Seed Dispersal and Frugivory:
Ecological Consequences for
Tree Populations and Bird Communities
Von der Fakultät für Mathematik, Informatik und Naturwissenschaften - Fachbereich 1 - der Rheinisch - Westfälischen Technischen Hochschule Aachen zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften genehmigte Dissertation
vorgelegt von Diplom-Biologin Bärbel Bleher aus Urach, jetzt Bad Urach
Berichter: Universitätsprofessor Dr. rer. nat. Ingolf Schuphan Universitätsprofessor Dr. rer. nat. Hermann Wagner
Tag der mündlichen Prüfung: 13. September 2000
If I know a song of Africa, of the giraffe and the African new moon lying on her back, of the plows in the fields and the sweaty faces of the coffee pickers, does Africa know a song of me? Will the air over the plain quiver with a color that I have had on, or the children invent a game in which my name is, or the full moon throw a shadow over the gravel of the drive that was like me, or will the eagles of the Ngong Hills look out for me? T. Blixen dedicated to my parents
CONTENTS
1. GENERAL INTRODUCTION 1
1.2 SEED DISPERSAL BY ANIMALS AND CONSEQUENCES FOR PLANTS 1 1.2 FRUIT AVAILABILITY AND CONSEQUENCES FOR FRUGIVOROUS ANIMALS 2 1.3 RELEVANCE FOR CONSERVATION 3 1.4 AIMS OF THESIS 4
2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES 7
2.1 INTRODUCTION 7 2.2 THE TREES 8 2.3 STUDY SITES 9 2.4 METHODS 10 2.4.1 FRUGIVORE DIVERSITY 10 2.4.2 TREE OBSERVATIONS 10 2.4.3 FRUIT TRAPS 10 2.5 RESULTS 11 2.6 DISCUSSION 16 2.7 SUMMARY 19
3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL, SEEDLING ESTABLISHMENT AND THE SPATIAL PATTERN OF SEEDLINGS AND TREES 21
3.1 INTRODUCTION 21 3.2 STUDY SITES AND SPECIES 22 3.2.1 STUDY SITES 22 3.2.2 STUDY SPECIES 23 3.3 METHODS 24 3.3.1 SEED DISPERSAL 24 3.3.1.1 Fruit traps 24 3.3.1.2 Tree observations 25 3.3.2 SEEDLING ESTABLISHMENT, DISPERSAL BENEFIT AND SEEDLING DISTRIBUTION 25 3.3.2.1 Seedling establishment 25 3.3.2.2 Benefit of seed dispersal 26 3.3.2.3 Seedling distribution 27 3.3.3 SPATIAL DISTRIBUTION OF TREES 28 3.3.3.1 Field data 28 3.3.3.2 Computer simulation 28 3.4 RESULTS 29 3.4.1 SEED DISPERSAL 29 3.4.2 SEEDLING ESTABLISHMENT, DISPERSAL BENEFIT AND SEEDLING DISTRIBUTION 32 3.4.2.1 Seedling establishment 32 3.4.2.2 Benefit of seed dispersal 33 3.4.2.3 Seedling distribution 33 3.4.3 SPATIAL DISTRIBUTION OF TREES 35 3.4.3.1 Field data 35 3.4.3.2 Computer simulation 36 3.5 DISCUSSION 37 3.6 SUMMARY 41
4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND THE SPATIAL PATTERN OF TREES 43
4.1 INTRODUCTION 43 4.2 METHODS 44 4.2.1 SIMULATION MODEL 44 4.2.1.1 Tree populations 44 4.2.1.2 Simulation parameters 45 4.2.1.3 Running the simulation 46 4.2.2 DETECTING SPATIAL PATTERNS 46 4.2.3 STATISTICAL ANALYSIS 47 4.3 RESULTS 47 4.3.1 FACTORS INFLUENCING SPATIAL PATTERNS 47 4.3.2 SEED DISPERSAL AND SPATIAL PATTERNS 49 4.3.3 BREEDING SYSTEM AND SPATIAL PATTERNS 50 4.3.4 TREE DENSITY AND SPATIAL PATTERNS 51 4.3.5 START POPULATION AND SPATIAL PATTERNS 51 4.3.6 COMBINED EFFECTS OF DISPERSAL DISTANCE AND TREE DENSITY 52 4.4 DISCUSSION 53 4.5 SUMMARY 56 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES FOR FRUGIVORES IN A DRY FOREST IN KWAZULU - NATAL, SOUTH AFRICA 59
5.1 INTRODUCTION 59 5.2 STUDY SITE 60 5.2.1 LOCATION 60 5.2.2 CLIMATE 61 5.2.3 VEGETATION 62 5.2.4 FRUGIVORE COMMUNITY 63 5.3 METHODS 63 5.3.1 FRUIT AVAILABILITY AND ABUNDANCE 63 5.3.2 FRUIT UTILIZATION BY THE FRUGIVORE COMMUNITY 64 5.4 RESULTS 64 5.4.1 FRUIT AVAILABILITY AND ABUNDANCE 64 5.4.2 FRUIT UTILIZATION BY THE FRUGIVORE COMMUNITY 67 5.4.2.1 Frugivore community 67 5.4.2.2 Plants used by frugivores 70 5.5 DISCUSSION 73 5.5.1 FRUIT AVAILABILITY 73 5.5.2 FRUGIVORE COMMUNITY 74 5.5.3 RESOURCE UTILIZATION AND KEYSTONE SPECIES 75 5.6 SUMMARY 77
6. GENERAL CONCLUSIONS 79
7. REFERENCES 83
8. APPENDIX 96
9. ACKNOWLEDGEMENTS 103
10. CURRICULUM VITAE 105
1. GENERAL INTRODUCTION ......
1. General Introduction
Mutualistic plant-animal interactions exist in a wide variety with one of the most important being seed dispersal or the transport of seeds away from a parent plant by animals (Howe & Westley 1988). Not only plants depend on animals for the dispersal of their seeds, animals, too, rely on plants for fruit as a food resource. Seed dispersal by animals and frugivory have reached their pinnacle in tropical forests, where a percentage of up to 90% of tree and shrub species produce fleshy fruits adapted to animal dispersal and eaten by a large number of vertebrates (Frankie et al. 1974, Foster 1982a, Howe & Smallwood 1982, Howe 1986). Howe & Westley (1988, p.105) state that "one of the exciting challenges of modern ecology is to acquire an understanding of the means by which plants and animals exploit each other to their mutualistic benefit". For this understanding both facets of the mutualistic interaction between animals providing seed dispersal and plants providing fruits as a food resource have to be examined and its consequences for their mutualistic partners determined.
1.2 Seed dispersal by animals and consequences for plants
The important role that animals, and especially birds, play in the seed dispersal of tropical plants is well documented (e.g. Howe & Estabrook 1977, Howe & Vande Kerckhove 1981, Howe 1986). In general, three hypotheses have been proposed to explain the benefit of seed dispersal for the plants (Howe & Smallwood 1982). The escape hypothesis assumes that seeds and seedlings might escape from disporportionate mortality near the parent due to pathogens, predators or seedling competition. The colonization hypothesis is applied when seed dispersal might allow the parent plants to establish their offspring in vacant sites. The directed dispersal hypothesis assumes that dispersing agents transport seeds to special microsites critical for germination and establishment. Many studies have been conducted to evaluate these hypotheses, however, although several have supported the first one, results remain controversial and hypotheses are not mutually exlusive (for a review see Clark & Clark 1984).
1 1. GENERAL INTRODUCTION ......
Whatever hypothesis might be applied, in general, the seed shadow or the spatial distribution of seeds around the parent plant shows a peak under and close to the parent tree and a steady decline away (Willson 1992). Fruit-eating animals might alter the shape of the seed shadow and consequently exert varying influence on plants depending on their behavior: animals may drop seeds under the parent plant during fruit handling, destroy them or disperse them from the vicinity of the parent plant. Only in the latter case do they benefit the plant as various studies have shown that survival of seeds and seedlings is higher further away from the parent plant (Janzen 1970, Connell 1971, Augspurger 1983, 1984, Howe et al. 1985). Consequently, seed dispersal is though to affect seedling establishment and the spatial pattern of seedlings and saplings (Fleming & Heithaus 1981, Howe 1986). Differential seedling distribution might be a critical determinant of offspring survival and influence density, spatial patterning and composition of plant communities in general (Fleming & Heithaus 1981, Coates-Estrada & Estrada 1986, Howe 1986). Fruit-eating animals therefore might have a significant influence on the population dynamics of tropical forest communities, however, studies linking animal-mediated seed dispersal and seedling distribution with the spatial pattern and dynamics of plant populations are rare (but see Fleming & Heithaus 1981).
1.2 Fruit availability and consequences for frugivorous animals
Complementing the diversity of fruits suitable for consumption by animals is an equivalent diversity of frugivores, i.e. fruit-eating birds and mammals that depend mainly on fleshy fruit. In tropical forests, frugivorous animals are the dominant group of vertebrates. The challenge of frugivores is to find, eat and subsist partly or entirely on fruits that are mostly deficient in protein, but rich in carbohydrates or lipids (Howe & Westley 1988). Fruit availability is one crucial factor influencing the frugivore community (Howe & Estabrook 1977, Thompson & Willson 1979). It appears to be highly variable also in the tropics and marked by periods of scarcity and abundance (Terborgh 1986a). Most phenological studies report seasonality with fruit being more limiting in some seasons than in others for most tropical ecosystems (e.g. Frankie et al. 1974, Foster 1982a). Consequently, frugivorous animals are facing seasonal irregularities in their food resources. Plants providing
2 1. GENERAL INTRODUCTION ...... fruits during periods of low general fruit production play an important role in maintaining entire frugivore communities (Howe 1977). These so-called "keystone resources" have been identified for several forests with figs being of importance (Leighton & Leighton 1983, Terborgh 1986a) as well as a variety of lipid-rich arillate species (Leighton & Leighton 1983, Gautier-Hion & Michaloud 1989). To make the keystone species concept more applicable for conservation, it was recently defined as "a species whose impacts on its community or ecosystem are large, and much larger than would be expected from its abundance" (Power & Mills 1995). Consequently, rather uncommon or rare species might be keystones as compared to abundant dominants. Phenological studies assessing fruit availability and identifying keystone species exist for various tropical forests in the Neotropics and South-East Asia, however, only a few are known for the African continent and none for Southern Africa.
1.3 Relevance for conservation
Increasing fragmentation and destruction of tropical ecosystems is known to not only affect biodiversity but also processes crucial for the maintenance of these ecosystems such as seed dispersal and frugivory. Fragmentation, for example, may alter the composition of bird assemblages and the relative contribution of some species as seed dispersers (Pizo 1997, Santos & Telleria 1994, 1997). Loss of dispersal agents might lead to a break-down in seed dispersal (Howe 1984a, Pizo 1997) and reduce seedling establishment and recruitment of fruiting trees. Consequently, disrupted seed dispersal has consequences for the regenerative potential of ecosystems and probably leads to changes in the abundance and spatial distribution of trees (Hubbell 1979), perhaps even to higher extinction probabilities of focal trees (Bond 1995, Tilman et al. 1994). In the long term, a break-down of seed dispersal processes might be a serious threat to tropical plant diversity (Bond 1995). Frugivory, on the other hand, might also be subject to influence by fragmentation, as frugivorous animals rely on fruiting plants for food, especially in times of general fruit scarcity. A loss of keystone resources critical for the maintenance of frugivore communities could result in a break-down of these communities and might push some dependent animal on the brink of extinction (Howe & Westley 1988). Removal of keystone species or change of
3 1. GENERAL INTRODUCTION ...... key processes such as seed dispersal could have wide-ranging effects on other species, processes and interactions (Meffe & Carroll 1994). Long-term consequences of both, loss of keystone seed dispersers and resources, might be widening circles of extinctions of mutually inter-dependent plants and animals following the disappearence of pivotal species (Futuyma 1973, May 1973, Howe 1977, Gilbert 1980). However, this scenario is mostly speculative and related empirical studies are still rare. It is therefore essential to look into mutualistic interactions such as seed dispersal and frugivory and investigate possible consequences of human-induced disturbance on the community and its dynamics.
1.4 Aims of thesis
In this thesis I investigated consequences of seed dispersal and frugivory for plant populations and bird communities in the dry subtropical forest of Oribi Gorge Nature Reserve (OGNR), KwaZulu - Natal, South Africa, from January 1997 to April 1999. One aim of the thesis was to determine the effect animals, especially birds, have on the dispersal of plants' seeds and consequently on seedlings and trees. A second aim was to understand patterns of seasonality in fruit availability and its use by the frugivore community. This thesis consists of four major chapters (chapters 2 to 5) which can be read independently. Each chapter is organized like a journal publication containing an introduction, followed by a method, result and discussion section and by a brief summary. The thesis closes with general conclusions. In chapter 2, 3 and 4, I focused on the exemplary seed dispersal system of typically bird-dispersed trees of the genus Commiphora (Burseraceae) and determined consequences of animal dispersal for plant populations. First, I studied the seed dispersal system of Commiphora harveyi in South Africa and compared results with those found for a related species, Commiphora guillaumini, in Madagascar (chapter 2). This comparative study was thought to be a "natural experiment" investigating the consequences of different frugivore diversity for seed dispersal with a high avian frugivore diversity in South Africa and a depauperate frugivore diversity in Madagascar. This study was conducted over one fruiting season. Second, I discussed not only influences of the frugivores for dispersal rates, but also
4 1. GENERAL INTRODUCTION ...... for seedling establishment, seedling distribution and the spatial pattern of trees extending the study over a second season (chapter 3). Third, results from the field study were evaluated by a computer model, simulating tree populations over time (chapter 4). I tested whether seed dispersal, especially offspring distribution and dispersal distance, and other factors such as breeding system and tree density had an influence on the formation of spatial patterns in tree populations. Finally, I investigated fruit abundance and availability in the same study site over 13 months, in order to determine annual periods of fruit scarcity and potentially important fruiting plants for the frugivore community during that period (chapter 5). The relative use of those plants by the local frugivore community was assessed and potential keystone resources identified.
5 6 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
2. Seed dispersal by birds in a South African and a Malagasy Commiphora species
2.1 Introduction
A large number of plant species rely on animals for the dispersal of their seeds. The percentage of animal-dispersed tree species can be as high as 90% in the tropics (Howe & Smallwood 1982). In spite of the widespread occurrence of animal-dispersed plants it is not known whether plants depend on particular animal species for the dispersal of their seeds. At one end of the interaction continuum, plant species can be dispersed by a multitude of fruit- eating animal species (Howe & DeSteven 1979, Coates-Estrada & Estrada 1988, Fleming & Williams 1990, Pizo 1997). Thus, the extirpation of one or a small number of animal species might not have consequences for seed dispersal because animal species are redundant and other species would fill the gap (Wheelwright & Orians 1982, Howe 1984b, Herrera 1985). At the other end, plant species might be dependant on only a few obligate dispersers whose absence could therefore reduce seed dispersal (Bond & Slingsby 1984). Hence it is important to study such dispersal systems for the eventuality of a disperser becoming extinct. Low rates of seed dispersal away from a parent tree might result in low rates of seedling establishment, as establishment and survival of seedlings is much lower under the crown than at a greater distance from the parent tree (Janzen et al. 1976, Augspurger 1983, 1984; Augspurger & Kelly 1984, Howe et al. 1985, Schupp 1988, Böhning-Gaese et al. 1999). Furthermore, seed dispersal could influence the abundance and chance of local extinction of a plant species (Howe 1977, 1984a) and affect the spatial distribution of trees (Hubbell 1979, Howe 1989). It is therefore important to ask whether regional differences in frugivore diversity, in the sense of species richness, have an effect on seed dispersal, and whether low frugivore diversity leads to low and potentially ineffective seed dispersal. One approach to answering these questions is to compare frugivory and seed dispersal in South Africa and in Madagascar. Although Africa and Madagascar are separated only by 400 km of open sea, the diversity of frugivorous birds differs considerably. In Africa, the diversity of frugivorous birds is high and comparable with other continents (Fleming et al.
7 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
1987). Madagascar, however, is "strikingly depauperate" (Fleming et al. 1987, Goodman & Ganzhorn 1997), and compared with Africa lacks a multitude of frugivorous taxa such as Bucerotidae, Coliidae, Musophagidae and Capitonidae (Langrand 1990, Maclean 1993). Ideally I wished to compare frugivory and seed dispersal in the same tree species in South Africa and Madagascar, but as 96% of the tree and shrub species in Madagascar are endemic (G.E.Schatz, pers. comm.) this was not possible. I therefore compared two species in the same genus (Commiphora, Burseraceae) with morphologically similar, typically bird- dispersed fruits. The questions asked were: 1. Does local diversity of frugivorous bird species differ between the South African and Malagasy study sites ? 2. Are there differences in tree visitation and seed dispersal rates between the South African and Malagasy Commiphora species ? 3. Does a low diversity of frugivorous birds in the Malagasy study site lead to low seed dispersal ?
2.2 The Trees
The study was conducted on Commiphora harveyi in South Africa and C. guillaumini, a Madagascar endemic. Both species are dioecious trees, the South African species flowering from October to December (Pooley 1994) and the Malagasy species from October to November (Rohner & Sorg 1986). The fruiting season of C. harveyi lasts from March to June (Pooley 1994, B. Bleher, pers. obs.), of C. guillaumini from January to April (Rohner & Sorg 1986). Both species bear roundish fruits that consist of an unpalatable greenish-reddish fleshy outer covering (exocarp and mesocarp) that splits into two halves exposing a single diaspore (Fig. 2.1). The diaspore consists of a brilliant black seed (South Africa: 7.1 x 5.7 x 5.1 mm, n = 20; Madagascar: 12.2 x 8.0 x 6.3 mm, n = 20) which is partly enveloped by a red fleshy aril- like endocarp (hereafter called an aril). Arils in C. harveyi are cup-shaped with four lobes (van der Walt 1986) (Fig. 2.1), aril shape in C. guillaumini is similar but lobes are absent (de la Bathie 1946) (Fig. 2.1). Before the fruit reaches complete maturity and splits open in C. harveyi, frugivorous tree visitors are able to open the outer coverings for access to seed and aril. In C. guillaumini, frugivores cannot open the outer coverings but have to wait until they split open by themselves and the diaspore is "displayed".
8 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
2.3 Study Sites
In South Africa, the study was conducted from February to July 1997 in Oribi Gorge Nature Reserve (OGNR) on the KwaZulu - Natal South Coast (Fig. 2.1). This 1850 ha nature reserve is located 110 km south of Durban, and 22 km inland from Port Shepstone, and is classified as coastal scarp forest (Cooper 1985). Average annual rainfall in the area is 1176 mm with the main rainfall season between October and March (Glen 1996). For further details on Oribi Gorge and coastal forests see Acocks (1988) and Glen (1996). In Madagascar, the study was carried out in February and March 1993 in a dry deciduous forest in western Madagascar (Fig. 2.1). The study site was the Kirindy Forest / CFPF, a 10 000 ha forestry concession of the Centre de Formation Professionelle Forestière de Morondava (CFPF) situated 60 km north of Morondava. Average rainfall in the study area is 799 mm with the main rainfall season between December and March (Sorg & Rohner 1996). Further information on the Kirindy Forest / CFPF is given in Ganzhorn & Sorg (1996) and Böhning-Gaese et al. (1995, 1996, 1999).
Fig. 2.1. Map of study sites (arrows) with respective fruits of the South African (Commiphora harveyi) and the Malagasy species (C. guillaumini). (f) whole fruit with outer coverings, (s) seed, (a) aril (redrawn after de la Bathie 1946 & van der Walt 1973).
9 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
2.4 Methods
2.4.1 Frugivore diversity
The number of frugivorous bird species was obtained for Oribi Gorge, South Africa from the unpublished bird list of the KwaZulu-Natal Nature Conservation Service, and for the Kirindy Forest / CFPF, Madagascar from Langrand (1990). For determining the degree of frugivory for each species Maclean (1993) and Langrand (1990) were used. Bird species were categorized as mainly frugivorous in both countries when fruit was the first item listed in the diet and as partly frugivorous when fruit was taken but was not the primary food source.
2.4.2 Tree observations
To obtain information about frugivorous tree visitors eight trees in South Africa and six trees in Madagascar were observed for two days each. Trees with comparatively large numbers of fruits were selected. However, this did not bias the results because the number of fruits did not affect the identity or the dispersal behavior of tree visitors (see results). Observations were conducted in three sessions from sunrise to 10:00 h, from 10:00 h to 14:30 h, and from 14:30 h to sunset (giving a total of 184 observation hours for C. harveyi and 156 hours for C. guillaumini). Observation sessions were randomly distributed over the study period. All visits by animals were recorded. For each visitor I noted the seed handling behavior and counted the number of seeds that were either dropped under the crown or dispersed away from the crown. The small number of other observations of animals only resting in the trees were removed from the data, so that analyses were limited to individuals that were observed foraging on arils. Night observations were conducted in four four-hour sessions in South Africa and in five three-hour sessions in Madagascar, using a night-vision scope.
2.4.3 Fruit traps
To obtain more accurate information on seed dispersal rates, fruit coverings and seeds were collected in fruit traps placed under the crowns of the same trees. In South Africa, five 1m² traps were placed at random positions under each tree. Traps were installed in February before fruit production had started and were removed in July after it ceased. They were monitored
10 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ...... three times a week. In Madagascar, large nets which covered the entire crown area were installed under the trees. Nets were open for four to five weeks from the middle of February until the end of March. They were installed after the first fruit coverings had dropped off the trees and were removed when there was still some fruit available. Nets were inspected four times daily. In both South Africa and Madagascar, the number of coverings and seeds were counted at each inspection. In addition to the seed dispersal data from tree observations, another estimate of the total percentage of seeds dispersed away from the crown could be obtained - as each fruit has two coverings and one seed - by calculating the difference between the number of seeds expected (half the number of coverings) and the number of seeds found in the nets. I observed neither rodents nor ants removing seeds from the traps.
2.5 Results
The number of frugivorous bird species recorded for the study sites in South Africa and Madagascar reflects the general pattern of avian frugivore diversity in the two countries. In Oribi Gorge, 6.2% (14 out of 226) of all recorded bird species are mainly, and 22.6% (51 out of 226) partly frugivorous. In the Kirindy Forest / CFPF, 3.5% (4 out of 114) of all observed bird species feed mainly, and 8.8% (10 out of 114) partly on fruits. In both countries, trees were visited primarily by birds. Birds constituted an average of 97.7% of all visitors in South Africa and 98.3% in Madagascar. I counted 25 ± 7 (if not otherwise noted mean ± 1 SE) visitors per tree per day to C. harveyi (range: 2.5 - 54.5, n = 8 trees), and 11 ± 4 to C. guillaumini (range: 4.0 - 25.5, n = 6; t-test: t = 1.5, df = 12, p = 0.16) (Fig. 2.2a), and the former was visited by more bird species than the latter. In South Africa, I recorded 13 bird and one primate species, with Crowned Hornbill (for scientific names see Table 2.1) being the most frequent visitor (Fig. 2.3a). Bird species ranged from 12 to 54 cm in body length. In Madagascar, four bird and one primate species were counted, with the Lesser Vasa Parrot being the most frequent visitor (Fig. 2.3a). Body length of bird species in Madagascar ranged from 10 to 54 cm. Most of the bird species are classified as mainly or partly frugivorous (Table 2.1). However, three of the bird species in South Africa and two in Madagascar are usually insectivores. I did not observe any nocturnal animals (e.g. fruit bats) feeding on arils.
