predation. competition, and the composition of …

E lt t l ~ t lt ~ ~L ~ ~ o t ~ ~ ~ o p I ~lt gt Il i gt pp 119-138 c- IM3 h i rhc Lccgtlilcil r l a c r b elf rnrt~zi

PREDATION COMPETITION AND THE COMPOSITION OF LARVAL ANURAN GUILDS

Ahcrrtrit Experimental manipulations of densities of the predatory salamanders lVoroph t ~ t r l r~~r~Ic ~irirlrccrrzctii~rctrlicand 41~1h~tot7ltrtirirlri~l significantly altered relative abundances of six species of larval anuriin in 22 artificial-pond communitie One competitively inferior anuran Hylo crirciPr was virtually excluded from predator-free control communities but survived best and occurred at greater relative abundances in communities containing high densities of l V o r o p h t l ~ t r l ~ ~ ~ t r A econd competitively inferior pecies Hyltr rcit ioci survived best at intermediate intensities of predation Each of four competitively superior species (Sctrphiopuc holhroohi Ktrt~tr r p l ~ e t ~ o c ~ c ~ p ~ c i l t r Bufi) t r r -rrctric and Hylci chryocelis) exhibited inverse relationhips between relative abundance and Vo-tophrl~crlt~lricdensity At~lhycrorltreliminated the entire anuran guild from tank communitie and had a much greater per capita impact on anuran guild composition than did hotophrhaln~ric

In most anuran species maximum and mean mass at metamorphoia were positively correlated with predator density suggesting an inverse relationship between intensities of predation and competition among tadpoles Low growth rates of most anul-an species in the absence of predators were correlated with high abundances of superior competitors These results indicate that predators mediated interspecific competition among larval anurans

Intermediate values of horophrhult711~density maximized the total production of anuran meta-morph biomass Biomass of metamorphs of each species varied in a species-specific manner with predation The propensity of Kcinii cp l~er loc~rph~rl t i for overwintering as tadpoles following a season of growth was also related to predator density Laboratory feeding experiments demonstrated that at lea[ two competitively superior anurans 5c~trpkii1p1rtholhrooXi and R ~ i j i )tcarrectric were especially vulnerable to predation by salamanders This suggested that V o t o ~ ~ k t ~ t i l t ~ ~ t r cmay preferentially remove competitively wperior anurans from pond communities allowing competitively inferior anuran to persit and to complete development successfully at moderate to high predator densities

he) i~r~rt ic Ambystoma Bufo c~on~n~r tn i r y con~l~ctirioncrr~ict~irr rliersiry ~ r i i l t i Hyla KO- tophthiilmu prrrlrrfion Rana Scaphiopus terll~ortrry po t~r l c

I N I KOI)ICTION is needed to determine whether the overriding impor-

A general sqnthesis of the roles of predation para- tance of predation is peculiar to space-limited species

sitism m u t ~ ~ a l i s m and competition in determining or constitutes a general theme of communitq organi-

communitq structure is a principal goal of community zation In particular the impact of predation on the (sensu Root ecologq Predation has received particular emphasis structure of g ~ ~ i l d s 1967) of vertebrates

in previous sqntheses (Paine 1966 Harper 1969 Con- remains largelq conjectural The composition of many

nell 1975 Menge and Sutherland 1976 Lubchenco guilds of lizards birds and mammals has been ex-

1978) Predators can enhance prey species richness bq plained as the outcome of interspecific competition

limiting abundances of superior competitors therebq (e g Mac Arthur 1958 Brown and Lieberman 1973

preventing monopolization of critical resources bq Pianka 1973 Codq 1974 Schoener 1974) However

competitivelq dominant species and precluding local an experimental studq of larval salamanders (Wilbur

competitive exclusion of inferior competitors (Paine 1972) and a comparative studq of piscine guild com-

1966 Daq ton 1971 1975) Predation also promotes the position (Zaret and Paine 1973) suggest that specie maqdominance of species or morphs which are either re- composition in some vertebrate g ~ ~ i l d s also de-

sistant to o r dependent on predation (Paine itnd V~tdas pend on predation

1969 Dodson 1974 Kerfoot 1975 Paine 1980 Zaret This paper describes differences in the species com-

1980) position of larval-anurnn guilds caused bq experimen-

The generalit) of previous syntheses is compro- tal manipulations of the abundance of predatory sal-

mised by their restriction to sessile space-limited amanders in irtificial-pond communities lhese results

species or aquatic invertebrates of low trophic status specifically address how predation and interspecific

Information from a greater diversity of communities competition maq interact to structure assemblages of mobile vertebrate prey and provide a distinct coun-

Manucript received 3 August 1981 revied I June 1982 terpoint to previous studies of analogous processes in

acceoted 3 Julc 1982 assemblages of sessile invertebrates and plants Fac- i re en t adbress Department of Biolog) College of the tors previouslq suggested to influence the species

Hol) Cross Worcester Maachusett 01610 USA composition of larval-amphibian assemblages include

I20 PETER J MORIN Ecological Monographs Vol 53 No 2

interpecific competition (Dumas 1964 Wiltshire and m) filled with rainwater They are acidic (pH = 40-Bull 1977) predation (Heyer et al 1975 Walters 1975) 65) and dilute (conductivity lt40 pSIcm at lgdegC) Many and ~nteractions between competition and predation ponds completely evaporate during periods of drought (Brockelman 1969 Wilbur 1972 1980) Preliminary re- which tend to occur in late summer and autumn These sults reported earlier (Morin 1981) for three anuran episodes of desiccation apparently exclude fish from species in the present study demonstrated that com- most ponds petition and predation interact to produce nonrandom In the Sandhills salamanders assume the role of patterns of species composition This paper extends dominant vertebrate predators where fish are absent the effects of predation to three additional species in Observations of predation by salamanders on amphib- the nura ran guild documents an inverse relation be- ian eggs and larvae are frequently correlated with the tween the intensities of predation and competition postbreeding exclusion of species from ponds (Walters among anurans and describes alternate patterns of 1975 Morin 1983) The broken-striped newt Notoph-larval development that permit successful metamor- thalmus viridescens dorsrilis enjoys a nearly ubiqui- phosis at high intensities of competition or predation tous distribution among 17 frequently observed ponds

Several other salamanders including Ambystomri ti- I H E EM NATuRAL TEMPORARY grinum Amhystomtr mabeii and Siren intermeditr are ON A N D ExPERIMENTAL ANALOGs less frequently encountered Notophthrilmirs is a gen-

Although my experiments were conducted in artifi- eralist predator that feeds on zooplankton large aquatic cia1 ponds they were inspired by the natural history invertebrates and the eggs and larvae of amphibians of temporary ponds in the Sandhills Game Manage- (Wood and Goodwin 1954 Christman and Franz 1973 men1 Area of Scotland and Richmond Counties in Mellors 1975 Walters 1975 Gill 1978) south-central North Carolina USA The following Rigorous experimental study of the impact of sala- overview of the natural history of natural ponds iden- mander predation demands control over potentially tifie4 potential sources of community organization confounded factors (especially variable colonization spec~ties the limitations of natural ponds as experi- and historical differences in species composition) that mental units and emphasizes the aspects of natural are difficult to separate control or manipulate in nat- ponds that could be mimicked in experimental artificial ural ponds Effective removals of Notophthtrlmlrs are ponds especially difficult in these natural ponds because these

The natural ponds are potential breeding sites for nonmigrating populations of newts cannot be excluded -25 species of frogs (Martof et al 1980) and up to 17 by interception at drift fences (Storm and Pimentel species of frogs may chorus and breed near a single 1954) Homing behavior also compromises the suc-pond during one evening Consequently larvae of many cessful augmentation of newt densities (Gill 1979) To species are forced into syntopy after breeding chorus- overcome the potential limitations imposed on exper- es that follow heavy rain creating rich and varied pos- iments conducted in restricted numbers of highly vari- sibilities for interspecific interactions There is little able natural ponds I manipulated salamander densi- evidence for habitat separation among larval anurans ties in artificial-pond communities reconstituted in in the Sandhills and aside from patterns of temporal cattle-watering tanks These analogs of temporary separation that vary greatly among years mechanisms ponds contained many of the species found in natural reducing ecological overlap among these or other lar- ponds but could be controlled manipulated and cen- val xnurans are not pronounced (Heyer 1974 1976) sused better than natural ponds Reconstructed fresh- In general tadpoles feed by rasping and filtering bac- water communities ranging from laboratory micro- teria phytoplankton periphyton and detritus (Was- cosms (Neil1 1974 1975) to artificial ponds (Hurlbert sersug 1975) To exploit temporary ponds successful- et al 1972 Johnson 1973) possess distinct advantages ly tadpoles must garner sufficient resources to for the empirical study of communities Construction metamorphose before ponds dry while simultaneous- of initially similar communities eliminates historical ly eading predators (Wilbur 1980) Successful mem- differences in species composition Reconstituted bership in a tadpole guild is best (and somewhat par- trophic webs can be engineered to contain fewer species adoxically) assayed by the successful metamorphosis than natural webs thereby revealing details of inter- and departure of tadpoles from the pond community specific interactions The realism of such reconstituted while unsuccessful members are those species that fail communities is confirmed by the successful develop- to complete larval development ment persistence and reproduction of component

Although most species can chorus and breed in each species (Morin 1983) pond in the Sandhills species composition of tadpoles Theory suggests that dispersal may influence broad and metamorphs varies conspicuously among ponds patterns of predator-mediated coexistence of prey This varixtion may reflect stochastic exploitation of species in collections of discrete habitats linked by ponds by ovipositing frogs However tadpoles of some migration (Caswell 1978) I have focused first on pat- species often fail to persist in ponds even after the terns of species composition and community dynamics input of thousands of viable eggs Most of the natural caused by processes operating strictly within com-ponds are shallow depressions (maximum depth lt 1 munities

J u n e 1981 P K F 1 ) 4 1 1 0 2 4 N I ) A N U K 4 N C I L I L I ) C O L I P O I I 1 0 2 111

1 reconstructed 22 pond communities in cylindrical galvanized steel cattle-watering tanks which were 152 m in diameter and 062 m in height The interior of each tank was painted with white epoxq enamel to retard rusting Each tank contained = 1 m of water drawn from the Durham municipal water suppl) The tanks were placed in an arraq of five rows of four tanks and a sixth row of two tanks in the Duke Universitq Botanq Plot located in Durham Count) North Car- olina lightlq fitting lids constructed of Fiberglas win- dow screen stapled to hexagonal wooden frames retained experimental organisms and excluded unwanted colonists

After filling the tanks with water in early March 1980 I added measured amounts of litter nutrients and macroph) tes to each tank Each tank received 50 g of Purina trout chow and 550 g of dry litter Trout chow provided nutrients grassy litter collected near a pond in Scotland County provided cover additional nu-trients and infusoria Fift) washed stems of the aquatic macrophq te M~iopll l lrrtrpi~lrlt lrrit i l were planted in the litter in each tank to provide additional spatial heterogeneit)

1 reconstituted simple temporary-pond trophic webs by inoculating the tanks with temporarq-pond organisms via a repeatable protocol Successive inocula-tions of each tank on 14 20 and 22 March 1980 es-tablished phytoplankton periphyton zooplankton and small invertebrate5 On each date an inoculum consisted of 650 mL of a well-mixed suspen5ion of organisms in pond water Inocula were drawn from a mix-ture of organisms collected bq tows of an 88-pm mesh plankton net through eight natural ponds in the Sand- hills 1 screened inocula to prevent uncontrolled introductions of large predatory insects or larval amphibians Tanks were inoculated within 94 h of organism collection Within 1 mo of inoculation cladocera and copepods became abundant and cleared much of the phqtoplankton and bacteria from the previously green and turbid water The diversity of common inverte-brate taxa that persisted and reproduced (see Appen- dix I ) emphasized the resemblance of tank communities to natural ponds

1 also introduced three taxa of macroinvel-tebl-ates a s alternate prey for salamanders 10 adult Syt~rrrc~lltr c~ i~~~ t r~hor l t r i t z i(Amphipoda G~tmmal-idae) per tank on 2 1 March and several adults and freshly deposited eggs of Hes~c~rocorirtrI and Siytrrtr ( I (Hemiptera Co- rixidae) on 22 March

1 experimentally manipulated one factor the initial densities of adult Yoro~~ztizcll~~lrrviridesccrls and larval I t~~hy t ro t i l c~rigritllltrz established in each tank Six treatments were defined by different initial densities

of salamanders Four treatments constituted an ex-perimental gradient in predator densit) and consisted of the addition of 0 2 4 or 8 iYoro~~izr i~ci l i t i~r to each tank with a sex ratio of I 1 in each population Each treatment was replicated in four tanks lwo additional treatments consisted of the addition of either both four At i rhy ro t~t r larvae and four V o r o ~ ~ i ~ r i ~ t i l n ~ i r tor on15 four Atilhtrotrlrr to remaining tanks lhese two treatments were each replicated in three tanks Individual salamanders and replicates of each treatment were ritndomly assigned to tanks in accord with a random- ized design for variance analysis

