Divergent roles of herbivory in eutrophying forests

Abstract

Ungulate populations are increasing across Europe with important implications for forest plant communities. Concurrently, atmospheric nitrogen (N) deposition continues to eutrophicate forests, threatening many rare, often more nutrient-efficient, plant species. These pressures may critically interact to shape biodiversity as in grassland and tundra systems, yet any potential interactions in forests remain poorly understood. Here, we combined vegetation resurveys from 52 sites across 13 European countries to test how changes in ungulate herbivory and eutrophication drive long-term changes in forest understorey communities. Increases in herbivory were associated with elevated temporal species turnover, however, identities of winner and loser species depended on N levels. Under low levels of N-deposition, herbivory favored threatened and small-ranged species while reducing the proportion of non-native and nutrient-demanding species. Yet all these trends were reversed under high levels of N-deposition. Herbivores also reduced shrub cover, likely exacerbating N effects by increasing light levels in the understorey. Eutrophication levels may therefore determine whether herbivory acts as a catalyst for the "N time bomb" or as a conservation tool in temperate forests.

Fig. 1. Spatial distribution of resurvey sites, N-deposition in the year 2000, and changes in (Δ) herbivory pressure across Central Europe.

Our analysis spans 52 resurvey sites with inter-census time spans ranging from 10 to 64 years (median: 47.5 years). Color of points corresponds to the magnitude of change in site herbivore pressure between the baseline survey and resurvey (Supplementary Data file 1). Total nitrogen deposition (wet and dry, reduced and oxidized) is calculated using the EMEP database for the year 2000 and displayed across a color gradient of light to dark blue representing lowest to highest values at a spatial resolution of 10 km.

Fig. 2. Higher herbivory decreased shrub layer cover, but not herb and tree layer cover.

Relationships between changes in (Δ) herbivory pressure and a Δ shrub layer cover, b Δ herb layer cover, and c Δ tree layer cover. All models included inter-census time span, site area, and baseline herbivory as covariates. Note two sites lacked shrub and tree cover and one site also lacked herb cover data so that there were n = 50 and n = 51 independent resurvey sites for a, c, and b, respectively. Lines and ribbons represent the posterior mean line and the 95% credible interval. Dashed regression lines represent statistically unclear relationships. Frequency distributions (density, boxplot and points) of the respective response variables are displayed alongside. Boxplots bound the interquartile range (IQR) divided by the median and whiskers extend up to a maximum of 1.5 × IQR beyond the box. Triangles indicate the mean. Horizontal lines at zero indicate no change. Source data are provided as a Source Data file.

Fig. 3. Herbivory increased temporal species turnover but was not clearly associated with changes in species richness.

Relationships between change in (Δ) herbivory pressure and a Δ species richness (number of species) and b temporal species turnover (unitless) at a study site. All models included inter-census time span, site area, and baseline herbivory as covariates, with n = 52 independent resurvey sites. Lines and ribbons represent the posterior mean line and the 95% credible interval. Dashed lines represent statistically unclear relationships. Frequency distributions (density, boxplot and points) of the respective response variables are displayed alongside. Boxplots bound the interquartile range (IQR) divided by the median and whiskers extend up to a maximum of 1.5 × IQR beyond the box. Triangles indicate the mean. Horizontal lines at zero indicate no change. Source data are provided as a Source Data file.

Fig. 4. Herbivory effects depend on N-deposition levels.

Relationships between change in (Δ) herbivory pressure and a Δ community-weighted mean N-number (CWM-N), c percentage change in (Δ %) non-native species, e Δ % red-listed species, and g Δ % small-ranged species. Frequency distributions (density, boxplot and points) of the respective response variables are displayed alongside. Boxplots bound the interquartile range (IQR) divided by the median and whiskers extend up to a maximum of 1.5 × IQR beyond the box. Triangles indicate the mean. Horizontal lines at zero indicate no change. Herbivory effects depend on N-deposition (b, d, f, h). Conditional effects of herbivory are depicted at the 10th (348 kg/ha; left) and 90th (1010 kg/ha; right) percentile of cumulative N-deposition in the data. There are n = 52 independent resurvey sites for all models. Lines and ribbons represent the posterior mean line and the 95% credible interval. Dashed lines represent statistically unclear relationships. Rugs in figure bottom in b, d, f, h depict the marginal distribution of the predictor. Cumulative N-deposition is calculated between the baseline and resurvey year per site. See Supplementary Fig. 3 and Supplementary Table 26−29 for model outputs of the effects of N-deposition alone. See Supplementary Fig. 4 for interaction effects on species richness change and exchange ratio. Source data are provided as a Source Data file.