popula- ofJalisco, the spiders were studied at Chamela tions will not vary in the laboratory among in- field station, owned and run by the Institute de dividuals from different populations. The com- Biologia, Universidad Nacional Autonoma de binationoflaboratoryandfieldobservationsthus Mexico (UNAM). This site is seasonallydrybut allows distinction between constrained, poten- nevercool. Spiders were found during the rainy tially genetically determined, parameters and season atall sites, but are facultativelybivoltine phenotypically plastic parameters. atboth coastal Veracruz sites (Higgins in press). METHODS There was no secondgeneration at these sites in 1989 and 1990, so data are presented from only Populations studied.—Spiders were observed the rainy season. in six sites spanning Mexico from the Veracruz Rates ofprey capture were estimated through coast to the Jalisco coast. Three sites were in trap-line surveys (Turnbull 1960). The spiders Veracruz: PlayaEscondida,Nanciyaga,andFor- in PlayaEscondidacapturedfewerpreythanthe tindelasFlores. Onesitewasinthehighaltitude spidersatothersites(Table 1)(Higginspers.obs.). desertvalleyofTehuacan, Puebla. Twositeswere Predation load, estimated as the proportion of west of the central plateau: Arroyo Frio, Mi- juveniles less than 0.5 cm leg I tibia + patella choacan, and Chamela, Jalisco. In 1989 I trav- lengththatabandoned intactorbwebs following eled to Playa Escondida, Nanciyaga, Fortin de predatorattack(asinHiggins 1992b),washigher las Flores, Arroyo Frio, and Chamela. In 1990, inTehuacan,ArroyoFrio, andChamela,butthe I studiedspidersinPlayaEscondida,Nanciyaga, differences were not significant (Higgins pers. Fortin de las Flores, Tehuacan, and Chamela. obs.). Because these sites varied only 1®latitude, there Itraveledalternatelyeastandwestfrom Mex- was no significant difference in photoperiod ico City, visitinginland sites both on my wayto amongthem. However, the sites variedin many and returning from the coastal sites. The visits environmental parameters including type and atthecoastalsiteswereslightlylongerthanvisits degreeofseasonality,andpreycapturerates(Ta- to the inland sites. The combination oflonger ble 1). visits at the end points and repeated visits at Playa Escondida and Nanciyaga are privately intermediate sites enhanced the probability of owned forest preserves about 13,5 km apart on observing molts by marked individuals. the Veracruz coast, separated by cattle ranches. Field ol)ser¥ations*~Field observations ofin- Bothhavewetclimateswithrelativelycoolwin- dividuals utilized the methodology previously ters.PlayaEscondidaisapproximately 1 kmfrom describedindetail (Higgins 1992a). Ateachsite, apreviousstudy site, thebiological station“Los censusesofspiderabundanceandsizeweremade Tuxtlas”(Higgins 1992a,b).TheN.davipespop- andwebsiteswereflagged. Measurements,made ulationatthelattersitedisappearedshortlyafter withHeliosneedle-tippedcalipers, includedspi- dispersalofjuvenilesearlyin 1989,andthestudy der leg I tibia + patella length (TPF, cm), ab- wascontinuedatPlayaEscondida.ThethirdVe- domen length, and abdomen width. Individuals racruzsite, FortindelasFlores, isamid-altitude largerthan0.5cmTPLweremarkedwithenamel areaofcoffeeplantations(Benton&Uetz 1986). paint on their legs. From abdomen length and This site is coolerand experiences strongerwin- width, the abdominal volume was estimated as HIGGINS--DEVELOPMENTOFJUVENILE NEPHILA CLAVIPES 109 Table L—Location,climateandrelativepreycaptureratesateachsite.Annualrainfallandmeantemperature weretaken fromthe nearestweatherstationreportedbyGarcia(1973) forall sitesexceptChamela. Datafrom Chamelacomefrom Bullock(1986 