Mammalian Browsers Disrupt Eco-Evolutionary Dynamics in a Forest Tree Restoration Planting

Abstract

Native and restored forests are increasingly impacted by pests and diseases, including large herbivores. While community- and species-level impacts of these tree enemies are often well-documented, there is little understanding of their influence on finer-scale eco-evolutionary processes. We here study the influence of large-mammal herbivory on the survival and height growth of trees in a mixed species restoration planting of the Australian forest trees, Eucalyptus ovata and E. pauciflora, in Tasmania, Australia. Common-garden field trials mixing the two species were compared in adjacent unbrowsed (fenced) and browsed (unfenced) plantings. The browsed planting was exposed to mammal browsing by native marsupials, as well as feral introduced European fallow deer (Dama dama). Each tree species was represented by open-pollinated families from 22 paired geographic areas, allowing the assessment of the effects of browsing on the species and population differences, as well as on family variation within each species. In the browsed planting, a marked reduction in species and population differences, as well as in family variance, was observed for both height growth and survival. The pattern of height growth and survival of the populations of both species also differed between browsing regimes, with significant changes of climate relationships involving both focal tree attributes detected. Our results argue for a major disruption of the eco-evolutionary dynamics of restored forests in the presence of browsing by large mammalian herbivores, at the observed period of the tree life cycle. Importantly for forest restoration and conservation in the face of global change, our results challenge the choice of tree populations for translocation based solely on predicted or observed relationships of their home-site climate with current and predicted future climates of the restoration sites, while emphasising the need for genetic diversity to provide future resilience of restored forests to both biotic and abiotic stresses.

Keywords: assisted migration; climate adaptation and modelling; differentiation of tree species and provenances; family variance; herbivore‐plant interactions; tree survival and height growth.

FIGURE 1

Maps of the Tasmanian geographic range of (a) Eucalyptus ovata and (b) E. pauciflora . The natural range, distribution of the sampled 22 paired geographic areas (black points; see Table S2) and the common‐garden trial site at Connorville (red box) are shown. The region mapped is indicated in the insert of Australia. (Images: R. Wiltshire and P.A. Harrison).

FIGURE 2

Long‐term climate patterns for the Connorville site where mixed species field trials of Eucalyptus ovata and E. pauciflora were established under two browsing regimes. The graphs show the 5‐year moving average mean annual temperature (TANN, °C) and annual precipitation (RANN, mm). The red line represents the overall average for the site (1911–2021), the blue line represents the historical climate average prior to the detectable signature of climate change in the southern hemisphere (Abram et al. ; pre‐industrial warming, 1911–1959), the green line corresponds to the climate average often used to represent the contemporary climate (1976–2005) and the black line corresponds to the climate average for the growing period from planting (2014–2021).

FIGURE 3

Least‐squares means for the expected probability of 8‐year‐old survival and the overall 3‐year‐old height growth estimated for mixed species field trials of Eucalyptus ovata and E. pauciflora , established in two browsing regimes—unbrowsed and browsed plantings. The 95% confidence intervals are depicted for the estimated least‐squares means, and the significance probabilities obtained from statistical F‐tests of the difference between species are presented for each browsing regime (see also Table S4).

FIGURE 4

GAMs modelling the relationship of 3‐year height growth or 8‐year tree survival with climate variables for Eucalyptus ovata and E. pauciflora , established in two browsing regimes—unbrowsed and browsed plantings. In the figures, the y‐axes represent values for the height or survival response variables standardised to a unit variance, and the x‐axes pertain to home‐site climate variables for populations of each species, namely: (a) maximum temperature of the warmest week (TMXWP, °C); and (b) isothermality (TISO, % diurnal temperature range/annual temperature span). The fitted curves (red line) and corresponding 95% confidence intervals (dashed back line) were estimated on either the height response scale, or the survival probability scale (via the inverse probit link function), prior to standardisation. The symbols illustrate the standardised least‐squares means for each geographic area originally estimated from the fitted mixed‐effects models (LMMs and GLMMs for height and survival, respectively). The (local) population most proximal to the Connorville common‐garden site is shown as an open symbol, as opposed to the remaining (non‐local) populations which are illustrated with closed symbols. For each figure, the blue arrow at the bottom indicates the average of the climate variable for the 3‐year or 8‐year growing period at the common‐garden site for height and survival, respectively, and the probability (p‐value) at the top refers to a hypothesis test undertaken to evaluate whether a fitted model was statistically significant (note that, for survival, statistical inference refers to the probit scale; see Table 2). The results from contrasts comparing unbrowsed and browsed plantings of the same species, as well as from contrasts comparing species within the same browsing regime, are given in Table 2.