Reduced availability of large seeds constrains Atlantic forest regeneration

Abstract

Secondary forests are expanding in defaunated fragmented tropical landscapes, but their resilience potential remains poorly understood. In this study we used a chronosequence of advancing (19–62-yr old) Atlantic forest regeneration following slash-and-burn agriculture to infer successional shifts in seed rain in terms of seed density, species richness, taxonomic and functional composition, and local spatial distribution. After monitoring seed rain during 12 months in 60 1-m² seed traps, we recorded over 400,000 seeds belonging to 180 morphospecies. From early to late-successional stage, seed rain decreased in density, increased in per capita species richness, gradually changed in species composition, and became less aggregated spatially. Regardless the age of forest stand, vertebrate-dispersed seeds accounted for 67–75% of all species recorded. Large-seeded species typical of old-growth forests, on the other hand, accounted for only 5–8% of the species recorded in the seed rain, a proportion around five times smaller than that reported for the old-growth forests of the same study site (31%). Our results suggest that the secondary forests considered, which are embedded in one of the largest (3500 ha) and best preserved remnant of the severely fragmented Atlantic forest of Northeast Brazil, may fail attaining older successional stages due to the reduced availability of large-seeded late-successional species. This regeneration constraint may be even stronger in smaller, more isolated forest remnants of the region, potentially reducing their ability to provide ecosystem services.

Introduction

Secondary forests are becoming increasingly common in the tropics as abandoned agricultural lands recover through secondary succession (FAO, 2009, Gardner et al., 2009). Global estimations indicate that around half of the world’s remaining tropical forest is second-growth forests or degraded old-growth forests affected by selective logging, road opening, fire, fragmentation, overhunting, and other anthropogenic forces (ITTO, 2002). Unfortunately, the rapid expansion of secondary tropical forests has been not paralleled by a similar understanding on their functioning. Recent studies highlight the potential of these novel forests to protect soils, cycle nutrients, support wildlife, store carbon, maintain watershed function, and mitigate species extinction (Wright and Muller-Landau, 2006, Chazdon et al., 2009, Lugo, 2009), but the ecological processes involved in the maintenance of these functions are not completely understood.

It is well-known that the continuous influx of native seeds is among the most important ecological processes involved in forest regeneration (Guariguata and Ostertag, 2001). Altered seed rain invariably results in biased seedling assemblages (e.g. Melo et al., 2006, Melo et al., 2007), which in turn result in impoverished tree assemblages (Laurance et al., 2006, Santos et al., 2008, Santos et al., 2010) and possibly entire communities with limited ability to maintain biodiversity and provide ecosystem services (see Gardner et al., 2009, Parry et al., 2009, Schindler et al., 2010). Nevertheless, seed rain remains scantly studied in secondary tropical forests (e.g. Barbosa and Pizo, 2006, Castillo and Ríos, 2008), particularly in older forest stands (>45-yr old) where community structure, composition, and functioning start to approach those of old-growth forests (Saldarriaga et al., 1988, Guariguata and Ostertag, 2001, Chazdon et al., 2007).

Seed arrival from adjacent areas assumes a disproportional importance in the first years of regeneration when deforestation is followed by fire (Uhl, 1982), given that fire eliminates the soil seed bank, pre-existing seedlings and saplings, resprouting of broken trees and other sources of plant recruitment (Yao et al., 1999). As regeneration advances and the first colonizers achieve maturity, seeds produced locally also contribute to forest recovery (Guariguata and Ostertag, 2001). Some decades later, the arrival of late-successional seeds from adjacent old-growth forests − most large seeds dispersed by large-bodied vertebrates (e.g. monkeys) − is critical to drive mid-successional stages toward older successional ones. Assuming that deforestation and forest fragmentation negatively affect large-bodied seed dispersers, reduce the amount of old-growth forests locally and regionally, and favor the hyper-proliferation of early-successional small-seeded pioneer trees (Silva and Tabarelli, 2000, Laurance et al., 2008, Tabarelli et al., 2010), it is likely that local reduction in the availability of large seeds constrains forest regeneration following slash-and-burn agriculture in fragmented neotropical landscapes. This phenomenon has been described for Atlantic forest edges (Melo et al., 2006) and helped to explain why these degraded habitats are not able to attain older successional stages or maintain old-growth forests (Santos et al., 2008, Santos et al., 2010, Tabarelli et al., 2008).

In this study we examined successional shifts in seed rain across 19, 32, and 62-yr old forest stands recovering after slash-and-burn agriculture in the Atlantic forest of Northeast Brazil, a severely fragmented region now reduced to less than 6% of its original pre-Colombian extent (Galindo-Leal and Câmara, 2003). We selected forest stands completely immersed into a large (3500 ha) forest remnant that still retains large tracts of old-growth forest, but not all large frugivores recorded for the Atlantic forest (see Methods). Every 15 days during 12 months we monitored seed rain in terms of seed density, species richness, taxonomic and functional composition, and spatial distribution at the forest stand level. We predicted that local seed density would decrease and per capita species richness would increase during forest succession due to the replacement of pioneer trees by a more diverse group of late-successional trees that reproduce less frequently and more scantly. The proportion of vertebrate-dispersed species was expected to be elevated at any forest stand because zoochory is a widespread dispersal syndrome in the local flora (see Oliveira et al., 2004, Santos et al., 2008). The proportion of large-seeded species in the seed rain was predicted to increase with forest age because active dispersal of large seeds in the study site is greater in old-growth than second-growth-like areas (Melo et al., 2006). As long described in the literature of tropical forest regeneration (reviewed by Guariguata and Ostertag, 2001), we expected gradual changes in species composition and floristic divergence during the regeneration process. Finally, because copious seed production is expected for forest stands dominated by pioneer plants (i.e. earlier successional forest stands) and more than 99% of the seeds in tropical forests usually fall down near parents (Terborgh et al., 2002, Bentos et al., 2008), we predicted a gradual reduction in spatial aggregation of seed rain, which may further minimize negative density dependence effects and result in more diverse seedling assemblages (Harms et al., 2000).