andpers. commun.). Temperaturesare forthegrowingseason. Noweather data are available forArroyo Frio. (§Mean prey capture per 12 diurnal hours per spider (Higgins, pers. obs.), determinedas in Higgins & Buskirk 1992.) Annual Mean Altitude rainfall temp, Prey Site Coordinates (m) (m) °C capture§ Playa Escondida 18°35'N, 95°W 0 4.5 26 low Nanciyaga 18°35'N, 95°W 100 4.5 26 high Fortin de las Flores 18°50'N, 97°W 1000 2.5 22 high Tehuacan 18°20'N, 97°30'W 1200 0.3 20 high Arroyo Frio 19°10'N, 101°30'W 1200 — Chamela 19°30'N, 105°W 50 0.7 high a cylinder. For a given TPL, abdomen volume lectedforlaboratorystudywereNanciyaga, For- is highly correlated with spider weight (Higgins tin de las Flores, Tehuacan and Chamela. The 1992a). Each individual found on a non-viscid spiders were maintained on three dimensional silk platform was assumed to be pre- or post- frames made of two intersecting 30 cm circles ecdysis(Higgins 1990). Thesizeoftheabdomen made offiberglass strips, and were free to move relative to the legs distinguished between these about the laboratory. From each site, spiders of conditions: pre-ecdysis individuals have large, 0.2-0.4 cm TPL were collected. Twenty-three distended abdomens whereas post-ecdysis indi- spiders were used in the experiment, as follows: vidualshavemuchsmallerabdomensrelativeto Nanciyaga(3females, 1 male),Fortin(3 females, leg length and carapace width. Collection ofex- 3 males), Tehuacan (2 females, 4 males), Cha- uviae provided additional data for the analysis mela (2 females, 5 males). The ratio ofjuvenile ofgrowth perecdysis. Post-ecdysisspiders often males to femalesdependedupon the exactdates hang the exuvium in the barrier webs near the ofcollection. Later in the season, small spiders hub connection, and TPL ofan exuvium is not are more likely to be males (pers. obs.). In ad- significantly different from the spider size in the dition, three spiders escaped prior to being previous instar (Higgins 1992a). When several marked during a trip to Fortin and Tehuacan. exuviae were present, I only measured the larg- These animals were included in the description est, from the most recent molt. ofweight gain during the intermolt interval. These data are used to compare the relation- The spiders were maintained 3-4 days with ships among abdomen volume, premolt TPL, onlywatertoincreasetheprobabilityofspinning and postmolt TPL within and among the pop- when released, and were offered food immedi- ulations over the entire life-cycle ofthe spiders. ately after a web was spun. Any spider that did Theseparametersalldescribethegrowthpermolt, not spin an orb within three days ofrelease in and as such are not strictly independent. How- the laboratory was not included in the experi- ever,becausetherewasachancethatspidersgain ment. Throughout the experiment, each spider weightbeyondthatrequiredtosuccessfullycom- wasofferedthreeDrosophilavirenseachday. For plete ecdysis (particularly relevant for penulti- a period ofone week, onlyD. melanogasterwas mate-instarindividuals), premoltabdomen vol- available; two ofthese were substituted foreach ume was compared as well as premolt and ofthe largerD. virens(fora total ofsix flies). At postmolt TPL. To describe growth patterns of theinitiationoftheexperiment,thespiderswere thespeciesinNorthAmerica,datafromprevious measured (TPL, abdomen length, abdomen studiesin Texas, USA, “LosTuxtlas”, Veracruz width), and these measures were repeated with Mexico,andPanama(Higgins 1992a)werecom- each molting. Abdomen volume was measured pared to those presented here. every other day during the intermolt interval to Common garden experiment.