Role of larval host plants in the climate-driven range expansion of the butterfly Polygonia c-album


and present address: Brigitte Braschler, Biodiversity and Macroecology Group, Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, UK. E-mail:


  • 1Some species have expanded their ranges during recent climate warming and the availability of breeding habitat and species’ dispersal ability are two important factors determining expansions. The exploitation of a wide range of larval host plants should increase an herbivorous insect species’ ability to track climate by increasing habitat availability. Therefore we investigated whether the performance of a species on different host plants changed towards its range boundary, and under warmer temperatures.
  • 2We studied the polyphagous butterfly Polygonia c-album, which is currently expanding its range in Britain and apparently has altered its host plant preference from Humulus lupulus to include other hosts (particularly Ulmus glabra and Urtica dioica). We investigated insect performance (development time, larval growth rate, adult size, survival) and adult flight morphology on these host plants under four rearing temperatures (18–28·5 °C) in populations from core and range margin sites.
  • 3In general, differences between core and margin populations were small compared with effects of rearing temperature and host plant. In terms of insect performance, host plants were generally ranked U. glabra  U. dioica > H. lupulus at all temperatures. Adult P. c-album can either enter diapause or develop directly and higher temperatures resulted in more directly developing adults, but lower survival rates (particularly on the original host H. lupulus) and smaller adult size.
  • 4Adult flight morphology of wild-caught individuals from range margin populations appeared to be related to increased dispersal potential relative to core populations. However, there was no difference in laboratory reared individuals, and conflicting results were obtained for different measures of flight morphology in relation to larval host plant and temperature effects, making conclusions about dispersal potential difficult.
  • 5Current range expansion of P. c-album is associated with the exploitation of more widespread host plants on which performance is improved. This study demonstrates how polyphagy may enhance the ability of species to track climate change. Our findings suggest that observed differences in climate-driven range shifts of generalist vs. specialist species may increase in the future and are likely to lead to greatly altered community composition.


During current climate warming, some insect species are expanding their distributions northwards to track climate changes (Parmesan & Yohe 2003; Root et al. 2003). However, not all species are responding and the ability of species to colonize newly available climatically suitable habitats beyond their range margin is affected by factors such as species’ dispersal ability and the availability of breeding habitat (Hill et al. 2001; Warren et al. 2001).

Understanding insect–host plant interactions and the causes and consequences of changing hosts is a central theme in ecology (Ehrlich & Raven 1964; Thompson & Pellmyr 1991). Shifts on to novel host plants are important in sympatric speciation (Berlocher & Feder 2002), in allowing species to colonize new habitats (Singer & Thomas 1996), and in allowing species to escape predators (Murphy 2004). Shifts in host plant preferences of insects also affect colonization rates within metapopulations (Hanski & Singer 2001), although changes in host plant use have rarely been considered in the context of species’ ability to track climate changes (Thomas et al. 2001). Many species have more restricted niches towards the edges of their ranges (Hengeveld & Haeck 1982; Brown 1984), but warmer temperatures might facilitate the exploitation of a wider range of larval host plants, or increase larval survival on certain host plants (Hellmann 2002). This would increase species’ abundance and habitat availability at range margins and thus would increase range expansion rates, but data are lacking.

Dispersal is important for successful colonization of new habitats and there is evidence that evolutionary increases in dispersal ability may occur in insect populations during range expansion (Thomas et al. 2001; Hughes, Hill & Dytham 2003; Simmons & Thomas 2004). This is because colonists are usually not a random selection of the source population but share a suite of traits associated with increased dispersal ability (Haag et al. 2005). Such increases in dispersal ability in margin populations would be expected to promote range shifts by increasing the likelihood of colonization of new habitat, and may result in unexpectedly rapid range expansion rates (Thomas et al. 2001). Not only may there be evolutionary changes in dispersal, but the expression of dispersal traits may vary under different environments. For example, insects developing on poor-quality hosts may show increased dispersal ability (Coll & Yuval 2004). Thus, the dispersal ability of individuals in margin populations may reflect not only colonization effects but also the host plants utilized by larvae, but this has not been considered previously.

