Tropical phenology in a time of change

Abstract

Phenology is increasingly recognised as an important indicator to measure the impacts of global environmental change. Changes to the phenology of tropical ecosystems are likely to have wide-reaching impacts on species, human society and even feedback onto climate. However, tropical phenology data are often unavailable and analyses have been constrained by dependence on geographically limited, noncircular indicators and lack of power in statistical analyses. This thesis addresses these challenges by making available and analysing for the first time a 32-year long record of monthly focal-crown observations (>1000 individuals of >80 species) from western equatorial Africa (Lopé National Park, Gabon). In Chapter 2, I developed a novel application of Fourier analysis to objectively and quantitatively describe flowering phenology at Lopé (856 trees of 70 species). I tested the power of this approach under different scenarios of data noise (regularity of the cycle and detectability of phenological events) and data length using both simulations and field data. Most individual trees monitored at Lopé flower at regular intervals (59%) and most species have dominant annual flowering modes (88%). I showed that at least six years of data are necessary to confidently detect flowering cycles using this method. In Chapter 3, I considered how both existing, and emerging, tropical phenology monitoring programs could be made most effective for change analyses by investigating major sources of noise in data collection. Using Fourier analyses of focal crown observations from Lopé (827 trees of 61 species) I showed that regular annual cycles are more common among reproductive than vegetative phenophases. Using expert knowledge and Generalized Linear Mixed Modelling I showed that experienced field observers can provide important information on major sources of noise in data collection and that observation length, phenophase visibility and phenophase duration are all important positive predictors of cycle detectability. In Chapter 4, I assessed how local weather has changed in western equatorial Africa using Wavelet analysis and Generalised Linear Mixed Models of the long-term weather record from Lopé (34 years of rainfall and temperature observations). Lopé is characterised by a cool, cloudy, long dry season that contrasts with two bright rainy seasons. Lopé has warmed at a rate of 0.23°C per decade (minimum daily temperature) and dried at a rate of -52mm per decade (total annual precipitation) since 1984. Interannual variation in rainfall and temperature is significantly influenced by global weather patterns such as the El Niño Southern Oscillation and the Atlantic Cold Tongue. Given this context of change, in Chapter 5 I selected focal-crown observations from a representative subset of canopy tree species at Lopé (108 trees of 8 species representative of 63% of total canopy volume) to assess seasonal and interannual variation in leaf phenology. I found that the tree community is evergreen with dominant species exchanging leaves incrementally and that new leaf development is suppressed during the long dry season. Using Generalised Linear Mixed Models I demonstrated that moisture, light and leaf herbivory are all important positive predictors of new leaf production at seasonal scales. The community-wide probability of leaf flush at Lopé has declined since 1986 and is most strongly predicted by the rise in atmospheric CO2. Finally, in Chapter 6 I applied the knowledge accumulated in the previous chapters to assess the impacts of fluctuating resource availability on commercialisation of Moabi Oil, a traditional non-timber forest product in west central Africa. I combined over 15 years’ scientific monitoring of the phenology of Baillonella toxisperma at Lopé National Park with interviews of indigenous knowledge of Moabi oil producers in rural Gabon, to describe the factors that influence Moabi harvest success and explore its impacts on the rest of the Moabi oil value chain. Because of the temporal and regional variability of wild Moabi fruit availability I recommended a multi-species approach to NTFP commercialisation in the Gabonese NP buffer zones. In summary, I have shown that regularly cycling phenology is common in tropical tree communities although a wide range of strategies is evident. The evidence from Lopé supports the idea that western equatorial Africa experiences a strongly seasonal environment with a uniquely light deficient long dry season and that this seasonality in environmental conditions directly impacts the phenology of the plant community. The potential stresses on the plant community associated with the long-term warming and drying trends at Lopé appear to be compensated by CO2 fertilisation and the characteristic light deficiency of the region which improve water use efficiency. This thesis answers numerous calls for more quantitative assessment of tropical phenology data by making available evidence from a previously unpublished longterm dataset. This thesis also serves to link the cycles of tropical forest productivity and reproduction to global socio-ecological issues such as forest regeneration, climate mediation and resource availability for threatened animal species and human forest users.