Elevation and Moths in a Central Eastern Queensland Rainforest

bs_bs_banner

Austral Ecology (2015) ••, ••–••

Elevation and moths in a central eastern Queensland rainforest

ERICA H. ODELL,* LOUISE A. ASHTON AND ROGER L. KITCHING Environmental Futures Research Institute and Griffith School of Environment, Griffith University, 170 Kessels Road, Brisbane, Queensland 4111, Australia (Email: erica.odell@griffithuni.edu.au)

Abstract Elevational gradients are powerful natural experiments for the investigation of ecological responses to changing climates. Automated modified Pennsylvania light traps were used to sample macro-moth assemblages for three consecutive nights at each of 24 sites ranging from 200 m asl to 1200 m asl within continuous tropical rainforest at Eungella, Queensland, Australia (21°S, 148°E). A total of 13 861 individual moths representing approximately 713 morphospecies and 10 045 individuals belonging to approximately 607 morphospecies where sampled during November 2013 and March 2014 respectively. Moth assemblages exhibited a strong elevational signal during both sampling seasons; we grouped these into lowland and upland assemblages.The dispersal pattern of moth assemblages across the landscape reflected the stratification of vegetation communities across elevation and correlated with shifts in eco-physical variables, most notably temperature and substrate organic matter. Regional historical biogeographical events likely contributed to the observed patterns.The analysis presented here identifies a set of statistically defined elevationally restricted moths which may be of use as part of a multi-taxon predictor set for monitoring future ecosystem level changes associated with elevation and, by implication, with climate.

Key words: diversity, elevation, moth, rainforest, stratiﬁcation.

INTRODUCTION have resulted in functional asynchronies among important biological events (Root et al. 2003; Kiers Over the past century, average global temperatures et al. 2010; Thackeray et al. 2010; Rafferty & Ives have increased by 0.78°C (IPCC 2013) and are pre- 2011; Bellard et al. 2012; Nakazawa & Doi 2012). dicted to continue to rise throughout the 21st century Temperature, solar radiation, precipitation, sea level (National Research Council 2008). Pragmatic climate as well as the timing and magnitude of extreme events projections estimate a global shift in average tempera- are all anticipated to change under global climate tures by 3–6°C within the coming century (IPCC change. The likely impacts on biological diversity of 2013), with some areas, such as mountains and high change in several of these variables can be investigated latitudes experiencing greater levels of warming than using elevational gradients. Elevational gradients are others (Diaz et al. 2003; Nogués-Bravo et al. 2007; powerful natural experimental systems for investigat- Kohler & Maselli 2009; World Bank 2012). Under- ing ecological responses to climatic inﬂuences (Körner standing the impacts and responses of ecosystems to 2007). They offer a unique opportunity to observe climate change is essential if measures to ameliorate potential responses of ecosystems to climate change by and predict the impacts of climate change are to be presenting naturally steep climatic gradients within effective (Shaver et al. 2000). relatively small geographical ranges and therefore A broad range of organisms from several bio- allow sampling from a narrow species pool. This sub- geographical regions have been shown to have stitution of space for time, however, is not without responded to recent changes in the global climate issues especially given the uncertainty in trajectories of (Parmesan 2007; Cook et al. 2008; Thackeray et al. future change. Nevertheless, by monitoring the 2010). Changes in species distributions and elevational distributions of in situ assemblages, we can phenological events are among the most widely begin to understand how ecosystems may respond to a observed (Williams et al. 2003; Thomas et al. 2004; globally changing climate. In contrast to locations in Chen et al. 2009; Hughes 2012). Shifts in the timing of the Northern Hemisphere, Australia, in general, lacks ﬂowering, bud burst, emergences, migrations and the necessary long-term historical datasets and nesting have been observed (Van Asch et al. 2007; phenological monitoring that have allowed for the Møller et al. 2008; Burkle et al. 2013), some of which detection of a range of biodiversity impacts of climate change to date (Hughes 2003). Recent climate *Corresponding author. warming in Australian is now well established with Accepted for publication April 2015. observable consequences for biodiversity (Steffen et al.

