Monitoring and modelling biodiversity in relation to the prominent agri-environment scheme of wildflower areas

Biodiversity loss is a major global crisis affecting every species group including insects, reptiles and birds, and is driven by several factors such as habitat loss and fragmentation.  Among the conservation actions designed to mitigate these problems are agri-environmental schemes (AES) such as wildflower areas.

In previous studies, AES have demonstrated some potential for supporting insect and bird species, but these findings have not been replicated consistently. In many regions AES are deployed opportunistically, not necessarily where they would be most effective. Attempts to improve conservation actions are hampered by a severe lack of data on many species’ groups and on the efficacy of conservation measures.

Until recently, most monitoring methods have been resource intensive which limits how many sites can be surveyed. This means that knowledge of species distributions is patchy, and that it is difficult to determine the efficacy of conservation measures. Some methods such as species distribution modelling can use even limited data to draw conclusions or inform conservation actions, but these generally focus on very broad phenomena rather than local factors. New monitoring and modelling methods may thus offer significant benefits for conservationists.

We performed three different studies focused on the state of Schleswig-Holstein, focused on biodiversity monitoring and modelling. Field work was conducted at 41 sites across Schleswig-Holstein employing multiple different methods, while modelling drew on our field results and from existing monitoring schemes within the state.

In Chapter 1, we designed, built and tested a machine learning classifier for Orthopteran insect sounds and assessed the efficacy of passive acoustic monitoring for these species using Audiomoth recorders.  In Chapter 2, we used local data to build a species distribution model for the locally threatened sand lizard Lacerta agilis and validated this model with field surveys. In Chapter 3, we built and tested a RangeShifter individual-based model using real-world population data and used landscape modelling to assess how changing the quantity and placement of wildflower areas might influence the abundance and distribution of the grey partridge Perdix perdix over 15 years.