Frugivory and seed dispersal in a fragmented Afromontane forest, Taita Hills, Kenya: can the mutualism between birds and trees persist?

Tropical forest loss, degradation and isolation are believed to generate long-term effects on species diversity and composition, community dynamics and ecosystem processes, and are regarded as a major cause of the current biodiversity crisis. Although there is ample evidence of impoverished biodiversity as a consequence of habitat fragmentation, its most profound effects may actually result from functional changes in ecological processes such as trophic interactions. Theoretically, the loss of one mutualistic partner could cause cascading extinctions. However, the diffuse character of many interactions ensures resilience to some degree, and effects of fragmentation on interactions and ecosystem processes are usually more subtle and complex. Understanding how fragmentation may affect ecosystems and communities through the disruption of biotic interactions is of critical importance toward mitigating adverse effects, yet empirical studies remain scarce. In this dissertation, I examined plant-disperser relationships as one example of a vital interaction that may be affected by fragmentation. More specifically, our aims were (i) to assess whether avian frugivores track fruits in space and time, and if this resource tracking behaviour is less evident in more isolated forested patches compared to larger patches in homogeneously forested landscapes and (ii) to determine if, and to what degree, seed dispersal and post-dispersal processes vary in fragmented landscapes and how this affects early recruitment. We answer these questions based on a high resolution study (one study area, questions partly answered based on one model tree species, three years of field data collection). The study area, the Taita Hills (South-East Kenya), is unique because (i) it is strongly isolated from other, similar mountains; (ii) it is severely fragmented (only 2% of the original habitat remains); (iii) a limited number of bird species act as seed disperser and (iv) there are long-term bird ringing data available. All these factors simplify the model system and enable an in-depth study of patterns and processes of the plant-frugivore mutualism in a fragmented landscape. We found no evidence for fruit tracking behaviour, neither in study plots surrounded by farmland nor in plots surrounded by forest (chapter 1), probably because the bird species in our study were not obligatory frugivores and have an opportunistic foraging strategy. Seed dispersal of the forest interior tree Xymalos monospora was strongly reduced in small and disturbed fragments, mainly due to the loss or decline of its main dispersers (chapter 2), which were forest specialist birds that are less mobile and more vulnerable to habitat loss and disturbance (chapter 1, 2). Because of reduced germination rate and speed of non-ingested seeds (chapter 3), and reduced germination and seedling establishment capacities under the parent tree (chapter 4), this reduced seed dispersal resulted in lower recruitment in small fragments (chapter 5). Even though seeds from forest interior species may be dispersed quite far into the matrix (chapter 3), inter-fragment movements were not observed and are considered to be extremely rare (chapter 3) as a result of extreme habitat loss in the Taita Hills, low connectivity and the absence of faithful long-distance dispersers (chapter 2). Recruitment outside the forest will likely fail due to desiccation of the seeds (chapter 5). However, exotic plantations immediately surrounding the indigenous forest patches are regularly visited by several frugivores (chapter 3) and may provide perches to birds and shelter for seeds and seedlings, protecting these from harsh environmental conditions prevalent in the hostile agricultural matrix, at least in years with a normal rainfall pattern (chapter 4, 5). Although our results suggest that both seed dispersal and fruit tracking behaviour by frugivores are severely affected in the fragmented landscape of the Taita Hills, one should be cautious about extrapolating these findings to other study areas and species, as the sensitivity of plant-frugivore interactions depend both on environmental characteristics (at patch and landscape-level) and on plant and frugivore species-specific traits. Our findings are promising in that the detrimental effects of fragmentation on fruiting plants and the frugivores that feed upon them in the Taita Hills are probably not (yet) irreversible. To counteract further detrimental fragmentation effects, we urgently need to preserve the habitat quality of the remnant forest stands, enlarge fragments or at least restore and conserve the exotic plantations surrounding them, and increase landscape connectivity