Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Contributed Paper Interactions between Ecological Traits and Host Plant Type Explain Distribution Change in Noctuid Moths NIINA MATTILA,∗ JANNE S. KOTIAHO,∗ † VEIJO KAITALA,‡ ATTE KOMONEN,§ AND JUSSI PÄIVINEN∗∗ ∗ Department of Biological and Environmental Science, University of Jyväskylä, P.O. Box 35, FI-40014, Finland †Natural History Museum, University of Jyväskylä, P.O. Box 35, FI-40014, Finland ‡Department of Biological and Environmental Science, University of Helsinki, P.O. Box 65, FI-00014, Finland §Faculty of Forest Sciences, University of Joensuu, P.O. Box 111, FI-80101, Finland ∗∗ Natural Heritage Services of Metsähallitus, Jyväskylä, P.O. Box 36, FI-40101, Finland Abstract: The ecological traits of species determine how well a species can withstand threats to which it is exposed. If these predisposing traits can be identified, species that are most at risk of decline can be identified and an understanding of the processes behind the declines can be gained. We sought to determine how body size, specificity of larval host plant, overwintering stage, type of host plant, and the interactions of these traits are related to the distribution change in noctuid moths. We used data derived from the literature and analyzed the effects of traits both separately and simultaneously in the same model. When we analyzed the traits separately, it seemed the most important determinants of distribution change were overwintering stage and type of host plant. Nevertheless, ecological traits are often correlated and the independent effect of each trait may not be seen in analyses in which traits are analyzed separately. When we accounted for other correlated traits, the results were substantially different. Only one trait (body size), but 3 interactions, explained distribution change. This finding suggests that distribution change is not determined by 1 or 2 traits; rather, the effect of the traits depends on other interacting traits. Such complexity makes it difficult to understand the processes behind distribution changes and emphasizes the need for basic ecological knowledge of species. With such basic knowledge, a more accurate picture of the factors causing distribution changes and increasing risk of extinction might be attainable. Keywords: distribution change, ecological traits, Lepidoptera, Noctuidae, predictive conservation science Las Interacciones entre Atributos Ecológicos y Tipo de Planta Huésped Explican el Cambio de Distribución de Mariposas Nocturnas Resumen: Los atributos ecológicos de las especies determinan qué tan bien una especie puede resistir las amenazas a que está expuesta. Si se pueden identificar estos atributos que contribuyen a la predisposición, las especies que están en mayor riesgo de declinación pueden ser identificadas y los procesos que subyacen en las declinaciones pueden ser entendidos. Buscamos determinar la relación entre el tamaño corporal, la especificidad de planta hospedera de larvas, la etapa hibernación, el tipo de planta huésped y las interacciones de estos atributos con el cambio de distribución de mariposas nocturnas de la familia Noctuidae. Utilizamos datos derivados de la literatura y analizamos los efectos de los atributos, tanto por separado como simultáneamente, en el mismo modelo. Cuando analizamos los atributos por separado, pareció que los determinantes del cambio de distribución más importantes fueron la etapa de hibernación y el tipo de planta huésped. Sin embargo, los atributos ecológicos a menudo están correlacionados y el efecto independiente de cada atributo puede no ser visto en análisis separados. Cuando consideramos atributos correlacionados, los resultados fueron sustancialmente diferentes. Solo un atributo (tamaño corporal), pero tres interacciones, explicaron el cambio de distribución. Este hallazgo sugiere que el cambio de distribución no está determinado ∗ email [email protected] Paper submitted May 26, 2008; revised manuscript accepted October 3, 2008. 1 Conservation Biology, Volume **, No. *, ***–*** C 2008 Society for Conservation Biology DOI: 10.1111/j.1523-1739.2008.01138.x Species’ Ecological Traits and Distribution Change 2 por uno o dos atributos individuales; más bien, el efecto de los atributos depende de otros atributos interrelacionados. Tal complejidad dificulta el entendimiento de los procesos que subyacen en los cambios de distribución y enfatiza la necesidad de conocer la ecologı́a básica de las especies. Con tal conocimiento básico, se puede obtener una visión más precisa de los factores que causan cambios en la distribución e incremento del riesgo de extinción. Palabras Clave: atributos ecológicos, cambio de distribución, ciencia de conservación predictiva, Lepidoptera, Noctuidae Introduction Understanding the underlying causes of distribution changes and extinction risk is of vital importance for predictive conservation