The arboreal crab Parasesarma leptosoma has been recently discovered at Mngazana, a southerly mangrove system in southern Africa, where crab tree preferences were studied using an indirect (browse leaf damage) and a direct (tree traps) method. The extent of crab induced leaf damage was compared for three mangrove species at two sites, one next to a tidal creek and one away from the creek. Using ANOVA, significant differences were found between tree species (P < 0.001) at different distances from the creek (P < 0.022). Crabs were found to occur on Rhizophora mucronata and Brugueira gymnorrhiza, but not on Avicennia marina. This reflected a gradient in browsing, from well-browsed R. mucronata (100% near the creek and 25.7% away from the creek), to medium browsing of B. gymnorrhiza (51.5% near and 0% away) and no browsing on A. marina (near or away). These differences could be explained in terms of palatability, as both R. mucronata and B. gymnorrhiza are salt excluders, while A. marina secretes salt from its leaves. Leaf consumption levels averaged between 1.73% and 2.6% of leaf area for R. mucronata and 0-1.76% for B. gymnorrhiza. For both R. mucronata and B. gymnorrhiza there was a significant correlation between the number of crabs caught directly and the amount of browse leaf damage (P < 0.01). Crab number was also significantly correlated with tree circumference for R. mucronata (r2 = 0.67) and B. gymnorrhiza (r2 = 0.76, P < 0.05), with crabs more prevalent on the former tree species and no crabs trapped on A. marina (91.7%, 38.3% and 0% catches, respectively), thus reflecting the results obtained by the indirect method. Total Nitrogen and Phosphate were measured for both sediment and leaves of the three mangrove species at the two sites. Leaf comparisons showed significant differences (P < 0.01) for both Total Nitrogen and Phosphate with R. mucronata having the highest values, followed by A. marina and lastly B. gymnorrhiza. Total Nitrogen was significantly higher for both B. gymnorrhiza and R. mucronata compared with A. marina, while leaf phosphate was significantly lower for B. gymnorrhiza when compared with both R. mucronata and A. marina. No significant differences were found for leaf nutrients between sites, with the exception of A. marina and R. mucronata Total Nitrogen, which was significantly higher at the near creek sites (P < 0.05). Sediment analysis showed no significant differences (P > 0.05) in either nutrients or median particle size. Thus, R. mucronata, especially near the creek, had higher nutrient value and was probably more palatable and could explain observed differences in crab distribution. Very little browse damage was encountered in saplings below 10 cm. Most poles chopped by the local communities are R. mucronata in the 15-20 cm category, which coincides with peak crab frequencies in the 15-25 cm size classes for R. mucronata and B. gymnorrhiza, so that this selective harvesting is affecting this crab population maximally. Predictions were made as to the effect of crab loss, tree replacement rate and alternatives to chopping, which would boost community socio-economic levels and reduce the anthropogenic pressure on this biodiverse southerly mangrove system.