The plant seed dispersal by animals is a central process for the self-maintenance and dynamics of tropical forests. This is also true for the Shimba Hills National Reserve in south-eastern Kenya, the object of this study. This work compares local endozoochory, seed predation, the disperser spectra of plants and the (diaspore) food spectra of dispersers respectively, the duration of dispersal and resulting distribution patterns, as well as the temporal fruit availability and its utilisation. Following the life-cycle of plants, the early stages in plant recruitment and their complex biocoenotic connexes are investigated. Research on post-dispersal stages includes the role and establishment of the soil seed bank, natural germination responses of seeds, and various aspects concerning safe sites of seedlings and saplings. In addition, contributions are made to the occurrence and some synecological impacts on plant maintenance of animal species, for some of which relatively little relevant information was available. Furthermore, the relevant theories and methods of the mentioned topics are critically discussed. The project is part of the research program 'mechanisms for the maintenance of tropical diversity' of the German Research Foundation. Attention will be paid to the fundamental ecological questions of this program, i.e. some stochastic versus deterministic processes, mechanisms and causalities will be uncovered. However, the main emphasis of this work is focused on seed dispersal (sensu lato), natural forest regeneration and an improved understanding of the function and maintenance of the local biocoenosis. The knowledge gained is provided in order to benefit the conservation of the Reserve. The various screenings of this study are also meant to help fill the East African gap in the global map of rarely known seed dispersal spectra in species-rich communities. The Shimba Hills National Reserve represents a coastal lowland refuge of about 220 km2 for over 1000 angiospermous plant species and an as of yet unknown large number of animal species. The main vegetation pattern is a mosaic of seasonal evergreen forests, forests in either stages of progress or regression as well as anthropogenic grasslands and pine plantations.
Diaspore utilisation, and dispersal respectively, for about 100 000 diaspores of 549 plant taxa by 37 animal taxa were investigated with the use of 2000 faeces analyses as well as feeding and dispersal observations, feeding experiments and a few gut investigations. Among the five most intensively investigated animal species, the elephant dispersed most plant species - followed by the baboon, the Garnett's galago (bushbaby) and the African civet respectively, and the rusty-spotted genet. Some results indicate other animal species also taking an important part in plant recruitment. It was possible to verify a large number of dispersal and/or diaspore utilisation interactions. The dispersal efficiency differed inter- and intraspecifically in terms of dispersed diaspores and species per defecation, among other factors. Many animals behaved selectively in various degrees as seed predators whereby dispersal was usually still assured. The evaluation of animal activities and the determination of diaspore passage times through the digestive tracts results in different dispersal and distribution patterns - at least when considering short biological time scales. Phenological analyses, based on numerous verified feeding and dispersal interactions, showed that various plant species provide fruits throughout the year, which are utilised and dispersed by many animal species. The total and some individual fruit availability varied temporally as well as spatially; still a number of specific main fruiting periods become apparent. A weak fruiting peak occurs towards the end of the major rainy season. Furthermore, the results indicate that even roaming animals only utilise less than half of the monthly available species with fruits. Generally, the results for both the plants' dispersal and the dispersers' food show niche overlapping as well as differences.
Soil seed bank investigations at the end of the dry and main rainy season showed a lower diaspore and species availability at the end of the rainy season, though the overall fruit availability in the area seems then at its highest point. In total, diaspores of over 60 taxa were found in three horizons up to 12 cm in depth. Non-viable seeds were included as an indicator of intact diaspores. Many forest species seem only able to regenerate via fresh seed rain and not due to persistent seed banks, but the analyses were troublesome and the relatively few and also locally restricted soil samples limit most conclusions. An improved understanding of the regeneration potential and dynamics in an area, however, was found to require an integration of different approaches. In addition, it is shown that seeds are incorporated into the ground by various means including the secondary seed dispersal activities of dung beetles. Field observations and experiments demonstrate that the depth of seeds in the substrate varies and that the burial depth can particularly be crucial for the germination and subsequent seedling establishment of small-seeded species. The composition, establishment, function and general importance of soil seed banks are discussed.
Germination response was tested for over 10% of the local species in 576 different seed germination trials, which involved about 22 000 diaspores of 168 plant species. Quiescence (> 95% of 112 successfully tested taxa) and occasionally individual dormancy are common. Only the seeds of one species were not quiescent and even able to germinate in the dryer (here termed 'non-quiescence'). The abundance of quiescence and rather short seed viability also explains the lack of persistent soil seed banks with forest species. External moisture appears to be one of the main factors triggering germination. Light was not required for the germination of most of the buried seeds. Similar responses were also frequent in the field and in bags filled with seed-containing dung. The trials with diaspores/seeds collected from plants and faeces for over a total of 150 plant species, showed that passage through a digestive system is not obligatory for the germination of the tested seeds. Simple frugivory or any fruit treatment appears for many fleshy-fruited plants to be as equally effective for seed germination as endozoochory, which includes these processes. There exists almost no evidence of tight coevolutive plant-animal relationships concerning seed dispersal and germination. The temporal variation in germination most likely results as an advantage for risk spreading and thus appears more important than a possibly altered germination response due to endozoochory. Considerable variation was found after long-term experiments or slightly changed repetitions. The general use and interpretation of germination tests is thoroughly discussed.
