Designing biodiverse urban gardens: ecological planning through computational techniques for pollinator promotion

Abstract

Urban green spaces have the potential to contribute to biodiversity, but they are often created for aesthetic purposes with out proper planning or consideration of the associated flora and fauna and the services they provide. This study applies a genetic algorithm to select optimal plant combinations, considering ecological characteristics (such as pollinator richness) and aesthetic characteristics (flowering phenology). Over 12 months, the flowering phenologies of 15 native plant species were recorded along with the floral visitors associated with them during their flowering period (spring and summer). Using this information, a qualitative plant–pollinator network was built, and genetic algorithm models were applied to it to identify plant species groups that maximize pollinator richness, flowering duration, or both characteristics simultaneously. Also, we evaluated what is the minimum number of species necessary to obtain these maximum values. In addition, surveys were conducted among landscape specialists to compare with the results of the genetic algorithm. A total of 37 species of floral visitors were recorded on the 15 focal plants. The minimum number of plants required when seeking to optimize pollinator richness was five species, only two species when optimizing flowering length, and six in the scenario that optimized both variables at the same time. When we compare the results of the applied model with the choices of landscaping specialists, these models differ, especially when choosing plants that maximize flowering time. Our results show that the genetic algorithm is a simple and accessible tool that allows planning urban green spaces, considering variables of different types.