science (Cardillo et al. 2004; Kotiaho et al. 2005). Without this understanding, scarce resources available for management of species may be misallocated. Because the biology of a species determines how well it can withstand the threats to which it is exposed (Cardillo et al. 2004), it is essential to understand the biological or ecological correlates of distribution changes. If the predisposing traits can be identified, we may be able to predict which species are biologically most vulnerable and hence initiate conservation efforts when their populations are still viable. A few researchers have identified ecological traits of species that can predict distribution change (e.g., Conrad et al. 2004; Kotiaho et al. 2005; Mattila et al. 2006). Nevertheless, even when it seems that some traits are connected to distribution change, the correct biological reasons why these traits predispose a species are still unknown. This is because in correlative studies, the trait under the study may not be the cause of the observed effect, but may be correlated with some other trait that is the actual cause. Unfortunately, the nature of these phenomena is such that they are not really amenable to manipulative experiments that would provide a better understanding of the causative pathways. Because of the potential correlated effects, results may be misinterpreted when trying to come up with biological explanations for why some traits have an effect on distribution change. By aiming at identifying the actual predisposing traits and connecting these to various threats important knowledge can be gained for use in conservation and management. Nevertheless, it is not always easy to explain these connections because of the cumulative effects of multiple threats and interactions between separate threats. In a previous study we reported on the distribution decline and on the ecological determinants of extinction risk and distribution decline in Finnish noctuid moths (Mattila et al. 2006). Comparable declines among noctuid moths have also been reported from the British Isles (Conrad et al. 2006). In our previous study we concluded that many of the ecological traits we studied had effects on extinction risk, but the only ecological trait that ex- Conservation Biology Volume **, No. *, 2008 plained the distribution change was the overwintering stage. Nevertheless, the overwintering stage is related to the type of larval host of macrolepidoptera (Niemelä et al. 1982; Virtanen & Neuvonen 1999), and in this study we sought to determine the effects of overwintering stage, body size, specificity of larval host plant, and type of host plant on the distribution change of noctuid moths. Methods Overall, 404 noctuid moth species (Lepidoptera, Noctuidae) occur in Finland (Huldén et al. 2000). We considered the 306 moth species that have a resident or fluctuating population in Finland (Huldén et al. 2000) and classified them according to their larval host plant (see later). One hundred sixty-three species had clearly definable host plant types, and we included only these species in our analyses. We tested ecological traits and host plant type separately and simultaneously in the same model to remove the correlated effects of other traits and to determine possible interactions between the traits. Distribution Change In the Atlas of Finnish Macrolepidoptera (Huldén et al. 2000), the distribution of species is given as the number of occupied 10 × 10 km grid cells in the national coordinate system. In the atlas distribution data are divided into observations before 1988 and observations between 1988 and 1997. We used the difference in these distributions (number of occupied 10 × 10 km grid cells) to estimate the distribution change for each species. In all analyses we expressed the change in relative scale in percentages such that positive percentages indicate increased distribution and negative values indicate reduced distributions. Because the atlas covers all reliable observations in the last 250 years, the timescales are unequal, the earlier being much greater. Nevertheless, there were more observations recorded during the later period (figure 14 in Huldén et al. 2000). We acknowledge that it is possible that the inequality of the time periods and the difference in observation activity may bias the overall magnitude of the distribution change. Nevertheless, this bias is unlikely to explain our main findings. We cannot Mattila et al. think of any simple mechanism for why the possible bias should correlate with the ecological traits. Type of Larval Host Plant and Species-Specific Ecological Traits We based type of larval host plant on Huldén et al. (2000). There were several types of host plants into which only a few species were classified. To retain even some power in our analyses, we included only species that used the 3 most abundant and easily demarcated types of host plants: