The compositional similarity of urban forests among the world's cities is scale dependent

04 February 2016

Aim We examined species composition of urban forests from local to global scales using occurrence and abundance information to determine how compositional similarity is defined across spatial scales. We predicted that urban forests have become more homogeneous world-wide, which should result in minimal scale dependence that is more pronounced for non-native species, especially when considering abundance information. Location Thirty-eight cities world-wide. Methods We estimated compositional dissimilarities of urban forests, including both spontaneous and cultivated trees, from local to global spatial scales using six dissimilarity metrics. We used redundancy analysis to determine how climate, geographic distance and anthropogenic factors are related to compositional dissimilarity among cities. These analyses were implemented for all species combined and for native and non-native species separately. Results The 38 cities contained a median of 77 tree species, with a greater percentage of these classified as native (median = 58%). The similarity of urban forests was scale dependent, declining as the spatial scale increased – an outcome that did not differ when considering native and non-native species separately. Climate, geographic distance and city age were the main factors describing variation in tree species composition among cities. The addition of abundance information resulted in lower dissimilarity across spatial scales. Main conclusions Compositional similarity of urban forests is a scale dependent phenomenon that is not affected by the presence or absence of non-native species, suggesting a limited role for biotic interchange in promoting homogenization. However, compositional similarity across spatial scales increased uniformly with the addition of abundance information, suggesting that patterns of abundance may have greater biological relevance when homogenization trends among urban forests are considered.