The effects of changing climates on seed production and seed viability on tropical plant species

Abstract

Climate change is now regarded as one of the most significant mechanisms contributing to the sixth mass extinction currently occurring on Earth. Increases in both minimum and maximum temperature, increases in the severity of extreme events such as severe tropical cyclones and drought episodes, changes in precipitation patterns and an increase in the duration of the dry season in seasonal tropical regions, are key threatening processes. Tropical forest communities in Australia are extremely vulnerable to these effects as phenological patterns in seed production, seed release and the transition from seed to seedling life stages are intimately timed to current rainfall patterns. Thus, the changes to existing rainfall patterns predicted under climate change scenarios threaten the future composition of these diverse forests. In this thesis I examine the phenophase of seed fall and seed production in a seasonal tropical forest in Cairns, Australia, and explore the germination requirements of a group of species that produce desiccation sensitive seeds. This group, in particular, are underrepresented in current studies, but are potentially most at risk from changes in moisture availability because increasingly dry conditions could substantially reduce numbers of seeds successfully germinating. To examine phenological patterns in response to seasonal and inter-annual differences in weather patterns, seed production was monitored weekly over four years via 72 seed traps within a 2ha permanent plot at the Smithfield Conservation Park (SCP). Concentration of species' mean dispersal periods within the wet season suggested intra-annual patterns in seed release corresponded to periods of high moisture availability. Inter-annual variability in seed rain also responded to moisture. My results demonstrate a decrease in seed production during a severe drought associated with the with very low moisture availabilities, with 24/42 species experiencing their lowest levels of seed production in association with 2002 El Nino event. Previous evidence has shown an increase in seed production in association with El Nino Southern Oscillation (ENSO) events in aseasonal and seasonally moist tropical forests. The results I present are counter to results previously shown from other parts of the world, being the first to report a negative impact of ENSO on seed output and suggest the ability to predict the effects of changing global climates on plant reproductive output may be location- and event-specific. Almost 50% of all seed plants produce desiccation sensitive seeds. In Australia, only three species had been identified prior to the beginning of this thesis within a seasonal tropical rainforests such as SCP; previous hypotheses suggest that these species should time seed release to periods within the year where seed survival prior to germination is high; i.e. that seed dispersal should coincide with wet season. Using the 100-seed Test I identified 24 species that produce desiccation sensitive seeds and tested the hypothesis that dispersal periods of desiccation sensitive seeded species were coincident with the wet season. Unlike similar forests, there was no support for this pattern. Only four species had mean dispersal dates in the wet season. A further four species had dispersed seed across the dry/wet season and were considered wet season dispersers. The remaining 16 species all had dispersal dates centred in the dry season. There was no evidence that release of desiccation sensitive seeds was concentrated within the wet season. Intensification of seasonal aridity may increase mortality in desiccation sensitive seeds. Reduction in rainfall and the intensification of dry season moisture deficit threaten to expose desiccation sensitive seeds to greater potential negative effects of desiccation. Coupled with the time of release, just how quickly potential to successfully germinate decreases when seeds are exposed to dry conditions will influence future recruitment success. To address this I investigated the rate of seed moisture loss in the same 24 species (see above) and tested the common hypothesis that seeds conform to a simple negative exponential model of moisture loss with time. A negative exponential model described moisture loss in 14 species, but was not the best model for the remaining 10 species. Moisture loss in eight species was best described by a double-negative exponential model, and by a double-linear model in the remaining two species. One way that desiccation sensitive seeds may reduce moisture loss and maintain viability is by reducing surface area to volume (SA:V) ratio. Indeed, desiccation sensitive seeds are generally larger than orthodox seeds, and this is proposed to reflect pressure to reduce SA:V ratio. One prediction from this is that seed mass should predict the rate of desiccation between and within species. I tested this hypothesis by calculating the time constant (t 0.368) for all seeds and used these values to test if seed mass could predict the rate of desiccation. Within species the time to a given state of desiccation could be predicted by seed mass for eight species. Between species there was no relationship between desiccation rate and seed mass. Combined with evidence above that no single simple model of moisture loss can explain pattern observed in all species, this evidence suggests that seed structural features may be more important than seed mass in prolonging desiccation. To investigate correlations among seed traits considered influential in the retention of viability prior to germination, I examined critical water content (Wr50), the mean seed water content when a cohort of seeds reaches 50% mortality, desiccation rate (t0.368), the time taken for each seed to reach a relative water content of 36.8%, mean time to germination (TG), the time taken for 50% of seeds within a cohort to germinate; and specific relative water content (WM), the proportion of available water to dry seed mass; two morphological traits: seed size (MD), dry seed mass and seed coat ratio (SCR), the proportional mass between the seed coat and dry seed mass; and one phenological variable: mean monthly rainfall at mean time of seed dispersal (RM) in sixteen plant species that produce desiccation sensitive seeds from a seasonal tropical forest in Cairns, Queensland. Regression analysis revealed a significant negative relationship between RM and SCR, indicating that species that had dispersal periods at times of high moisture availability invested proportionally less into seed coats. No other pairwise trait combinations were significantly related. I also examined trait combinations in a multivariate context via principal component analysis (PCA). PCA revealed two axes of trait space. The first axis was associated with the traits SCR, TG, t0.368, RM and WM, and explained 44.1% of the variation between species. There were strong positive loadings for SCR, TG and t0.368, and strong negative loadings for RM and WM. Traits with strong loadings on the second axis (total variance 21%) were MD and Wr50 and WM. There were strong positive loadings for Wr50 and strong negative loadings for MD and WM. Most importantly; my findings reveal a first principal component axis of trait variability among desiccation sensitive seeds that is orthogonal to size. The implication of this finding is that seed size alone may not account for pre-germination viability in desiccation sensitive seeds. Seed size alone has been considered one of the principal dimensions defining plant life-history strategy space, but this work has almost exclusively been conducted outside of moist tropical areas and used few species that produce desiccation sensitive seeds. My results suggest that the traits I have measured might be important in surviving pre-germination environmental conditions, independent of the advantages and/or disadvantages that size has been shown to confer on seed survival and seedling establishment, and should be considered in attempts to predict long-term persistence and management of these species under expected changes in moisture availabilities. Throughout this thesis I describe how species exhibit different responses to a decrease in moisture availability. It is now certain that these conditions will be experienced in coming years and decades. My results show that we may anticipate a decrease in seed production associated with years of extreme dry conditions, and differential levels of susceptibility to pre-germination mortality of seeds due to desiccation will influence species differently. For those species most susceptible, the predicted impacts of climate change have the potential to eliminate some species from current vegetation communities and it is expected that many species will experience either a reduction in range size, or at the most extreme extinction. If we wish to conserve many of these species, management plans will need to be developed. I performed the work detailed in this thesis in the hope that these findings will be instructive prioritising conservation management decisions. I encourage further research to investigate the recruitment of desiccation sensitive seeds under a range of drought conditions.