species feeding on herbs (n = 65), grasses (n = 59), and broadleaved trees (n = 39). In addition to the overwintering stage, we included body size (measured as male wing span) and larval–host plant specificity in our analyses. For some species, there were no data for some of the traits; thus, sample sizes varied between the analyses. We based overwintering stages on the Atlas of Finnish Macrolepidoptera (Huldén et al. 2000). Overwintering stages of Finnish noctuid moths are egg, larva, pupa, and adult. The very few species overwintering as adults were not included in our analyses. We also excluded species that overwinter twice or an unspecified number of times to avoid problems in classifying the overwintering stages. We derived average body size (wing span) from Mikkola and Jalas (1977, 1979). Mean body size was that of 25 individuals. In some rare species, we used measurements from fewer individuals. We arbitrarily decided to use male size, but this is unlikely to have any effect on the results because male and female body size correlate strongly and positively (Pearson correlation, r = 0.98, n = 163, p < 0.001). Specificity of larval host plants was classified into 3 categories: monophagous, oligophagous, and polyphagous species. Monophagous species use only one host plant species, oligophagous species use plant species in one genus, and polyphagous species use plant species from more than one genus. Data on larval host plants were derived from Mikkola and Jalas (1977, 1979) and Huldén et al. (2000). Phylogeny Closely related species tend to share many characters through common descent rather than through independent evolution. Thus, statistical methods that treat such characters as independent may be problematic (Harvey & Pagel 1991; Harvey & Purvis 1991). The method of independent contrasts can be used with phylogenetic information to transform interspecific data into values that can be analyzed with standard statistical methods (Harvey & Pagel 1991; Purvis & Rambaut 1995). Nevertheless, if a phylogeny is not properly resolved, as is the case in noctuid moths, any correction based on it may be incorrect. Moreover, what is often not recognized is that there are a few rather strict assumptions 3 that must be met before the phylogenetically independent contrasts are in fact phylogenetically independent (Freckleton 2000). If the assumptions cannot be met, one cannot rely on the results from phylogenetic corrections. For these reasons, we derived our results from the original species data without correcting for phylogeny. Overall, we are relatively confident our results were not biased by the current phylogeny. This is because in our previous manuscript on noctuid moths (Mattila et al. 2006), we applied the method of phylogenetically independent contrasts (Purvis & Rambaut 1995) whenever it was possible, and in these analyses correcting for phylogeny did not change the conclusions that were based on the uncorrected analyses. Statistical Analyses We used analysis of variance (ANOVA) and linear regression models to analyze the effects of each trait separately. These were followed by pairwise comparisons (Tukey test). In simultaneous analyses in which the effects of other traits were controlled for, we used analysis of covariance (ANCOVA). We entered all main effects and interactions up to 3-way interactions into the model and then applied a stepwise removal of nonsignificant interactions. Removal was initiated from the least significant interaction at the highest-order interactions and removal ceased when the first interaction reached the set probability level. When there were interactions between traits, simple-effects tests followed by appropriate pairwise comparisons were used to investigate the interactions in detail. Comparisons became excessively numerous, so to maintain the clarity of presentation, we report only results for which the probability level was <0.060. We conducted statistical analyses in SPSS (version 13.0 for Windows; SPSS, Chicago, Illinois). Critique of Atlas Data Butterfly atlas data are often criticized because they usually do not include recorder effort. Without this information, observations can accumulate for certain areas (Dennis et al. 1999). Records of observations are lowest in areas where the human population is most sparse and greatest where it is dense (Heath et al. 1984; Dennis & Williams 1986). Long-term monitoring schemes can also bias the observed distribution (Pollard & Yates 1993). In 1993 a moth-monitoring scheme was initiated in Finland (Väisänen 1993). About 100 monitoring locations were evenly distributed across Finland, which may have a slight balancing effect if earlier observations were focused on particular areas. Overall, we believe that in Finland there are fewer problems with biases in observations than in many other countries because Finnish legislation includes a legal concept of "everyman’s