11 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
60
a 50
40
30
# visitors 20
10
0 South Africa Madagascar b 100 80
60
40
20 % seeds handled 0 South Africa Madagascar c 80 60
40
20 % seeds dispersed 0 South Africa Madagascar
Fig. 2.2. Comparison of frugivory and seed dispersal in Commiphora harveyi (South Africa, n = 8 trees) and C. guillaumini (Madagascar, n = 6 trees). a. number of visitors per day and tree (mean ± 1 SE) (data from tree observations). b. total percentage of seeds handled per tree (mean ± 1 SE) (data from tree observations). c. total percentage of seeds per tree dispersed (mean ± 1 SE) (data from fruit traps).
The mean proportion of seeds per tree handled by visitors was similar in both countries, i.e. 87.1 ± 1.2% in South Africa (range: 80.3 - 91.1%, n = 8 trees), compared with 89.7 ± 3.3% in Madagascar (range: 74.2 - 95.9%, n = 6; t-test: t = -0.8, df = 12, p = 0.43) (Fig. 2.2b). However, whereas the total percentage of seeds dispersed away from the parent tree (as obtained from fruit trap data) was high in South Africa, with an average of 66.1 ± 5.2% (range: 43.6 - 81.4%, n = 8), the situation in Madagascar was completely different (Fig. 2.2c).
12 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ...... Although 89.7% of all seeds were handled (Fig. 2.2b), only 9.0 ± 2.6% were dispersed (range: -1.0 - 15.2%, n = 6; t-test: t = 8.8, df = 12, p < 0.0001) (Fig. 2.2c). The negative dispersal rate of -1.0% for one tree indicates that more seeds were "imported" than "exported". However, this number might be an artifact because fruit coverings and seeds did not drop simultaneously, and seed dispersal rates could not be calculated with absolute precision (Böhning-Gaese et al. 1995).
South Africa Madagascar a 40 80 30 60
20 40 % visits 10 20
0 0 ch rw st fw we or sb sm bu rt ti ba fd gt lv cj gv wv si 60 100
b 50 80 40 60 30 40 20 10 20 % seeds handled 0 0 ch rw st fw we or sb sm bu rt ti ba fd gt lv cj gv wv si 60 100 c 50 80 40 60 30 20 40 10 20
% seeds dispersed 0 0 ch rw st fw we or sb sm bu rt ti ba fd gt lv cj gv wv si
Fig. 2.3. Importance of different animal species in visiting trees, handling seeds, and dispersing seeds in Commiphora harveyi (South Africa, n = 8 trees) and C. guillaumini (Madagascar, n = 6 trees) (all data from tree observations). a. visitation rate by species per tree (mean ± 1 SE); b. seed handling rate by species per tree (mean ± 1 SE); c. seed dispersal rate by species per tree (mean ± 1 SE), e.g. Crowned Hornbills were responsible for 43% of all seeds dispersed. Grey bars: seeds dispersed by swallowing, shaded bars: seeds dispersed in the bill. In South Africa, Forest Weavers (fw) always, and Sombre Bulbuls (sb) in 16.7% of all dispersal events, dispersed seeds in the bill. Data are not normally distributed, however, as most medians show a value of zero, mean and standard error is used. For abbreviations of species names see Table 2.1.
13 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
Tab. 2.1. List of bird and primate species recorded as visitors to Commiphora harveyi (South Africa) and C. guillaumini (Madagascar) with their respective diet and seed handling behavior. For each country the order of the species reflects decreasing visitation rates (see Fig. 2.2a). Average body length data of bird species obtained for South Africa from Maclean (1993) and for Madagascar from Langrand (1990). Diet data obtained for the South African species from Lawes (1991) and Maclean (1993), and for the Malagasy species from Harcourt & Thornback (1990) and Langrand (1990). mf = mainly frugivorous, pf = partly frugivorous, i = insectivorous, a = swallows and disperses seed, b = disperses seed in bill, c = removes aril and drops seed under crown, d = not able to open fruit.
Tree visitor Scientific name Body Diet Hand- Abbr. length (cm) ling South Africa Birds: Crowned Hornbill Tockus alboterminatus 52 mf a ch Redbilled Woodhoopoe Phoeniculus purpureus 33 i a rw Blackbellied Starling Lamprotornis corruscus 20.5 mf a st Forest Weaver Ploceus bicolor 17.5 pf c, b fw Cape White-eye Zosterops pallidus 11.5 pf d we Blackheaded Oriole Oriolus larvatus 24.5 pf a or Sombre Bulbul Andropadus importunus 21 pf a, b sb Blackeyed Bulbul Pycnonotus barbatus 21 mf a bu Redfronted Tinker Barbet Pogoniulus pusillus 12 mf a rt Southern Black Tit Parus niger 15.5 i c ti Blackcollared Barbet Lybius torquatus 19.5 mf a ba Forktailed Drongo Dicrurus adimilis 23.8 i a fd Goldenrumped Tinker Barbet Pogoniulus bilineatus 12 mf a gt Primates: Samango Monkey Cercopithecus mitis - mf a sm
Madagascar Birds: Lesser Vasa Parrot Coracopsis nigra 35 mf c, b lv Common Jery Neomixis tenella 10 i c cj Greater Vasa Parrot Coracopsis vasa 50 mf c gv Whiteheaded Vanga Leptopterus viridis 20 i c wv Primates: Sifaka Propithecus verreauxi - mf c si
14 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
The reasons for the differences in dispersal rates were the different seed handling behavior and dispersal rates of individual animal species (as obtained from tree observation data) in South Africa and Madagascar (Fig. 2.3b, c, Table 2.1). In South Africa, a total of ten bird species dispersed seeds by swallowing them and therefore carrying them away from the tree (Table 2.1). Seed dispersal rates per species, i.e. seed dispersal rates with respect to each species contributing to the dispersal of seeds, were highest for the Crowned Hornbill (43.1%, Fig. 2.3c). The three most important dispersers, the Crowned Hornbill, Redbilled Wood- hoopoe and Redfronted Tinker Barbet, dispersed 75% of all seeds, the other species dispersed less than 10% of the seeds each. Two species in South Africa, the Forest Weaver and Southern Black Tit, removed the aril and dropped the seed under the crown, therefore not contributing to dispersal (Table 2.1). The Sombre Bulbul and Forest Weaver were occasionally observed dispersing seeds by carrying them away in their bills, but this did not represent their main handling method (Table 2.1). The Cape White-eye visited the trees but did not manage to open the fruits (Table 2.1). All species recorded in the Malagasy trees stripped the arils off the seeds and usually dropped the seeds under the crown (Table 2.1). Seed dispersal took only place when Lesser Vasa Parrots took off from a tree with a seed still in their bills (on 47% of their visits). Therefore, seed dispersal rates were highest for the Lesser Vasa Parrot (100%, Fig. 2.3c), the only species observed dispersing seeds. Average number of fruits produced per tree was larger in C. harveyi than in C. guillaumini (C. harveyi: median 3774 fruits per tree, range 1200 - 15231, n = 8 trees; C. guillaumini: median 678, range 345 - 4591, n = 6 trees; Wilcoxon test: z = -2.52, p = 0.0118). I tested whether fruit production had an influence on the number of visitors and the total percentage of seeds handled and dispersed. This was not the case, neither for South Africa nor for Madagascar (p > 0.6). Furthermore, fruit production did not affect the number of species visiting the trees, handling seeds and dispersing seeds (p > 0.7). As these results could be due to small sample size, I tested whether differences in the total percentage of dispersed seeds were influenced by fruit production and country (Fig. 2.4). Only country contributed significantly to the differences in the percentage of dispersed seeds (ANCOVA: F2,11 = 43.91, 2 p < 0.0001, R = 88.9, country: F1,11 = 39.40, p < 0.0001, log [fruit production]: F1,11 = 2.28, p = 0.16).
15 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
100
80
60
40
20 % seeds dispersed 0
2 3 4 5 log (fruit production)
Fig. 2.4. Total percentage of seeds dispersed (as obtained from fruit trap data) as a function of (log) fruit production in Commiphora harveyi (South Africa, n = 8 trees, l) and Commiphora guillaumini (Madagascar, n = 6 trees, ¡).
2.6 Discussion
The comparison of frugivory and seed dispersal by birds in South Africa and Madagascar demonstrated that regional differences in frugivore diversity, and especially in seed handling behavior significantly affected seed dispersal. As might be expected from the higher diversity of frugivorous birds in South Africa, the South African species C. harveyi was visited by 13, the Malagasy species C. guillaumini by only four bird species. Furthermore, 10 of the South African but none of the Malagasy species dispersed seeds by swallowing them. Consequently, seed dispersal rates in Madagascar were 7.4 times lower than in South Africa. The low dispersal rates of the Malagasy species were caused by a local lack of "legitimate" dispersers that swallow seeds and disperse them effectively away from the parent tree. Thus, the combination of frugivore diversity, which provides the plant with a certain variety of possible dispersers, and efficient handling is crucial for seed dispersal. One might argue that only the effectiveness of seed removal but not of seed dispersal can be addressed in the present study as the fate of the removed seeds is not known. Several
16 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ...... factors influence seed dispersal effectiveness after removal of seeds from the parent tree. One factor is that seeds being swallowed have to survive the passage through frugivore guts. In C. harveyi, the main disperser was the Crowned Hornbill. In general, hornbills are known to provide high-quality dispersal for many trees in African and Asian forests (see Kemp 1995). Their effectiveness as seed dispersers comes from the fact that they move seeds away from the parent trees and that most seeds are not damaged during ingestion (Leighton & Leighton 1983, Becker & Wong 1985, Gautier-Hion et al. 1985, Whitney et al. 1998). In the Malagasy Commiphora species, the Lesser Vasa Parrot as main disperser did not swallow seeds but dropped them to the ground after aril removal. Seeds were not damaged despite the fact that, in general, parrots are known rather to be seed predators than seed dispersers (Higgins 1979, Coates-Estrada et al. 1993). Therefore, in both countries the main disperser of Commiphora seeds very probably did not have a negative impact on the fate of the seeds following their removal from the parent tree. What are possible alternative explanations, besides frugivore diversity, for the differences in tree visitation and seed dispersal rates between South Africa and Madagascar ? Important variables influencing the composition of avian frugivore assemblages at fruiting plants are, e.g. fruit structure, size and abundance (Pratt & Stiles 1985). Fruit structure is very similar in both species, although there are differences in diaspore size. Smaller diaspores attract more species of birds, and have a higher probability of being dispersed away from the crown because they can be swallowed by birds with a smaller gape width (Wheelwright 1985, 1993; Jordano 1987a, b; Levey 1987). Diaspore size in Commiphora is mainly determined by seed size because the aril covers only part of the seed. The South African species has smaller seeds than the Malagasy species (average maximum diameter South African species: 5.7 mm, Malagasy species: 8.0 mm, Fig. 2.1). However, these differences in seed size are relatively small, and seeds of both species range in the lower end of the fruit size distribution for frugivores as recorded by, e.g. Wheelwright (1985: Fig. 1) or Pratt & Stiles (1985). Furthermore, observations made at the same study site in a second study year during the same season demonstrated that the somewhat larger seeds of the Malagasy Commiphora species were readily swallowed by the Madagascar Bulbul Hypsipetes madagascariensis and Crested Drongo Dicrurus forficatus (Böhning-Gaese et al. 1999). However, as these species are uncommon at the study site and dispersal events by them extremely rare, they are rather unimportant seed dispersers.
17 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ......
Another factor influencing dispersal rates might be differences between the two tree species in South Africa and Madagascar in the mean number of fruits produced. However, for both species the results show that this was not the case and that the total percentage of dispersed seeds, as well as the number of species visiting trees, handling seeds and dispersing seeds, did not depend on fruit production. These results are consistent with other studies that often reported little or no influence by the number of fruits produced on the percentage of dispersed seeds (Howe & Smallwood 1982, Jordano 1987a, Laska & Stiles 1994). Thus, fruit characteristics like structure, size, and abundance do not appear to be the reason for the profound differences in seed dispersal rates found between the South African and Malagasy Commiphora species. However, as only one species in each country was studied, these results apply only to the specific study site and tree species. To be able to make generalizations, further studies on other species have to be carried out. Seed dispersal rates of 66% as found for Commiphora harveyi are comparable with other tropical tree species with similar phenology and fruit morphology. For example, seed dispersal rates were 91% for Casearia corymbosa in Costa Rica (Howe & Vande Kerckhove 1979), 76% for Virola sebifera in Panama (Howe 1981), 66% for Virola surinamensis in Panama (Howe & Vande Kerckhove 1981), 45% for Cymbopetalum baillonii in Mexico (Coates-Estrada & Estrada 1988), and 70-97% for Bursera simaruba in Mexico (Greenberg et al. 1995). Seeds of the South African Commiphora were dispersed by ten bird species, Casearia corymbosa by 12 (Howe & Vande Kerckhove 1979), Virola sebifera by six (Howe 1981), Virola surinamensis by seven (Howe & Vande Kerckhove 1981), Cymbopetalum baillonii by 20 (Coates-Estrada & Estrada 1988), and Bursera simaruba by ten bird species (Greenberg et al. 1995). By comparison, seed dispersal rates of 9.0% recorded for the Malagasy C. guillaumini are extremely low, and its seeds were dispersed only by one species (see Casearia corymbosa, Howe 1977). To my knowledge the only other published studies of frugivores visiting fruiting trees on Madagascar are those by Scharfe & Schlund (1996), Goodman et al. (1997) and Dew & Wright (1998). All three studies show that lemurs play a prominent role in seed dispersal in Madagascar. Fruits of a Ficus species were eaten by four bird and four lemur species (Goodman et al. 1997), fruits of Poupartia sylvatica by two bird and four lemur species and fruits of Berchemia discolor by six lemur species (Scharfe & Schlund 1996). The island is, in
18 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ...... general, depauperate in frugivorous birds in comparison with other tropical areas (Langrand 1990, Goodman & Ganzhorn 1997, Goodman et al. 1997). This does not appear to result from the extinction of frugivorous bird species either in historical or in paleontological time (Langrand 1990, Goodman & Rakotozafy 1997). However, it can be linked to a lower general fruit availability and a reduced diversity and density of Ficus species, which is often an important keystone group for numerous tropical frugivores (Goodman & Ganzhorn 1997). To conclude, the studies in South Africa and Madagascar demonstrated that low diversity of fruit-eating and seed-dispersing animal species can have profound consequences for seed dispersal. The low number of frugivorous bird species in Madagascar, and particularly the local absence of birds that swallow seeds, led to extremely low rates of seed dispersal. Other studies also confirm that a reduced frugivore community, e.g. through forest fragmentation, can lead to a reduced number of dispersers (Howe 1984a, Pizo 1997). Furthermore, low rates of seed dispersal can result in low rates of seedling establishment. Seeds of the Malagasy Commiphora species that were dropped under the crown had a 36 times lower probability of getting established as seedlings than seeds dispersed away from the crown (Böhning-Gaese et al. 1999). It remains to be seen if low rates of seedling establishment have consequences for the abundance and probability of extinction of a tree (Bond 1995). However, there is some evidence to indicate that dispersal-generated patterns of seedling establishment can have an influence on the spatial distribution of adult trees (Janzen 1970, Connell 1971, Hubbell 1979).
2.7 Summary
The diversity of fruit-eating bird species in Africa is high and comparable to other continents; however it is strikingly depauperate in Madagascar. To address the question of whether regional differences in the diversity of frugivorous bird species have an influence on seed dispersal I compared the tree visitation and seed dispersal rates of a South African and a Malagasy species of Commiphora (Burseraceae). In keeping with the higher diversity of frugivorous birds on the African continent, the South African Commiphora species was visited by 13, the Malagasy species by only four bird species. In each of the two countries only
19 2. SEED DISPERSAL BY BIRDS IN A SOUTH AFRICAN AND A MALAGASY COMMIPHORA SPECIES ...... one primate species visited the trees. Ten of the South African but none of the Malagasy bird species dispersed seeds by swallowing them. Other species either dropped the seed under the parent tree or occasionally carried the seeds away in their bills. Consequently, the percentage of seeds dispersed away from the crown in South Africa was 66% and in Madagascar 9%. The results demonstrate that regional differences in frugivore diversity, and especially in seed handling behavior, can lead to pronounced differences in seed dispersal away from the parent tree.
20 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
3. Consequences of frugivore diversity for seed dispersal, seedling establishment and the spatial pattern of seedlings and trees
3.1 Introduction
Especially in the tropics up to 90% of tree species rely on frugivorous animals for the dispersal of their seeds (Howe & Smallwood 1982). Although many studies exist on frugivory and seed dispersal (reviews by Howe & Smallwood 1982, Willson 1992), little research on the consequences of seed dispersal by frugivorous animals on seedling establishment and on the spatial distribution of seedlings and trees has been conducted (but see Fleming & Heithaus 1981). For animal-dispersed plant species the number of dispersing frugivores could be a major factor for the functioning and non-functioning of their dispersal. If plant species are dispersed by a large variety of frugivorous animal species (see e.g. Howe & De Steven 1979, Coates-Estrada & Estrada 1988, Fleming & Williams 1990), loss of only a few frugivores might not have consequences for seed dispersal as other frugivores could fill the gap (Wheelwright & Orians 1982, Howe 1984a, Herrera 1985). However, if plant species depend on only a few frugivorous animals, their lack could lead to a failure in seed-dispersal (Bond & Slingsby 1984, Böhning-Gaese et al. 1999). Declines in frugivore diversity due to habitat fragmentation have been reported by Pizo (1997) and Santos & Telleria (1997). Disrupted seed dispersal might reduce seedling establishment as it is reported to be lower under the parent tree as compared to greater distances from the tree (Janzen et al. 1976,
Augspurger 1983, Howe et al. 1985, Schupp 1988). Consequently, frugivore reduction could affect the abundance and extinction probability of trees (Bond 1995). Furthermore, changes in the spatial pattern of seedling establishment might have consequences for the spatial distribution of trees (Janzen 1970, Connell 1971, Hubbell 1979). Thus, frugivore diversity might have severe consequences for the regeneration potential, the dynamics and the floristic composition of forests. It is therefore important for the conservation of forest ecosystems to investigate whether frugivore diversity indeed has an influence on plant populations.
21 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
Ideally I would like to manipulate experimentally frugivore diversity to study the long- term effects on seedling establishment and the spatial distribution of seedlings and trees. However, this is not possible especially when studying tree establishment. Therefore, a "natural experiment" was used by comparing seed dispersal of two tree species in two areas with different degree of frugivore diversity, Africa and Madagascar. The diversity of frugivorous birds is markedly different in both countries with a high diversity in Africa and a depauperate frugivore community in Madagascar (Goodman & Ganzhorn 1997, Goodman et al. 1997), where important frugivorous bird groups like e.g. hornbills, turakos, and barbets are missing (Fleming et al. 1987, Langrand 1990). This general pattern of avian frugivore diversity in both areas is reflected by the avian frugivore community at our study sites, Oribi Gorge Nature Reserve in South Africa and the Kirindy Forest / CFPF in Madagascar. In Oribi Gorge 14 (of a total of 226) bird species are mainly frugivorous, whereas at the Kirindy Forest / CFPF only four bird species (of 114) feed mainly on fruit (see chapter 2). Originally, the study intended to investigate the same tree species with typically bird-dispersed fruits in both sites in South Africa and Madagascar. However, since 96% of the tree and shrub species in Madagascar are endemic (G. E. Schatz, pers. comm.), I studied two species of the same genus, Commiphora (Burseraceae), and asked whether regional differences in frugivore diversity can be linked to seed dispersal, seedling establishment and the spatial distribution of seedlings and trees.
3.2 Study Sites and Species
3.2.1 Study Sites
The studies were carried out at the KwaZulu / Natal South Coast in South Africa and in western Madagascar (Fig. 3.1). In South Africa, the study site was Oribi Gorge NR, a 1850 ha Nature Reserve (30°43 S, 30°15‘ E), located 110 km south of Durban and 21 km inland from Port Shepstone. The Reserve is situated in a 280 m deep gorge whose steep slopes are covered with coastal scarp forest (Cooper 1985); it is characterised by subtropical climate with an average rainfall of 1176 mm (Glen 1996). The study site in Madagascar was the Kirindy
22 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
Forest / CFPF (20°03’ S, 44°39’ E), a 10 000 ha forestry concession of the Centre Professionelle Forestière de Morondava (CFPF), located 60 km north of Morondava. It is characterised by tropical climate with an average annual rainfall of 730 mm (Sorg & Rohner 1996) and the vegetation is classified as dry deciduous forest. The studies in South Africa were carried out from March to June 1997, 1998 and in March 1999, in Madagascar in February and March 1993, 1994 and 1995. Further information on Oribi Gorge is given in Acocks (1988) and Glen (1996) and on the Kirindy Forest in Ganzhorn & Sorg (1996) and Böhning-Gaese et al. (1995, 1996, 1999).
Fig. 3.1. Map of study sites (arrows), study tree species and their corresponding fruits in South Africa and Madagascar. (f) whole fruit with outer coverings, (s) seed, (a) aril (redrawn after de la Bathie 1946 and van der Walt 1973).
3.2.2 Study Species
I studied two species of the same genus Commiphora, Commiphora harveyi in South Africa and Commiphora guillaumini in Madagascar (Fam. Burseraceae). Both species make common canopy trees in the respective study sites. They are dioecious, with the South African
23 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ...... species flowering from October to December and fruiting fom March to June, with a median number of 1879 produced fruits per tree (range = 324-15231, n = 15 trees). The Malagasy species flowers from October to November and bears fruits from January to April with a median number of 747 fruits per tree (range = 345-4591, n = 12 trees). Both species have morphologically similar, typically bird-dispersed roundish fruits consisting of an unpalatable greenish-reddish fleshy outer covering (mesocarp and exocarp) that splits into two halves to expose a single diaspore (Fig. 3.1). The diaspore consists of a brilliant black seed (South Africa: 7.1 x 5.7 x 5.1 mm, n = 20; Madagascar: 12.2 x 8.0 x 6.3 mm, n = 20) which is partly enveloped by a red fleshy aril-like endocarp (hereafter called an aril, Fig. 3.1). Arils in the
South African species are cup-shaped with four lobes (van der Walt 1986), in Madagascar lobes are absent (de la Bathie 1946) (Fig. 3.1). In South Africa, frugivorous tree visitors are able to open the outer coverings to gain access to seed and aril before the fruits reach maturity and split open by themselves. In the Malagasy species, frugivores cannot open the outer coverings but have to wait until they split and the diaspore is displayed. Seed germination in the South African species takes place from September to November the same year (germination trials and field observations), in Madagascar from October to January.
3.3 Methods
3.3.1 Seed dispersal
3.3.1.1 Fruit traps For information on seed dispersal rates, fruit traps were established under tree crowns of eight trees in 1997 and seven trees in 1998 in South Africa and under six trees each year 1994 and 1995 in Madagascar. Different tree individuals were used in the two study years. The chosen trees had on average higher fruit crops than the other trees in the areas, however, this did not bias the results (see chapter 2). In South Africa five 1 m² traps were installed at random positions under each tree, in Madagascar large nets were installed which covered the entire crown area. Traps in South Africa were monitored three times a week, in
24 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
Madagascar four times daily, and the number of coverings and seeds were counted at each inspection. The percentage of fruits handled was calculated as the difference between number of seeds expected (half the number of coverings) and the number of seeds with untouched aril found in the traps. The percentage of seeds dispersed was obtained by calculating the difference between number of seeds expected (half the number of coverings) and the number of seeds found in the nets. Dispersal rates were pooled for the two study years for South Africa and Madagascar as there were no differences between the years.
