Long-term changes in the primary productivity of migratory caribou (Rangifer tarandus) calving grounds and summer pasture on the Quebec-Labrador Peninsula (Northeastern Canada): the mixed influences of climate change and caribou herbivory

Abstract

At high herbivore density, herbivory can reduce forage abundance, potentially contributing to habitat degradation and driving changes in herbivore population size or range use, in accordance with the exploitation ecosystem hypothesis. The migratory Rivière-George caribou herd (RGH, Rangifer tarandus) of the Quebec-Labrador Peninsula (Canada) has experienced a large decline in population size since the population peaked in the early 1990s, with similarly large changes in seasonal range use. Demographic changes are suspected to have influenced forage abundance and caribou range use through density-dependent interactions between caribou and their habitat. We used the Normalized Difference Vegetation Index (NDVI) to examine relationships between RGH caribou density and range productivity from 1991 to 2011. A modelling approach was used to control for the response of climate and to isolate the influence of caribou herbivory on primary productivity. Significant negative relationships were identified between caribou density and primary productivity, after controlling for climatic variation, for the global RGH calving grounds (r² = 0.54–0.55) and summer range (r² = 0.42–0.51), but not for the “core” ranges, where caribou density was highest. Positive temporal trends in primary productivity appeared to reflect the decline in RGH population size, suggesting vegetation recovery following reductions in caribou abundance. Climate warming (of up to + 1.5 °C per decade) was most responsible for the strong positive trends in primary productivity observed over the 1991–2011 period, but decreases in RGH herbivory likely also contributed to the increases in range productivity. Forage access likely improved over the study period, which may have influenced RGH range use and habitat selection.

Appendix

See Tables 1, 2, 3, 4, Figs. 1, 2, 3, 4, 5 and 6.

Table 1 Candidate models for cumulative growing season NDVI (cNDVI) in the Rivière-George caribou (Rangifer tarandus) herd global 1991–2011 calving grounds, defined by the 100% minimum convex polygon of adult female caribou locations during the calving period

Table 2 Candidate models for cumulative growing season NDVI (cNDVI) in the Rivière-George caribou (Rangifer tarandus) herd global 1991–2011 summer range, defined by the 100% minimum convex polygon of caribou locations during the summer period

Table 3 Trends in primary productivity (cumulative growing season NDVI or cNDVI), cNDVI-climate model residuals, and climatic variables for the 1991–2011 period in the Rivière-George caribou (Rangifer tarandus) herd global calving grounds, global summer range, and 1991–2000 and 2001–2011 core calving areas

Table 4 Linear regression results for the relationships between caribou density (caribou km⁻²) and mean cumulative growing season NDVI (cNDVI)-climate model residuals for the different seasonal ranges and core areas of the Rivière-George caribou (Rangifer tarandus) herd

Fig. 1

The global calving grounds and summer range of the Rivière-George caribou (Rangifer tarandus) herd over the 1991–2011 period, as defined by minimum convex polygons (MCPs), clipped to the coastline. The calving grounds MCP is based on the locations of collared females, whereas the summer range MCP is based on the locations of both collared males and females

Fig. 2

A loess smoothing spline fitted to aerial survey estimates of Rivière-George caribou (Rangifer tarandus) herd population size (black data points) to produce annual population size estimates. Error bars represent confidence intervals (α = 0.10) associated with the aerial survey data. The grey shaded area represents the 1991–2011 study period

Fig. 3

The distribution of the Rivière-George caribou (Rangifer tarandus) herd in Northern Quebec and Labrador during the calving periods for a 1991–2000 and c 2001–2011, and during the summer periods for b 1991–2000 and d 2001–2011. The gradient from light grey to black represents the number of overlapping individual Brownian Bridge (BB) 95% home ranges, with darker shades representing areas with a greater degree of overlap; each BB represents the seasonal home range of one collared individual. Data from all collared individuals were used to delimit the summer range, but only data from adult females were used to delimit the calving grounds. The total number of BBs varied as follows: a 166, b 147, c 140, d 194. Core areas, shown in white crosshatch, represent areas with ≥ 70% of the maximum number of BB overlaps: a ≥ 13 BB overlaps, b ≥ 43, c ≥ 35, d ≥ 99. Note that areas with < 3 BB overlaps are excluded from the maps

Fig. 4

Interannual changes in cumulative growing season NDVI (cNDVI)-climate model residuals, cNDVI, mean summer temperature, and growing season length in a the global 1991–2011 calving grounds, b the global 1991–2011 summer range, c the 1991–2000 core calving area, and d the 2001–2011 core calving area of the Rivière-George caribou (Rangifer tarandus) herd. Values represent means calculated from all non-water pixels within each area

Fig. 5

The relationships between mean cumulative growing season NDVI (cNDVI)-climate model residuals and estimates of Rivière-George herd caribou (Rangifer tarandus) density, lagged by 0–2 years with respect to the model residual data, for a the global 1991–2011 calving grounds, b the global 1991–2011 summer range, c the 1991–2000 core calving area, and d the 2001–2011 core calving area. Sample size is consistently n = 21 for a and b, with one data point for each year of the 1991–2011 period, and n = 10 for c, with one data point for each year of caribou presence in the core area. Sample size varies depending on the time lag used for d, from n = 12 with no time lag to n = 10 for a 2 year lag. The density estimates for the global 1991–2011 calving grounds refer only to adult females, but the other density estimates refer to all individuals. r² values are provided for all significant relationships (α = 0.05); these relationships were also tested for 3–6 year lags (Table 4 in Appendix)

Fig. 6

Interannual changes in cumulative growing season NDVI (cNDVI)-climate model residuals and estimates of caribou (Rangifer tarandus) density or Rivière-George herd population size for a the global 1991–2011 calving grounds and summer range, and b the 1991–2000 and 2001–2011 core calving areas. Caribou density estimates refer only to adult females. The time series for the two core calving areas can be separated into two periods based on the density estimates: caribou present and caribou absent. Model residual and density values represent means calculated from all non-water pixels within each area