right." Everyman’s right allows people right of access to almost any land without requiring the landowner’s permission. Consequently, it Conservation Biology Volume **, No. *, 2008 Species’ Ecological Traits and Distribution Change 4 Table 1. Pairwise comparisons (Tukey test) between overwintering stages and host types on distribution change of noctuid moths. Comparison Mean difference SE p 11.47 24.57 13.10 −13.99 −21.81 −7.82 5.37 5.42 4.98 4.93 5.55 5.66 0.085 <0.001 0.025 0.014 <0.001 0.353 Egg–larva Egg–pupa Larva–pupa Herbs–grasses Herbs–broadleaved trees Grasses–broadleaved trees is likely that in Finland observations are more evenly distributed and not accumulated into areas of public use. In addition, the Lepidoptera fauna of Finland is considered one of the best known in the world (Lepidopterological Society of Finland 2005). In addition, Finland has a large number of lepidopterists in proportion to the population (Lepidopterological Society of Finland 2005), and observation activity has increased over time (Huldén et al. 2000). An additional complication in the atlas data is that declines are only detected when species are lost from the entire grid square. There can be several populations of given species within each square, and a species may have to decline to a small fraction of its former abundance before declines are noticed on distribution maps (Thomas & Abery 1995). For this reason, the grid methodology (e.g., Mattila et al. 2006) is likely to underestimate the true magnitude of moth declines. Results Separate Analyses for Each Trait Type of host plant was related to distribution change (ANOVA, F 2,160 = 8.56, p < 0.001, R2 = 0.10). Species feeding on herbs declined more than species feeding on trees or grasses, but there was no difference between species feeding on trees and grasses (Table 1). Overwintering stage also had an effect on distribution change Table 2. Analysis of covariance of the effects of ecological traits on distribution change in noctuid moths (R 2 = 0.33). Source Overwintering stage Host plant type Body size Larval specificity Host plant type∗ overwintering stage Host plant type∗ larval specificity Overwintering∗ larval specificity Error Conservation Biology Volume **, No. *, 2008 MS F p η2 2 2 1 2 4 1824.65 1162.09 4894.66 1844.74 2440.54 2.93 1.87 7.86 2.96 3.92 0.057 0.159 0.006 0.055 0.005 0.039 0.025 0.052 0.040 0.099 4 2529.56 4.06 0.004 0.102 4 2288.27 3.67 0.007 0.093 143 622.88 df Table 3. Number of species in each combination category of larval specificity and overwintering stage and larval specificity and host plant type. Larval specificity Monophagous Oligophagous Polyphagous Overwintering stage, n egg, 6 larva, 2 pupa, 8 egg, 7 larva, 2 pupa, 4 egg, 31 larva, 57 pupa, 46 Host plant type, n herbs, 6 grasses, 6 trees, 4 herbs, 4 grasses, 5 trees, 4 herbs, 55 grasses, 48 trees, 31 (ANOVA, F 2,160 = 10.42, p < 0.001, R2 = 0.12). Pupal overwinterers declined more than egg or larval overwinterers, but there was no difference between egg and larval overwinterers (Table 1). Larval specificity and male size had no effect on distribution change (ANOVA, F 2,160 = 0.03, p = 0.974; linear regression F 1,161 = 1.03, p = 0.312 and slope B = −0.345, SE 0.340, respectively). Simultaneous Analysis for All Traits After controlling for the other traits, body size had an effect on distribution change (Table 2): larger species declined more than smaller ones (B = −0.902 [SE 0.322], t = −2.803, p = 0.006). Simultaneous analysis revealed 3 significant interactions: host plant type and overwintering stage; type of host plant and larval specificity; and overwintering stage and larval specificity (Table 2). The number of species in each trait-combination category is in Tables 3 and 4. Type of host plant affected distribution change in larval overwinterers (simple effects test, F 2,143 = 4.131, p = 0.018) such that species feeding on herbs or on broadleaved trees declined more than species feeding on grasses (pairwise comparisons, p = 0.024 and p = 0.038, respectively). Overwintering stage affected the distribution change in species feeding on broadleaved trees (simple effects test, F 2,143 = 6.110, p = 0.003). Larval and pupal Table 4. Number of species in each combination category of overwintering stage and host plant type. Overwintering stage Egg Larva Pupa Host plant type, n herbs, 6 grasses, 18 trees, 20 herbs, 24 grasses, 35 trees, 2 herbs, 35 grasses, 4 trees, 17 Mattila et al. overwinterers declined more than egg overwinterers (pairwise comparisons p = 0.005 and p = 0.004, respectively). Type of host plant had an effect on distribution change in monophagous species (simple effects test, F 2,143 = 4.236, p = 0.016) such that species feeding on herbs declined more than species feeding on grasses or on broadleaved trees (pairwise comparisons p = 0.005 and p = 0.022, respectively). Larval specificity had an effect on