As climate change continues to dominate the international environmental agenda, phenology â€“ the study of the timing of recurring biological events â€“ has received increasing research attention, leading to an emerging consensus that phenology can be viewed as an â€˜early warning systemâ€™ for climate change impact.
A multidisciplinary science involving many branches of ecology, geography and remote sensing, phenology to date has lacked a coherent methodological text. This new synthesis, including contributions from many of the worldâ€™s leading phenologists, therefore fills a critical gap in the current biological literature. Providing critiques of current methods, as well as detailing novel and emerging methodologies, the book, with its extensive suite of references, provides readers with an understanding of both the theoretical basis and the potential applications required to adopt and adapt new analytical and design methods.
An invaluable source book for researchers and students in ecology and climate change science, the book also provides a useful reference for practitioners in a range of sectors, including human health, fisheries, forestry, agriculture and natural resource management.
Students studying reproductive plant biology (e.g. Plant Reproductive Ecology offered by the Institute of Ecology at University of Georgia, USA), phenology community and those newly discovering the topics from a climate change point of view, interested and advanced researchers either collaborating with statisticians or those that are able to understand and apply the methods themselves, statisticians being faced with requests on handling phenological datasets to get the overview of currently used approaches as well as their statistical limitations.
Dedication.- Contributing Authors.- Forward.- 1. Introduction and overview. 1.1 History. 1.2 Current issues in phenology. 1.3 Aims of this book; MR Keatley, IL Hudson.- 2. Global Framework for data collection â€“ data bases, data availability, future networks, online databases; E Koch.- 3. Seasonality as a core business of phenology. 3.1 Seasons as genuine phenological units. 3.2 Seasonal patterns describe the annual rhythm. 3.3 The phenological season diagram. 3.4 Seasons at a glance; F Jeanneret, T Rutishauser.- 4. Societal adaptation options to changes in phenology. 4.1 Introduction. 4.2 Phenological changes: impact and required adaptation. 4.2.1 Primary Production Sectors. 4.2.2 Public Health. 4.3 Successful adaptation requires answers to four questions. 4.4 Contribution of phenological networks to the adaptation process. 4.4.1 Phenological monitoring. 4.4.2 Phenological analysis. 4.4 Conclusions; AJH van Vliet.- 5. The influence of sampling method, sample size, and frequency of observations on plant phenological patterns and interpretation in tropical forest trees. 5.1. Introduction. 5.2. Methods. 5.2.1. Frequency of observations. 5.2.2 Sample size. 5.2.3. Comparison of sampling and estimation methods. 5.2.4. Phenological observations. 5.2.5. Data analyses. 5.3. Results. 5.3.1. Frequency of observations. 5.3.2 Sample Size. 5.3.3. Comparison of sampling methods. 5.4. Discussion and concluding remarks; LPC Morellato et al.- 6. Regression and causality. 6.1 Introduction. 6.2 An example dataset. 6.3 Linear regression. 6.4 Polynomial regression. 6.5 Some alternative ways of identifying trends. 6.6 Effects of starting year, end year and duration. 6.7 Multiple regression: Relations with temperature. 6.8 Comparing slopes. 6.9 Final thoughts; T Sparks, P Tryjanowski.- 7. Combining messy phenological time series. 7.1 Introduction. 7.2 Linear Models of phenological time series. 7.2.1 Linear Models. 7.2.2 Fixed and mixed effects models. 7.2.3 Practical issues and the R pheno-package. 7.2.4 Outlier detection. 7.2.5 Gaussian normals. 7.3 Applications. 7.3.1 Gaussian Normals. 7.3.2 Station effects. 7.4 Summary; J Schaber et al.- 8. Phenology for topoclimatological surveys and large-scale mapping. 8.1 Phenology in space and time. 8.1.1 At the crossroad of interdisciplinarity. 8.1.2 Sources of data acquisition. 8.1.3 Available phenology data for survey. 8.2 Network and survey data for mapping. 8.2.1 The space: phenology for survey. 8.2.2 Classical phenological maps at medium and large scale. 8.3 Mapping in detail: Topoclimatic scale. 8.3.1 A special network in mountainous areas. 8.3.2 Topo scale maps â€“ a genuine and unique product of phenology. 8.4 Interpolation, extrapolation and spatial modeling. 8.4.1 Phenology mapped at mesoscale. 8.4.2 Modelled phenological maps in GIS. 8.4.3 Toposcale maps in Switzerland. 8.4.4 Urban phenology â€“ surveys in a special type of space. 8.5 Future phenological mapping; F Jeanneret, T Rutishauser.- 9. Spatio-temporal statistical methods for modelling land surface phenology. 