3.3.1.2 Tree observations For information on frugivorous tree visitors, observations were carried out on the same trees that were used for fruit traps in South Africa and Madagascar (total number of hours: South Africa 1997: 184 h, 1998: 161 h, Madagascar 1994: 156 h, 1995: 156 h). Observations were conducted in three blocks from sunrise to 10:00 h, from 10:00 h to 14:30 h, and from 14:30 h to sunset. Each tree was observed for two days with observation blocks randomly distributed over the entire study period. All tree visitors were recorded and data on the seed handling behavior and the number of seeds dropped under the crown or dispersed away from the parent tree noted. Night observations were conducted in 4 x 4 h-sessions in South Africa and in 5 x 3 h-sessions in Madagascar using a night vision scope. Data on visitation, handling and dispersal rates of species and number of visitors were pooled for the two study years, both for South Africa and Madagascar.
3.3.2 Seedling establishment, dispersal benefit and seedling distribution
3.3.2.1 Seedling establishment For information on seedling establishment, seedlings were mapped along transects in South Africa and Madagascar. Transects were run through the areas where observation trees were found. Seedlings below a height of 40 cm were mapped 1 m to the right and 1 m to the left of the transects resulting in a mapped area of 2600 m2 in South Africa and of 6875 m2 in Madagascar each season. Mapping was done in the year following the tree observations, just before the onset of the new fruiting season in South Africa (February to March 1998 and
25 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
1999) and during the fruiting season in Madagascar (Feburary to March 1994 and 1995). In South Africa, newly established seedlings (i. e. first-year seedlings) were clearly identifiable by two typical cotyledons and non-woody stems when compared to seedlings older than one year with pinnate leaves and woody stems. In Madagascar, woodiness of seedling stems was noted only in the second study year for the last 76% of the seedlings (130 out of 170 seedlings). For South Africa, data on number of mapped seedlings were pooled for the two study years as there were no differences between years. For Madagascar, pooling of data was only possible for number of all seedlings. When calculating numbers of first-year seedlings and older seedlings separately only second year data were available. As a measure of survival probability of first-year seedlings to an older seedling stage the formula (number of seedlings older than one year) / (number of seedlings older than one year + number of first-year seedlings) was used (Ricklefs 1997).
3.3.2.2 Benefit of seed dispersal The increase in probability of getting established away from the parent tree due to dispersal was called dispersal benefit. To quantify the benefit of animal dispersal on seedling establishment I compared the number of seeds and the number of seedlings of C. harveyi in South Africa and C. guillaumini in Madagascar, under and away from the trees. For the calculation of dispersal benefit in South Africa see Fig. 3.2 with x as the total number of seeds produced by a tree, b as the proportion of seeds per tree dispersed by animals and a as the proportion of seeds dropped under the tree (=1 - b) (a and b from fruit traps). The number of seeds under the crown is cexp = a * x and away from the trees dexp = b * x. If there is no dispersal benefit, then the expected distribution of seedlings is reflected by the distribution of seeds. cobs is the actual observed number of seedlings under crowns and dobs the one of seedlings away from the trees (taken from the data described in 3.3.2.3 Seedling distribution). The proportion of seeds falling under the tree and getting established as new seedling can then be calculated as e = cobs / cexp, the proportion of seeds getting established away from the tree as f = dobs / dexp. Consequently, the benefit of seed dispersal as the increase in probability of getting established away from parent trees can be calculated as the ratio of both products, i.e. f
/ e = (a * dobs) / (b * cobs) (Fig. 3.2). Note, that the total number of seeds x does not matter for the calculation of dispersal benefit. The computation of dispersal benefit in Madagascar varies
26 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ...... slightly as secondary dispersers contribute to benefit as well (see Böhning-Gaese et al. 1999). Dispersal benefit was calculated for first-year seedlings found in South Africa and Madagascar. For South Africa, data were pooled for both study years. For Madagascar, benefit could only be calculated for the second study year as no data on age of seedlings were available for the first year (see 3.3.2.1 Seedling establishment).
Fig. 3.2. Flow chart of seed dispersal process and calculation of dispersal benefit for Commiphora harveyi (South Africa). Starting with x seeds on a tree, an average proportion of b seeds is dispersed away from the tree, whereas a proportion of a seeds is dropped under the crown, leading to an expected number of dexp = b*x seedlings away and cexp = a*x under the tree. cobs is the number of seedlings actually observed under trees and dobs the number found away from trees. Consequently e = cobs / cexp is the percentage of seeds that get established as seedlings under and f = dobs / dexp away from the trees. The dispersal benefit, i.e. the increase in probability to get established is the ratio f / e = (a * dobs) / (b * cobs).
3.3.2.3 Seedling distribution For information on seedling distribution, for each seedling found the distance to the next female Commiphora crown was measured in both study sites. For South Africa, data were
27 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ...... pooled for both study years for the distribution of all seedlings, first-year seedlings and seedlings older than one year. For Madagascar, pooling of data was only possible for the distribution of all seedlings. For distribution of first-year seedlings and seedlings older than one year, only second year data were available.
3.3.3 Spatial distribution of trees
3.3.3.1 Field data For information on the spatial distribution of Commiphora trees in the two study sites, the T- Square method was used (Ludwig & Reynolds 1988). For this method, first, the distance of a random point (r) to the nearest Commiphora tree (t1) was measured. Then, a line perpendicular to the line rt1 at t1 was drawn, and the distance from tree t1 to its nearest neighbour t2 beyond the “half-plane“ created by this perpendicular was measured. The spatial distribution of trees could be determined by the distribution of the values of the two distances rt1 and t1t2 and by calculating an index C derived from the ratio of the squared distances rt1 and t1t2. The value of C ranges from 0 to 1 with 0.5 for random patterns, < 0.5 for uniform patterns and > 0.5 for clumped patterns (Ludwig & Reynolds 1988). A test statistics z is used to test significant departure from random pattern. 40 random points were used along a transect system running through the study site in South Africa and 38 in Madagascar.
3.3.3.2 Computer simulation To test whether seed dispersal influences the spatial pattern of tree populations, an individual- based simulation model was developed under Borland Delphi 4 (1998) to simulate seed dispersal of trees based on empirical evidence from field data. In the simulation a virtual population of 300 trees (crown diameter of trees 6 m) was distributed randomly in a virtual 25 ha plot ressembling the field situation between South Africa and Madagascar (tree density: 12 trees per hectare). Female trees of a dioecious population reached an age of 20 to 70 simulation time steps (in the following years) and were then replaced by their reproductively active offspring, male trees died without being replaced. The number of offspring for each female tree was drawn from an exponential distribution with a mean of 2.2 (De Steven 1994).
28 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
Dispersal distance as the distance between female tree and offspring was drawn from a log- normal distribution (ressembling the seedling distribution in South Africa) or from a negative exponential distribution (ressembling the seedling distribution in Madagascar). Mean dispersal distances for each simulation ranged from 0 m (seeds falling under the parent tree) to 50 m. For each combination of parameters 30 replications were carried out. After a simulation duration of 300 years the spatial pattern of the final tree populations was quantified using the T-Square method with 40 random points.
3.4 Results
3.4.1 Seed dispersal
In both countries mainly birds were observed visiting the trees (South Africa: 96.4% ± 7.4% of the visitors; if not otherwise noted mean ± 1 SD, range = 77.9%-100%, n = 15 trees; Madagascar: 99.1% ± 2.3% of the visitors, range = 92.2%-100%, n = 12 trees). The South African tree species was visited by an average of 19 ± 17 individual animals per tree per day (range = 1.5-54.5, n = 15 trees) and the Malagasy species by 14 ± 7 tree visitors per tree per day (range = 4.0-25.5, n = 12 trees). I found no significant differences in the number of visitors between countries (t-test: t = 0.9, df = 25, p = 0.36). In terms of species number, the South African Commiphora trees were visited by more species than the Malagasy ones (Tab. 3.1). In South Africa I documented 15 bird and one primate species, compared to only six bird and one lemur species in Madagascar. All visiting species, both in South Africa and Madagascar, were diurnal. The most frequent tree visitors in South Africa were the Redbilled Woodhoopoe (27.0%) (for Latin names see Tab. 3.1) and the Crowned Hornbill (22.8%) and in Madagascar the Lesser Vasa Parrot (74.6%, Tab. 3.1). The mean proportion of fruits per tree which was handled by visitors was similar in both countries with a handling rate of 90.1% ± 4.5% in South Africa (range = 80.3%-96.8%, n = 15 trees) and 93.2% ± 6.8% in Madagascar (range = 74.2%-97.9%, n = 12 trees). However, whereas handling rates were similar in both countries (t-test: t = -1.4, df = 25, p = 0.17), seed dispersal rates were not. The total percentage of seeds dispersed away from the
29 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ...... parent tree was high in South Africa (Fig. 3.3), with an average of 70.8% ± 13.4% (range = 43.6%-85.7%, n = 15 trees). Dispersal rate in Madagascar was extremely low (Fig. 3.3) with only 7.9% ± 5.6% of the seeds dispersed (range = -1.0%-16.3%, n = 12 trees). The difference in dispersal rates between the two countries was highly significant (t-test: t = 15.1, df = 25, p < 0.0001).
Tab. 3.1. List of visitors in Commiphora harveyi (South Africa) and C. guillaumini (Madagascar) trees with their seed handling behavior, their mean percentages of visits per tree per day and their mean percentages of dispersed seeds per tree, e.g. Redbilled Woodhoopoes were responsible for 29.6% of all seeds dispersed. Visitation and dispersal rates for species were calculated from tree observations. a = seed swallowed and dispersed, b = seed dispersed in beak, c = aril removed and seed dropped under the crown.
Tree visitor hand- % of visits % seeds disp. ling per tree per tree South Africa Redbilled Woodhoopoe Phoeniculus purpureus a 27.0 29.6 Crowned Hornbill Tockus alboterminatus a 22.8 34.0 Blackbellied Starling Lamprotornis corruscus a 7.0 6.0 Blackcollared Barbet Lybius torquatus a 6.7 10.1 Redwinged Starling Onychognathus morio a 6.2 0.3 Cape White-eye Zosterops pallidus c 5.6 - Forest Weaver Ploceus bicolor b, c 5.5 0.1 Blackeyed Bulbul Pycnonotus barbatus a 3.8 3.5 Samango Monkey Cercopithecus mitis a 3.6 5.1 Blackheaded Oriole Oriolus larvatus a 3.4 1.2 Redfronted Tinker Barbet Pogoniulus pusillus a 2.1 - Goldenrumped Tinker Barbet Pogoniulus bilineatus a 1.8 1.1 Sombre Bulbul Andropadus importunus a, b 1.7 0.7 Grey Coocooshrike Coracina caesia ? 1.4 - Southern Black Tit Parus niger c 0.8 - Forktailed Drongo Dicrurus adsimilis a 0.2 0.03 Total number of species 16 12
Madagascar Lesser Vasa Parrot Coracopsis nigra b, c 74.6 87.1 Common Jery Neomixis tenella c 19.4 - Greater Vasa Parrot Coracopsis vasa c 3.1 - Whiteheaded Vanga Leptopterus viridis c 1.4 - Sifaka Propithecus verreauxi c 0.9 - Crested Drongo Dicrurus forficatus a 0.4 10.4 Madagascar Bulbul Hypsipetes madagascariensis a 0.3 2.5 Total number of species 7 3
30 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
100
80
60
40
20 % dispersed seeds
0 South Africa Madagascar
Fig. 3.3. Comparison of total percentage of seeds dispersed per tree (mean ± 1 SE) between Commiphora harveyi (South Africa, n = 15 trees) and C. guillaumini (Madagascar, n = 12 trees) (data obtained from fruit traps).
These differences in total dispersal rates between the two countries were due to differences in frugivore diversity and the seed handling behavior of individual frugivorous species. In South Africa, altogether 11 bird and one primate species dispersed seeds by swallowing them (Tab. 3.1). Three bird species removed the aril and dropped the seed under the crown, i.e. the Forest Weaver, the Southern Black Tit and the Cape White-eye (Tab. 3.1). Two bird species (Forest Weaver and Sombre Bulbul) occasionally dispersed seeds by carrying them away in their beaks, however this did not represent their main handling method (Tab. 3.1). Seed dispersal rates were highest for the Crowned Hornbill (34.0%) and the Redbilled Woodhoopoe (29.6%) (Tab. 3.1). More than 70% of all seeds were dispersed by the Crowned Hornbill, Redbilled Woodhoopoe and Blackcollared Barbet. In Madagascar, four bird species and one lemur species stripped the arils off the seed and dropped the seed under the crown therefore not contributing to seed dispersal (Tab. 3.1). Two bird species, the Crested Drongo and the Madagascar Bulbul, dispersed seeds by swallowing them. However, these species were very rare visitors. Thus, the main portion of seeds (87.1%) was dispersed by the Lesser Vasa Parrot by occasionally carrying the seeds out of the crowns in its beak (Tab. 3.1).
31 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
The average crop size per tree was larger in the South African tree species than in the
Malagasy one independent of the study year (Wilcoxon test: z = -3.0, p = 0.0027, nSouth Africa =
15 trees, nMadagascar = 12 trees). Possibly the larger crop size in South Africa caused the differences in visitation and dispersal rates between the two countries. An ANCOVA testing the effect of crop size and country revealed a crop size effect for the number of visitors (F1,26 =
9.02, p = 0.0062), the number of visiting species (F1,26 = 8.74, p = 0.0069) and the number of dispersing species (F1,26 = 8.04, p = 0.0092). A country effect was only found for the number of dispersing species (F1,26 = 7.53, p = 0.0113). For the percentage of seeds dispersed, crop size did not contribute to the differences (F1,26 = 1.53, p = 0.23), however there was a strong country effect (F1,26 = 176.26, p = 0.0001).
3.4.2 Seedling establishment, dispersal benefit and seedling distribution
3.4.2.1 Seedling establishment
In South Africa, a total of 138 seedlings were mapped in the two years (n1.year = 61, n2. year = 77) corresponding to 265 seedlings per hectare per year (Fig. 3.4). Of these, 64% were newly established first-year seedlings. In Madagascar, I found a total of 266 seedlings in the two years (n1.year = 96, n2. year = 170). Of the 246 seedlings per hectare in the second year 85% were first-year seedlings (Fig. 3.4). Number of total seedlings found per hectare and year was similar in both countries, however, when comparing numbers of older seedlings, Madagascar showed a much lower number (Chi2 = 16.8, df = 1, p < 0.0001). Survival probability of first- year seedlings was 36 % in South Africa, but only 15 % in Madagascar which indicates lower mortality of first-year seedlings in South Africa.
32 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
300 all seedlings 250 first-year seedlings seedlings > 1 year 200
150
100 # seedlings / ha
50
0 South Africa Madagascar
Fig. 3.4. Number of total seedlings, first-year seedlings and seedlings older than one year found per hectare in the South African and Malagasy study site. For South Africa, data are pooled over two years, for Madagascar, data of only the second year were used for first-year seedlings and seedlings older than one year.
3.4.2.2 Benefit of seed dispersal To calculate seed dispersal benefit I compared the expected number of first-year seedlings and the actually observed number of first-year seedlings under and away from Commiphora trees.
In South Africa, I found three first-year-seedlings (cobs) under and 41 first-year seedlings away from the trees (dobs). Seeds getting dispersed had a 5.6 times higher probability of getting established as seedlings than seeds under parent trees (for calculation see 3.3.2.2). In
Madagascar, 29 first-year-seedlings (cobs) were found under and two first-year seedlings away from the trees (dobs). The respective dispersal benefit was 79.9.
3.4.2.3 Seedling distribution Seedling distributions differed markedly between South Africa and Madagascar and reflected the pattern of seed dispersal in both countries (Fig. 3.5). In the South African study site, the majority of all seedlings was found away from the crown of the closest female Commiphora tree with a median distance of 21.1 m (q1 = 6.4 m, q3 = 25 m, n = 138 seedlings pooling both study years) (Fig. 3.5a). In Madagascar, most seedlings were found in close vicinity to a female Commiphora crown with a median distance of 0.9 m (q1 = 0 m, q3 = 2.9 m, n = 226
33 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ...... seedlings pooling both study years) (Fig. 3.5b). The difference in the median distance between the two countries was highly significant (Wilcoxon test: z = 12.1, p < 0.0001, nSouth Africa =
138, nMadagascar = 226). When investigating first-year seedlings and seedling older than one year separately (Fig. 3.5c, d), I found similar trends with seedlings in South Africa standing further away from trees and in Madagascar close and under Commiphora trees Furthermore, I compared distributions between first-year seedlings and seedlings older than one year in both countries. In South Africa, older seedlings were found further away from the nearest female Commiphora crown (median = 25.0 m, q1 = 13.1 m, q3 = 25.0 m, n = 50 seedlings) when compared to first-year seedlings (median = 18.3 m, q1 = 5.5 m, q3 = 25.0 m, n
= 88, Wilcoxon test: z = 2.19, p = 0.028, n1. year seedlings = 88, nolder seedlings = 50) (Fig. 3.5b). Similarly, in Madagascar older seedlings were found further away from the nearest female
Commiphora crown (only data on second study year available) (median = 2.7 m, q1 = 1.2 m, q3 = 7.1 m, n = 19 seedlings) when compared to first-year seedlings (median = 0.8 m, q1 = 0 m, q3 = 4.6 m, n = 111 seedlings, Wilcoxon test: z = 3.08, p = 0.0020, n1. year seedlings = 111, nolder seedlings = 19).
South Africa Madagascar 60 120 a b 40 80
20 40 # seedlings
0 0 all seedlings 0 5 10 15 20 25 >25 0 5 10 15 20 25 >25 first-year seedlings 60 seedlings > 1 year 30 c d 40 20
20 10 # seedlings
0 0 0 5 10 15 20 25 >25 0 5 10 15 20 25 >25 distance from crown (m) distance from crown (m)
Fig. 3.5. Spatial distribution of all seedlings (a, b) and of first-year seedlings and seedlings older than one year (c, d) found in relation to distance from the nearest tree crown for the South African and Malagasy study site. For South Africa, data of the two study seasons are pooled, for Madagascar, data for all seedlings of the two study seasons are pooled, data on first-year seedlings and seedlings older than one year were only available for the second study season.
34 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ...... 3.4.3 Spatial distribution of trees
3.4.3.1 Field data
Distances measured from random points to the nearest Commiphora trees (rt1) (x-axis in Fig. 3.6) were similar for the South African and the Malagasy study site with a median value of
14.4 m (q1 = 6.5 m, q3 = 22.1 m, n = 40 random points) and 12.0 m (q1 = 6.1 m, q3 = 22.1 m, n = 38 random points) respectively. However, distances from first to second neighbour trees
(t1t2) (y-axis in Fig. 3.6) were significantly higher for South Africa with a median of 25.0 m
(q1 = 13.5 m, q3 = 53.0 m) when compared to Madagascar with 11.9 m (q1 = 4.6 m, q3 = 17.8 m, Wilcoxon test: z = 4.7, p < 0.0001, nSouth Africa = 40, nMadagascar = 38). This indicates that Commiphora tree density was roughly the same in both study sites whereas clumping was more pronounced in the Malagasy study site. Calculation of the C-index revealed that trees at the South African study site were significantly uniform distributed (C = 0.36, z = -3.02, p < 0.01), whereas trees at the Malagasy study site were significantly clumped (C = 0.67, z = 3.72, p < 0.001) (Fig. 3.6).
South Africa Madagascar 60 uniform 2 t 1 40 random
distance t 20 clumped
0 0 20 40 60 distance rt1
Fig. 3.6. Spatial pattern of Commiphora trees in the South African (n = 40 random points) and Malagasy (n = 38 random points) study site, as measured by the T-Square- Method. Diagonal line is the expected distribution of both distances for randomly spaced populations. X-values are similar for both sites, however y-values differ significantly. Therefore data points from the South African study site are distributed more above the diagonal line (uniform distribution), data points from the Malagasy study site are distributed more below the diagonal line (clumped distribution).
35 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
3.4.3.2 Computer simulation In my simulation model, dispersal distances strongly influenced the spatial pattern of tree populations. C-indices declined with increasing dispersal distance both for a negative exponential as well as for a log-normal seedling distribution (regression: negative exponential distribution: b = -0.0066, p < 0.0001, r2 = 74.49%, n = 180; log-normal distribution: b = - 0.0076, p < 0.0001, r2 = 78.53%, n = 180) (Fig. 3.7). This indicates that the further seeds get dispersed away from parent trees the less clumped the spatial pattern of the offspring population will become. Differences in results for a negative exponential and a log-normal seedling distribution are found for low dispersal distances (1 m, 2 m, 6 m and 15 m from tree crown) in such a way that log-normal seedling distributions resulted in significantly higher clumping of populations compared to exponential seedling distributions (Fig. 3.7, t-test: t < - 2.5, df = 58, p < 0.05).
0.9
0.7 *** *** clumped ** 0.5
random *
Spatial pattern [C-index] 0.3 uniform 1 10 100 log (dispersal distance) [m]
Fig. 3.7. Spatial pattern of trees as a function of dispersal distance. Mean ± 1 SD (n = 30 replications) of C-indices after simulation with dispersal distances drawn from an exponential distribution (see left inlay, l) and a log-normal distribution (see right inlay, ¡). C-indices range from 0 to 1 with 0.5 for random patterns, < 0.5 for uniform and > 0.5 for clumped patterns. Dashed horizontal lines represent significance limit (p < 0.05) for departure from random pattern. *** p < 0.001, ** p < 0.01, * p < 0.05.
