Geographic patterns of seed mass are associated with climate factors, but relationships vary between species.

Abstract

Seed size is a critical life history attribute with fitness effects that cascade throughout the lifespan of plants. Interspecific studies repeatedly report a negative correlation between seed mass and latitude. Yet, despite its importance, little is known about geographic variation in seed size within species' ranges. To improve our understanding of intraspecific geographic variation in seed size, we collected and weighed seeds by maternal line from 8 to 17 populations of seven herbaceous plant species spanning large geographic areas, and measured a dispersal trait, awn length, for two grass species. We examined the overall relationship between seed mass and latitude, then divided the data into species-specific subsets to compare the fit of three models to explain seed mass and awn length: (1) latitude and longitude, (2) long-term climate, and (3) collection-year weather. Like previous work, we found a negative relationship between interspecific seed mass and latitude. However, the best-fit models explaining seed size and awn length differed between individual species and often included significant interaction terms. For all species, the best model was either long-term or collection-year climate data instead of latitude and longitude. Intraspecific geographic patterns for seed traits were remarkably inconsistent, covarying both negatively and positively with temperature and precipitation. The only apparent generalization is that annual species' seed mass corresponded more with collection-year weather while perennial species covaried more with long-term climate. Overall, this study suggests that the scale of climate variation that molds seed traits is highly species-specific.