This dataset contains measurements of ethylene production rates in five "cryptic" nitrogen-fixing niches from a study investigating rates and controls of nitrogen fixation in regenerating lodgepole pine forest plots in the Greater Yellowstone Ecosystem (GYE). Those five cryptic niches are free-living N fixation by asymbiotic heterotrophic bacteria in pine litter ("litter"), decay class 4 and 5 coarse woody debris ("wood"), surface mineral soil (upper 2 cm of A horizon; "soil"), and associative N fixation in ground mosses (consortia of Polytrichum and/or Pleurozium species on the soil surface; "moss") and lichens (consortia of Peltigera or Cladonia species on the soil surface; "lichen"). The measurements were made using samples collected from the field sites in Yellowstone National Park and either measured immediately after collection or taken back to the University of Montana and measured under experimental manipulations of temperature and moisture (Climate experiment). The field sites are all dominated by lodgepole pine that regenerated following stand-replacing fires in 1988. We collected samples of the five cryptic niches from each plot in May, August, and October 2022 and measured in situ actual rates (no water added) of ethylene production using a modified acetylene reduction assay (ARA). When exposed to acetylene gas, the enzyme nitrogenase, which is responsible for the reduction of N2 into ammonia, converts acetylene to ethylene gas, thus providing a measurable proxy for N fixation activity. Ethylene content in each sample was analyzed using a Shimadzu GC-2014 gas chromatograph equipped with a flame ionization detector. Vegetation samples were oven dried at 65°C, and soil samples were dried at 105°C for 48 hours and reweighed to determine gravimetric moisture content. To convert acetylene reduction (AR) rates to N fixation rates, we used the geometric mean of 15N2-calibrated R ratios for each niche as well as the minimum and maximum R ratios published in Soper et al. (2021; "Is 3 the magic ratio?") to calculate ranges of N fixation inputs.
Scaling field-based N fixation rate measurements
We used plot-level abundance, biomass, or mass/area data for each niche paired with ARA measurements to generate niche-specific spatial N fixation rates. Briefly, we estimated the percentage area cover of lichen and moss in each plot then paired the N fixation rates as measured by ARA with the known sample area of our sample collection tubes (4.91 cm2). For litter, we obtained dry biomass/area estimates per plot by collected litter from within 900cm2 throughout each plot then drying the sample to a constant mass. For soil, scaled N fixation rates (kg N ha-1 y-1) were calculated using the rate of N fixed per unit dry mass of soil and average sample bulk density (g cm-3) in each plot measured to a depth of 2 cm. However, we assumed an active soil N fixation depth of 10 cm in all spatially scaled estimates. We calculated plot-level dead wood biomass estimates for both decay classes using established protocols to upscale N fixation rates in decaying wood. Estimates of each niche’s biomass or area coverage are included in this data package.
We measured N fixation rates every two months within the snow-free period to calculate seasonal (“spring”, “summer”, “fall”) and ultimately annual N fixation rates. However, for the summer N fixation estimate, the study plots received 4.17 mm of rain the day after our initial summer sampling concluded. Thus, we used this opportunity to explore the effects of a mid-summer pulse of rain on N fixation by re-sampling all plots/niches within 24 hours of the rain event. As a result, the summer N fixation estimate was calculated using the mean of N fixation rates measured during the two sampling periods (“august” and “august after rain event”). Annual N fixation rates were calculated first using mean rates at the plot level scaled on an hourly basis using 1460 hours as a two-month long “seasonal” period. The seasonal rates were then summed to estimate an annual rate.
Climate experiment
Associative N fixation (1st round) — To explore the effects of temperature and moisture on associative N fixation rates (lichen and moss), we conducted a full-factorial (moisture x temperature) incubation experiment in the laboratory. In May 2022, we collected moss and lichen samples (96 per niche) from the five study sites. Samples were randomly organized into groups of six replicates, and assigned one of four moisture treatments: no treatment, two hours of drying at 35°C, three and a half hours of drying at 35°C, or treated with 1.2 mL of de-ionized water. The treatments were designed to create a gradient of moisture levels across all samples in each niche using the upper range of daily maximum temperatures in YNP (35°C). After treatment, samples were kept in partial sunlight in a greenhouse for 24-48 hours until the experiment began. Groups of 24 samples from each niche were then incubated in controlled environmental chambers at 5°C, 15°C, 25°C, or 35°C with 154-159 umol m-2 s-1 of photosynthetically active radiation to simulate understory light. This was done to acclimate the sample to their experimental treatments prior to ARA. Following a 24-hour incubation period acetylene reduction rates were measured over a 24 hour incubation period. Following the experiment, gravimetric moisture contents were obtained for each sample. Samples were then sorted into groups of 3-5 samples based on their actual moisture contents for analysis of ethylene production rates across the different levels of actual moisture content and temperature.
Heterotrophic N fixation (2nd round) — To explore the effects of temperature and moisture on heterotrophic N fixation rates (soil, litter, and wood), we conducted a similar but separate incubation experiment. In August 2022, we collected samples of mineral soil, decomposing plant litter, and decaying wood (160 per niche) from our field sites. Samples were transported to the University of Montana within 48 hours of sample collection. Given the dry conditions at the time of sampling, the collected samples were used to represent the lowest moisture conditions. Dry samples were treated with four different levels of de-ionized water to create a range of moisture contents. Samples were incubated at the same temperatures and processed in the same manner as described above. Experimental data are reported as hourly rates of AR per treatment; we did not attempt to convert or scale measurements.
Statistical Analysis
Included in this dataset are three scripts used to analyze our field measurement data and our climate experiment data. Field-based N fixation rates (by season) were calculated using plot level (n = 5 per niche) means for each niche. While we used our seasonal sampling design to scale up N fixation rates to the annual scale, we did not test for statistical differences in scaled seasonal rates as our measurement frequency within each season was not enough to justify this. Instead, we only test for differences in AR rates between sampling periods. We used non-parametric Kruskal-Wallis tests followed by Dunn’s post-hoc tests to examine AR rate differences, and average sample moisture content differences between sampling periods (including before and after rain comparisons) and niches (α = 0.05).
To analyze rates in the climate experiments, we first assessed whether temperature, moisture and their interaction increased the probability of a sample being non-detectable (“inactive”) or active by fitting a negative binomial generalized linear model using a logit link function, with AR rate as a binary response (0 or > 0) and temperature and moisture content as covariates. Only AR rates among lichen were more likely to be above zero at higher temperatures; other niches showed no increase in non-zero probability with higher temperatures or moisture contents. As we could not account for other factors that may drive zero vs non-zero data, we excluded inactive samples from our analysis. “Active” (non-zero) samples met assumptions of normality when ln-transformed. Thus, we tested effects of temperature as a second order polynomial on active AR rates with square root-transformed moisture content as a continuous covariate using Analysis of Covariance (ANCOVA, α = 0.05). All analyses were performed using R statistical software (see “Other Entities” in this Dataset).
