The large-crowned emergent tree Microberlinia bisulcata dominates rainforest groves at Korup, along with two co-dominants Tetraberlinia bifoliolata and T. korupensis. M. bisulcata has a pronounced modal size frequency distribution around ~110 cm stem diameter: its recruitment potential is very poor. It is a long-lived light-demanding species, one of many found in African forests. Tetraberlinia species lack modality, are more shade-tolerant and recruit better. All three species are ectomycorrhizal. M. bisulcata dominates grove basal area, even though it has similar numbers of trees (≥ 50 cm stem diameter) as each of the other two species. This situation presented a conundrum which prompted a long-term study of grove dynamics. Enumerations of two plots (82.5 and 56.25 ha) between 1990 and 2010 showed mortality and recruitment of M. bisulcata to be very low (both rates ~0.2 %/yr) compared with Tetraberlinia (2.4, 0.8 %/yr), and M. bisulcata grows twice as fast as the Tetraberlinia. Ordinations indicated that these three species determined community structure by their strong negative associations whilst other species showed almost none. Ranked species abundance curves fitted the Zipf-Mandelbrot model well, and allowed ‘over-dominance’ of M. bisulcata to be estimated. Spatial analysis indicated strong repulsion by clusters of large (50 – < 100 cm) and very large (> 100 cm) M. bisulcata of their own medium-sized (10 – < 50 cm) trees and all sizes of Tetraberlinia. This was interpreted as competition by M. bisulcata increasing its dominance, but also inhibition of its own replacement potential. Stem coring showed a modal age of ~200 year for M. bisulcata, but with large size variation (50 – 150 cm). Fifty-year model projections suggested little change in medium, decreases in large, and increases in very large trees of M. bisulcata, accompanied by overall decreases in medium and large trees of Tetraberlinia species. Realistically increasing very-large-tree mortality led to grove collapse without short-term replacement. M. bisulcata most likely depends on climatic events to rebuild its stands: the ratio of disturbance interval to median species’ longevity is important. A new theory of transient dominance explains how M. bisulcata may be cycling in abundance over time and displaying non-equilibrium dynamics.
Enumerations of permanent forest plots
There are six files and a species-codes list. The format is ANSI text (space-delimited). For the P-plot the large trees (stem diameter ≥ 50 cm) alive in the first (1991) and second (2005) enumerations are given in p91lar.txt and p05lar.txt. The subplot columns (‘Col’) are numbered 1 – 33 (from east to west), and the rows (‘Row’) lettered ‘A’ – ‘J’ (from north to south: see Fig. 2a in main paper). The X- and Y-coordinates (‘EE’ and ‘NN’) are in metres from the SW-corner (origin) of the plot. ‘Tag’ is the permanent tree number, and ‘Code’ the eight-character designations of the species’ names. Almost always, the first four characters of this code are the starting letters of the genus name and the second four those of its epithet (and where species are not known to species level of the form “sp1", for example, or “spnv”). Note one line is for indeterminates “Indet”! The full list of Latin names are given in codes.txt: authorities and family names are listed in Appendix B to the paper. Recording periods for the enumerations of the P-plot, with their median dates in 1991 and 2005, the P-plot extensions around 2000, and of the NW-plot, with median date in 2003, are all to be found in the main paper. ‘Diam1' (in 1991) and ‘Diam2' (in 2005) are the tree stem diameters in centimetres. The exact heights of measurement necessarily differed across trees: these are not given here – refer to the Methods in main paper and Appendix A there for details). Where a tree had a substituted (estimated) diameter (see Appendix A for reasons and methods used) this is indicated by a value of 1 in the column ‘Substn’ (1991-measured trees only). In p91lar.txt, ‘Stat2' indicates alive (1) or dead (0) condition by 2005, whilst in p05lar.txt it shows whether a tree was a survivor from 1991 (1) or a recruit into the size class by 2005 (2). (‘Stat1' by definition was always 1.) The same system of variable names is applied for the NW-plot large trees in nw03lar.txt. Here ‘Col’ runs again east to west from subplot columns 35–49, and ‘Row’ north to south as JN–XN (see Fig. 2b in main paper; and conforming to the overall grove map in Fig. 1 ibid.): there has been no main second enumeration of the NW-plot (except for the two part censusses in 2009 and 2001 – see main paper). The P-plot extensions are in pe00lar.txt. The six extensions travel north and south of their corresponding P-plot edge positions: coordinates (‘E’ and ‘N’) are for within-subplot distances only. Enumerations of medium-sized trees (stem diameter 10 –< 50 cm) are listed in p91med.txt and nw03med.txt. ‘Col’ and ‘Row’ again define the subplots measured, ‘Code’ and ‘Diam1' have same meaning as for large-tree files. For the P-plot, ‘Stat2' indicates whether the tree by 2005 was alive, either remaining in the medium size class (1) or advancing into the large one (–1), or was dead (0). Recruitment was not recorded. The NW-plot has had no second enumeration of medium trees so there is no status column. For the P-plot where the variable ‘Substn’ has a value of 1, the tree diameter was substituted (estimated): see Appendix A to main paper for reasons and methods used.
