Body size variation among invertebrates inhabiting different canopy microhabitat: flower visitors are smaller

Ecological Entomology (2013), 38, 101–111 DOI: 10.1111/j.1365-2311.2012.01410.x

Body size variation among invertebrates inhabitingdifferent canopy microhabitat: flower visitors aresmallerC A R L W . W A R D H A U G H ,1 W I L L E D W A R D S1,2 and N I G E L E .S T O R K 3 1School of Marine and Tropical Biology, James Cook University, Smithfield, Australia, 2Centre for Tropical

Environmental and Sustainability Science, James Cook University, Cairns, Australia, and 3Environmental Futures Centre, Griffith

School of Environment, Griffith University, Nathan, Australia

Abstract. 1. Factors such as reproductive fitness, climatic tolerance, predationpressure, energetic requirements and the quality and quantity of food sources allcorrelate with invertebrate body sizes.

2. This study examines body size variation between an invertebrate communityinhabiting five different microhabitats (mature leaves, new leaves, flowers, fruit andsuspended dead wood) that differ in quality, quantity, and availability in an Australiantropical rainforest canopy.

3. Mean body size varied significantly between invertebrate and beetle feeding guildsacross microhabitats. Phylogenetically independent contrasts revealed that invertebratetaxonomic groups were significantly smaller on flowers than on mature and new leaves.Size differences between microhabitats were most pronounced among herbivoroustaxa (Hemiptera, Lepidoptera). In particular, the immature stages or those groups thatdevelop on flowers were significantly smaller on flowers and larger on leaves thanexpected. Taxonomic groups with many strong flying species, especially those thatcomplete larval development on resources other than flowers, typically showed nodifferences in body size across microhabitats.

4. There are a number of potential hypotheses for the smaller body sizes offlower visitors, including: (i) differences in the physical sizes of the microhabitats;(ii) variation in time-dependent mortality risks that influence development times; and(iii) differences in the nutritional quality of the microhabitats, which can influencebody size via metabolic pathways.

5. The findings of this study do not support hypothesis (i) (with the possibleexception of one or two predatory groups). It is suggested that hypotheses (ii (time-dependent mortality factors) and particularly (iii) (nutritional variation) may be thebest avenues for future study as the main drivers of body size differences betweenmicrohabitats.Key words. Australian Canopy Crane, Daintree rainforest observatory, microhabitatdifferentiation, wet tropics.

Introduction

Body size is a fundamental species trait. Development times(Klingenberg & Spence, 1997; Blanckenhorn, 2000), fecundity(Honek, 1993; Wardhaugh & Didham, 2005), dispersal ability

Correspondence: Carl Wardhaugh, School of Marine and TropicalBiology, James Cook University, Cairns Campus, McGregor Road,Smithfield, Queensland 4870, Australia. E-mail: [email protected]

(Forkner et al., 2008), physiological performance (Wasserman& Mitter, 1978; May, 1979; Willmer & Unwin, 1981), compet-itiveness and vulnerability to predation (Connor & Taverner,1997; Beckerman et al., 2010) are all strongly correlated withbody size. Body size patterns also show distinctive relation-ships with species richness and abundance in multispeciesassemblages (Blackburn et al., 1990, 1993a; Stork & Black-burn, 1993; Siemann et al., 1996, 1999a), and can have animportant influence on community structure (Damuth, 1981;Morse et al., 1988; Blackburn et al., 1990, 1993a,b; Stork &

© 2013 The Royal Entomological Society 101

102 Carl W. Wardhaugh, Will Edwards and Nigel E. Stork

Blackburn, 1993; Blackburn & Gaston, 1994; Lindstrom et al.,1994; Nylin & Gotthard, 1998; Novotny & Basset, 1999; Sav-age et al., 2004).

Indeed, invertebrate community body size patterns havebeen found to correlate with a number of ecological factors,including successional status, disturbance and specialisation(Siemann et al., 1999b; Ribera et al., 2001; Braun et al., 2004;Lovei & Magura, 2006). For example, habitat disturbance gen-erally coincides with a decrease in body size (Braun et al.,2004), while body size has also been found to decrease throughhabitat succession (Siemann et al., 1999b). The latter relation-ship is influenced by differences in the efficiency with whichgenerally smaller-bodied specialist and larger-bodied general-ist herbivores process early and late successional plant species(Siemann et al., 1999b). Late successional plant species areless palatable than fast-growing, early successional species,so they tend to support a more specialised herbivore faunathat includes more small-bodied herbivores (Siemann et al.,1999b). Within single sites, body size has also been found tovary with respect to microhabitat utilisation. Stork and Black-burn (1993) showed that the mean body size of invertebrates inIndonesian rainforest was smallest among communities inhab-iting the soil and leaf litter, intermediate for species inhabitingthe herb layer and tree trunks, and largest among those insectsin the canopy.

Within species, large individuals often achieve higher repro-ductive fitness and have greater environmental tolerances thansmaller individuals (Shine, 1989; but see McLachlan, 1986;Ohgushi, 1996; Klingenberg & Spence, 1997; Nylin & Got-thard, 1998), so some directional selection should operatetoward larger body sizes. Indeed, it is generally accepted thatsexual selection in males and fecundity selection in femalesprimarily act to promote increased body size (Blanckenhorn,2000). However, selection against larger body sizes also exists.Development times usually increase with body size, potentiallyincreasing exposure of vulnerable immature stages to time-dependent mortality risks, such as predation, adverse weatherconditions or food shortages (Haggstrom & Larsson, 1995;Bernays, 1997; Williams, 1999; Blanckenhorn, 2000; but seeClancy & Price, 1987; Leather & Walsh, 1993; Nylin & Got-thard, 1998). Further, habitat constraints can also limit bodysize. For example, endoparasitic species are constrained bythe physical size of their hosts (Fox et al., 1996). Bonal andMunoz (2009) showed that body size of the seed weevil, Cur-culio elephas (Coleoptera: Curculionidae), was limited by thesize of the acorns within which their larvae develop. Resourcelimitation may also exert some influence at the scale of hostplant size. Small plants in general tend to result in reducedadult sizes among insects that complete development on them(Thompson, 1983; Dixon et al., 1995).

For particular taxonomic groups, locomotion, energy require-ments, feeding ecologies, and respiration impose the upper andlower limits in body size (Blackburn & Gaston, 1994). Forexample, the lower limit to body size in endothermic birds andmammals is set by the energetic requirements for maintaininginternal body temperatures (Pough et al., 1999), while uppersize in terrestrial arthropods is limited by diffusion efficienciesof tracheal respiratory systems under given atmospheric O2

concentrations (Schmidt-Neilsen, 1984). Novotny and Wilson(1997) proposed that the minimum body sizes of xylem-feedingHemiptera were constrained by the energetic cost of feeding onxylem fluid that is under negative tension and must be physi-cally pumped from the plant. Xylem-feeding insects thereforeneed to be large enough that the work required to extract xylemfluid is less than the nutritive benefits gained by consumingit. Xylem fluid is also nutritionally poor (Mattson, 1980) andmust be consumed in large quantities in order to extract enoughnitrogenous compounds to facilitate growth, which may alsofavour larger insects with longer digestive tracts that can absorball the available nutrients.

