Wavelet Signatures of Climate and Flowering: Identification of Species Groupings

Abstract

Phenology: the timing of biological events (life stages such as flowering, fruiting, bird arrival underpins or influences many different ecological processes (Dunlop and Brown 2008; Forrest and Miller-Rushing 2010). These processes also have a significant role in shaping society’s values (e.g. on human health, biodiversity, forestry, agriculture and tourism (Beggs 2004; Fitter and Fitter 2002; van Vliet 2010)). Since the 1990s, primarily because of climate change (Keatley & Hudson 2010, Parmesan 2006, Root et al., 2008, Schwartz 2003, Sparks 1995, 2002) phenological time series have been used to determine and report the impacts of global warming in both natural and managed systems (Menzel et al. 2006; Rosenzweig et al. 2008; Sparks et al. 2005). Determining trends in relation to long-term climate, however, is not easy as trends can be confounded by short-term interannual trends. Hence not only are long term records required, but also needed is the development of novel statistical methods which can deal with confounding factors (Badeck et al. 2004; Hudson 2010; Hudson & Keatley 2010a). Wavelet methods (Daubechies, 1992) have been extensively applied to many arenas (eg. to the study of change in European spring temperatures (Palus et al. 2005) and rainfall (Koch and Markovic 2007), changes in vegetation cover (Lu et al. 2007), and to brain imaging (Bullmore et al. 2003; Sendur et al. 2007). It is the ability of wavelets to cope with nonstationary data: to deconstruct a time series into its subcomponents and remove noise; to accommodate multi-scale information, and to minimize correlation and time-dependency in data (Cornish et al. 2006; Gencay et al. 2001; Percival and Walden 2000; Vidakovic (1999)) that have added to their popularity. As phenological time series are usually non-stationary and noisy, and as such wavelet methods present as a useful analytic method (Hudson et al., 2005, Hudson 2010, Hudson et al., 2010a,b) for examining phenological records and for the determination of possibly changing climatic impacts on flowering, at an annual and across years basis. The utility of wavelets in investigating the relationship of flowering to climate (three temperature variants and rainfall) is shown in this chapter by examining the flowering intensity time series records of eight Australian eucalypts – namely, E. camaldulensis,