Ecological restoration can be viewed as guided community assembly, with seed mixes serving as inputs to build realized plant communities. Commercially available seed mixes are an important tool for restoring degraded habitats, such as tallgrass prairie, one of the most endangered habitats on earth. However, little is known about how commercial mixes compare to existing remnant and restored plant communities in terms of species composition and diversity. Using data from companies across the Midwestern United States, we quantified biodiversity of commercially available mixes, and compared these diversity measures to remnant and restored tallgrass prairie plant communities. We analyzed multiple facets of biodiversity, including species richness, coefficient of conservatism, bloom-time coverage (phenological diversity), and the degree of evolutionary relatedness between species (phylogenetic diversity). Each of these components of diversity contributes to important ecosystem functions, and therefore, analyzing multiple diversity measures allows us to draw connections between mixes and potential ecosystem functions of their resultant prairies. We also studied whether seed mix price is predictive of diversity. Lastly, we used machine learning to design seed mixes that meet multiple biodiversity objectives simultaneously, at realistic price points.
In all, we studied 67 seed mixes from 14 companies, comprised of 215 species from 36 families. Richness of restored mixes ranged from 4-91 species (mean 34 spp.). Commercial seed mixes were less diverse than remnant prairies in terms of native species richness, and phylogenetic diversity (MPD), but did not differ from restored prairies. On the other hand, commercial seed mixes had higher mean coefficients of conservatism than both remnant and restored prairies. In addition, analysis of seed mix cost and diversity metrics showed that there are greater ecological returns with increasing financial investment. That is, more-expensive mixes had higher taxonomic diversity, conservatism, and phylogenetic diversity. Designed seed mixes were constrained by the commercial availability of species and/or groups in the same way that typical mixes would be (i.e., relatively few spring blooming prairie species are commercially available, regardless of cost, therefore bloom-time coverage of designed mixes was constrained). Based on our results, we propose a tool to aid in species selection for restoration under multiple biodiversity objectives and cost constraints.