Ecological Archives E093-137-A5

Anton J. Flügge, Sofia C. Olhede, and David J. Murrell. 2012. The memory of spatial patterns: changes in local abundance and aggregation in a tropical forest. Ecology 93:1540–1549. http://dx.doi.org/10.1890/11-1004.1

Appendix E. Analysis of the effect of habitat dependency on the results.

Introduction

Here we investigate the potential role of habitat selection in explaining the variation in aggregation caused by abundance and recent changes in abundance, by excluding species that have been shown to have habitat preferences. Harms et al. (2001) tested 171 BCI species (those that had more than 65 individuals in the 1990 census) on whether they showed any habitat dependencies. Of these, 61 did not show any dependency with one of the 5 habitat types in the study (swamp, low plateau, high plateau, slope, streamside), and 49 of these species are included in our study, 35 of which are top- or mid-canopy tree species.

Effect of local abundance on aggregation

There is no significant relationship between local abundance and aggregation in the 49 species identified by Harms et al. (2001) as not having any habitat dependency (Fig. E1). Fitting a weighted linear regression model to the log aggregation with the log abundance as main effect explains 5.6% of the variation in the 49 species data (R2), and 7.2% for only the 35 canopy species. This result is not significant, with more than 17% of the 10000 bootstrap samples showing a stronger correlation in the complete set (P = 0.17) and P = 0.15 for the set restricted to only the canopy species.

Effect of changes in local abundance in the BCI data

We fitted the weighted linear regression model for all species using both the log abundance and the log relative abundance changes as main effects to predict the log aggregation indices. This shows that the current local abundance and the changes in local abundance can explain 12.1% of the variation in the aggregation indices (R2) for the abundance changes between 2000 and 2005, and 18.9% of the variation when considering the period between 1985 and 2005. Despite this, the results are not significantly better than the results obtained for the model using only the current abundance as main effect (P = 0.36 for the period 2000–2005; and P = 0.18 for the period 1985–2005).

If we restrict our analyses to the 35 top- and mid-canopy tree species, we find that the model can fit the data much better. For only the canopy species, the model including local abundance changes can explain 32.1% of the variation in log aggregation (P = 0.061) for the 2000 to 2005 time span, and 32.4% (P = 0.055) for the 1985 to 2005 time span (see also Fig. E2).

Discussion

The variation explained for the non-habitat dependent canopy species is slightly less than for all canopy species when looking at the 5-year period, but slightly larger for the 20-year period. Because of the smaller sample set we have less statistical power, but in principle the results for the non-habitat dependent species seem to be in line with the results for all species. That we fail to find the relationship between aggregation and abundance is also due to the fact that only very few of the most common species are non-habitat dependent, our sub-sample in this analysis is therefore biased towards rare species and represents a smaller range of abundances.

Fig. E1. The relationship between local abundance and population aggregation for the 49 species from BCI that meet our selection criteria that show no habitat dependency (top-canopy trees Δ; middle canopy trees O; understory trees ; shrubs *). The thick dashed gray line shows the best linear fit of the data of all species, the other lines show the best linear fit for the species of the four different growth types (solid for shrubs, dashed-dotted for understory trees, dashed for mid-canopy trees, and dotted for top-canopy trees). Using bootstrapping methods, we find the slopes are not statistically different from 0 (see main text for a discussion on our methods).



Fig. E2. The relationship between recent change in local abundance and aggregation for the 35 top- and mid- canopy species that meet our selection criteria and have not been shown to have any habitat dependency by Harms et al. (2001). Plotted is the logarithm of the relative changes in local abundance (between (a) 2000 and 2005 and (b) 1985 and 2005) against the residual log aggregation values after subtraction of the constant effect and the effect of current abundance. These were obtained from a linear regression model with the logarithm of the local abundance and the logarithm of the relative changes in local abundance as main effects. The black line shows the remaining effect of the logarithm of the relative changes in local abundance. Top-canopy trees are depicted by Δ and mid-canopy trees by O. See main text for more details about our methods.


Literature Cited

Harms, K. E., R. Condit, S. P. Hubbell, and R. B. Foster. 2001. Habitat Associations of Trees and Shrubs in a 50-Ha Neotropical Forest Plot. Journal of Ecology 89:947–959.


[Back to E093-137]