The quality, quantity and availability of food may ultimatelybe responsible for the body size range exhibited within andbetween many species (Tammaru, 1998; Ergon et al., 2004;Pfenning et al., 2007). Metabolic rate is inversely related tobody size (weight) (Elgar & Harvey, 1987; Nagy, 1987; Westet al., 2002; Brown et al., 2004), resulting in a general require-ment for small species to feed on more nutritionally con-centrated food sources than larger species (Horsfield, 1977;Augner, 1995; Behmer, 2009). As such, there is an expecta-tion that body size distributions should differ between habitattypes, due to inherent differences in the nutritional qualityof various food sources. Substantial differences in the bodysize distributions for species inhabiting different broad habi-tat types (soil, leaf litter, herb layer, tree trunks, and canopy)have been reported in tropical rainforests in Indonesia (Stork& Blackburn, 1993). Community-level body size variation hasalso been observed between sites with respect to disturbancelevel (Braun et al., 2004) or successional status (Siemannet al., 1999b). However, as far as we are aware, no one hasexamined body size variation at a scale as fine as betweenmicrohabitats within single trees. Nevertheless, there is reasonto suspect differences at this resolution. First, species abun-dances are negatively related to body size (Damuth, 1981,2007; Blackburn et al., 1990, 1993a; Stork & Blackburn, 1993;Brown et al., 2004). Previous work examining the invertebratecommunity in an Australian rainforest canopy has shown thatabundance varies considerably between different microhabi-tats (mature and new leaves, flowers, fruit and suspended deadwood) (Wardhaugh et al., 2012a,b). In particular, flowers sup-port densities of invertebrates that are orders of magnitudegreater than other microhabitats. Thus, if the abundance–bodysize relationship holds true within microhabitats on a single treespecies, body size distributions should be very much smalleron flowers in comparison to the other microhabitats. Here, wetest this hypothesis by examining body size variation withinand between different invertebrate taxonomic groups and feed-ing guilds collected from each microhabitat, and explore thisin more detail using beetle taxa alone in separate analysis.

Materials and Methods

Study site

All fieldwork was conducted using the Australian CanopyCrane at the Daintree Rainforest Observatory (a Long-TermEcological Research site; www.jcu.edu.au/canopycrane/), near

© 2013 The Royal Entomological Society, Ecological Entomology, 38, 101–111

Canopy invertebrate body size variation 103

Cape Tribulation (16◦17′S, 145◦29′E), Queensland, Australia(Stork, 2007). The crane is situated approximately 40 m abovesea level (asl) and >300 m from the forest edge in complexmesophyll vine forest (Tracey, 1982) that is contiguous withthe extensive lowland and upland rainforests of the DaintreeNational Park and Wet Tropics World Heritage Area (rangingfrom 0 to >1300 m asl). Approximately 1 ha of rainforestcontaining 745 individual trees [>10 cm diameter at breastheight (dbh)] from 82 species and 34 families is accessiblefrom the crane gondola [based on a recent (2009) survey at thecrane site which updates previously published data (Laidlawet al., 2007)]. The canopy is noticeably uneven in height,varying from 10 to 35 m. Although some rain does fall eachmonth (the lowest average monthly rainfall occurs in August;80 mm), there is a distinctive wet season from November toApril (the highest average monthly rainfall occurs in March;550 mm). The 50-year average annual precipitation at CapeTribulation is 3926 mm (Hopkins et al., 1996).

Sampling

Invertebrates were sampled from five microhabitats: matureleaves, new leaves, flowers, fruit and suspended dead woodfrom 23 locally common canopy plant species. The hosttree species selected represent a broad range of taxonomicrelatedness, growth pattern, phenology, distribution, size, andabundance. In addition to woody trees (19 species), two speciesof palms and two species of lianas were sampled (Table 1).

These species comprise 435/745 individuals and >70% ofthe basal area of all trees >10 cm dbh in the ∼1 ha area offorest under the crane arc (Laidlaw et al., 2007). One to threeindividuals of each host species were sampled each month for1 year (May 2008 to May 2009). Sampling did not occur inOctober 2008 due to the temporary unavailability of the crane.Invertebrate sampling was carried out by precision beatingof the microhabitat over a small beating sheet (0.23 m2).Each microhabitat on each replicate tree was sampled for10 min. In general, trees that were flowering, fruiting and/orleaf flushing were selected wherever possible, to maximisethe number and temporal distribution of samples from thesemore ephemeral microhabitats. Cross-contamination betweenmicrohabitat samples was kept to a minimum through the useof a small beating sheet that could be precisely positioned andby only sampling microhabitats that were discretely partitionedon host trees.

Body size measurements

Each beetle species (372 morphospecies) and every otherindividual invertebrate (30 039 individuals) was measuredfrom the front of the labrum to either the tip of the abdomen(excluding cerci or ovipositors) or the end of the elytra forsome Coleoptera (whichever is longer) using a calibratedgraticule. For beetle species, the mean size was calculatedfrom a sample of up to five individuals. The beetle faunawas chosen for a species-level examination of body size in

Table 1. The canopy plant species sampled, including family and growth habit.

Habit Family Species Trees on siteNo. of matureleaf samples

No. of newleaf samples

No. of flowersamples

Trees Lauraceae Endiandra microneura 22 20 4 0Cryptocarya mackinnoniana 16 14 6 0Cryptocarya grandis 7 1 2 2Cryptocarya hypospodia 1 3 0 1

Myrtaceae Acmena graveolens 16 19 5 5Syzygium sayeri 9 20 3 6Syzygium gustavioides 8 10 11 22

Meliaceae Dysoxylum papuanum 12 21 4 2Dysoxylum pettigrewianum 9 19 5 0

Euphorbiaceae Cleistanthus myrianthus 90 23 1 0Apocynaceae Alstonia scholaris 61 20 3 0Elaeocarpaceae Elaeocarpus angustifolius 7 22 0 2

Elaeocarpus bancrofti 1 11 1 1Cunoniaceae Gillbeea whypallana 5 3 3 2Proteaceae Cardwellia sublimis 14 20 6 2

Musgravia heterophylla 7 0 0 1Sterculiaceae Argyrodendron peralatum 17 16 4 7Myristicaceae Myristica insipida 59 18 2 3Fabaceae Castanospermum australe 8 22 4 2

Lianas Entada phaseoloides — 17 7 2Convolvulaceae Merremia peltata — 19 3 5

Palms Arecaceae Normanbya normanbyi 59 23 4 14Archontophoenix alexandrae 7 22 0 3

The number of individuals (>10 cm diameter at breast height) of each tree species accessible to the crane, and the total number of samples(individual trees across all time periods) taken from mature leaves, new leaves, and flowers for each tree species.



relation to microhabitat utilisation because of its ecologicaldiversity and high species richness (Grove & Stork, 2000). Alladult beetles (Coleoptera) were pinned or pointed and sortedto morphospecies (hereafter referred to as species). Specieswere compared with previous collections from the site (Stork& Grimbacher, 2006) and were critically evaluated by C.W.Wardhaugh and N.E. Stork. All samples are stored at JamesCook University, Cairns.

