Temperature-Dependent Alterations in Host Use Drive Rapid Range Expansion in a Butterfly

See allHide authors and affiliations

Science  25 May 2012:
Vol. 336, Issue 6084, pp. 1028-1030
DOI: 10.1126/science.1216980


Responses of species to climate change are extremely variable, perhaps because of climate-related changes to interactions among species. We show that temperature-related changes in the dependence of the butterfly Aricia agestis on different larval host plants have facilitated rapid range expansion. Historically, the butterfly was largely restricted to a single plant species, Helianthemum nummularium, but recent warmer conditions have enabled the butterfly to increasingly use the more widespread plant species Geranium molle. This has resulted in a substantial increase in available habitat and rapid range expansion by the butterfly (79 kilometers northward in Britain in 20 years). Interactions among species are often seen as constraints on species’ responses to climate change, but we show that temperature-dependent changes to interspecific interactions can also facilitate change.

Many species are altering their ranges in response to climate warming (1), but patterns of expansion vary greatly among species (2, 3). Some species have retreated where they might have been expected to expand, whereas others have expanded considerably faster than expected, based on the rate of climate change (3). There are many potential explanations for variation in patterns of range change, including habitat availability (4), land-use change (which can cause retractions), and dispersal ability (2). Laboratory and modeling studies also suggest that altered interspecific interactions could represent a major source of variation in determining range changes (510). Specifically, such interactions with other species are most commonly regarded as additional constraints because they may limit species to a narrower set of physical conditions (and hence narrower geographic ranges) than their fundamental, climatic niches might otherwise allow. Here, we provide field evidence of responses within individual populations and across geographic ranges to show that changing biotic interactions can accelerate rather than constrain distribution change.

Our study species is the brown argus butterfly, Aricia agestis, which has spread northward in Great Britain by ~79 km in 20 years (Fig. 1A) (11), which is 2.3 times faster than the average expansion rate documented for species globally (median 16.9 km per decade) (3). The butterfly reaches its northern range boundary in Britain, where it was scarce and declining in the 1980s (12) and largely restricted to using rockrose (Cistaceae, Helianthemum nummularium) as its larval host plant, in calcareous grasslands (13). Since then, the butterfly has rapidly extended its distribution northward (Fig. 1A) (14, 15), which is unusual for a species previously considered to be relatively specialized and sedentary (2), and colonized large areas where rockrose is absent. We report here how temperature-dependent changes in the butterfly’s association with a different larval host plant species underpin this unexpectedly rapid range expansion.

Fig. 1

Distribution and density changes of the brown argus butterfly. (A) Occurrence of brown argus in 10 × 10 km grid squares that contain rockrose (red) and Geraniaceae only (blue). Records from the period 1970–1987 [before the increase in use of Geraniaceae (B)] are in dark shades; new 10 × 10 km squares colonized in the period 1988–2009 are in light shades. (B) Increase in the fraction of brown argus distributional records from 10 × 10 km grid squares where only Geraniaceae hosts are present. (C) Mean annual density (count per kilometer) of brown argus in rockrose (solid line, solid symbols) and Geraniaceae (dashed line, open symbols) sites from 1976–2009. (D) Same as for (C), but separating 1995 to 2009 population counts into numbers for first (squares) and second (circles) adult flight periods each year. Photographs are (E) Brown argus [Jim Asher/Butterfly Conservation]; (F) rockrose [Rachel Pateman]; and (G) dove’s-foot cranesbill [Alison Jukes].

Plant species in the Geraniaceae family are used by the butterfly in continental Europe (16) but were rarely used historically in Britain (13), despite them being widespread. Using data on the distribution of brown argus collected by amateur butterfly recorders over the past four decades, we found that during the 1990s there was an increase in the fraction of all occurrences of the butterfly in locations where the only available host plants were in the family Geraniaceae (Spearman’s rank correlation between year and fraction of 10 × 10 km resolution observation records in Geraniaceae-only areas: rs = 0.82, n = 35 years, P < 0.001) (Fig. 1B) (11). This has been achieved through an increase in occurrence in Geraniaceae-only areas because occurrence on rockrose does not appear to have declined. Increases in the occurrence of brown argus in Geraniaceae-only areas were associated with warm summers [linear regression of the between-year change in the fraction of brown argus records in Geraniaceae-only grid squares versus mean summer temperature (MST): equation 1, change = (0.03 × MST) − 0.47; Pearson r = 0.45, F1,32 = 8.35, P = 0.007] (11).

To establish the population-dynamic basis for this distributional change in host plant use, we analyzed the effect of climate on brown argus populations associated with different larval host plants, on the basis of count data by volunteers from over 200 fixed transects in Britain (17). Warmer summers result in higher brown argus population densities on both rockrose and Geraniaceae [analysis of covariance, relationship between annual population growth rate and MST, with host as categorical variable: F1,35 = 6.85, P = 0.013; equation 2, population growth on rockrose = (0.21 × MST) − 3.30; equation 3, population growth on Geraniaceae = (0.33 × MST) − 5.18], with no effect of mean annual temperature and little or no effect of winter temperature (11). Thus, butterfly populations on both host plant types have benefited from the recent increase in frequency of warm summers.