IYoro~~ i~ r l~c~ l t i l i r c were collected from and Ainhjtot~~rr Grassq Pond in Scotland County I Y o r o ~ ~ i z r ~ t r l ~ t ~ ~ ~ t were dipnetted on 19 March 1980 Developing III-hytrottrtr embrq os were collected in Februarq 1980 Hatchling Al~hytrointrwere reared in the laboratorq on a diet of zooplankton amphipods and isopods Y o r o ~ ~ r ~ t r l t r ~ ~ r ~with a mean mass of 12 18 mg were added to appropriate tanks on 22 March Atz~trot11t1 with a mean mass of 979 mg were added to tanks on 27 March before the addition of tadpoles Atr~hgtctotltr of this size were invulnerable to predation b) Yo-roIl lr1tlt~1l~

Initiallq identical tadpole guilds were established in all tanks by adding hatchlings of six species of anurans to the tanks on three s ~ ~ c c e s s i v e dates in 1980 Hatch-lings were obtained by collecting egg masses or am- plectant frogs in the Sandhills and Durham Count) Numbers of each species added to each tank were 100 Ktrt~cl spilrrroc~c~~iltrItr 900 Scrli~iriopirc 11olhtooXi and 300 H ~ l t icrrrc(fbron 27 March 300 Bid ) trrrttr ic on 2 Maq and 150 HyIt i cilrcoc~~lic and 150 Hyltr grtrr io~rron 24 Maq Three successive introductions corresponded to the natural phenologq of breeding choruses in the Sandhills Rather than artificially con- trive additions of all species on 1 d 1 preserved the natural phenologq and potential advantages and dis- advantages conferred bq the natural timing of breeding lhese six species were the most abundant species breeding in the Sandhills in 1980 and were assumed to be the major potential members of the larval-anuran guild

Hatchlings from multiple clutches of a species were mixed before addition to I-andomize potential effects of genetic variation among sibships (lravis 1980) Counted hatchlings were randomly assigned to each tank and all introductions on each date were made within 1 h This protocol ensured that initial inti-aspe-cific densities did not vary among tanks or treatments

Interspecific differences in initial anuritn densities arose from unavoidable differences in availabilities of viitble hatchlings however experimentttl densities of salitmanders and tadpoles were well within the range of densities estimated in natural ponds Amphibian densities were sampled with a screen drop box sampler similar to that used by Turnipseed and Altig ( 1975) Densities of representative species in the Sandhills in April 1979 (Table 1 ) indicate that initial experimental

--

T A R IE I Mean densities (tstandard error) of common amphibians sampled in natural ponds during late April and earl) May 1979 Units of density are the number of individu~ls per cubic metre 1 m i the average volume of a cattle tank Means ar-e based on six haphazar-d samples per pond The sampler enclosed an area of 05 m and samples were taken it seceral depths uithin each pond Experi- mental densities established in the cattle tanks are given at the bottom of the table for comparison

Mean density ( i i ~ )

VOro[ll-Pond 1 1 HyI(i crircii~ Riirlci spp

Wagram 19 i 13 119 t 49 145 t 74 Jog I 22 t 22 240 t 149 00 t 00 Jog 2 00 t 00 3626 t 1264 2784 t 1356 Mistletoe 63 t 42 4640 t 1264 00 t 00 Deep 104 t 39 00 t 00 792 t 158 Sandbou l 00 t 00 00 i 00 133 t 45+ Ditch I 17 i 17 784 t 452 00 t 00 SWI- 22 t 22 492 i 241 00 t 00 N EI- 00 t 00 00 i 00 00 t 00 PAEOR 00 t 00t 3194 2 447 210 i 71 Grassy 155 t 25 377 t 235 00 t 00 Weed 00 t 00$ 2557 t 702 00 t 00 Slirnington 40 t 23 1301 t 708 00 t 00 S q ~ ~ a r e 80 t 29 284 i 185 00 t 00 Cattle tanks 0 2 4 8 300 100

fadpoles of Ilt(irici c [ ) l r t ~ r r o c ~ c ~ ~ ~ l ~ ~ i ~ i Overwinter-ing tadpoles of K e i r ~ c i c~cii~ririiir Known to be present in pond but not found in six sam-

ples

densities of ttidpoles and newts were not excessive Counts of 41tlh~ro1l~ eggs in iri~~i~lrotl small pond (NET) of known area yielded an estimate of 8 hatchlings mof benthos in 1980 In compirison A-t~cto~tlcrdensities in the tanks corresponded to ~ 7 2 larvae m2

I completelq censused the final species composition of each tadpole g ~ ~ i l d metamor-by daily collecting ill phosed froglets from each tank Metamorphosis was defined bq the emergence of at least one forelimb Be- cause surviving a n ~ ~ - a n s completelywere censused rather than incompletely sampled nieasu~-ed attributes of anuran populations ( w c h as mean mass at metamorphosis) were described by complete distributions and were therefore not subject to sampling error I intel-changeably use the terms abundance and densitq to d e x r i b e total numbers of lnurzins censured from each tank because tmAs were similar in olume and area

Captured froglets were taken to the label-atol-q where the towel-dried wet mass of each frog was measured to the nearest milligram Metamorphs were retained until tail I-esol-ption was completed when each frog was weighed again and released in Dul-ham or Scot- land County

I terminated the tank expel-inients on 18 November 1980 1 mo after anuran nietamorphois ceased Manq but not all natural ponds had dried bq this date indicating that the experiments duration col-respond-

M O K I N t c ~ i l ~ i g ~ c ~ lhlonogr~ph V u l ( 3 K O 2

ed to the time available for larval development in the Sandhills At termination six tanks still contained populations of large overwintering Klirlci tadpoles that exhibited no indication of imminent metamorphosis Amphibians large arthropods litter and mticl-ophytes were harvested from the tanks with a large dip net A[-thl-opods were preserved for later enumeration R(itlci tadpoles were counted and their towel-dried wet mass was determined to the nearest milligram

I ~ ~ s e dlaboratory trials to detect interspecific difference among tiidpoles in vulnerability to predation bq Vorol111rllcil1r11(and A t t ~ h y t r o ~ ~ ( r are-Experimental nas were co e~-ed polyethylene pans (30 cm long x 23 cm wide x 19 cm deep) containing a 9 cm deep layer of washed sand 7 i of water and a plastic a q ~ ~ i i r i ~ ~ n ~ plant for cover In each experimentiil replicate 90 similarly sized tadpoles of each of two or three species were kidded to a pan Ttidpoles were sorted into size- clases with s iees and tadpoles of a single size-class were used to remove biases caused by interspecific differences in size Small recentlq hatched tadpoles were ~ ~ s e d in all experiments because pl-edatol--related niortalitq in natural pop~llations is greatest shortlq if- ter hatching (Herreid and Kinney 1966 Calef 1973 Licht 1974) Aftel- tadpole addition ii single predator was added to each pan Predators fed for 2 h and were then I-emoved Surviving tadpoles were identified to species and counted Frequencies of predation on each species were determined by subtl-action Prey densities and the duration of feeding trials ensured that predators could feed to satiation on i single preq species Salamanders were haphazardlq selected from laboratory stocAs maintained on diets of Purina trout chow (Ralston-Purina Saint Louis Missouri) and a q ~ ~ a t i cinvel-tebl-ate Saliimanders were not fed for at least 24 h before each feeding t l - i d ltidpoles of nine species (Ko~lci olrc~roccl~lrrrlo IlolhrooliSc~crl~l~iol~ci H r ~ f i )rc~rct~ic H clriccicr~crttlrrcri r~icricirer Hlcr c~r~cjfir and I o11-H yrcirio er I clr~cocclic clc~o~ri)were obtained by collecting eggs and tadpoles from natural ponds in Scotland and Durham Counties North Carolina Combinations of species used in different trials were determined by the availability of species of similar size

Results of differential predation experiments were anilyzed with standard techniques for frequency analysis (SoAal and Rohl 1981 ) Numbers of consumed tadpoles were summed separatelq bq species over till in-dependent replicates of each experiment Thece sums defined frequencies of predation on each species in a given experiment Significant deviations (tested with t chi-square statistic) of observed frequencies of pl-e- dation from frequencies expected under the null hq- pothesis of indiscriminant predation on all species were ~ l s e d to infer species-specific differences in vulnera- bilitq to predators

I his sectlon brretlq ]dentifie the maln hqpothee reted t o detect and decl-ibe the effects of s~t lamander predation on both the entire kinuran girild and it coni- poneni populations Specilic hbpothees ancl statistical les t are justified ancl detailed further in Appendix 11

1 ) Do stltniitnder treatments alter the final compo- sitlon of unul-an girilds G~ l i ld composition in each tank + ~ sdelined b) the Lectol P = (11 11- 11 11 11 11 i here 11 is the fraction ( o r r e l a t i ~ e ttbunclancei of all s i rn i lng anurans cenwsecl from tank j belong-rng to specles r A 41ngle-ftctor MANO A wis used to cletect I hether Pdiffered significant11 aniong pred- ittor treatments ancl a discriminant filnction anal ) i s identitied specie4 whoe relatihe abundance contrib- irted to differences aniong treatments Because the relative abundance4 ([I) in each tank sum to one (and thus Ire Ilnear I ) clependent) the M 4 N 0 4 wa5 per- fol-med on Lectors of fihe rather than ix pecies

2 1 Do sali~mancter alter the cornbinect densit) of all s~ r r ~ iv ingtadpoles An ANOVA of the total numbel of anurans censused from each tank tested whether-wlamancter treatment4 intluenced the total densitq of t a d p o l e s s u r v i i ng a n d p r e s u n i a b l ) c o m p e t i n g th~oughout the experiment

-3) [lo alaniincters affect the surival of each anuritn pecies Separate ANOVA4 for the effect of salamander treatments on the ilrvitl of each snecies i l -lustl-ittecl the basis for changes in the I - e l a t i ~ e abun-dance o r doniinince of species

4 1 Doe salanitnder pr-edation alter a pec t s of the 1 t r lt i l development of each pecie that norniallq re-f e c t the intensitq of competition experienced during deelopment Separa te MANOVcs for each specie tested whether combined patterns of mean mass at metxnorpho is ( in milligr-am) mean Ittral period tin dabs ) and mean growth rates tin niilligrims per dab) differed among treatments for populations of each species These measures are fitnes components of Iiir- a1 frog (Wilbur 1972 Smith-Gill and Gill 19781 reductions in mean mass 01- mean growth rates 01- in-creases in mean larval periods indicate increased intensitie of competit ion

i) Is in inclex of the intensti) of competit ion (mean g r o ~ r t h l-iitel colreltted with the abundttnces of sur-~ i v i n g potential competitors in each tank o e r all trettnients A canonical correlation analbsis tested whether the mean growth rate = ( of 4pecies i in t a n k j as colrelated with a canonical triable 1=

CVl - c2 2J 7 cjvjJ + c4VJ + + C6V where V is the densitq of species i in tank and the coefficient5 c are chosen to nitximize the correlation between ( and Iwithin tank o t e r all communities This anttlqsis is directlq tnalogous to multiple regression techniques for estimating niultispecies competitive relations (Emlen I ) but i t relies on a correl~-t i e a p p r o a c h b e c a u s e i tbundi tnces o f po ten t i a l

compet i tors were not experinientallq manipulated Specie correlated in abundance with C are identified a s probable competitors bb negitie product moment correlttions between value of 1 and densitie4 of SLII-

~ i v i n g potential competitors 6) Doe predation influence the total production of

metitmorph biomas An ANOVA of the effects of predator t leatments on the cornblned b ~ o n i a s of all iinurtn censused from each tank teted whether pred- a tors influenced the export of secondarb production froni tank ancl suggesteel whether t h ~ n n ~ n g b) pred~t- tors reduced competition and enhanced pr-ocluction

7 ) Does predation alter the di5tribution of g~l i ld biomass among pecies T h e relative clistribution of biomass among species in each tank defined a vectol-If = OII~ 12 I I I I() where I is the fraction of total g~l i ld biomass collecteel from tank j contr-ibuted bq species i 4 M A N O V A on $1 inilo-got14 to the analbsis of relktt i~e abundance decrihecl abohe indicated whether guild bionias wt distrib-uted cliffel entlq among specles in ctlftel-ent trettments provieling an alternate measure of the dominance of specie based on their abilitq to extract nutrient froni tank c o m m ~ ~ n i t i e s

All tatitical itnal)se were performeel with proce- dure of the Statistictl Analbsis Sqstem (Helwig ancl Council 1979)

