Predicting where and when possum browse will kill trees
Fuchsia trees killed by possum browse at Haast, South Westland. Image – Caroline Thomson.
Possums have been implicated as major drivers of the loss of biodiversity in forests in New Zealand. Although possums favour non-foliar foods such as energy-rich flowers and fruits (and such browsing may hinder long-term forest regeneration), foliage browsing is a significant cause of dieback and mortality of native tree species.
Untangling quantitative, causative relationships between possum density, observable damage and tree mortality is difficult, not least because such data are highly variable, both within and among forest sites. For example, while both Landcare Research and the Department of Conservation use the Foliar Browse Index (FBI) method to assess forest tree condition, it provides only snapshots of forest health. Pen Holland and colleagues have been using FBI data to formulate and parameterise a mathematical model of foliage growth, turnover and consumption, (Fig.) to answer two questions:
- How much of a tree’s foliage do possums have to eat in order to kill it?
- To what level does foliage consumption by possums have to be reduced in order to protect trees at a particular location from browse-induced mortality?
The model framework is based on interacting processes at different spatial scales, from individual leaves to tree canopies, through to large areas of mixed forest.
Browse damage occurs when part of a leaf is eaten by a possum, and the FBI records the ratio of whole to partially browsed leaves in the tree canopy. For this reason, the model incorporates the within-canopy foraging strategy of possums, such as browsing entire or partial leaves indiscriminately, or only browsing entire leaves, leaf tips or petioles, or new growth.
The timescale is also important. While trees may be able to regenerate after a one-off, severe, defoliation event (e.g. storm damage), continual and preferential browsing by possums can have a detrimental effect on foliage growth rates and lead to tree death. The model has been used to quantify how foliage growth rates of kāmahi change with browsing damage, and to estimate mean leaf life span from FBI data. This is a crucial part of the framework, since leaf turnover removes evidence of historical browse.
At the tree scale, light browsing does not generally make any difference to canopy health, but heavy browsing clearly does so. When browsing exceeds some threshold, the model shows that the tree is unable to regrow foliage fast enough to replace the leaves lost to both browse and leaf-fall, and total defoliation and death are inevitable. This proves that browsing alone can kill kāmahi, with an average sized tree dying if more than about 6,000 leaves or about 10% of its foliage (<1 kg dry weight, or <2% of a typical possum’s annual foliage consumption) are eaten annually. Once a tree has passed the browsing threshold, possum control may not be enough to reverse canopy decline.
Possums choose to eat foliage from an individual tree depending on the palatability, nutritional quality and toxin levels of its foliage, and the variety of other foods available. Hence, to predict how possum damage drives tree mortality at a site means large-scale foraging strategies must be incorporated into the model. FBI data alone cannot tell managers why spatial browsing patterns are so heterogeneous, but such data can be used to estimate foraging behaviours. For example, the proportion of kāmahi at a site showing no sign of browse can be used to estimate the degree of preference for individual kāmahi trees of similar size. If more large trees than small show heavy browse damage (as is normally the case), there must be an underlying preference for feeding in the canopies of large trees, over and above the proportion of food such trees provide within an area.
To predict tree mortality at a new location, minimal new data are required, including diameter at breast height, foliage cover, and browsing damage indices from a sample of trees, and a single estimate of possum density. Pen and her colleagues used such data from two North Island locations to test the model’s predictions, and found that their model reproduces annual kāmahi mortality patterns well. Future development of the model will broaden the scope to more native tree species, and incorporate drivers of possum population and foraging dynamics.
This work was funded by Landcare Research’s Capability Funding.
