The Heating of Prairie Systems project provided research experience to undergraduates at the University of Oregon. While the original research design and methods for this experiment were created and implemented by the Bridgham lab, students from the Environmental Leadership Program (ELP) have played a vital role in collecting and analyzing phenological data in various ways, following strict protocol under the supervision of a project manager. The 2016, 2017, and 2018 Climate & Phenology ELP teams visited the Willow Creek study site, located in Eugene, Oregon, during April and May of each year to collect phenological data.
Methods
Each of the three different sites is divided into 20 plots (60 total across the three sites) that were randomly assigned 4 climatic treatments (control, drought, heating, heating plus precipitation) and 5 replicates of each treatment. The plots are circular, 3m in diameter and broken up into a grid coordinate system. During spring of 2016-18, the Climate & Phenology teams recorded phenological data for each of the focal species, counting buds, flowers, reproductive stems and flowering stalks, and recording presence of senescence.
In addition, the teams also measured the normalized difference vegetation index, or NDVI, using the Crop Circle ACS-430, a handheld remote sensing device. NDVI measures the amount of green vegetation. This measurement is quantifiable 
evidence of the overall photosynthetic activity, measuring plant canopy index, a direct indicator of plant primary productivity. Six NDVI measurements were taken at each experimental plot.
Each year, the ELP team had an additional focus. In 2018, the team studied plant demography by conducting censuses and germination counts. The 2017 team looked at the interaction of the treatments and competition on plant flowering. The western half of all the plots were weeded to distinguish the performance of the 12 species when they are in competition with other invasive or robust prairie plants.
The 2016 team performed pollinator surveys to see if the treatments had effects on pollinator visitation rates. This entailed performing 5 minute surveys for each plot if flowering was present. The team only recorded a pollinator as having visited a flower if it was observed on the stigma end of the flower. Three categories were used to classify pollinators: bees, flies, and other.
For the 2016 team, pollinator data was only collected when the weather conditions were appropriate, specifically, winds that are between 0 and 2 on the Beaufort Wind Scale. The team did not conduct surveys when it was raining. Additionally, cloud cover and temperatures were recorded before and after surveys.
Data Analysis
To analyze the phenology data for the 2015-2016 trial, the team compared timings of first flowering, first senescence, and first seeding for the 15 focal species by treatment. The team also compared peak flowering based on the highest number of flowers a species had within a plot for each of the three treatment groups.
Individual plant reproduction measurements were calculated by counting the reproduction of two randomly chosen flower stems, taking the average, and multiplying the average by the total number of flowering stems. The total reproduction was calculated from the sum of each individual plant’s reproduction. The Sidalcea malviflora data was analyzed based on treatment. This included inferring about how the basal area, number of flowers, and number of leaves were affected using analysis of variance (ANOVA) with a Tukey’s post-hoc test, using an alpha value of 0.05.
For each plot, the five NDVI measurements were averaged to obtain a single value for each plot for each recorded day. The data was then averaged based on treatment group. Then, the team examined change over time within the NDVI data and how the different treatments compared to each other. Additionally, a one way ANOVA with a Tukey’s post-hoc test was performed for each day the NDVI was recorded to determine if there was any significant statistical difference between treatment groups.