2009; IPCC 2014).The establishment of baseline dis- The region’s mountains are frequently covered by a cloud tribution data and the identification of potential pre- cap. This produces condensation which helps maintain a dictor species remains an essential need if our consistently moister environment in the upland forest, par- understanding of climate change, monitoring and ticularly in the dry season. Cloud forests are generally thought management responses are to be sensible and timely of as crucial for the maintenance of high elevation rainforest structure and the associated arthropod assemblages of upper (Hughes 2000; Huddleston 2012). The present paper elevational zones (Nadkarni et al. 2000; Oosterhoorn & contributes one such baseline. Kappelle 2000; Williams-Linera 2002). Direct precipitation Indicator species or groups of species whose behav- from cloud caps mitigates moisture loss and provides a stable iour mimics that of a larger set of variables are useful moist environment suitable for the development and mainte- for their ability to convey information about environ- nance of montane rainforest. Notophyll vine forests (Webb mental change (Bohac 1999; Niemi & McDonald 1978), characteristic of low nutrient soils, dominate higher 2004; Trindade-Filho & Loyola 2011). Individual elevations and gradate into mesophyll forests (Webb 1978) at species may be used to detect environmental changes lower elevations. Some baseline information is available on the but, when community level changes are probable, sets vegetation, of which most is derived from a permanent plot of organisms are likely to be more powerful and useful maintained by the Commonwealth Scientific and Industrial (Kitching et al. 2000; Kitching & Ashton 2013). Inver- Research Organisation (CSIRO) (Graham 2006) located to the south of the Clarke Range on basaltic soils. Much of the tebrates are widely accepted as powerful indicators of northern section of the Clarke Range, however, has developed ecosystem states. Aquatic taxa such as Ephemeroptera, upon an extensive granitic landscape (early Miocene to Plecoptera and Trichoptera have been widely used Palaeocene) (Graham 2006). Few studies provide detail on with considerable success to assess stream quality and vegetation composition or structure for forests on these gra- environmental health of rivers (Blattenberger et al. nitic soils. Little if any information is available on the associ- 2000; Carignan & Villard 2002; Wright et al. 2007; ated invertebrates. EHMP 2008). In terrestrial ecosystems, Lepidoptera (moths and butterflies) and Coleoptera (beetles) are commonly employed as indicator groups to measure Study design environmental quality (Niemelä et al. 2000; Rainio & Niemelä 2003; Maleque et al. 2009; Gerlach et al. The methodology of this study follows those of the Investigating 2013). The present study examines elevation-related the Biodiversity of Soil and Canopy Arthropods (IBISCA) changes in a Queensland tropical rainforest in order Queensland project (Ashton et al. 2011; Kitching et al. 2011). to select statistically defined sets of night-flying The continuity in sampling techniques allows for uncompro- Lepidoptera (moths) that may be used to monitor mised comparisons of data between Eungella and previous future ecosystem level changes. studies (see Ashton et al. 2011; Ashton 2013).The comparisons drawn among these studies, uncompromised by the use of incompatible methodologies, contributes to the development of a more general understanding of ecosystem responses to eleva- METHODS tion and climate, and how these respond to changing climate. This study was carried out along an elevational gradient at Eungella National Park to investigate changes in moth Study site assemblages between 200 m asl and 1200 m asl. Sampling occurred among elevational bands at 200 m elevation inter- Eungella (21°S, 148°E), located 80 km west of Mackay on vals at 200 m asl, 400 m asl, 600 m asl, 800 m asl, 1000 m asl the Clarke Range, is the largest patch of remnant rainforest and 1200 m asl. Four plots spaced at least 500 m apart were between the Wet Tropics of far northern Queensland and the established within each elevational band (Fig. 1, see Appen- Macpherson Range and Tweed Caldera in the subtropical dix S1 for site coordinates). However, in contrast to the south-east of Queensland and north-eastern New South IBISCA Queensland project, the spatial positioning of the Wales. The rainforests of the Eungella massif present a con- plots within each elevational band were not placed in a line to tinuum ranging from lowland rainforest at about 200 m asl to account for spatial autocorrelation factors. montane rainforest above 1100 m asl. In contrast to the forests of the Wet Tropics and the Tweed Caldera, the rainforest of Eungella have been little studied ecologically. Eungella’s rainforest massif is known as an area of Light trapping endemism, with numerous local species including vertebrates such as the Eungella honeyeater (Lichenostomus hindwoodi), Moths were sampled using modified Pennsylvania light traps the orange speckled forest skink (Eulamprus luteilateralis), the (Kitching et al. 2005).These use a vertically mounted 8-watt Eungella day frog (Taudactylus eungellensis) and the, now actinic fluorescent tube (Philips, 350–410 nm) that emits extinct, Eungella gastric brooding frog (Rheobatrachus short wave blue (and white) light which is surrounded by a vitellinus). A high level of endemism has similarly been trio of transparent vanes which intercept the flight paths of observed in the region’s invertebrates, with Eungella- attracted moths.These moths are thus directed into a bucket producing unique suites of species in multiple studies suspended below the light. Moths are euthanized in situ using (Kitching et al. 2001, 2004; Majer et al. 2001). dichlorvos impregnated resin strips. The traps are powered doi:10.1111/aec.12272 © 2015 Ecological Society of Australia MOTH ASSEMBLAGES IN A CENTRAL QLD FOREST 3

Fig. 1. Topographic map of Eungella study sites. Contours occur at 100 m intervals. ▲ = 200 m asl, ▼ = 400 m asl, ■ = 600 m asl, ● = 800 m asl, ◆ = 1000 m asl, ✶=1200 m asl. by a single 12-volt battery and are operated via a timer cut-off ensured the sampling of a large proportion of all enabling precise, synchronized sampling for a 13-h period Lepidoptera while avoiding the technical (and taxonomic) between 6 pm and 7 am. difficulties associated with the handling of some micro- Two traps were used to sample the moth fauna at each site, lepidopteran groups, as the expertise and taxonomic informa- one positioned in the understory and the other suspended in tion is absent for the Australian fauna. All pyraloid moths, the canopy. Previous work has shown marked contrast however, were processed regardless of size. Individual moths between the ground and canopy fauna of rainforest moths were sorted into morpho-species, with representative indi- (Basset et al. 1992; Schulze et al. 2001; Brehm 2007). The viduals of each morpho-species mounted and included in a separation of traps ensures a broad sampling of the noctur- numbered reference collection. Subsequent encounters with nally active moth fauna and provides an adequate represen- easily recognizable morpho-species were simply recorded and tation of individuals from different strata. the moths discarded. Individuals were identified to the lowest possible taxonomic level using available collections at Griffith University and the Queensland Museum in addition to pub- Sampling design lished sources (see Common 1990). In total 288 single-night trap samples were processed during November 2013 and The moth fauna was sampled twice, once at the beginning of March 2014. the wet season (November 2013) and again at its end (March 2014).A pilot study using a similar but not identical set of sites was conducted in March 2013. Ground and canopy traps at Vegetation survey each site were run simultaneously for three nights, with each night’s catch collected the following morning and transferred At each site, 20 × 20 m vegetation plots were established. All to a field laboratory for processing.All moths with a fore-wing trees with a diameter at breast height (DBH) greater than 5 cm length equal to or greater than 1 cm were processed. This within this plot were marked with a permanent aluminium tag

