In this post Markus Eichhorn discusses his new article ‘Effects of deer on woodland structure revealed through terrestrial laser scanning

About the video: Three-dimensional reconstruction of a transect from Wyre Forest, an area of high deer density. The central 10 X 50 m plot is surrounded by a large number of points which were not used in the analyses. Survey apparatus is still visible. Points are coloured with visual spectrum returns, but these were not collected at the same time as positional information, and it was not always possible to assign a colour to each point. Hence there are many white patches; this is not snow! Video prepared by Joe Ryding.

In our new paper we reveal that forests with high or low deer densities differ in their structure not only in the understorey, where deer are able to browse, but throughout the whole canopy profile. We demonstrate this using terrestrial laser scanning, a novel means of documenting the three-dimensional structure of complex habitats with many potential applications. The work was part of the WoodMAD project*, funded by Defra, the raison d’etre of which was to understand the drivers of decline in British native woodland birds**. Our paper doesn’t directly address birds, but by demonstrating profound differences in habitat structure, we anticipate that there will be effects on not only on birds but small mammals, insects and many other species which depend on particular aspects of forest architecture. Tree recruitment and regeneration of forest stands are also likely to be affected.

Our view of forests has historically been constrained by our terrestrial perspective. Forest mensuration traditionally focusses on measurements that can be taken at ground level, with only crude metrics for the canopy above. While airborne remote sensing has opened new horizons in describing tree canopies, it is the fine-scale structure of forests beneath the canopy which determines their suitability as habitat for a wide range of species.

laser scanner
The laser scanner in action in the forest. Photo by Joe Ryding, who led the field surveys.

Terrestrial laser scanning allows high-resolution three-dimensional sampling of forest architecture. The technique has been increasingly used in forests over recent years, largely for determining wood volume in production forestry or estimating carbon stocks. Here we demonstrate that it can also be used to measure the distribution of foliage. The challenge is to use this new and powerful technique to extract new metrics with which to describe and assess complex habitats.

We set out to test whether there were consistent differences between forests with high and low deer densities, or which had either been managed (normally by stand thinning) or left unmanaged. To achieve this we selected 40 lowland woodlands of similar composition in two regions to allow for a fully factorial comparison. Terrestrial laser scanning was conducted in sample plots in each of these, along with confirmatory deer density surveys.

figure_2.eps
Distribution of foliage material with height in the 40 woodland plots. This figure reverses the axes relative to the version in the original paper to allow for a more intuitive reading. In the paper height is on the abcissa to better reflect the statistical analyses. Note that Ffridd Mathrafal plot 4 was managed.

Our main finding will not surprise most readers: areas with high deer densities (>10 per km2) contain much less foliage in the understorey (0.5–2 m above ground) than in areas with low deer densities (<1 per km2). Our findings are an advance on this because we can measure this empirically as a 68% reduction in foliage volume at these canopy heights.

More surprising was that the apparent effects of deer extend throughout the whole canopy profile; high-deer forests were 5 m taller on average. This cannot be attributed to direct browsing of deer, but might be related to their impacts on the recruitment of tree seedlings and saplings. In a comparative survey we cannot definitively state that these differences were caused by deer, nor distinguish cause from effect***. Further experimental evidence would be desirable, of course, but maintaining stable deer populations at the timescales over which forest structure develops is effectively impossible.

There are, however, a number of indicators which support the inference that deer are responsible for these gross changes in forest structure. The same pattern was replicated in plots in two regions, separated by hundreds of miles. We could not attribute the differences to forest composition or edaphic factors. They were only seen in the distribution of foliage, not in the stem material, which instead indicated regional differences. Finally, our study adds to a compelling pattern of evidence from around the world (especially in North America) of profound shifts in woodland structure when deer densities are at high levels.

figure_3.eps
Distribution of stem material with height in the 40 woodland plots.

In contrast, we detected relatively minor differences between managed and unmanaged sites. This could simply reflect the inconsistency of management interventions, which varied in timing and intensity. What we can say, however, is that their impacts on overall forest structure were minimal, and greatly outweighed in our data by associations with deer. The implication for woodland management is therefore that attempts to create complex understorey habitats are more likely to be successful by controlling deer than by doing so manually.

How deer populations might be reduced is beyond the scope of our work, but is considered in this blog post which explores the options. Our work adds to the evidence base that unchecked deer populations are responsible for structural alterations in our native woodlands whose effects on a wide range of flora and fauna are likely to be equally profound.

* WoodMAD stands for Woodland Management and Deer. We’re very proud of the acronym.

** Full report and recommendations.

*** Between the authors we came up with a number of hypotheses for what might be driving the canopy height differences, to which the editor added another, many of which were mutually incompatible, and none of which had anything other than guesswork to support them. I’m therefore withholding any speculation on the actual causes and suggest that it’s an area for further investigation.

Advertisements