Guild assignment

Analysis of the guild composition of the arthropod faunaof the five microhabitats sampled here is discussed elsewhere(Wardhaugh et al., 2012a; C.W. Wardhaugh, unpublished)and our guild assignments follow Moran and Southwood(1982), Stork (1987) and Wardhaugh et al. (2012a), includingassigning ants to a separate guild because of their frequent

omnivorous or opportunistic feeding behaviour (see Table 2for a summary of guild assignments). Tourists are defined asthose invertebrate groups that do not feed on plants or relatedresources and include only the Diptera among the groupsincluded in the taxonomic analysis.

Statistical analyses

Since different food sources can influence invertebrate bodysizes, we first tested whether invertebrate and beetle feedingguilds differed in body size between each other and on differentmicrohabitats using two-way anova. All temporal sampleswere pooled and each feeding guild on each microhabitat oneach tree species was considered a replicate for this analysis(e.g., the mean body size of ants on new leaves on Cardwelliasublimis represents a single replicate). Since sampling effort

Table 2. The mean body size (mm) of invertebrates within each taxonomic group across the five microhabitats.

Flowers Mature leaf New leaf Fruit Wood

PredatorsAcari 0.37− 0.45+ 0.40 0.41 0.43Hymenoptera 1.52 1.31 1.37 — —Araneae 2.12− 2.49 2.90+ 2.30 2.18Larvae (Neuroptera) 4.21 2.88 — — —Coleoptera 1.74− 2.24+ 1.81 — —

FormicidaeAdults 5.58+ 4.32− 5.24+ 5.11 5.30Larvae — 3.00 3.27 2.97 —

SaprophagesBlattodea 5.02 5.39 6.34 — 6.23Collembola 1.20 1.30 1.14 1.50 1.21Coleoptera 1.77 2.63 — 1.7 —

HerbivoresLarvae (Lepidoptera) 4.24− 9.36+ 7.19 3.98 4.93Cicadellidae nymph 2.59 3.56 2.79 — —Gastropoda — 3.57 3.86 — —Mesophyll feeders 3.39− 7.38+ 6.49 — —Mesophyll nymph 1.88− 3.06 5.29 — —Orthoptera 7.85 8.93 12.71 — 8.59Phloem feeders 1.62− 3.85+ 3.01Phloem nymph 0.61− 1.87+ 1.69+ — —Thysanoptera 1.03− 1.03 0.79− 1.14 4.69+Coleoptera 2.95 3.02 3.15 3.03 —

FungivoresPsocoptera 1.54 1.73 1.69 1.70 1.64Coleoptera 1.66 1.44− 1.43 1.29 1.94

XylophagesColeoptera 9.75 6.37 — — —

TouristsDiptera 1.76 1.67 1.64 — —

UnknownLarvae (Coleoptera) 1.50 2.27+ — — 1.43Larvae (Diptera) 1.63 1.18− 0.93− — —

Total invertebrates 1.79− 2.96+ 3.29+ 2.50 2.92+Measurements are shown only for those groups where at least 10 individuals were collected from that particular microhabitat. Numbers in boldwith − signs signify those taxonomic groups that are smaller than expected under randomness on that particular microhabitat, while numbers inbold with + signs signify those groups that are larger than expected. Numbers that are not in bold did not differ from random expectation in meanbody size.



was not even across tree species (see Table 1), this approachreduced the likelihood of relatively large or small host-specificspecies on single host trees skewing the results. Beetles wereanalysed at both the species level (i.e. a single body sizemeasure per species) and at the individual level (i.e. thebody size of each individual was incorporated, identical tothe invertebrate analysis). Different microhabitats also differin the physical size of habitat units (i.e. individual leavesor flowers). To test for the influence of microhabitat size,the relationships between flower and leaf size (dry weightbiomass) and the body sizes of each of the 26 taxonomicgroups and eight feeding guilds (with n ≥ 10 individuals on atleast two microhabitats) were explored using linear regression.The minimum abundance limit of 10 individuals on at leasttwo microhabitats was chosen as a compromise betweenincluding a maximum number of groups for comparisonswhile excluding those with insufficient abundances to detectsignificant differences in body size, if they exist.

Next we tested whether individual taxonomic groupsdiffered in body size between microhabitats. Whereas theinitial two-way anova described above tests for broaddifferences in body size between microhabitats and feedingguilds, this second analysis was undertaken to identifywhich taxonomic groups, if any, were significantly smalleror larger on each microhabitat. For this we carried out arandomisation procedure to identify significant deviations fromrandom expectation in body size of invertebrate taxonomicgroups inhabiting different canopy microhabitats. We choseto use a randomisation procedure as opposed to an anovabecause, while some distributions allowed for use of a simpleF -statistic, others did not, and across the entire datasetno single transformation could be uniformly implemented.The randomisation process was therefore used to generatedistributions of the test statistic (for identifying taxa that werelarger or smaller than expected) that were unique and specificto each taxa, derived from the actual dataset, and without therequirement that the assumptions of the standard anova modelbe met (Whitlock & Schluter, 2008).

The randomisation process involved taking the entire inver-tebrate sample (40 374 individuals) and randomly assigningeach individual to microhabitat types to generate an expectedmean body size of each taxonomic group on each microhabi-tat. Randomisations were carried out within taxonomic groups.The procedure was iterated 1000 times. The mean body sizeof each taxonomic group on each microhabitat was recordedafter each iteration to generate frequency distributions of meanbody size per taxonomic group per microhabitat, assuming thatall invertebrates were equally likely to be found in all micro-habitats. We compared the observed (true) mean body size ofeach taxonomic group against the first and 1000th values in theordered set of values generated in the randomisation procedure[to ensure Bonferroni-adjusted (0.05/25 tests) α = 0.002 forall conclusions regarding significance] as a test for greater orsmaller mean body sizes of particular taxonomic groups withinmicrohabitats. Observed body sizes above the 1000th valuerepresent significantly larger body sizes for that particular tax-onomic group on that particular microhabitat, while observedbody sizes under the first value represent significant smaller

body sizes. Analyses were restricted to those taxonomic groupswhere abundance was ≥10 on at least two microhabitats.

