英文摘要: | There is little consensus as to why there is so much variation in the rates at which different species’ geographic ranges expand in response to climate warming1, 2. Here we show that the relative importance of species’ abundance trends and habitat availability for British butterfly species vary over time. Species with high habitat availability expanded more rapidly from the 1970s to mid-1990s, when abundances were generally stable, whereas habitat availability effects were confined to the subset of species with stable abundances from the mid-1990s to 2009, when abundance trends were generally declining. This suggests that stable (or positive) abundance trends are a prerequisite for range expansion. Given that species’ abundance trends vary over time3 for non-climatic as well as climatic reasons, assessment of abundance trends will help improve predictions of species’ responses to climate change, and help us to understand the likely success of different conservation strategies for facilitating their expansions.
Identifying species’ traits associated with rapid range expansions in response to climate change provides insight into the conservation strategies most likely to be successful4. However, such understanding may be difficult to attain, given that the ability of species’ traits, such as reproductive rate, to explain responses to climate change is frequently low2. Previous studies suggest that the expansion of species’ distributions across landscapes will depend on species’ dispersal abilities1, 5, 6, the availability of habitat7 and population abundance trends, which determine the supply of migrants to colonize new locations8. Species’ population and distribution trends will also be affected by interactions between traits and the environment, thus predictions of range expansions may be limited if habitat availability and population trends are not considered simultaneously. Furthermore, abundance trends vary over time3, associated with variability in climate warming9 and habitat quality and quantity10, so it might be expected that the relative importance of predictors of distribution changes also vary over time. Here, we consider the roles of abundance trends, habitat availability and dispersal capacity in the range changes of 25 British butterfly species during two periods. Distribution changes were measured between blocks of time (1970–1982 to 1995–1999 and then 1995–1999 to 2005–2009) to ensure sufficient data to record distribution changes in a robust manner (1970–1982, 1995–1999 and 2005–2009 represent periods with intensive recording; >1,220,000 distribution records and >262,000 abundance transect records). Butterflies are an ideal group for this analysis. Not only are there more long-term species-specific data sets than any other poikilothermic animal group worldwide, but most between-species variation in expansion rates exists within taxonomic groups rather than between groups9 and so our conclusions are likely to be relevant to other taxa. Average annual temperature increased at a rate of 0.03 °C yr−1 in the first study period (1970–1982 to 1995–1999) and 0.01 °C yr−1 in the second (1995–1999 to 2005–2009). We expected the lower rate of temperature increase in the second period to have relatively little effect on rates of distribution change owing to climate distribution lags11, 12 and indeed species showed idiosyncratic responses to climate warming; some expanded their ranges in both periods, some in only one period and some retracted in both periods13 (Supplementary Table 1). We studied 25 southerly distributed butterfly species that have the potential to extend their distributions under climate change (migrants, northern and ubiquitous species were excluded, further exclusions were due to insufficient data). We quantified changes in distribution area using the Butterflies for the New Millennium (BNM) data set14 as the percentage change in the number of 10 km grid squares occupied per year, to account for the different lengths of study periods and different initial species’ range sizes. Changes in abundance were calculated using the UK Butterfly Monitoring Scheme (UKBMS) transect data set15 by regressing abundance indices from continuously occupied transect sites (sites at which a species was present every year during the study period) against year16, to give percentage change in abundance per year for each species. We used a rank mobility score17 to represent species’ dispersal ability (derived from expert opinion). Habitat availability was calculated by combining remote-sensed land cover18, 19 estimates with expert assessments of species’ habitat associations14 (see Supplementary Information). We considered only the availability of habitat in the 10 km grid squares that the species colonized during each period, thus focussing measures on those areas where species’ distributions were changing. It was not possible to quantify landscape change over time because annual habitat data are not available and the categorization of land-cover data in the two study periods has changed18. We employed an information-theoretic approach to identify the best models for explaining distribution changes. For each study period separately, we constructed general linear models to assess distribution changes against all three variables (abundance trends, habitat availability, dispersal ability) including their interactions, and Akaike information criterion (AICc) values and Akaike weights were used to determine the best fitting models. When ΔAICc<2, models are considered to be of equal strength20 so model averaging was used. (Incorporation of phylogenetic correlations did not improve the fit of models so we do not present phylogenetic analyses; see Supplementary Information.) In the earlier period, nine species expanded their distribution area (mean change = 0.8% yr−1±0.1 s.e.m.) and 16 species retracted (mean change = −2% yr−1±0.2 s.e.m.). The abundance trends of species were generally stable in permanently occupied sites (mean abundance change = −0.5% yr−1±1.75 s.e.m.). The best fitting models included habitat availability and dispersal ability, but not abundance (Table 1a). Habitat availability was the most important explanatory variable (R2 = 0.35, Supplementary Table 4a); range expansions were greatest for species with high habitat availability (Fig. 1a). Dispersal ability was much less important and in models where it was included it showed a negative relationship. This unexpected relationship suggests that once habitat availability was accounted for, less dispersive species did not fare any worse than more dispersive species.
- Mattila, N., Kaitala, V., Komonen, A., PäIvinen, J. & Kotiaho, J. S. Ecological correlates of distribution change and range shift in butterflies. Insect Conserv. Divers. 4, 239–246 (2011).
- Angert, A. L. et al. Do species’ traits predict recent shifts at expanding range edges? Ecol. Lett. 14, 677–689 (2011).
- Brereton, T., Roy, D. B., Middlebrook, I., Botham, M. & Warren, M. The development of butterfly indicators in the United Kingdom and assessments in 2010. J. Insect Conserv. 15, 139–151 (2011).
- Arribas, P. et al. Evaluating drivers of vulnerability to climate change: a guide for insect conservation strategies. Glob. Change Biol. 18, 2135–2146 (2012).
- Warren, M. S. et al. Rapid responses of British butterflies to opposing forces of climate and habitat change. Nature 414, 65–69 (2001).
- Gaston, K. J. & Blackburn, T. M. Large-scale dynamics in colonization and extinction for breeding birds in Britain. J. Anim. Ecol. 71, 390–399 (2002).
- Hill, J. K. et al. Responses of butterflies to twentieth century climate warming: Implications for future ranges. Proc. R. Soc. Lond. B 269, 2163–2171 (2002).
- Newton, I. Links between the abundance and distribution of birds. Ecography 20, 137–145 (1997).
URL:
| http://www.nature.com/nclimate/journal/v4/n2/full/nclimate2086.html
|