Species-specific phenological responses to changing climate are reshuffling the timing of species interactions, however, we don’t fully understand the consequences of these changes for species’ population dynamics and community composition. In this study, we experimentally manipulated the timing of germination for five annual plant species from southern California and used pairwise competition experiments and coexistence theory to quantify how phenological shifts may impact species interactions and coexistence. We found that phenological shifts may help promote coexistence when they confer an advantage for competitively inferior species, but in other cases promote dominance by competitively superior species. Earlier germination generally increased species’ performance relative to competitors, but the relative changes in intra-and inter-specific interactions caused more complex effects on niche and fitness differences. Phenological differences tended to reduce stabilizing niche differences for many species pairs and reduced overall coexistence probabilities. While phenological differences among species have typically been considered a form of niche partitioning, it seems increasingly likely that phenological offsets could destabilize species coexistence. The net effects of changing phenology on species coexistence will depend on the complex combinations of effects on intra- and inter-specific interactions, which remain challenging to predict.

We studied five study species: two non-native grasses, Bromus diandrus (BRDI) and Bromus rubens (BRRU), and three native forbs, Lasthenia gracilis (LAGR), Nemophila menziesii (NEME), and Plantago erecta (PLER). The native species are common members of the herbaceous community, and the two non-native grasses are both species that are well-known to become abundant at the expense of local native species diversity and abundance.

These data come from a common garden competition experiment at the University of California, Riverside’s Agricultural Operations facilities, in Riverside, CA. The experiment was designed to calculate pairwise interaction coefficients (including intra-specific interactions) for all combinations of the five species, yielding 10 unique species pairs. Each study species was sown into three plots (1m²) at low, medium and high densities, to form a competitive density gradient for each species. Into each plot we sowed focal individuals of each species, establishing all pairwise combinations of competitive interactions. The relative timing of germination between the focal species and competitors was manipulated by offsetting the introduction of seeds by three weeks in both directions. These species of annual plants all have fast germination rates when they are exposed to enough water (within 3-4 days), as they are adapted to germinate with the beginning of seasonal rains during the cool wet winters of southern California. Thus, we watered them after sowing seeds, causing 3-week differences in germination dates.  In "focal first" plots, the focal seeds were sown and then three weeks later the competitor species were sown (Figure S1). Similarly, "background first" plots were established by waiting three weeks after sowing competitor species to introduce the focal species. Finally, a "same time" treatment was established by sowing focal plants and competitors at the same time, mid-way between the first and last dates.  Each species was also sown into plots without competitors to help estimate the performance of each species in the absence of competition. This design resulted in a total of 2850 focal individuals in 114 plots (5 species x 3 background densities x 2 replicates x 3 phenology treatments + 24 “no competition” plots). The garden was irrigated and weeded regularly to maintain suitable growing conditions and as “weed-free” pairwise competition plots as possible.

The reproductive output of focal individuals was measured at the end of the growing season, along with the number and identity of competitors within a 20cm diameter neighborhood. As seed counts were logistically impossible, we counted flowers of all focals and obtained estimates of number of seeds per flower for each species. Prior to counting a total of flowers, 10 flowers per species had seeds extracted and counted to determine an average seed count per flower. These quantities were used to convert flower counts to seed estimates for each focal plant. Seed germination rates were estimated from a combination of buried seeds in nylon mesh bags, and germination trials in a greenhouse. We tried a variety of seed survival rates to mirror the typical pattern that the native species in this ecosystem have longer-lived seed banks than the non-native grasses, however, all results were quantitatively and qualitatively robust to very different seed survival values.

Thus, the core dataset contains individual plant seed production as a function of the composition and density of neighbors.