36 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
3.5 Discussion
This comparative seed dispersal study in South Africa and Madagascar presents evidence that regional differences in frugivore diversity and behavior can have consequences for seed dispersal rates, seedling establishment and the spatial distribution of seedlings and trees. Due to a high avian frugivore diversity in South Africa, seeds of the South African tree species were dispersed by 11 bird species (mainly by swallowing seeds), consequently leading to a high dispersal rate of 70.8%. Correspondigly, established seedlings could be found away from Commiphora trees and trees were uniformely spaced. In contrast, the low avian frugivore diversity in Madagascar was reflected by only three bird species dispersing seeds (mainly by carrying seeds away in their beak), resulting in an extremely low dispersal rate of 7.9%. Correspondingly, most seedlings established under and close to the Commiphora trees and trees were clumped in the study site. One possible alternative explanation for the differences in dispersal rates between South Africa and Madagascar could be differences in fruit size (Pratt & Stiles 1985). The somewhat smaller diaspores of the South African Commiphora species could possibly attract more bird species and have a higher probability of being dispersed because they can be swallowed by birds with smaller gape width (Wheelwright 1985, 1993, Jordano 1987a, b, Levey 1987). However, differences in diaspore size are very small (South Africa: 7.1 x 5.7 x 5.1 mm, n = 20; Madagascar: 12.2 x 8.0 x 6.3 mm, n = 20, Fig. 3.1). In addition, observations demonstrated that the larger diaspores of the Malagasy Commiphora species can be readily swallowed by the Madagascar Bulbul and Crested Drongo (Böhning-Gaese et al. 1999). However, these species were rare in the study site and did not contribute much to dispersal. Thus, fruit size does not seem to be the reason for the differences in dispersal rates between the two countries. Another factor that could be responsible for differences in dispersal rates might be differences in fruit crop size between the two tree species in South Africa and Madagascar (Pratt & Stiles 1985). For both species, however, the results show that the total percentage of dispersed seeds did not depend on fruit crop size. These results are consistent with other studies that often reported little or no influence of the number of fruits produced on the percentage of dispersed seeds (Howe & Smallwood 1982, Jordano 1987a, Laska & Stiles
37 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
1994). Thus, crop size does not seem to be the reason for the profound differences in seed dispersal rates found between the South African and Malagasy species. Frugivore diversity and seed dispersal rates as found in the South African study site are similar to studies in other African and Neotropical forests (for details see chapter 2) whereas avian frugivore diversity and dispersal rates for the Malagasy study site were exceptionally low. A depauperate avian frugivore community for Madagascar was already reported from other studies with mainly lemurs instead of birds contributing to seed dispersal of tree species
(Scharfe & Schlund 1996, Dew & Wright 1998, Böhning et al. 1999, Ganzhorn et al. 1999). My data indicate that regional differences in frugivore diversity and behavior not only affect seed dispersal rates but also seedling establishment, dispersal benefit and seedling distribution. Other studies report that a reduced frugivore diversity, e.g. in habitat fragments, can lead to a decrease in dispersal efficiency and to a lower seedling recruitment (Santos & Telleria 1994, 1997). Establishment of first-year seedlings was similar in the two study sites with 169 newly established seedlings per hectare per year in South Africa and 209 in Madagascar (Fig. 3.4). However, when assuming that seeds of the Malagasy tree species might be dispersed at the same high rates as in the South African species, i.e. 70.8% instead of 7.9%, I could expect much more first-year seedlings in Madagascar. The number of seedlings in the Malagasy study site that can be expected with South African dispersal rates can be calculated with the formula (cobs * (aSouth Africa / aMadagascar)) + (dobs * (aSouth Africa / aMadagascar)) /
(cobs + dobs) (for description of parameters see Figure 3.2 and 3.3.2.2 Benefit of seed dispersal). This results in an up to 8.7 higher seedling establishment in Madagascar, i.e. 1818 seedlings instead of 209. Seedling mortality of first-year seedlings was higher in Madagascar compared to South Africa (35% vs. 15%) which could be due to a general high mortality of seedlings under parent trees as indicated by a shift of older seedlings away from the trees (Fig. 3.5d). Dispersal benefit as the increase in probability of a seed getting established as seedling when dispersed was much higher in Madagascar compared to South Africa (factor 79.9 vs. 5.6). This difference could be due to differences in frugivore deposition behavior. In South Africa, newly established seedlings were often found in aggregations under trees that were probably used as perches by birds. This could result in an increasing density-dependant mortality of seedlings away from trees and decreasing dispersal benefit. In addition, in Madagascar seedling density under parent trees was very high, probably leading to high seedling mortality under parent trees and increasing dispersal benefit.
38 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
Differences in seedling establishment might have consequences for the abundance of adult trees. Relevant studies are scarce, although their importance for conservation issues is often stressed. In a study by De Steven (1994), the size of a new seedling cohort was shown to be proportional to their later contribution to the seedling population, and the observed seedling and sapling dynamics were consistent with recent population trends of trees. Seedling distributions in both study sites matched well with the dispersal rates. A high seed dispersal rate in the South African study site was correlated with a log-normal seedling distribution with a median of 21 m away from the next Commiphora crown whereas in the Malagasy study site a low dispersal rate resulted in a negative exponential seedling distribution with a median (0.9 m) close to the next Commiphora tree (Fig. 3.5). Results in both study sites point towards a higher mortality of seedlings under and near the tree crown compared to seedlings further away, resulting in a shift of older seedlings away from the parent trees. This is supported by the data about high dispersal benefit. These findings support Janzen (1970, 1971) and Connell’s (1971) hypothesis of higher seedling mortality beneath parent trees due to seed predation, herbivory or intraspecific competition. In general, several studies have supported the Janzen-Connell-model (Wright 1983, Augspurger 1984, De Steven & Putz 1984, Howe et al. 1985) although results often remain controversial (Howe & Smallwood 1982, Clark & Clark 1984). Besides tree abundance, also spatial patterns of plant populations might be affected by seed dispersal and seedling distribution. This is indicated by the field data and computer simulations. In the South African study site, Commiphora trees were uniformely distributed, whereas trees in the Malagasy study site had a clumped pattern (Fig. 3.6). Many studies of tropical forests demonstrate that tree species are mostly clumped or randomly distributed rather than uniform (Lang et al. 1971, Hubbell 1979, Forman & Hahn 1980, Strasberg 1996, Parthasarathy & Karthiekeyan 1997). What could have caused the uniform pattern of the South African tree species ? One reason could be the fact that the T-square index of spatial pattern, C, which was used in this study is sensitive to detecting uniform pattern in relation to other indices (Diggle et al. 1976). In addition, my simulation data indicate that C-values of 0.36 can occur as statistical outliers when in fact a random pattern should be detected. The simulations support my field data indicating that the spatial pattern of tree populations is indeed influenced by the average dispersal distance with which seeds are
39 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ...... dispersed. So far, hardly any study linked dispersal distances with the spatial pattern of trees
(but see Fleming & Heithaus 1981). Hubbell (1979) linked seed dispersal mode with the spatial pattern of trees suggesting that mammal-dispersed plant species are more clumped than bird- or wind-dispersed species. In my simulation, greater dispersal distances resulted in less clumped and more random patterns. Alternatively, differences in seedling distribution and spatial patterns of trees between South Africa and Madagascar could be due to differences in habitat heterogeneity and seed and seedling predation, factors I did not account for in my simulation. With regard to habitat heterogeneity, the clumped distribution of trees in Madagascar might be caused by a more patchy distribution of suitable microhabitats for seedling and tree establishment in Madagascar as compared to South Africa. Potentially important microhabitat variables such as soil water content, nutrients, or vegetation cover have not been systematically mapped in the two study sites. However, from visual inspection, the study site in South Africa appears to be far more heterogenous than the study site in Madagascar. Species composition of the forest in the South African site and vegetation cover varies markedly with distance from river, slope and exposition. In contrast, the topography of the study site in Madagascar is fairly level and species composition and vegetation cover is fairly homogenous throughout the study area. Thus, the spatial distribution of suitable microhabitats appears not to be the reason for the different spatial distributions of trees. Therefore, seed dispersal distance appears to be the most parsimonious explanation for observed spatial patterns. To conclude, field data on two exemplary dispersal systems in South Africa and Madagascar under different frugivore regimes as well as simulation data on the effect of dispersal on spatial structure of plant populations, suggest that frugivore diversity and behavior has not only consequences for seed dispersal of plants, but also for seedling establishment and the spatial distribution of seedlings and trees. However, results of "natural experiments" have to be assessed carefully as other factors can never be completely ruled out. As I studied only one species in each country, these results, of course, first of all apply to the specific study site and tree species. Thus, to make generalizations, further studies on other study sites and species are needed. However, my data suggest that frugivores appear to affect not only seed dispersal of concerned plants but in a wider frame also the structure and dynamic of tropical forest ecosystems.
40 3. CONSEQUENCES OF FRUGIVORE DIVERSITY FOR SEED DISPERSAL ......
3.6 Summary
Many plants depend on frugivorous animals for the dispersal of their seeds. However, it is only poorly known whether regional differences in frugivore diversity have consequences for seed dispersal, seedling establishment, and the spatial distribution of seedlings and trees. This comparative seed dispersal study investigated the consequences of regional differences in frugivore diversity for two tree species of the genus Commiphora. Commiphora harveyi was studied in South Africa where avian frugivore diversity is high, C. guillaumini was studied in Madagascar where a depauperate avian frugivore community is found. In both study sites, visitation rates, seed handling and dispersal rates in Commiphora trees were quantified by tree observations and fruit traps for two consecutive years. Seedlings were mapped and the spatial distribution of trees quantified. In both study sites, fruits were mainly eaten by birds with the Crowned Hornbill (Tockus alboterminatus) and the Redbilled Woodhoopoe (Phoeniculus purpureus) being the major dispersers in the South African site and the Lesser Vasa Parrot (Coracopsis nigra) in the Malagasy site. Total seed dispersal rates in South Africa were significantly higher compared to Madagascar (70.8% vs. 7.9%) due to a lack of effective dispersers swallowing seeds in Madagascar. Seed dispersal benefit, i.e. the increase in the probability to get established as seedling away from parent trees due to dispersal was much higher in Madagascar (factor 79.9) compared to South Africa (factor 5.6). Corresponding to different seed dispersal rates, seedlings in South Africa were found at relatively large distances from the nearest Commiphora tree in South Africa (median distance 21.0 m), whereas in Madagascar seedlings were found mostly under and close to the nearest Commiphora tree (median distance 0.9 m). Finally, Commiphora trees in the Malagasy study site were clumped, in the South African study site more randomly distributed. An individual- based simulation model indicated that seed dispersal influences the spatial pattern of tree populations with low seed dispersal distances leading to clumped populations and high distances resulting in randomly distributed populations. These results suggest that regional differences in frugivore diversity and behavior strongly affect seed dispersal of trees, seedling establishment and the spatial distribution of seedlings and trees.
41 42 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
4. Seed dispersal, breeding system, tree density and the spatial pattern of trees
4.1 Introduction
The spatial pattern of tree populations has been investigated in a number of studies (Poore
1968, Hubbell 1979, Forman & Hahn 1980, Strasberg 1996). However, studies focussing on possible factors driving the spatial patterns of trees are scarce. In the tropics, most tree species are known to occur in low densities (see e.g. Pitman et al. 1999) with mostly clumped spatial distributions (see e.g. Hubbell 1979) resulting in small local breeding populations. It is important to study the factors influencing spatial patterns as small breeding populations of tropical trees are at great risk in the face of deforestation and habitat fragmentation (Ackerly et al. 1990). Fragmentation might disturb biological processes such as seed dispersal and seedling establishment (Turner 1996, Laurance & Bieregaard 1997, Santos & Telleria 1997, Bleher & Böhning-Gaese, unpubl. manuscript). Fragmentation also increases the spatial isolation of plant populations (Young et al. 1996,) and therefore affects the genetic structure of populations (Hamilton 1999). Potential factors influencing the spatial pattern of tree populations might be seed dispersal and the location and density of possible seed sources. Seed dispersal by animals plays an important role especially in the tropics (Howe & Smallwood 1982). The probability that seeds will be dispersed and deposited at long distances from the parent plant might be determined by frugivorous animals who consequently alter the shape of the seed shadow by extending its tail (Portnoy & Willson 1993). Furthermore, seed dispersal is assumed to influence seedling establishment as, in general, the survival of seeds and seedlings is reported to be higher away from the parent trees (Janzen et al. 1976, Augspurger 1983, 1984; Howe et al. 1985, Schupp 1988). The interplay of both, seed deposition probability due to dispersal and establishment probability due to differential seed and seedling mortality, is thought to influence the spatial pattern of seedlings and saplings (Fleming & Heithaus 1981, Howe 1986). This might have consequences for the spatial distribution of adult tree populations as well, however, studies linking seed dispersal and seedling distribution with the spatial pattern of plant populations are rare (but see Fleming &
43 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
Heithaus 1981, Bleher & Böhning-Gaese, unpubl. manuscript). In the present study, I tried to find evidence that seed dispersal distance and offspring distribution affect the spatial pattern of tree populations. Other possible factors responsible for the formation of spatial patterns are the location and density of possible seed sources as determined by breeding system, tree density and spatial pattern of parent population. However, evidence for this is also scarce (Hubbell 1979, Nanami et al. 1999). Many tropical tree species are known to have dioecious breeding systems and to occur mostly in low densities, whereas temperate tree species are rather wind- dispersed monoecious species in high-density populations (Hubbell 1979, Ashton 1984, Bawa & Krugman 1991). Dioecious breeding systems in the tropics with only halve of the tree population contributing to seed dispersal, and low tree density might affect the spatial distribution of offspring generations. I studied the effect of both factors on the spatial patterns of plant populations. In general, field studies linking possible influential factors with the spatial pattern of tree populations result in current snap-shots since population dynamics in trees over generations can only be investigated with great difficulty. To provide insight in possible factors driving the formation of spatial patterns of tree populations over time, I used an individual-based simulation model. The aim of the model was to simulate tree populations over time and investigate if seed dispersal distance, offspring distribution, breeding system, tree density and the spatial pattern of the parent plant population (i.e. random, clumped or uniform distribution) influence the spatial pattern of trees.
4.2 Methods
4.2.1 Simulation model
4.2.1.1 Tree populations An individual-based, spatially explicit simulation model was constructed in Borland Delphi 4 (1998). It simulates population dynamics of seed-dispersing tree populations over time. The
44 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ...... tree populations (each tree with a crown diameter of 6 m) are distributed on a 25 ha plot (500 x 500 m) which was assumed to be of homogenous habitat. Each tree is assigned a death year (maximum life span 70 years) and for each reproductive tree a number of offspring is drawn from a negative exponential distribution with a mean of either 1.0 or 2.2 (see below).
4.2.1.2 Simulation parameters Simulations were run with different parameter combinations. Parameters studied were dispersal distance, offspring distribution, breeding system, tree density and the spatial pattern of the parent population. 1. dispersal distance: dispersal distance in my model represents the distance between parent tree and offspring tree of the next generation, but not the average distance with which seeds are dispersed from the parent tree as it is normally used. Therefore, I do not simulate the distance with which seeds are dispersed, but the distance at which reproductive offspring trees get established. Dispersal direction was assumed to be random. 2. offspring distribution: distribution of offspring represents the spatial distribution of offspring in relation to their parent tree which had the form of either a negative exponential or a log-normal curve. Both curves are reported for seed shadows, and seedling distributions (Willson 1993, Bleher & Böhning-Gaese, unpubl. manuscript). I used these two distribution curves for my simulations as two extreme forms of offspring distributions. For a negative exponential distribution, the peak is much closer to the parent tree crown as compared to a log-normal distribution. The tail of the offspring distribution is longer for the negative exponential distribution as for the log-normal distribution. A given dispersal distance determines the distance of the peak from the parent tree. Offspring being dispersed outside the virtual 25 ha plot were assumed to be lost. 3. tree density: this parameter represents the number of trees on the 25 ha plot. 4. breeding system: populations were either assumed to be monoecious with all trees reproducing (mean number of offspring 1.0) or dioecious with 50% of the population, i.e. only female trees, reproducing (mean number of offspring 2.2). 5. spatial pattern of parent populations: this parameter represents the intial distribution of trees before the start of the simulation, with trees being either randomly, uniformly or clumped distributed on the plot.
45 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
For a pre-evaluation of the importance of the parameter studied, only two extreme values of dispersal distance (low (2 m) and high (50 m)) and tree density (low (2 trees / ha) and high (40 trees / ha)) were used for the simulations.
4.2.1.3 Running the simulation For each parameter combination I run 30 simulation replications each over 300 time steps (hereafter called years). In the simulation, trees successively died when reaching their assigned death year. Dying trees "dispersed" their assigned number of offspring with given dispersal distances. Dispersed offspring became reproductive trees of the future population. The approach of parent trees dying and dispersing seeds only once at the end of their life was used to keep the tree population stable. However, I do not think that this approach lead to biases in results. Maximum overlap of individual tree crowns was 1.3 * radius of tree crown, adding a component of density-dependant competition.
4.2.2 Detecting spatial patterns
After each simulation, the spatial pattern of the tree population was determined using the T- Square-method (Ludwig & Reynolds 1988). First, the distance of 40 random points (r) in the
25 ha plot to the nearest tree (t1) was measured (rt1). Then, a line perpendicular to the line rt1 at t1 was drawn, and the distance from tree t1 to its nearest neighbour t2 beyond the "half- plane" created by this perpendicular was measured. The spatial distribution of trees could be determined by calculating an index C derived from the ratio of the squared distances rt1 and t1t2. The value of C ranges from 0 to 1 with 0.5 for random patterns, < 0.5 for uniform patterns and > 0.5 for clumped patterns. A test statistics z is used to test departure from random pattern. The T-Square-method was used as it is assumed to be sensitive towards uniform distributions (Diggle et al. 1976) and less scale-dependant compared to quadrat sampling methods. However, for an evaluation of the method, I compared results obtained with the C- index with those obtained with the Morisita index (Morisita 1959).
46 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ...... 4.2.3 Statistical analysis
Mean C-indices for each parameter combination were obtained by averaging over 30 replications. The importance of dispersal distance, offspring distribution, breeding system, tree density, and start pattern for the spatial pattern of tree populations was evaluated by a multivariate ANOVA. To test the effect of the most important factors dispersal distance and tree density on the spatial distribution, I performed regression analyses varying these factors with a finer resolution of values. For all statistical analysis JMP (1995) was used.
4.3 Results
4.3.1 Factors influencing spatial patterns
An example of spatial pattern formation over 300 simulated years is shown in Figure 4.1, starting with clumped, random and uniform start populations (see left snap-shots). Using an average dispersal distance of 25 m in the simulation, a stable random end pattern developed after 100 simulated years, independant of the spatial pattern of the start populations. Therefore, a simulation time of 300 years seems to be adequate to obtain stable end patterns.
Fig. 4.1. Spatial patterns of tree populations measured as C-index (mean ± SD) over 300 simulated years for populations starting from a random, clumped and a uniform pattern. Left snap-shots illustrate examples for each initial spatial pattern. Dispersal distance 25 m, offspring distribution: negative exponential, breeding system: dioecious, tree density: 12-13 trees / ha. Dotted lines reflect significance limit for departure from random patterns. Each mean calculated from 30 replications. 47 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
Dispersal distance proved to be one of the dominant factors driving the formation of spatial patterns in my model, whereas offspring distribution had only weak effects (Fig. 4.2, Tab. 4.1). Breeding system had a medium effect on spatial patterns, whereas tree density appeared to be another important factor driving spatial patterns (Fig. 4.2, Tab. 4.1). The spatial pattern of start populations had only weak effects (Fig. 4.2, Tab. 4.1). In the following, all effects are described explicitely. Additionally, the main factors dispersal distance, breeding system and tree density were studied in more detail, with a random start pattern and a negative exponential offspring distribution.
1.0 dispersal distance low (2 m)
0.8 clumped
0.6
random 0.4
nexp lno nexp lno nexp lno nexp lno 1.0 dispersal distance low (2) high (50) low (2) high (50high m)(50) low (2) high (40) 0.8 Spatial pattern [C-index] clumped
0.6
random 0.4
nexp lno nexp lno nexp lno nexp lno offspring distribution mono dioe mono dioe breeding system low (2 trees / ha) high (40 trees / ha) tree density inital spatial pattern clumped random uniform
Fig. 4.2. Spatial patterns of tree populations measured as C-index (mean ± SD) after 300 simulated years under various factor combinations. Dotted lines reflect significance limit for departure from random patterns. Each mean calculated from 30 replications. nexp - negative exponential, lno - log-normal, mono - monoecious, dioe - dioecious.
48 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ...... Tab. 4.1. Multivariate ANOVA evaluating the effect of dispersal distance, offspring distribution, breeding system, tree density and inital spatial pattern on the spatial pattern of tree populations. Given are degrees of freedom, test statistics F and p.
variable df F p dispersal distance 1 4031.52 0.0000 offspring distribution 1 8.68 0.0033 breeding system 1 142.43 0.0001 tree density 1 3388.91 0.0000 initial spatial pattern 2 7.40 0.0006
4.3.2 Seed dispersal and spatial patterns
Dispersal distance was one of the dominant factors influencing spatial patterns of tree populations. Populations were highly clumped when dispersal distances were low i.e. when offspring grew in immediate vicinity to the parent trees (Fig. 4.3). Clumping declined with increasing dispersal distance which holds for different tree densities (Fig. 4.3). The decline in clumping was gradual in low-density-populations resulting in a lower degree of clumping with high dispersal distance. In populations with higher densities, I found a rapid fall-off in clumping reaching a stable plateau of random distribution.
1.0
tree density 0.4 trees / ha 0.8 0.8
4 clumped 0.6 40 random Spatial pattern [C-index] 0.4
uniform 0 20 40 60 80 100 Dispersal distance (m)
Fig. 4.3. Spatial pattern of tree populations (mean C-index) after 300 simulated years as a function of dispersal distance at four different tree population densities. Smoothed curves produced through cubic spline interpolation (SigmaPlot 1997). Offspring distribution: negative exponential, breeding system: dioecious, initial spatial pattern: random. Dotted lines reflect significance limit for departure from random patterns. Each mean calculated from 30 replications. 49 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
In contrast to dispersal distance, offspring distribution had only weak effects on spatial patterns (Fig. 4.2, Tab. 4.1). Only in dioecious high-density tree populations with low dispersal distance, populations with log-normal offspring distribution resulted in a higher degree of clumping than populations with negative exponential distribution (Fig. 4.2).
4.3.3 Breeding system and spatial patterns
Breeding system proved to be a medium effect influencing spatial patterns. With a constant medium dispersal distance of 20 m, tree populations with a dioecious breeding system always developed higher clumping than populations with a monoecious breeding system (Fig. 4.4). The only exceptions were high-density populations with high dispersal distance resulting in a random distribution independant of breeding system (Fig. 4.2).
tree density 1.0 * 0.4 trees / ha 4 trees / ha 20 trees / ha 0.9 *** 40 trees / ha
0.8 ***
0.7 *** Spatial pattern [C-index]
0.6 clumped
0.5 monoecious dioecious random
Fig. 4.4. Spatial pattern of tree populations measured as C-index (mean ± SD) after 300 simulated years for monoecious and dioecious tree populations at four different tree densities. Dispersal distance: 20 m, offspring distribution: negative exponential, initial spatial pattern: random. Dotted line reflects significance limit for departure from random patterns. Each mean calculated from 30 replications.
50 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
4.3.4 Tree density and spatial patterns
Tree density was another dominant factor influencing the spatial pattern of tree populations. Highest clumping was found in low-density populations (Fig. 4.5). The degree of clumping declined with increasing tree density which holds for different dispersal distances (Fig. 4.5). In simulations with low dispersal distance a gradual decline of clumping can be found resulting in a lower degree of clumping with higher tree density. In populations with high dispersal distance, the decline was steep in the beginning reaching a stable plateau of random distribution.
1.0
dispersal distance 0.8 1 m 5 m
clumped 20 m 0.6 100 m random
Spatial pattern [C-index] 0.4
uniform 0 10 20 30 40 Tree density (trees / ha)
Fig. 4.5. Spatial pattern of tree populations (mean C-index) after 300 simulated years as a function of tree density at four different dispersal distances. Smoothed curves produced through cubic spline interpolation (SigmaPlot 1997). Offspring distribution: negative exponential, breeding system: dioecious, initial spatial pattern: random. Dotted lines reflect significance limit for departure from random patterns. Each mean calculated from 30 replications.
4.3.5 Start population and spatial patterns
The spatial pattern of the start population had only a small effect on the spatial pattern of the end populations (Fig. 4.2, Tab. 4.1). Only in monoecious low-density tree populations with
51 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ...... low dispersal distance, a higher degree of clumping was developed from a clumped start population when compared to a random or uniform one (Fig. 4.2).
