Skip to main content
Dryad logo

California subalpine forest post-fire diversity and productivity


Brodie, Emily (2021), California subalpine forest post-fire diversity and productivity, Dryad, Dataset,


High severity fire may promote or reduce plant understory diversity in forests. However, few empirical studies test long-standing theoretical predictions that productivity may help to explain observed variation in post-fire plant diversity. Support for the influence of productivity on disturbance-diversity relationships is found predominantly in experimental grasslands, while tests over large areas with natural disturbance and productivity gradients are few and have yielded inconsistent results. Here, we measured the response of post-fire understory plant diversity to natural gradients of fire severity and productivity in a large-scale observational study in California’s subalpine forests. We found that plant species richness increased with increasing fire severity and that this trend was stronger at high productivity. We used plant traits to investigate whether release from competition might contribute to increasing diversity and found that short-lived and far-dispersing species benefited more from high severity fire than their long-lived and near-dispersing counterparts. For far-dispersing species only, the benefit from high severity fire was stronger in high productivity plots where unburned species richness was lowest. Our results support theoretical connections between fire severity, productivity and plant communities that are key to predicting the consequences of increasing fire severity and frequency on diversity in the coming decades.


Fire and plot selection

Over the course of three summers (July – September, 2017 – 2019), we sampled 248 plots across 13 fires in subalpine forest (see Appendix S1: Table S1 for fire information). We identified fires that contained each of six fire severity classes (unburned, low, low-moderate, moderate, high-moderate, and high) corresponding to fire-caused basal area mortality (0%, 0-25%, 25-50%, 50-75%, 75-90%, and 90-100% respectively) using the remotely sensed relativized differenced normalized burn ratio (RdNBR; Miller and Thode 2007). Relativized fire severity measures such as RdNBR report relative change such that areas of both sparser and denser vegetation can experience the full fire severity spectrum. This helps to decouple the correlation between fire severity and productivity that can arise with absolute measures of biomass change or mortality (Pausas and Bradstock 2007). We sampled all 10 subalpine fires in the study region that contained six fire severity classes and were within two days hike of a trailhead as well as 3 smaller fires with fewer severity classes. Time since fire at the time of sampling ranged relatively evenly from 2-17 years.

We used a stratified random sampling design, placing 405 square meter circular plots at the crosshairs of a 200x200 meter grid in upland areas with no recorded history of previous fire or logging. We stratified plots across preliminary fire severity classes (from RdNBR) as well as aspect (see Appendix S1: Supplemental Methods & Figure S1) and estimated final plot-level fire severity based on fire-caused basal area mortality as described above as well as relative fuel consumption (see Appendix S1: Table S2 for fire severity class descriptions). We subsequently censused all vascular plant species in the plot by systematically searching the plot area. Species richness was low in our study area (median 10), making it easy to locate all species.

Remotely sensed environmental variables

We used NDVI, a remotely sensed measure of photosynthetic activity, to represent plot-level productivity. NDVI has strong theoretical and empirical links to Gross Primary Productivity (GPP; Glenn et al. 2008), and robust correlations have been found between NDVI and tree radial growth measurements in other energy-limited systems like boreal forest (Beck et al. 2011).  NDVI derived from 30x30m Landsat imagery is commonly used to represent productivity in montane forest (Burkle et al. 2015) and a recent study in Sierra Nevada subalpine forest found that it declined with mortality in whitebark pine stands and that inter-annual variability in NDVI tracked snow accumulation throughout a 30 year study period (Potter and Dolanc 2016). Using Google Earth Engine, we removed snow and cloud pixels from Landsat V imagery and calculated the average of maximum annual NDVI for the years 1995 through 2000 (Pettorelli et al. 2005, Gorelick et al. 2017). This six-year time period predates sampled fires, represents both wet and dry years in the study region, and its average correlates well with total aboveground tree biomass calculated for 37 unburned plots (Appendix S1: Fig. S1). We calculated Heat Load Index, a single metric accounting for slope, aspect, and some elements of slope position and shadow (essentially total plot-level solar radiation), using the R package spatialEco (McCune and Keon, 2002).

Plant trait data

We compiled available information on plant life history and dispersal syndrome for species in our dataset using the University of California Jepson Herbarium (, the USDA Forest Service Fire Effects Information System (, JSTOR Global Plants (, and our own familiarity with the species. We characterized species life history as either annual/biennial, short-lived perennial (non-woody perennial without rhizomes, stolons, bulbs or root storage organs) or long-lived perennial (woody perennials and/or perennials with rhizomes, stolons, bulbs, or other storage organs), and species dispersal syndrome as either far-dispersing (vertebrate or wind dispersal syndrome) or near-dispersing (unassisted dispersal syndrome) (Supplement S1: Table S3). Species dispersal syndrome was classified as vertebrate-dispersing if the reproductive structures were fleshy or adhesive, wind-dispersing if the reproductive structure included a pappus, and unassisted if the reproductive structure had none of the previously described characteristics.

Usage Notes

See README file.


National Science Foundation Graduate Research Award, Award: 1650042

USDA Forest Service Region Five