Area threshold and trait–environment associations of butterfly assemblages in the Zhoushan Archipelago, China. (15th December 2020)

• Record Type: Journal Article
• Authors: Chen, Chuanwu; Xu, Aichun; Wang, Yanping
• Is Part Of: Journal of biogeography. Volume 48:Number 4(2021)
• Page Start: 785
• Page End: 797
• Publication Date: 2020-12-15
• Subjects: area thresholds -- body weight -- butterflies -- data transformations -- life‐history traits -- small‐island effect -- species–area relationship -- trait–environment associations -- wing loading -- wingspan; Biogeography -- Periodicals; 578.09
• DOI: 10.1111/jbi.14037
• Languages: English
• ISSNs: 0305-0270

Abstract

Abstract: Aim: An area threshold is usually considered as the point of abrupt change in the species–area relationship (SAR). Despite their widespread existence, few studies have considered the influence of data transformations on their presence or the mechanisms that generate them. The aims of this study were thus to test for the existence of an area threshold in butterfly assemblages and to compare trait–environment associations between small and large islands split by the threshold. Location: Zhoushan Archipelago, China. Taxon: Butterflies. Methods: We used the line‐transect method to survey butterflies on 42 islands from July to August 2014. We also collected data on three commonly used ecological traits of butterflies (wingspan, wing loading and body weight). We transformed the butterfly datasets into three types (log‐log, semi‐log and untransformed) and fitted each of them using a set of breakpoint regression models. We then split all islands into small and large islands based on the area threshold and analysed the trait–environment associations for each island group. Results: Two‐slope breakpoint models performed best for log‐log and semi‐log transformed datasets. However, the former exhibited a steep‐shallow shape with an area threshold of 0.820 km 2 and the latter was in shallow‐steep shape with a threshold of 17.022 km 2 . By contrast, no threshold was detected when using the untransformed power model. Moreover, for small islands, small area and large isolation were positively correlated with species with large wing loading and body weight. For large islands, however, the perimeter to area ratio (PAR) was significantly and positively related to wing loading. Main conclusions: We found that data transformations may influence the detection of area thresholds. Trait–environment analyses can further provide novel and mechanistic insights into processes generating area thresholds. We urge that the influence of data transformations and ecological traits of species should be explicitly considered in future studies.