General declines in Mediterranean butterflies over the last two decades are modulated by species traits

Species' responses to environmental changes are highly idiosyncratic and context-dependent. Although intrinsic traits (i.e. those that define species niches) may play a key role, little empirical evidence exists regarding their relationship to demographic responses. We used data for 66 butterfly species representing five ecological and two life-history traits to study the effect these factors have on population growth rates and variations in populations. Using a novel methodological approach, we provide here improved estimates of population change. Our results reveal declines in 70% and increases in 23% of the studied species, clear evidence of more serious population declines in Catalan butterflies than those that have previously been reported. Declines were associated with species' degree of habitat specialisation and the number of generations. For all species, fluctuations were greater within than between years and, on average, the latter was 1.5 times greater. Our results indicated that habitat specialists and multivoltine species are more likely to suffer severe annual fluctuations in population abundance; and that multivoltine species and extreme larval specialists had the most marked fluctuations within seasons. We also found higher resilience to environmental changes in generalist species, which is concordant with biotic homogenisation in disturbed communities. However, among the declining species there were also many generalists, which indicates a potential general reduction in this group that goes beyond faunal homogenisation. Given butterflies are biodiversity indicators, these patterns are a possible reflection of an overall impoverishment in biodiversity. •70% of Mediterranean butterflies show declining populations.•Declines are linked to species traits, being steeper in specialists and multivoltine species.•Seasonal and annual fluctuation of populations are highest in multivoltine species.•Models accounting for detection probability improve estimates from monitoring data.