PHOSPHORUS DYNAMICS IN A LOWLAND AFRICAN RAIN-FOREST - THE INFLUENCE OF ECTOMYCORRHIZAL TREES

In the southern part of Korup National Park, Cameroon, nine replicate plots (40 x 80 m) in forest with low abundances (5-15% of the basal ar ea of all trees greater than or equal to 30 cm gbh [girth at breast he ight]; LEM) and nine with high abundances (45-68%; HEM) of ectomycorrh izal caesalpiniaceous legumes were selected from an earlier enumeratio n along an east-west 5-km transect. The codominant caesalps in HEM plo ts were Microberlinia bisulcata, Tetraberlinia bifoliolata, and T. mor eliana, and ordination showed the LEM and HEM plots to be floristicall y distinct. The HEM plots lay in a large well-defined patch of these c aesalps, set within surrounding LEM forest. A further contrast to HEM forest was achieved by selecting six LEM and six VLEM plots (very low, approximate to 0% ectomycorrhizal trees) on a second transect 12 km t o the north. Korup has an annual rainfall of 5180 mm with one very dis tinct 3-mo dry season (December-February). The soils are sandy, acidic , and very phosphorus poor. The aim was to determine whether LEM and H EM plots differed in their litter and soil phosphorus status and the c haracteristics of their phosphorus cycles. On 13 occasions between Aug ust 1988 and September 1990, litter and soils in the surface root laye r and the mineral layer 5 cm below were sampled on the main transect a nd analyzed for phosphorus fractions. This analysis distinguished betw een inorganic and organic forms and provided various measures of labil ity. Nitrogen and carbon contents, pH, and moisture contents of each l ayer, depth of the root layer, and amounts of litter (i.e., litter mas s) and soil were the other main variables. In a subset of plots, leaf litterfall was collected over the period and analyzed for phosphorus a nd nitrogen, and two litter-bag experiments in early wet seasons estim ated rates of litter disappearance. Root biomass and change with depth were estimated from pit samples within the plots. HEM plots had sligh tly higher mean basal area of trees than LEM plots (32.3 and 27.1 m(2) /ha, respectively) and higher fine root (less than or equal to 5 mm di ameter) biomasses (519 and 364 g/m(2), 0-5 cm), but the same mean litt er mass (155 g/m(2)) and annual leaf litterfall (720 g/m(2)) and very similar disappearance rates (1.82 g.m(-2).d(-1) on litterfall/mass bas is; t(1/2) of 53 d from litter bags). Litter mass (and litterfall) rea ched a clear peak in the mid-dry season in both HEM and LEM plots (sli ghtly earlier in HEM), almost completely disappearing by mid-wet seaso n. The phosphorus concentrations in HEM falling leaf litter was much g reater than in LEM (801 and 676 mu g/g) leading to greater inputs (1.7 9 and 1.35 mg.m(-2).d(-1)), while for nitrogen the concentrations and inputs were very similar (means 17.8 mg/g, 35.3 mg.m(-2).d(-1)). An im portant feature was the marked peak in phosphorus (but not nitrogen) c oncentration in litter in the HEM plots only in those years that did n ot follow a mast fruiting. Depth of the root layer was greater in the HEM than LEM plots (8.1 and 4.2 mm), as was carbon content (5.5 and 4. 0%), but bulk density was less (0.85 and 1.30 g/cm(3)). Moisture conte nts, which tracked the seasons in all three layers, were slightly high er in HEM than LEM, as were clay and silt contents, but pH values were lower (4.19 and 4.59, root layer). The relative differences in the ei ght soil phosphorus fractions (resin-, bicarbonate- and NaOH-extractab le, and chloroform-labile, inorganic, and organic) plus residual and t otal phosphorus were remarkably consistent between HEM and LEM plots. In the root layer total phosphorus was much higher in HEM than LEM plo ts (309 and 186 mu g/g, respectively) and likewise in the mineral laye r (192 and 119 mu g/g, respectively): HEM/LEM ratios of approximate to 1.6 were maintained across nearly all fractions. Covariance analysis indicated that the basal areas of LEM and HEM plots could not account for the differences in phosphorus concentrations. Different fractions followed different trends with time, but these did not match seasons a nd showed, for the more labile inorganic phosphorus fractions especial ly, a linear decline over the study period. Interactions between fores t type and date (split-plot repeated-measures analysis of variance) we re rarely significant even though date itself invariably was. In marke d contrast, nitrogen fractions did not decline, and labile organic nit rogen showed clear seasonal peaks. Differences between the LEM forest on the two transects (two occasions in common) were small and inconsis tent, supporting the wider contrast. with HEM plots. Principal compone nts analysis of the phosphorus and nitrogen fractions showed a strong discrimination between LEM and HEM plots. Calculation with amounts of phosphorus in the top 6 cm of the soil showed very similar patterns to the concentrations. Large ectomycorrhizal trees appeared to have incr eased the depth of the surface root layer, its phosphorus content, the labile phosphorus fraction (notably its organic component), and to ha ve enhanced phosphorus cycling. The role of ectomycorrhizas in this pr ocess is discussed. The decline in labile phosphorus is explained by a phenological and climatic ectomycorrhizal response (PACER) hypothesis that highlights the relationship between phosphorus demand in mast fr uiting years and soil phosphorus concentrations. The importance of thi s in adaptations by ectomycorrhizal caesalps to this strongly seasonal and phosphorus-poor site is considered.