4.3.6 Combined effects of dispersal distance and tree density
When regressing the C-index on the main factors dispersal distance and tree density, and on the interaction term, I found a significant effect of all three factors on the spatial distribution of tree populations (regression: whole model: F = 2907.98, p < 0.0001, r2 = 83.88%; [log] dispersal distance: F = 652.41, p < 0.0001; [log] tree density: F = 2102.97, p < 0.0001; interaction term: F = 45.55, p < 0.0001). A contour plot with the spatial pattern as a function of dispersal distance and tree density is shown in Figure 4.6. Tree populations with medium to high tree density in combination with medium to high dispersal distances always resulted in a random distribution, whereas low-density populations with seeds being dispersed at low distances resulted in clumped distributions.
Fig. 4.6: Contour plot of the spatial pattern (mean C-index) as a function of dispersal distance and tree density. Numbers in plot refer to mean C-indices as a measure of spatial pattern. The darker the shades the more clumped the pattern.
52 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
4.4 Discussion
My simulation model provides evidence that the formation of spatial pattern in tree populations is mainly influenced by dispersal distance, breeding system and tree density. The lower dispersal distance and tree density, the higher the degree of clumping in offspring populations. As for breeding sytem, dioecious populations were almost always more clumped than monoecious ones. Shape of the offspring distribution and start pattern did not play an important role for the spatial pattern of tree populations. There are only few other studies linking spatial pattern of tree populations with seed dispersal. In a simulation model by Tilman et al. (1997) the extent of clumping in species in general was also found to be a function of dispersal range which supports my results. However, most other studies focus rather on dispersal mode or dispersal rates than on dispersal distance. Ashton (1969) found in a Malaysian rainforest that clumping was most pronounced in tree families in which means of dispersal did not exist or were unreliable (e.g. in Dipterocarpaceae). Clumping was least pronounced in the small-seeded wind-dispersed families Apocynaceae and Leguminosae and in animal-dispersed families such as Meliaceae,
Burseraceae, Sapotacae and Moraceae. Hubbell (1979) linked the spatial distribution of trees in a Costa Rican rainforest to the mode of seed dispersal. Mammal-dispersed species were found to be more clumped than bird- or wind-dispersed ones. Assuming that different dispersal modes might reflect different average dispersal distances (mammal-dispersal resulting in lower average dispersal distances compared to bird- and wind-dispersal) the above findings support my simulation data that seed dispersal distance influences the spatial pattern of trees. In a comparative study (chapter 3, see Tab. 4.2), more randomly distributed tree populations in a South African coastal dry forest were correlated with high dispersal rates due to a rich frugivore community. In contrast, clumped tree populations in a Malagasy dry forest were correlated with low dispersal rates due to a depauperate frugivore community. Strasberg (1996), also, put down clumped distribution of trees of a lowland rainforest on La Réunion to a lack of seed dispersers. Several studies stress the linkage between seed dispersal and the shape of the seedling distribution (Augspurger 1984, Clark & Clark 1984, Houle 1998, Nanami et al. 1999). Seedling establishment is thought to affect adult distributions as well (Janzen 1970, Connell
53 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
1971, Hubbell 1979), however there is not much empirical evidence. In my model, the shape of offspring distribution was not an important factor influencing spatial patterns of populations. Some studies report that the spatial relationship between dispersal-generated seed rain and seedling distribution is concordant deducing a predominance of clumping in trees (Hubbell 1979). In other studies a discordant relationship is discussed due to differential mortality of seeds and seedlings in relation to distance from parent trees resulting possibly in a uniform spatial pattern of trees (Janzen 1970, Connell 1971). The transition from seedling to sapling and adult tree distribution is rarely studied and yields different results. Fleming & Heithaus (1981) discuss a general thinning-out of the distribution with increasing age of the plants. Nanami et al. (1999) report a clumped pattern of seedlings and saplings followed by a rather regular distribution of adult trees in Podocarpus nagi. Strasberg (1996) assumes that high concentration of seedlings near adults and low frequency of young plants and small trees away from mature trees results in clumped distributions of adult trees.
Tab. 4.2. Regional avian frugivore diversity (mainly frugivorous species of total species), seed dispersal system, seedling establishment and spatial pattern of seedlings and trees in a comparative study between a South African and a Malagasy Commiphora tree species (Bleher & Böhning-Gaese, unpubl. manuscript).
South Africa, Madagascar, Oribi Gorge NR Kirindy Forest / CFPF, Commiphora harveyi Commiphora guillaumini avian frugivore diversity 14 of 226 4 of 114 seed dispersing bird species 11 3 dispersal rate 70.8% 7.9% seedling establishment 169 seedlings / ha 209 seedlings / ha seedling distribution away from trees, under and near trees, log-normal negative exponential spatial pattern of trees not clumped clumped
The effects of breeding system on the spatial structure of populations has not been studied in detail. In general, a large proportion of up to 22% of dioecy has been reported for tropical forests, as compared to a lower proportion in temperate forests (review in Bawa & Opler 1975, but see also Ashton 1969, Bawa 1974). In my simulation, breeding system influenced the spatial pattern of tree populations with dioecious populations usually showing
54 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ...... higher clumping compared to monoecious populations. The number of male and female plants are assumed to be the same in a dioecious population as compared to a monoecious one, therefore only half of the parent plants contribute to seed dispersal. Correspondingly, Nanami et al. (1999) reported that dioecy affects the general spatial heterogenity of plant density in tree populations and in combination with a lack of dispersal leads to clumping of seedlings and saplings around female trees. However, Hubbell (1979) found no indication that dioecious tree species were more clumped that monoecious ones. Besides seed dispersal and breeding system, tree density proved to be another important factor in my model affecting spatial patterns. My results that low-density populations were more clumped support Hubbell (1979) who found, too, that rare species are more clumped than common species. In general, most tropical forest trees occur at very low densities (Black et al. 1950, Hubbell & Foster 1983, Pitman et al. 1999). Hubbell & Foster (1983) report from BCI, Panama, that more than 50% of the tree species occur at densities of only 2-3 indiduals per hectare, a number comparable with the tree density used in my simulation. Especially for low-density populations the rate of population increase may depend on successful dispersal of seeds which therefore plays an important role for plants recovering from losses or near extinction (see e.g. Clark & Clark 1981). My results support the general pattern that tropical tree species are mostly clumped or randomly distributed rather than regularly (Poore 1968, Ashton 1976, Hubbell 1979, Forman
& Hahn 1980, Strasberg 1996, Parthasarathy & Karthiekeyan 1997). Uniform tree patterns seem to be very rare regardless of geographical location or forest type (Armesto et al. 1986). Most studies cited, however, focused at the spatial pattern of abundant species rather than on uncommon species which make most of the high species diversity in the tropics (but see Hubbell 1979). Furthermore, many studies were conducted on a very small scale (1.5 - 4 ha) except for Poore (1968, 26 ha) and Hubbell (1979, 13.4 ha) and used different methods for detecting spatial patterns. However, it is known that detecting spatial patterns is very much dependant on the right choice of scale and method (Hurlbert 1979). Various tests are suggested in the literature to measure the deviation of an observed spatial distribution from a theoretical random distribution (Pielou 1969, Greig-Smith 1983). The T-Square-Method used in this study is not frequently applied, however, the obtained C-index is known to be sensitive for detecting uniform patterns (Diggle et al. 1976). For an evaluation of the T-Square-method,
55 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ......
I compared results obtained with the C-index and the Morisita index. Results were very similar and the general findings of my study remained the same. There are various other factors which might influence the spatial pattern of tree populations (Harper 1977), which were not considered in my model, like abiotic (Austin et al. 1972, Ashton 1976, in Forman & Hahn 1980) and edaphic factors (Hubbell & Foster 1986, Ashton 1988, Clark et al. 1999) resulting in different habitat heterogeneity, competition of seedlings (Fleming & Heithaus 1981), seed predation and seed deposition behaviour (e.g. under sleeping sites, Juillot 1997; under bird perches, Pui et al. 1998). However, my simulation model with only one main factors, dispersal distance, can explain the preponderance of clumped and random distributions in tropical low-density tree populations without any additional assumptions. Clumping might be an unavoidable result of dispersal in any habitat even if it is homogenous (Tilman et al. 1997). Future simulation models including additional factors like habitat heterogeneity might provide further insight into the processes driving the formation of spatial patterns in tree populations.
4.5 Summary
Tropical tree populations in low densities and with clumped spatial distributions are at risk in the face of fragmentation. It is therefore important to understand factors driving spatial patterns of tree populations. Seed dispersal might influence the spatial pattern of seedlings and offspring populations of plants. Other possible factors affecting spatial patterns could be breeding system, tree density and the spatial pattern of the parent population. However, relevant studies are scarce. I studied the effect of these factors on the spatial pattern of tree populations over time in an individual-based simulation model. Dispersal distance and tree density were the main factors influencing spatial patterns. Low dispersal distance resulted in highly clumped tree populations and an increase in dispersal distance lead to a decline in clumping towards more randomly spaced populations. Populations with low tree density developed highly clumped patterns, too, and an increase in density lead to a decline in clumping. Breeding system had a medium effect on spatial patterns with dioecious populations usually developing higher clumping than monoecious ones. Offspring distribution
56 4. SEED DISPERSAL, BREEDING SYSTEM, TREE DENSITY AND SPATIAL PATTERN OF TREES ...... and the spatial start pattern had only small effects on the spatial pattern of populations. My results demonstrate that a simulation model with low tree density and dispersal distance as the single factor produces predominantly clumped patterns as found in many tropical tree species.
57 58 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
5. Fruit availability and keystone plant species for frugivores in a dry forest in KwaZulu - Natal, South Africa
5.1 Introduction
"Scarcity and abundance is the key consideration to understand the ecology of tropical forests." Terborgh (1986a, p. 331)
The abundance and diversity of fruits in the tropics is correlated with a diverse animal fauna relying on fruits as food resource (Orians 1969, Karr 1976). Frugivorous animals are the dominant group of vertebrates in tropical forests and, in some areas, make up over 80% of the mammalian and avian biomass (Gautier-Hion et al. 1985, Terborgh 1986b, Fleming et al. 1987). The seasonal pattern of fruit availability and abundance is thought to be a key factor in the evolution of diversity of the frugivore community (Orians 1969, Karr 1976). Considerable variation in fruit availability has been reported for tropical plant species with respect to timing, duration and frequency of fruiting occuring at many levels like season, year, habitat and species (e.g. Frankie et al. 1974, Opler et al. 1980). However, a universal tendency of tropical forests to fluctuate seasonally in fruit production with abundance peaks and periods of scarcity has been established for the Neotropics (e.g. Frankie et al. 1974, Foster 1982a, Terborgh 1986a), South-East Asia (e.g. Raemakers et al. 1980, Leighton & Leighton 1983) and for the African continent (e.g. Lieberman 1982, Gautier-Hion et al. 1985). Fruiting plants that sustain frugivores during times of general fruit scarcity are called "keystone species" (Leighton & Leighton 1983, Terborgh 1986a, Lambert & Marshall 1991). They are of great ecological significance because they appear to set the carrying capacity of the frugivore community (Terborgh 1986b). Figs (genus Ficus) have been identified as keystone plants for frugivore communities in South American and South-East Asian tropical forests during lean periods (e.g. Leighton & Leighton 1983, Terborgh 1986a). However, they do not appear to be keystone resources for frugivores in a Gabon rainforest (Gautier-Hion & Michaloud 1989), where instead plants with arillate fruits take that place as reported somewhere else as well (Leighton & Leighton 1983).
59 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
Frugivores that rely on these keystone resources may be vulnerable to changes in fruit supply resulting from the influence of climatic change on plant phenology (Foster 1982b, Corlett & LaFrankie 1998), from logging (Leighton & Leighton 1983) and from increasing deforestation and fragmentation. This is especially true for tropical forests, where most trees occur in low densities, resulting in highly spatio-temporal patchiness of fruit resources (Terborgh & Winter 1980). It is therefore of critical importance for conservation of tropical forests to improve the knowledge about fruiting phenology and identify vital keystone resources at the level of individual species and communities. To date, only few phenological studies on the African continent exist such as the ones by Hall & Swaine (1981) and Lieberman (1982) in West Africa and by White (1994) in Central Africa. However studies on Southern African forests are missing. Of particular interest is the question whether figs are indeed possible keystone species for frugivorous animals. Although a widespread feature of tropical forest ecosystems, keystone species might be different at different sites. In the present study, I monitored fruiting phenology in a dry forest in KwaZulu - Natal, South Africa, over 13 months to assess fruit availaibility and to determine periods of low fruit production. Furthermore, I investigated seasonal fruit resource utilization by the local frugivore community and identified keystone plant species according to the concept by Power & Mills (1995). This keystone species concept defines a keystone species as "a species whose impact on its community or ecosysten are large, and much larger than would be expected from its abundance".
5.2 Study Site
5.2.1 Location
Our study site was Oribi Gorge Nature Reserve (OGNR) at the South Coast of KwaZulu - Natal, South Africa, located 110 km south of Durban and 21 km inland from Port Shepstone (30°41 - 30°45’ S, 30°10‘ - 30°18.5 E) (Figure 5.1, small map). The 1850 ha Reserve is situated in a 280 m deep gorge which is cut out of Sandstone of the Mzikaba Formation (Thomas et al. 1992) by the Umzimkulwana river. OGNR is under protection of the KwaZulu - Natal Nature Conservation Service (KZNNCS).
60 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
Fig. 5.1. Map of the study site Oribi Gorge Nature Reserve, its location in South Africa (inset) and the six fruit trails used for monitoring fruit abundance and resource utilization.
5.2.2 Climate
The climate in Oribi Gorge is characterised as subtropical with temperatures always above 10
° C and with an average annual rainfall of approximately 1176 mm (Glen 1996). Rainfall peaks in summer in November and February / March whereas consistently low rainfall is found in winter from May to August. For data on temperature and rainfall during the study period see Figure 5.2 (data from SASA, Mount Edgecombe, data on July and August are missing).
61 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
250 30
200
20 150
100
rainfall (mm) 10 temperature (°C) 50
0 0 J A S O N D J F M A M J J 1997 1998 month
Fig. 5.2. Mean monthly rainfall (mm) (bars) and mean (D ), minimum (o) and maximum (l ) monthly temperatures (°C) from July 1997 to July 1998, measured at Paddock (data from the SASA weather station). Data from July and August 1997 are lacking.
5.2.3 Vegetation
The dominant vegetation type in OGNR is dry forest classified as coastal scarp forest (Cooper 1985) which covers the steep slopes of the gorge. Coastal scarp forest is found mainly on South and East facing slopes of high coastal ridges and slopes of deep gorges along the KwaZulu - Natal and Transkei Coast and today covers approximately 21 200 ha (McDevette et al. 1989). Oribi Gorge Nature Reserve is an important conservation area in the so-called Pondoland centre, a region characterised by high plant diversity and endemism, with Cape elements as well as tropical influences in its fauna (Glen 1996). A number of 211 woody plant species is recorded for OGNR (KZNNCS checklist and own records). Besides coastal scarp forest with mostly closed canopy, patches of open woodland can be found on the drier North - facing slopes of the gorge. Along the cliffs, the main vegetation type is grassland. Although the Reserve is surounded by large sugar cane farms and plantations, the vegetation in the gorge has been only slightly disturbed by cutting and gathering of plants, and by the spread of weeds (Glen 1996). For a more detailed vegetation description on OGNR see Glen (1996).
62 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
5.2.4 Frugivore community
The number of bird species recorded for Oribi Gorge Nature Reserve is 226, about one fourth of the total number of bird species recorded for Southern Africa (Maclean 1993). Of the 226 bird species 6.2 % (14) feed mainly, and 22.6% (51) partly on fruit (Maclean 1993). Primates recorded in the Reserve are the Vervet Monkey (Cercopithecus aethiops) and the Samango Monkey (Cercopithecus mitis), both of which feed mainly on fruit, while the Chacma Baboon (Papio ursinus) is omnivorous but includes fruit in its diet (Stuart & Stuart 1995). The Blue Duiker (Philantomba monticola), a small antelope, also occurs in the Reserve and is reported to be mainly frugivorous (Stuart & Stuart 1995).
5.3 Methods
5.3.1 Fruit availability and abundance
Data on fruit production was obtained from July 1997 to July 1998 by monitoring fruit trails following Chapman et al. (1994). For fruit trails, six existing trails in the Reserve were used with a total length of 14.5 km (Fig. 5.1). The trails were representative for the Reserve’s vegetation running through all major vegetation types such as forest (e.g. Hoopoe Falls Trail, Samango Falls Trail), open woodland (Nkonka Trail) and grassland (Mziki Trail). All trails were monitored every two weeks during the morning and early afternoon. All indigenous plants bearing fruit which might be consumed by frugivorous animals were identified 5 m to the left and 5 m to the right of the trail (total monitored area 15.95 ha); identification of plants was carried out with Pooley (1994, 1998), Coates Palgrave (1995) and van Wyk & van Wyk (1997). Exotic species growing in the Reserve were not included in the census. For each plant in fruit the following parameters were noted: life form (tree, shrub, herb, creeper or parasite); presence and number of fruits (ripe and unripe) estimated in situ on a logarithmical scale (1 to 10, 10 to 100, 100 to 1000, 1000 to 10000, > 10000 fruits); degree of ripeness of fruits estimated on a scale from 0 (no fruits ripe) to 1 (all fruits ripe) in 0.25 steps and fruit type (arillate, fig, other). For each species I determined the fruiting period as the period the species had ripe fruit. I calculated the density of each species per hectare and its
63 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ...... median crop size using the arithmetic mean of the logarithmic fruit number class boundaries (i.e. 5 fruits, 50 fruits, 500 fruits, 5000 fruits, 50000 fruits).
5.3.2 Fruit utilization by the frugivore community
Information on fruit utilization by frugivorous animals was obtained during monitoring of the fruit trails. For each observation of animals eating fruit, species and number of individual animals was noted. Please note that all numbers of frugivores are only animals observed in fruiting plants. For plant species consumed by frugivores fruits were collected and weighed fresh. The annual fruit biomass (pulp and seed) per hectare produced by the plant species was calculated from the average fresh fruit weight and the median fruit crop (see Fruit availability and abundance). All statistical analysis were performed using JMP (1995).
5.4 Results
5.4.1 Fruit availability and abundance
I recorded 96 plant species with fruit (ripe and unripe) over the 13-month-study period comprising 59 tree, 26 shrub, 7 creeper, 2 herb, 1 strangler fig and 1 parasite species (for the complete species list see 8. Appendix). Plant species belonged to 46 families with the largest ones being Euphorbiaceae and Rubiaceae (7 species each) (see 8. Appendix). Of the 96 plant species recorded 91 were seen with ripe fruits. Plant species bearing ripe fruit were found all year-round with a minimum number of 16 species bearing ripe fruit in any given month. However, fruiting was seasonal with a peak in early September 1997 just before the start of the rainy season (27 species) and a second more pronounced peak between March and May 1998 at the end of the rainy season (40 species) (Fig. 5.3a). Both fruiting peaks were clearly seen in tree species, however, a less pronounced pattern was found for shrubs and creepers (Fig. 5.3a). Periods of low fruit production
64 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ...... appeared to occur during the dry winter season from June to August, and in summer in between the two rainfall peaks in December and January. Unripe fruit was available all year- round with lower availability only during the winter dry season.
50 a all species 40 trees shrubs creepers 30
20
10 # species with ripe fruit 0
J A S O N D J F M A M J J 1997 1998
b total individuals 200 trees shrubs creepers 150
100
50 # plants with ripe fruit
0 J A S O N D J F M A M J J 1997 1998 month
Fig. 5.3. a. Number of plant species monitored in OGNR with ripe fruits between July 1997 and July 1998. Cumulative numbers given for all plant species, and non-cumulative numbers for tree, shrub and creeper species separately. b. Individual plants monitored with ripe fruits between July 1997 and July 1998 cumulative for all plants, and non- cumulative for trees, shrubs, and creepers separately.
65 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
The abundance of fruiting plants (940 individuals with 598 trees, 258 shrubs, 60 creepers, 19 strangler figs, 4 herbs and 1 parasite) showed a seasonal pattern of fruit production similar to that of species numbers (Fig. 5.3b). Abundance of fruiting plants peaked in October 1997 just before the onset of the rainy season (103 individuals) and between March and May 1998 (maximum of 212 individuals in May) following high rainfall in February and March (Fig. 5.3b). Trees showed both peaks, whereas shrubs peaked only once before the rainy season (Fig. 5.3b). Fruit availability was reduced during the long dry season in July and August and during the rainy season in November and December. These periods are in the following called periods of fruit scarcity. In periods of fruit scarcity, only trees contributed substantially to numbers of fruiting plant species (>70% of all individuals). During low fruit production in July 1997 10 tree species were found fruiting, with Chaetachme aristata (8 trees) and Ficus thonningii (7) being the most frequent ones. During low fruit production in November 1997, 10 tree species were found with ripe fruit, with Chaetachme aristata (10 trees) and Harpephyllum caffrum (9) being the most frequent ones. During low fruit production in July 1998, I found 20 tree species fruiting with Ficus thonningii (9 trees) and Chaetachme aristata (8) being the most frequent species. Fruiting phenologies of various tree species are shown in Figure 5.4. Many species such as e.g. Commiphora harveyi peaked in the late rainy season extending into the dry season (from February to April) (Fig. 5.4). Ficus thonningii and other Ficus species fruited irregularly over the year, however, with peaks during the dry winter period (July) (Fig. 5.4). Examples for tree species with fruiting peaks during the dry winter season (June and July) when fruit availability was low, were Macaranga capensis and Schefflera umbellifera (Fig. 5.4). Examples for tree species fruiting in November and December when fruit availability was also low, were Protorhus longifolia and Harpephyllum caffrum (Fig. 5.4).
66 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
40 10 Commiphora Protorhus 8 30 harveyi longifolia 6 20 4 10 2 # plants with ripe fruit 0 0 J A S O N D J F M A M J J J A S O N D J F M A M J J 10 12 Ficus Harpehyllum 8 thonningi 10 caffrum 8 6 6 4 4 2 2 # plants with ripe fruit 0 0 J A S O N D J F M A M J J J A S O N D J F M A M J J 3 3 Macaranga Schefflera capensis umbellifera 2 2
1 1 # plants with ripe fruit
0 0 J A S O N D J F M A M J J J A S O N D J F M A M J J 1997 1998 1997 1998 month month
Fig. 5.4. Fruiting phenologies of typical tree species used by frugivores.
5.4.2 Fruit utilization by the frugivore community
5.4.2.1 Frugivore community I observed 20 animal species with 18 bird and 2 monkey species eating fruits (see Table 5.1). 16 of the observed animal species are considered to be mainly or partly dependant on fruit, whereas four species are generally considered as insectivorous (Table 5.1). Species numbers
67 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ...... peaked between February and May with a maximum number of 12 animal species in May (Fig. 5.5a). The increase in species number appears to be mainly due to an addition of insectivorous animal species such as e.g. the Redbilled Woodhoopoe observed feeding on fruit only in this time of the year (Fig. 5.5a).
Tab. 5.1. Animal species observed eating fruit from July 1997 to July 1998. Given is their respective diet (from Maclean 1993, Stuart & Stuart 1995), the number of individuals observed, the number of months the animal species was observed eating fruits, and the number of plant species the animal used. mf mainly frugivorous, pf partly frugivorous, i insectivorous.
