Theory of seed mix design with applications to ecological restoration

A major factor hindering ecological restoration is uncertainty about which plant species will best establish. We account for this uncertainty in the design of seed mixes, though our developments are relevant to other planting mixes (e.g. root stock). We view seed mixes as being comprised of one or more species groups (e.g. shrubs, grasses, and nitrogen fixers). We mathematically establish that chances of relatively low densities decline as a species group's seeding rate (e.g. 100 seeds/m2) is divided more evenly among more species. This decline is sharpest if survival probabilities vary widely among species. To determine how much survival probabilities typically vary, we studied grasses commonly seeded in Great Plains grasslands and Mediterranean annual grasslands in the western United States. Survival probabilities varied extensively, so the chances of low densities declined markedly with increasing seeded species numbers. In the Great Plains, the chances of establishment failures (0 plants/m2) were 50% when the seeding rate was allocated to one species versus 0% when the seeding rate was divided evenly among five or more species. Similarly, in Mediterranean annual grasslands, the chances of very low densities (≤1.0 plants/m2) declined from 24% when one species was seeded to 0% when three or more species were seeded. The seeding rate for each plant group should be divided as evenly as possible among as many species as practical. Compared to increasing seeding rates to provide greater densities, dividing fixed rates more evenly among more species could prove less expensive.