Habitat fragmentation is a major threat to biodiversity and it is up to conservationists to determine the amount of habitat needed to preserve a species. The process of decreasing size and increasing isolation of habitats is very evident when looking at semi-natural pastures. In Sweden 82% of the semi-natural pastures have disappeared since 1880 (Angelstam et. al. 1993). The decline strongly affects those species that are dependent on low- nutrient, unfertilised pastures. This includes butterflies, which are now rapidly declining all around Europe, mostly due to the intensification of agriculture and change in land-use. Most butterfly species can be considered as living in metapopulations, where groups of local populations are connected through dispersal and immigration between the groups and where the population size fluctuates due to extinction and recolonization (Hanski et. al. 1995). It is expected that many species in a community of butterflies respond to a decrease in area of a preferred habitat in the whole landscape. Several researchers have shown that patch size alone cannot predict a population’s persistence, the landscape surrounding the patch is also important (Thomas et. al. 1992; Bergman & Landin 2001).
The aim of this study was to investigate how the butterfly community changes in a coniferous dominated landscape with different amount of meadows and pastures surrounding each butterfly site. The butterfly diversity was expected to be lower in pastures with small amount of semi-natural pastures in the surrounding landscape. Besides the landscape factors the relative importance of the local habitat quality was also examined.
The study was conducted in Östergötland, Sweden, and the study area consisted of semi-natural pastures with deciduous forest situated in a matrix of coniferous forest with small amount of arable fields. In total 60 pastures were selected and for each study site the total area of meadows and pastures in the landscape within three different radii (500, 2000 and 5000 m) were calculated and used as landscape variables. Nine local habitat factors thought to affect the butterfly community were also recorded. Butterfly recordings were conducted five times on each site between May- September 2004. The transect-line method by Pollard (1977) was used to record the butterflies. The statistical analyses were performed with the CANOCO 4.5 software using multivariate methods based on linear assumptions.
A Principal Component Analysis (PCA) was conducted to describe structures in species data. The results showed clear patterns in butterfly composition and the relative importance of all environmental variables was inferred from the intra-set correlation values. The first principal component (PC1) explained 25.6% of the variation in species data. The environmental variables that had highest correlation with PC1 were two local habitat factors, nectar supply and sward height. The three landscape variables had relatively low correlation values with the butterfly communities. Other factors had equal or more importance for the butterfly composition. Each landscape variable was further analysed in a partial Redundancy Analysis (pRDA). The local habitat factors were inserted as covariables so that the difference in site quality was extracted, thereby showing only the explanatory power of each landscape factor. It was only the amount of semi-natural pastures at the 5000 m scale that significantly explained some of the variation in butterfly composition. Overall the variation explained by the landscape variable at 5000 m was small, indicating that other factors were also important for the butterfly community. Hence, the importance of the landscape factors in this study was rather weak in contrast to what was expected. Since many butterfly species have a dispersal range around 1000 m and even up to 3000 m , the amount of habitat at 5000 m scale is relevant for the butterfly community. It has also been concluded that a minimum of 15-20 well connected patches are required for long-term persistence of a metapopulation of butterflies (Hanski & Gilpin 1997). In order to preserve that many patches, the 5000 m scale is necessary. Besides the effect of the landscape on the butterfly composition this study also showed that several local habitat factors were of importance. It was found that rather isolated patches had a rich butterfly fauna and that semi-natural pastures in a coniferous dominated landscape generally had more butterfly species compared with an agricultural landscape. This is probably due to the combination of a diverse matrix and a more recent history in decline of semi-natural pastures.
In summary, the butterfly communities in a coniferous landscape seem to be affected both by habitat quality and to some extent also by the landscape quality. This has implications for conservation management. It is very important to support a traditional land-use with a continuous grazing regime for the semi-natural pastures. In a matrix of coniferous forest it is efficient to continue to manage more isolated pastures since they show high butterfly diversity. Still, it may be important to consider the whole landscape in conservation strategies for long-term persistence of a rich butterfly fauna.
Responsible for this page: Agneta Johansson
Last updated: 05/19/05