Species Latin names Diet No of No of No of plant individuals months species used BUCEROTIDAE Trumpeter Hornbill Bycanistes bucinator mf 47 11 6 Crowned Hornbill Tockus alboterminatus mf 24 6 6 CAMPEPHAGIDAE Grey Cuckooshrike Coracina caesia i 1 1 1 DICRURIDAE Forktailed Drongo Dicrurus adimilis i 5 2 2 LYBIIDAE Blackcollared Barbet Lybius torquatus mf 11 5 4 Redfronted Tinker Barbet Pogoniulus pusillus mf 2 2 2 MUSOPHAGIDAE Knysna Lourie Tauraco corythaix mf 55 12 12 ORIOLIDAE Blackheaded Oriole Oriolus larvatus pf 5 4 2 PARIDAE Southern Black Tit Parus niger i 3 2 2 PHOENICULIDAE Redbilled Woodhoopoe Phoeniculus purpureus i 40 4 2 PLOCEIDAE Thickbilled Weaver Amblyospiza albifrons pf 3 1 1 PYCNONOTIDAE Blackeyed Bulbul Pycnonotus barbatus mf 5 1 2 Sombre Bulbul Andropadus importunus pf 3 1 1 STURNIDAE Blackbellied Starling Lamprotornis corruscus mf 125 2 3 Redwinged Starling Onychognathus morio mf 100 5 4 Glossy Starling Lamprotornis nitens pf 12 2 2 TURDIDAE Olive Thrush Turdus olivaceus pf 3 1 2 ZOSTEROPIDAE Cape White-eye Zosterops pallidus pf 14 3 4 CERCOPITHECIDAE Samango Monkey Cercopithecus mitis mf 15 3 6 Vervet Monkey Cercopithecus aethiops mf 4 1 1
68 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
14 a 12 10 8 6 4 # animal species 2 0 J A S O N D J F M A M J J insectivorous 1997 1998 partly frugivorous mainly frugivorous b 140 120 100 80 60 40
# individual animals 20 0 J A S O N D J F M A M J J 1997 1998 month Fig.5.5. a. Number of animal species and b. number of individual animals observed feeding on fruit from July 1997 to July 1998 considered as either insectivorous, partly or mainly frugivorous in their diet.
Considering the number of observations, I recorded a total number of 477 animals eating fruits with the most frequent species being the Blackbellied Starling (26.2 %), the Redwinged Starling (21.0%), the Knysna Lourie (11.5%) and the Trumpeter Hornbill (9.9 %), all species mainly depending on fruit (see Table 5.1). Observations of fruit-eating animals were most abundant between February and May 1998 corresponding to the increase in fruit availability, with a maximum number of 132 fruit-eating animals in February 1998 (Fig. 5.5b). The increase in abundance in fruit-eating animals was only partly due to an addition of insectivorous species, but mostly due to an increase in observations of mainly frugivorous animals, especially of Blackbellied and Redwinged Starlings, both occuring in large flocks (Fig. 5.5b). 69 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
5.4.2.2 Plants used by frugivores Of a total of 96 plant species fruiting along fruit trails during our study, birds or monkeys were observed eating fruits of 23 plant species, predominantly trees (19 species) (Table 5.2). Most plant species were used by the Knysna Lourie (12), the Trumpeter Hornbill, Crowned Hornbill and Samango Monkey (each 6). Plant species in which most animal species and individuals were observed were Ficus thonningii with 13 animal species and 130 individuals recorded and Commiphora harveyi with 7 animal species and 156 individuals (Table 5.2). Both species accounted for 60% of all observed individuals (Table 5.2). All other plant species were each used by less than 5 animal species and less than 33 individuals (Table 5.2). Ficus thonningii produced highest biomass per hectare and year with 32 kg, whereas lowest biomass was produced by Commiphora woodii with 0.01 kg (Tab. 5.2).
Tab. 5.2. Plant species used by animals from July 1997 to July 1998, with the number of animal species and individuals observed eating fruit, fruit weight and the calculated annual biomass per hectare.
Plant species No of animal No of indiv. Fruit Biomass species animals weight (g) (kg/ ha) observed observed Ficus thonningii 13 130 1.1 31.86 Commiphora harveyi 7 156 0.3 0.60 Macaranga capensis 5 11 0.3 2.07 Ficus sur 4 16 23.6 5.70 Protorhus longifolia 3 25 0.6 1.86 Commiphora woodii 3 18 0.5 0.01 Harpephyllum caffrum 3 8 4.1 3.11 Cassipourea gummiflua 3 6 0.4 15.52 Celtis africana 2 33 0.3 0.36 Chaetachme aristata 2 20 2.3 1.27 Croton sylvaticus 2 11 1.0 0.21 Ekebergia capensis 2 8 4.0 0.41 Trema orientalis 2 7 0.2 0.03 Trichilia dregeana 2 4 2.1 0.29 Cryptocarya myrtifolia 2 3 4.0 0.28 Phoenix reclinata 2 3 0.7 0.22 Schefflera umbellifera 2 3 0.2 2.07 Cussonia nicholsoni 1 7 1.1 3.41 Bersama swynni 1 2 - - Ficus burt-davyi 1 2 0.2 0.13 Phyllantus myrtaceus 1 2 - - Halleria lucida 1 1 1.1 0.08 Olea capensis 1 1 0.3 0.14
70 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
Animals were observed eating fruit in Ficus thonningii for the longest number of months (i.e. 10 months), indicating that this species provided fruit year-round (i.e. 10 months) which corresponds to the fruit phenology data (see 8. Appendix). In most other tree species, fruit utilization by animals was only observed over a few months indicating that these species provided only highly seasonal fruit resources. In fact, the number of months a plant species was observed with ripe fruit was correlated with the number of months I observed animals eating fruit in this species (Spearman: r = 0.5899, p = 0.0030, n = 23). The high attractivity of certain plant species to animals, i.e. high numbers of animals eating fruit in plant species, might be due to high plant species abundance, long fruiting periods of plant species, the fact that plants were fruiting during periods of general fruit scarcity and consequently more attractive to animals, large crop sizes or fruit type. I conducted the analysis with 91 plant species, i.e. the plant species which were found to have ripe fruit at some stage during the study period. I found a significant influence of crop size and fruit type in such a way that plant species with higher crop size and with arillate fruits attracted more animals (Tab. 5.3, Tab. 5.4). The higher attraction of animals by species with fig fruits as compared to other fruit types was only marginally significant, probably due to low sample size of Ficus species (4 Ficus species) (Tab. 5.4). Plant species abundance, length of fruiting period or fruiting during general scarcity had no influence on the species attractivity to animals (Tab. 5.3).
Tab. 5.3. Multivariate ANCOVA evaluating the effect of plant species abundance (log), number of months a plant species fruited, fruiting during general fruit scarcity, species crop size (log) and fruit type (arillate, fig or other) on the number of animals (log[animals+1]) observed feeding on fruit of the plant species. Error df = 84, r2 = 39.5 %. Given are degrees of freedom, test statistics F and p.
effect df F p whole model 6 9.14 <0.0001 log [plant species abundance] 1 1.76 0.19 # months fruiting 1 2.01 0.16 fruiting during scarcity 1 0.11 0.74 log [crop size] 1 20.70 <0.0001 fruit type 2 5.77 0.0045
71 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ...... Tab. 5.4. Contrasts between the three fruit types "arillate", "fig" and "other" within the ANCOVA of Table 5.3. Given are sample sizes, least square means after controlling for the other variables in the model (see Tab. 5.3), the test statistics t and p.
arillate fruit fig fruit other fruit type n 9 4 78 least square mean 0.61 0.60 0.19 standard error 0.13 0.21 0.05
t = 1.89, p = 0.063
t = 2.95, p < 0.01
The plant species Ficus thonningii, Protorhus longifolia, Chaetachme aristata, Ficus sur and Macaranga capensis, attracting a high number of animals eating fruit disproportional to the plant`s abundance and fruiting during periods of general fruit scarcity appear to be keystone species in the sense of Power & Mills (1995) (Fig. 5.6). In general Ficus species and plant species producing arillate fruits during times of fruit scarcity can be seen as keystone groups.
Fig. 5.6. Number of animals per plant species observed feeding on fruit (log) and plant species density (log) for species fruiting in periods of general fruit scarcity (l) and for species fruiting outside periods of scarcity (o). Data points falling along the regression line represent plant species which attracted animal numbers proportional to their abundance. Data points falling above the regression line and above a certain absolute threshold represent plant species which attracted disproportional high animal numbers relative to their abundance and can be seen as keystone species (e.g. F.t., P.l.). C.a. Chaetachme aristata, F.t. Ficus thoningii, F.s. Ficus sur, M.c. Macaranga capensis, P.l. Protorhus longifolia.
72 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ......
5.5 Discussion
5.5.1 Fruit availability
Our study presents data on fruit availability and abundance in a South African dry forest. Although some fruit was provided throughout the year, fruiting was seasonal with a fruiting peak before the start of the rainy season and a more pronounced peak at the end of the rainy season. Fruit production was low during the long dry winter season and during a 2-month period in the rainy season. Most phenological studies in the tropics report a seasonality in fruit availability, both for rainforests as well as dry forests (e.g. Frankie et al. 1974, Foster 1982a), however, there seems to be much variation in fruit availability between sites and habitats. In general, the rainy season with its hot and moist conditions is thought to favor fruit ripening with rainfall rather than temperature being the major ecological factor causing fluctuations in fruit production (Corlett 1998). Similar data as for South Africa are found for a tropical dry deciduous forest in Costa Rica where fruit availability peaked before the first rainy season and after the second rainy season, whereas a period of low fruit activity was found in the following short dry season (Frankie et al. 1974). Similarly, in a tropical dry forest in Southern India a fruiting peak was found during the late rainy season extending into the early dry season (Murali & Sukumar 1994). Phenological studies on the African continent are known mainly from western and central Africa, but less so for Southern Africa. These studies report fruiting peaks in the rainy season for a dry forest in Ghana (Lieberman 1982) and a rainforest in Gabon (White 1994), peaks after the rainy season for a rainforest in Gabon (Gautier-Hion & Michaloud 1989), or peaks just before the onset of the rainy season for an evergreen forest in Malawi (Dowsett- Lemaire 1988). Periods of low fruit production are reported mostly for the dry seasons (Dowsett-Lemaire 1988, Gautier-Hion & Michaloud 1989, White 1994). However, peak fruiting in the dry season is reported for rainforests at Ivory Coast (Alexandre 1980) and Ghana (Hall & Swaine 1981). I found altogether 96 fruiting plant species over the year with a high percentage (> 60%) being trees. However, the actual number of woody plants with fruit for animal consumption is higher in the Reserve as some species which might provide important fruit resources did not fruit along our fruit trails or not during our study period. A total number of 211 woody species
73 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ...... is recorded for OGNR (KZNNCS checklist) with an estimated 66% (140 species) bearing fruit adapted to animal dispersal. Our estimated percentage appears higher than the one found by Knight & Siegfried (1983) with 58% for the Southern African tree flora probably due to the inclusion of shrubs in our census. Other percentages given for animal dispersed plant species on the African continent are 75% for a dry deciduous forest in Ghana (White 1994), 76% for the Tai rainforest in Ivory Coast (Alexandre 1980) and 80% for an evergreen forest in Malawi (Dowsett-Lemaire 1988).
5.5.2 Frugivore community
In our study the frugivore community observed eating fruit consisted of 18 bird and 2 monkey species. Animal species number and abundance peaked between February and March corresponding to the high fruit availability. The increased observations of birds categorized as insectivorous is reported by other studies as well, as insectivorous species might become seasonally frugivorous (e.g. see Bates 1992). Animal species most frequently observed were Blackbellied and Redwinged Starling due to their occurence in large flocks. Animals seen over the longest time and using the highest number of plant species were Knysna Lourie (family Musophagidae), Trumpeter Hornbill and Crowned Hornbill (both family Bucerotidae), families which are known to be highly frugivorous (Leighton & Leighton 1983, Kemp 1995, Sun et al. 1997, O’Brian 1997). Animal species numbers observed in fruiting plants were in general lower compared to the observations in a South African dune forest (Frost 1980) and in an evergreen forest in Malawi (Dowsett-Lemaire 1988). For example, fruits of the tree species Ekebergia capensis were observed being eaten by only two animal species during our study, by nine species in a South African Dune forest (Frost 1980) and by four species in an evergreen forest in Malawi (Dowsett-Lemaire 1988). The question therefore arises whether random observations along fruit trails as carried out in our study yield representative results when compared to long-term systematic observations such as the ones by Frost (1980) and Dowsett-Lemaire (1988). In a seed dispersal study on Commiphora harveyi at the same study site over two fruiting seasons with systematic long-term observations of 345 hours I recorded altogether 13 animal species feeding on the fruits (Bleher & Böhning-Gaese, unpubl. manuscript), whereas in the present study, the recorded species number in C. harveyi trees was
74 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ...... seven. However, the seven species accounted for 71% of all observations in the long-term study and only rare visitors were missed in random observations. Therefore our results appear to be representative for the most abundant frugivores.
5.5.3 Resource utilization and keystone species
Data on resource utilization by the frugivore community was gathered on only 24% of the plants found fruiting (23 species) predominantly on large trees. Although the shrubs Cryptocarya wyliei and Psychotria capensis, and the understory tree Englerophytum natalense occured in highest densities, I did not observe any animals feeding on their fruits. Similar results with fewer animal species observed in small trees, climbers and epiphytes as compared to large trees are reported from a study by Dowsett-Lemaire (1988). Two of the 23 plant species, the trees Ficus thonningii and Commiphora harveyi, were found to attract a high number of bird species and individuals (66%), whereas for most other plant species feeding records involved fewer than three animal species. The small numbers of bird species observed might reflect the rarity of the plant species, short fruiting periods, the fact that the plant species fruited during periods of overall high fruit availability, small crop size, or an unattractive fruit type, however, only crop size and fruit type was found to influence animal numbers (Table 5.3). A species that makes an unusually strong contribution to another species, to community structure or processes, is called "keystone species", as for example a unique food source that fruits during seasons of fruit scarcity for tropical frugivores (Meffe & Caroll 1994). A first concept of "keystone species" was drawn by Terborgh (1986a) who identified figs as keystones in a Peruian rainforest due to 1. low interannual variation in fruit production and a regular timing of the fruiting period and 2. a high rate of consumption by frugivores. In addition, figs provided fruit year-round as a result of asynchronous fruiting at the population level making them a comparatively constant reliable source of food whereas other fruit species are distinctly seasonal (Frankie et al. 1974, Janzen 1979, Foster 1982a, Leighton & Leighton 1983). The keystone species concept is very important for nature conservation where priorities must be set in the effort to conserve species and habitats (Power & Mills 1995). Therefore, the need for a clearer and more operational concept resulted in a concept redefinition with a
75 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ...... keystone species defined as "a species whose impacts on its community or ecosystem are large, and much larger than would be expected from its abundance" (Power & Mills 1995). Furthermore, keystones according to this concept have a total impact whose magnitude exceed some absolute threshold (Power & Mills 1995). Consequently, Ficus thonningii emerges as clear keystone species at our study site in the sense of Power & Mills (1995). Figs have been seen as keystone species for many tropical forests (Leighton & Leighton 1983, Terborgh 1986a, Raemakers et al. 1980). However, Gautier-Hion & Michaloud (1989) provided evidence that figs could not be considered as keystone species for a tropical forest in Gabon due to their occurence at low densities (1.5 trees per hectare for 20 species), their unpredictable fruiting patterns and low crop production, therefore being infrequently eaten by most animal species. In contrast to that, fig trees at our study site were very common trees at high densities (4.7 individuals per hectare for four species), fruiting asychronously nearly year-round, providing large fruit crops with high biomass and attracting a total of 15 bird species and one monkey species. Fig trees therefore provided a staple diet for the frugivore community over the whole year, especially in periods of fruit scarcity. Ficus is reported to be the largest genus in Natal with 19 species (Pooley 1994). Its importance for frugivores is stressed by the fact that large fig trees are often found standing isolated on sugar cane farms and plantations all over the coast attracting a large variety of animal species from far away when in fruit. Although keystone species appear to be site-specific (Menge et al. 1994), figs in general might be considered as a keystone group for frugivores in South Africa, at least so for coastal forests. Other identified keystone species at our study site according to the definition by Power & Mills (1995) were Protorhus longifolia, Chaetachme aristata and Macaranga capensis. However, whereas the removal of e.g. Macaranga capensis trees might not have severe consequences for the frugivore community due to their low density at the study site, the removal of Ficus trees might drastically influence the community. This has already been shown for Umfolozi Nature Reserve, South Africa, where Ficus sycomorus populations were destroyed by Cyclone Demoina in 1984, consequently leading to a local disappearance of frugivorous bird species. To conclude, identification of keystone plant species for frugivores seems especially important in areas were there are marked seasonal periods of low fruit production and where
76 5. FRUIT AVAILABILITY AND KEYSTONE PLANT SPECIES IN SOUTH AFRICA ...... escape of frugivores to other forests for food is constrained. As for coastal forests in KwaZulu - Natal, South Africa, they today cover only approximately 21 000 ha (McDevette et al. 1989, unpublished KZNNCS report) and are only found in Reserves, especially in deep rather inaccessible gorges, and on a few private farms. Therefore, the knowledge of crucial keystone resources is essential for sound management decisions.
5.6 Summary
I investigated fruit phenology and availability from July 1997 to July 1998 in the dry forest of Oribi Gorge Nature Reserve, South Africa, and studied its seasonal utilization by the local frugivore community. Along fruit trails of a total of 14.5 km I monitored fruiting plants every two weeks and made observations of fruit-eating animals. A total of 96 plant species predominantly tree species (59 species) were recorded bearing fruit with a total of 940 individuals. Although some fruit was available all year-round, both fruiting species richness and fruit abundance showed a strong seasonality with a peak just before the rainy season for trees (October, November), and a second even more pronounced peak after the rainy season (March to May) for trees, shrubs and creepers. Fruits of 23 species were observed being eaten by animals, i.e. by 18 bird and two monkey species, during the study period. Species richness and abundance of animals eating fruit corresponded with the main peak of fruiting availability and showed a maximum from February to May. This maximum was caused by a number of insectivorous species feeding on fruit, in addition to an increase in observations of mainly and partly frugivorous species. Certain tree species attracted a high percentage of the animal species and individuals disproportional to their abundance. These species appear to be keystone species at our study site in the sense of Power & Mills (1995). The keystone species attracting highest animal numbers was Ficus thonningii. Fig trees (4 species) occured in overall high density, fruited asynchronously during most of the year, even during periods of low general fruit production, and provided high overall fruit biomass. Other keystone species identified according to Power & Mills (1995) were e.g. Protorhus longifolia and Macaranga capensis. These species were much sought after by the frugivore commuity, despite their low overall density.
77 78 6. GENERAL CONCLUSIONS ......
6. General Conclusions
In tropical forests, a high percentage of up to 90% of tree and shrub species produce fleshy fruits adapted to animal dispersal and eaten by a large number of frugivorous vertebrates. On the one hand, seed dispersal is important for the plants' fitness as survival of seeds and seedlings appears to be enhanced further away from the parent plant. On the other hand, frugivorous animals rely on fruits as food resource and therefore often face seasonal fluctuations in fruit availability. In this thesis, I studied the consequences of seed dispersal by animals for plant populations and of seasonality of fruit availability for the avian frugivore community in the dry subtropical forest of Oribi Gorge Nature Reserve, South Africa. In a first approach I focused on the seed dispersal system of a typically bird-dispersed tree species, Commiphora harveyi (Burseraceae) in South Africa where a high avian frugivore diversity is found. Results were compared with those of the related species Commiphora guillaumini in Madagascar where a depauperate avian frugivore diversity prevails. I addressed the question of whether regional differences in the diversity of frugivorous birds have an influence on seed dispersal, seedling establishment and distribution and, finally, on the spatial pattern of adult tree populations. In keeping with the higher diversity of frugivorous birds on the African continent, the South African Commiphora species was visited by more animal species than the Malagasy species. Fruits were mainly eaten by birds with the Crowned Hornbill (Tockus alboterminatus) and the Redbilled Woodhoopoe (Phoeniculus purpureus) being the major dispersers in the South African site and the Lesser Vasa Parrot (Coracopsis nigra) in the Malagasy site. Total seed dispersal rates in South Africa were significantly higher compared to Madagascar (70.8% vs. 7.9%) due to a lack of effective dispersers swallowing seeds in Madagascar. Whereas at the South African study site most animals dispersed seeds by swallowing them, at the Malagasy site most animals either dropped the seed under the parent tree or occasionally carried the seeds away in their bills. Seedling establishment was equally high at both sites, however, mortality of newly established seedlings was much higher in Madagascar. Seed dispersal benefit, i.e. the increase in the probability to get established as seedling away from parent trees due to dispersal was much higher in Madagascar (factor 79.9) compared to South Africa (factor 5.6). Corresponding to different seed dispersal rates, seedlings in South Africa were found at relatively large
79 6. GENERAL CONCLUSIONS ...... distances from the nearest Commiphora tree in South Africa (median distance 21.0 m), whereas in Madagascar seedlings were found mostly under and close to the nearest Commiphora tree (median distance 0.9 m). Finally, Commiphora trees in the Malagasy study site were clumped, in the South African study site more randomly distributed. These results suggest that regional differences in frugivore diversity and behavior strongly affect seed dispersal of trees, seedling establishment and the spatial distribution of seedlings and trees. To complement the results obtained from my field data, I developed an individual- based spatially explicit simulation model and tested in computer simulations whether seed dispersal, particularly offspring distribution and dispersal distance, influences the spatial pattern of tree populations. In addition, the influence of breeding system, tree density and the spatial start pattern of the parent population on spatial patterns was investigated. Dispersal distance and tree density were the main factors influencing spatial patterns. Low dispersal distance resulted in highly clumped tree populations and an increase in dispersal distance lead to a decline in clumping towards more randomly spaced populations. Populations with low tree density developed highly clumped patterns, too, and an increase in density lead to a decline in clumping. Breeding system had a medium effect on spatial patterns with dioecious populations usually developing higher clumping than monoecious ones. Offspring distribution and the spatial start pattern had only small effects on the spatial pattern of tree populations. Results from the simulation model demonstrate that dispersal distance as single factor in low- density populations could generate predominantly clumped patterns in tree populations as found in many tropical forests. In a second approach, I investigated seasonality in fruit availability and abundance over the year and determined fruiting peaks and periods of fruit scarcity. The utilization of available resources by the local frugivore community was investigated and possible keystone resources identified. I found 96 plant in fruit, with 58 tree, 28 shrub, seven creeper, two herb, one strangler and one parasite species and a total of 940 individuals. Although some fruit was available all year-round, both, species number and abundance showed a strong seasonality with a fruiting peak for trees just before the rainy season (October, November), and a second more pronounced peak for trees, shrubs and creepers after the rainy season (March to May). Periods of low fruit production occured during the dry winter season (July, August) and during the rainy season (December, January). Of 96 plant species, fruit of 23 species were observed to be eaten by animals, i.e. 18 bird and 2 monkey species, over the whole study
80 6. GENERAL CONCLUSIONS ...... period. Species number and abundance of frugivores corresponded with the main peak of fruiting availability and showed a maximum from February to May with an increase in insectivorous species feeding on fruit. The tree species Ficus thonningii occuring in high density at the study site, fruiting nearly all-year round and producing highest biomass, attracted most animals. This species appeared to be a staple diet during periods of general fruit scarcity and is considered as important keystone resource in South African coastal forests. Other keystone species identified were tree species occuring in low density but, nevertheless, attracting many frugivorous animals during periods of fruit scarcity. The results obtained in this thesis have important implications for conservation. First, the exemplary seed dispersal study showed that low avian frugivore diversity in Madagascar as compared to high diversity in South Africa lead to lower dispersal rates and higher mortality of seedlings under and near the parent trees. Consequently, the spatial pattern of seedlings and trees was influenced as well and it is assumed that it might also affect the population dynamics of the tree populations. This strongly suggests that a reduction of frugivore diversity e.g. due to habitat fragmentation can have profound consequences for animal-dispersed plants and might affect the structure, floristic composition and dynamics of forest ecosystems. Consequences might be a reduced establishment of adult trees and in the worst scenario a local extinction of plant species due to a break-down of seed dispersal and establishment processes. These results stress the importance of shifting the focus of research and conservation from single-species to key processes and interactions in the community. Monitoring of key processes like seed dispersal might be an useful indicator system of ecosystem disturbance, thus providing an important basis for sound management decisions in nature conservation in the future. Second, the phenological study confirmed that fruit production in tropical forests is subject to seasonal fluctuations. Consequently, frugivorous animals face seasonal low fruit availability and are therefore dependant on a few important keystone plant species providing fruits such as fig trees in Oribi Gorge Nature Reserve. The removal of these fig trees e.g. due to habitat destruction or climate change could lead to a failure of the local frugivore community. These results suggest that the identification of keystone resources represents valuable knowledge for conservation, especially in a more and more fragmented landscape.
