Multi-species pollination interactions in a Kenyan savannah ecosystem

Research output: ThesisDoctoral Thesis


Previous work on African acacias has shown that co-flowering species (those that flower in the same place at the same time) partition the activity of shared pollinators, and so avoid competition for pollination. The main aim of this thesis is to assess the evidence for temporal structuring of pollinator activity at a second African site, Mpala, in north central Kenya. I address this issue both for a guild of acacia species, and go beyond previous work to examine daily temporal patterning at the level of flowering communities. This second approach involves the generation of pollination webs for different periods of time within a single day, and the use of null modelling to compare temporal structure in real and randomised visitation data. I replicated this approach over sites and seasons, and generated the first pollination webs for any African savannah habitat.
Analysis of patterns within acacias requires identification of sets that coflower, and so could potentially compete for pollination. Analysis of a long term dataset (1999-2005) revealed extensive co-flowering across Mpala acacia species, but little consistency in co-flowering species across years and sites. Previous work in Tanzania found co-flowering acacias to show high synchrony in timing of daily pollen release, and significant overdispersion (regularity in spacing) of species pollen release peaks through the day. This pattern is as predicted for the partitioning of a resource (shared pollinators) along a resource axis (daily time) by competitive displacement. Activity of shared pollinators tracked pollen release across the coflowering acacias, resulting in partitioning of pollinator activity within the acacia assemblage. In contrast, I found Mpala acacias to show relatively low intraspecific synchrony in dehiscence. Further, although species' pollen release peaks ranged through the day between dawn and dusk, their distribution showed no significant signature of competitive displacement.
Mpala acacias share visitor species, particularly bees and syrphid flies. Visits to flower heads tracked the timing of pollen availability, illustrating the potential for bottom-up' influences in this system. Thus, whilst coflowering Mpala acacias could potentially partition shared pollinators in daily time through divergence in the timing of dehiscence, no evidence for such a mechanism was found. Possible reasons for absence of such a pattern at Mpala are discussed.
I quantified flower-visitor interactions for two replicate flowering communities at four seasonal time points in 2004. Flowering plant species, visitor species and the interactions between them varied through seasonal time and between sites. Mpala acacias shared visitors with many other plant species, illustrating the value of adopting a community perspective. A novel null-modelling approach found all seasonal datasets to show significant daily temporal structure, resulting from concentration of activity by specific groups of flower visitors within a subset of the four daily time periods sampled. Consideration of patterns of floral resource provision in specific plant taxa (e.g. Malvaceae) showed that at least some of this temporal structuring was the result of bottom-up' control. My data were not adequately resolved to assess community-wide evidence for partitioning of shared pollinators, but the presence of temporal structure means that this remains a possibility. Further work is required to assess the potential of 'top-down influences (such as bee nesting cycles or visitor thermal physiology) in structuring daily temporal patterns. The significance of my results for other types of interaction webs is discussed.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • University of Edinburgh
  • Stone, Graham, Supervisor, External person
Award date31 Dec 2007
Publication statusPublished - 31 Dec 2007
Externally publishedYes


Dive into the research topics of 'Multi-species pollination interactions in a Kenyan savannah ecosystem'. Together they form a unique fingerprint.

Cite this