Ecological thresholds in herb communities for the management of suburban fragmented forests

Abstract

Fragmentation is a major cause of biodiversity loss. The relationship between fragmentation and biodiversity might be non-linear, due to the presence of ecological thresholds; however, the application of the threshold concept on the conservation of herb biodiversity has received limited attention. We assessed the relationship between a pool of woodland indicators and features of forest fragments (area, isolation, shape), and we evaluated whether ecological thresholds might provide guidelines for conservation. We performed our study in a human dominated area of northern Italy. In order to obtain a group of floristic indicators, we chose perennial herbs characteristic of the phytosociological syntaxa Fagetalia sylvaticae, Carpinion betuli and Erythronio-Carpinion, with suitable Landolt L and H index, obtained from previous relevés in reference and high-quality forests. Subsequently, we analyzed the relationship between richness of indicators and patch features recorded in a highly fragmented landscape. Generalized linear models showed that the number of forest indicators was positively related to fragment size and connectivity. Generalized Additive Models showed that the relationship between indicator richness and fragment size was non-linear, with the presence of two thresholds. The number of forest indicators increased until 35–40ha; above this size the richness remained constant. Fragments above this threshold should be the target of conservation. Furthermore, the relationship between area and number of indicators was less steep in patches smaller than 1–1.5ha. Patches between these two thresholds can be optimal targets of management actions, as small size increases can greatly improve their value for biodiversity. The identification of two thresholds in the relationship between patch area and indicators suggests an abrupt response to changing landscape features and provide quantitative targets for conservation.