Preclation b) larhal A ~ ~ l ~ ~ t o l ~ r ressentiitllq cleleted the entire anuran guild froni tank communities Tacl- pole failed to survihe through metamorphosis in half of the tank containing AIJIIY~IOII~~ In the remaining three tank two three and eight tadpoles s ~ ~ r ~ i e d and nietanior-phosed These tadpoles includeel I O H ~ l r i

v~rr~iocoRri11o ~ ~ l r c ~ ~ o c ~ c ~ ) l ~ o l o flgtl(ecctr(fijt anclI I I f l ~ l r r clrlcocc~lic T h e preence o r tbence of fc3~11- V ~ ~ O ~ I I ~ I ~ I ~ I I I ~ wasin addition t o four - I ~ I I~ ~OI~ I~ I not related to complete exclusion of tadpole Because few 01- n o tadpoles su r i ed in these tanks the) were not included in the analysis of remaining treatments which conipl-ised a salarnande~-densit) gradient froni predator-free tanks to tanks containing eight Vo top l~ -rl1tr11~1rrcH e r e a f ~ e r differences among treatment4 ascribeel t o salamander densit) refer only t o differences in o r o p l ~ ~ l ~ i r l ~ ~ ~ r t cdensit)

Differences in salamander densit) produced striking difference in the final specie composition of the anuran gi~i ld ( Fig I ) Metamorphs of Yc tiplriol~rrc 1101- 111ooXi predominated in salktmandel--free tanks K o ~ ~ r i ~ ~ l ~ c ~ r o c ~ c ~ p l ~ i l o(combined nietamorph and o ~ e r w i n - tering tadpoles) Brcfi) tc~~ccrric and f ~ l i i (Igt coc olic also persisted at model-ate r e l ~ ~ t i v e abundances in the absence of z~lamitnders Metaniorphs of f l gt l c r c~crcjligtr and Hgtl(goriocci were notably rare in predator-free

PETFK J 1 1 0 K l N Ecolog~c i~ lhfvnogt-~ph Val 53 Yo 2

S c a ~ h l o ~ u sholbrookl Hyla cruclfer Rana sphenocephala

Bufo terrestr~s ~_yla~ h r y ~ o c e h b l a gratiosa y W tKKtEf

0 2 4 8 Newt D e n s ~ t y (no [ t a n k )

F I ~ I Mean relative abundances of surv~ving c e n s ~ ~ s e d anurans at each level of or0pl1rl1cil1~711~d e n s ~ t ) R e l a t i ~ e nbundances ~ n d ~ c a t e anuran in each tank correspond~ng to each species Relative the percent of the total number of s ~ l r ~ ~ ~ i n g abundance of R(ir~ciinclude cornb~ned abundances of rnetamorphs and overwinter~ng tadpole Histograms for each pecle Lire Identified by the keq at the top of the figure 411 means are based on four replicate communltles at each level of oroplrrlr~ilr~~rc~denvt)

tanks despite moderate initial relative abundances as hatchlings ( 2 5 and 1257 respectively of all introduced hatchlings)

In tanks containing eight 2oo~1hlztrlt1ricanuran relative abundances differed greatly from the pattern observed in the predator-free tanks (Fig I ) Meta-morph4 of H ~ l r r crrrcjfir predominated Species metamorphosing at high to moderate relative abundance in the predator-free tanks ( S IzolhrooLi R plzc~t~o-c~cpl~trlrrB rct~cricand I f c~l~r~occ~lis ) were relatively rare at high salamander densities Relative abundances of metamorphosing H grrriostr were consistently low at all 2oopl1tlzrrlt~zr~cdensities

Differences in species composition between ex-tremes of the newt den5ity gradient did not correspond to alternate discrete states of guild structure Instead there was a continuum of guild structure With in-

creasing salamander density numerical dominance bq Scrrpl~iopric 11olh1oohi gradually shifted to numerical dominance bq Ifyltr crrrcifi~r This shift was accom- panied by declining relative abundances of R pI~e-tzoc~c~pl~trlrr I f c~hlsoc~c~lit re-B rerrc~srric and in sponse to increased salamander density

A MANOVA of vectors of angularly transformed relative abundances demonstrated that predator-mediated differences in anuran guild composition were statistically significant ( P lt 01 Tables 2 and 3 ) Hence variation density signifi- in 2010~1l1tl1trltiirr~ cantly influenced overall patterns of anuran species composition The discriminant function analqi of anuran relative abundances (Table 3 ) indicated that differences in species composition detected by the MAN- OVA corresponded to differences among treatment in relative abundances of H crrlciir 5 IrolhrooLi R

T A B IF 2 Newt density and error matrices of sum of squares and cros product for the k1AKOV4 of final anuran relat~ve abundance Diagonal elements runnlng from upper left to lower right correpond to univariate urns of squares for eparate ANOVAs of the relat~ve abundances of each specie

-

Sums of squares or cros products

klatrix (dt) Species

Error 11hrooXi rrcifir Ilc~rrOc~c~~lltllLi rrosrric trtio 5 L l

N e u t dens~t (3) ~IhrooXi rrcifir ~ h ~ ~ l l O c ~ ~ ~ [ ~ h ~ l ( l rrcJ5 rris ciriocci

June 1981 PREDArlON 4 h D A h U R A h GUIIll COMPOSITION 1 2 5

T X H IE 3 Multivariate test criterion and summary of discriminant function analysis for differences in anuran relative abundances among neut treatments Wilks Crite-ion = 176) P lt 0 1 ~ i ~ ~ r i ~ i ~function and correlations between discriminant function scores and transformed relative abundance of anurans are g i ~ e n

Species Coefficients Correlations

c ~ ~ l r o ~ o c ~ c ~ ~ ~ l ~ r r l ~ r and Brtfi) rcrrcrtris Relative abundances of these species were significantly correlated with values of discriminant function scores for each tank and therefore were primarily responsible for differences in species composition among treatments

The total number of surviving anurans of all species censured from each tank was inversely related to Yo-toplltlrtr1t1lrir density (Fig P i = 927 1 lt 002) High newt denities significantly reduced total abundances and densities of potentially competing tadpoles comprising the anuran guild

Intermediate denities of predators maximized the

function that maximized differences among treatments -( ~ ~ b l ~5 ) ~h~~~ species contributed to signif icant dif-ferences in relative biomass among predator treat-~ ~ ~ ments S 11olhrooki accounted for most of the anuran biomass harvested from predator-free tanks but guild biomass at greater predator densities was the product of more equitable contributions by metamorphs of all species

Biomasses of 5 I lolhroohi and B tcrrcstric were inversely related to newt density Biomasses of K sp l~~t loc~cpl~r r l t r were atand H grrrriovrr maximized intermediate densities of newts Biomass of H crrc-

c j f i~rwas positively related to newt density while biomass of H c~l1ryvoc~clivwas unaffected by salamander density The interplay between mean abundance and mean mass generating these species-specific patterns of biomass is summarized in Fig 4 For some species predator-mediated decreases in abundance were offset by predator-mediated increases in mean mass pro-ducing constant or increasing biomass despite lower abundances of metamorphs in the presence of predators Predator densities that maximized total guild biomas did not simultaneously maximize the biomass of each species (Fig 3 )

Predator density significantly influenced the survival total production of metamorph biomass in anuran g ~ ~ i l d s of all species except H clrrycoccllic(Table 6 ) Survival (Fig 3 Pi = 561 F lt OH Guild biomass was greater at moderate predator densities than in predator-free tanks despite the survival of greater total numbers of metamorphs in the absence of newts

Predator treatments also altered relative contributions of each species to guild biomass (Fig 3) A M A N O V A of vectors of angularly transformed relative biomass demonstrated that these differences were significant (Tables 4 and 5 ) A discriminant function analysis indicated that relative biomasses of S l lol-

hrooLi t i crrrcjfi~rand R c ~ ~ l i c ~ r ~ o c ~ c ~ ~ ~ l l r ~ l ~ rwere significantly correlated with scores for the discriminant

of S I lolhroohi K cpl7ct~oc~c~pl~trlrrand B rc~rrc1ctric

three species predominating in the predator-free tanks was inversely related to predator density Reductions in relative abundances of these species corresponded to low survival where newts were abundant In contrast survival of I f crricjfi~rwas very low in predator- free tanks but significantly greater and 5imilar in tanks containing two four or eight Y iiriticsccrzc Its relative abundance increased despite moderate but constant survival in the presence of newts because survival of other guild members declined precipitously with increasing predator density

Newt D e n s ~ t v ( n o tank

F I G 7 Relation between final abundance of all cen5used anuran (including o~erwintering R(it7ci)and initial density of predatory oroplrrl~cil~tririn each of four tanks Final densitie of uriving anurans declined significantly in response to experimental increaes in Votop l~t l~c i l r~~ i i~density

PETER J MORIN t c o l o g c i l h lonogl-aph Vol 5 3 K O 2

-H_ chrysoce l~s H gratiosa

- AY r 1504

I 0 2 4 8 Newt Denslty ( n o t a n k )

FIG3 Mean biomass of metamorph~ of each species censused from four replicates of each l V o t o ~ ~ i l r l ~ ( i l r ~ ~density treatment Biomass of each pecies is tacked to reflect the sequence of metamorphosis in each treatment (specie5 metamorphosing firt at the base specie metamorphosing last at the top) Numbers in parentheses indicate the mean number of metamorphs contributing to total guild biomass in each treatment

Survival of H grcrtiorcl was maximal at intermediate fined two groups of anuran species One group tended levels of newt density (Table 6 ) A combination of to survive best in the absence of h o top~r ~tr l r~ r and density-dependent and density-independent factors included S ~o lhrooh i R sp~c~tioc~c~pi t~Itr B tcrre-apparently caused its low survival in predator-free rri and H c~~voc~elisThe second group of species tanks H ~rtrtioctrlarvae grew slowly in predator-free survived best at moderate or high newt densities and tanks (Table 6) but persisted at high densities through included H c r l ic ( f i~and H lt~trrriocrr early October However floundering dying tadpoles Mean mass at metamorphosis increased significantly of H rtrriorr appeared at the water surface in late in response to increasing newt density for all species October coincident with a drop in water temperature except S I~ollrooki(Table 6 ) Marked depressions of to below 10degC All tadpoles of H yrtrtioctr died within mean mass at metamorphosis of H crrrcifir and H several days of this observation Overwintering tad- gttrriorr in predator-free tanks coincided with their poles of R ~ ~ ~ I ~ r ~ t i o c ~ e p i t ~ I ( ~ reduced survival in the absence of predators in the same tanks showed no signs of distress H gtntiotr tadpoles from less Frequency histograms of mass at metamorphosis in dense populations with more rapid growth and devel- populations of H crliciir (Fig 5 ) unambiguously il- opment metamorphosed long before the onset of de- lustrate that increases in mean mass at metamorphosis clining water temperature however few escaped pre- at greater predator densities correspond to increases dation at high densities of newts in minimum mean and maximum mass in each pop-

Patterns of survival in response to newt density de- ulation Differences among replicates simply reflect the

4T ~ B L E Newt density and error matrices of sums of squares and cross products for the MANOVA of relative biomass of anuran metamorphs Format a s in Table 2

Sums of squares or cross products

Sctrphioprrt Hy Itr Birf) Rtrt~tr H~lri Matrix (df) Species ho h~ook i c~ricifcr T T t i sn l~r t~oc~e~~~ir t r~ ~ ~ i r l i o s(i

Error ( 12) holhrooki H c r ~ i c ~ i ~ ~ B tr~rrsrric K (phet~ocrphrilri H fi~crriostr

Newt density (3) holh~ooli H crlicifi~r B t rrrr~ ~ tri R sphrt~ocephcrltr H ~rtrrio tr

June 1983 P R t L ) ITION 4 U D IULK I U GLILL) COMPOSI 11ON

T A R IE 5 Multivariate test criterion and summarq of discriminant function analysi for differences in relative biomas of anuran metamorph among newt density treatments Wilks Criterion = 01 1676 P 01 Discriminant function coefficients and correlation5 between discriminant function cores and transformed relative bioma of each species are given


error variance associated with responses to each treatment There was little or no overlap between distributions of mass at metamorphosis at low and high predator densities Nonoverlapping size di5tribution5 showed that increases in mean mass at metamorphosis could not arise from size-dependent truncations of a common underlying distribution that would exist in the untruncated form in the absence of predators Rather predator-mediated increases in mean mass occurred because survivors at high predator densities attained greater sizes than did most survivors at low predator densities

Ihere was a strong positive con-elation between mean and maximum mass measured for H ctrrcifir in each tank (1 = 99 P 0 1 ) demonstrating that mean mass accurately reflected differences in maximum mass among populations Mean and maximum mas5es were significantly and positively correlated in populations of all remaining species (Y I iolhroohi r = 94 P 01 K cp l c~ t loc~e l~ l (~ I ( r r = 80 P - 01 B rctrcc-rris I = 6 1 P O5 H c~coccli r = 84 P 01 H crrrtiotrr 1 = 93 P 0 1 ) underscoring the interspecific generality of this predator-mediated pattern These correlations demonstrated that differences in mean mass among treatments con-ehponded to differences in maximum mass and shifts in entire size distributions in response to predation

Mean developmental periods of three species ( K cpllc~troc~c~pl~rrltr and H grrrriosrr) wereH c~1rjcoc~clis shortened significantly at high salamander densitie (Table 6 ) Larval periods of the three remaining species exhibited a similar though nonsignificant trend