Lastly, since broad comparisons of body size distributionsbetween microhabitats that group taxa into a single mean esti-mate will be confounded by phylogeny (as in analysis 1), wealso attempted to correct for this using phylogenetically inde-pendent contrasts (PICs) (Felsenstein, 1985). PICs were usedto test for coordinated changes in body size and microhabitatutilisation across multiple independent divergences. Positiveoutcomes in PICs reveal an evolutionary change in a particu-lar variable (e.g. body size) across multiple taxa in responseto a common factor (e.g. microhabitat utilisation). Each con-trast involved comparing the mean body size of individualsfrom each taxonomic group between microhabitats. The nullhypothesis for the test is that divergences in one aspect (micro-habitat usage) are not associated with divergences in another(body size), and mean difference in body size between differ-ent microhabitats was thus tested against a null hypothesis ofzero using a paired t-test. Fruit and dead wood invertebratecommunities were omitted from this analysis due to the lownumbers of invertebrates collected from these microhabitats.

Because beetle taxa were categorised at a much finerresolution (family and sub-family) than all other groups,we used a similar procedure of a set of phylogeneticallyindependent contrasts using data for beetles. Each contrastinvolved comparing the mean body size of species withineach beetle family on each microhabitat. Due to low speciesrichness and family-level diversity on new leaves, fruit anddead wood, only the single contrast comparing flower-visitingspecies and mature leaf-inhabiting species was possible. Allbody size measurements were log10(x + 1)-transformed priorto analyses to normalise the data.

Results

Body size variation between feeding guilds and microhabitats

A two-way anova with body size as the response vari-able showed a significant interaction between microhabi-tat identity and invertebrate feeding guild (F22, 422 = 3.69,P < 0.0001). Saprophagous invertebrates on flowers weresmaller than those collected from mature leaves, new leavesand suspended dead wood. Herbivores on flowers were alsosignificantly smaller than herbivores on all of the othermicrohabitats. Mean body size (mm ± SE) decreased inthe following order [no significant difference (=), signifi-cant difference (>)]: xylophages (6.61 ± 1.18) = ants (4.69 ±0.19) > saprophages (3.89 ± 0.23) = herbivores (3.34 ±0.18) > fungivores (1.61 ± 0.04) = tourists (1.79 ± 0.1)= predators (1.95 ± 0.09) = unknown (1.93 ± 0.14) (Fig. 1).

When beetle species alone were considered, again there wasa significant interaction between feeding guild and microhabitaton body size (F19, 214 = 2.183, P = 0.0039) (Fig. 2a). Whenanalysed at the individual abundance level, there was asignificant interaction between feeding guild and microhabitatin beetle body size (F19, 214 = 2.043, P = 0.0078) (Fig. 2b).In both the species-level and individual abundance analyses,fungivorous beetles were small on new leaves and large on



Fig. 1. The mean body size (± SE) of each invertebrate guild in thepooled sample. Letters indicate significant differences in size betweenfeeding guilds.

(a)

(b)

Fig. 2. The mean body size (± SE) of each feeding guild at thelevel for: (a) beetle species; (b) all individual beetles. Letters indicatesignificant differences in size between feeding guilds.

suspended dead wood, while herbivores showed the oppositetrend. The mean body size (mm ± SE) of beetle speciesdecreased in the order xylophages (7.28 ± 1.04) > herbivores(3.89 ± 0.32) > predators (2.55 ± 0.24) = fungivores (1.9 ±0.1) (Fig. 2a). Similar differences in the body sizes of feedingguilds were found when abundance was incorporated and allindividual beetles were analysed (Fig. 2b).

Body size variation of taxonomic groups betweenmicrohabitats

There was an overall trend for taxa to be smaller onflowers, especially among herbivores. Six of the 10 herbivorous

taxonomic groups were significantly smaller than expectedon flowers based on the randomisation model (summarisedin Table 2), and included in this group were three categoriesof immature taxa (Lepidoptera caterpillars, mesophyll-feedingand phloem-feeding Hemiptera nymphs). Three of the fivepredatory groups (Acari, Araneae and predatory Coleoptera)were also significantly smaller than expected on flowers. AdultFormicidae were the only group for which body size wassignificantly larger than expected on flowers. No taxonomicgroups differed significantly from random expectation inbody size on fruit and suspended dead wood, with theexception of Thysanoptera on suspended dead wood, whichwere larger than expected. Most remaining taxonomic groups,including saprophagic groups, Hymenoptera, Diptera and mostColeoptera guilds, showed no significant deviations fromrandom expectation in body size on any particular microhabitat(Table 2).

Body size and microhabitat biomass

Mean body size among flower visitors was not related toflower size (dry weight biomass) for any of the 26 invertebrategroups or eight feeding guilds (all P > 0.1), with the exceptionof fungivorous beetles, which showed a positive relationship(F1,16 = 11.8793, R2 = 0.4261, P = 0.0033). However, whena Bonferroni-adjusted significance level of 0.002 was applied,this significant result disappeared. Leaf size (dry weightbiomass) also had no effect on the body sizes of the foliageinvertebrate community either as a whole (F1,20 = 0.3273,R2 = 0.016, P = 0.57) or when different taxonomic groupsor feeding guilds were analysed separately (all P > 0.05).

Phylogenetically independent contrasts

Phylogenetically independent contrasts revealed a coordi-nated and repeated change in body size with microhabitatfor invertebrate taxonomic groups, which were significantlysmaller on flowers compared with mature leaves (t19 =2.36, P = 0.03) (Fig. 3a), and new leaves (t18 = 2.64, P =0.017) (Fig. 3c). PICs revealed no difference in body sizebetween invertebrate taxonomic groups inhabiting mature andnew leaves (t19 = 0.35, P = 0.73) (Fig. 3b). Between beetlespecies within the same families, PICs showed no differencein body size between flowers and mature leaves (t19 = 1.24,P = 0.23) (Fig. 4).

Discussion

These results showed that the distribution of body sizes forinvertebrates differed significantly between microhabitats, andthat invertebrates collected from flowers were (on average)smaller than those collected from other microhabitats. Thefindings support the hypotheses that body size varies betweencanopy microhabitats and that invertebrates on flowers aresmaller than invertebrates inhabiting other microhabitat types.PICs carried out between beetle species on flowers and mature



(a)

(b)

(c)

Fig. 3. The mean body size (mm) [log10(x + 1)] of invertebratetaxonomic groups in phylogenetically independent contrasts between:(a) mature leaf and flower-visiting taxa; (b) mature leaf and new leaftaxa; and (c) new leaf and flower-visiting taxa. The trend lines in eachgraph represent the null hypothesis of no difference in mean bodysize between microhabitats (y = x ). Points above the line representtaxonomic groups that are larger on the microhabitat represented onthe y-axis, while points below the line represent groups that are largeron the microhabitat on the x -axis.

leaves showed no differences in body size associated withmicrohabitat identity. Half of the taxonomic groups (six outof 12) for which mean body size was identified as beingsignificantly different from that expected under randomnesswere herbivorous, and half of these (three out of six) wereimmature stages (larvae and nymphs). All of the herbivoroustaxa that differed significantly from random expectation inmean body size were smaller on flowers than on other micro-habitats. This pattern suggests that size differences amongmany invertebrate groups could be linked to factors associ-ated with larval growth, since body size differences were notdetected in many groups that develop elsewhere and onlyvisit the focal microhabitats as adults (e.g. Hymenoptera,

Fig. 4. Phylogenetically independent contrasts in mean body size(mm) [log10(x + 1)] of flower-visiting beetle species and confamilialspecies inhabiting mature leaves. The trend line represents the nullhypothesis of no difference in mean body size between microhabitats(y = x ). Points above the line represent beetle families that arelarger on the microhabitat represented on the y-axis (mature leaves),while points below the line represent families that are larger on themicrohabitat on the x -axis (flowers).