distribution change in species feeding on herbs (simple effects test, F 2,143 = 4.468, p = 0.013) and on species feeding on broadleaved trees (F 2,143 = 4.472, p = 0.013). For species feeding on herbs, monophagous species declined more than oligophagous or polyphagous species (pairwise comparisons p = 0.006 and p = 0.007, respectively). In species feeding on broadleaved trees, oligophagous species declined more than monophagous or polyphagous species (pairwise comparisons p = 0.034 and p = 0.003, respectively). Overwintering stage had an effect on distribution change of monophagous (simple effects test, F 2,143 = 3.546, p = 0.031), oligophagous (F 2,143 = 3.554, p = 0.031), and polyphagous (F 2,143 = 8.653, p < 0.001) species. In monophagous species larval overwinterers declined more than pupal overwinterers (pairwise comparisons p = 0.009), whereas in oligophagous species, pupal overwinterers declined more than egg overwinterers (p = 0.009). In polyphagous species larval and pupal overwinterers declined more than egg overwinterers (pairwise comparisons p = 0.033 and p < 0.001, respectively). The effect of larval specificity on distribution change was nearly significant in egg and in larval overwinterers (simple effects test, F 2,143 = 2.912, p = 0.058 and F 2,143 = 2.919, p = 0.057, respectively) and clearly significant in pupal overwinterers (simple effects test, F 2,143 = 6.032, p = 0.003). In egg overwinterers monophagous species declined more than oligophagous or polyphagous species (pairwise comparisons p = 0.043 and p = 0.024, respectively). In larval overwinterers monophagous species declined more than polyphagous species (pairwise comparisons p = 0.024). In pupal overwinterers oligophagous species declined more than monophagous or polyphagous species (pairwise comparisons p = 0.001 and p = 0.045, respectively), and polyphagous species declined more than monophagous species (p = 0.011). Discussion When we analyzed the traits separately, it seemed that overwintering stage and type of host plant were both important determinants of the distribution change. The ef- 5 fect of overwintering stage may be connected to climate change because there is a gradual change of the overwintering stages over a latitudinal gradient (Virtanen & Neuvonen 1999). Species overwintering as eggs are usually southern species (Virtanen & Neuvonen 1999) that are limited by the climate and thus may have benefited from warming climate. On the other hand, the reason species feeding on trees seem to have succeeded better may be due to changes in grassland and forest management (e.g., the widespread decrease of open habitats and afforestation [Mikkola 1997]). Nevertheless, overwintering stage and type of host plant are correlated (Niemelä et al. 1982; Virtanen & Neuvonen 1999), and these kinds of correlations among traits may confound conclusions about their role in influencing the observed changes in lepidopteran fauna. Thus, traits should also be analyzed simultaneously to partial out the correlated effects and to determine the possible interaction effects between the traits. When we analyzed all traits in a single model, factors contributing to the distribution change were much more complicated than our first analysis led us to believe. Distribution change was not similarly affected by overwintering stage in all host plant types; the role of larval specificity seemed to depend on type of host plant and on overwintering stage. Indeed, the distribution change was determined by body size and by 3 interactions: type of host plant and overwintering stage; type of host plant and larval specificity; and overwintering stage and larval specificity. Type of Host Plant and Overwintering Stage Among tree-feeding species, egg overwinterers were the most successful. Warming climate may have increased the advantage of this trait combination. Species that overwinter as an egg and feed on trees are often not covered by insulating snow because eggs are laid on trees (Virtanen & Neuvonen 1999). Without insulating snow cover these species are more prone to climatic variations and may thus benefit from warming climate, whereas thinning snow cover may decrease the overwintering success of pupae and larvae, stages that are usually covered by snow. Egg overwinterers tend to fly late in the summer or in the autumn, and warmer weather during autumns may benefit these species when adults have a longer time to reproduce and disperse. Egg overwinterers may also be advantaged by warmer springs because species that lay their eggs on trees do not lose contact with their host plants and are ready to feed in the spring when leaves start to develop. Among larval overwinterers, type of host plant had an effect such that species feeding on grasses declined the least. There were only 2 larval overwinterers that feed on trees in the data set, so the difference between species feeding on trees and on grass was