Modelling patterns of pollinator species richness and diversity using satellite image texture

Area of Science:

• Ecology
• Remote Sensing
• Biodiversity Research

Background:

• Standardized field sampling for species richness and diversity is costly and time-consuming.
• Satellite remote sensing (RS) offers a cost-effective method for collecting extensive spatial data to model biodiversity.
• Spatial habitat heterogeneity, often reflected in image texture, is a key determinant of species distribution and diversity.

Purpose of the Study:

• To assess the utility of image texture measures from Landsat-TM imagery as a proxy for spatial habitat heterogeneity in modeling local pollinator biodiversity.
• To test the ability of texture features to predict bee diversity and species richness using a multimodel inference approach.
• To compare the predictive power of texture metrics for bee biodiversity with findings from previous avian biodiversity studies.

Main Methods:

• Utilized four years of bee monitoring data (2010-2013) from six 4 × 4 km field sites in Central Germany.
• Derived texture features from Landsat-TM imagery to represent spatial habitat heterogeneity.
• Employed a multimodel inference approach to model local pollinator biodiversity based on texture features.

Main Results:

• Texture features, particularly first-order entropy and terrain roughness, showed some ability to reflect patterns of bee diversity and species richness.
• Texture measurements explained only 3-5% of the variability in the final biodiversity models, which accounted for up to 60% of the total variability.
• Results were consistent across different bee groups (bumble bees, solitary bees), but predictive power was lower than reported for avian biodiversity.

Conclusions:

• Satellite imagery textures show potential for modeling bee biodiversity patterns, but their explanatory power is limited.
• Landscape configuration heterogeneity captured by texture may be less functionally important for wild bees than plant community composition.
• Taxa-specific variability exists in the effectiveness of texture metrics for biodiversity modeling, highlighting the need for nuanced approaches.