Mean growth rates of S ~ o l h t o o h iwere consistently high and varied little in response to predator density (Table 6 ) Mean growth rates of H crricjfir were con- si5tently less than growth rates of other species at each level of 5alamander density Mean growth rates of If crtrtiotr ranged from low values in the predator-free tanks to the highest rates observed at the highest newt densities Mean growth rates of all species except Stir-pliopr indicated that tadpoles metamorphosed at larger sizes and after shorter periods of development as densitie5 of horopzrl~rrltnrr increased MANOVAs

0 1 2 LoglO Mean Densty (no t a n k )

Flc 4 Relation between the logarithm of mean metamorph abundance (number per tank) and the logarithm of mean metamorph mass (in milligrams) for each anuran species at each level of l V o r o l ~ h i l ~ c ~ l r ~ ~ ~ ~ density Trajectories for each species originate at mean values for the predator-free tanks and proceed through respective values for denities of two four and eight newts per tank Dashed lines indicate combination of abundance and mean mass qielding equal values of biomass Rs = Ktrt~tr pl7c~t7oc~c~1hciIc1 =H Y Hyltr grci- tioctr YI = S~~trpl~iojprrs Hyltr clfqco- holhrooXi Hc11 = cclic B1 = Rl~t i j orrc~ctric Hc = Hyltr crlrcifir

for mean mass at metamorphosis mean larval period duration and mean growth rate indicated that predator-mediated differences described above were significant for all species except S I lolhrooki Thus univariate trends suggested in Table 6 were confirmed by more conservative multivariate tests

Newt density also influenced the likelihood of overwintering by tadpoles of Rrrtrtr c p l ~ c t i o c ~ c ~ ~ ~ r t r I ( ~ (Table 7 ) Rrrt~tr p ~ c ~ t i o c ~ c ~ p ~ t ~ I t ~ was the only species with overwintering tadpoles All other species completed development before the termination of the experiment in November Overwintering K(rt7rr tadpoles were restricted to tanks containing two or no newts Over- wintering tadpoles were more abundant and of smaller mean mass where newts were absent than in tanks containing two newts All Rnticr populations in predator-free tanks contained some overwintering in-dividuals while only half the populations in tanks containing two newts had overwintering tadpoles Ap- proximately 1 mo elapsed between the collection of the last Rtrtitr metamorph and the termination of the

128 PETER J hlORlN tltcological M o n o g r a ~ h Vol 5 3 No 2

6 T ~ B L E Means of population means for statistics measured at each level of newt density The ANOVA column indicates approximate levels of significance from univariate tests for effects of newt density on population means The MANOVA column gives conservative levels of significance ( P ) for simultaneous differences among treatments in mean mass at tall resorption length of larval period and mean growth rate of each species jV indicates the number of populations producing survivors

l V r ~ i o ~ ~ ~ t l ~ c i l r ~ ~ ~ ~ sdensit) (no per tank) -

P -

Species Variable 0 2 4 8 ANOVA MANOVA

I~olbrooki Survival ( r i1 Mass (mg) Larval period ( d ) Growth rate (mgd) v Survival ( r i l Mass (mg) Larval period ( d ) Growth rate (mgd) lc

K pll(~~roc~c~plc~ltr Survival ( C )

Ma (mg) Larval p e ~ i o d (dl Growth rate (mgld) V

R cJrrcJtric Survival ( C )

Mass (mg) Larval period ( d ) Growth rate (mgd) lc

H ltrciiio11 Survival (7) Mass (mg) Larval period ( d ) Growth rate (mgd) lc

H (~ryocrlic Survival (I Mass (mg) Larval period ( d ) Growth rate ( m u d ) V

experiment At lower newt densitie5 the distribution species Canonical correlation analyse5 summarized in of larval period duration for Kr~itr would have been Table 8 describe correlations between mean growth bimodal with peaks of metamorphosis in 1980 and rates of each species and densities of potential com-1981 separated by I winter hiatus in development petitors meawred in each tank For example a sig-

nificant canonical correlation indicated that mean growth rates of H c r r i c ~ i i l r in each tank were positively correlated with values of a canonical variable

Mean growth rates of most species were negatively which was a linear combination of the final abun- correlated with final censused abundances of surviving dances of potential competitors In turn values of the

TABLE7 density on overwintering tadpole of Ktr~~tr Surviving Kcitlci columnsEffects of corophih(i~~i~ cpl~c~t~oc~rp~i~Ici indicate mean numbers of metamorph and tadpole censued from each tank summed through the termination of the experiment Entries in tadpole statistic columns refer only to cohort of overwintering tadpoles Column headed V denote the number of K(~tlripopulations included in each tabulation

C I ~ ~ I ~ I - Surviving Kcltlcl p~c~t~occ~p~ciIci Tadpole statistics -rllc~lnl~r

denvt) Metamorphs Tadpoles Mean mass (no per tank) per tank per tank V No per tank (mg)

PRF I IATlOh I U D I U L K I U GL111-D COhlPOSITIOU 119

8 N E W T S-- d j a-

_ - 4 N E W T S

a

2 N E W T S

0 N E W T S

i 160 260 360 4io Mass (mg)

FIG 5 Frequency histogram of mas at tail rewrpt ion in all populations of lyci crtrcifi~r that produced meta-morph One population in the predator-free treatment produced no metumorph

canonical variable were correlated strongly and negatively with densities of ~urviving K c - j ~ l i c ~ t ~ o c ~ c p l ~ ( i I t ~ and S 11olhrooLi Values of the same canonical variable were correlated moderately and negatively with densities of surviving B tcrrcrri and H clrrgtocelic The latter two species were added to the community after K c j~ l rc~r~oc~c~p l ( i c iand S ~ o l h r o o X iand potentially interacted with H crric(r for less time The canonical variable was not negatively correlated with intraspecific density despite considerable variation in density of H crrrcjfgtramong tanks Thus mean growth rates of H crrrcifgtr were negatively correlated with final densities of four species These four species also survived best in the same predator-free tanks where H crricijkr survived poorly

Similarly mean growth rates of H grtrtiocer were correlated negatively with den~i t i es of surviving C I ~ o l h ~ ~ o o L i B ~t~rrtltriand HR ~ ~ l ~ e r ~ o c ~ c ~ j ~ l t c r l r r clrrcocclic As with H crrrcifi~rmean growth rates of H lt~lr~tiotrrwere not significantly negatively cor-related with final intraspecific density Mean growth rates of both H ltltirtiocc~and H crrrcifgtr were negatively correlated with densities of ar~urans that predominated in the same predator-free tanks from which If grtrrioci and H ct~icfirwere essentially excluded

Growth rates of the four remaining species were negativel correlated with both intraspecific and interspecific densities in contrast to the preceding pattern Mean gl-0~4th rates of 1rolhrooLi and R plrcj-t~oc ep l~c l l t rwere highly cnrrel~ted with respective intraspecific densities and wet-e also moderately correlated with densities of B rc~rrc~ctricor H clrtocc~li Mean growth rates of H cltocc~licwere correlated strongly with densities of R cp l~e~~oc~cp l l c i l r rand S IzolhrooLi and more moderately correlated with intriispecific densit) Mean growth rates of B tcrretrti

T A H Ic 8 Correlations between final densitie of anurans and canonical variable best correlated with mean growth rates of each pecie Entries in each column are product-moment correlations between cenued nnuran densities and value of cinonical variable best correlated with growth [-ate of species heading each column V indicate the n ~ t m b e r of surcicing population tvailable for each anal) is indicates cor-[-elation ignificant at the Oi level

5 lio)rooLi H crrrcitijr K c[~ilc~tioc~cli(iI(i B fcrrcfri H c~ilr~oc~clit H qri~iioo

C~tnonicul correlation Significance

-

Canonical var-iable best corr-eluted with mean gr-owth rates of

-

Correlation between anur-an densit) and canonical variable 8 8 - Xi 85- 93 -

35 37 76- 59 9 3 88- 69 - 88

6 1 - - i2ii- 6 2

58- 60 1 7 73

2X 2 3 21 I 0

91 89 88 96 0016 0089 0595 0001

15 16 13 16

1- n i I 9 Differential culnerabilitq of tadpole to wlamander predation Combinations of nnurun species offered to predator in eparate experiments are lited in the column headed prey Fr-equencq indicate the n ~ t m b e r of each species consumed in V replicates of each experiment Deciationj from equal predation on uv~ilahle preq specie a re ev~luated with the accompunqing 4 statitics indicates significant deciation (P 01) hs indicates not significant ( P - US)

PIey F I equencq Y df P

(predator - ~iriclcgtcccr~c)lVoro~)rri~cil~t~~~ R c~ic~rioc~c~i~iclo P fric~ricirci 17 I 27 1 0 H cr~rciir R c ~ ~ i ~ ~ r i o i ~ ~ ~ ~ ~ i ~ i ~ i 73 IS H crirtir 96 R c ~ ~ i ~ ~ r i o c ~ ~ ~ ~ ~ o ~ i 30 1 0 P rri~riiit(i 35 R ~ioioccl~iiciIii 74 l I0 I5 B fcrrctiris C iolhrooLi 63 R )ii(~~ot~(~)ii(iI(i 27 1 5 b icrrcrric 82 R c[~lic~ri~~~~c~[~ioI(i SiohrooLi 12 132 IS B trrrcctrir iiolhrooLi 97 87 15 H cir coct~li H lt~riiiioii H cititlcrorti 95 129 105 IS R c ~ ~ i c ~ r i i ~ c ~ ~ ~ ~ i i i I i i l I0 I5 f yr(iiioo 135 R ~ i i c r i oc c l~ i c i i c i H circocc~ic 80 86 10 6 icrrciri H yrcirioo 78 89 15 B circ~rc~ic~ir H i~citiocci 208 218 IS

(predator Arti)c i o r i ( i iiqritii(rri) R ~ ~ ~ i r c ~ ~ r i ) c ~ c ~ ~ ~ i ~ i ~ i I t crirc(fir P fri t~ri(i t i i 3 3 4 32 10 R ~ ~ i c ~ r i i ~ t ~ c ~ ~ i i ( i i i Btcrrcctri S itolh~ooLi 9 33 43 8 R c~icriocci~iidii 3 0 7 H yrcirioci 37

tended to be negatively correlated with den5ities of captured and held in the buccal cavit) for 5everal sec- specie that survived best in the absence of newts but onds Rejected Kcrriti tadpoles usually swam away from this trend WI marginally nonsignificant Mean grouth these encounters with At~lycrotutr and were appar-rates of these four species were not negatively corre- ently i~nharmed b) aborted attacks lated with densities of H crricifit or f I yrciriocri Over 111 tanks mean growth rates of all species were negatively correlated only with final densities of the species Experimental manipulation5 of p r e d ~ ~ t o r density in that survived best in predator-free tiinks tank communitie5 dramiitically altered the final species

composition of larval-anuriin guilds Direct negative effects of predators on the persistence to metamorphosis of four species caused part of the difference in

Anuran species differed in vulnerability to predation g ~ ~ i i dcomposition among treatments Reduced relative b) VoropIrrlrrilrrri and Ar)ll~trot)lcr (Table 9 ) When abundances of S IlolhrooXi R cplietloc~cpltrItr B ter- confronted with various combinations of equally abun- lctrric and f I c~lrcocclic it high newt densities were dant S Irolhrooli K c ~ l i c ~ t i o c ~ c ~ l ~ c i l t i the direct results of increased consumption by pred- and R tczrrcJc-~ r i c consistently consumed ators However enhanced survival and relative abun- l otoplrtl~cilt~~ri~ more 5 Iiolh~~i~oXi dances of H ctrcifi~r and H ltrrcitio~tr in the presence than either R trtrc~ttrit or K cphoioc~rplr- trltr and consumed more R rc~rrcgttrritthan R cplrctlo- of newts could not be reconciled with direct negative cc~litrltr This pattern held whether the three species effects of predators Similarly the nearly complete ex- were presented simultaneously to ~oropl~tlrcrlrt~ir~c clusion of H crric(ir and f I grcirioctr from predator- or in pairwise combinations Feeding trials using other free tanks could not be attributed to predation Ample prey indicaled that various hylid tadpoles (Hyltr and evidence indicated that predator-mediated shifts in an- Pccrrdtrcti spp) did not differ in vulnerability to Yo- ilran dominance were consequences of inverse rela- topllrlrtrl~~~ric tions between intensities of salamander predation and predation Similarly hylid tiidpoles did not differ from Rtitltr c~lic~~~occpiciltr competition among tadpoles and complementary in- in vulnerability In general 2oro1irtlicil1~lrit selectively preyed on tad- terspecitic differences among iinurans in the ability to poles of Scciplrioprr lolhrooli and Blifi) trtrcttri~ iln- exploit communities successfully under extremes of der most circumstance5 and indiscriminately preyed competition and predation First I discuss the statis- on tadpoles of other species as they were encountered tical and biological evidence for predatol--mediated