and assigned a unique identification number.A record of all trees Relationships among moth assemblages and environmen- and their circumference is available and will be published in due tal data were compared using the Mantel type Spearman course. Species identification was completed by Dr. W. rank correlation (found in the RELATE function of MacDonald (Queensland Herbarium, Toowong). PRIMER) with 999 permutations. A vector overlay based on Pearson’s correlation (more than 0.9) (Gerstman 2003) was superimposed on the MDS ordination and isolated promi- Temperature nent plant species and families which may be possibly responsible for driving the distribution patterns of the moth assemblages. Local elevational climatic data were obtained using elec- In addition to these multivariate analyses values of Fisher’s tronic data loggers (i-Buttons) suspended within ventilated diversity index (α) were calculated for each elevation. Fish- sections of polyester piping from November 2013 to January er’s index is a parametric index combining species richness 2015. These were deployed at three out of four sites at each and logged values of species abundances. The index is sup- elevation. Two i-Buttons remained in situ at each of the posedly independent of sample size once that sample size selected plots: one in the understory and the other positioned exceeds 1000 individuals (Kempton & Taylor 1976). The in the canopy. Each device was synchronized and pro- value has been widely used in previous studies of moth grammed to sample the local temperature on an hourly basis. assemblages (Brehm et al. 2003; Brehm & Fiedler 2005; Axmacher & Fiedler 2008) and was calculated here for ease of comparison with those studies. Soil To test the completeness of our sampling, we calculated values of the abundance coverage estimator for each eleva- Five soil core samples were randomly collected within each tion and the location as a whole. Abundance-based coverage site. All leaf litter and surface organic matter was first moved estimators approximate the true number of species in a given aside before obtaining a 6 × 6 × 15 cm deep soil core. These area by taking rare species into consideration (Chao et al. five cores were then thoroughly mixed and a subsample 2005; Colwell et al. 2012). This method has been recom- extracted to represent each plot. A range of physical, chemi- mended for datasets with large numbers of rare species, such cal and mineralogical characteristics of the soil was tested by as this one (Chao et al. 2000; Colwell et al. 2012) and have Phosyn Analytical Pty Ltd (Andrews, Queensland, Australia) been shown to be less biased than incidence-based estimators with all extraction protocols following those outlined in for simulated mobile organisms (Brose & Martinez 2004). Rayment and Higginson (1992). In order to select a suite of statistically robust and easily monitored predictor species, we categorized morpho-species as either ‘common’ or ‘rare’ using the algorithm P = (1–1/ N Data analysis n) based on a 95% chance of repeat encounter (see Novotny et al. 2007). Morpho-species whose chance of repeat Combined canopy and ground data were pooled across all encounter was greater than 95% were designated as three trap nights to form a single data point per site. Species common, and those lower than 95% as rare. All species with data were converted to proportions and square root trans- an encounter rate less than the calculated cut-off were formed before analysis. Plymouth Routines in Multivariate excluded for consideration as a potential predictor species. Ecological Research (PRIMER) 6 (Clarke & Gorley 2006) Predictor species and elevationally restricted species were was used to investigate patterns in moth assemblages across identified using the IndVal package offered in R Studio elevational zones. Permutational multivariate analysis of (RStudio 2013). This analysis identifies elevationally variance (PERMANOVA), found in the PERMANOVA + restricted species based on distributions, fidelity to add-on software of PRIMER 6, was used to test for signifi- elevational zone and repeat capture across sites within an cant differences in moth assemblages across the six elevational zone. For example, a species that occurs only at a elevational bands.The dissimilarity in assemblages was deter- single elevation (or a number of adjacent elevations) and mined with the Bray–Curtis dissimilarity index (Clarke & appears in all sites from that elevation (or elevations) is Gorley 2006) and transformed into multidimensional scaling ranked highly as a potential indicator. In this fashion, an (MDS) ordinations calculated with 100 random restarts to ‘indicator group’ of species that could form the basis of illustrate the relationship of moth assemblages among differ- future monitoring protocols can be developed, and the dis- ent sites. tributional changes of moths over time tracked (Kitching & The degree to which the spatial distribution of the plots Ashton 2013). IndVal analyses were executed on all possible influenced and/or explained the datasets, in comparison to combinations of elevational zonation and performed using elevational differences was also explored. Longitudes and 9999 iterations. latitudes of site locations were entered into PRIMER 6 to construct a Principal components analysis (PCA) ordination. The resulting x and y coordinates were used to produce a RESULTS between-site distance matrix based on measures of Euclidean distance. Mantel tests (available as the RELATE function of PRIMER 6) calculated correlation coefficients and associ- Vegetation ated probabilities between the distance matrices of the moth assemblages and site locations. This same analysis was then A total of 1567 stems with a DBH greater than 5 cm re-run using elevational coordinates. were surveyed along the elevational transect. These doi:10.1111/aec.12272 © 2015 Ecological Society of Australia MOTH ASSEMBLAGES IN A CENTRAL QLD FOREST 5

Table 1. Summary of vegetation and environmental parameters of the Eungella elevational transect

200 m 400 m 600 m 800 m 1000 m 1200 m

Stem count 254 206 285 319 260 243 Species richness 48 42 45 31 43 24 No. of families 24 23 24 17 18 15 Average organic matter (%) 9.9 10.1 12.0 12.2 16.6 15.7 Median temperature (°C) 19.4 18.7 17.7 16.6 15.3 14.5

Fig. 2. Multidimensional scaling (MDS) ordination based on the Euclidean distance index of Eungella vegetation communities. Significance was confirmed using permutational multivariate analysis of variance (PERMANOVA) (P < 0.05). Vegetation communities split similarly to moth assemblages into ‘lowland’ and ‘upland’ assemblages; this is represented by the dotted line. ▲ = 200m asl, ▼ = 400m asl, ■ = 600m asl, ● = 800m asl, ◆ = 1000m asl, ✶ = 1200m asl. consisted of 110 species from 40 families (see Table 1 six adjacent elevations along the elevational gradient in for a summary). Patterns in vegetation communities Eungella. A second sampling event in March 2014 explored using the Euclidian distance index showed a encountered a further 10 045 individuals comprising clear and significant stratification of communities of 607 morpho-species (Table 2). Abundance across elevation (P < 0.001) (Fig. 2). Correlation of coverage-based estimator (ACE) analysis indicated the vegetation data to the spatial distribution of sites that this was approximately 76–80% of the total esti- was determined using the Mantel type Spearman rank mated richness for the region. Individual ACE values correlation and found to be largely independent of of moth fauna at each of the investigated elevations distribution (Rho = 0.42, P < 0.0001). (Table 2) showed a slight (but insignificant), mid elevational peak in species richness. Values of Fisher’s Temperature alpha paralleled this result. Abundance data also showed a mid elevational peak at 800 m before sharply Temperature decreased with increasing elevation; declining. Both high (1000 and 1200 m) and low however, little thermal variation existed between the (200 m) elevations had similar abundances. 200 m asl and 400 m asl sites (Fig. 3, Appendix S2). Microclimate thermal stratification occurring between the canopy and the understory was minimal. Moth assemblages along an elevation gradient

Summary statistics of macro-moths along an Ordination plots of moths sampled in November 2013 elevational gradient and March 2014 (Fig. 4) show clear elevational stratiﬁcation. PERMANOVA analyses indicate a sig- During November 2013, a total of 13 861 individuals niﬁcant relationship between elevation and moth representing 713 morpho-species was sampled from assemblages for both sampling seasons (P < 0.05).

°C) 21

19 Temperature (

15 200m 400m 600m 800m 1000m 1200m Elevaon

Fig. 3. Median understory and canopy temperatures from March 2013 to July 2014. The bars represent error. Under- story = ○, canopy = ●.