Global climates are predicted to continue warming by up to 5·8 °C by the end of the twenty-first century (IPCC 2001), and so understanding how insect development is affected by temperature and host plant use, and how patterns of host plant use interact with temperature, may be crucial for predicting species’ ranges in the future. In this study, we investigate the polyphagous butterfly Polygonia c-album (L.), which is showing the greatest range expansion of any butterfly in Britain during current climate warming (Asher et al. 2001). In Britain, this species apparently has altered its host plant preference from hop Humulus lupulus L. to include other hosts, particularly Wych elm Ulmus glabra Hudson and common nettle Urtica dioica L. (Pratt 1986, 1987). In this study, we investigate how the flight morphology and performance of the butterfly is affected by temperature and host plant use. We test the hypothesis that butterfly performance on U. glabra and U. dioica is improved under warmer temperatures, and we investigate whether or not the effects of temperature and performance on different host plants differ between populations from a site within the core of the distribution and a site at the expanding range margin.

Materials and methods

the study species

In the early nineteenth century, Polygonia c-album was widespread in Britain but subsequently underwent dramatic range retraction (Pratt 1986, 1987). This contraction may have been associated with a reduction in the cultivation of hops H. lupulus (the main larval host plant), but climate cooling at the end of the nineteenth century may also have been important, especially as retraction occurred even in areas where hops were still widely grown (Pratt 1987). From the 1940s onwards, the climate has been warming and P. c-album has successfully re-colonized most of England and Wales (Fig. 1). This expansion has been associated with a shift in the preferred host plant from H. lupulus to include the closely related and widespread species Urtica dioica and Ulmus glabra (Pratt 1986, 1987), host plants that are widely used by P. c-album in other parts of Europe (e.g. Nylin & Janz 1993). Polygonia c-album overwinters as an adult and develops through either one or two generations per year in Britain depending on location. Voltinism is primarily determined by environmental factors (e.g. day length, host plant, temperature; Wiklund, Wickman & Nylin 1992; Wedell, Nylin & Janz 1997), although genetic effects may also play a role.

Figure 1.

Recent range expansion and present distribution of Polygonia c-album in Britain, and location of study sites. Distribution records (10-km grid resolution) are plotted for two time periods corresponding with the publication of two butterfly atlases (Heath, Pollard & Thomas 1984; Asher et al. 2001). Collection sites are from north to south: Newcastle (re-colonized in 1995), York (1993), Nottingham (1985) and Wye Valley (core site). Females from York and Wye Valley were used in the laboratory experiments, wild-caught individuals from all sites were used to determine differences in flight morphology between core and range margin sites. Histogram bars show the number of 10-km Ordnance Survey grid cells with records of the species for each decade of the twentieth century (P. c-album distribution data provided from the Butterflies for the New Millennium project, courtesy of Butterfly Conservation and Biological Records Centre).

insect material

During April 2004, adult females were collected from two sites in Britain; a ‘core’ site in south Wales (Wye Valley, Ordnance Survey 10-km grid reference ST 59; Fig. 1) where P. c-album apparently persisted during the nineteenth century range contraction (Pratt 1987), and a margin site in northern England (York, SE 53; Fig. 1) which according to recent sightings was re-colonized in 1993 (Biological Records Centre data, CEH Monks Wood). Four females from each site were used to establish breeding lines for experiments. We also collected males from each study site, as well as from two additional sites in central England (Nottingham, SK 63, re-colonized in 1985; Fig. 1), and in north-east England (Newcastle, NZ 15, re-colonized in 1995; Fig. 1). Overall, we collected 12 females (four from Wye Valley, and eight from York) and 35 males (six from Wye Valley, six from Nottingham, 12 from York, and 11 from Newcastle). All wild-caught individuals were used to investigate differences in adult flight morphology among sites in relation to range expansion. Four females from the York site did not lay fertile eggs and were also included in this analysis.

experimental design

Wild-collected females were allowed to lay eggs on potted U. dioica plants in a greenhouse. In order to produce sufficient insect material for the rearing experiments and to maintain genetic diversity, F1 female offspring from wild-caught females were mated with F1 male offspring from a different mother, but the same site, to create nine different F2 families per site.

