Singular Spectrum Analytic (SSA) Decomposition and Reconstruction of Flowering: Signatures of Climatic Impacts

• Published in: Environmental modeling & assessment 2017-02, Vol.22 (1), p.37-52
• Main Authors: Hudson, Irene L., Keatley, Marie R.
• Format: Article
• Language: English
• Subjects: Analysis; Applications of Mathematics; Assessments; Climate; Correlation analysis; Decomposition; Earth and Environmental Science; Eigenvalues; El Nino; Environment; Eucalyptus; Flowering; Flowers & plants; Gene expression; Influence; Math. Appl. in Environmental Science; Mathematical Modeling and Industrial Mathematics; Operations Research/Decision Theory; Phenology; Rainfall; Reconstruction; Signatures; Spectrum analysis; Studies; Temperature effects; Time series; Trends
• ISSN: 1420-2026; 1573-2967
• DOI: 10.1007/s10666-016-9516-4

An empirical approach for the decomposition and reconstruction of long-term flowering records of eight eucalypt species is presented. Results from singular spectrum analysis (SSA) allow for characterisation of the dynamic and complex flowering patterns in response to temperature and rainfall throughout the year. SSA identified trend, annual, biennial and other sub-components of flowering. The ability to discriminate flowering and climate relationships is demonstrated based on SSA (cross-)correlation analysis. SSA also identified the cyclical influence of temperature and rainfall on peak flowering. For each species, there is, on average, 6 months of the annual cycle when temperature positively influences flowering and 6 months when the influence of temperature is negative. For all species, rainfall exerts a negative influence when temperature is positive. Investigation of short-term and long-term lags of climate on flowering provided best-case climatic scenarios for each species’ flowering; e.g. more intense peak flowering is likely in Eucalyptus leucoxylon when cool, wet conditions coincide with peak flowering and is further enhanced if the preceding autumn and winter were warm and dry, and the previous spring and summer cool and wet. Three clear species groupings, according to similar SSA (cross-)correlation signatures, were identified: (1) E. leucoxylon and E. tricarpa ; (2) E. camaldulensis , E. melliodora and E. polyanthemos and (3) E. goniocalyx , E. microcarpa and E. macrorhyncha. Lastly, change point years for flowering based on SSA sub-components in four of the species seem to align with years of major shift in global ENSO signal (1951/1957/1958) as indicated by the extended multivariate ENSO index.