Patterns of vulnerability to predation b) Al~h)ro- competition among larval anuran5 1 then consider the ~ ~ r r relevance of predator-mediated competition amongwere similar to patterns described for ~oroplirlltil- r)rr i (Table 9) Unlike ~oroplitllt~ltirit Atih~crottlti anurans to emerging theories of cornmunit) structure discriminated against tiidpoles of R c~hoioceplicilci Finally I describe patterns of Iiirviil development that when they u e r e presented with Ptrrrtlcic~ri and fIgtltr may allow some species to tolerate competition and tadpoles 1 observed that Attll~~totici frequently re- I suggest constraints making other anurans either jected tadpoles of Rtirlci vplic~rloc~c~~licrIci predator-dependent or fugitive species after they were

June 19X i PKFL) 1I ION 1 U I l - ULK-U CrLIIdIl COIlPOITION I 3 l

Ev~dence presented above for an inverse relation between salamander dens~ty and competition among tadpoles is indirect but compelling A rich literature documents that competition among tadpoles is density dependent and illustrates that increased conipetition depresses mean niass at metamorphosis while pro- longing periods of larval development (Brockelnian 1969 Wilbur 1972 1976 1977t1 h 1980 1982 Wilbur and Collins 1973 DeBenedictis 1974 Wiltshire and Bull 1977 Smith-Gill and Gill 1978 Steinwascher 1979otr h Travis 1980) 1 inferred that intriispecific differences in niass at metamorphosis larval period and growth rate among predator treatments were manifes- tations of predator-mediated differences in the intensity of anuran competition Density-dependent competition among tiidpoles is expected to be less intense where fewer txipoles e x a p e predation and survive t o compete The interpretation of reduced intensities of anuran competition at high salamander densities is consitent with the invere relationship found between densities of surviving potentially competing tadpoles and predatol-a

Hypothe5es that d o not invoke variable intensities of conipetition among tadpoles cannot adequately ac- count for observed predator-mediated differences in anusan growth and development For example increases in mean mass at metamorphosis might result from selective predation by newts on smaller tadpoles and increasingly extreme truncations of potentially identical size distributions The crucial difference between t h ~ s s~ze-dependent predlt~on h ~ p o t h e s ~ s and the predation-dependent competition hypothesis is that size-dependent truncations will result in broad overlap among pop~~la t ions in frequent) di5tributions of mass and will not increae maximum niass In contrast if tadpoles are released from competition at high predator densities mo5t individuals should grow to a larger size than their counterparts at low predator densities Distribution of H crlrcifigtrmass at low and high pr-edator densities barely overlapped Truncation of I crrrcifir size distributions in the predatol--free tanks could not produce shifts in maximum ma observed it high newt densities Thus increases in mean mass were assumed to reflect a PI-edator-mediated release from intense competition in the predator-free tanks I analyzed population means instead of maxima because means contained information on all survivors while maxima reflected the per-formiince of single individu-

rigorously identified by negative correlations between densities of surviving iinurans (potential competitors) and mean growth rates of species responding to competition Tadpole densities were not directly manipulated b) experiment but predation and other factors generated considel-able variation in final anuran densities among tanks Relative strengths of correlation5 between intra- and interspecific densities and mean growth rates were iissumed to correspond to relative strengths of competition exerted by each species Strong negative correlation5 indicated high covariance between densities and growth rates

The negative correlation between mean growth rates of F crrrcifir and F ltlrtrtiocrr and densities of surviving S IrolhrooXi R cpietroc~epic~l(iR tcrrc2c tric and H crcocclic indicates that these latter four species were the source of intense competition contributing to the demise of H crricifi~rand I yrtrtiocci in predator-free tiinks The fact that mean growth rates of H crriciJrr and H ~rtrriocti were not negatively correlated with their respective intraspecific densities suggests negligable intrispecific competition in these two 5pecies Mean growth rates of the four species that survived best in predator-free tanks were nega-tivel) correlated both with respective intraspecific densities and densitie of other members of this group but were not correlated with denities of I crrrcjOt or H yttiriovci I suggest that the two groups of species distinguished by high survival or near exclusion in the absence of newts correspond to competitivel) superior specie5 (S iroIl~r~ooXi R rcrrr~crricR c ~ i c ~ t i o c ~ c ~ ~ ~ i r ( r ~ r

and I civoc~lic)capable of high s~~rvivi i l or ~ L I C - cessful development despite high tadpole densities and competitively inferior species (H crricjc~r and I f yrcr-tioctr) incapable of persisting to metamorphosis at high tadpole densities The assumption of poor competitive ability of I crrrcjfi~rand I ltlrtrtioctr was 5~1pported b) the absence of negative correlations between their den5ities and growth rates of any anurans b their near exclusion from predator-free tanks where conipetition appeared to be most intense and b) their dra- matically enhanced growth where predators reduced the densities of other anurans

Mean growth rates of H c~lr~~coc~cliaand H yrcr-

tiostr were negatively correlated with densities of S iolhrooXi even though S IlolhrooXi metamorphosed from all tanks several weeks before hatchlings of these late-breeding HJlci s l ~ pwere added to the communities Thi5 relation might be exp1lined as a stitistical artifiict but it might also reflect in intriguing effect of earl-breeding S i~olhrooXi on the growth of late-

als depended on sample size and thus cons t i t~~ted breeding species rather biased measures of average Iiirval ittributes The statistical artifact hypothesis invokes the posi- Means accurately reflected differences in maxima tive correlation between tinal densities of S IiolhrooXi among population given the positive correlations be- and K cp~oroc~rpiriltrThis hypothesis also recog- tween means and maxima nizes that werelarval periods of R cpirc~t~oc~cpitrItr

Becti~e conipetltlon imong ttdpole 1s density de- much longer than those of S IrolhrooAi resulting in pendent I ~nferred that competing specles could be extensive temporal overlap between K cplctrocc~pl-

irlir and late-breeding species Negative correlations of I ) I ~ c I I ~ ~ ~ c I I I ~ I I ~ ~between densities R and mean

growth rates of H qtcitioc~and H c~llroc~olisuggest that K cplct~oc~rplciltr ma) have competed with late- breeding species In this case densities of S Irol-hiooi would be negatively correlated with mean grouth rates of late-breeding f f I ( ~simply because densities of T I~olhrooXiand R p l rc t ioc~r ~ ~c~Ic~ were positively correlated A more 5peculative alternate hypothesis is that earl~ybreeding species may have a lin-gering impact on re5ource availability subsequently influencing the growth of species exploiting a pond later in the same year (Seale 1980) Such priority effects are an interesting topic for future I-esearch

Previous st~ldies have demonstrated that lal-val density may have nonadditive and nonlinear effect5 on intensity of competition among tadpoles (Wilbur 1972 1976 197711 h Wilbur and Collins 1973 Smith-Gill and Gill 1978) Such effects have been cited as evidence for the ~ntdequacy of simple a d d ~ t ~ v e linear models of dens~t)-dependent compet~tion but this In-

Ecologicil Monograph Vol 5N o 2

ing tadpoles and other potential competitors were most abundant in predator-free tanks where reduced growth rates prevented Krrtitr from attaining minimum meta- morphic size in a single season At high predator densities R(itlci exhibited enhanced growth and easily at- t a ined min imum m e t a m o r p h i c s i z e P r e d a t i o n apparentl) influenced competition among Kr~trtr tad-poles to the extent that cohorts with initially similar density required different number5 of years to complete larval development

Alternate responses of anuran guild composition to experimentally varied intensities of predation suppo~-t theories of community organization that invoke com- plementar) effects of inversely varying intensities of competition and predation (Brooks and Dodson 1965 Paine 1966 Connell 1975 Menge and Sutherland 1976 Lubchenco 1978) High premetamorphic mortalit) of two species ( H grrrtiotrr and H crrrcjfir) appx-ently resulted from intense competition in the absence of

terpretitlon hts been questioned by other s ( P o m e ~ ~ t n t z predators These competitive exclusions would not be 1981 Vandermeer 1981 ) In tank communities canonical correlation between mean growth rate and simple lineal- functions of anuran abundances ranged between 88 and 96 These high col-relations indicated that simple lineal- functions of competitor densities described much of the variance in mean gl-outh rates among tanks M) ana lys~sd ~ dnot lnclude tests to1 the presence of nonlinear o r nonadditive effects but even if such effects occurred simple linear functions of final anuran abundances satisfactorily accounted for most of the observed variation in inferred competition

Canonical correlation analyses also suggested that interspecific competition among tadpoles was strongly asymmetrical Growth rates of H c~rrrc~(fitand H ~ r t i -tioci were 5trongl) depressed by high abundances of four competitively superior species but growth rates of the foul- competitively superior species appeared unrelated to abundances of I f c~rrrc~jfer and I f grrl- t io c~ Similar instances of ~ l n e q ~ l a l competition among anurans have been desc~ibed in other studie (Wilt-shire and Bull 1977 Wilbur 1982) The apparent absence of intraspecific density dependence in growth rates of I f c~rrrc~(fgtr and I f grrrrioctr also suggests that interspecific competition may outweigh the effects of intraspecific competition in weak competitors

The restriction of overwintering tadpoles of Krltltl pl~c~t~oc~e~li tr l tr ofto tanks containing low densitie predators and high densities of tadpoles also support5 predator-mediated competition Collins ( 1979) suggested that the likelihood of overwintering by Ktrtltr c~trtechciirt~rrtadpoles may be influenced b) climate timing of breeding and resource availability The fir5t two factors did not vary among tank communities leaving predator-mediated differences in resource availability as a likely source of variation in overwintering by Rirtrir plret~ocepIztrltr tadpole5 Overwinter-

predicted by community theories based only on predation (e g Heyer et al 1975 for anurans Dodson 1974 Zal-et 1980 for aquatic invertebrates) and illus- trate that competition can effectively structure g~lilds when predators are absent

Predator-mediated change in anul-an guild composition provided an equall) important demonstration of predations efficacy in structuring g~lilds of vertebrate pre) Patterns of anuran relative abundance arising from competition in predator-free tanks bore little resemblance to patterns observed at moderate to high predator densities Complete shifts in the identity of dominant species among treatments emphasized that persistence in the presence of predators was unrelated to competitive ability It was unlikely that interspecitic competition determined predations influence on species composition (Wilbur 1980 1982) because competitively superior species failed to maintain their dominance at high predator densities On the contrar) competitions impact on guild composition was con- ditional on the intensity of predation on tadpoles Pre- dation on hatchling tadpoles ameliorated interspecific competition b) reducing the densities of anurans that survived to compete Further at least two competitively superior species (5 Izoll~rooLiand B tttrecttic) were more vulnerable to predators than were competitivel) inferior species Differentially severe predation on competitively dominant species can selectively re- d ~ ~ c etheir abundance in the guild and simultaneously release inferior competitors from competitive limitation This mechanisms importance has been previously recognized and emphasized in other communities (Brooks and Dodson 1965 Lubchenco 1978) since it can shift dominance from predator-vulnerable competitively superior species in predator-free habitats to less vulnerable competitively inferior species at high

133 June 1983 P R E D A T I O N A N D A N U R A N GUIL2D COMPOSI I ION

intensities o f predation However poor survival o f apparently unpreferred Ktrnr~ tadpoles at high predator densities suggests that reduced vulnerability to predators provides an imperfect refuge at best

Inversely varying intensities o f competition and predation have been suggested to control species com-position in various communities o f sessile space-limited organisms (Paine 1966 1980 Harper 1969 Connell 1975 Menge and Sutherland 1976 Lubchenco 1978) Such communities differ greatly in taxonomic com-position spatial complexity and general habitat fea- tures from temporary-pond communities and their experimental analogs Convergent responses o f such disparate assemblages to manipulations o f predator abundance provide strong support for the generality o f predations impact on specie composition over a variety o f habitats and organisms The structuring o f anuran guilds by predation demonstrates that such e f - fects are not limited to communities dominated by sessile species or invertebrates

Previous studies have suggested several factors that may determine the relative impact o f competition and predation on species composition in a given locale In intertidal communities prey often tolerate harsh en- vironments better than do predators (Connell 1961 1975 Menge and Sutherland 1976) Where environ- mental stress more readily excludes predators than prey competition may predictably influence prey species composition Predator limitation by environ- mental stress may be less important in pond commu-nities Adult Notopl~r~trlt~tr( withstandeasily the physiological stress imposed by the drying o f temporary ponds while their larval amphibian prey are excluded from ponds by desiccation Adult Noropllthtrl- tnrs retire beneath moist litter in the bottom o f drying ponds while larval amphibians must metamorphose or perish under identically harsh conditions When ponds refill adult newts persist at predrought densities and pose an unmoderated threat to larval amphibians entering the community via renewed anuran breeding The apparent harshness o f ephemeral ponds creates a refuge for tadpoles from predatory fish (Heyer et al 1975 Wilbur 1980) but this reduction in the intensity o f predation is relative rather than absolute because predatory salamanders and insects abound in fish-free ephemeral ponds (Walters 1975)

