Climate effects and temperature thresholds for Eucalypt flowering: a GAMLSS ZIP approach

Long-term phenological studies are limited in Australia, and meta-analytic studies of these almost nonexistent (Chambers and Keatley, 2010a,b, Hudson, 2010a, Hudson and Keatley, 2010a). Eucalypts form the focus of known Australian phyto-phenological studies, as they are the dominant species both in a botanical and economic sense. This study extends the Generalised Additive Model for Location, Scale and Shape (GAMLSS) approach to incorporate the zero-inflated (ZIP) Poisson family and to study the flowering records of 8 eucalypt species, Eucalyptus camaldulensis, E. goniocalyx, E. leucoxylon, E. macrorhyncha, E. melliodora, E. microcarpa, E. polyanthemos and E. tricarpaa - with the aim, in part, of detecting non-linear responses to climate. Regardless of the cyclicity of flowering, each species flowering is shown to be significantly influenced by temperature and this effect is non-linear. The main driver for flowering in E. leucoxylon, E. macrorhyncha, E. camaldulensis and E. melliodora is minimum temperature with flowering intensity less for E. leucoxylon for warmer minimum temperatures, while flowering intensities increase for the other 3 species in a similar climatic environment; with minimum temperatures in the region under study having decreased significantly by 0.1°C between 1998 and 2007. Maximum daily temperature has increased by 0.6°C and mean daily temperature by 0.3°C. More intense flowering is evident in E. goniocalyx and E. microcarpa as well as E. polyanthemos, which are positively influenced by mean temperature (for E. goniocalyx and E. microcarpa) and maximum temperature (for E. polyanthemos). Flowering in E. tricarpa is expected to be less intense over this period, as it is negatively influenced by maximum temperature and positively influenced by minimum temperatures (after accounting for maximum temperature). Base threshold temperatures were similar for E. camaldulensis, E. melliodora, and E. macrorhyncha, between 8.3°C and 9.3°C; and 9.8°C for E. tricarpa. By contrast E. microcarpa has a higher threshold temperature of 16.5°C, similar E. polyanthemos (17.1°C) and E. goniocalyx (13.6°C) - indicating their flowering is in a heat-demanding developmental period (Wielgolaski, 1999). E. polyanthemos has the highest base temperature but the shortest interval, 1 month, between the pre-determined start date and peak flowering. E. leucoxylon commences flowering the latest, with the lowest temperature requirement for development - as supported by its lowest threshold temperature (3.3°C). Of the eight species upper threshold temperatures (when a phase ceases) only one has previously been determined for E. leucoxylon (Hudson et al., 2003). Flowering of all species was positively and significantly correlated with last month's flowering (except for E. macrorhyncha); and with flowering 11-12 months earlier for 4 of the species. There is a clear cycling of the direction of effects of the short term (< 6 months) flowering state lags for 3 of the 8 species - E. camaldulensis, E. melliodora and E. polyanthemos. These 3 species had 2-3 month short term lag effects of past flowering (apart from the positive lag 1 effect) that negatively impact on current flowering. For E. polyanthemos there is also a highly significant but opposite, positive 4 month lag effect. These eight species are shown to be significantly influenced by temperature and as a consequence their flowering phenology will possibly change in response to climate change, with changes in temperature likely to translate to changes in both the timing of flowering commencement and intensity. GAMLSS analysis demonstrates the same contemporaneous effect of climate on flowering for E. tricarpa and E. leucoxylon, which constitutes one species pairing; for E. goniocalyx, E. microcarpa and E. macrorhyncha; and for E. camaldulensis, E.melliodora and E. polyanthemos, both species triples whose members were shown recently to flower synchronously (Hudson et al., 2011a,b). GAMLSS are thus able to assist in delineating the unique climatic signatures for species which synchronise flowering.