Cuestionarios

All natural environments are characterised by spatial and temporal heterogeneity. Moreover, environments are expected to become increasingly unpredictable as a consequence of global climate change. In my dissertation, I have explored multiple strategies that allow plants to spread their risk in space (seed dispersal) and time (dormancy and perenniality), as well as reduce their risk in unfavourable/unreliable pollinator environments (breeding system). My research

demonstrates that interactions between dispersal, dormancy and breeding systems in the context of life history, geographic position and climatic unpredictability are clearly complex and often contradict theoretical expectations.

My findings support the existence of two discrete syndromes among annual South African daisies: high selfing ability associated with good dispersal and obligate outcrossing associated with lower dispersal ability. This is consistent with the hypothesis that selection for colonisation success drives the evolution of an association between these traits. However, no general effect of range position on breeding system or dispersal traits was evident. This suggests selection on both breeding system and dispersal traits act consistently across these species‘ distribution ranges. Selfing ability has probably evolved in tandem with dispersal ability, most likely because autogamy offers reproductive assurance to dispersal-prone individuals that are more likely to experience conditions of pollen limitation, regardless of range position.

I also did not detect an effect of range position on relative investment in dispersal vs. dormancy in seed heteromorphic, annual Dimorphotheca species. Moreover, I found no support for the expectation that bet-hedging through relative investment in dormancy should increase in climatically unpredictable sites. This could reflect a strong influence of other local

environmental factors on fruit production (for example variation in soil nutrient availability or pollinator environment), obscuring the pattern among populations across broad geographic

114

gradients. Alternatively, selection on the production of dispersive, non-dormant propagules vs. non-dispersive, dormant propagules may be exerted by factors other than temporal heterogeneity and range edge proximity.

Interestingly, I show that the effect of life history strategy on dispersal and dormancy is not consistent. Longevity is an alternative temporal risk-reducing strategy and therefore theory predicts that it negates the need for dormancy. In support of this, I found that perennials tended to produce few dormant propagules and that annuals tended to produce many. In annuals, the importance of delayed germination as bet-hedging strategy, especially in arid environments, is well documented. Dispersal on the other hand was more strongly affected by phylogenetic relatedness than by life history. For example, perennial Dimorphotheca invested more in the production of dispersive fruit compared to annuals, which supports the prediction that dispersal is favoured in perennial plants to avoid kin competition, to increase the probability of

recruitment of scarce sites or to operate as alternative risk-reducing strategy to dormancy. However, I show that this pattern is not consistent across different genera, indicating the presence of phylogenetic structure in traits that affect wind dispersal ability.

Across 27 daisy species, controlling for phylogeny, I found evidence for negative

covariation between dispersal and dormancy. Moreover, this pattern was consistent across annual vs. perennial species, suggesting that it is not only driven by life history effects. This is

consistent with the prediction for a trade-off between traits that affect dispersal and dormancy. Negative covariation between dispersal and dormancy of different species in the same

environment may reflect interactions between a temporal storage effect and a spatial storage effect (e.g. involving competition-colonisation trade-offs) (Chesson, 2000a; b; Snyder & Chesson, 2004; Facelli et al., 2005), which can facilitate coexistence of multiple species when individual species respond differently to environmental variation (Buoro & Carlson, 2014). However, in contrast to expectations, I show that this trade-off is not necessarily expressed at the population-level and, apart from seed heteromorphic species, the individual-level. In the seed heteromorphic Dimorphotheca, individual fruit that are highly dormant do not possess structures for wind dispersal and vice versa, which is consistent with the idea that physiological or weight constraints may drive a trade-off. Spatial and temporal dispersal as alternative risk-reducing

115

strategies have important consequences for population dynamics and species persistence,

emphasising the need for further studies that integrate risk-spreading trade-offs and improve our understanding of the causes, consequences and constraints on their evolution. Indeed, my overall findings suggest that dispersal in space and time may be selected for by entirely different

selection pressures.