We used the relationships described above between temperature and butterfly distribution (equation 1) and abundance (equations 2 and 3) to hindcast changes in the relative use of different host plants (from distribution data) and in relative annual population growth rates on different host plants (from transect counts) for the 19th and 20th centuries in Britain, the period for which fine-scale historical climate data are available (Fig. 2) (11). The climate over most of this period was cooler than at present, and our models predict a retraction in the distribution of brown argus away from Geraniaceae in most decades and lower relative population growth rates on Geraniaceae. Both the distribution and population growth rate analyses indicate that the rockrose host plant was more favorable than was Geraniaceae under these cooler conditions. Historical records of brown argus support this prediction, with only 19% of occurrences in Geraniaceae-only areas during the cool beginning of the 20th century (1900–1929), increasing to 30% during the warmer 1930s and 1940s and dropping back down to 16% during the cooler period from 1950 to 1989 (insufficient records are available from the 19th century to assess the relative occurrence in Geraniaceae-only areas) (11). In the past two decades, as the frequency of warm summers has increased, our models (equations 1 to 3) lead to the expectation that brown argus will have experienced higher population performance on Geraniaceae than on rockrose and have expanded in its distribution onto Geraniaceae (Fig. 2), as has been observed (Fig. 1).

Fig. 2

Estimated decadal net increase (positive) or decrease (negative) in the fraction of all brown argus occurrences associated with Geraniaceae sites in the past (solid line, solid circles). Difference in estimated annual population growth rates, averaged across decades, between Geraniaceae and rockrose sites (mean for Geraniaceae sites minus mean for rockrose sites) (long dashed line, open circles; positive values indicate higher relative population growth on Geraniaceae, and negative values indicate higher performance on rockrose). MST for each decade (short red dashed line and triangles). Dates on scale bar refer to first year of each decade for which estimates have been calculated.

The brown argus is likely to have survived past, cooler periods predominantly as localized populations in warm sites that contain large rockrose populations. Rockrose achieves high local densities compared with the annual dove’s-foot cranesbill (Geranium molle) (18, 19), the most frequently used of the Geraniaceae host plants (15) [rockrose averages 23 times greater ground cover than dove’s-foot cranesbill per 100 × 100 m sample area that contains host plants (fig. S1) (11)]. The long-lived perennial rockrose also has more stable populations than does the annual and ruderal dove’s-foot cranesbill. These differences between the plants enable rockrose to support larger (Fig. 1, C and D) and more stable brown argus populations [Levene’s test for equality of variances: W = 10.97, n = 2 30-generation sequences, P = 0.002 (Fig. 1D)]. Moreover, rockrose frequently grows in areas of short turf on southerly facing slopes (fig. S2) (11, 20), which provide warm microclimates [southerly aspects receive greater direct radiation and achieve higher maximum summer temperatures (21)]. As recently as the early 1980s, the brown argus was mainly associated with rockrose populations on sheltered south-facing slopes (12). The few historical records of Geraniaceae-feeding populations from this period were predominantly in sand dunes (13), which also provide warm microclimates.

Summer temperatures in Britain from 1990 to 2009 were on average 0.78°C warmer than between 1800 and 1989, and this is likely to have increased the thermal suitability of sites for brown argus, especially those that are not southerly facing. This would have increased the ability of Geraniaceae-containing sites to support brown argus population growth; there was a 5.3-fold increase in brown argus population density in Geraniaceae sites between 1976–1985 and 2000–2009 (Spearman’s rank correlation between year and density on Geraniaceae: rs = 0.76, n = 34 years, P < 0.001) (Fig. 1C). In contrast, no increase in overall population density occurred at rockrose sites (Spearman’s rank correlation between year and density on rockrose: rs = 0.25, n = 34 years, P = 0.162; 1.1-fold density increase from 1976–1985 to 2000–2009) (Fig. 1C), even though butterfly abundance increased temporarily during warm summers. This suggests that other factors limit population density on rockrose (supplementary text).

Based on 100 × 100 m grid squares with records of host plants, dove’s-foot cranesbill is 4 to 17 times more widespread than is rockrose in counties where rapid expansion has taken place (Fig. 3) (11). Once the brown argus can establish populations on cranesbill, the high frequency of cranesbill populations in the landscape permit it to spread between populations of this host plant without the need for long-distance dispersal. The butterfly’s capacity to use Geranicaeae has been aided by the spread of butterfly phenotypes that readily select Geraniaceae plants for egg-laying (15) and by a degree of escape from natural enemies (parasitoids) associated with historical rockrose sites (22). These processes have come together to generate an unexpectedly rapid transformation in the metapopulation dynamics of the butterfly from a highly localized distribution associated with southerly facing rockrose-containing calcareous grasslands to widespread use of virtually any grassland with rockrose or Geranicaeae host plants. Ecological and evolutionary adjustments by the butterfly, interacting with alternative host plants that differ in their niches and life-history traits, have resulted in rapid range expansion of this previously rare and declining butterfly. We suggest that altered interactions among species do not necessarily constrain distribution changes but can facilitate expansions.

Fig. 3

Availability of rockrose and cranesbill in the landscape. 100 × 100 m grid squares with records of rockrose (red symbols), dove’s-foot cranesbill (blue symbols), or both species (purple symbols) in two well-recorded counties: (A) Bedfordshire and (B) Suffolk. Rapid range expansion took place in Bedfordshire and Suffolk, associated with the increased use of dove’s-foot cranesbill and other Geraniaceae.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S4

Tables S1 and S2

References (2328)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: Work was funded through NERC Ecology and Hydrology Funding Initiative grant NE/E012035. Butterfly distribution data were derived from a database of records submitted by volunteers (, and densities were derived from UK Butterfly Monitoring Scheme data ( Both schemes are operated by Butterfly Conservation and the Centre for Ecology & Hydrology and funded by a consortium of government agencies. Plant data were derived from a database of volunteer records managed by the Botanical Society for the British Isles ( We are grateful to the volunteers who collected the original butterfly and plant data. Temperature data were derived from the Central England Temperature (CET) data set ( Digital Elevation Models were provided by the NERC Earth Observation Data Centre ( Original data collected by authors is presented in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article