Table 2. Summary of catches from combined ground and canopy traps over three sampling nights during (a) November 2013 and (b) March 2014

(a) November 2013

200 m 400 m 600 m 800 m 1000 m 1200 m

Abundance 1591 1845 1258 2723 1553 1075 Richness 284 (9.24) 304 (10.04) 208 (9.81) 221 (9.03) 189 (7.11) 155 (8.15) ACE Mean 389 441 334 388 265 278 Fisher’s Alpha 99.9 103.6 71.0 56.8 56.4 49.7 Erebidae abundance 47 88 35 183 196 148 richness 34 34 16 37 29 18 Geometridae abundance 231 267 235 435 424 257 richness 107 98 63 91 94 65 Noctuidae abundance 192 201 192 549 188 178 richness 64 79 46 74 59 52 Pyraloidea abundance 789 743 338 228 255 135 richness 137 144 73 69 60 32 Other abundance 332 546 458 1328 490 357 richness 144 148 98 115 103 83

(b) March 2014

200 m 400 m 600 m 800 m 1000 m 1200 m

Abundance 1628 2535 2783 3805 1524 1586 Richness 319 (10.82) 370 (11.3) 352 (11.6) 309 (9.93) 221 (8.93) 210 (9.83) ACE Mean 446 557 474 475 322 338 Fisher’s Alpha 118.6 119.3 106.7 79.4 71.8 64.9 Erebidae abundance 106 134 109 427 189 147 richness 30 60 42 51 37 22 Geometridae abundance 257 480 553 495 488 469 richness 94 145 161 108 99 110 Noctuidae abundance 185 289 220 820 157 192 richness 106 133 94 154 83 82 Pyraloidea abundance 728 1065 1223 896 407 512 richness 169 190 165 112 102 67 Other abundance 352 567 678 1167 283 266 richness 127 152 151 110 73 69

Abundance and richness are summarized for ﬁve moth taxa or groups of taxa; Erebidae, Geometridae, Noctuidae, Pyraloidea and ‘Others’. ‘Others’ is the sum total of all target families less the Erebidae, Geometrica, Noctuidae and Pyraloidea. Numbers in brackets are standard deviations of richness. Estimated species richness has been approximated by abundance-based coverage estimator (ACE mean) and Fisher’s Alpha estimates diversity. doi:10.1111/aec.12272 © 2015 Ecological Society of Australia MOTH ASSEMBLAGES IN A CENTRAL QLD FOREST 7

Fig. 4. Multidimensional scaling (MDS) ordination plots based on the Bray–Curtis dissimilarity index of moths assemblages in (a) March 2014 and (b) November 2013. Ground and canopy samples from individual sites were pooled across all three sampling nights to form a single data point. The significance of the observed patterns were confirmed using permutational multivariate analysis of variance (PERMANOVA); (a) P = 0.001, (b) P = 0.001. Super imposed vectors are environmental parameters significantly correlated to assemblage composition (median temperature P < 0.001, organic matter P < 0.005). The dotted line indicates the clear demarcation between ‘lowland’ and ‘upland’ sites for both seasons sampled. ▲ = 200m asl, ▼ = 400m asl, ■ = 600m asl, ● = 800m asl, ◆ = 1000m asl, ✶=1200m asl.

Spatial auto-correlation analyses performed in 0.70 from the IndVal analysis as being sufficiently PRIMER 6 indicate that this distribution of moths is restricted and common at specific elevations to form better explained by changes associated with elevation part of our predictor set (Table 3). In this fashion, 18 (Nov13 Rho = 0.7, Mar14 Rho = 0.7) than by spatial morpho-species were identified as strong indicators of distribution (Nov13 Rho = 0.3, Mar14 Rho = 0.6). elevation during March 2014 (Table 3): four of which Temperature changes associated with elevation were were confined to above 1000 m asl and another 5 identified as the most important environmental vari- which only occurred at 1200 m asl. In November able, particularly for lower elevations. Organic matter 2013, 10 morpho-species were selected via this surpassed median temperature as the most important method (Table 3, Appendix S3). These species are determinant at high elevations.The composition of the nominated to form part of a multi-taxon predictor set vegetation assemblages was also similarly explained by that can be used to assess ecosystem level changes median temperature and organic matter. associated with climate. A more stringent benchmark, The similarity between moth assemblages and veg- such as an IndVal of 0.80, should be used if a more etation communities was lower than expected during robust predictor set is required.These species are illus- both seasons (Nov13 Rho = 0.4, Mar14 Rho = 0.5). trated in Appendix S3. The preliminary study conducted in March 2013 indicated a significantly higher correlation between moth assemblages and vegetation communities DISCUSSION (Mar13Rho = 0.9). Nevertheless, both moth and vegetation assemblages were both best explained by the This study has examined elevation related changes in same two environmental variables: median tempera- moth assemblages along a transect in Eungella ture and organic matter. National Park, Queensland. Characteristic assemblages of moths occurred at each elevation. These can be separated into two clusters: ‘lowland’ (200 m– Moth predictor set 600 m asl) and ‘upland’ (800 m–1200 m asl).We have confirmed this in two seasons (November 2013 and Using the formula of Novotny et al. (2007) (see March 2014) and identified a suite of species with above), ‘common’ morpho-species were defined as indicator potential. those represented by 70 individuals or more. Only morpho-species with 70 or more individuals were retained for the IndVal analysis. As per McGeoch et al. Assemblage stratification (2002), the benchmark of 70% was adopted to desig- nate morpho-species as ‘strong’ indicators. We identi- Our results confirm that elevational gradients are pow- fied those species with a value greater than or equal to erful natural experimental systems in which the

Table 3. Common morpho-species sampled during (a) November 2013 and (b) March 2014 identiﬁed by the IndVal analysis with an indicator value greater than 0.7

(a) November 2013

Family Species IndVal Probability Fig. 200 m 400 m 600 m 800 m 1000 m 1200 m

Noctuidae Caradrina sp. 0.9028 0.0010 D Crambidae Nacoleia oncophragma 0.7373 0.0270 V Crambidae Margarosticha australis 0.9410 0.0002 U Noctuidae Proteuxoa tibiata 0.8006 0.0214 C Nolidae Meceda mansueta 0.8776 0.0214 B Erebidae Scoliacma nana 0.9485 0.0002 G Erebidae Eilema sp. 1.0000 0.0002 H Geometridae Ectropis sp. 0.9253 0.0002 O Notodontidae Syntypistis opaca 0.9776 0.0002 F Crambidae Parotis atlitalis 0.8448 0.0076 P