We used a split-brood design to examine the effects of host plant and temperature on performance and flight morphology. Upon hatching, first instar larvae from each of the 18 families were split equally among the three different host plants (H. lupulus, U. dioica and U. glabra) and four different temperatures (18 °C, 21·5 °C, 25 °C and 28·5 °C (±1 °C); Sanyo MLR-350 growth cabinets). Approximately eight to 11 larvae of the same age from the same female were reared together in plastic containers and provided daily with excess freshly cut young leaves of host plants in vials of water (yielding 18 containers with larvae (nine containers per site) in each of the 12 temperature by host plant treatments). Occasionally when larvae fed very heavily, or when wilting was observed, leaves were replaced up to twice daily. All leaves were from young plants and were thus likely to be of high nutritional quality. All larvae were reared under a long photoperiod (20L : 4D) to maximize the incidence of nondiapausing individuals. Upon pupation, individuals were weighed and reared individually in plastic tubes under the same light and temperature conditions as larvae. Adults were killed by freezing within 12 h of emergence after allowing time for wing expansion and release of pupal waste products. After thawing, adult body parts were dried at 60 °C for 24 h and then weighed on a Sartorius electrobalance (sensitivity 0·1 µg). The sex and adult morph (directly developing vs. diapausing) of all individuals were determined. Directly developing morphs can be identified easily by their lighter coloration.

measures of insect growth and flight morphology

Direct measures of dispersal are difficult to obtain in insects (particularly in highly mobile species such as P. c-album) and many studies have thus inferred dispersal potential from indirect measures of adult flight morphology. In butterflies, individuals with greater flight ability generally have relatively larger, broader thoraxes (comprising predominantly flight muscle, e.g. Berwaerts, Van Dyck & Aerts 2002). Therefore measures of adult flight morphology were used in this study as an index of dispersal ability. We measured thorax shape (width divided by length) in both wild-caught and laboratory reared individuals. Thorax mass is likely to change with age and so we only measured this variable in laboratory reared individuals that were killed on emergence.

The following variables were used to assess insect performance in different treatments: larval growth rate (pupal fresh mass divided by time from hatching to pupation), development time (time from hatching of the larva until emergence of adult), adult dry mass, abdomen dry mass and survival (proportion surviving from first instar larva to adult). Adult mass is likely to be a good indicator of host plant quality with larger butterflies developing on higher quality plants being expected to be more fecund, better competitors for mates, and able to disperse over longer distances. Repeated measurements for all mass variables proved to be reliable (less than 5% difference). Physiological trade-offs between flight and reproduction are expected such that individuals with relatively large thoraxes are likely to be less fecund (Hughes et al. 2003; but see Hanski, Saastamoinen & Ovaskainen 2006). Growth rate in butterflies is related to temperature (Wedell et al. 1997; Gotthard 2004) and host plant quality (Janz, Nylin & Wedell 1994). It also affects developmental pathway (diapause vs. direct development; Wedell et al. 1997) and thus will affect population growth rate and size, and therefore range expansion.

statistical analysis

General Linear Mixed Models were used to examine the effect of population (core vs. margin), temperature (18 °C, 21·5 °C, 25 °C, 28·5 °C), host plant (H. lupulus, U. glabra, U. dioica), sex, and developmental pathway (direct development vs. diapause) on variables associated with dispersal ability (thorax shape, relative thorax mass) and performance (development time, adult mass, larval growth rate, relative abdomen mass). Additional analyses on pupal fresh weight and time to pupation were also performed but generally gave the same results as those for adult mass and time to adult development and are thus not presented in depth. Interaction effects between population and host plant, population and temperature, host plant and temperature, and the threefold interaction between population, host plant and temperature were also included in analyses. Family effects were nested within the population term and included as random effects. Tukey–Kramer tests were used to test for differences between treatments. Some variables showed allometric relationships and so we included total adult body mass as a covariate in analyses of thorax mass and abdomen mass.