Diptera, various Coleoptera). Body size also varied signifi-cantly between different invertebrate feeding guilds, a find-ing that was consistent when beetles alone were analysed.The general pattern found in both sets of analyses was fora decrease in body size across feeding guilds in the order:xylophages > herbivores > fungivores = predators. The onlyexception was the saprophages, which varied markedly in bodysize between the analyses of invertebrate feeding guilds andbeetle feeding guilds.

Why are flower visitors small and leaf inhabitants large?

Four potential hypotheses to explain the differences in bodysize found between canopy microhabitats, and in particularwhy flower-associated invertebrates are smaller than theirrelatives on leaves, have been proposed. These are: (i) time-dependent mortality factors; (ii) nutritive and metabolicexplanations; (iii) the physical size of microhabitats; and (iv)life history stage and microhabitat switching. Each of thesehypotheses is primarily concerned with larval development,as final adult size is influenced predominantly by processesoperating on immature stages which carry out most lifetimefeeding and are subject to the highest mortality rates in mostspecies.

Body size in invertebrates is often influenced by diet dur-ing the larval stages, both over evolutionary time and withinsingle generations (Scriber & Slansky, 1981; Tammaru, 1998).During juvenile development, increased nutritional quality offood often results in increased growth rates (Heisswolf et al.,2005; Cornelissen & Stiling, 2006) and larger adult bodysizes (Tammaru, 1998; but see Cornelissen & Stiling, 2006),which usually have positive effects on survival and reproduc-tion (Ohgushi, 1996; Awmack & Leather, 2002). But imma-ture stages of invertebrates are typically less mobile thanadults and experience high mortality rates due to predation



or food shortages (Haggstrom & Larsson, 1995; Bernays,1997; Williams, 1999; Blanckenhorn, 2000; but see Clancy& Price, 1987; Leather & Walsh, 1993; Nylin & Gotthard,1998). Under these conditions, selection should favour rapidgrowth in order to minimise time spent in the vulnerablelarval stage (Nylin & Gotthard, 1998). This creates conflictbetween the advantages of prolonging the larval stage, andthereby attaining a larger adult size, and the presumed repro-ductive benefits conferred by larger size, and shortening thelarval period to minimise juvenile mortality risks. Life histo-ries (including developmental periods) of individual speciesrepresent the outcome of the compromise between theseextremes.

Time-dependent mortality factors

Flowers harbour higher densities of predators (mites, spi-ders) than other microhabitats (C.W. Wardhaugh et al., unpub-lished), which could make flowers a particularly danger-ous place for developing herbivorous insects. Romero andVasconcellos-Neto (2004) provide tangential evidence for thestrength of predation pressure that may exist on flowers. Theyshowed that predation of herbivorous insects due to flower-dwelling spiders on Trichogoniopsis adenanther (Asteraceae)was great enough to exert positive effects on seed productionby reducing herbivore numbers and thus damage to reproduc-tive structures. Insects that develop on flowers may thereforeexperience very high rates of predation, and thus short develop-ment times may be advantageous. A similar argument could bemade for competition. Flowers are locations of very high den-sities of invertebrates (Wardhaugh et al., 2012b), so resourcecompetition could also be intense. In this case, rapid develop-ment could reduce time spent competing for limited resources.Since significant deviations in body size were prevalent amongthe same immature and adult insect groups (e.g., Hemipteragroups), it is possible that differences in body size betweenmicrohabitats are driven by the necessity for rapid developmentof insects on flowers in response to increased vulnerability tojuvenile mortality. Conversely, insects on leaves may be freeto prolong larval development in order to grow larger due torelatively higher survival rates resulting from less predationand competition.

The ephemeral nature of flowers compared with leaves couldalso impose limits on development time for florivorous insects.Most trees do not flower continuously, or even for prolongedperiods (Bawa et al., 2003). In this study, most trees wererecorded as flowering during a single sampling period (32/44individuals), and just five trees flowered on more than twoconsecutive sampling periods (i.e. for more than 8 weeks; aSyzygium sayeri, an Argyrodendron peralatum, and an Acmenagraveolens for three consecutive months and two Syzygiumgustavioides trees which flowered almost continuously). Singleinflorescences last for shorter time periods, and individualflowers last little more than 1 day for most generalist insect-pollinated plant species (Primack, 1985). Developing onflowers could thus constrain development as a function ofresource availability. Insects that develop on flowers may

therefore have short larval or nymphal stages, resulting insmall adult body sizes. By contrast, individual leaves ontropical rainforest trees last from several months to severalyears (Coley, 1988). Consequently, foliage-feeding species,even those that develop on a single leaf (e.g. leaf miners),may not be subject to the same development time constraintsas flower-feeding species.

Nutritive and metabolic explanations

Small species cannot survive on poor-quality food becausethe energetic costs of obtaining the meal outweigh thenutritional benefits (see Novotny & Wilson, 1997). Flowers aregenerally a more nutritious food source than leaves (Carisey& Bauce, 1997; Irwin et al., 2004), which may allow for themaintenance of very small body sizes. This could explainwhy many herbivorous groups were smallest on flowers andlargest on leaves. The overall pattern in changes in meanbody size among feeding guilds supports this hypothesis(Mattson, 1980). For example, for both the entire invertebrateassemblage and the beetle species considered alone, wefound mean body size to decrease across feeding guilds inthe order xylophages > herbivores > fungivores = predators.This order represents increases in nutritional quality ofthe resources used by each group; wood < living plantmaterial < animal prey = fungi. Similar patterns in body sizewere found by Grimbacher and Stork (2007), who showedthat xylophagous and herbivorous beetle species captured atthe same site (Daintree Rainforest Observatory) were largerthan predatory or fungivorous species.

The physical size of microhabitats

While ecological factors can affect body sizes as discussedearlier, it is possible that differences in body size betweenmicrohabitats simply reflect differences in the physical sizesof those microhabitats (Kirk, 1991; Dixon et al., 1995). Wefound little support for this proposition. For all tree speciesexamined in this study, flower sizes were very small comparedwith leaf sizes (only Merremia peltata flowers were largerin dry weight biomass than the smallest mature leaves).There was no relationship between leaf or flower size andthe body size of any invertebrate group. This is in contrastto Kirk (1991), who found that British species of thrips,pollen beetles (Nitidulidae) and flies (Tephritidae) all increasedsignificantly in body size with the increasing size of their hostflowers.

