Inferring local processes from macro‐scale phenological pattern: a comparison of two methods

Abstract

Summary1. Understanding the processes responsible for macro-scale spatial and temporal phenological patterns is a critical step in developing predictive phenological models. While phenological responses may involve the integration of multiple environmental cues, the spring phenology of many plant and animal species appears to be especially sensitive to temperature. 2. As a result of the success of citizen science schemes in mobilizing amateur naturalists, for some parts of the world, there now exist extensive data sets of phenological timings, spanning many species, locations and years. In macroecology, two types of models -time windows and growing degree-days -are widely used to predict phenology on the basis of temperature. 3. Here, we compare the performance of the two methods in predicting spatiotemporal variation in the timing of Quercus robur first leafing. The methods agree on the time at which leafing becomes sensitive to temperature and provide weak support for a delay in initiation of thermal sensitivity with increasing latitude due to a day-length requirement. Both methods explain c. 50% of the variation in first dates and identify plasticity, rather than local adaptation, as the major cause of spatial covariation between temperature and phenology. For a 1°C rise in spring temperatures we predict that a plastic response of first leafing will give rise to an advance of about seven days. 4. Synthesis: Time-window and growing degree-day methods provide remarkably congruent insights into the processes underpinning geographic variation in Quercus robur first leafing dates. We find that a spatially invariant plastic response to temperature dominates spatiotemporal phenological variation, which means that it may be reasonable to substitute space for time to project how this species will respond to climate change. This study demonstrates the contribution that top-down macroecological approaches can make to our understanding of the processes that give rise to intraspecific phenological variation.

Estimating the ability of plants to plastically track temperature‐mediated shifts in the spring phenological optimum

“…Three of the species included here, pedunculate oak, garlic mustard and cuckooflower, have been subject to earlier work in a simpler version of our framework (Phillimore et al, 2012(Phillimore et al, , 2013. Our estimates of species' plasticity are similar to those obtained in previous studies.…”

“…Chevin et al, 2010;Hadfield, 2016). Where growing degree-day and linear reaction norm approaches have been applied to the same data sets, they have been shown to provide similar insights into phenological cues and responses (Phillimore et al, 2013;Roberts et al, 2015). We use a sliding-window approach to identify the window(s) during which mean temperature best predicts phenology.…”

“…Experimental studies reveal that photoperiod plays a key role in some species (Caffarra & Donnelly, 2011), although the precise nature of any interactions between photoperiod and the response to forcing and/or chilling temperatures is not known for most species (Polgar & Primack, 2011;Vitasse & Basler, 2013). While longitudinal data from a single site are not informative about the role that photoperiod plays, spatial replication of longitudinal data across latitudes may be (Phillimore et al, 2013), and such data are often collected by citizen science schemes.…”

Multiple pathways mediate the effects of climate change on maternal reproductive traits in a red deer population

“…This approach has been widely used across a range of study systems (e.g., Estrella et al 2007, Brommer et al 2008, Husby et al 2010, Phillimore et al 2013, and provides an easily implemented and easily interpreted means of assessing the impact of climate at different time points. However, it also clearly has some limitations, firstly because it assumes that all days within a critical window exert equal influence on the trait, and secondly because the arbitrary nature of the cutoff points used (for example, the start and end of a week, as used here) may not be biologically realistic (Sims et al 2007, van de Pol andCockburn 2011).…”

“…Previously, the effects of climate change on the Rum study population have been examined using ''growing degree-days'' (GDD), the cumulative sum of the daily mean temperatures above a threshold during a given period, as a measure of climate, assuming a relevant temperature threshold for GDD of 58C (Moyes et al 2011). Here we used raw weather data, rather than GDD because we wanted to consider other climatic measures: both maximum and minimum temperature, as well as precipitation and wind speed; further, it has been shown in other systems that findings based on using temperature in a time-window approach gives congruent results with a GDD model (Phillimore et al 2013); for further discussion regarding choice of weather variables, see the Appendix.…”