To examine the effects of rearing treatments on the propensity of individuals to enter diapause, a Generalized Linear Mixed Model similar to those described above was used but with a binomial error distribution and a logit link function. A simplified version of this model was used for analysing the effects of population, rearing temperature, and host plant on survival as the sex was known only for individuals that reached pupation, and developmental pathway was known only for individuals that reached adulthood. Family effects were nested within the population term and included as random effects in the analysis of developmental pathway but could not be included in the analysis of survival as the design of the experiment did not allow differentiation between family and treatment effects in this respect.

Where appropriate, variables were ln-transformed for analyses. SAS vs. 8·1 was used for all analyses. The glimmix macro was used for Generalized Linear Mixed Models with a binomial error distribution (Littell et al. 1996). We sampled only one core and one margin location, and so analysing data for individual P. c-album may lead to pseudo-replication. However, P. c-album is highly mobile and so sampled individuals are probably representative of a relatively large geographical area.


In total, we analysed data from 1252 F2 laboratory reared individuals and from 47 wild-caught adults. Overall, females had broader thoraxes than males (F1,1038 = 9·06, P = 0·0027) but males had relatively heavier thoraxes than females (F1,1035 = 1172·67, P < 0·0001), indicating no clear pattern in our indices of dispersal ability between the sexes. Larval growth rates did not differ between the sexes, but adult males completed their development faster than females (F1,1035 = 15·92, P < 0·0001) because they were smaller than females (F1,1036 = 66·08, P < 0·0001).

factors affecting developmental pathway

Wild-caught individuals were collected in spring and so were the diapausing morph, although directly developing individuals have been reported from all study sites. In the rearing experiments, there was no significant difference between core and margin populations in the percentage of individuals that entered diapause (45·8% vs. 31·9%; F1,16 = 2·40, P = 0·1409). Individuals entering diapause had broader and relatively heavier thoraxes (thorax shape: F1,1038 = 16·02, P < 0·001; thorax mass: F1,1035 = 25·76, P < 0·0001) indicating greater dispersal ability. Males had a greater propensity to enter diapause compared with females (51·4% vs. 24·3%; F1,1039 = 107·42, P < 0·0001; Fig. 2), and adults entering diapause had lower larval growth rates (F1,1038 = 35·12, P ≤ 0·0001), and longer development times (F1,1035 = 287·64, P < 0·0001).

Figure 2.

Proportion of directly developing P. c-album adults from a range margin (hollow bars) and core (solid bars) population reared under four different temperatures and on three different host plants. Upper row: females, lower row: males. Ug = U. glabra, Hl = H. lupulus, Ud = U. dioica.

Overall, the host plant did not directly affect the propensity of individuals to diapause, although this was approaching significance (F2,1039 = 2·96, P = 0·0525), but higher temperatures greatly increased the likelihood that individuals developed directly (F3,1039 = 52·06, P < 0·0001; Fig. 2). A significant host plant by temperature interaction showed that at higher temperatures individuals were more likely to develop directly on U. glabra and U. dioica, but at lower temperatures individuals were more likely to develop directly on H. lupulus (host plant × temperature interaction: F6,1039 = 6·89, P ≤ 0·0001; Fig. 2).

factors affecting adult flight morphology

In wild-caught individuals, there was an increase in thorax breadth in individuals from populations from the core to the range margin, suggesting increased dispersal ability at the range margin (regression for males from four sites, weighted by sample size; F1,3 = 52·90, P = 0·018, R2 = 0·96; Fig. 3). This effect was not evident in wild-caught females (t-test comparing females from two study sites; t10 = 1·50, P = 0·16; Fig. 3), although this analysis lacked power due to the small sample sizes and few sites sampled.

Figure 3.

Thorax shape of wild-caught individuals from a core site and three newly colonized sites. Higher values correspond to relatively broader thoraxes and indicate greater dispersal ability. Mean values with standard errors for males (solid symbols) and females (hollow symbols) are shown separately. Data for females are displaced slightly for clarity. Numbers by data points indicate sample sizes.

