Abstract:Biodiversity is one of the key factors affecting ecosystem services, and there are very close connections between agricultural ecosystem and species diversity. The expansion of agricultural land and enhancement of land intensive utilization cause great pressure on the protection of agricultural landscape biodiversity. In highly intensive agriculture landscape, woodlands are important habitats for biodiversity protection. Reduction of the vegetation coverage can lead to changes of habitat patch composition and configuration, which mainly including the decrease of mean patch size and increase of patch isolation degree and number of patches, then impacts species dispersal and settlement. Landscape heterogeneity can affect a variety of ecosystem processes, functions and ecosystem services. The composition and structure of biological habitats in agricultural landscape have an important role in maintaining species diversity. This also makes the transformation of biodiversity protection approaches from species centered to multifunctional landscape planning. Though the composition of landscape habitat patches is crucial for species survival, whether organisms can migrate successfully from one habitat patch to another or not also depends on the landscape connectivity between the two patches. The landscape connectivity level is determined by landscape structure and organisms' behavior characteristics together. To avoid species loss in landscape level, enhancing landscape connectivity becomes particularly important. The aim of this study is to investigate the contribution of the proportion of woodlands area (PA), number of patches (NP), mean patch size (PS), mean patch isolation (PI) for habitat availability (characterizing by probability index of connectivity, PC) along a gradient of woodlands area. Zhengzhou, Henan Province, a typical agricultural region in the middle and lower reaches of the Yellow River, was selected as the case study area. The mixed-effects models (Multi-model Inference, MMI) were applied to model the effects of woodlands landscape characteristics and habitat availability. The explanatory variables are landscape characteristics, namely PA, NP, PS, PI, and eight hypothetical species dispersal abilities (10、50、100、250、500、1000、2000 m and 3000 m). 260 landscapes units are classified into four scenarios according to the proportion of woodlands area, high ( > 50%), intermediate (50%-30%), low (30%-10%) and very low ( < 10%). The analytical results show that, in regions with high proportion of woodlands area ( > 50%), the proportion of woodlands area is the most important factor that affects the degree of habitat availability. In regions with intermediate proportion of woodlands area (50%-30%), besides the proportion of woodlands area, mean patch isolation and mean patch size are the main factors that must be considered. In regions with low and very low proportion of woodlands area ( < 30%), depending on the different species dispersal abilities, it is need to consider the effect of mean patch size and number of patches. For landscape management, we suggest that, conservation efforts should focus on maintaining the current forest, especially in landscapes with high proportion of woodlands area. While in landscapes with intermediate and low proportion of woodlands area, restoration should aim at enhancing habitat availability and reducing isolation. In the regions with very low proportion of woodlands area, managers must put the habitat restoration in the first place to avoid local extinction.