Spatial dispersion patterns and conspecific density is expected to strongly affect the fecundity of individuals in multi-species co-flowering communities. In support of this, my results underscore the importance of heterospecific interference and mate availability on fecundity. Both of these mechanisms are affected by plant density and dispersion, and operate independently of quantitative variation in pollinator visitation rates to flowers. Indeed, my findings emphasise the importance of including both pollinator observations and fecundity measures to tease apart the contributions of different pollinator-mediated interactions in communities. Community structure is also important: at low abundance and scattered dispersion patterns, individuals in my

experimental arrays performed poorly in terms of fruit set. Self-compatibility, however, ensured consistent fruit set and may provide a mechanism to enhance fecundity for species with scattered distributions in a community. Such scattered distributions may be evident in species with highly dispersive propagules, which is in accord with the association between selfing and high dispersal among annual South African daisies that I established in Chapter 2.

Taken together, my research illustrates that dispersal, dormancy and seed

heteromorphism may function as alternative risk-reducing strategies, enabling plants to persist in unpredictable environments. However, I show that the effects of longevity and phylogenetic relatedness are significant, and that studies focussing on covariation in dispersal and dormancy need to take into account the role of life history strategies and evolutionary relationships. My findings also highlight the importance of selfing ability as a risk-reducing strategy in biotically unfavourable or unpredictable environments. For example, selfing ability may be advantageous when mates are limited following long-distance dispersal or when individuals occur at low relative abundance in a community, or when the probability of heterospecific interference is high.

116

My research sheds some light on the many contradicting hypotheses that exist to predict and explain interactions among various risk-reducing strategies. However, it is also evident that the current theory literature is inadequate to explain the complexities observed in southern African daisies. On the other hand, studies of this nature are limited and do not consistently provide support for these theoretical predictions. Future research will benefit from additional empirical tests of dispersal-dormancy theoretical predictions, particularly studies which

simultaneously test for the influence of local determinants (e.g. pollen and resource availability) on fruit set, which could strongly influence the strength of selection to shape allocation patterns. For example, the absence of certain predicted patterns or trade-offs at the level of populations within species could be due to an absence of selection across space, as implied in some of my research chapters. However, one of the most insightful contributions to the study of range margins over the last couple of decades has been the idea that marginal populations just might not be at their adaptive optimum because of the influx of genes via migration from the species‘ core populations (Kirkpatrick & Barton, 1997; see Sexton et al., 2009 for review). This idea might be explored in the context of the present study‘s findings, and the possibility entertained that natural selection might not be as powerful to draw populations to their optimum as is sometimes presumed, and as is implied by most of the theoretical models cited in this dissertation.

Risk-reducing strategies may be especially important as environments become

increasingly unpredictable due to global climate change. Understanding the effects of climate change on biodiversity poses a major challenge to biologists in the 21st century, particularly in species-rich regions where many species may face the risk of extinction due to the loss of

suitable habitat, range shifts, etc. (Hannah, Midgley, & Millar, 2002; Hannah, Midgley, Lovejoy, et al., 2002). The Succulent Karoo biome and the Cape Floristic Region are both counted among the world‘s 25 biodiversity hotspots (Myers et al., 2000) – areas of remarkably high levels of endemism and species richness – but both are under imminent threat from climate change (Malcolm et al., 2006; Midgley & Thuiller, 2007). Some of the predicted changes for these regions include range shifts, range contractions and increasingly unpredictable rainfall events (Midgley et al., 2003; Midgley & Thuiller, 2007). Moreover, anthropogenic effects may drive habitat fragmentation which may affect pollinator diversity and consequently plant reproductive

117

success (Donaldson et al., 2002). Understanding how organisms might respond to these threats is imperative. Yet, surprisingly little information is available on the pollination biology and

dispersal of Namaqualand plants and future research will benefit greatly from a more in depth understanding of the region‘s unique ecology.

Increased awareness of risk-reducing strategies has important conservation implications (Eriksson, 2000; Buoro & Carlson, 2014). From this perspective, I suggest that Namaqualand daisies exhibiting the selfing dispersive syndrome (Chapter 2) may be at an advantage compared to those exhibiting the outcrossing/low dispersal syndrome. The latter may be particularly vulnerable to range shifts/contractions and changes in the pollinator environment, because they essentially exhibit ―specialist‖ breeding system and dispersal behaviour (Bond, 1994; E Kisdi, 2002). Their only compensation may be the ability to hedge their bets in time through dormancy (Chapter 5), although I found little evidence for increased investment in dormancy in relation to increased climatic unpredictability. This research contributes to our understanding of the ecology of spatial and temporal risk-reducing strategies and the intricate relationships among these strategies that may enable plants to persist in changing environments.

118