The classic proposition o f Hairston et al ( 1960) and its extension by Menge and Sutherland (1976) suggests that abundances of species in all but the highest trophic levels are potentially limited below carrying ca- pacity by predators It i s consistent to expect that competition occurs among predatory salamanders oc- cupying the top trophic levels in temporary ponds (Wilbur 1973 Morin 1983) or among predatory star-fish in intertidal communities (Menge 19721 while predation controls the species composition o f their re-spective prey In qualitative agreement with Menge and Sutherland (1976) the impact o f competition on

anuran guild composition clearly depended on whether anurans constituted the top trophic level (predator- free tanks) or a subordinate trophic level (tanks containing salamanders) Factors favoring dense populations o f ~Votopltl~trlt~~~ pondsin natural remain speculative Noropl~rl~trlt~~rc density in natural ponds may not be predictable from in situ pond conditions and immigration may primarily influence the local abundance o f adult newts (Gill 1978 1979) but the potential importance o f newts as keystone predators is unambiguous where populations become estab-lished

The specific impact o f predators on anuran guild composition depended on the density and species o f salamander Notophrhtrln~rr~ shifted dominance from competitively superior to competitively inferior anurans without altering anuran diversity within tanks (Morin 1982) In contrast Atzhysronltr deleted the anuran guild obviously reducing anuran diversity Sim- ilar events are frequently observed in two ponds in the Sandhills where A tiyrinrrtl regularly occurs Many anurans oviposit in these ponds but large tadpoles and metamorphs are seldom subsequently collected Dif- ferent short-term vulnerabilities o f anurans to Atr11~yc- ton111 in the laboratory (Table 9) did not provide an effective refuge from this rapidly growing predator This suggests that Anlhysto~~ltr can overcome their apparent dislike for Ktrnt~ tadpoles when other preferred prey are absent Anlhy~tot~lrr attain a much larger size than Notol~lltltrln~~l (10-20 vs 3-4 g for Noro~~hrhtrl- 215)and most tadpoles with the occasional excep- tion o f H grtrtioctr cannot attain a refuge in body size from thi large predator Four Anlhyttotltr per tank had a far greater per capita impact on tadpole guild composition than four Notophtll~lnlci This result clearly demonstrated that top predators at least among salamander species may differ drastically in their effects on community organization In the Sandhills the small number o f ponds successfully exploited by Anl- hyston~tr probably limits its global importance I t s po-tentially extreme impact i s moderated further because drought and predation by Noro~l~r~~lllr(lt on eggs and small larvae (Morin 1983) readily excludes Anlhytro- t ~ l t rfrom natural ponds

Theories relating predation to community organization often specify relations between predation intensity and univariate measures o f species composition especially species richness and species diversity (Paine 1966 Addicott 1974 Lubchenco 1978) Other theories are framed in terms o f predator-mediated changes in the dominance or relative abundance o f prey species (Brooks and Dodson 1965 Zaret 1980) without specific reference to species richness or diversity Different measures o f community composition used in previous studies complicate comparisons o f predator effects among communiites Specifically alternate measures o f species composition (dominance diversity and richness) differ substantially in their resolution of dif-

I 34 PETER J

ferences among communities In tank communities manipulations of predators reversed patterns of anuran relative abundance but neither species diversity nor species richness differed among treatments (Morin 1981)

This failure illustrates one serious shortcoming of univariate indices of community structure (also see Hurlbert 1971) Species diversity and species richness obscure simple but important shifts in dominance among species In contrast vectors of relative abundance preserve this i~formation by linking relative abundance to the identity of each species Multivariate analyses of vectors of relative abundance remain sensitive to shifts in dominance of species regardless of changes in species number or diversity Multivariate analyses are also sensitive to correlated changes in relative abundances of two or more species Thus a single MANOVA can detect complex differences in the composition of entire multispecies assemblages that might easily be missed by univariate tests and mea-sures

Interspecific differences in larval growth and development illustrated various mechanisms that permit anuran persistence under high intensities of competition These mechanisms include low sensitivity to tadpole density plastic size at metamorphosis and the ability to overwinter as larvae under conditions unfavorable to rapid growth

S(trplliolxi 11olhtooXi survived well at high com-petitor densities because of its unparalleled competitive ability illustrated by rapid growth and development that responded only slightly to competi tor densities This pattern was consistent with a high efficiency of resource utilization and low sensitivity to competitor density and contrasted with the great range of density-dependent growth rates of the other five species Rapid larval growth is a hallmark of this genus (Richmond 1947 Bragg 1965) and represents an ad- aptation for exploiting highly ephemeral but produc- tive ponds (Wassersug 1975 Wilbur 1980) Intense se- lection for rapid growth in ephemeral ponds might effectively preadapt Sc~rrl~lziopci~as a superior competitor by selecting for high efficiencie of re5ource utilization However the high vulnerability of Sc~rr-plliopri to predators suggests its susceptibility to exclusion from natural ponds by salamanders Superior competitive ability and low sensitivity to density argue against competitive exclusion of Sc~rrl~llioprrfrom natural ponds by any of the anurans con5idered in this study

Hlti c~llrocc~lic and Rlltfi) tetrotri exhibited greater sensitivity to tadpole density than 5 IlolhrooXi yet both did metamorphose in number from predator-free tanks where competition was intense In these species platic size at metamorphosis permitted successful

metamorphosis at reduced size where competition reduced growth rates These species retained the ability to capitalize on resources and metamorphose at increased size where predators were abundant and competition was less intense but also appeared quite vulnerable to predators

Rri~lrr c~llrnoceplltrltr displayed density-dependent growth and development coupled with a relatively large minimum size at metamorphosis Overwintering in the tadpole stage permitted persistence in tanks where competition was so intense that larvae could not attain minimum size for metamorphosis in a single season of growth In many years natural ponds exploited by Rtrnc~ pllrnocc~plltrItr dry in late summer or autumn illustrating a major shortcoming of overwintering Overwintering is clearly dangerous unless tadpoles overwinter in more permanent ponds or unless stochastic events such as occasional autumn storms prevent ponds from drying

Hylrr ctricii~r and Hyltr ~ r r ~ r i o s t r may both rely on predators to reduce competition among tadpoles and ensure their larval success In the Sandhills spring- breeding H crrrc(fgtrand summer-breeding H ytrr-rioc~are the most abundant anurans metamorphosing from Grassy Pond which contains high densities of salamanders (Table I R a l ~ s ~ ~ l ~ c ~ ~ ~ o c ~ c ~ ~ ~ l ~ r ~ I t ~) consis-tently fails to metamo~phose from the same pond despite the annual input of roughly equivalent numbers of eggs These observations are consistent with patterns in tank communities Hyla ctlic~(fir is broadly di5tributed among natural ponds perhaps reflecting the distribution of Norop11rl1rrl1~1rrc cwr-among ponds H cjigtmight also persist as a fugitive species in predator-free ponds if other superior competitors are absent by chance Comparison of growth rates among species suggests that H crlicifgtt has a low efficiency of resource utilization even under reduced competitor density

H yrtrtior~like R ~ p l z c ~ n o c ~ c ~ ~ ~ I ~ t ~ I r ~ had a large minimum size at metamolphosis that was not readily attained at high intensities of competition However it grew rapidly in tanks where predators were abundant and readily attained a refuge in size from Noto~~ l z t l~ r r l -t~1rrsThe apparent failure of H grtrtio~tr to overwinter corresponds to the observation that small stunted tadpoles of this species can be collected in some natural ponds in autumn but are never subsequently collected from the same ponds in winter or early spring H y~rlriocilike H crlrcjfgtr is predicted to be most successful in natural ponds where predators limit abundance of superior competitors

4lthough salamanders reduce the probability that hatchlings of most species will survive to metamorphose these predators also appear to enhance the fitness of the tadpoles that survive Fitness is enhanced because mass at metamorphosis is negatively corre-lated with tadpole density and tadpole density is a decreasing function of predator density Large size at

June 1981 PRI+IgtAI ION 4 N I l 4 N U K 4 N GUILD COlPOITION I35

metamorphosis may be advantageous because large metamorphs have an initial growth advantage that may lead to early attainment of reproductive size (Wilbur 1972) Large size may also enhance reproductive out- put given the positive correlation between adult size and fecundity in anurans (Salthe and Duellm~tn 1973 Berven 1981 ) Furthermore abbreviated periods of larval development in the presence of predators make mortality by drought less likely since tadpoles spend less time in unpredictktbly ephemeral ponds Density dependence tnd the pltst~c~ty of tmph~blan development permit competition and predation to influence patterns of offspring number and quality above and beyond the evolutionary constraints imposed by the inverse relation between the initial size and number of propagules (Wilbur 1 9 7 7 ~ ) In addition to controlling itnuran guild composition competition and predation interact to generate intraspecific differences in offspring number and quality within generations that rival the magnitude of interspecific differences in life history parameters ascribed to long-term evolutionary pressures

Ihi paper I haed on a dissertation subm~tted in partial fulfillment of the requirement for the PhD at Duke Uni- verity 1 thank the memhers of mq doctoral committee Jo- eph R B a ~ l e ) Donald S Burdick Daniel A Livingstone H Freder~k Nijhout and Henry M Wilbur for t h e ~ r asis- tance in this endeavor Manq individuals helped by collecting frogs or main ta~n~ng the experiments including Joe Bailey Ilianne Campbell Chi-i Chanlbers SaIIgt Gladtone Sara lev^ Jack Longino Marsha Morin Alec Motten Kurt

Steinuascher Dave S m ~ t h And) Taqlor Joe Iravi Sara Via and Henr) Wilbur Dougla E Gill made man) helpful editorial ~ ~ g g e t i o n and comments hq three anon)mous re- i iewers significantl) ~mproved an earlier draft of thi paper

l his reearch was supported bq National Science Foun- dation Grunts DEB 7682620 and DEB 791 1539 to H M Wil- bur Computing fund were provided bq the Department of Zoologq Duke Universit) 1 was supported by a James B Duke Fellowship I teaching asitantship from the Depart- ment of Zoolog) and National Science Foundation Grant DEB 791 1539 to H M Wilhur

L I T E R A TL R E C I T L D

Addicott J F 1974 Predation and prey community true-ture an experimental tudq of the effect of moquito larvae on the protozoan communities of pitcher plants Ecologq 55475492

Berven K A 1981 Mate choice in the h o o d frog K(itici ~ ( r t i c Evolution 35707-722 (1

Bragg 4 N 196i Gnome of the night Univenitq of P e n n s ~ l v a n ~ a Pennqlvania U S 4 h e Ph~lade lph~a

Brockelman W Y 1969 4 n analys~s of denit effect and predation in Btrto ~ttt1c~ric~trt1tic 50tadpole Ecology 63 2-644

Brook J L and S 1 Dodwn 1965 Predation bod) size and compos~tion of plankton Science 15028-35

Brown J H and G A Lieherman 1973 Resource uti- I~rat ion and coexistence of eed-eating desert rodent in and dune habitats Ecologq 54788-797

Calef G W 1971 Natural mortality of tadpole in a pop-ulation of K(itici citirorci Ecolog) 54741-7iX

Caswell H 1978 Predator mediated coexistence a non-equilih~ium model 4merican Naturalit 112 127-154

Christman S P and LA R Fwnz 1973 Feeding habits of the striped newt Vor~l)l~rl~(iltrirc pcrrri(irtr Journal of Herpetology 7 133-135

Cod) M LA 1974 Competition and the structure of bird communities Princeton Universitq Pres Princeton Neh Jersey USA

Collin J P 1979 Intrapopulation variation in the bod) size at metamorphosis and timing of metamorphosi in the bullfrog R(itlci c~circhci(iti(i Ecologq 60738-749

Connell J H 1961 Effects of competition predation by 711(ii I~il)illirc and other factor on natural populations of the barnacle Bcilcinrr h~ilrit~oitlc~ Ecological Monograph 3161-104

1975 Some mechanisms producing tructure In natural communitie a model and ome evidence from field experiments Page 4hW9U tt1 M LA Cod) and J Din-mond ed i ton Ecology and evolution of communit~e Belknap Pres Cambridge Musachusetts USA

Duqton P K 1971 Competition disturbance and communit) organIration the provision and subsequent utili-zation of space in a rock) intertidal cornmunit) Ecological Monograph 413i 1-389

197i Experimental evaluation of ecological dom- Inance in u rock) intertidal algal communitq Ecological Monograph 45 137- 159

DeBenedict~s P A 1974 Interspecific competition be-tween tadpoles of Reitlei pipir t~s and Reitlei l~~ciriccian exper~mental field tudq Ecological Monograph 44 179- 151

Dodson S 1 1974 Zooplankton competition and pred~t- tlon an exper~mentul test of the size-efficiencq h) pothesis Ecology 55605-6 13

Duma P C 1964 Specie palr allopatrq In the genera R(iti(i and PI~ryt~oottrci Ecolog) 45 178-181

Emlen J M 1981 Field est imat~on of competition intensity Iheoreticul Population Biologq 19275-789

Gill Igt E 1978 The metapopulation ecology of the red- potted neht Voroplirliciltti~~~iritlcsic~ti(Rnfinesque) Ecological Monograph 48 145-166

1979 Denit) dependence and homing behavior in adult red-potted newt Vofopiirl~~iltirr iirttlt~ctn (Rafi- nesque) Ecologq 60800-8 13

H ilrton N G F E Smith and L B Slohodkin 1960 Communit) structure population control and competi-tion American Naturalit 94421425

Hallett J G and S L Pimm 1979 Direct estimation of competition American Naturalist 113593-600

