Abstract
Aim The species–area ( SAR ) and species–time relationships ( STR ) are of vital importance in community ecology. Previous studies suggest that a unified, general species–time–area relationship ( STAR ) may hold, with non‐independent scaling of richness across space and time. Most STAR studies to date have considered species accumulation curves in relatively homogeneous habitats. Here, we test the generality of the STAR in an island system and assess how factors other than area influence species richness, accumulation and turnover through time. Location Thousand Island Lake, China. Methods We surveyed bird communities on 36 islands using line transects, and calculated annual species richness of breeding birds from 2007 to 2015. We built island STAR models at island (island STAR ; ISTAR ) and transect levels (local community–time–area relationship; LCTAR ). We employed partial correlations and multiple regressions to examine potential influences of island attributes other than area (i.e. isolation, edge effect and habitat richness) on slopes of STR s. Results ISTAR and LCTAR models explained 88.8% and 83.1% of total variance, respectively, and both models have a negative space–time interaction. Richness scales comparably in space and time, for both whole‐island and transect‐level analyses. The partial correlation analysis showed that distance to mainland and perimeter‐to‐area ratio are significantly positively correlated with the time scalar ( w ), and habitat richness and w are negatively correlated. Multiple regression models identify perimeter‐to‐area ratio as particularly influential. Main conclusions The STAR pattern generalized to an island system where species turnover is high, indicating an interdependency of time and space in determining species richness. Islands have attributes other than area that influence patterns of species accumulation and turnover through time. Ecologists should consider the interdependence of space and time when characterizing species richness patterns.