Abstract
Our possibility to appropriately detect, interpret and respond to climate-driven phenological changes depends on our ability to model and predict the changes. This ability may be hampered by non-linearity in climate-phenological relations, and by spatiotemporal variability and scale mismatches of climate and phenological data. A modeling methodology capable of handling such complexities can be a powerful tool for phenological change projection. Here we develop such a methodology using citizen scientists’ observations of first flight dates for orange tip butterflies (Anthocharis cardamines) in three areas extending along a steep climate gradient. The developed methodology links point data of first flight observations to calculated cumulative degree-days until first flight based on gridded temperature data. Using this methodology we identify and quantify a first flight model that is consistent across different regions, data support scales and assumptions of subgrid variability and observation bias. Model application to observed warming over the past 60 years demonstrates the model usefulness for assessment of climate-driven first flight change. The cross-regional consistency of the model implies predictive capability for future changes, and calls for further application and testing of analogous modeling approaches to other species, phenological variables and parts of the world.
Highlights
Phenological changes due to climate change have the potential to negatively affect species conservation and interactions, and may result in decrease of biodiversity [1]
In order to model the variation of first flight dates (FF) dates for an insect species, spatially within and among regions and temporally as climate changes in a region, we hypothesize the existence of a species-specific but climate- and region-independent value (DDC) of cumulative DD that must on average be achieved until an insect is ready for flight
Considering all spatiotemporal data points (Fig 1B–1D and Fig 2A–2C), the investigation of the spatial variability assumption (SVA) case includes a total of 185 observations of FF distributed over 40 temperature grid cells for the Medelpad/Ångermanland region, 906 observations over 33 grid cells for the Sörmland/Stockholm region, and 474 observations over 28 grid cells for the Skåne region
Summary
Phenological changes due to climate change have the potential to negatively affect species conservation and interactions, and may result in decrease of biodiversity [1]. Predictions of phenological change due to climate change are often based on models that relate observed changes in some phenological event with some phenologically decoupled measure of climate change The latter may be expressed as change in average temperature [2,3,4] or in cumulative degree-days (DD) over a fixed time period at some geographical location [5,6]. This implies that the relevant accumulation time for DD changes along with the climate-driven FF change for a species Such non-linearity may violate assumptions of independence between driving climate variables and predicted phenological events [7]. We address these needs by developing a methodology to modeling climatedriven FF change that accounts for the non-linearity in FF dependence on cumulative DD until FF, and links associated spatiotemporally variable climate and phenological data with different data support scales. The methodology application includes model use for interpreting and understanding long-term climate-driven FF change over the last 60 years in three Swedish regions along a steep climate gradient
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
