Mule deer impede Pando's recovery: Implications for aspen resilience from a single-genotype forest

Abstract

Aspen ecosystems (upland Populus-dominated forests) support diverse species assemblages in many parts of the northern hemisphere, yet are imperiled by common stressors. Extended drought, fire suppression, human development, and chronic herbivory serve to limit the sustainability of this keystone species. Here we assess conditions at a renowned quaking aspen (Populus tremuloides) grove-purportedly the largest living organism on earth-with ramifications for aspen biogeography globally. The "Pando" clone is 43 ha and estimated to contain 47,000 genetically identical aspen ramets. This iconic forest is threatened in particular by herbivory, and current management activities aim to reverse the potential for type conversion, likely to a non-forest state. We set out to gauge agents affecting recent deterioration through a network of monitoring plots and by examining a chronosequence of historic aerial photos to better understand the timing of putative departure from a sustainable course. Sixty-five permanent forest monitoring plots were located in three management regimes existing within Pando: no fencing, fencing with active and passive treatments, fencing with passive-only treatment. At each sample plot we measured live and dead mature trees, stem recruitment and regeneration, forest and shrub cover, browse level, and feces counts as a surrogate for ungulate presence. Ordination results indicate that aspen regeneration was the strongest indicator of overall forest conditions at Pando, and that mule deer (Odocoileus hemionus) presence strongly impacts successful regeneration. Additionally, fencing with active/passive treatments yielded the most robust regeneration levels; however, a fence penetrable by ungulates in the passive-only treatment most likely played a role in this outcome. The aerial photo sequence depicts various human intrusions over the past seven decades, but perhaps most telling, a decline in self-replacement beginning 30-40 years ago. Aspen communities in many locations in North American and Europe are impacted by unchecked herbivory. The Pando clone presents a unique opportunity for understanding browse mechanisms in a forest where tree genotype, closely aligned with growth and chemical defense, is uniform.

Fig 1. Map depicting the study area within the Pando aspen clone, Utah, USA.

Sixty-five sample plots were randomly distributed across the study area, with equal portions located within No Fence, 2013 Fence, and 2014 Fence management regimes. Base image courtesy of USDA Aerial Photography Field Office.

Fig 2. A seventy-two year aerial photo chronosequence showing forest cover change within the Pando aspen clone, Utah, USA.

Photos were georectified using ArcMap^® software to ensure accurate scale and location alignment. Yellow polygon depicts the boundary of the 43 ha clone as projected over each photo year. Base image courtesy of USDA Aerial Photography Field Office.

Fig 3. A joint plot depicts the results of nonmetric multidimensional scaling (NMS) ordination on a matrix of 65 plots by 16 monitoring variables.

Highly correlated environmental variables (r² ≥ 0.3) are overlaid on the ordination to show relationships to primary axes. Vectors indicate direction and strength (length) of these factors in the ordination space defined by plot values of all measured variables. Variables shown are: deer pellet groups ha^-1 (Deerh), aspen recruitment stems ha^-1 (Recruith), percent aspen cover (Pasp), live mature trees ha^-1 (TPHlive), trees ha^-1 (TPH), and aspen regeneration stems ha^-1 (Regenh). Triangles show locations of plot scores in “data space” within the NMS ordination.

Fig 4. Box plots depicting a significant difference (χ² = 37.10, p < 0.0001) in terms of the Kruskal–Wallis test for differences between treatments (groups) of regeneration ha^-1.

Output from Kruskal–Wallis test is shown in Wilcoxon mean scores on the y-axis (SAS^®). Whiskers show minimum and maximum values, boxes represent 25–75% data ranges, horizontal lines within boxes are medians (no line indicates Wilcoxon score of zero), and diamond symbols are means. Results are considered significant where a Monte Carlo-simulated chi-square test using 10,000 runs produced an estimated p-value of <0.05.

Fig 5. Wilcoxon–Mann–Whitney U test results displayed in box plots showing no significant difference (χ² = 1.37, p < 0.24) in regeneration ha^-1 between No Fence and 2014 Fence treatment groups.

Output from the test is shown in Wilcoxon mean scores on the y-axis (SAS^®). Whiskers show minimum and maximum values (no line indicates Wilcoxon score of zero), boxes represent 25–75% data ranges, horizontal lines within boxes are medians, and diamond symbols are means.