Most tree species sampled produced dense inflorescences,which greatly increases the amount of flowers available in theimmediate area, despite the small size of individual flowers.Thus, it is not necessarily the case that individual flowers (orleaves) are the appropriate level to consider habitat size/bodysize relationships, since aggregations of individual flowerswithin inflorescences (and individual leaves on branches) resultin a large total biomass in a small local area. The physical sizeof flowers and leaves may thus only restrict the maximum sizesof endophagous species, with little effect on externally feeding



species, which constituted the majority of the invertebratescollected in this study. For example, we found that therewere no differences in body size between microhabitatsamong strong flying groups such as Diptera, Hymenoptera andmany Coleoptera feeding guilds. However, spiders were alsosignificantly smaller than expected on flowers, which could bethe result of selection favouring the ability to hide among orwithin small flowers. Indeed, most spiders on flowers weresit-and-wait hunting crab spiders (Thomisidae), which rely oncamouflage to hunt flower visitors (Llandres et al., 2011). Itis possible, therefore, that spiders and perhaps some otherinvertebrate groups may be smaller on flowers because flowersare physically small.

Life history stage and microhabitat switching

Lastly, differences in body size of immature taxa betweenmicrohabitats could be due to the use of different resourcesduring larval or nymphal development. For instance, smallearly instars may be more vulnerable to chemical defences thatare prevalent within the foliage (van Dam et al., 2001), but notwithin floral tissues (e.g., Carisey & Bauce, 1997). Variationin susceptibility to chemical defences dependent on life-historystage could therefore facilitate diet switching. Other changesin feeding mode are necessitated by growing too large for aparticular resource. For example, many caterpillars switch fromleaf mining to case-bearing or leaf rolling when they exhausttheir food supply by growing too large for the leaf they inhabit(Gaston et al., 1991). Immature chewing herbivores (such ascaterpillars) may also need to attain a given size before theirmouthparts and related muscles are physically able to processtoughened leaf material (Bernays, 1986; Bernays & Janzen,1988). Early instars may therefore feed on flowers beforeswitching to foliage once large enough to mechanically handleleaf material (see Bernays, 1986).

However, body size differences between microhabitats maybe more likely to be the result of different microhabitats sup-porting different species of invertebrates (which is consistentwith hypotheses i, ii, and iii), rather than different microhabi-tats supporting different-sized individuals of the same species(hypothesis iv). For example, for phloem- and mesophyll-feeding Hemiptera, both adults and nymphs were significantlysmaller than expected on flowers, indicating that the differ-ences are due to true differences in body sizes between speciesthat occur on flowers and those that occur on leaves. These dif-ferences are also responsible for the significantly larger bodysizes of Thysanoptera on dead wood compared with flowersand leaves, and the significant deviations from expected bodysizes of spiders and ants on flowers and leaves. Dead branchessupported a large and distinctive species of Thysanoptera(adults were typically over 10 mm in length, compared with0.5–2 mm for most other species) that was not found on anyother microhabitat (C. W. Wardhaugh, pers. obs.), and thiswas probably the main driver of body size differences betweenmicrohabitats for Thysanoptera. In the case of spiders, flow-ers supported large numbers of small crab spiders, while thefoliage supported a wide variety of larger, actively huntingspecies (C. W. Wardhaugh, pers. obs.).

The most important group for which microhabitat switchingis likely to be a factor is the ants. Ants were larger thanexpected on flowers and new leaves, and smaller than expectedon mature leaves, which is the opposite pattern to thatobserved for most other groups that varied significantly in bodysize between microhabitats. While ants were rare on flowers(C.W. Wardhaugh et al., unpublished), the most abundantspecies collected from this microhabitat and new leaves wasthe large green weaver ant, Oecophylla smaragdina. Thisdominant species is found predominantly on leaves, whereit constructs large arboreal nests and tends to honeydew-producing trophobionts that favour new foliage (Bluthgen &Fiedler, 2002). But workers will visit all parts of a tree in theirregular foraging. While O. smaragdina was also abundant onmature leaves, this microhabitat was also utilised by manysmaller species that do not tend trophobionts (thus there maybe fewer of them on new leaves), and did not forage amongflowers to the extent that O. smaragdina did. Size differencesin ants between microhabitats may therefore be the resultof the non-random identity of ant species that move frommature leaves to new leaves and flowers when these areavailable.

Conclusions

This exploratory study represents the first demonstrationof body size differences of invertebrate taxonomic groupsbetween different canopy microhabitats. In particular, weshowed that invertebrates sampled from flowers were sig-nificantly smaller in body size than invertebrates collectedfrom other microhabitats. This pattern was demonstrated acrossa number of mostly herbivorous taxonomic groups, indicat-ing that size differences may be related to feeding biology.There are a number of potential (and not necessarily exclu-sive) hypotheses to explain this pattern. We suggest that two ofthese represent the most likely hypotheses for further investiga-tion. These are time-dependent mortality factors and nutritionalvariation, since they can be applied to both adult and imma-ture life history stages. The challenge now is to thoroughlytest these hypotheses using manipulative experiments to sepa-rate and identify the relative strengths of the effects of thesepossible causative factors. In particular, identifying whetherselection operates to reduce body size on flowers or increasebody size on leaves, or both, will be a valuable avenue forfuture work.

Acknowledgements

We thank Cassandra Nichols, Andrew Thompson, Shane Kelly,and Russell Holmes at the Australian Canopy Crane forallowing access to the canopy, and Peter Grimbacher for hishelp with identifying beetles. Financial support for fieldworkwas provided by a MTSRF grant. CWW was supported by agrant from the Skyrail Rainforest Foundation and an AustralianPostgraduate Award. We also thank two anonymous refereeswhose comments greatly improved the manuscript.



References

Augner, M. (1995) Low nutritive quality as a plant defence: effects ofherbivore-mediated interactions. Evolutionary Ecology, 9, 605–616.

Awmack, C.S. & Leather, S.R. (2002) Host plant quality and fecundityin herbivorous insects. Annual Review of Entomology, 47, 817–844.

Bawa, K.S., Kang, H. & Grayum, M.H. (2003) Relationships amongtime, frequency, and duration of flowering in tropical rain foresttrees. American Journal of Botany, 90, 877–887.

Beckerman, A.P., Rodgers, G.M. & Dennis, S.R. (2010) The reactionnorm of size and age at maturity under multiple predator risk.Journal of Animal Ecology, 79, 1069–1076.

Behmer, S.T. (2009) Insect herbivore nutrient regulation. AnnualReview of Entomology, 54, 165–187.