(b) March 2014

Family Species IndVal Probability Fig. 200 m 400 m 600 m 800 m 1000 m 1200 m

Crambidae Margarosticha australis 0.8710 0.0016 U Erebidae Lambula pristina 0.9299 0.0002 J Noctuidae Hydrillodes lentalis 0.7383 0.0176 E Geometridae Idaea sp. 0.7358 0.0216 N Crambidae Nacoleia oncophragma 0.8241 0.0108 V Erebidae Camptomatyx sp. 0.8500 0.0024 K Geometridae Taxeotis epigaea 0.9375 0.0002 M Erebidae Scoliacma nana 0.9375 0.0002 G Erebidae Eilema sp. 0.7815 0.0064 H Crambidae Hyalobathra crenulata 0.7680 0.0034 R Notodontidae Syntypistis chloropasta 0.7882 0.0002 W Geometridae Polyacme dissimilis 0.7333 0.0010 L Crambidae Palpita horakae 0.7978 0.0002 Q Pyralidae Lacalma sp. 0.8730 0.0048 T Noctuidae Achaea janata 0.7100 0.0432 A Erebidae Hesychopa chionora 0.7379 0.0056 I Crambidae Herpetogramma licarsisalis 0.7873 0.0440 S Notodontidae Syntypistis opaca 0.8119 0.0016 F

These individuals are strongly indicative of elevation and form a multi-taxon predictor set for future monitoring of elevational change within Eungella National Park. ecological responses of changing climates may be reflect the stratification of sets of plant species. This investigated. A set of inter-related environmental vari- association assumes two things. First, that most ables change predictably along elevational gradients moths are more or less host plant specific and, and may well, in turn drive biotic turnover (Raich et al. second that adult moths have low mobility. Existing 2006; Barry 2008; Sun et al. 2009; Lewandowska et al. information on moths and their food-plants suggests 2012). The existence of clearly defined sets of species that most species are restricted to a single genus, or of moths at different elevations suggests that there are small group of genera, of host plants (Common deterministic processes involved in community assem- 1990; Novotny et al. 2003), although there are a bly at different elevations. The challenge is to identify small number which exhibit extreme polyphagy the drivers of the elevation-specific process (or pro- (Robinson et al. 2001). The second assumption of cesses) of assembly. restricted vagility is generally accepted to be true, For the Lepidoptera, larval food plants have been with notable exceptions of some robust larger moths widely promoted as the main determinants of moth and well-known migratory species (Williams 1930; ranges (Novotny et al. 2007; Mattila et al. 2009). 1937; Johnson 1969; Beck and Kitching 2007). This carries the implication that the presence of a Our vegetation surveys indicated a similar split in particular adult moth is an indication that its larval lowland and upland floral communities suggesting food plant is close by (Novotny et al. 2003). Accord- some correlation between moth and vegetation ingly, any stratification in moth assemblages should communities. doi:10.1111/aec.12272 © 2015 Ecological Society of Australia MOTH ASSEMBLAGES IN A CENTRAL QLD FOREST 9

Ecologically speaking, the upper and lower The moth predictor set elevational boundaries of a species distribution is determined, at least in part, by a species’ capacity to The suite of moths nominated here as potentially match its climatic tolerances to the climatic profile of useful indicators of climatic change is tentative. the landscape (Hodkinson et al. 1999; Hodkinson Further, complementary, experimental studies and 2005). In reality, these boundaries are often con- investigations into moth life histories are needed to tracted, or more rarely expanded, to reflect the confirm that these species are indeed proxies for biotic availability of resources to support survival and and environmental shifts.The paucity of data on moth establishment of a population (Barton 1993). life histories currently constrains these demands. Elevation-related changes in the landscape matrix, General physiological considerations suggest that such as the loss of habitat surface area and changes to moths, either directly or through their food plants, are the environment both in terms of its physicochemical likely to be sensitive to environmental changes, and properties and available resources, influence the pres- these responses may serve as analogues for climate ence and survivorship of species at different elevations change (Janzen 1967; Addo-Bediako et al. 2000; (Damken et al. 2012). Such changes doubtlessly Deutsch et al. 2008). The indicator species identified underlie the transition of Eungella’s rainforest from here demonstrate high levels of fidelity to particular mesophyll forest at low elevations to complex elevational zones suggesting a high future probability notophyll vine forest at high elevations. This, in turn, of repeated capture at current (or shifted) elevational will drive the turnover of moths across elevation bands during future sampling efforts. Those species through the food plant mechanism previously restricted to upper elevations may be vulnerable to discussed. local extinctions under future warming scenarios as no The observed moth and vegetation communities at local higher elevational destinations are available to Eungella cannot be wholly understood without con- them (see Ashton et al. 2011; Kitching & Ashton sidering the regions biogeographic history. Little is 2013).The highest elevations will likely become occu- known about the specific biogeographic history of pied by ‘refugee’ species migrating from lower eleva- Eungella, and subsequently, how this biogeographic tions in an attempt to match their thermal tolerances history has affected the evolution, diversity and distri- and environmental needs with the climatic profile of bution of flora and fauna in this region. We can, the landscape. More challenging, however, is any however, infer general information from broader scale attempt to predict the future composition of the biotic biogeographic historical information, such as that communities at the lowest elevations. It is unknown available about the eastern Queensland coast and which species will tolerate or adapt to the new and neighbouring rainforests. During the Quaternary, Aus- unfamiliar environmental conditions. Here we may tralia experienced several glacial and interglacial expect to find novel ecosystems (Hobbs et al. 2009) to periods (Williams et al. 2006), which, at times, gave arise of currently unknown (and unpredictable) com- rise to land bridges and fragmented landscapes. Rain- position and functionality. forests which previously covered the Australian land- The results described here mirror those of other scape retracted heavily during glacial periods giving plant and arthropod taxa sampled during this study rise to more arid habitats (Kershaw et al. 2007). and will be published in due course. These results Several regions of rainforest, however, persisted along should be combined with the other taxonomic groups eastern Queensland (Macphail 1997). Paleontological to develop a multi taxon predictor set which should be records from other eastern Queensland rainforests used, and may be more powerful, for the future moni- indicate a dramatic increase in rainforest flora and toring of ecological changes in composition and fauna in the late Miocene suggesting these remaining functioning. The specificity of species in conjunction patches of rainforest acted as refugia (Hekel 1972; with the high degree of variability in climatic and Hocknull 2005). During interglacial periods, however, environmental conditions across landscapes limits this these rainforests expanded and reconnected (Kershaw set of nominated indicator species to use within et al. 2007).These series of events may have acted as a Eungella National Park. species pump and facilitated the assembly of different This study represents the fourth elevational transect communities across elevation. Furthermore, fossil in a much larger body of cross latitudinal work sam- records from south of Eungella at Mt Etna show bio- pling moth fauna in Queensland (see Ashton et al. geographic links to Papua New Guinea suggesting a 2011; Ashton et al. this volume; Ashton 2013) and mixing of species throughout the central and northern elsewhere. The results presented here contribute to Queensland coast (Hocknull 2005). These historical necessary baseline data of lepidopteran assemblages changes in environmental conditions and the refugia across elevation in Eungella National Park, Queens- provided by these forests offer some explanation for land, Australia. Future monitoring programmes and the observed patterns in species richness and commu- additional complementary research should expand on nity assembly at Eungella. and draw information from the results of this

