Relationships between flowering phenology and community composition in an experimental restoration of northwest prairies

Phenology, the timing of biological life cycles, is a key indicator of global climatic change, with numerous studies showing that species' phenologies are shifting in response to climate change. Despite general trends (e.g., warming causing earlier arrival of spring events such as leaf-out and flowering onset), studies repeatedly reveal that phenological changes tend to be species-specific and thus may alter species interactions. Less studied is the potential feedback between biotic interactions and phenology and the impacts on species’ fitness. To understand the consequences of shifting phenology for species and communities, we need to quantify how phenology and competition interact to affect species’ fitness. Here, we studied the potentially interacting effects of species’ phenology and competition on plant fecundity (as a proxy for fitness). We sowed seeds of various species combinations to test how variation in competitor species richness, identities, and densities affect the phenology and fecundity of an annual wildflower, Clarkia purpurea, using an unconventional experimental design to encourage public engagement with the experiment. We found that C. purpurea’s flowering phenology varied with competitor identity and competitor species richness, that fecundity was negatively correlated with competitor density but not species richness, and that the strength of competition tended to vary by competitor identity but appeared unrelated to the relative phenology of the competitor. These findings offer unique evidence that competitive interactions may impact plant phenology and fecundity in complex ways and could influence species’ persistence and coexistence conditions in our changing global environment.

Methods
We conducted our research along the south bank of the Willamette River in Eugene, Oregon (44.05 ºN, 123.07 ºS, approximately 130m elevation). This region has a Mediterranean climate with warm, dry summers and cool, wet winters (12.06 ºC average annual temperature, 961.1 mm average annual precipitation). Historically, the research site was composed primarily of riparian hardwood forest and upland prairie vegetation before being degraded by industrial land use (e.g., railroad development, gravel mining, and fill deposit) starting in the mid-19th century. Since 1989, it has been under the management of the University of Oregon and is currently part of a 24-acre area designated as the Willamette River Natural Area (WRNA) which aims to promote habitat restoration, education, research, and recreation opportunities (Krueger 2022). Drawing on our site’s history of habitat loss and its location within the WRNA, we aimed to conduct this experiment in a way which informed local restoration efforts and engaged the community in the study of phenology and environmental change (Fig. 2).  

Study System

Following the restoration goals specified for our site by the WRNA management plan, our study focused on plants native to Willamette Valley northwest prairies. This ecosystem type has been reduced to less than two percent of its pre-settlement area, consequently making it one of the most threatened ecosystem types nationally (Willamette Valley Oak and Prairie Cooperative 2020). Prior to the onset of our experiment, the dominant plant species at the research site were non-native shrubs (primarily Rubus armeniacus), grasses, and forbs. To prepare for sowing our experiment, we mowed, covered the site with plastic (i.e., solarized) to reduce the existing seedbank, and used spot torching to remove remaining plant material. We then sowed seed of nine annual prairie forbs species of restoration interest: Collinsia grandiflora (COLLIN), Plectritis congesta (PLECON), Plagiobothrys figuratus (PLAFIG), Clarkia purpurea (CLAPUR), Collomia grandiflora (COLLOM), Madia sativa (MADSAT), Gilia capitata (GILCAP), Epilobium densiflorum (EPIDEN), and Navarretia squarrosa (NAVSQU). Although these species are relatively common species of northwest prairies and may have been historically present at our site, they were not abundant (or likely present at all) in the immediate area of the study, which is heavily invaded by non-native species. We used annuals as their short life cycles are conducive to short-term experiments asking questions about competitive interactions, phenology, and fecundity. We selected annuals that are commonly used for restoration in our region and vary in their respective flowering phenology (Fig. S3). This allowed us to answer our question regarding competition between species with different flowering times and created an extended spring/summer flowering display along the pathway, thereby drawing public interest in the experiment.

Experimental Design

We created two experimental planting areas: a 30m-long planting strip and a 20m-diameter planting circle, each consisting of plots varying in composition, density, and richness of species against which C. purpurea would be competing (Fig. 2 a-c). Although not extensively discussed here, it is worth noting that this experiment was designed to test the study’s ecological questions while also engaging the public in questions of phenology, species coexistence, and climate/environmental change (Fig. 2d). The planting strip contained 16  plots of annual prairie plant monocultures as well as one 10 density gradient plot with C. purpurea and C. grandiflora and four  reduced competition plots containing C. purpurea growing without immediate neighbors (Fig. 2c). The “reduced competition” plots help for fitting competition models (Hart et al. 2018). The additional purpose of the planting strip was to display the species in monocultures, roughly in order of flowering times, for public education. The planting circle featured multiple rings of varying species richness: monocultures, two-species plots, three-species plots, and a center core containing all nine of our study species (Fig. 2b). The multispecies plots contained C. purpurea with different unique combinations of our study species. Seed quantities for each species were estimated by weight to try to obtain an equal number of germinants of each species (at a target density of 300 plants per square meter, adjusting seed quantity by estimated germination rates). However, final densities for all analyses were based on counts of actual germinants. Initial seed sowing occurred in January 2023 with some reseeding in March 2023. Throughout the duration of the experiment, we manually removed weeds from the plots.

Fecundity and Phenology Measurements

To assess how competitive neighborhoods impact the flowering phenology and fecundity of C. purpurea, we marked 125 focal individuals of C. purpurea within the two planting areas, capturing a gradient of competitor density and species richness. We quantified the competitive neighborhood of each focal plant as the identity and number of plants rooted within a 10cm radius of the focal. We monitored the phenology of focal plants from the beginning of May at a frequency of three times per week prior to the start of C. purpurea’s flowering season and daily henceforth, until the end of August. C. purpurea produces bowl-shaped flowers; an open flower was defined as one in which all four whorls (sepals, petals, stamens, and pistil) were fully visible. Once the first focal plant was observed in flower, we visited plots daily to record the presence of open flowers on all focal plants. We derived first flower dates for our other eight study species from observations of the monoculture plots of each species in the planting strip.

Once the C. purpurea focal plants matured and reached the end of their reproductive periods, we counted the number of fruits present on each plant. We converted these to a measure of fecundity (i.e., seed counts) by collecting and dissecting fruits from focal plants in backgrounds of varying competitor densities. To determine whether average seed-per-fruit counts varied based on competitor density, we then fit a linear regression. Since there was a significant difference between seed counts per fruit within the different competitor density environments (linear model, F2,43 = 4.57;   , we calculated mean seeds-per-fruit estimates in three competitor density categories (zero, medium, and high densities, corresponding to competitor densities of 0, < 25, and >= 25 respectively). Finally, we used these estimates to convert each plant’s fruit count into a fecundity measurement which accounted for the density of its competitors. This fecundity measurement for each plant was used as a proxy for plant fitness since C. purpurea is an annual species.