Differential responses of heather and red grouse to long-term spatio-temporal variation in sheep grazing

Retrospective comparisons of competing demographic models give clarity from “messy” management on a Scottish grouse moor

“…Conversely, the relative evidence for the two models containing a buzzard effect increased (with biannual assessments, for instance, by the post‐breeding assessment in 2018 the combined weight of the two buzzard models was 0.88 versus 0.11 for the model with harrier effects only), which strongly favors the killing hypothesis over the scavenging one. The inclusion of habitat as a factor limiting pre‐breeding grouse density in three out of the seven models in each model set tested the hypothesis that habitat, either on its own (baseline + habitat model) or in combination with other factors (non‐protected predator + habitat model, buzzard + habitat model), was an important determinant of grouse numbers in the counted areas during the period of the study. Heather cover at Langholm fell by a third between 1997 and 2009 before starting to recover (Ludwig, Aebischer, Richardson, et al, 2020; Figure 3a), and superficially the counts of red grouse during the long‐term study (Figure 1) appeared to follow the same pattern. However, the two simple habitat models (baseline + habitat, NPP + habitat) were rapidly downweighted in the model comparison process across all three model sets (Figure 6).…”

“…The non‐breeding‐season models predicted pre‐breeding counts ($$$ {D}_{\mathrm{pre}}^t $$$) as a function of the previous post‐breeding count ($$$ {D}_{\mathrm{post}}^{t-1} $$$), the bag density from fall shooting ( B t −1 ), and summer and winter survival ($$$ {S}_s^{\mathrm{ad}},{S}_w $$$; Appendix S1: Table S1). We structured models as basic “BIDE” (births, immigration, deaths, and emigration) models of population growth, although we ignored immigration and emigration because our study site was isolated from other grouse moors and regarded as effectively closed from a population dynamics perspective (Ludwig, Aebischer, Bubb, Roos, & Baines, 2018; Ludwig, Aebischer, Richardson, et al, 2020). In all years, demographic parameters derived from Langholm Moor were influenced by multiple factors whose effects we sought to evaluate (Figure 1).…”

“…Red grouse were counted in a 42-km 2 F I G U R E 1 Temporal distribution of management strategies (dashed lines for heather burning and sheep grazing represent reduced levels of those disturbances; keepering occurred during gray-shaded periods and not during unshaded periods), raptor population levels (dashed line for buzzards represents moderate levels), weather events, and mean pre-and post-breeding densities of red grouse at Langholm Moor in southwest Scotland during 1992-2019. Density data from Ludwig, Aebischer, Richardson, et al (2020).…”

“…Gamekeepers used medicated grit (Newborn & Foster, 2002) during 2008-2014 to control the intestinal nematode T. tenuis, which is known to cause periodic population declines in grouse (Hudson et al, 1998). Initial reductions in sheep numbers from 1990 did not prevent heather decline, but further reductions over a larger area from 2011 led to heather recovery, relative to 1990 levels, by 2015 (Ludwig, Aebischer, Richardson, et al, 2020). Weather and raptor population dynamics added further complexity to this managed system (Figure 1), including two periods in which half (1998)(1999) or all (2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015) of the hen harrier broods were managed with diversionary feeding to reduce the impact of harriers on grouse productivity (Ludwig, McCluskie, Keane, Barlow, et al, 2018;Redpath et al, 2001).…”

“…We therefore extended the baseline model by including habitat effects only in the non-breeding-season component and followed methods of Redpath and Thirgood (1997) to model adult winter survival as a function of carrying capacity. We developed an index of carrying capacity using the proportion of quadrats with heather present from vegetation surveys conducted on the areas used to count grouse in 1997, 2012, 2015(Ludwig, Aebischer, Richardson, et al, 2020, combined with an estimate of change between 1988 and 1994 from Redpath and Thirgood (1997) 3a). We calculated carrying capacity as the proportion, H, of heather extent relative to the estimated extent in 1992, the first year of our study, and in the predictive model (Equation 6) set S w ¼ HS w,dd .…”