© 2015 Ecological Society of Australia doi:10.1111/aec.12272 10 E. H. ODELL ET AL. investigation. Baseline datasets such as these are essen- Brose U. & Martinez N. (2004) Estimating the richness of tial tools for the accurate identification of future shifts species with variable mobility. Oikos 105, 292–300. in local ecological regimes. Burkle L. A., Marlin J. C. & Knight T. M. (2013) Plant- pollinator interactions over 120 years: loss of species, co-occurrence, and function. Science 339, 1611–15. ACKNOWLEDGEMENTS Carignan V. & Villard M.-A. (2002) Selecting indicator species to monitor ecological integrity: a review. Environ. Monit. Assess. 78, 45–61. This study was funded by the Mackay Regional City Chao A., Chazdon R. L., Colwell R. K. & Shen T.-J. (2005) A Council and conducted in conjunction with Griffith new statistical approach for assessing similarity of species University School of Environment and Environmen- composition with incidence and abundance data. Ecol. Lett. tal Futures Research Institute. Countless thanks are 8, 148–59. owed to all associated scientists and community Chao A., Hwang W.-H., Chen Y.-C. & Kuo C.-Y. (2000) Esti- participants for volunteering their time to this mating the number of shared species in two communities. Stat. Sinica 10, 227–46. project. Special recognition and appreciation is owed Chen I.-C., Shiu H.-J., Benedick S. et al. (2009) Elevation to Dr. Bill MacDonald for his generous botanical increases in moth assemblages over 42 years on a tropical contributions. mountain. Proc. Natl. Acad. Sci. U S A 106, 1479–83. Clarke K. & Gorley R. (2006) Primer v.6 – user manual/tutorial. Primer E Ltd. Plymouth. REFERENCES Colwell R. K., Chao A., Gotelli N. J. et al. (2012) Models and estimators linking individual-based and sample-based rar- Addo-Bediako A., Chown S. L. & Gaston K. J. (2000) Thermal efaction, extrapolation, and comparison of assemblages. tolerance, climatic variability and latitude. Proceed. R. Soc. J. Plant Ecol. 5, 3–21. Lond. Ser. B Biol. Sci. 267, 739–45. Common I. F. B. (1990) The Moths of Australia. CSIRO Publi- Ashton L. (2013) Moths and mountains: diversity, altitude and cations, Collingwood, Victoria. latitude (PhD thesis). Griffith University. Cook B. I., Cook E. R., Huth P. C., Thompson J. E., Forster A. Ashton L. A., Kitching R., Maunsell S. C., Bito D. & Putland D. & Smiley D. (2008) A cross-taxa phenological dataset from A. (2011) Macrolepidopteran assemblages along an Mohonk Lake, NY and its relationship to climate. Int. J. altitudinal gradient in subtropical rainforest – exploring Climatol. 28, 1369–83. indicators of climate change. Mem. Queensl. Mus., Nat. 55, Damken C., Perry G. & Beggs J. R. (2012) Complex habitat 375–89. changes along elevational gradients interact with resource Axmacher J. C. & Fiedler K. (2008) Habitat type modifies geom- requirements of insect specialist herbivores. Ecosphere 3, etry of elevational diversity gradients in geometrid moths 123. (Lepidoptera Geometridae) on Mt Kilimanjaro, Tanzania. Deutsch C. A., Tewksbury J. J., Huey R. B. et al. (2008) Impacts Trop. Zool. 21, 243–51. of climate warming on terrestrial ectotherms across latitude. Barry R. G. (2008) Mountains Weather and Climate. Cambridge Proc. Natl. Acad. Sci. U S A 105, 6668–72. University Press, Cambridge, UK. Diaz H., Grosjean M. & Graumlich L. (2003) ClimateVariability Barton A. M. (1993) Factors controlling plant distributions: and Change in High Elevation Regions: Past, Present and drought, competition, and fire in montane pines in Arizona. Future. Clim. Change 59, 1–4. Ecol. Monogr. 63, 367–97. EHMP (2008) Ecosystem Health Monitoring Program annual tech- Basset Y., Aberlenc H.-P. & Delvare G. (1992) Abundance and nical report (eds P. Maxwell, P. Hough, B. Mayger et al.) stratification of foliage arthropods in a lowland rain forest of South East Queensland Healthy Waterways Partnership, Cameroon. Ecol. Entomol. 17, 310–18. Brisbane. Beck J. & Kitching I. J. (2007) Correlates of range size and Gerlach J., Samways M. & Pryke J. (2013) Terrestrial inverte- dispersal ability: a comparative analysis of sphingid moths brates as bioindicators: an overview of available taxonomic from the Indo-Australian tropics. Global Ecol. Biogeogr. 16, groups. J. Insect. Conserv. 17, 831–50. 341–9. Gerstman B. B. (2003) StatPrimer (Version 6.4). San Jose State Bellard C., Bertelsmeier C., Leadley P., Thuiller W. & University, San Jose. Courchamp F. (2012) Impacts of climate change on the Graham A. W. (2006) The CSIRO Rainforest Permanent Plots of future of biodiversity. Ecol. Lett. 15, 365–77. North Queensland: Site, Structural, Floristic and Edaphic Blattenberger M., Knepp R., Lingle E., Malar A. & Mullins N. Descriptions. Cooperative Research Center forTropical Rain- (2000) Diversity of E.P.T. taxa in relation to land use along forest Ecology and Management, Cairns. Great Through Creek. J. Ecol. Res. 2, 47–52. Hekel H. (1972) Pollen and Spore Assemblages from Queensland Bohac J. (1999) Staphylinid beetles as bioindicators. Agric., Tertiary Sediments. Queensland Geological Survey, Brisbane. Ecosyst. Environ. 74, 357–72. Hobbs R. J., Higgs E. & Harris J. A. (2009) Novel ecosystems: Brehm G. (2007) Contrasting patterns of vertical stratification in implications for conservation and restoration. Trends Ecol. two moth families in a Costa Rican lowland rain forest. Basic Evol. 24, 599–605. Appl. Ecol. 8, 44–54. Hocknull S. (2005) Ecological succession during the late Brehm G. & Fiedler K. (2005) Diversity and community struc- Cainozoic of central eastern Queensland: extinction of a ture of geometrid moths of disturbed habitat in a montane diverse rainforest community. Mem. Queensl. Mus., Nat. 51, area in the Ecuadorian Andes. J. Res. Lepid. 38, 1–14. 39–122. Brehm G., Süssenbach D. & Fiedler K. (2003) Unique Hodkinson I. D. (2005) Terrestrial insects along elevation gra- elevational diversity patterns of geometrid moths in an dients: species and community responses to altitude. Biol. Andean montane rainforest. Ecography 26, 456–66. Rev. 80, 489–513. doi:10.1111/aec.12272 © 2015 Ecological Society of Australia MOTH ASSEMBLAGES IN A CENTRAL QLD FOREST 11