—In 1990,juve- monitorweightgain. Themajorityofthespiders nilespidersfrom twowetandtwodrysiteswere were held for two molts; one individual from broughtintothelaboratory. Thepopulationsse- Fortin failed to molt a second time in the lab- 110 THEJOURNALOFARACHNOLOGY o Playa Escondida Nanciyaga A Fortin 0.5 Tehuacan Arroyo Frio • Chamela 0.0 0.2 0.4 0.6 0.8 1.0 premolt abdomen volume, cc Figure 1.—Postmolt size (leg I tibia + patella length, TPL) as a function ofpremolt abdomen volume at all sites. Onlydatafromjuvenilesandfemalesmoltingto maturityareincludedinthegraph. Thecurverepresents a fit to the entire data set,y = 1.66 x (R2 = 0.99). oratory. Observations also distinguished those RESULTS days when the spiders were foraging and had partially or wholly renewed orbs from the days Field observations. volume and immediatelypre- orpost-molt, whenthe spiders molting: The relationships among premolt ab- were not actively foraging. Data presented here domenvolume, premoltTPLandpostmoltTPL concernthe growth perecdysis, length ofthe in- vary little within or among the six populations termolt,daysspentforagingduringtheintermolt studied. PostmoltTPLisrelatedtotheabdomen and pattern ofabdomen volumegain duringthe volume by a concave function, approximately a intermolt period. function of the cube root of abdomen volume Statistical analyses*—The developmental pa- (Fig. 1). The data are insufficient to allow com- rameters examined in this study are dependent parison among the sites because recording pre- uponthesize, TPL, oftheindividual. Therefore, moltabdomenvolumeandpostmoltTPLforthe allanalysestestedfora significantregressionbe- same individual was unlikely in the field. tween TPLand the measurementin question. If Thepremoltabdomenvolumeisafunctionof the regression analysis was significant for each premoltTPL, and males and females moltingto population, furtheranalysistestedforsignificant sexual maturity do so at a lower abdomen vol- variationintheslopeoftheregressionlinesamong ume compared tojuveniles molting tojuvenile populations.Iftherewasasignificantinteraction instars. ThedatafromFortinincludedthegreat- ofpopulation and TPL, indicating difference in est number of observations of molts to sexual slope,thentheanalysiswashalted(Sokal&Rohlf maturity (Fig. 2). Penultimate instar males and 1981). Ifthe interaction terms were not signifi- females were identifiedas follows: almost all fe- cant, a final analysis ofcovariance (ANCOVA) malesofTPLgreaterthan 1.0 cm aremoltingto with TPLascovariatetested forvariation in the sexualmaturity(pers.obs.)andpenultimatemales altitude of the line (y intercept). Lastly, if no have swollen palpi. After correcting for heter- difference was found due to population ortreat- oscedasticitybytakingthesquare-rootofthede- mentinabiologicallyimportantvariable, apos- pendent variable (cube root of abdomen vol- teriori power tests were calculated to determine ume),ANCOVAofthedatafromFortinrevealed the minimum percent difference in the slope or thatthedifferencesamongjuveniles,penultimate interceptthatcouldhavebeendetectedwiththese instarmales,andpenultimateinstarfemaleswere data. significant(TPL:F(,_52)= 1475.6,P < 0.001;sex/ HIGGINS-==DEVELOPMENTOFJUVENILE NEPHILA CLAVIPES 111 Figure 2.—The cube root ofpremolt abdomen volume as a function ofpremolt TPL (leg I tibia + patella length) for spiders in Fortin. These data include males molting to sexual maturity (= A) and juveniles and females(= o). The points above premolt TPL = 1.0 cm are females moltingto sexual maturity. a=ge:13F.(725,,52)P= <8.805.,00P1)<. 