In laboratory reared butterflies, adult flight morphology (relative thorax mass, thorax shape) did not differ significantly between core and margin populations. However, larval host plant and temperature had significant but opposite effects on relative thorax mass and thorax shape. Thus butterflies reared under higher temperatures and on H. lupulus had relatively broader thoraxes (temperature effect, F3,1038 = 11·39, P < 0·0001; host plant effect, F2,1038 = 9·24, P = 0·0001) but relatively small thoraxes (temperature effect, F3,1035 = 22·66, P < 0·0001; host plant effect, F2,1035 = 23·65, P < 0·0001). This makes it difficult to deduce the impacts of host plant and temperature on flight morphology. There were also significant interactions between host plant and temperature on thorax shape (host plant × temperature interaction: F6,1038 = 4·13, P = 0·0004) and on relative thorax mass (F6,1035 = 10·16, P < 0·0001). Thus, individuals reared on H. lupulus had increasingly broader but smaller thoraxes at higher temperatures. There was also a significant interaction between population and larval host plant on relative thorax mass (population × host plant interaction: F2,1035 = 4·44, P = 0·0120). Thus individuals from the core had relatively larger thoraxes on U. glabra but no such difference was evident in the margin population.

factors affecting insect performance

Survival decreased with increasing temperature (F3,1150 = 6·22, P = 0·0003; Fig. 4) and was poorest on H. lupulus compared with the other two host plants (F2,1150 = 30·03, P < 0·0001). Lower survival on H. lupulus was most evident at high temperatures (28·5 °C: 8·1% survival on H. lupulus vs. 44·6% on U. glabra and 49·0% on U. dioica; Fig. 4) while survival was similarly high on all three host plants at low temperatures (e.g. 18 °C: 60·6% survival on H. lupulus vs. 63·6% on U. glabra and 67·6% on U. dioica; temperature × host plant interaction: F6,1150 = 2·83, P = 0·0096; Fig. 4).

Figure 4.

Survival of P. c-album from a range margin (hollow bars) and core (solid bars) population reared under four different temperatures and on three different host plants. Ug = U. glabra, Hl = H. lupulus, Ud = U. dioica.

In general, effects of temperature and host plant were similar between core and margin populations. Overall, performance on the novel host plants was much better than on H. lupulus regardless of temperature treatment or population. However, there were some differences in the ranking of the two novel larval host plants. While there was no difference in performance on either U. glabra or U. dioica in the core population, development time on U. glabra was shorter than on U. dioica at the range margin (Tukey–Kramer tests for difference between novel host plants; core: t1035 = 0·25, P = 0·99, margin: t1035 = 6·25, P < 0·0001). This resulted in higher growth rates on U. glabra than on U. dioica in the margin population (Tukey–Kramer tests for difference between novel host plants; core: t1038 = 2·26, P = 0·21, margin: t1038 = 7·39, P < 0·0001). These findings support anecdotal reports of the butterflies’ preference for U. glabra in northern Britain. However, there were no differences in adult mass in either population on the two novel host plants, although adult mass was always much larger than on the original host H. lupulus. Increased temperatures resulted in higher larval growth rates (F3,1038 = 195·77, P < 0·0001), decreased development times (F3,1035 = 1841·09, P < 0·0001), and smaller adults (F3,1036 = 94·26, P < 0·0001). This indicated a trade-off between development time and adult size. However, individuals reared on H. lupulus had the slowest larval growth rates, the longest development times, but also produced the smallest adults (Table 1).

Table 1.  Performance of P. c-album on three different host plants (H. lupulus = original host, U. glabra and U. dioica are new hosts). Results from Generalized Linear Models on ln-transformed variables. Back-transformed mean ± SEs are shown
 Larval host plantd.f.FP
H. lupulusU. glabraU. dioica
Larval growth rate (mg day−1) 11·0 ± 1·0 16·1 ± 1·0 14·6 ± 1·02163·55< 0·0001
Time to pupation (days) 20·6 ± 1·0 17·2 ± 1·0 18·5 ± 1·02 47·06< 0·0001
Pupal fresh mass (mg)226·2 ± 1·0276·5 ± 1·0270·0 ± 1·02131·18< 0·0001
Development time (days) 31·3 ± 1·0 26·2 ± 1·0 27·1 ± 1·02 28·52< 0·0001
Adult dry mass (mg) 32·8 ± 1·0 44·8 ± 1·0 46·3 ± 1·02252·20< 0·0001