enumerations.zip
Tree stem growth data
Sixteen small files provide the stem diameters (‘Diamp’, cm) at the start of intervals (p), absolute growth rates, agr (‘Agrp-q’, cm/yr), and – as appropriate – species’ codes (‘Code’). The format is ANSI text (space-delimited). Fourteen files concern one or more of the three main caesalp species (Microberlinia bisulcata, Tetraberlinia bifoliolata and T. korupensis), for different time intervals and different size classes as uniquely indicated in their file names. Mean (± SE) of the agr-values are those given in Table 8 of the main paper. The two remaining files concern ‘other’ (Az) species measured in the P-plot and at the Isangele Road (IR) site for all large and some medium trees respectively. The former has species’ codes; the latter not because identifications were not certain. The Az mean (± SE) agr-values appear in the main paper text. A small script in R (agr_in) allows all of these files to be easily read into data-frames. To find relative growth rates, rgr, the time intervals will need to be applied to agr-values, and the resulting increments used to find final diameters and hence the rgr-values. The interval lengths are given in Table 8 ibid.
treegrowth.zip
Data files for spatial statistics analysis
Although these data could be found from the main plot data sets with a little light programming, the coordinates of the large and medium trees of the three principal species (Mb, Microberlinia bisulcata; Tb, Tetraberlinia bifoliolata; and Tk, T. korupensis) are given more simply here in six files, one for each species in each of the P- and NW-plots. Each file has the same format (ANSI text, tab-delimited), with coordinates (‘x’ and ‘y’, m), stem diameter (‘diam’, cm), the size class (‘dcl’; 1 = medium [10 – < 50 cm], 2 = large [≥ 50 cm]) and species code (as immediately above). This allows spatial analysis of each species individually, or for large and medium trees separately, as well as same-species large/medium bivariate analyses. However, to run (a) analyses on the large trees divided into two more size classes (see main paper text), the median must be first found, and then trees labelled ‘diam’ < or ≥ the median; or (b) different-species large/medium bivariate analyses, the medium trees of one species needs be recombined with large trees of either of the other two (six possible combinations).
spatstata.zip
Input files for ordination analysis
The six essential files here relate to the main ordination results and Fig. 3 of the main paper. They are based on the data sets for large trees (stem diameter ≥ 50 cm) in the P- and NW plots, viz. p91lar, p05lar and nw03lar, and applying the two alternative 1-ha grids (‘xa’, origin of plot [offset = 0, in title line]; ‘xb’, origin shifted [=1]: see main paper text for explanation). The format is ‘Cornell condensed’: each set of lines starting with the sample number (here 1-ha subplots) has several pairs of values – one the species’ number and the other its abundance (in this case basal area dm2/ha [‘ba/d’ = 1 in title line]), for those species present in that subplot. At the end of the file are listed the species’ codes (refer to codes.txt) and sample numbers (80 for P-plot, 49 for NW-plot). Rare species were not excluded. Note: the format line may need adjusting slightly to suit different programs; here it is set for PC-ORD; McCune & Grace 2002).
ordindata.zip