Safe sites for germination and seedling establishment were evaluated through the use of examples. The examples were based on field observations, germination tests, feeding and direct plant damage trials, heat treatments of seeds, and seedling/sapling exposure inside and outside fenced field plots. Epigeal seedlings are common (65% of 62 species), often expose the remainder of the seed and seem still susceptible to seed predators. However, at least for the sole tested species (Tabernaemontana pachysiphon), seedlings can regenerate from a rest meristem in the hypocotyl after the testa (including the cotyledons) and the apical meristem have been removed. Field observations and one experiment involving over 600 transplanted seedlings and young saplings from eight species showed a high mortality for these early plant stages. Neither habitat nor soil type, microclimatic conditions nor protection by fencing usually had an obvious general influence on their chances of survival. There was no evidence of larger herbivores causing damage in these trials or creating major bottlenecks in the field. Seedling mortality was also high in the forest, though drought and fire seem particularly critical at more exposed sites. Only a few seeds of a few species, among them an introduced one, germinated after relatively high dry heat treatment, which simulated temperatures possible during grassland burning. No seed of any species germinated after exposure at 150O C for 20 min. Seedlings of the sole tested forest tree species (Zanha golungensis) survived drought conditions better than either the still non-germinating seeds or germinating seeds, albeit all three stages had been kept dry for over three weeks. Safe sites are barely predictable due to numerous dangers, particularly for early plant stages. Potential safe sites, as provided by various dispersers, also appear quite variable. Nevertheless, notably elephant boli serve as important seed beds in which plants can even grow at disturbed and sun-dried sites.
Local animals were observed and partly more intensely investigated for their occurrence and their positive and negative impacts on plant recruitment and forest maintenance. Special attention is paid to biocoenologically relevant aspects of viverrids, elephants, primates, fruit bats, reptiles, ruminants, suids, rodents (i.e. giant rats), birds (i.e. hornbills), other vertebrates and some invertebrates (i.e. dung beetles). Furthermore, direct and indirect seed dispersal by carnivores (the latter was verified for snakes with extreme gut conditions), feeding associations, the impact of animal behaviour and potential diaspore hoarding in the tropics are addressed. A main result is that in the Shimba Hills most plant species benefit from (endo-) zoochory. Furthermore, most animal species more or less participate in both forest regeneration and plant damage. Interspecific latrine use by viverrids has been encountered and recorded with camera traps. Pioneer effects for forest regeneration can be caused by many widely scattered elephant boli as well as by the use and sharing of latrines at a few exposed and disturbed sites. Moreover, latrine use as well as the exceptional defecation of some intact seeds has been verified for giant rats and has also been observed for a suni in an enclosure at the Bamburi Nature Trial. In addition, the probably endangered and rare bushy-tailed mongooses and palm civets have been documented in the Shimba Hills.
General discussion
The more detailed an aspect has been investigated, the more possibilities of variation became apparent. Unfortunately, testing ecological theories is always limited and generalisations about the impact of any animal species are often largely impossible. In practice, particularly the recently discussed so-called 'effectiveness of dispersers' is problematic to assess, because in reality too many factors must be taken into account. From a theoretical point of view, several causes of seed dispersal are summarised. Among the regularly considered non-mutually exclusive dispersal impacts, 'colonisation' is common and also frequently provided by larger mammals, which have often been assumed as ineffective due to clumped seed deposition. 'Directed dispersal' to potential safe sites is also not restricted to highly specialised dispersal systems as described before. This mechanism is typical for endozoochory and clumped excrements, mainly due to the secondary post-dispersal activities of dung beetles and ants. The widely assumed density-dependent mortality after clumped deposition seems less critical due to common secondary dispersal processes, among other factors. Endozoochory is particularly common in the tropics and dung-dependent secondary dispersal frequently occurs, though it has been so far largely ignored. Therefore, this mechanism has been termed here as 'coprochory'. As a new observation, occasionally even roundish seeds from larger mammalian faeces are buried by 'rollers' instead of dung balls. Embedding of seeds in a germination substrate appears often crucial for avoiding lethal desiccation at unprotected regeneration sites. In this area and most likely in others too, larger mammals regularly provide long-distance dispersal. Seeds and seedlings in their faeces may sometimes suffer density-dependent mortality, but the large faeces can also invite secondary dispersal or may form a safe site itself. Resulting seed shadows and supports for establishment as well as variation in the plant-disperser system and stochastic impacts should, in future discussions about density- and distance-related seed and seedling mortality, be more respected. The multiple diaspore utilisation by a large array of animals and the 'polyzoochory', which has been proved for many plant species, further reject the still widely accepted hypothesis of specialised dispersal systems or the meanwhile largely dismissed idea of close coevolution between plants and dispersers. Existing dispersal syndromes have little in common with various findings of this study. Recently suggested dispersal syndrome analyses are misguiding and better avoided, because for zoochory they are usually based on unsuitable data. Forest maintenance depends on a number of processes and mechanisms such as the investigated animal behaviours, primary and secondary seed dispersal (coprochory), quiescence and (individual) dormancy, among others. General theories about forest regeneration and some potential bottlenecks are discussed. The forests of the Shimba Hills are often considered to be endangered supposedly by too many elephants. Apparently, however, there exist different views about how forests are maintained and what they should look like. For the central, closely observed parts of the Reserve, overall forest rejuvenation was subjectively found to be quite stable during the course of this project. This comparative study shows that elephants are particularly important for seed dispersal and forest regeneration in the Shimba Hills. The gradual forest exploitation and the fierce grassland burning appear similarly or even more critical for the future of the forests than elephants alone. The Shimba Hills are highly important due to their extraordinary biodiversity, their function as a water catchment area and their many other potentials. Some difficult ecological problems and conceivable solutions have been elucidated. The abundant quiescence and short seed viability, which go along with the largely missing persistent seed banks, the strong dependence on an intact community for seed dispersal and forest maintenance as well as the documentation of additional species that require protection, are further arguments provided by this study that urge and justify thorough conservation. The main emphasis about the area should be laid on the forest-related biocoenosis and this important East African lowland refuge is recommended for an upgrading to National Park status.