not necessarily general. Conservation Biology Volume **, No. *, 2008 Species’ Ecological Traits and Distribution Change 6 The difference in distribution change among larval overwinterers that feed on herbs and on grass might be related to the unequal decline of herbs and grasses caused by changes in land use (Mikkola 1997). Why this is not shown among other overwintering stages is a question that requires further attention. Host Plant Type and Larval Specificity In polyphagous species it seemed that type of host plant type was not important because all polyphagous species succeeded quite well, whereas in monophagous species those feeding on herbs declined the most. This may again be explained by changes in land use because herbs have suffered especially from changes in land use (Mikkola 1997). Being dependent on only one host plant makes species more prone to extinction (Koh et al. 2004; Kotiaho et al. 2005; Mattila et al. 2006), and stability of resource availability affects extinction risk of moth populations (Nieminen 1996). Herbs and grasses are more likely to suffer from instability than large, more tolerant, and deep-rooted trees. It is also predicted that the abundance of deciduous trees increases as climate warms (Kuusisto et al. 1996). Larval specificity is also connected to dispersal ability (Warren et al. 2001) and monophagous moth species have been reported to have lower migration rates than oligophagous or polyphagous species (Nieminen et al. 1999). In addition, dispersal ability can be an important factor in distribution change, but unfortunately data on dispersal ability of noctuid moths are not available. In butterflies the general trend seems to be that mobile generalist species are increasing (Warren et al. 2001; Kotiaho et al. 2005), and this same pattern may be true in noctuid moths as well. Overwintering Stage and Larval Specificity In oligophagous and polyphagous species egg overwinterers fared best, whereas in monophagous species egg overwinterers declined the most. We have no comprehensive explanation for these results, but overwintering as an egg is a southern adaptation (Virtanen & Neuvonen 1999) and southern species at the edge of their distribution may have an ability to track the warming climate. Polyphagous and oligophagous species have better dispersal ability (Warren et al. 2001; Komonen et al. 2004) and use a wider array of resources because they feed on several species. These facts may explain the better success of polyphagous and oligophagous species. Body Size Larger-sized moths were predisposed to distribution decline. This effect was revealed only when analyzed simultaneously with the other traits. It is generally assumed that large species are more sensitive to habitat fragmentation because of greater space use and food-resource re- Conservation Biology Volume **, No. *, 2008 quirements (Cyr 2000), but this assumption ignores the fact that large species may also, because of their higher mobility (Komonen et al. 2004), be able to use multiple patches (Hambäck et al. 2007). Therefore, small and large species may be affected by different aspects of fragmentation: small species may be most affected by the path size reduction, whereas large species may be more affected by the total amount of suitable habitat in the landscape (Hämbäck et al. 2007). Body size is connected to specificity of larval host plant in geometrid and noctuid moths, and more specialized species are smaller (Niemelä et al. 1981; Lindström et al. 1994). This may lead to the conclusion that smaller size predisposes moths to distribution decline. Nevertheless, when the effect of larval specificity was removed, the result was just the opposite; larger size actually predisposed a group to distribution decline. Conclusions When the traits of noctuid moths were analyzed separately, overwintering stage and type of host plant affected distribution change. Nevertheless, when we analyzed the traits simultaneously, interesting interactions were revealed and body size predicted distribution change. This suggests that distribution change is not determined by only a single trait and that the effect of certain traits may depend on the other interacting traits. Such complexity makes it difficult to understand the processes behind distribution changes and emphasizes the need for basic ecological knowledge of species. With such basic knowledge, a more accurate picture of the factors causing distribution changes and risk of extinction may be attainable. Acknowledgments We thank the "Monday coffee club" members, M. Luoto, and 2 anonymous referees for their valuable comments on the manuscript. This study was supported by the Academy of Finland and by the Center of Excellence in Evolutionary Research. Literature Cited Cardillo, M., A. Purvis, W. Sechrest, J. L. Gittleman, J. Bielby, and G. M. Mace. 2004. Human population density and extinction risk in the world’s carnivores. Public Library of Science Biology 2:e197. Conrad, K. F., I. P. Woiwod, M. Parsons, R. Fox, and M. S. Warren. 