0S.e0p0a1r;atientceroamcptaironi:soF(n2,o52f) coGmrpoawrtehdpweirtheicndyasnisd:aGmroonwgthpoppeurlaetcidoynssisuswiansg males and females molting to sexual maturity regression analysis ofpostmolt TPL on premolt indicated that the function ofpremolt abdomen TPL. The slope ofthe regression line is an in- volume on premolt TPL have the same slope dicationoftherateofsize-specificgrowth. Molt- d(Fi(fi,fe1r5)en=ce2i.n28s,lopnes,).power test: 4.1% detectable riantgetoofsegxruoawltmhaattureictdyyswiassbpercaeussuemepdosttomaoflftectTtPhLe Datafromtheremainingpopulationsincluded iscorrelatedwithpremoltabdomenvolume,and fewobservationsofspidersmoltingto maturity, whether the spiders were molting to maturity so the comparison of the function of premolt influencedthepremoltabdomenvolume.There- abdomen volume on premolt TPL among sites fore,theobservationsofmoltingjuveniles,males utilized only data from juvenile molts. Arroyo moltingto maturity, and molting females larger Frio data were excluded from the final analysis than LO cm TPL were considered separately. because few spiders were observed in the size Within sites, preliminary regression analyses range of0.5 cm-1.0 cm TPL. ANCOVA ofpre- showedthatpremoltandpostmoltTPLweresig- molt abdomen volume (to the 0.15 power) with nificantlycorrelatedforjuvenilemoltsatallsites TPLascovariate revealedthattherewasno dif- (all P < 0.003) (Fig. 3). Individual ANCOVA f1e.r7e5,ncnes,inposwleorpeteastm:o1n.g8%aldletfeivcetasbilteesdi(fFf(e4r^e1n07c)e)=.= jwuevreenirluenmtoolcthsecbketfworeedinffyeeraernsceastitnhegrsoitwetshsdtuurdiinedg Theinteractiontermwasdroppedfromthe final inbothyears: Nanciyaga, Playa Escondida, For- ANCOVA, and this test showed that spiders in tin de las Flores, and Chamela. No significant Chamela molted at a slightly but significantly differences were found (all P > 0.12); therefore, lower premolt abdomen volume for their size datafrom 1989 and 1990werecombinedforthe [ANCOVA. TPL: F^,, ,„) = 4283.4, P < 0.001; remaining tests. Due to small numbers of ob- msietle:a:F(y4,= 0=,307.6+3,0,P54<(0T.P0L0)1;;rreemgraeisnsiinognss:itCehs:a-y sfoerrvtahteioannsa,lytsheesdoaftagrforwotmhbdoutrhinygeamroslwtesrteopsoeoxlueadl = 0.32 + 0.55 (TPL)]. maturity. Preliminary regression analyses of 112 THEJOURNALOFARACHNOLOGY 2.0 1 Tehuacan PlayaEscondida ArroyoFrio 1.5 - O <9 1.0 GD 0.5- 0 1989 O Tehuacan A 1990 ArroyoFrio males 0.0 _| 2 0' 0.0 0.5 1.0 1.5 . Chamela Nanciyaga OO premoltTPL, cm A A o 1.5 AO 1.0 / /A 0.5 - O 1989 0 1989 A 1990 A 1990 males 0.0 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 premoltTPL, cm Figure 3.—Growth per ecdysis, determined as the relationship between premolt TPL (leg I tibia + patella length) and postmolt TPL for all populations observed. Arroyo Frio (1989) and Tehuacan (1990) are plotted together, all other plots contain data from two years. Males moltingto maturity are indicated by solid squares (). postmolt TPL on premolt TPL were significant tween age and size classes Ouvenile, male or fe- formalesfromPlayaEscondida,Nanciyaga, and male) were restricted to a few sites. Data from Fortin, and for females from Nanciyaga {P < Nanciyagaindicatedthattherewasnodifference 0.02). in growth per ecdysis betweenjuveniles and fe- Travelprecludedcollectingcompletedatasets males molting to maturity (no interaction effect for all populations, therefore comparison be- P < 0=.000.101;,mnast;uAriNtCy:OVF,A,.jgT)P=L:2.F0„1.,58)n=s).15I6n7.a1l,l Table2.