Bernays, E.A. (1986) Diet-induced head allometry among foliage-chewing insects and its importance for graminivores. Science, 231,495–497.

Bernays, E.A. (1997) Feeding by lepidopteran larvae is dangerous.Ecological Entomology, 22, 121–123.

Bernays, E.A. & Janzen, D.H. (1988) Saturniid and sphingid caterpil-lars: two ways to eat leaves. Ecology, 69, 1153–1160.

Blackburn, T.M. & Gaston, K.J. (1994) Animal body size distribu-tions: patterns, mechanisms and implications. Trends in Ecology andEvolution, 9, 471–474.

Blackburn, T.M., Harvey, P.H. & Pagel, M.D. (1990) Species number,population density and body size relationships in natural communi-ties. Journal of Animal Ecology, 59, 335–345.

Blackburn, T.M., Brown, V.K., Doube, B.M., Greenwood, J.J.D.,Lawton, J.H. & Stork, N.E. (1993a) The relationship betweenabundance and body size in natural animal assemblages. Journalof Animal Ecology, 62, 519–528.

Blackburn, T.M., Gaston, K.J. & Stork, N.E. (1993b) Temporaldynamics of body size of beetles on oaks: a cautionary tale. Eco-logical Entomology, 18, 399–401.

Blanckenhorn, W.U. (2000) The evolution of body size: what keepsorganisms small? The Quarterly Review of Biology, 75, 385–407.

Bluthgen, N. & Fiedler, K. (2002) Interactions between weaverants Oecophylla smaragdina, homopterans, trees and lianas in anAustralian rain forest canopy. Journal of Animal Ecology, 71,793–801.

Bonal, R. & Munoz, A. (2009) Seed weevils living on the edge:pressures and conflicts over body size in the endoparasitic Curculiolarvae. Ecological Entomology, 34, 304–309.

Braun, S.D., Jones, T.H. & Perner, J. (2004) Shifting average bodysize during regeneration after pollution – a case study using groundbeetle assemblages. Ecological Entomology, 29, 543–554.

Brown, J.H., Gillooly, J.F., Allen, A.P., Savage, V.M. & West, G.B.(2004) Toward a metabolic theory of ecology. Ecology, 85,1771–1789.

Carisey, N. & Bauce, E. (1997) Impact of balsam fir flowering onpollen and foliage biochemistry in relation to spruce budwormgrowth, development and food utilization. Entomologia Experimen-talis et Applicata, 85, 17–31.

Clancy, K.M. & Price, P.W. (1987) Rapid herbivore growth enhancesenemy attack: sublethal plant defenses remain a paradox. Ecology,68, 733–737.

Coley, P.D. (1988) Effects of plant growth rate and leaf lifetime on heamount and type of anti-herbivore defense. Oecologia, 74, 531–536.

Connor, E.F. & Taverner, M.P. (1997) The evolution and adaptivesignificance of the leaf-mining habit. Oikos, 79, 6–25.

Cornelissen, T. & Stiling, P. (2006) Does low nutritional quality actas a plant defence? An experimental test of the slow-growth, high-mortality hypothesis. Ecological Entomology, 31, 32–40.

van Dam, N.M., Hermanau, U. & Baldwin, I.T. (2001) Instar-specificsensitivity of specialist Manduca sexta larvae to induced defencesin their host plant Nicotiana attenuate. Ecological Entomology, 26,578–586.

Damuth, J. (1981) Population density and body size in mammals.Nature, 290, 699–700.

Damuth, J. (2007) A macroevolutionary explanation for energyequivalence in the scaling of body size and population density. TheAmerican Naturalist, 169, 621–631.

Dixon, A.F.G., Kindlmann, P. & Jarosik, V. (1995) Body sizedistribution in aphids: relative surface area of specific plantstructures. Ecological Entomology, 20, 111–117.

Elgar, M.A. & Harvey, P.H. (1987) Basal metabolic rates in mammals:allometry, phylogeny and ecology. Functional Ecology, 1, 25–36.

Ergon, T., Speakman, J.R., Scantlebury, M., Cavanagh, R. & Lam-bin, X. (2004) Optimal body size and energy expenditure duringwinter: why are voles smaller in declining populations? The Amer-ican Naturalist, 163, 442–457.

Felsenstein, J. (1985) Phylogenies and the comparative method. TheAmerican Naturalist, 125, 1–15.

Forkner, R.E., Marquis, R.J., Lill, J.T. & Le Corff, J. (2008) Timingis everything? Phenological synchrony and population variabilityin leaf-chewing herbivores of Quercus. Ecological Entomology, 33,276–285.

Fox, C.W., Martin, J.D., Thakar, M.S. & Mousseau, T.A. (1996)Clutch size manipulation in two seed beetles: consequences forprogeny fitness. Oecologia, 108, 88–94.

Gaston, K.J., Reavey, D. & Valladares, G.R. (1991) Changes infeeding habit as caterpillars grow. Ecological Entomology, 16,339–344.

Grimbacher, P.S. & Stork, N.E. (2007) Vertical stratification of feedingguilds and body size in beetle assemblages from as Australiantropical rainforest. Austral Ecology, 32, 77–85.

Grove, S.J. & Stork, N.E. (2000) An inordinate fondness for beetles.Invertebrate Taxonomy, 14, 733–739.

Haggstrom, H. & Larsson, S. (1995) Slow larval growth on asuboptimal willow results in high predation mortality in the leafbeetle Galerucella lineola. Oecologia, 104, 308–315.

Heisswolf, A., Obermaier, E. & Poethke, H.J. (2005) Selection oflarge host plants for oviposition by a monophagous leaf beetle:nutritional quality or enemy-free space? Ecological Entomology, 30,299–306.

Honek, A. (1993) Intraspecific variation in body size and fecundity ininsects: a general relationship. Oikos, 66, 483–492.

Hopkins, M.S., Head, J., Ash, J.E., Hewett, R.K. & Graham, A.W.(1996) Evidence of a Holocene and continuing recent expansionof lowland rain forest in humid, tropical north Queensland. Journalof Biogeography, 23, 737–745.

Horsfield, D. (1977) Relationships between feeding of Philaenusspumarius (L.) and the amino acid concentration in the xylem sap.Ecological Entomology, 2, 259–266.

Irwin, R.E., Adler, L.S. & Brody, A.K. (2004) The dual role offloral traits: pollinator attraction and plant defense. Ecology, 85,1503–1511.

Kirk, W.D.J. (1991) The size relationship between insects and theirhosts. Ecological Entomology, 16, 351–359.

Klingenberg, C.P. & Spence, J.R. (1997) On the role of body size forlife-history evolution. Ecological Entomology, 22, 55–68.

Laidlaw, M., Kitching, R.L., Goodall, K., Small, A. & Stork, N.E.(2007) Temporal and spatial variation in an Australian tropicalrainforest. Austral Ecology, 32, 10–20.