2004. Long-term population trends in widespread British moths. Journal of Insect Conservation 8:119–136. Conrad, K. F, M. S. Warren, R. Fox, M. S. Parsons, and I. P. Woiwod. 2006. Rapid declines of common, widespread British moths provide evidence of an insect biodiversity crisis. Biological Conservation 132:279–291. Cyr, H. 2000. The allometry of population density and inter-annual variability. Pages 267–295 in J. H. Brown and G. B. West, editors. Scaling in biology. Oxford University Press, Oxford, United Kingdom. Mattila et al. Dennis, R. L. H., and W. R. Williams. 1986. Butterfly diversity. Regressing and a little latitude. Antenna 10:108–112. Dennis, R. L. H., T. H. Sparks, and P. B. Hardy. 1999. Bias in butterfly distribution maps: the effects of sampling effort. Journal of Insect Conservation 3:33–42. Freckleton, R. P. 2000. Phylogenetic tests of ecological and evolutionary hypotheses: checking for the phylogenetic independence. Functional Ecology 14:129–134. Hambäck, P. A., K. S. Summerville, I. Steffan-Dewenter, J. Krauss, G. Englund, and T. O. Crist. 2007. Habitat specialisation, body-size and family identity explain Lepidopteran density-area relationships in a cross-continental comparison. Proceedings of the National Academy of Sciences of the United States of America 104:8368– 8373. Harvey, P. H., and M. D. Pagel. 1991. Comparative method in evolutionary biology. Oxford University Press, Oxford, United Kingdom. Harvey, P. H., and A. Purvis. 1991. Comparative methods for explaining adaptations. Nature 351:619–624. Heath, J., E. Pollard, and J. A. Thomas. 1984. Atlas of butterflies in Britain and Ireland. Viking, London. Huldén, L., A. Albrecht, J. Itämies, P. Malinen, and J. Wettenhovi. 2000. Atlas of Finnish Macrolepidoptera. Lepidopterological Society of Finland, Finnish Museum of Natural History, Helsinki. Koh, L. P., N. S. Sodhi, and B. W. Brook. 2004. Co-extinctions of tropical butterflies and their host plants. Biotropica 36:272–274. Komonen, A., A. Grapputo, V. Kaitala, J. S. Kotiaho, and J. Päivinen. 2004. The role of niche breath, resource availability and range position on the life history of butterflies. Oikos 105:41–54. Kotiaho, J. S., V. Kaitala, A. Komonen, and J. Päivinen. 2005. Predicting the risk of extinction from shared ecological characteristics. Proceedings of the National Academy of Sciences of the United States of America 102:1963–1967. Kuusisto, E., L. Kauppi, and P. Heikinheimo, editors. 1996. Ilmastonmuutos ja Suomi. Yliopistopaino, Helsinki. Lepidopterological Society of Finland. 2005. Perhoset harrastuksena. Suomen Perhostutkijain Seura ry, Helsinki, Finland. Available from www.perhostutkijainseura.fi/sps_harrastus.html (accessed August 2008). 7 Lindström, J., L. Kaila, and P. Niemelä. 1994. Polyphagy and adult body size in geometrid moths. Oecologia 98:130–132. Mattila, N., V. Kaitala, A. Komonen, J. S. Kotiaho, and J. Päivinen. 2006. Ecological determinants of distribution decline and risk of extinction in moths. Conservation Biology 20:1161–1168. Mikkola, K. 1997. Population trends of Finnish Lepidoptera during 1961–1996. Entomologica Fennica 8:121–143. Mikkola, K., and I. Jalas. 1977. Suomen perhoset. Yökköset 1. Otava, Helsinki. Mikkola, K., and I. Jalas. 1979. Suomen perhoset. Yökköset 2. Otava, Helsinki. Niemelä, P., S. Hanhimäki, and R. Mannila. 1981. The relationship of adult size in noctuid moths (Lepidoptera, Noctuidae) to breadth of diet and growth form of host plants. Annales Entomologica Fennica 47:17–20. Niemelä, P., J. Tahvanainen, J. Sorjonen, T. Hokkanen, and S. Neuvonen. 1982. The influence of host plant growth form and phenology on the life strategies of Finnish macrolepidopterous larvae. Oikos 39:164– 170. Nieminen, M. 1996. Risk of population extinction in moths: effect on host plant characteristics. Oikos 76:475–484. Nieminen, M., H. Rita, and P. Uuvana. 1999. Body size and migration rate in moths. Ecography 22:697–707. Pollard, E., and T. J. Yates. 1993. Monitoring butterflies for ecology and conservation. Chapman and Hall, London. Purvis, A., and A. Rambaut. 1995. Comparative analysis by independent contrasts (CAIC): an Apple Macintosh application for analysing comparative data. Computer Applications in the Biosciences 11:247– 251. Thomas, C. D., and J. C. G. Abery. 1995. Estimating rates of butterfly decline from distribution maps: the effect of scale. Biological Conservation 73:59–65. Virtanen, T., and S. Neuvonen. 1999. Climate change and macrolepidopteran biodiversity in Finland. Global Change Science 1:439–448. Väisänen, R. 1993. Valtakunnallinen yöperhosseuranta. Baptria 18:9– 11. Warren, M. S., et al. 2001. Rapid responses of British butterflies to opposing forces of climate and habitat change. Nature 414:65–69. Conservation Biology Volume **, No. *, 2008