Determinants of Betula spp. invasion of lowland heath
1. The invasion of Betula spp. (B.pendula and B.pubescens) triggers a phase transition
between lowland heath and scrub vegetation states. Transition to scrub is currently the
most serious threat to the conservation of UK lowland heath; it has been found to
correlate with soil phosphorus sorption capacity (PSC) at the landscape scale.
2. It was hypothesised that Betula invasion of Lowland heath was limited by numerous
factors that may be subdivided into safe-site and seed limitation and that phosphorus
availability was a key axis in the determination of a Betula safe-site.
3. An implicit assumption of earlier research was that PSC affected the P available to
invading plants. An approach combining observation, statistical modelling and
experimentation found that PSC affected the retention and availability of phosphorus in
heathland soils. Synthesis of these findings with the published literature suggests that the
direct effect of PSC on P-availability is small compared to indirect effects on organic
matter (SOM) accumulation and vegetation cycling.
4. These findings allowed the hypothesis that P-availability affects the likelihood of
invasion to be tested within an experimental framework. P-availability, seed rain and
disturbance were experimentally manipulated in a multifactorial field experiment on a
wet heath ecosystem in a stable, uninvaded heath area. Betula seedling densities and
numerous covariates were also measured. It was assumed throughout this research that
Betula seedling densities were indicative of the likelihood of heath-scrub transition.
5. Analysis of deviance found that all three treatments had significant effects on Betula
seedling densities with seed availability proving to be the single greatest limitation at the
site. Conversion of the experimental treatment factors into a continuous form allowed for
a more detailed description of the phase transitional area within the site. The most
significant descriptors of seedling densities were seed-rain, various plant neighbour
variables and P-availability.
6. Replication of this experiment with fewer treatment levels and replicates at two additional
sites found that the identity of the factors controlling Betula seedling densities was
broadly similar, (e.g. vegetation and seed availability factors played a role at all sites) but
that their relative contributions to within-site heterogeneity varied widely.
7. A single statistical model was fitted to data collected from three experimental sites. The
model, which explained 59.8% of the deviance in seedling densities, describes Betula
colonisation as a function of biomass density, necromass density, vegetation height, seed
bank density, phosphorus availability, and to a lesser extent, soil water content. The form
of the fitted relationships was complex with numerous interaction and polynomial terms.
If the model is applicable to a wider range of conditions then it may be concluded that it is heathlands close to seed sources and in the degenerate state, and possibly those
subjected to severe bums, that are the most likely to shift to the Betula scrub state. At
larger scales these conditions are probably most common in low management intensity,
high phosphorus sorption capacity (PSC) regions.
8. Validification of the combined-site model was attempted using data that was equivalent to
that used in the fitting of the model and which was collected over two 5ha grids of 130
sampling points and covering a wider range of heathland environments. Low predicted
and actual seedling densities prevented formal testing of the models accuracy but
predictions were qualitatively accurate, despite extrapolation.
9. Exploration of spatial heterogeneity in the determinant factors using the aforementioned
spatial grid data, was achieved with geostatistics. This revealed that seed bank densities
and edaphic factors displayed small-scale patchiness of around 50m while vegetation
factors possessed longer ranging autocorrelation resulting in single across-site gradients.
The proportion of spatially structured variance in the variables, within the studied range
(17.6-150m), was low for many factors. The pattern of these factors is discussed with
reference to the exclusion of Betula colonists and management recommendations are
10. It is concluded that the combined-site model provides a coarse but fairly accurate
definition of the phase-transitional area between lowland heath and Betula scrub
ecosystem states and that the determinant factors are at larger scales governed by
landscape-scale processes (e.g. soil (PSC), climate and management regimes). These
factors may explain regional differences in the scale and extent of scrub transition.