Hodkinson I. D., Bird J., Miles J. E., Bale J. S. & Lennon J. J. Environment (CDE), Geographica Bernensia, Swiss Agency (1999) Climatic signals in the life histories of insects: the for Development and Cooperation (SDC), Bern. distribution and abundance of heather psyllids (Strophingia Körner C. (2007) The use of ‘altitude’ in ecological research. spp.) in the UK. Funct. Ecol. 13, 83–95. Trends Ecol. Evol. 22, 569–74. Huddleston N. (2012) Climate Change: Evidence, Impacts & Lewandowska A. M., Breithaupt P., Hillebrand H., Hoppe Choices. National Research Council, National Academy of H.-G., Jürgens K. & Sommer U. (2012) Responses of Science, Washington, DC. primary productivity to increased temperature and phyto- Hughes L. (2000) Biological consequences of global warming: is plankton diversity. J. Sea Res. 72, 87–93. the signal already apparent? Trends Ecol. Evol. 15, 56–61. McGeoch M. A., Van Rensburg B. J. & Botes A. (2002) The Hughes L. (2003) Climate change and Australia: trends, projec- verification and application of bioindicators: a case study of tions and impacts. Austral Ecol. 28, 423–43. dung beetles in a savanna ecosystem. J. Appl. Ecol. 39, 661– Hughes L. (2012) Climate change impacts on species interac- 72. tions: assessing the threat of cascading extinctions. In: Macphail M. K. (1997) Late neogene climates in Australia: fossil Saving a Million Species (ed. L. Hannah) pp. 337–59. Island pollen- and spore-based estimates in retrospect and Press and the Center for Resource Economics, Washington prospect. Aust. J. Botany 45, 425–64. DC. Majer J. D., Kitching R. L., Heterick B. E., Hurley K. & Brennan IPCC (2013) The Physical Science Basis. Contribution of Working K. E. C. (2001) North-south patterns within arboreal ant Group I to the Fifth Assessment Report of the Intergovernmental assemblages from rain forests in Eastern Australia. Panel on Climate Change (eds T. F. Stocker, D. Qin, G.-K. Biotropica 33, 643–61. Plattner et al.) p. 1535. Cambridge University Press, Cam- Maleque M. A., Maeto K. & Ishii H. T. (2009) Arthropods as bridge, UK and New York, NY, USA. bioindicators of sustainable forest management, with a IPCC (2014) Climate C\Change 2014: Impacts, Adaptation, and focus on plantation forests. Appl. Entomol. Zool. (Jpn.) 44, Vulnerability. Contribution of Working Group II to the Fifth 1–11. Assessment Report of the Intergovernmental Panel on Climate Mattila N., Kotiaho J. S., Kaitala V., Komonen A. & Päivinen J. Change (eds C. B. Field, V. R. Barros, D. J. Dokken et al.). (2009) Interactions between ecological traits and host plant Cambridge University Press, Cambridge, UK and New type explain distribution change in noctuid moths. Conserv. York, NY, USA. Biol. 23, 703–9. Janzen D. (1967) Why mountain passes are higher in the tropics. Møller A. P., Rubolini D. & Lehikoinen E. (2008) Populations of Am. Nat. 101, 233–50. migratory bird species that did not show a phenological Johnson C. G. (1969) Migration and dispersal of insects by flight. response to climate change are declining. Proc. Natl. Acad. Methuen and Co. Ltd, London. Sci. U S A 105, 16195–200. Kempton R. A. & Taylor L. R. (1976) Models and statistics for Nadkarni N. M., Lawton R. O., Clark K. L., Matelson T. J. & species diversity. Nature 262, 818–20. Schaefer D. (2000) Ecosystem ecology and forest dynamics. Kershaw A. P., Bretherton S. C. & van der Kaars S. (2007) A In: Monteverde: Ecology and Conservation of a Tropical Cloud complete pollen record of the last 230 ka from Lynch’s Forest (eds N. M. Nadkarni & N. T. Wheelwright) pp. 303– Crater, north-eastern Australia. Palaeogeogr. Palaeoclimatol. 50. Oxford University Press, Oxford. Palaeoecol. 251, 23–45. Nakazawa T. & Doi H. (2012) A perspective on match/mismatch Kiers T. E., Palmer T. M., Ives A. R., Bruno J. F. & Bronstein J. of phenology in community contexts. Oikos 121, 489–95. L. (2010) Mutualisms in a changing world: an evolutionary National Research Council (2008) Ecological Impacts of Climate perspective. Ecol. Lett. 13, 1459–74. Change. The National Academies Press, Washington, DC. Kitching R. & Ashton L. A. (2013) Predictor sets and biodiver- Niemelä J., Kotze J., Ashworth A. et al. (2000) The search for sity assessment: the evolution and application of an idea. common anthropogenic impacts on biodiversity: a global Pac. Conserv. Biol. 19, 418–26. network. J. Insect. Conserv. 4, 3–9. Kitching R. L., Bickel D., Creagh A. C., Hurley K. & Symonds Niemi G. J. & McDonald M. E. (2004) Application of ecological C. (2004) The biodiversity of Diptera in Old World rain indicators. Ann. Rev. Ecol., Evol. Syst. 35, 89–111. forest surveys: a comparative faunistic analysis. J. Biogeogr. Nogués-Bravo D., Araújo M. B., Errea M. P. & Martínez-Rica J. 31, 1185–200. P. (2007) Exposure of global mountain systems to climate Kitching R. L., Boulter S. L., Vickerman G., Laidlaw M. J., warming during the 21st century. Glob. Environ. Change 17, Hurley K. L. & Grimbacher P. S. (2005) The Comparative 420–8. Assessment of Arthropod and Tree Biodiversity in Old-World Novotny V., Basset Y., Scott E. M. & Kitching R. L. (2003) Rainforests. Cooperative Research Centre for Tropical Rain- Herbivore assemblages and their food resources. In: Arthro- forest Ecology and Management, Cairns. pods of Tropical Forests. Cambridge University Press, Cam- Kitching R. L., Orr A. G., Thalib L., Mitchell H., Hopkins M. S. bridge, pp. 40–53. & Graham A. W. (2000) Moth assemblages as indicators of Novotny V., Miller S. E., Hulcr J. et al. (2007) Low beta diversity environmental quality in remnants of upland Australian rain of herbivorous insects in tropical forests. Nature 448, 692–5. forest. J. Appl. Ecol. 37, 284–97. Oosterhoorn M. & Kappelle M. (2000) Vegetation structure and Kitching R., Li D. & Stork N. (2001) Assessing biodiversity composition along an interior-edge-exterior gradient in a ‘sampling packages’: how similar are arthropod assemblages Costa Rican montane cloud forest. For. Ecol. Manage. 126, in different tropical rainforests? Biodiv. Conserv. 10, 793– 291–307. 813. Parmesan C. (2007) Influences of species, latitudes and meth- Kitching R., Putland D., Ashton L. A. et al. (2011) Detecting odologies on estimates of phenological response to global biodiversity changes along climatic gradients: the IBISCA- warming. Glob. Change Biol. 13, 1860–72. Queensland Project. Mem. Queensl. Mus., Nat. 55, 235–50. Rafferty N. E. & Ives A. R. (2011) Effects of experimental shifts Kohler T. & Maselli D. (2009) Mountains and Climate Change: in flowering phenology on plant-pollinator interactions. From Understanding to Action. Centre for Development and Ecol. Lett. 14, 69–74.