81 82 7. REFERENCES ......
7. References
Ackerly, DD., JM Rankin-de-Merona, & WA Rodrigues (1990) Tree densities and sex ratios in breeding populations of dioecious Central Amazonian Myristicaceae. Journal of Tropical Ecology 6: 239-248.
Acocks, JPH (1988) Veld types of South Africa. Memoirs of the Botanical Survey of Southern Africa 57: 1-146.
Alexandre, DY (1980) Charactère saisonnier de la fructification dans un forêt hygrophile de Côte d’Ivoire. Revue d’Ecologie (Terre et Vie) 34: 335-350.
Armesto, JJ, JD Mitchell, & C Villagran (1986) A comparison of spatial patterns of trees in some tropical and temperate forests. Biotropica 18: 1-11.
Ashton, PS (1969) Speciation among tropical forest trees: some deductions in the light of recent evidence. Biological Journal of the Linnean Society 1: 155-169.
Ashton, PS (1976) Mixed Dipterocarp forest and its variation with habitat in the Malayan lowlands: a re-evaluation at Pasoh. Malayan Forester 39: 56-65.
Ashton, PS (1984) Biosystematics of tropical forest plants: A problem of rare species. Pp. 497-518 in WF Grant, ed. Plant Biosystematics. Academic Press, Toronto.
Ashton, PS (1988) Dipterocarp biology as a window to the understanding of tropical forest structure. Annual Reviews in Ecology and Systematics 19: 347-370.
Augspurger, CK (1983) Seed dispersal of the tropical tree, Platypodium elegans, and the escape of its seedlings from fungal pathogens. Journal of Ecology 71: 759-771.
Augspurger, CK (1984) Seedling survival of tropical tree species: interactions of dispersal distance, light-gaps, and pathogens. Ecology 65: 1705-1712.
Augspurger, CK, & CK Kelly (1984). Pathogen mortality of tropical tree seedlings: experimental studies of the effects of dispersal distance, seedling density, and light conditions. Oecologia 61: 211-217.
Austin, MP, PS Ashton, & P Greig-Smith (1972) The application of quantitative methods to vegetation survey. III. A re-examination of rain forest data from Brunei. Journal of Ecology 60: 305-324.
83 7. REFERENCES ......
Bates, JM (1992) Frugivory on Bursera microphylla (Burseraceae) by wintering gray vireos (Vireo vicinior, Vireonidae) in the coastal deserts of Sonora, Mexico. The Southwestern Naturalist 37: 252-258.
Bawa, KS (1974) Breeding systems of tree species of a lowland tropical community. Evolution 28: 85-92.
Bawa, KS, & SL Krugman. (1991) Reproductive biology and genetics of tropical trees in relation to conservation and management. Pp. 119-136 in A Gómez-Pompa, TC Whitmore & M Hadley, eds. Rainforest regeneration and management. Unesco + Parthenon Publ. Group Ltd. MAB Series 6.
Bawa, KS., & PA Opler (1975) Dioecism in tropical forest trees. Evolution 29: 167-179.
Becker, P, & M Wong (1985) Seed dispersal, seed predation, and juvenile mortality of Aglaia sp. (Meliaceae) in lowland dipterocarp rainforest. Biotropica 17: 230-237.
Black, GA., T Dobzhansky, & C Pavan (1950) Some attempts to estimate species diversity and population densities of trees in Amazonian forests. Botanical Gazette 111: 413-425.
Bond, WJ (1995) Assessing the risk of plant extinction due to pollinator and disperser failure. Pp. 131-146 in JH Lawton & RM May, eds. Extinction rates. Oxford.
Bond, WJ, & P Slingsby (1984) Collapse of an ant-plant mutualism: The Argentine ant, Iridomyrmex humilis and myrmecochorous Protaceae. Ecology 65: 1031-1037.
Böhning-Gaese, K, BH Gaese, & SB Rabemanantsoa (1995) Seed dispersal by frugivorous tree visitors in the Malagasy tree species Commiphora guillaumini. Ecotropica 1: 41-50.
Böhning-Gaese, K, JF Burkhardt, & J Schmid (1996) Seed dispersal in the tree Commiphora guillaumini: a combination of ornithochory and myrmecochory in a dry tropical forest in western Madagascar. Pp. 305-310 in JU Ganzhorn & J-P Sorg, eds. Ecology and economy of a tropical dry forest in Madagascar. Primate Report Special Issue 46-1.
Böhning-Gaese, K, BH Gaese, & SB Rabemanantsoa (1999) Importance of primary and secondary seed dispersal in the Malagasy tree Commiphora guillaumini. Ecology 80: 821-832.
Borland Delphi 4 (1998) Entwickler-Handbuch. Inprise, Langen.
84 7. REFERENCES ......
Chapman, CA, RW Wrangham, & LJ Chapman (1994) Indices of habitat-wide fruit abundance in tropical forests. Biotropica 26: 160-171.
Clark, DA, & DB Clark (1981) Effects of seed dispersal by animals on the regeneration of Bursera graveolens (Burseraceae) on Santa Fe Island, Galapagos. Oecologia 49: 73-75.
Clark, DA, & DB Clark (1984) Spacing dynamics of a tropical rain forest tree: evaluation of the Janzen-Connell model. The American Naturalist 124: 769-788.
Clark, DB, MW Palmer, & DA Clark (1999) Edaphic factors and the landscape-scale distributions of tropical rain forest trees. Ecology 80: 2662-2675.
Coates Palgrave, K (1995) Trees of Southern Africa. Struik Publishers, Cape Town.
Coates-Estrada, R, & A Estrada (1986) Fruiting and frugivores at a strangler fig in the tropical rain forest of Los Tuxtlas, Mexico. Journal of Tropical Ecology 2: 349-357.
Coates-Estrada, R, & A Estrada (1988) Frugivory and seed dispersal in Cymbopetalum baillonii (Annonaceae) at Los Tuxtlas, Mexico. Journal of Tropical Ecology 4: 157-172.
Coates-Estrada, R, A Estrada, & D Meritt Jr (1993) Foraging by parrots (Amazona autumnalis) on fruits of Stemmadenia donnell-smithii (Apocynaceae) in the tropical rain forest of Los Tuxtlas, Mexico. Journal of Tropical Ecology 9: 121-124.
Connell, JH (1971) On the role of natural enemies in preventing competitive exclusion in some marine animals and in rain forest trees. Pp. 298-310 in PJ Den Boer & GR Gradwell, eds. Dynamics of populations. Centre for Agricultural Publishing and Documentation, Wageningen.
Cooper, KH (1985) The conservation status of indigenous forests in the Transvaal, Natal and O.F.S., South Africa. Wildlife Society of Southern Africa, Durban, South Africa.
Corlett, RT (1998) Frugivory and seed dispersal by vertebrates in the Oriental (Indomalayan) Region. Biological Reviews 73: 413-448.
Corlett, RT, & JV LaFrankie (1998) Potential impacts of climate change on tropical Asian forests through an influence in phenology. Climate Change 39: 439-453. de la Bathie, HP (1946) Flore de Madagascar et des Comores. 106e famille. Burseracées. Tananarive, Madagascar.
85 7. REFERENCES ......
De Steven, D (1994) Tropical tree seedling dynamics: recruitment patterns and their population consequences for three canopy species in Panama. Journal of Tropical Ecology 10: 369-383.
De Steven, D, & FE Putz (1984) Impact of mammals on early recruitment of a tropical canopy tree, Dipteryx panamensis, in Panama. Oikos 43: 207-216.
Dew, JL, & P Wright (1998) Frugivory and seed dispersal by four species of primates in Madagascar’s eastern rain forest. Biotropica 30: 425-437.
Diggle, PJ, J Besag, & JT Gleaves (1976) Statistical analysis of spatial point patterns by means of distance methods. Biometrics 32: 659-667.
Dowsett-Lemaire, F (1988) Fruit choice and seed dissemination by birds and mammals in the evergreen forests of upland Malawi. Revue d’Ecologie (Terre et Vie) 43: 251-281.
Fleming, TH, & ER Heithaus (1981) Frugivorous bats, seed shadows, and the structure of tropical forests. Reproductive Botany: 45-53.
Fleming, TH, R Breitwitsch, & GH Whitesides (1987) Patterns of tropical vertebrate frugivore diversity. Annual Reviews in Ecology and Systematics 18: 91-109.
Fleming, TH, & CF Williams (1990). Phenology, seed dispersal, and recruitment in Cecropia peltata (Moraceae) in Costa Rican tropical dry forest. Journal of Tropical Ecology 6:163-178.
Forman, RTT, & DC Hahn (1980) Spatial patterns of trees in a Carribean semievergreen forest. Ecology 61:1267-1274.
Foster, RB (1982a) The seasonal rythm of fruit fall on Barro Colorado Island. Pp. 151-172 in ED Leigh, AS Rand, & DM Windsor, eds. The ecology of a tropical forest: seasonal rythms and long term changes. Smithsonian Institution Press, Washington DC.
Foster, RB (1982b) Famine on Barro Colorado island. Pp. 201-212 in ED Leigh, AS Rand, & DM Windsor, eds. The ecology of a tropical forest: seasonal rythms and long term changes. Smithsonian Institution Press, Washington DC.
Frankie, GW, HG Baker, & PA Opler (1974) Comparative phenological studies of trees in tropical wet and dry forests in the lowland of Costa Rica. Journal of Ecology 62: 881- 919.
86 7. REFERENCES ......
Frost, PGH (1980) Fruit-frugivore interactions in a South African Coastal Dune forest. Acta XVII. Congressus International Ornithologici: 1179-1184.
Futuyma, D (1973) Community structure and stability in constant environments. The American Naturalist 107: 443-446.
Ganzhorn, JU, & J-P Sorg, eds (1996) Ecology and economy of a tropical dry forest in Madagascar. Primate Report Special Issue 46-1.
Ganzhorn, JU, J Fietz, E Rakotovao, D Schwab, & D Zinner (1999) Lemurs and the regeneration of dry deciduous forest in Madagascar. Conservation Biology 13: 1-12.
Gautier-Hion, A, J-M Duplantier, R Quris, F Feer, C Sourd, J-P Decoux, G Dubost, L Emmons, C Erard, P Hecketsweiler, A Moungazi, C Roussilhon, & J-M Thiollay (1985) Fruit characters as a basis of fruit choice and seed dispersal in a tropical forest vertebrate community. Oecologia 65: 324-337.
Gautier-Hion, A, & G Michaloud (1989) Are figs always keystone resources for tropical frugivorous vertebrates ? A test in Gabon. Ecology 70: 1826-1833.
Gilbert, LE (1980) Food web organization and the conservation of neotropical diversity. Pp. 11-33 in ME Soulé & BA Wilcox, eds. Conservation biology: an evolutionary- ecological perspective. Sinauer Associates, Sunderland, Massachusetts.
Glen, HF (1996) A description of the vegetation of Oribi Gorge Nature Reserve, Natal, Part I. Trees in South Africa 46: 18-27.
Goodman, SM, & JU Ganzhorn (1997) Rarity of figs (Ficus) on Madagascar and its relationship to a depauperate frugivore community. Revue d'Ecologie - la Terre et la Vie 52: 321-329.
Goodman, SM, JU Ganzhorn, & L Wilmé (1997) Observations at a Ficus tree in Malagasy Humid Forest. Biotropica 29:480-488.
Goodman, SM, & LMA Rakotozafy (1997) Subfossil birds from coastal sites in western and southwestern Madagascar. Pp. 257-279 in Goodman, SM & BD Patterson, eds. Natural change and human impact in Madagascar. Washington.
87 7. REFERENCES ......
Greenberg, R, MS Foster, & L Marquez-Valdelamar (1995) The role of the White-eyed Vireo in the dispersal of Bursera fruit on the Yucatan Peninsula. Journal of Tropical Ecology 11: 619-639.
Greig-Smith, P (1983) Quantitative plant ecology. 3rd ed. Blackwell, London.
Hall, JB, & MD Swaine (1981) Distribution and ecology of vascular plants in a tropical rain forest: forest vegetation in Ghana. W.Junk, The Hague.
Hamilton, MB (1999) Tropical tree gene flow and seed dispersal. Nature 401:129.
Harcourt, C, & J Thornback (1990) Lemurs of Madagascar and the Comores. IUCN, Gland.
Harper, JL (1977) Population biology of plants. Academic Press, London.
Herrera, CM (1985) Determinants of plant-animal coevolution: the case of mutualistic dispersal of seeds by vertebrates. Oikos 44: 132-141.
Higgins, ML (1979) Intensity of seed predation on Brosimum utile by Mealy Parrots (Amazona farinosa). Biotropica 11: 80.
Houle, G (1998) Seed dispersal and seedling recruitment of Betula alleghaniensis: spatial inconsistency in time. Ecology 79: 807-818.
Howe, HF (1977) Bird activity and seed-dispersal of a tropical wet forest tree. Ecology 58: 539-550.
Howe, HF (1981) Dispersal of a Neotropical nutmeg (Virola sebifera) by birds. Auk 98: 88- 98.
Howe, HF (1984a) Implications of seed dispersal by animals for tropical reserve management. Biological Conservation 30: 261-281.
Howe, HF (1984b) Constraints on the evolution of mutualism. The American Naturalist 123: 764-777.
Howe, HF (1986) Seed dispersal by fruit-eating birds and mammals. Pp. 123-189 in DR Murray, ed. Seed dispersal. Academic Press, Australia.
Howe, HF (1989) Scatter- and clump-dispersal and seedling demography: hypothesis and implications. Oecologia 79: 417-426.
88 7. REFERENCES ......
Howe, HF., & GF Estabrook (1977) On intraspecific competition for avian dispersers in tropical trees. The American Naturalist 111: 817-832.
Howe, HF, & GA Vande Kerckhove (1979) Fecundity and seed dispersal of a tropical tree. Ecology 60: 180-189.
Howe, HF, & D De Steven (1979) Fruit production, migrant bird visitation, and seed dispersal of Guarea glabra in Panama. Oecologia 39: 185-196.
Howe, HF, & GA Vande Kerckhove (1981) Removal of wild nutmeg (Virola surinamensis) crops by birds. Ecology 62: 1093-1106.
Howe, HF, & J Smallwood (1982) Ecology of seed dispersal. Annual Reviews in Ecology and Systematics 13: 201-228.
Howe, HF, EW Schupp, & LC Westley (1985) Early consequences of seed dispersal for a neotropical tree (Virola surinamensis). Ecology 66: 781-791.
Howe, HF, & LC Westley (1988) Ecological relationships of plants and animals. Oxford University Press.
Hubbell, SP (1979) Tree dispersion, abundance, and diversity in a tropical dry forest. Science 203: 1299-1309.
Hubbell, SP, & RB Foster (1983) Diversity of canopy trees in a neotropical forest and implications for conservation. Pp. 25-41 in SL Sutton, TC Whitmore & AC Chadwick, eds. The tropical rain forest: ecology and management. Blackwell, Oxford.
Hubbell, SP, & RB Foster (1986) Commonness and rarity in a neotropical forest: Implications for tropical tree conservation. Pp. 205-231 in ME Soulé, ed. Conservation biology: The science of scarcity and diversity. Sinauer Associates, Sunderland, Massachusetts.
Hurlbert, SH (1979) Spatial distribution of the montane unicorn. Oikos 58:257-271.
Janzen, DH (1970) Herbivores and the number of tree species in tropical forests. The American Naturalist 104: 501-528.
Janzen, DH (1971) Seed predation by animals. Annual Review of Ecology and Systematics 2: 465-492.
Janzen, DH (1979) How to be a fig. Annual Reviews of Ecology and Systematics 10: 13-51.
89 7. REFERENCES ......
Janzen, DH, GA Miller, J Hackforth-Jones, CM Pond, K Hooper, & DP Janos (1976) Two Costa-Rican bat-generated seed shadows of Andira inermis (Leguminosae). Ecology 57: 1068-1075.
JMP (1995) Statistics and Graphics Guide. Version 3.1. SAS Institute Inc., Cary, NC, USA.
Jordano, P (1987a) Avian fruit removal: effects of fruit variation, crop size, and insect damage. Ecology 68: 1711-1723.
Jordano, P (1987b) Frugivory, external morphology and digestive system in Mediterranean sylviid warblers Sylvia spp. Ibis 129: 175-189.
Juillot, C (1997) Impact of seed dispersal by red howler monkeys Alouatta seniculus on the seedling population in the understorey of tropical rain forest. Journal of Ecology 85: 431-440.
Karr, JR (1976) Seasonality, resource availability, and community diversity in tropical bird communities. The American Naturalist 110: 973-994.
Kemp, AC (1995) The hornbills: Bucerotiformes. Bird fanilies of the world. Oxford University Press, Oxford.
Knight, RS, & WR Siegfried (1983) Inter-relationships between type, size and colour of fruits and dispersal in Southern African Trees. Oecologia 56: 405-412.
Lambert, FR, & AG Marshall (1991) Keystone characteristics of bird-dispersed Ficus in a Malaysian lowland rain forest. Journal of Ecology 79: 793-809.
Lang, GE, DH Knight, & DA Anderson (1971) Sampling the density of tree species with quadrats in a species-rich tropical forest. Forest Science 17: 395-400.
Langrand, O (1990) Birds of Madagascar. Yale University Press, New Haven, Connecticut, USA.
Laska, MS, & EW Stiles (1994) Effects of fruit crop size on intensity of fruit removal in Viburnum prunifolium (Caprifoliaceae). Oikos 69: 199-202.
Laurance, WF, & RO Bieregaard, eds (1997) Tropical forest remnants ecology, management, and conservation of fragmented communities. University of Chicago Press.
90 7. REFERENCES ......
Lawes, MJ (1991) Diet of samango monkeys (Cercopithecus mitis erythrarchus) in the Cape Vidal dune forest, South Africa. Journal of Zoology 224: 149-173.
Leighton, M, & DR Leighton (1983) Vertebrate responses to fruiting seasonality within a Borneon rain forest. Pp. 181-196 in SL Sutton, TC Whitmore, & AC Chadwick, eds. Tropical rain forest: ecology and management. British Ecological Society, special publication no. 2. Blackwell Scientific Publication, Oxford.
Levey, DJ (1987) Seed size and fruit-handling techniques of avian frugivores. The American Naturalist 129: 471-485.
Lieberman, D (1982) Seasonality and phenology in a dry tropical forest in Ghana. Journal of Ecology 70: 791-806.
Ludwig, JA, & JF Reynolds (1988) Statistical Ecology. Wiley & Sons, New York.
Maclean, GL (1993) Roberts‘ birds of Southern Africa. New Holland Publishers, London.
May, R (1973) Stability and complexity in model ecosystems. Princeton University Press, Princeton, New Jersey.
Meffe, GK, & CR Carroll (1994) Principles of conservation biology. Sinauer Associates, Sunderland, Massachusetts.
Menge, BA, EL Berlow, CA Blanchette, SA Navarrete, & SB Yamada (1994) The keystone species concept: variation in interaction strength in a rocky intertidal habitat. Ecological Monographs 64: 249-286.
Morisita, M (1959) Measuring of the dispersion of individual and analysis of the distributional patterns. Memoirs Faculty Science, Kyushu University Series F. (Biol.) 2: 215-235.
Murali, KS, & R Sukumar (1994) Reproductive phenology of a tropical dry forest in Mudumalai, southern India. Journal of Ecology 82: 759-767.
Nanami, S, H Kawaguchi, & T Yamakura (1999) Dioecy-induced spatial patterns of two codominant tree species, Podocarpus nagi and Neolitsea aciculata. Journal of Ecology 87:678-687.
O’Brian, TG (1997) Behavioural ecology of the North Sulawesi Tarictic Hornbill Penelopides exarhatus exarhatus during the breeding season. Ibis 139: 97-101.
91 7. REFERENCES ......
Orians, G (1969) The number of bird species in some tropical forests. Ecology 50: 783-801
Opler, PA, GW Frankie, & HG Baker (1980) Comparative phenological studies of treelet and shrub species in tropical wet and dry forests in the lowlands of Costa Rica. Journal of Ecology 68: 167-188.
Parthasarathy, N, & R Karthiekeyan (1997). Plant biodiversity inventory and conservation of two tropical dry evergreen forests on the Coromandel coast, South India. Biodiversity and Conservation 6: 1063-1083.
Pielou, EC (1969) An introduction to mathematical ecology. Wiley, New York.
Pitman, NCA, J Terborgh, MR Silman, & P Nuez (1999) Tree species distributions in an upper Amazonian forest. Ecology 80: 2651-2661.
Pizo, MA (1997) Seed dispersal and predation in two populations of Cabralea canjerana (Meliaceae) in the Atlantic Forest of southeastern Brazil. Journal of Tropical Ecology 13: 559-578.
Pooley, E (1994) The complete field guide to the trees of Natal, Zululand & Transkei. Natal Flora Publications Trust, Durban, South Africa.
Pooley, E (1998) A field guide to wild flowers. KwaZulu-Natal and the Eastern Region. Natal Flora Publications Trust, Durban, South Africa.
Poore, MED (1968) Studies in Malaysian rainforest 1. The forest on triassic sediments in Jengka forest reserve. Journal of Ecology 56: 143-196.
Portnoy, S, & MF Willson (1993) Seed dispersal curves: behavior of the tail of the distribution. Evolutionary Ecology 7: 25-44.
Power, ME, & LS Mills (1995) The keystone cops meet in Hilo. Trends in Ecology and Evolution 10: 182-184.
Pratt, TK, & EW Stiles (1985) The influence of fruit size and structure on composition of frugivore assemblages in New Guinea. Biotropica 17: 314-321.