Leather, S.R. & Walsh, P.J. (1993) Sub-lethal plant defences: theparadox remains. Oecologia, 93, 153–155.



Lindstrom, J., Kaila, L. & Niemela, P. (1994) Polyphagy and adultbody size in geometrid moths. Oecologia, 98, 130–132.

Llandres, A.L., Gawryszewski, F.M., Heiling, A.M. & Herberstein,M.E. (2011) The effect of colour variation in predators on thebehaviour of pollinators: Australian crab spiders and native bees.Ecological Entomology, 36, 72–81.

Lovei, G.L. & Magura, T. (2006) Body size changes in ground beetleassemblages – a reanalysis of Braun et al. (2004)’s data. EcologicalEntomology, 31, 411–414.

Mattson, W.J. (1980) Herbivory in relation to plant nitrogen content.Annual Review of Ecology and Systematics, 11, 119–161.

May, M.L. (1979) Insect thermoregulation. Annual Review of Ento-mology, 24, 313–349.

McLachlan, A.J. (1986) Survival of the smallest: advantages andcosts of small size in flying animals. Ecological Entomology, 11,237–240.

Moran, V.C. & Southwood, T.R.E. (1982) The guild composition ofarthropod communities in trees. Journal of Animal Ecology, 51,289–306.

Morse, D.R., Stork, N.E. & Lawton, J.H. (1988) Species number,species abundance and body length relationships of arboreal beetlesin Bornean lowland rain forest trees. Ecological Entomology, 13,25–37.

Nagy, K.A. (1987) Field metabolic rate and food requirement scalingin mammals and birds. Ecological Monographs, 57, 111–128.

Novotny, V. & Basset, Y. (1999) Body size and host plant special-ization: a relationship from a community of herbivorous insects onFicus from Papua New Guinea. Journal of Tropical Biology, 15,315–328.

Novotny, V. & Wilson, M.R. (1997) Why are there no smallspecies among xylem-sucking insects? Evolutionary Ecology, 11,419–437.

Nylin, S. & Gotthard, K. (1998) Plasticity in life-history traits. AnnualReview of Entomology, 43, 63–83.

Ohgushi, T. (1996) Consequences of adult size for survival and repro-ductive performance in a herbivorous ladybird beetle. EcologicalEntomology, 21, 47–55.

Pfenning, B., Poethke, H.J. & Hovestadt, T. (2007) Dealing withtime constraints on development: the effect of food availability.Ecological Entomology, 32, 273–278.

Pough, F.H., Janis, C.M. & Heiser, J.B. (1999) Vertebrate Life, 5thedn. Prentice-Hall, Upper Saddle River, New Jersey.

Primack, R.B. (1985) Longevity of individual flowers. Annual Reviewof Ecology and Systematics, 16, 15–37.

Ribera, I., Doledec, S., Downie, I.S. & Foster, G.N. (2001) Effectof land disturbance and stress on species traits of ground beetleassemblages. Ecology, 82, 1112–1129.

Romero, G.Q. & Vasconcellos-Neto, J. (2004) Beneficial effectsof flower-dwelling predators on their host plant. Ecology, 85,446–457.

Savage, V.M., Gillooly, J.F., Brown, J.H., West, G.B. & Charnov,E.L. (2004) Effects of body size and temperature on populationgrowth. The American Naturalist, 163, 429–441.

Schmidt-Neilsen, K. (1984) Scaling: Why is Animal Size so Important?Cambridge University Press, Cambridge, U.K.

Scriber, J.M. & Slansky, F. Jr (1981) The nutritional ecology ofimmature insects. Annual Review of Entomology, 26, 183–211.

Shine, R. (1989) Ecological causes for the evolution of sexual sizedimorphism: a review of the evidence. Quarterly Review of Biology,64, 419–461.

Siemann, E., Tilman, D. & Haarstad, J. (1996) Insect species diversity,abundance and body size relationships. Nature, 380, 704–706.

Siemann, E., Tilman, D. & Haarstad, J. (1999a) Abundance, diversityand body size: patterns from a grassland arthropod community.Journal of Animal Ecology, 68, 824–835.

Siemann, E., Haarstad, J. & Tilman, D. (1999b) Dynamics of plantand arthropod diversity during old field succession. Ecography, 22,406–414.

Stork, N.E. (1987) Guild structure of arthropods from Bornean rainforest trees. Ecological Entomology, 12, 69–80.

Stork, N.E. (2007) Australian tropical forest canopy crane: new toolsfor new frontiers. Austral Ecology, 32, 4–9.

Stork, N.E. & Blackburn, T.M. (1993) Abundance, body size andbiomass of arthropods in tropical forest. Oikos, 67, 483–489.

Stork, N.E. & Grimbacher, P.S. (2006) Beetle assemblages from anAustralian tropical rainforest show that the canopy and the groundstrata contribute equally to biodiversity. Proceedings of the RoyalSociety of London Series B , 273, 1969–1975.

Tammaru, T. (1998) Determination of adult size in a folivorous moth:constraints at instar level? Ecological Entomology, 23, 80–89.

Thompson, J.N. (1983) Selection pressures on phytophagous insectsfeeding on small host plants. Oikos, 40, 438–444.

Tracey, J.G. (1982) The Vegetation of the Humid Tropical Region ofNorth Queensland. CSIRO, Melbourne, Australia.

Wardhaugh, C.W. & Didham, R.K. (2005) Density-dependent effectson the reproductive fitness of the New Zealand beech scale insect(Ultracoelostoma assimile) across multiple spatial scales. EcologicalEntomology, 30, 733–738.

Wardhaugh, C.W., Stork, N.E. & Edwards, W. (2012a) Feeding guildstructure of beetles on Australian tropical rainforest trees reflectsmicrohabitat resource availability. Journal of Animal Ecology, 81,1086–1094.

Wardhaugh, C.W., Stork, N.E., Edwards, W. & Grimbacher, P.S.(2012b) The overlooked biodiversity of flower-visiting invertebrates.PLoS ONE, 7, e45796.

Wasserman, S.S. & Mitter, C. (1978) The relationship of body sizeto breadth of diet in some Lepidoptera. Ecological Entomology, 3,155–160.

West, G.B., Brown, J.H. & Enquist, B.J. (2002) A general model forontogenetic growth. Nature, 413, 628–631.

Whitlock, M. & Schluter, D. (2008) The Analysis of Biological Data,1st edn. Roberts and Company Publishers, Greenwood Village,Colorado.

Williams, I.S. (1999) Slow-growth, high-mortality – a general hypoth-esis, or is it? Ecological Entomology, 24, 490–495.

Willmer, P.G. & Unwin, D.M. (1981) Field analyses of insect heatbudgets: reflectance, size and heating rates. Oecologia, 50, 250–255.

Accepted 18 October 2012


Documents

Body size variation among invertebrates inhabiting different canopy microhabitat: flower visitors are smaller