Abstract

Despite the importance of tropical forest fragmentation on carbon balance, most of our knowledge comes from few sites in the Amazon and disregard long-term underlying processes related to landscape configuration. Accurate estimation of fragmentation effects should account for additive edge effects and edge age. Here we investigated those effects on C stock and forest structure (density, height, basal area) in fragments (13 to 362ha) of forest with≥70years old, surrounded by pasture, in the Brazilian Atlantic forest region. We measured 5297 stems sampled in four categories replicated in eight fragments: fragment interiors (>110m from edges); old (>50years) corner edges (< 50m); old straight edges; and new (ca. 44years) straight edges. Aboveground biomass was estimated from tree height and diameter at breast height, and converted to carbon. Carbon stock was highly variable between categories, scoring from 10.44Mgha−1 up to 107.59Mgha−1 (average of 41.27±23Mgha−1). Contrary to our expectations, interior plots did not have higher carbon stock, basal area or tree stem density than edges, but only taller trees. We found no significant effects of edge age or additive edge effects on carbon stocks. These results suggest that edge effects in the Atlantic rainforest may differ from those observed in more recently fragmented tropical forests, such as the Amazonian forest. We hypothesize that in heavily human-modified landscapes, more extensive edge effects combined with other human disturbances on tree mortality and carbon stock may contribute to overall high levels of degradation, reducing differences between edge and interior habitats. Existing models based on Amazonian forest data may underestimate the true impacts of fragmentation on carbon storage in landscapes with an old history of human disturbance.