In adult butterflies, thorax mass and abdomen mass comprise c. 70% of total mass and therefore conditions that resulted in increased investment in the thorax resulted in decreased investment in the abdomen. Thus, relative abdomen mass was smallest in individuals reared under low temperatures (F3,1035 = 31·90, P < 0·0001) and on U. glabra (F2,1035 = 33·88, P < 0·0001). However, a significant interaction between host plant and rearing temperature showed that the ranking of the host plants changed with temperature. Individuals on H. lupulus were generally small in terms of total mass (Table 1), but had increasingly larger abdomens with increasing temperatures compared with the other larval host plants (temperature × host plant interaction: F6,1035 = 13·94, P < 0·0001). Thus at the highest temperatures, individuals on H. lupulus invested more in their abdomens compared with those on U. glabra and U. dioica. A significant population × host plant interaction indicated that ranking of larval host plants also differed between the two populations; individuals in the core had relatively large abdomens on both H. lupulus and U. dioica, whereas at the margin they only had relatively large abdomens on U. dioica (F2,1035 = 4·93, P = 0·0074).


Polygonia c-album has shown the fastest rate of range expansion of any resident butterfly species in Britain during recent climate warming and is one of only a few species that appears to be tracking climate changes (Warren et al. 2001). In this study, we showed that both adult flight morphology and insect performance depended on the host plant used, as well as on temperature during development (Janz et al. 1994; Wedell et al. 1997). In addition, there were interaction effects between temperature and host plant, as well as between host plant and population, showing that the consequences of developing on different host plants may vary across the species’ range, and with climate change. Overall, both novel hosts were better than H. lupulus but U. glabra was the superior host in the margin population while there were few differences between U. glabra and U. dioica in the core population. Interestingly, feeding on H. lupulus led to greatest investment in the abdomen in the core population, but was inferior to U. dioica in the margin population. Thus the only instance where we found that the original host plant may be preferable in some respect to the novel hosts was in the core population.

factors affecting flight morphology

Individuals from recently founded populations at the expanding range margin would be expected to have greater dispersal ability compared with individuals from core populations (Thomas et al. 2001; Travis & Dytham 2002; Simmons & Thomas 2004). Indeed flight morphology in wild-caught males indicated higher dispersal ability at the range margin in this study. This was apparently due to different environmental conditions at the study sites (e.g. habitat quality) rather than due to any evolutionary changes during range expansion because differences in flight morphology were not evident between core and margin populations when individuals were reared under controlled conditions in the laboratory. Moreover, in laboratory reared individuals, measures of thorax shape and relative thorax mass gave conflicting results. Although measures of flight morphology have been successfully used in the past to deduce dispersal potential (Berwaerts et al. 2002) our findings suggest that the relation between thorax mass and shape is not straightforward in this species. Wild-caught material was obtained from sites spanning a greater range of distances from the range margin, and thus may have had increased genetic variation compared with laboratory reared individuals, which may have affected our results. However, given that few differences between core and margin populations were detected in any of our measures of insect performance it is possible that evolutionary changes during range expansion play a minor role in such a mobile species (Nylin et al. 2005). The absence of a clear adaptation to the original host in the core population could also be explained by high mobility and gene flow back into the core.

propensity to enter diapause

In Britain, adults emerging in summer can either enter diapause or develop through a second generation before diapausing. The ability to develop through two generations per year will increase population growth rates (Crozier 2004) and thus rates of range expansion (Fric & Konvicka 2002). In our experiment, diapausing individuals were more likely under cooler temperatures. Previous studies on P. c-album have shown that host plants affect the propensity to enter diapause through effects on larval growth rates (Wedell et al. 1997). The absence of a significant host plant effect on diapause in our study may indicate that for British P. c-album there is often sufficient time for a second generation even when using a suboptimal host such as H. lupulus.

