Greater biomass from Arctic greening absorbs increased grazing pressure from a large herbivore

Arctic warming is causing widespread “greening” of tundra ecosystems. What this means for plant-herbivore relations, including the grazing pressure herbivores exert on increasingly productive tundra ecosystems, is poorly understood. Svalbard is one of the fastest warming places on Earth, with concomitant increases in both forage biomass and reindeer numbers. In 11 years between 1998 and 2023, we measured grass biomass and the proportion of shoots grazed in mesic grass-dominated tundra to evaluate whether increased forage biomass of grass absorbed the grazing pressure of more reindeer. Also, we used GPS data from adult female reindeer (2009-2023) to identify if grazing pressure was relieved by spillover into other habitats. During the study period, reindeer abundance, estimated by annual capture-mark-recapture, tripled, while grass biomass only doubled. Grazing pressure increased from 4% to 8%, which was lower than expected from the increased reindeer density. This discrepancy was not caused by spillover into other habitats, but rather by increased grazing in higher-biomass patches that have emerged with summer warming. Our findings support the notion that increased summer forage has contributed to Svalbard reindeer population growth, notably by making available higher biomass grass swards that allow for greater food offtake.

Methods
Grazing data

In September 1997, we established 13 plots (5 m ×5 m) in grass sward habitats in Semmeldalen, Savalbard to monitor grazing pressure on grasses. Plots were randomly selected among separate patches of varying shoot density of grasses in grass sward habitat and were > 20 m apart from each other. In 2021 an additional 15 plots were established in grass sward habitat near the original plots. In late July and August (range: July 29 – August 24), we measured density of shoots and number of shoots grazed (only measuring true grasses and, thus, excluding rushes and sedges). Within each 5 m ×5 m plot, we sampled shoot density by randomly selecting 15 subplots of 10 cm ×10 cm where we counted the number of grass shoots and the number of shoots grazed. Subplots were selected by tossing a 10 cm × 10 cm frame and counting the number of grazed and ungrazed grass shoots where it landed; thus, locations of subplots varied during each sampling period. To estimate the available grass biomass of a plot, we determined average shoot mass by collecting 25 randomly selected, ungrazed, grass shoots from within each 5 m × 5 m plot at each time of measurement. We dried shoot samples at 60 °C and placed them in a desiccator for a minimum of 48 hours before weighing them to the nearest mg to get the average mass per shoot. Mean shoot mass was then multiplied by shoot density and then multiplied by 100 (since measurements were taken at the 10 cm × 10 cm subplot) to estimate grass biomass within each plot (g/m2; biomass = shoot count × mean shoot mass × 100). Plots were monitored annually in 1998-2005 (except in 2003), 2010, and 2021-2023, for a total of 170 plot-level observations of grass biomass and grazing. 

Reindeer population and GPS-collar data

Since 1995, a sample of calves and adult female reindeer were captured and marked with a unique collar band and ear tags each spring (March-April; for details see Albon et al. 2017). All capture and handling was performed under licenses and permits issued by the Norwegian National Research Authority (license nr. 22/5068) and the Governor of Svalbard (license nr. 16/01632-25).  In late July and early August, an annual census of reindeer abundance was conducted. by observers walking through the valleys to record marked and unmarked animals.  Population estimates were derived from  an integrated population model using these capture–mark–recapture data in winter and observations from the annual summer population census.

Since 2009, between 12 and 48 captured adult females were fitted with GPS collars (Vectronic Aerospace GmbH) during capture events (see for details Loe et al. 2016), resulting in GPS data from 135 different individuals. GPS collars remained on individuals for 1-7 consecutive years (n = 419 animal years), recording locations at rates ranging between 1 and 10 h. In total, 319,774 GPS locations were recorded from 2009 to 2023.