Chance versus determinism in canopy gap regeneration in coastal scarp forest in South Africa

Question: Is tree regeneration in canopy gaps characterized by chance or predictable establishment. Location: Coastal scarp forests, Umzimvubu district, Eastern Cape Province, South Africa. Methods: Estimation of richness of gap-filling species across canopy gaps of different size. Data are compared with regeneration under the canopy. Probability of self-replacement of gap forming species is calculated. Results: Forest area under natural gap phase was 7.8%, caused mostly by windthrow (54%). The abundance and average size of gaps (87.8 m(2)) suggests that species diversity may be maintained by gap dynamics. However, only four of 53 gap-filler species displayed gap size specialization and these were pioneer species. An additional 13 species were more common in larger gaps but there was no gradient in composition of gap-filler species across gap size (p = 0.61). Probabilities of self-replacement in a gap were low (< 0.3) and common canopy species were equally abundant in gaps and the understorey. Species composition in gaps showed no pattern of variation, i.e. was unpredictable. which suggests absence of a successional sequence within tree-fall gaps. There was also only a slight increase in species richness in gaps at intermediate levels of disturbance. Conclusions: Coastal scarp forest appears not to comprise tightly co-evolved, niche-differentiated tree species. Unpredictable species composition in gaps may be a chance effect of recruitment limitation of species from the species pool. Chance establishment slows competitive exclusion and may maintain tree diversity in these forests. These data suggest that current levels (3 gaps per ha) of selective tree harvesting may not cause a reduction in species richness in this forest.