ranking of larval host plants

In relation to insect performance, host plants were generally ranked U. glabra  U. dioica > H. lupulus and this ranking was not significantly affected by temperature. Differences between U. glabra and U. dioica were relatively small compared with large differences between these two host plants and H. lupulus, which was the poorest host plant. Individuals on U. glabra completed their development the fastest, and individuals reared on U. glabra and U. dioica also had higher survival and larger adult size than those on H. lupulus, confirming that U. glabra and U. dioica are high ranking host plants for P. c-album (Nylin & Janz 1993; Janz et al. 1994; Nylin & Janz 1996; Nylin, Bergstrom & Janz 2000). Higher temperatures also increased the proportion of directly developing individuals on both of these host plants. Thus overall we would expect the use of U. glabra and U. dioica to increase the rate of range expansion of P. c-album relative to using H. lupulus, especially under increasing temperatures.

There is little information on the degree to which P. c-album utilized U. dioica or U. glabra historically in Britain except for anecdotal evidence that the first generation may have used U. dioica early in the season when H. lupulus had not yet developed sufficiently (Pratt 1987). However, the ability to exploit different host plants may be facilitated if larvae retain the ability to develop on former host plants even when the oviposition preferences of the adults change (Janz, Nyblom & Nylin 2001). For example, P. c-album is now exploiting U. dioica, which is an ancestral host plant that is exploited by many other closely related Nymphalini species and is also used by many European populations of P. c-album (Janz et al. 2001). In addition, the three host plants examined in this study are all closely related and comprise an ancestral host plant clade for Polygonia species (Weingartner, Wahlberg & Nylin 2006) and this may also explain the shift on to new host plants by this species.

P. c-album uses different host plants throughout its European range, and the degree of specialization on different host plants varies geographically (Janz 1998). H. lupulus is apparently the historically preferred host plant in Britain (Pratt 1986, 1987), but our results showed it is generally a poor host plant in both core and margin sites. However, an historical advantage of H. lupulus may have been its wide availability through cultivation for the beer industry, particularly if such sites were situated in warm locations (e.g. south-facing slopes). In addition, our laboratory reared larvae were provided with high-quality young leaves from a single location, but plant quality may vary across the butterfly range, or with temperature, and this deserves further study.

Further range expansion of P. c-album will be influenced by the distribution of the host plant species within Britain. H. lupulus is common in the south of Britain where P. c-album is now ubiquitous, but is rare in the north where the other host plants occur widely (Preston, Pearman & Dines 2002). H. lupulus may increase its distribution with climate warming, but exploitation of U. dioica and U. glabra that occur widely in northern Britain will allow the mobile insect to colonize new areas without waiting for range expansion of its original host. Thus future range expansion of P. c-album does not appear to be limited by host plant availability and the present range expansion is likely to continue.

future range changes

Species are predicted to shift their distributions polewards in order to track future climate warming (Hill, Thomas & Huntley 1999; Hill et al. 2002; Thomas et al. 2004). However, range shifts are likely to be limited to mobile generalists, such as P. c-album, where range expansion is not limited by the loss and fragmentation of natural habitats (Warren et al. 2001). Results from this study illustrate the flexibility of polyphagous generalist species and demonstrate how the incorporation of novel host plants into larval diets may result in species having greatly enhanced abilities to track climatic changes. Our results indicate that this flexibility may lead to unexpectedly rapid range expansion in generalist species, and that current host plant preferences may underestimate future range changes in some species. The degree to which similar host-plant shifts will occur in other species remains to be seen, but such shifts are likely to be the exception rather than the rule. What is more clear is that tracking of twenty-first century climate warming is likely to be restricted to generalist and mobile species of relatively low conservation value, and that this may lead to greatly altered community composition in the future (Menéndez et al. 2006).


We would like to thank the Forestry Commission and Tilhill Forestry Ltd for permission to work at their sites. Sören Nylin and Nina Wedell gave helpful advice on insect rearing, and Tim Yardley helped setting up the experiment. We thank Butterfly Conservation and the Biological Records Centre (CEH-Monks Wood) for P. c-album distribution data. Comments by Thomas Merckx and two anonymous referees improved an earlier version of the manuscript. The project was funded by NERC.