~ANCOVAofgrowthpermoltofjuveniles three Veracruz sites, males were observed molt- and males molting to maturity in Playa Escondida, ing to maturity. ANCOVA showed significantly Nanciyaga, and Fortin delas Flores. (*P < 0.01, **P lower growth per ecdysis (slope of the line) in < 0.001) males molting to maturity than in molting ju- veniles less than 1.0 cm premolt TPL (Table 2). Factor df Fratio Comparisons made among sites for juvenile Premolt TPL 419.3** molts revealed no difference among sites in 1 Site 2 1.26 growth perecdysiseitherin slope orin intercept MSMiaattteuurrXiittpyyrexmoplrtemToPlLt TPL 211 327...298881* (tpseolcwotpeaebr:letFesd(t5if21f0e.8r)9e%=ncdee1;-t9ei8cn,ttearnbcsl,eeptpd:iofwfeerre2n1t3ce)es)=t(0T1.a-94b8%l,edn3es)-,. Site X maturity 2 0.26 However, these regression analyses obscure a Error 133 slight non-linearity of the data. Closer exami- HIGGINS-DEVELOPMENTOFJUVENILE NEPHILA CLAVIPES 113 1.0- A PlayaEscondida A Tehuacan 0 Nanciyaga • Arroyo Frio 0.8- ^•i A Fortin Chamela CL <M 0.6- CHL s 0.4- _ il 0 0.2- o- O.oJ 0.0 0.5 1.0 1.5 premolt TPL, cm Figure 4.—The growth statistic Cas a function ofpremoltTPL (leg I tibia + patella length). The value ofC equal to the average slope ofthe equations ofgrowth perecdysis (Table 3) is indicated by the dotted line. nationofvariationinthegrowthperecdysiswas studies ofpopulationsat Barro Colorado Island, permitted by calculation ofa growth statistic, C Panama, at Los Tuxtlas, Veracruz, Mexico, and =ln(TPL2/TPLl).Thisgrowthstatisticisrelated insoutheasternTexas,USA(Higgins 1992a). The to Huxley’s growth equation (1972 p. 6) as C = olderdata set did not distinguish males molting In(aG) when dt = 1 instar, but it is not the same tosexualmaturity.Therefore,theANCOVAwas ask. MeanCisafunctionoftheregressionequa- run with juveniles ofless than 1.0 cm premolt tion slope (In(slope) = mean C). This transfor- TPL and males. There was no difference among mation revealed that although C was indepen- any of the tropical populations (no interaction dent of premolt TPL for larger spiders, for effect: F(7 355) = 1.67, ns, power test: 2.2% de- individuals ofthe first and second instars (TPL tectable difference; ANCOVA. TPL: F(, 372) = < 0.3) Cwas strongly dependent upon premolt 19,370.4,P<0.001;site:F(7372)=0.99, ns,power TPL (Fig. 4). The data from all populations ap- test: 4.7%detectabledifference). Inclusionofob- pear to fall on the same curve. The high values servations from the University of Houston ofCforsmall spidersreflectthelargechanges in Coastal Center in Galveston County, Texas, re- TPLatecdysis. Spiderlingswith premoltTPLof sulted in a significantsiteeffect on slope, reflect- 0.05 cm often molt to TPL of0.11 cm, an in- ingthesignificantlylowerslopeofthegrowthper crease of 100% or C = 0.8.m. ecdysis for the Texas population (ANCOVA. The data concerningjuvenile growth per ec- TPL: F(, 425) = 5776.6, P < 0.001; site: F(8 425) = dysiscanbecomparedtodatacollectedinearlier 1.14, ns; site x TPL: F(8,425) = 4.91, P < 0.001). Table 3.—Regression equations forjuvenile growth per ecdysis, where initial TPL is less than 1.0 cm. (**P < 0001 . ) Regression Regression Site n intercept slope P2 F(regression) Playa Escondida 31 0.068 1.24 0.99 2671.1** Nanciyaga 55 0.051 1.29 0.97 1959.7** Fortin de las Rores 41 0.043 1.32 0.99 3150.0** Tehuacan 24 0.053 1.19 0.96 591.0** Arroyo Frio 17 0.078 1.25 0.99 1423.0** Chamela 52 0.047 1.29 0.99 4743.5** 114 THE JOURNALOFARACHNOLOGY 08-1 >• . 0.4n > • 0.7- 1 0.6H • o •m 0.3 - 4*0 0.5H • * E oA O o ® ° ° CL 0.4- 0.3- 0.2 0.1 0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6 premoltTPL, cm premoltTPL, cm Figure 5,—Growthinthelaboratorybyspidersfrom fourpopulations. Datafromtwomoltsareplotted,with the first molt indicated by open symbols and the second indicated by closed symbols. Squares = Chamela; diamonds = Tehuacan; triangles = Fortin; circles = Nanciyaga. The arrows indicate values forthe male from Nanciyagathatdelayed molting. Graph a = Growth perecdysis in the laboratory, plottedaspostmoltTPL(leg I tibia + patella length) vs. premolt TPL. Graph b = Premolt abdomen volume as a function ofpremolt TPL. Common garden experiment.