Pui, H, A Fabre, & T Gauquelin (1998) Spatial distribution of seedlings and young plants of Iryanthera hostmannii (Benth.) Warb. in French Guyana tropical rain forest. Comptes Rendus de L’Academie des Sciences Serie III - Sciences de la Vie 321: 429-435.
92 7. REFERENCES ......
Raemakers, JJ, FPG Aldrich-Blake, & JF Payne (1980) The forest. Pp. 26-29 in CJ Chivers, ed. Malayan forest primates. Plenum , New York.
Ricklefs, RE (1997) Comparative demography of new world populations of thrushes (Turdus spp.). Ecological Monographs 67: 23-43.
Rohner, U, & J-P Sorg (1986) Observations phénologiques en foret dense sèche, Vol. 1. Fiche technique of the CFPF, No. 12. Centre de Formation Professionelle Forestière, Morondava, Madagascar.
Santos, T, & JL Telleria (1994) Influence of forest fragmentation on seed consumption and dispersal of Spanish juniper Juniperus thurifera. Biological Conservation 70: 129-134.
Santos, T, & JL Telleria (1997) Vertebrate predation on Holm Oak, Quercus ilex, acorns in a fragmented habitat: effects on seedling recruitment. Forest Ecology and Management 98: 181-187.
Scharfe, F, & W Schlund (1996) Seed removal by lemurs in a dry deciduous forest of western Madagascar. Pp. 295-304 in JU Ganzhorn & J-P Sorg, eds. Ecology and economy of a tropical dry forest in Madagascar. Primate Report Special Issue 46-1.
Schupp, EW (1988) Seed and early seedling predation in the forest understory and in treefall gaps. Oikos 51: 71-78.
SigmaPlot (1997) User’s Manual. Version 4.0. SPSS Inc., Chicago, USA.
Sorg, J-P, & U Rohner (1996) Climate and tree phenology of the dry deciduous forest of the Kirindy Forest. Pp. 57-80 in JU Ganzhorn & J-P Sorg, eds. Ecology and economy of a tropical dry forest in Madagascar. Primate Report Special Issue 46-1.
Strasberg, D (1996) Diversity, size composition and spatial aggregation among trees on a 1-ha rain forest plot at La Réunion. Biodiversity and Conservation 5: 825-840.
Stuart, C, & S Stuart (1995) Field guide to the mammals of Southern Africa. Struik Publishers, Cape Town.
Sun, C, TC Moermond, & TJ Givnish (1997) Nutritional determinants of diets in three touracos in a tropical montane forest. The Auk 114: 200-211.
93 7. REFERENCES ......
Terborgh, J (1986a) Keystone plant resources in the tropical forest. Pp. 330-344 in ME Soulé, ed. Conservation biology: The source of scarcity and diversity. Sinauer Associates, Sunderland, Massachusetts.
Terborgh, J (1986b) Community aspects of frugivory in tropical forests. Pp. 371-384 in A Estrada & TH Fleming, eds. Frugivores and seed dispersal. Sinauer Associates, Sunderland, Massachusetts.
Terborgh, J, & B Winter (1980) Some causes of extinction. Pp. 119-133 in ME Soulé & BA Wilcox, eds. Conservation biology: an evolutionary ecological perspective. Sinauer Associates, Sunderland, Massachusetts.
Thomas, RJ, CGA Marshall, MK Watkeys, FJ Fitch, & JA Miller (1992) K-Ar and 40AR/39Ar dating of the Natal Group, Southeast Africa: a post Pan-African molasse ? Journal of African Earth Science 15: 453-471.
Thompson, JN, & MF Willson (1979) Evolution of temperate fruit / bird interactions: phenological strategies. Evolution 33: 973-982.
Tilman, D, CL Lehman, & P Kareiva (1997) Population dynamics in spatial habitats. Pp. 3-20 in D Tilman & P Kareiva, eds. Spatial ecology. The role of space in population dynamics and interspecific interactions. Princeton University Press, Princeton, New Jersey.
Turner, IM (1996) Species loss in fragments of tropical rain forest: a review of the evidence. Journal of Applied Ecology 33: 200-219. van der Walt, JJA (1973) The South African species of Commiphora. Bothalia 11: 53-102. van der Walt, JJA (1986) Burseraceae. Flora of Southern Africa. 18: 5-34.
Van Wyk, B, & P Van Wyk (1997) Field guide to trees of Southern Africa. Struik Publishers, Cape Town.
Wheelwright, NT (1985) Fruit size, gape width, and the diets of fruit-eating birds. Ecology 66: 808-818.
Wheelwright, NT (1993) Fruit size in a tropical tree species: variation, preferences by birds, and heritability. Vegetatio 107/108: 163-174.
94 7. REFERENCES ......
Wheelwright, NT, & GH Orians (1982) Seed dispersal by animals: contrasts with pollen dispersal, problems of terminology, and constraints on coevolution. The American Naturalist 119: 402-413.
White, LJT (1994) Patterns of fruit-fall phenology in the Lopé Reserve, Gabon. Journal of Tropical Ecology 10: 289-312.
Whitney, KD, MK Fogiel, AM Lamperti, KM Holbrook, DJ Stauffer, BD Hardesty, VD Parker, & TB Smith (1998) Seed dispersal by Ceratogymna hornbills in the Dja Reserve, Cameroon. Journal of Tropical Ecology 14: 351-371.
Willson, MF (1992) The ecology of seed dispersal. Pp. 61-85 in M Fenner, ed. Seeds. The ecology of regeneration in plant communities. CAB International, Wallingford, U.K.
Willson, MF (1993) Dispersal mode, seed shadows, and colonization patterns. Vegetatio 107 / 108: 261-280.
Wright, SJ (1983) The dispersion of eggs by a bruchid beetle among Scheelea palm seeds and the effect of distance to the parent palm. Ecology 64: 1016-1021.
Young, A, T Boyle, & T Brown (1996) The population genetic consequences of habitat fragmentation for plants. Trends in Ecology and Evolution 11: 413-418.
95 . 8. A 8. Appendix PPENDIX List of plant species in fruit from July 1997 to July 1998 with respective family. Authorities after Pooley (1994, 1998). Given for each species are life form (H herb, SH shrub, ST small tree, MT medium-sized tree, TT tall tree, SF strangler fig, CR creeper, P parasite), density per hectare, fruiting period as the period when the plant had ripe fruits, the number of months the plant species was observed bearing ripe fruit, the note whether the species fruited during periods of general fruit scarcity (i.e. July 1997 to August 1997, November 1997 to December 1997, July 1998)(+) or outside fruit scarcity (-), median fruit crop, the note whether the fruit was arillate, a fig or other and observations of animals eating fruit. Additional observations of animals at other occasions during the study period are given in brackets. bb = Blackcollared Barbet, bs = Blackbellied Starling, bu = Blackeyed Bulbul, ch = Crowned Hornbill, co = Grey Cuckooshrike, fd = Forktailed Drongo, fw = Forest Weaver, gp = Green Pigeon, gs = Glossy Starling, lo = Knysna Lourie, or = Blackheaded Oriole, ot = Olive Thrush, rs = Redwinged Starling, rtb = Redfronted Tinker Barbet, sa = Samango Monkey, sb = Sombre Bulbul, st = Spotted Thrush, th = Trumpeter Hornbill, ti = Southern Black Tit, tw = Thickbilled Weaver, we = White-eye, wh = Redbilled Woodhoopoe. Plant species Life Den- Fruiting No.of Fruit Fruit Fruit Fruits eaten by form sity period months during crop type animals / ha fruiting scarcity
96 AMARYLLIDACEAE Clivia miniata H 0.19 aug 1.5 + 6 other - ANACARDIACEAE Harpephyllum caffrum Bernh. MT-TT 1.38 jul-jan 9.5 + 550 other lo,sa,th Protorhus longifolia (Bernh.) Engl. MT-TT 0.56 oct-jan 3.0 + 5500 other bs,ch,rs Rhus chirindensis Bak.f. SH,ST 0.31 feb-may 5.0 - 550 other (we) Rhus gueinzi Sond. ST 0.31 jul-oct 4.0 + 5500 other - Rhus lucida L. SH 0.19 nov-jan 2.5 + 5500 other (we) Sclerocarya birrea(A.Richt.) Hochst. ST-MT 0.38 mar-apr 1.5 - 303 other - ANNONACEAE Monanthotaxis caffra (Sond.) Verdc. SH 1.32 mar-may 3.0 - 6 other - Uvaria caffra Benth. SH 0.13 apr 0.5 - 6 other - APOCYNACEAE Acokanthera oppositifolia (Lam.) Codd ST 0.06 jan 0.5 - 55 other - Carissa bispinosa (L.) Desf. ex Brenan ST 0.13 aug,mar,jun 1.5 + 55 other - ARALIACEAE Cussonia nicholsonii Strey MT-TT 0.56 jul-oct,jun-jul 5.0 + 5500 other th Schefflera umbellifera (Sond.) Baill. TT 0.19 aug-sep,jul 2.5 + 55000 other lo,rs,(th) Plant species Life Den- Fruiting No.of Fruit Fruit Fruit Fruits eaten by form sity period months during crop type animals / ha fruiting scarcity ARECACEAE Phoenix reclinata Jacq. MT-TT 0.56 mar-jul 4.5 - 550 other sa,ot ASPARAGACEAE Asparagus falcatus CR 0.19 mar-jul 2.5 + 550 other - ASTERACEAE Chrysantemoides monilifera SH 0.38 may-jul 3.0 + 303 other - subsp. rotundata (L.)T.Norl. BURSERACEAE Commiphora harveyi (Engl.) Engl. ST-TT 3.64 feb-jul 6.0 - 550 aril bb,bs,ch,fd,rs,ti,wh (bu),(fw),(or),(sa) Commiphora woodii Engl. MT-TT 0.31 feb-may 3.0 - 55 aril bb,or,wh 97 CAPPARACEAE Bachmannia woodii (Oliv.) Gilg ST 0.06 may 0.5 - 55 other - Capparis tomentosa Lam. CR 0.56 feb-may 4.5 - 55 other - Maerua racemulosa (A.DC.) Gilg Ben. SH,ST 0.88 sep-dec 3.0 + 55 other - CELASTRACEAE Maytenus peduncularis (Sond.) Loes. SH 0.06 jul,oct 1.0 + 55 aril - Pterocelastrus tricuspidatus (Lam.) Sond. SH 0.06 feb-mar 1.5 - 55 other - CLUSIACEAE Garcinia gerrardii Harv. ex Sim TT 0.06 may 1.0 - 55 other - CUCURBITACEAE Coccinia palmata CR 1.44 jul-sep,nov-jul 8.5 + 6 other (fw) DRACAENACEAE Dracaena aletriformis (Haw.) Bos ST 0.88 mar-jun 4.0 - 55 other - 8. A
EBENACEAE . Diospyros whyteana (Hiern) F.White SH 0.25 oct-mar 3.5 + 55 other - Euclea natalensis A.DC. ST-MT 0.31 mar-jul 4.5 - 5500 other - PPENDIX ERYTHROXYLACEAE Erythroxylum pictum E.Mey. ex Sond. ST 1.5 nov-jul 8.5 - 55 other - . 8. A Appendix (continued) PPENDIX Plant species Life Den- Fruiting No.of Fruit Fruit Fruit Fruits eaten by form sity period months during crop type animals / ha fruiting scarcity EUPHORBIACEAE Bridelia micrantha (Hochst.) Baill. ST-T 0.13 mar 1.0 - 303 other - Croton sylvaticus Muell.Arg. MT-TT 0.38 jul,jan-may 5.5 - 550 other ch,lo Drypetes arguta (Muell.Arg.) Hutch. ST 0.82 mar-jun 4.0 - 55 other - Macaranga capensis (Baill.) Benth. ex Sim TT 0.13 may-jul 2.5 + 55000 other bu,lo,sa,th,we Phyllanthus myrtaceus SH 0.06 may 1.0 - 5500 other lo Sapium integerrimum (Hochst.) J. Leonard TT 0.13 jun,jul 1.5 + 303 other (ch) Suregada africana (Sond.) Kuntze ST 0.06 sep 1.0 - 550 other - FLACOURTIACEAE Xylotheca kraussiana Hochst. ST 0.38 feb-mar 1.5 - 55 aril - ICACINACEAE
98 Apodytes dimidiata E.Mey.ex.Arn MT-TT 0.25 mar-jun 3.0 - 550 aril - Apodytes abbottii SH 0.38 jan,mar,apr-may 2.5 - 55 aril - IRIDACEAE Crocosmia aurea H 0.06 apr 0.5 - 6 other - LAURACEAE Cryptocarya latifolia Sond. MT-TT 0.13 feb-mar 2.0 - 550 other - Cryptocarya myrtifolia Stapf MT-TT 0.13 aug-oct,jan-feb 3.5 + 550 other lo,th Cryptocarya woodii Engl. ST 0.06 mar 0.5 - 55 other - Cryptocarya wyliei Stapf SH 4.45 jul-aug,jan-jul 9.0 + 6 other - LOGANIACEAE Strychnos mitis S.Moore ST-MT 0.19 feb-mar 1.5 - 550 other - LORANTHACEAE Erianthemum dregei P 0.06 aug-oct,dec,may-jul 5.5 + 550 other - MELASTOMATACEAE Memecylon natalensis Markg. ST 0.06 aug 0.5 + 55 other - MELIACEAE Ekebergia capensis Sparrm. MT-TT 0.19 mar-jun 4.0 - 550 other lo,th Ekebergia pterophylla (C.DC.) Hofmeyr ST-MT 0.75 mar-may 3.0 - 550 other - Plant species Life Den- Fruiting No.of Fruit Fruit Fruit Fruits eaten by form sity period months during crop type animals / ha fruiting scarcity MELIACEAE Trichilia dregeana Sond. TT 0.25 aug-oct,jul 3.0 + 550 aril ch,th,(gp) Turrea floribunda Hochst. SH,ST 0.31 mar-jun 2.5 - 55 other - MELIANTHACEAE Bersama swinnyi Phill. MT-TT 0.25 june-jul 1.5 + 55 aril ch MORACEAE Ficus burtt-davyi Hutch. SF 1.19 all year 7.5 + 550 fig lo Ficus ingens (Miq.) Miq. ST-MT 0.31 sep-oct,feb,jun-jul 4.0 + 5500 fig - Ficus sur Forssk. MT-TT 0.44 jul-sep,nov-dec, 7.5 + 550 fig lo,gs,bb,(th) may-jul 99 Ficus thonningii Bl. MT-TT 2.76 jul-sep,dec-jan, 9.0 + 5500 fig co,bb,bs,bu,fd,lo, mar-jul or,rtb,sa,sb,th,ti, we,(gs),(ot),(rs) MYRSINACEAE Rapanea melanophloeos (L.) Mez ST 0.06 - ? ? 5500 other - MYRTACEAE Syzygium cordatum Hochst. ST-MT 0.13 jul 0.5 + 303 other - OCHNACEAE Ochna arborea Burch. Ex DC. ST 0.38 aug-sep 1.5 - 55 other - Ochna serrulata (Hochst.) Walp. ST 0.19 sep-oct 2.0 - 55 other - OLEACEAE Olea capensis subs. enervis (Harv. Ex SH 0.88 apr-may 1.0 - 550 other lo C.H.Wr.) Verdoorn 8. A PHYTOLACCACEAE .
Phytolacca dodecandra L`Herit. CR 0.25 aug-oct 2.0 + 550 other - PPENDIX RHIZOPHORACEAE Cassipourea gummiflua Tul. MT-TT 0.94 jul-sep,apr-jun 4.0 + 0.8 aril ot,sa,we . 8. A Appendix (continued) PPENDIX Plant species Life Den- Fruiting No.of Fruit Fruit Fruit Fruits eaten by form sity period months during crop type animals / ha fruiting scarcity RUBIACEAE Burchellia bubalina (L.f.) Sims. SH,ST 2.13 all year 13.0 + 55 other - Canthium inerme (L.f.) Kuntze ST 0.25 dec-jan,apr-may 2.5 - 550 other - Gardenia thunbergia L.f. ST 0.13 - ? ? 30 other - Hyperacanthus amoenus (Sims) Bridson ST 0.5 - ? ? 30 other - Psychotria capensis (Eckl) Vatke SH, ST 3.51 dec-may 5.5 - 550 other - Rothmannia globosa (Hochst.) Keay ST-MT 1 jan-may 4.5 - 55 other - Tricalysia capensis (Meisn. ex Hochst.) Sim SH,ST 0.82 aug-sep,jan-feb, 4.0 - 55 other - mar,apr,may RUTACEAE Calodendrum capense (L.f.) Thunb. TT 0.13 - ? ? 550 other - 100 Teclea gerrardii Verdoorn ST-MT 0.63 jan-jun 5.5 - 550 other - Vepris lanceolata (Lam.) G.Don MT 0.13 mar-may 2.0 - 3025 other - SANTALACEAE Colpoon compressum Berg. SH 0.13 jul-aug 1.5 + 30 other - SAPINDACEAE Allophyllus dregeanus (Sond.) De Winter ST 0.06 jun-jul 1.0 + 550 other - Hippobromus pauciflorus (L.f.) Radlk. ST 0.13 oct-nov 1.0 + 550 other - SAPOTACEAE Englerophytum natalense Krause ST-MT 4.14 sep-nov,mar-may 4.5 - 55 other - Mimusops obovata Sonder ST-MT 0.31 nov-jan,mar-apr,jul 4.0 + 55 other - Sideroxylon inerme L. ST 0.13 nov-dec 1.5 + 303 other - Vitellariopsis marginata (N.E.Br.)Aubrev. MT) 0.13 aug-feb 7.0 + 55 other - SCROPHULARIACEAE Halleria lucida L. SH,ST 1.32 aug-nov,jul 5.0 + 55 other rtb SMILACEAE Smilax anceps CR 0.31 june-jul 1.5 + 55 other - Plant species Life Den- Fruiting No.of Fruit Fruit Fruit Fruits eaten by form sity period months during crop type animals / ha fruiting scarcity SOLANACEAE Solanum aculeastrum Dun. SH 1.44 all year 12.0 + 55 other - Solanum coccineum SH 1.19 all year 13.0 + 55 other - Solanum diplo-sinuatum SH 0.19 aug-oct 3.0 + 6 other - Solanum giganteum Jacq. SH,ST 0.13 dec,mar,jun-jul 3.0 + 550 other - THYMELACEAE Peddiea africana Harv. ST 0.38 jul,sep-dec,mar-jul 8.5 + 6 other - TILIACEAE Grewia lasioscarpa E.Mey. ex Harv. SH,ST 0.06 jul-aug 1.0 + 55 other -
101 Grewia occidentalis L. SH,ST 1.07 jan-apr 3.0 - 550 other - ULMACEAE Celtis africana N.L.Burm. MT-TT 2.19 dec-may 5.5 - 550 other ch,rs Chaetachme aristata Plach. MT-TT 1.00 jul-dec,may-jul 11.5 + 550 other lo,st,tw Trema orientalis (L.) Bl. ST 0.31 apr-jul 3.5 - 550 other we VERBENACEAE Clerodendrum glabrum E.Mey. ST 0.06 - ? - 6 other - VITACEAE Cyphostemma hypoleucum CR 0.82 jan,mar-may 3.0 - 55 other - Rhoicissus rhomboidea (E.Mey. ex Harv.) CR 0.19 Jun 1.0 - 55 other - Planch. Rhoicissus tridentata (L.f.) Wild & Drum. SH 0.06 Mar 0.5 - 550 other (lo) 8. A . PPENDIX 102 9. ACKNOWLEDGEMENTS ......
9. Acknowledgements
This study would not have been possible without the help and support of many people. Dr. Katrin Böhning-Gaese was an excellent teacher giving continuous support. She introduced me to the scientific world; I learned much from her about research, data analysis, teaching, and German academia. She taught me to be critical, and how to stay focused on the main points and not get distracted by small things. I could not have wished for a better teacher and supervisor. Prof. Dr. Hermann Wagner kindly gave me the opportunity to complete this thesis in his institute and let me use all the institute`s facilities. Prof. Dr. Ingolf Schuphan kindly took over the referat of the thesis. I am grateful to the Deutscher Akademischer Austauschdienst and the Deutsche Forschungsgemeinschaft (Bo 1221/7-1) for financing this research. I thank my collegue and friend Reik Oberrath, who shared euphory and doubts on the scientific world, and on writing simulation models and papers with me. Laughing with him was the best stress reliever. He together with Nicole Lemoine provided a fine working atmosphere. All my collegues at the Institute of Biology II in Aachen provided an easy working atmosphere and always lots of coffee, chocolate and help in all kinds of difficulties. I enjoyed working there very much. I am thankful to Tony Abbott, Helga Gaube, Hugh Glen, David Johnson, Mike Lawes, Nicole Lemoine, Doug Levey, Jörg Lippert, Reik Oberrath, Christina Potgieter and Liliane Ruess for many constructive comments on parts of the thesis. Sandra Brill and Anja Johnen helped with graphical problems, Marianne Dohms with photographs. Help from Hans- Joachim Poethke, Thomas Hovestadt, Udo Hommen, Silke Hein and Birgit Seifert was essential during development of the simulation model at the "Modelling and Simulation in Ecology and Conservation" workshop in Pesina. I am grateful to the KwaZulu - Natal Nature Conservation Service, South Africa, for the permission to work in Oribi Gorge and overall support. My KZNNCS supervisor and friend David Johnson shared his extensive knowledge on trees and birds with me. He and his wife Sally always made me feel welcome. Mornè du Plessis helped to get the field study started. Fieldwork benefited greatly from the support of Nigel and Sue Anderson and the game
103 9. ACKNOWLEDGEMENTS ...... guards at Oribi Gorge, especially Julius and Jabulane. Christina Potgieter helped tremendously with phenology monitoring and moral support in the field. I am very much indepted to the families in and arround Paddock, South Africa, for continuous support and friendship. My deepest thanks goes to the Buhrs - to Helga, Walter, Goma, Sonia, Nadine, Michele & Karen - for adopting me in their family. They nurtured me after bee attacks and lots of other small and big crises in the field and always made me feel at home. Lynne and Errol Milton always had a cup of tea after field work; I learned much about baby birds and other creatures from Lynne. Lots of thanks also go to William Allison, Di & Tony Browne, Yvonne & Martin Fronnemann, Bill Hocking, Cathy Key, Grace & Gus Robinson, and to Rieke & Kurt-Günter Tiedemann. I also thank Mark-Oliver, Roland, Romy & Tobias, Sabine, Steve, Tina, Ursula and Uli for their all-enduring friendship. As for Phil, only Phil knows what he is owed. Finally, I thank my parents for their love and support.
104 10. CURRICULUM VITAE ......
10. Curriculum vitae
08.01.1967 geboren in Urach, Baden-Württemberg
1973 - 1977 Grundschule Urach
1977 - 1986 Graf-Eberhard-Gymnasium, Bad Urach
1986 Abitur
1986 - 1995 Studium der Biologie an der Eberhard-Karls-Universität Tübingen
1994 -1995 Diplomarbeit in der Abteilung für Verhaltensphysiologie,
Universität Tübingen
1995 Abschluß zur Diplom-Biologin
1996 - 2000 Promotion am Institut für Biologie II, Lehrstuhl für Zoologie / Tier-
physiologie, RWTH Aachen
1997 - 1999 im Rahmen der Promotion zweijähriger Aufenthalt in Südafrika mit
Hilfe eines Doktorandenstipendiums vom Deutschen Akademischen
Austauschdienst
105