Raich J.W., Russell A. E., Kitayama K., Parton W. J. & Vitousek Williams C. B. (1930) The migration of butterflies. Oliver and P. M. (2006) Temperature influences carbon accumulation Boyd, Edinburgh and London. in moist tropical forests. Ecology 87, 76–87. Williams C. B. (1937) The migrations of day-flying moths of the Rainio J. & Niemelä J. (2003) Ground beetles (Coleoptera: genus Urania in tropical America. Proceedings of the Royal Carabidae) as bioindicators. Biodivers.Conserv. 12, 487–506. Entomological Society of London Series A, General Entomology Rayment G. E. & Higginson F. R. (1992) Australian Laboratory 12, 141–7. Handbook of Soil and Water Chemical Methods. Inkata Press, Williams S. E., Bolitho E. E. & Fox S. (2003) Climate change in Melbourne. Australian tropical rainforests: an impending environmental Robinson G. S., Ackery P. R., Kitching I. J., Beccaloni G. W. & catastrophe. Proceed. R. Soc. Lond. Seri. B Biol. Sci. 270, Hernandez L. M. (2001) Hostplants of the Moth and Butterfly 1887–92. Caterpillars of the Oriental Region. Natural History Museum Williams-Linera G. (2002) Tree species richness complementa- in association with Southdene Sdn Bhd, Oxford. rity, disturbance and fragmentation in a Mexican tropical Root T. L., Price J. T., Hall K. R., Schneider S. H., Rosenzweig montane cloud forest. Biodiver. Conserv. 11, 1825–43. C. & Pounds J. A. (2003) Fingerprints of global warming on Williams N. J., Harle K. J., Gale S. J. & Heijnis H. (2006) The wild animals and plants. Nature 421, 57–60. vegetation history of the last glacial–interglacial cycle in RStudio (2013) RStudio: Integrated Development Environment for eastern New South Wales, Australia. J Quaternary Sci. 21, R (Version 0.97). RStudio, Boston. 735–50. Schulze C. H., Linsenmair K. E. & Fiedler K. (2001) World Bank (2012) Turn down the heat: why a 4°C warmer Understorey versus canopy: patterns of vertical stratification world must be avoided. In: A report for theWorld Bank by the and diversity among Lepidoptera in a Bornean rain forest. Potsdam Institute for Climate Impact Research and Climate Plant Ecol. 153, 133–52. Analytics (eds H. J. Schellnhuber, W. Hare, O. Serdeczny Shaver G. R., Canadell J., Chapin F. S. et al. (2000) Global et al.) International Bank for Reconstruction and Develop- warming and terrestrial ecosystems: a conceptual framework ment and The World Bank, Washington, DC. for analysis. Bioscience 50, 871–82. Wright I., Davies P., Wilks D., Findlay S. & Taylor M. (2007) Steffen W., Burbidge A. A., Hughes L. et al. (2009) Australia’s Aquatic macroinvertebrates in urban waterways: comparing Biodiversity and Climate Change. CSIRO Publishing, ecosystem health in natural reference and urban streams. In Collingwood. Proceedings of the 5th Australian Stream Management Sun B., Zheng X. Q., Hu F. et al. (2009) Effect of temperature, Conference Australian rivers: making a difference. Charles rainfall and soil properties on farmland soil nitrification. Stuart University, Thurgoona. New South Wales. Huan Jing Ke Xue 30, 206–13. Thackeray S. J., Sparks T. H., Frederiksen M. et al. (2010) Trophic level asynchrony in rates of phenological change for marine, freshwater and terrestrial environments. Glob. SUPPORTING INFORMATION Change Biol. 16, 3304–13. Thomas C. D., Cameron A., Green R. E. et al. (2004) Extinction Additional Supporting Information may be found in risk from climate change. Nature 427, 145–6. the online version of this article at the publisher’s Trindade-Filho J. & Loyola R. D. (2011) Performance and con- web-site: sistency of indicator groups in two biodiversity hotspots. PLoS ONE 6, e19746. Appendix S1. Locality and elevation of each study Van Asch M., Van Tienderen P. H., Holleman L. J. M. & Visser site. M. E. (2007) Predicting adaptation of phenology in Appendix S2. Daily understory temperatures at each response to climate change, an insect herbivore example. Glob. Change Biol. 13, 1596–604. elevation. Webb L. J. (1978) A general classification of Australian Appendix S3. Examples of elevationally restricted rainforests. Aust. Plants 9, 349–63. moth species collected along the Eungella transect.