9.1 Introduction. 9.2 Thresholds. 9.2.1 Vegetation Indices (VI) thresholds. 9.3 Derivatives. 9.3.1 Greatest increase/decrease in VI. 9.3.3 Camelback phenology algorithm. 9.4 Smoothing function and model fits. 9.4.1 Autoregressive moving average. 9.4.2 Fourier analysis. 9.4.3 Principal component analysis. 9.5 Model fit. 9.5.1 Logistic Models. 9.5.2 Gaussian Local Functions. 9.5.3 Models based on growing degree-days. 9.6 Case study. 9.6.1 Thresholds. 9.6.2 Moving Average. 9.6.3 Derivatives. 9.6.4 Model fitting. 9.6 Synopsis. 9.6.1 A nomenclature is needed. 9.6.2 Uncertain error structures. 9.6.3 Limits of fitted models. 9.6.4 Parochial perspectives. 9.6.5 The challenge of water-limited systems. 9.6.6 Challenges ahead; KM de Beurs, GM Henebry.- 10. Climatic influences on the flowering phenology of four Eucalypts: a GAMLSS approach. 10.1 Introduction. 10.2 Data and Methods. 10.2.1 Phenological and climate data. 10.2.2 GAMLSS Method. 10.2.3 Model selection. 10.3 Results. 10.3.1 Thresholds for start and finish of flowering. 10.4 Discussion. 10.5 Conclusion; IL Hudson et al.- 11. Bayesian methods in phenology. 11.1 Introduction. 11.2 Bayes theory. 11.3 Examples of Bayes theory in phenological research. 11.3.1 Time series analysis by Bayesian non parametric estimation. 11.4 Correlation of phenological data with temperature. 11.4.1 Overview. 11.4.2 Methods. 11.4.3 Examples. 11.5 Discussion. 11.6 Conclusion; C Schleip et al.- 12. Smoothing methods. 12.1 Introduction. 12.2 An overview of regression methods for exploring relationships with weather data. 12.3 Improving on regression by smoothing. 12.4 Some mathematical detail. 12.5 Choosing the degree of smoothing. 12.6 Alternative methods. 12.7 Extending the method. 12.8 Software. 12.9 Conclusions; AMI Roberts.- 13. Accounting for correlated error structure within phenological data: a case study of trend analysis of snowdrop flowering. 13.1 Introduction. 13.2 Trend detection and regional-level analyses with phenological data. 13.3 Statistical analysis of snowdrop flowering data. 13.4 Results. 13.5 Discussion; N Kelly.- 14. Modelling the flowering of four Eucalypt eucalypts species using new Mixture Transition Distribution models. 14.1 Introduction. 14.2 History and approach. 14.2.1 The models. 14.3 Data sets. 14.3 Results. 14.4 Discussion; IL Hudson et al.- 15. Life history mediated responses to weather, phenology and large-scale population patterns. 15.1 Introduction. 15.2 Space-time synchrony of phenology events. 15.2.1 Threshold-triggered phenology. 15.2.2 Phenology of mast seeding. 15.3 Discussion; E Ranta et al.- 16. Applications of circular statistics in plant phenology: a case studies approach. 16.1. Introduction. 16.2. Circular Statistics. 16.2.1. Definition of circular scale and circular distribution. 16.2.2. Graphing and describing phenology using circular distributions. 16.3. Applications of circular statistics to phenological data. 16.3.1. The estimation of phenological variables using descriptive circular statistics and the vector r. 16.3.2. Synchrony and aggregation of phenological activity. 16.3.3. Testing hypotheses for circular distributions and comparing phenological patterns. 16.4. Further circular methods: a quick overview. 16.5. Circular statistics or time series analyses? A brief comment. 16.6. Concluding remarks and perspectives on the application of circular statistics; LPC Morellato et al.- 17. Wavelet analysis of flowering and climatic niche identification. 17.1 Introduction. 17.2 General motivation for wavelets analysis. 17.3 Methods. 17.3.1 Continuous versus Discrete Wavelet Transform (CWT vs DWT). 17.3.2 The Discrete Wavelet Transform (DWT). 17.2.2 Maximal Overlap DWT (MODWT). 17.3.4 Multiresolution Analysis using DWT. 17.3.5 Wavelet cross â€“ correlation and correlation. 17.4 Data analyses. 17.4.1 Phenological data. 17.5 Results. 17.5.1 MODWT-MRA. 17.5.2 Correlation with temperature and rainfall. 17.5.3 Wavelet cross-correlation with temperature and rainfall variables. 17.5.4 Level 3 wavelet cross-correlations. 17.6 Discussion. 17.6.1 Subcomponents. 17.6.2 Temperature and rainfall correlations. 17.6.3 Temperature and rainfall cross-correlations. 17.7 Conclusion; IL Hudson et al.- 18. Singular spectrum analysis: climatic niche identification. 18.1 Introduction. 18.1.2 SSA vs SSA MTM. 18.1.3 Examples of SSA-MTM. 18.2 Methods. 18.2.1 Stage 1 decomposition. 18.2.2 Stage 2 reconstruction. 18.2.3 ET groupings. 18.2.4 Assessment of any ET groups by the w-correlation plot. 18.2.5 Considerations. 18.2.6 Methods. 18.3 Results. 18.3.2 Climate. 18.3.3 Phenological data. 18.3.4 Reconstructed series 2 and 3. 18.3.5 Correlation and cross-correlations between species and weather reconstructed series 2 and 3. 18.3.6 Reconstructed series 4 and 5. 18.3.7 Reconstructed series 6 and 7. 