Harper J L 1969 The role of predation in tepetational dkersi tq Page 48-61 it1 G M woodhell editor Diver- sity and stabilitq in ecological )tern Brookhaven Na-tional Laboratories Upton New York USA

Helwig J T and K A Council editor 1979 SAS users guide 1979 edition SAS Intitute Raleigh North Caroli- na USA

Herreid C F and S Kinneq 1966 Survival of Alaskan wood frog (K(it~ci cgtl~ciric(i lari ae Ecologq 47 1039- 104 1

He le r W R 1974 Niche measurements of frog larvae from a seawnal t rop~cal location In I hailand Ecolog) 55 h i 1-656

1976 Studlea of larval amphibian habitat pnrtition- ing Smithsonian Contributions to Zoologq 242 1-77

Heler W R R W hlcDinrmid and D L Weigmann 1975 l adpole predation and pond habitats in the tropic Bio- tropica 7 100-1 I I

Hurlhert S H 1971 The nonconcept of pecies divenit a critique and alternate parameters Ecolog) 52577-i86

Hi~rlhert S H J Zedler and D Fairhank 1972 Eco-system alteration by mosquitofish (C(inihr~ici clffint) predation Science 175639-64 1

Johnson D M 1973 Predation b) damselfly naiad on cladoceran population fluctuating ~ntenyity Ecology 54 25 1-269

Kerfoot W C 1975 The divergence of adjacent population Ecolog) 56 1298-13 13

L ~ c h t 1 E 1974 Survival of emhrgos tadpole and adult of the frog R ~ i ~ c r cirrrorcc ctrrror~r and Kcirici ~)rcjriocc prc- ioci )mputric in outhmetern Bri t~h Columbia Can21- dian Journal of Zoolog) 52613-027

Lilbchenco J 1978 Plant pecies divers~ty in a marine intertidal comrnunit) importance of herbivore food pref- erence and algal competitive nbilitie American Naturalist 11221-39

Mac Arthur R H 1958 Population ecologq of some war- blers of northeatern coniferou forest Ecolog) 39599- 619

Martof B W M Palmer J R Baile) and J R Harrison 111 1980 Amphibian and reptiles of the Carolinas and Virginia I he University of North Carolina Pres Chapel Hill North Carolina USA

Mellor W K 1975 Selective predat~on of ephippial Dripilnier and the resistence of ephippial egg to digestion Ecolog) 56974-980

Menge B A 1972 Competition for food between two intertidal startish pecie and i t effect on bod) size and feeding Ecology 53635-634

Menge B and J P Sutherlnnd 1976 Species diversit) gradients synthesis of the roles of predation competition and temporal heterogeneit) American Naturalit 11035 1-369

Morin P J 19x1 Predator) salamanders reverse the outcome of competition among three pecle of anuran tad-poles Science 212 1283-1286

1982 Predation and the compos~tion of temporal-) pond communitie5 Disertation Duke University Dur-ham North Carolina U S 4

19x3 in prtJs Competitive and predatory interactions in natural and experimental populat ion~ of Vo-oi~i~tiiciliiir~ tlorcili and i~irillc~cc~ri Aitihtoiti(i ir~r- i i t r i ~ i Copeio

Morriwn D F 1976 Multivariate statistical methods McGraw-Hill New York New York U S 4

Neill W E 1973 The communit) matrix and the inter- dependence of the competition coefficients American Nat- uralist 10839)1108

1975 Experimental studies of microcrustacean community compwition and efficiency of rewurce utili- ration Ecolog) 56809-826

P a ~ n e R T 1966 Food web complexity and specie di- tersity American Naturalist 10065-85

1980 Food mebs linkage interaction trength and community infrastructure Journal of Animal Ecology 49 667-685

Paine R T and R 1 Vada 1969 The effects of grazing hy sect u r c h ~ n S t r o n ~ ~ y l i c ~ c i i r o i ~ spp on henthic algtl population Lgtimnology and Oceanography 14710-739

Pianka E R 1973 The structure of lizard communities Annual Review of Ecology and Systematics 453-74

Pomerantz M J 19x1 Do higher-order interactions in compet~tion sytems really exist American Naturalist 117 538-591

Richmond N D 1947 Life hitory of Sccrpiiioprrs iiol- hrooXi (Harlan) Part I larval development and behavior Ecologq 2853-67

Root R B 1967 The niche exploitat~on pattern of the Blue-gre) Gnat Catcher Ecological Monograph 37317- 350

Salthe S N and W E Duellman 1973 Quantitatice constraints associated with reproducti~e mode in anurans Pages 229-249 iti J L Vial editor Eolutionar) biolog) of the

Ecolog~cll Monographr Vol 5 3 Un 2

anurans Un~versitv of Miswuri Pres Columbia Mis-ouri U S 4

Scheffi H 1959 1 he analbhis of variance J Wilev and Sons New York New ~ o r k USA

Schoenel- I M 1974 Rewurce partitioning in ecological communities Science 18527-39

Seale D B 1980 lnfli~ence of amphibian larvae on pri- mar) production nutrient flux and competition in a pond ecos) tem Ecology 61 153 1-1550

Smith-Gill S J and D E Gill 1978 Curvilinearitie in the competit~on equation an experiment with ranid tadpoles American Naturulit 112557-570

Soknl R R and F J Rohlf 1981 Biometr) W H Free- man San Francisco California USA

Steinwuscher K 1979cr Competitive interaction among tadpoles responses to reource level Ecology 60 1172- 1183

197 Host-parasite interaction as a potential pop- illation-regulating mechanism Ecolog) 60XXJ-890

Storm R M and R A Pimentel 1954 4 method for stud)ing amphihian breeding populations Herpetologica 10lhl-166

l imm N H 1975 Multivariate nnnl)i mith application in Education and P3qchology Wadsworth Belmont Cal- ifornia USA

Iravis J 1980 P h e n o t ~ p i c variation and the outcome of intenpecific competition in h)lid tadpoles Evolution 34 30-50

Turnipseed G and R Altig 1975 Population densit) and age structure of three species of hqlid tadpole Journal of Herpetolog) 92X7-291

Vnndermeer J 1981 4 further note on communit) model American Nutural~st 117379-380

Walter B 1975 Studies of interspecific predation mithin an amphibian comrnunit) Journal of Herpetolog) 9267-279

Wassersug R J 1975 The adaptive significance of the tadpole tage with comment5 on the maintenance of complex life cqcles American Zoologit 15405-417

Wilbur H M 1972 Competition predation and the true-

ture of the ri~h~sroitici-Reii~cirylciicci community Ecol- og) 533-21

1970 Density-dependent apect of metumorphois in iiihyor~ici and Rciilci syli(iictr Ecolog) 57 1289- 1296

1977cr Denit)-dependent aspects of growth and metamorphosis in Br(fi) cinitric~ciiiir Ecolog) 58 196200

1977h Interactions of food level and population d e n s ~ t ) In Krinci s111riricri Ecology 58206209

1977c Propagule Ize numher and diperion pattern in -tilh~sroilci and Aclcpicic American Naturalist 1 11 43-68

19x0 Complex life cycle Annual Review of Ecol- ogy and Sy3tematics 1167-93

1982 Competition between tadpole of Hyl(i fri~- orcili and Hylci qrcitioci in laboratory experiment Ecol- ogy 63278-281

Wilbur H M and J P Collin 1973 Ecolog~cal apect of amphihian metamorphosis Science 182 1305-1314

Wiltshire D J and C M Bull 1977 Potential competitive interaction betmeen larvae of P~irdopiiryiic~ hihroiii and P c~i tr i i t rc irr~iorc i f ( (Anurn Leptodactlyidne) 4utra- lian Journal of Zoolog) 25449454

Wood J T and 0 K Goodwin 1954 Observations on the abundance food and feeding behavior of the newt lotol~lirlrcili~~i~ (Rafinesque) in iridc~cetl ririclc~cc~ti Virginia Journal of the Elisha Mitchell Scientific Society 7027-30

Zaret T M 1980 Predation and freshwater communities Yale Unicersity Pres New Haven Connecticut USA

Zaret T M and R T Paine 1973 Specie introduction in a tropical lake Science 182449355

June I983 PRtL) I ION 411 A L K 4 L L I l L) ( OhlPOjI I ION I

APPENDIX I Lit of common in~er tebra te tax) obserked in the arraq of

2 2 cattle-watering tank

Hc~c~x~roc~iwic~p ilt~cocip B I I ~ I I ~ I ~ I5p Girri5 p zctltiic p L-e1cc Ollil p 7 lr~rttiot~c~c~ri~hciticiri lic~to~~c~licisp

Piircit~ctric1 5 nonp11rcittrc1ric~ rr-Parametric tech-nique for kariance analqsis (ANOVA MANOVA) are ro- bust to heteroscedasticity (inequality of Lariances) and de- partures from normality (Scheffe 1959) Thi propert) ha been okerlooked by man) ecologist Scheffe (1959) point O L I ~that asumption of normalit) and homoscedaticity were originall) made to ease the derivation of tatistical theory but moderate depart~lres from these assumptions hace little impact on the abilit) of the analqsis to detect differences among treatment means Parametric technique are more powerful than analogous nonparametric tests ( i e the) can reol e mallel- differences among treatment at a gicen level of replic~~tion) can teat a greater diversity of a prioriand hypothee For analogou5 reasons tranformation of original data are uua11y unnecearq and hake heen ued par- inglq in my analqi

ed with hIANOVA Finall) the multivariate apploach emph~i- ize that the entire guild with its potential for correlated responws of species was the ~lni t of s t ~ ~ d y

Timm ( 1975) or Morrison ( 1976) provide detailed description of the MANOVA analysi Significance was asse ed by Wilks Criterion defined by [determinant determinant ( F - H ) ] where H is the matrix of urns of squares and cross products calculated among treatment means of all variable and r is the matrix of sums of quare and cros products calculated within treatment ti and F are analogou to hypothesis and error sum of squares in a unicariate ANOVA Detection of I 5ignificant treatment effect with a MANOVA Indicate that differences exist among treatments but doe not indicate which cariables in the analyis contribute to difference Variables contributing most to difference among treatments can be identified b) a disc~iminant function analysi of anuran relatike abundances in different treatment Coefficients of each original variable in the discriminant function are gicen bq the eigenvector corresponding to the largest eigencalue of r H Thi function maximizes difference among treatments and minimizes differences among replicate within treatments Variables ( i e relative abundances of species) contributing to significant differences among treatments are identified by evaluating product moment correlations between values of the dixriminant function ca lc~~la ted from anuran relative abundances in each tank and original anuran relatike abundances (Timm 1975) This approach is preferable to the inspection of discriminant function coefficients be- caue the coefficient are a l ~ sensitive to the magnitude and Lariance of original kariables (Timm 1975)

Aticii o f l~o l~c~lor io t~ -Ca~~se of differences in rotitic~c relatice abundance of species among treatment were in-ferred from effect of predator density on four statitic measured for each population of each anuran species survival mean mas at tail resorption ( in milligrams) mean larkal period tin day ) and mean growth rate (in milligrams per daq) defined b) mean mass at tail resorption divided by mean larval period duration Anal)es were performed on population means which will tend to be normally distributed even if the original data were not i a the central limit theorem Because exact normalit) is not required for the analysis of variance tScheffe 1959) these data were not transformed

Okerwintering tadpoles of R I I ~ I I I pre5ented~ ~ ~ ~ c ~ t l o c ~ c ~ l ~ ~ ~ i l c i

special problems for the analysis of population statistic L a -cal period di5tribution5 in six population with overwintering tadpoles were potentially bimodal (cohorts metamorphosing in 1980 and potentially also in 1981) Only the 1980 cohort

Zllri~~iri(ire o t l ~ y i m~~l t ivar i -if I( c ot~l~oiriot~-A ate description of anuran guild composition offered obciou ~dkantages oker a brute-force pecie bq specie analls is of change in g ~ ~ i l d structure Anuran guild composition wa repl-e5ented bq the vectol P = (11 i l l 1 11) whel-e p wa the fraction of all urcicing anutan cenued from tank j belonging to pecies i Guild composition can be geometrically idealized a s a directional vector in a six-dimen5ional space delimited by axe denoting the relative abundance of each anuran pecies The central hqpothesi was tested with a ingle MANOVA which indicated i a one tet whether vectors of anuran relatike abundances differed among tanks containing different densities of salamander Relatike abundances were a n g ~ ~ l a r l ) transformed to reduce further the likelihood of multicolinearit) A univariate approach would require 5everal separate ANOVAs (one for each species) increasing the likelihood of type 1 error accumulat- ing from multiple test The univariate approach was also inensitie to correlated change in relatice abundances of different species while correlated differences can be detect-

wa used to estimate larcal period duration providing a con- istent comparison with the other species Number of okerwintering tadpoles and metamorph were summed to de- tribe relatice abundance at the end of the experiment accuratelq reflecting the number of R ~ i n ~ i sureicing to interact with other species Abundance and mean masse of okerwintering tadpoles were also analqzed separate11 to docu- ment effect of predation

