Response of pollinator taxa to fire is consistent with historic fire regimes in the Sierra Nevada and mediated through floral richness†

Abstract

Many fire-prone forests are experiencing wildfires that burn outside the historical range of variation in extent and severity. These fires impact pollinators and the ecosystem services they provide, but how the effects of fire are mediated by burn severity in different habitats is not well understood. We used generalized linear mixed models in a Bayesian framework to model the abundance of pollinators as a function of burn severity, habitat, and floral resources in post-fire, mid-elevation, conifer forest, and meadow in the Sierra Nevada, California. Although most species-level effects were not significant, we found highly consistent negative impacts of burn severity in meadows where pollinators were most abundant, with only hummingbirds and some butterfly families responding positively to burn severity in meadows. Moderate-severity fire tended to increase the abundance of most pollinator taxa in upland forest habitat, indicating that even in large fires that burn primarily at high- and moderate-severity patches may be associated with improved habitat conditions for pollinator species in upland forest. Nearly all pollinator taxa responded positively to floral richness but not necessarily to floral abundance. Given that much of the Sierra Nevada is predicted to burn at high severity, limiting high-severity effects in meadow and upland habitats may help conserve pollinator communities whereas low- to moderate-severity fire may be needed in both systems.

Keywords: bees; burn severity; butterfly; fire; hummingbird; pollinator.

FIGURE 1

Map of the study area showing the region of the Sierra Nevada, California, where the King Fire burned in 2014; the inset shows the location of the King fire in California. (a) Green points are meadow, and black points are upland forest sites located in unburned, low‐moderate, and high‐severity burn classes. (b) Each site (square) consists of multiple 20‐m‐radius pollinator plots (circles) that were at least 100 m apart. (c) Pollinators were surveyed within each circular plot (circle), and plants in bloom were surveyed in eight randomly located 1‐m² quadrats (squares) within each plot. Base layer sources: Esri, USGS, NOAA.

FIGURE 2

Pollinator abundance by year, burn‐habitat class, and taxon, during surveys in the Sierra Nevada, California, 2 and 3 years after the 2014 King Fire. Moderate‐severity habitat was only sampled in 2017. Butterflies were only sampled in 2016, and other insects (including bees, wasps, flies, true bugs, and beetles) were only sampled in 2017. Note the difference in scales on the x‐axis; different taxa were sampled with different methodologies, precluding among‐taxon comparison of abundance (see main text for details).

FIGURE 3

Factors influencing the abundance of pollinator taxa in meadow and upland habitat in the Sierra Nevada, California, 2 and 3 years after the 2014 King Fire, estimated using negative binomial generalized linear mixed models (single‐species models for butterflies and hummingbirds, multi‐species models for bumblebees and other insects). Covariate coefficients for habitat type and cover and richness of blooming plant species visited at least 10 times by pollinators for (a) the bumblebee community on average and individual species, (b) the community of other flower‐visiting insects on average and individual species, (c) butterfly families, and (d) hummingbirds. Negative coefficients represent a decline in abundance; positive coefficients indicate an increase in abundance with increasing covariate value. Coefficients with significant effects (i.e., 95% Bayesian credible intervals do not overlap 0) are indicated by the darker shaded points. Species codes are in Table A5.

FIGURE 4

Factors influencing abundance of pollinator taxa in meadow and upland in the Sierra Nevada, California, 2 and 3 years after the 2014 King Fire, estimated with negative binomial generalized linear mixed models (single‐species models for butterflies and hummingbirds, multi‐species models for bumblebees and other insects). Covariate coefficients for the interaction of burn severity and habitat, and the quadratic term for burn severity in upland habitat only for (a) the bumblebee community on average and individual species, (b) the community of other insects on average and individual species, (c) butterfly families (not sampled in moderate‐severity habitat, so no quadratic burn effect), and (d) hummingbirds. Negative coefficients represent a decline in abundance; positive coefficients indicate an increase in abundance with increasing burn severity. Coefficients with significant effects (i.e., 95% Bayesian credible intervals do not overlap 0) are indicated by the darker shaded points. Species codes are in Table A5.

FIGURE A1

Photos taken at upland sites 3 years after the 2014 King Fire that were characterized as (a) unburned, (b) moderate‐severity, and (c) high‐severity burn classes.

FIGURE A2

Correlation between meadow and upland habitats and a. floral abundance and b. floral richness in the Sierra Nevada, California, 2 and 3 years after the King Fire.