18.4 Discussion. 18.4.1 Cycles. 18.4.2 Temperature and rainfall correlations. 18.4.3 Advantages and limitations of SSA. 18.5 Conclusions; IL Hudson et al.- 19. Herbarium collections and photographic images: Alternative data sources for phenological research. 19.1 Introduction. 19.2 Constraints to linking phenology with climate. 19.3 Weather and climate. 19.4 Using phenology to assess complex relationships between species and climate through time. 19.5 Historical data sets. 19.5.1 Herbarium collections. 19.5.2 Collection effort. 19.5.3 Completing the picture. 19.5.4 Digital photography. 19.6 Cautionary points for analysis and interpretation. 19.7 Two Australian case studies. 19.7.1 Phenological trends among Australian alpine species: using herbarium records to identify climate-change indicators. 19.7.2 Tracking phenological shifts and evolutionary impacts relating to climate change. 19.8 Initial exploratory data analysis. 19.8.1 Simple linear regression. 19.9 Limitations of standard methods to detect trend. 19.9.1 Non-linearity of trends. 19.9.2 Need for formal change point analyses. 19.10 Introduction on GAMLSS. 19.10.1 GAMLSS methods. 19.10.2 GAMLSS and Change point results. 19.11 Discussion. 19.11.1 Potential for change in hybridization dynamics; F MacGillivray et al.- 20. Meta-analysis and its application in phenological research: a review and new statistical approaches. 20.1 Introduction. 20.1.1 Why Meta-analysis in global climate change research? 20.1.2 Definition. 20.1.3 History of meta-analyses and syntheses. 20.2 Case studies. 20.2.1 Case study 1: Meta-analysis: three decades and 29,000 natural systems. 20.2.2 Case studies 2-4: Influences of methods on estimates of phenological response to global warming (Parmesan 2007) and the first fingerprints of global warming (Root et al. 2003, Parmesan and Yohe 2003). 20.2.3 Case study 5: European phenological response to climate change (Menzel et al. 2006). 20.2.4 Case study 6. Human modified temperatures induce species changes (Root et al. 2005). 20.3 Limitations and future directions. 20.3.1 Scarcity of global data. 20.3.2 Integrated data archiving and phenological networks. 20.3.3 Shortness of records. 20.3.4 Direct attribution studies. 20.3.5 Publication bias and consensus. 20.3.6 Binary indicators. 20.3.7 Future perspectives and projections. 20.3.8 Change point analyses. 20.4 Increased statistical sophistication. 20.4.1 Why we need to move beyond regression. 20.4.2 Non linearity of phenological response: implications to modelling and meta-analysis. 20.4.3 Generalised Additive Model for Location, Scale and Shape (GAMLSS), penalised spline signal regression and Bayesian nonparametric function estimation: 3 approaches to non-linear response. 20.5 Epidemiological perspectives and relevant studies. 20.5.1 Dose-response functionals and Bayesian hierarchical (BH) meta-analysis. 20.5.2 Bayesian hierarchical distributed lag models (BHDLMs). 20.5.3 Approach and justification of new methods. 20.6 Modelling of non-linear phenological response over time. 20.6.1 Generalised Additive Model for Location, Scale and Shape (GAMLSS) models for non-linear response. 20.6.2 Non-linear functionals with temperature: slope â€“ threshold forms from cubic splines. 20.6.3 Regression methods: from simple linear, multiple linear, stepwise regression to P-spline signal (PSR) regression. 20.6.4 Smoothed profile of regression coefficients in current and preceding year. 20.6.5 GAMLSS interconnection with PSR and splines in general: computer routines. 20.6.6 Modelling the functional behaviour of phenological time series: Bayesian nonparametric regression. 20.6.7 Understanding long-term ecological change: modelling current and preceding year climate. 20.6.8 Cyclic correlational and regressor profiles of past climate on flowering: links with PSR and wavelets. 20.7 Accounting for non-linearity in meta-analysis via Fractional polynomials and Spline Regression - an epidemiological application. 20.7.1 Nonparametric (smoothing) regression. 20.7.2 Parametric approaches. 20.8 Possible phenological applications. 20.8.1 Bayesian hierarchical models (BHMs): a climate change and health meta-analysis. 20.8.2 Threshold-slope model â€“ large variability amongst the cities (Baccini et al. 2008). 20.8.3 Second stage: Bayesian random effects meta-analysis (Baccini et al. 2008). 20.9 Bayesian hierarchical distributed lag models - an epidemiological application (Peng et al. 2009). 20.9.1 Single-location to across locations: a national across county averaged distributed lag function Âµ. 20.10 Towards a unified approach: semiparametric regression. 20.10.1 Mixed model approach to semiparametric regression: handling nonparametric functionals. 20.11. Discussion on increased statistical sophistication. 20.12 Conclusions; IL Hudson.-