Population statistics dexr ibed above have been argued to reflect fitness components of larkal anurans (Wilbur 1972 1980 Smith-Gill and Gill 1978) Although mean growth rate was completel) 5pecified by mean m a s at tail resorption and mean larval-period duration its presentation and analysis had heuritic calue in dexribing interactions between these a - pects of larcal development S u r ~ i v a l was known in all population and wa angularly tran5formed before anal) sis Analysis of surkikal demonstrated whether changes in relative abundance of a species corresponded to change5 mle1q in its survical or also reliected changes in surcical of other species For example a 5pecies may increase in relatike abundance without its survical increasing if surcikal of all other 5pecies in the guild declines Conerelq equivalent differences in surcical of all species among treatments will not generate differences in relative abundance Both survical

and relative abundance are required to describe effect of treatments on anuran 5pecies composition adequatelq

Population with zero s u r ~ i v a l had missing d u e s for other statistics Thu urcival was analyzed separatelq from re-maining population tatitics Because mean growth rate mean mas at tail resorption and mean larcal period were potentially correlated a MANOVA simultaneou~l) tested for effects of predator densit) on these statistics Miing slue for- different pecies did not coincide in the same tank ne-cessitating separate MANOVAs for each species Univariate results uggested which variable contributed to significant

multivariate tests but were not used to infer statistical significance

Differences in growth and de~elopment among treatments were ued to infer dilterences in the intensit) of competition experienced by tadpoles Mas at metamorphosis and growth rate are negaticely correlated with the intensit) of interan- uran competition while larval period duration is positively correlated with intensit) of competition (Brockelman 1969 Wilbur 1972 1976 197711 h 1980 Wilbur and Collins 1973 Wilthire and Bull 1977 Travis 1980) 1 hace used significant depression in mean mass at tail resorption or mean growth rate and significant increase in mean larval period duration to infer greater intensities of competition in ome treatment

1t1firctlcc~ ~fcr~tirl~ctiri~c~ tclti~~oIc-I used rcl(itions ~itirotllt~ canonic~l correlation analyis to describe patterns of correlation concisely between final abundances of potential intra- and interspecific competitors and mean growth rates of each specie Canonical correlation anal) sis is a computational an- alog of multiple regression that i specifically deigned for eculuating pattern of multiple correlation between two ~ I O L I ~

of variables (see Timm 119751 and Morrison [I9761 for further details of the analqsii Like multiple regression canonical correlation etablishe a linear combination of indepen-dent Lariables (in thi case the final denities of each species of anuran in each tank) that covaries best with Lalues of a dependent cariable (in this case the mean growth rate of a par t ic~~larspecies) The difference between the two ap- proache i one of interpretation Multiple regression is interpreted bq examining the coefficient of the independent ariable in the regresion equation However if two or more independent Lariables are highl) correlated and are also related to I dependent cariable on14 one independent Lariable will tend to carr) heavy weight in the regression equation (Sokal and Rohlf 1981) Thi is an artifact of the orthogonalization of independent variables in multiple

regresion and arbitrarily obscures other correlation be-tween independent and dependent variable In con-trat canonical correlation i interpreted by examining correlations between independent variables and their linear comhination (the canonical variable) best correlated with the dependent Lariable This approach p r e s e n e s pattern of correlation between cariables without arbitraril) eliminating cariables from the analqsi A correlative approach is also more appropriate than multiple regression when independent variables are not experimentall) manip~~la ted (Sokal and Rohlf 198 1 )

In my anallsis of densit)-dependent growth of each specie a linear combination of final densitie of six anuran 5pecies ( the canonical variable) maximized correlations between values of the canonical ariable and mean growth rate o e r all tanks The correlation between calues of canonical cariables and mean growth rate is called the canonical correlation and i5 analogous to R in a regression analysis A significant canonical correlation (as evaluated with Bartletts chi-square approximation [Timm 19751) confirmed correlations between final tadpole den5ities and mean growth rate Product mo-ment correlations between values of the linear combination of anuran abundances bet correlated with growth rates and final densities of anuran species in each tank identified species w h m e densities stronglq cocaried with growth rate Signif-icant negative correlations between densities and canonical cariable identified potential competitors i e 5pecies whose densitie were negative11 correlated with growth rates Ap- proximate intensities of competition exerted b) each species were inferred from the strength of negative correlation between anuran denhitie and canonical ariable

Thi use of canonical correlation was directlq analogous to multiple regression techniques proposed for the inference of interspecific competition (Hallett and Pimm 1979 Emlen 1981) These approaches asume that inter5pecific competition can be inferred bq regressing the abundance of a partic~llar specie against abundances of potential competitors in man) sampled habitats Multiple regression can be ued to infer the exis- tence of negative correlation between the abundance of one species and it potential interspecific competitor but it is obciousl) impos5ible to detect intraspecific competition b) regresing intrapecific densit) against itself By sing mean growth rate as an index of competitive stress it was pos~ib le to detect n e g a t i ~ e correlations between an index of compet- i t i ~ etres and both intraspecific and intenpecific densit)

I20 PETER J MORIN Ecological Monographs Vol 53 No 2

interpecific competition (Dumas 1964 Wiltshire and m) filled with rainwater They are acidic (pH = 40-Bull 1977) predation (Heyer et al 1975 Walters 1975) 65) and dilute (conductivity lt40 pSIcm at lgdegC) Many and ~nteractions between competition and predation ponds completely evaporate during periods of drought (Brockelman 1969 Wilbur 1972 1980) Preliminary re- which tend to occur in late summer and autumn These sults reported earlier (Morin 1981) for three anuran episodes of desiccation apparently exclude fish from species in the present study demonstrated that com- most ponds petition and predation interact to produce nonrandom In the Sandhills salamanders assume the role of patterns of species composition This paper extends dominant vertebrate predators where fish are absent the effects of predation to three additional species in Observations of predation by salamanders on amphib- the nura ran guild documents an inverse relation be- ian eggs and larvae are frequently correlated with the tween the intensities of predation and competition postbreeding exclusion of species from ponds (Walters among anurans and describes alternate patterns of 1975 Morin 1983) The broken-striped newt Notoph-larval development that permit successful metamor- thalmus viridescens dorsrilis enjoys a nearly ubiqui- phosis at high intensities of competition or predation tous distribution among 17 frequently observed ponds

Several other salamanders including Ambystomri ti- I H E EM NATuRAL TEMPORARY grinum Amhystomtr mabeii and Siren intermeditr are ON A N D ExPERIMENTAL ANALOGs less frequently encountered Notophthrilmirs is a gen-

Although my experiments were conducted in artifi- eralist predator that feeds on zooplankton large aquatic cia1 ponds they were inspired by the natural history invertebrates and the eggs and larvae of amphibians of temporary ponds in the Sandhills Game Manage- (Wood and Goodwin 1954 Christman and Franz 1973 men1 Area of Scotland and Richmond Counties in Mellors 1975 Walters 1975 Gill 1978) south-central North Carolina USA The following Rigorous experimental study of the impact of sala- overview of the natural history of natural ponds iden- mander predation demands control over potentially tifie4 potential sources of community organization confounded factors (especially variable colonization spec~ties the limitations of natural ponds as experi- and historical differences in species composition) that mental units and emphasizes the aspects of natural are difficult to separate control or manipulate in nat- ponds that could be mimicked in experimental artificial ural ponds Effective removals of Notophthtrlmlrs are ponds especially difficult in these natural ponds because these

The natural ponds are potential breeding sites for nonmigrating populations of newts cannot be excluded -25 species of frogs (Martof et al 1980) and up to 17 by interception at drift fences (Storm and Pimentel species of frogs may chorus and breed near a single 1954) Homing behavior also compromises the suc-pond during one evening Consequently larvae of many cessful augmentation of newt densities (Gill 1979) To species are forced into syntopy after breeding chorus- overcome the potential limitations imposed on exper- es that follow heavy rain creating rich and varied pos- iments conducted in restricted numbers of highly vari- sibilities for interspecific interactions There is little able natural ponds I manipulated salamander densi- evidence for habitat separation among larval anurans ties in artificial-pond communities reconstituted in in the Sandhills and aside from patterns of temporal cattle-watering tanks These analogs of temporary separation that vary greatly among years mechanisms ponds contained many of the species found in natural reducing ecological overlap among these or other lar- ponds but could be controlled manipulated and cen- val xnurans are not pronounced (Heyer 1974 1976) sused better than natural ponds Reconstructed fresh- In general tadpoles feed by rasping and filtering bac- water communities ranging from laboratory micro- teria phytoplankton periphyton and detritus (Was- cosms (Neil1 1974 1975) to artificial ponds (Hurlbert sersug 1975) To exploit temporary ponds successful- et al 1972 Johnson 1973) possess distinct advantages ly tadpoles must garner sufficient resources to for the empirical study of communities Construction metamorphose before ponds dry while simultaneous- of initially similar communities eliminates historical ly eading predators (Wilbur 1980) Successful mem- differences in species composition Reconstituted bership in a tadpole guild is best (and somewhat par- trophic webs can be engineered to contain fewer species adoxically) assayed by the successful metamorphosis than natural webs thereby revealing details of inter- and departure of tadpoles from the pond community specific interactions The realism of such reconstituted while unsuccessful members are those species that fail communities is confirmed by the successful develop- to complete larval development ment persistence and reproduction of component

Although most species can chorus and breed in each species (Morin 1983) pond in the Sandhills species composition of tadpoles Theory suggests that dispersal may influence broad and metamorphs varies conspicuously among ponds patterns of predator-mediated coexistence of prey This varixtion may reflect stochastic exploitation of species in collections of discrete habitats linked by ponds by ovipositing frogs However tadpoles of some migration (Caswell 1978) I have focused first on pat- species often fail to persist in ponds even after the terns of species composition and community dynamics input of thousands of viable eggs Most of the natural caused by processes operating strictly within com-ponds are shallow depressions (maximum depth lt 1 munities












--






ples



































-























- AY r 1504
























P -

















_ - 4 N E W T S

a

2 N E W T S

0 N E W T S

i 160 260 360 4io Mass (mg)








-


-


35 37 76- 59 9 3 88- 69 - 88

6 1 - - i2ii- 6 2

58- 60 1 7 73

2X 2 3 21 I 0

91 89 88 96 0016 0089 0595 0001

15 16 13 16






































I 34 PETER J





































































































































--






ples



































-























- AY r 1504
























P -

















_ - 4 N E W T S

a

2 N E W T S

0 N E W T S

i 160 260 360 4io Mass (mg)








-


-


35 37 76- 59 9 3 88- 69 - 88

6 1 - - i2ii- 6 2

58- 60 1 7 73

2X 2 3 21 I 0

91 89 88 96 0016 0089 0595 0001

15 16 13 16






































I 34 PETER J


























































































































--






ples



































-























- AY r 1504
























P -

















_ - 4 N E W T S

a

2 N E W T S

0 N E W T S

i 160 260 360 4io Mass (mg)








-


-


35 37 76- 59 9 3 88- 69 - 88

6 1 - - i2ii- 6 2

58- 60 1 7 73

2X 2 3 21 I 0

91 89 88 96 0016 0089 0595 0001

15 16 13 16






































I 34 PETER J





























1975 Some mechanisms producing tructure In natural communitie a model and ome evidence from field experiments hW9U tt1 M LA Cod) and J Din-mond ed i ton Ecology and evolution of communit~e Belknap Pres Cambridge Musachusetts USA























































































































-























- AY r 1504
























P -

















_ - 4 N E W T S

a

2 N E W T S

0 N E W T S

i 160 260 360 4io Mass (mg)








-


-


35 37 76- 59 9 3 88- 69 - 88

6 1 - - i2ii- 6 2

58- 60 1 7 73

2X 2 3 21 I 0

91 89 88 96 0016 0089 0595 0001

15 16 13 16






































I 34 PETER J












































































































































- AY r 1504
























P -

















_ - 4 N E W T S

a

2 N E W T S

0 N E W T S

i 160 260 360 4io Mass (mg)








-


-


35 37 76- 59 9 3 88- 69 - 88

6 1 - - i2ii- 6 2

58- 60 1 7 73

2X 2 3 21 I 0

91 89 88 96 0016 0089 0595 0001

15 16 13 16






































I 34 PETER J








































































































































P -

















_ - 4 N E W T S

a

2 N E W T S

0 N E W T S

i 160 260 360 4io Mass (mg)








-


-


35 37 76- 59 9 3 88- 69 - 88

6 1 - - i2ii- 6 2

58- 60 1 7 73

2X 2 3 21 I 0

91 89 88 96 0016 0089 0595 0001

15 16 13 16






































I 34 PETER J




























































































































_ - 4 N E W T S

a

2 N E W T S

0 N E W T S

i 160 260 360 4io Mass (mg)








-


-


35 37 76- 59 9 3 88- 69 - 88

6 1 - - i2ii- 6 2

58- 60 1 7 73

2X 2 3 21 I 0

91 89 88 96 0016 0089 0595 0001

15 16 13 16






































I 34 PETER J































































































































































I 34 PETER J


























































































































predation. competition, and the composition of …

Documents