—Twenty-three premolt TPL (0.2-0.5 cm). Preliminary analysis spiders with TPLof0.2-0.4 cm were brought to revealedno significantdifferencein slopeamong the laboratory from Nanciyaga, Fortin, Tehu- sites orbetween conditions (field orlaboratory), acan, and Chamela and held for one complete norasignificantinteractionofsiteandcondition intermoltcycle(twomolts).Theentirestudylast- (ANCOVA: site x TPL: F(3 gg) = 0.43, ns; con- eindTfPrLomweJruenevetroifOicetdobbyerm,ysaenldf,wvhairlieatailloncchaaunsgeeds dF(i3t.i6o8n) =X T0-P1L4:,Fn(s,).68)F=in0a.l10A,NnsC;OsiVteAx tceosntdiintgiofno:r by different persons making measurements pro- primary effects showed no significant variation ddhuaocdmeetdonibvneoclreuexmacesl.euddIenedrpraforrrtoiicmnultathrhe,edaeansattlaiyfmsraistoimooftnswinoocdfraeatabes-s- d=<ue00..6t05o0,1s;intses)i.toer:cFo(n3d7i5)t=ion1.(5T2,PLn:s;Fc(,on7d5i)t=io4n5:0.F(9,,7P5) ingabdomenvolume,resultinginremovaloffive Several parameters describing the intermolt observations. interval were collected from the laboratory ani- Atotal of13 maleswere includedinthe study mals (Table 4). ANOVA showed no significant andfiveofthemmoltedtomaturityinthesecond difference among populations in mean TPL fol- moltinthelaboratory: onefromNanciyaga,two lowing the first molt in the laboratory, although from Fortin andtwo from Tehuacan. Therefore, Nanciyaga and Fortin individuals were slightly analyses included the parameter ofjuvenile vy. larger. Thetotalintermoltintervalandthenum- maturation molt where appropriate. ber ofdays foraging between molts were not af- Utilizing both observed molts for each indi- fected by these slight differences in size (regres- = wv(Fiaidsgu.asl5i,ag)n.riefAigcrNaenCstsOfiooVrnAeaanocafhlygpsroiopswutolhfatgpireoronwe(tcahdllypsPiesr<se0hc.do0yw0se3id)s s=1i.o335n.,43on,fs;ntsor)te.aglTrheisenstiteoortnmaloolfitndtadeyurrsmaoftloitroanig:nitnTegrP:vLaT:lPaLFn:(,dFd(,5,a)y,s4) no significant difference among these popula- foraginginthelaboratoryvariedamongsitesbut tions or betweenjuvenile and maturation molts wasnotaffectedbywhethertheindividualmolt- (nointeractioneffects,P > 0. ANCOVA.TPL: ed to sexual maturity (ANOVA ofintermolt in- dF0(.a,5t5a3,8)wne=sr;e2m3ca0ot.um9r,paaPtrieo<dn:0to.F0(3,0713;3o)bp=so1ep;0ru.vl0aa0tt4ii,oonnns:s)o.fTmho33el)st=es tFF(e(,r3,v,1a65l)).==s4i1t..e66:94,,FP(n3s;=,6A)0=N.0O23;V.6mA9a,tuoPrfidt=ayy:0s.F(0f,3or;,a5g)mi=antgu0,r.i0sit0ty2e:,: in the field from the same sites and the same ns). Differences between Nanciyaga and the re- HIGGINS-DEVELOPMENTOFJUVENILE NEPHILA CLAVIPES 115 Table 4.—Intermolt duration in the laboratory for spiders from four populations. The TPL (leg 1 tibia + patellalength)reportedisthemeasurementfollowingthefirstmoltinthelaboratory. Lettersrefertostatistically similarvalues among sites. Site n TPL ± 1 SD Total days ± 1 SD Days foraging ± 1 SD Nanciyaga 4 0.48 ± 0.08 26.8 ± 8.0 (a) 24.3 ± 9.0 (c) Fortin de las Flores 6 0.47 ± 0.04 18.8 ± 4.0 (b) 15.4 ± 2.5 (d) Tehuacan 6 0.40 ±0.13 17.3 ± 4.3 (b) 14.8 ± 3.9 (d) Chamela 7 0.40 ± 0.04 18.7 ± 3.1 (b) 13.2 ± 1.7 (d) mainingsitesappeardueto onemalefromNan- parameter can result in the same adult size, but ciyaga that took over 30 days to complete the the conditions under which each varies may be intermolt interval and molt to maturity, twice distinct. Genetic variation or phenotypic plas- the usual intermolt duration for spiders ofthis ticity can lead to differences in development size. within and among populations, but the evolu- The abdomen volume gain in the laboratory tionaryconsequencesofeach sourceofvariation wasindependentofsite. Inthefirstmolt, spiders are distinct (Pease & Bull 1988). Longitudinal from Tehuacan molted at a significantly higher observations ofjuvenile growth are a first step premoltabdomenvolume(nointeractionaffects; towardsdetermininghowenvironmentalfactors ANCOVA. site: F^, = 5.42, P = 0.01). All generate differences in adult size, and whether ,3) spidersmoltedatthesamerelativeabdomenvol- these differences are the result ofphenotypic re- ume in the second molt (no interaction affects; sponse to the environment or genetic variation ANCOVA. TPL: F„, = 43.4, P < 0.001; site: amongindividualswithinoramonghabitats.The ,6) F(3,,6) = 2.61, ns; maturity: F(, ,6) = 1-3; ns) (Fig. results of the studies ofN. davipes imply that 5b). Because there was no difference in the sec- somedevelopmentalparametersarehighlyplas- ond molt in premolt abdomen volume among tic while growth per ecdysis is constrained and sites or between molts to maturity andjuvenile may be genetically determined (Higgins 1992a, molts, data from all individuals held fora com- presentstudy). Suchinformationisnotavailable pleteintermoltcycle(includingthreeofunknown from the census data presented in past arach- origin)were combinedto describe the pattern of nological studies without making basic assump- abdomen volume increase over the intermolt. tions concerning developmental processes. Therelativechangeinabdomenvolume[ln(av(d)/ Traditionally, field measures of growth uti- av (0)], where d = day and 0 = day ofmolt, was lized either frequency distributions of a single plotted against time for spiders molting within measure, such as carapace width, or the regres- 20 days, for spiders molting in 20-26 days, and sion oftwo allometricbody parts ofindividuals. for the individual from Nanciyaga requiring 36 Thedifficultiesofdetermininggrowthandinstar days (Fig. 6). The general trend was for the rate numberfromtheformermeasurehavebeenrec- ofabdomen volume increase to slow as the spi- ognized (Polis & Sisson 1990); however, the lat- ders approached the next molt. The individual teranalysisalsopresentsincompleteinformation from Nanciyaga that took longerbetween molts (Teissier 1960). Inhisformulationdx/dt = aGx, did achieve a greater premolt abdomen volume Huxley assumes constant growth per unit time andgrew slightly more at ecdysis than the other if the environmental factors represented by G spiders,asindicatedinFigs. 5aand5bbyarrows. are constant (1972, p. 6). In order to determine = DISCUSSION the rate ofgrowth orthe growth perecdysis (/ 1 instar), one must assume that the individuals In order to interpret variation in phenology are movingalongthetrajectorydescribedbythe and size at maturity, the proximal developmen- allometric relationships at a constant rate of tal causes ofthe variation must be identified. In growth per ecdysis, an assumption that may be arthropods, variation in two developmental pa- invalid ifthe growth perecdysisresponds to en- rameters can lead to differences in size at ma- vironmental factors (G). This assumption is in- turity: there may be variation in the change in valid for the linyphiid Linyphia triangularis size at each molt, or there may be variation in ClerckandmaybeinvalidforthelycosidLycosa thenumberofjuvenilemolts.Variationineither helluo (Turnbull 1962, Uetz et al. 1992). Turn- THEJOURNALOFARACHNOLOGY 116 volumeO) abdomen / volume (abdomen 0 5 ° 15 20 25 In —— T— —r —I T I T I 0 5 10 15 20 25 30 35 Day Figure 6,—Mean and SD ofabdomen volume gain in the laboratory over time. Spiders from all sites were grouped accordingto the duration ofthe intermolt.