Detecting temporal changes in the temperature sensitivity of spring phenology with global warming: Application of machine learning in phenological model

Dai, Wujun; Jin, Huiying; Zhang, Yuhong; Liu, Tong; Zhou, Zhiqiang

doi:10.1016/j.agrformet.2019.107702

Cited by 35 publications

(26 citation statements)

References 87 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fundamentally rising temperature should alter many biological processes, making robust methods for identifying these changes critical. In spring plant phenology, where declining sensitivities are often reported (Fu et al, 2015;Piao et al, 2017;Dai et al, 2019), warming may increase the role of 'chilling' (determined mainly by winter temperatures) and daylength (Laube et al, 2014;Zohner et al, 2016)-potentially increasing the thermal sum required for leafout at lower values of these cues (Polgar et al, 2014;Zohner et al, 2017;Flynn and Wolkovich, 2018). Adjusting our simulations to match this model yielded shifts in sensitivities with warming.…”

Section: Main Textmentioning

confidence: 93%

“…Climate change has reshaped biological processes around the globe, with shifts in the timing of major life history events (phenology), carbon dynamics and other ecosystem processes (IPCC, 2014). With rising temperatures, a growing body of literature has documented changes in temperature sensitivity-the magnitude of a biological response scaled per • C. Many studies have found declining responses to temperature in recent decades (Fu et al, 2015;Güsewell et al, 2017;Piao et al, 2017;Dai et al, 2019) or lower sensitivities in warmer, urban areas (Meng et al, 2020).…”

Section: Main Textmentioning

confidence: 99%

See 1 more Smart Citation

A simple explanation for declining temperature sensitivity with warming

Wolkovich

Auerbach

Chamberlain

et al. 2021

Preprint

View full text Add to dashboard Cite

Temperature sensitivity—the magnitude of a biological response per °C—is a fundamental concept across scientific disciplines, especially biology, where temperature determines the rate of many plant, animal and ecosystem processes. Recently, a growing body of literature in global change biology has found temperature sensitivities decline as temperatures rise (Fuet al., 2015; Güsewell et al., 2017; Piao et al., 2017; Chen et al., 2019; Dai et al., 2019). Such observations have been used to suggest climate change is reshaping biological processes, with major implications for forecasts of future change. Here we present a simple alternative explanation for observed declining sensitivities: the use of linear models to estimate non-linear temperature responses. We show how linear estimates of sensitivities will appear to decline with warming for events that occur after a cumulative thermal threshold is met—a common model for many biological events. Corrections for the non-linearity of temperature response in simulated data and long-term phenological data from Europe remove the apparent decline. Our results show that rising temperatures combined with linear estimates based on calendar time produce observations of declining sensitivity—without any shift in the underlying biology. Current methods may thus undermine efforts to identify when and how warming will reshape biological processes.Significance statementRecently a growing body of literature has observed declining temperature sensitivities of plant leafout and other events with higher temperatures. Such results suggest that climate change is already reshaping fundamental biological processes. These temperature sensitivities are often estimated as the magnitude of a biological response per °C from linear regression. The underlying model for many events—that a critical threshold of warmth must be reached to trigger the event—however, is non-linear. We show that this mismatch between the statistical and biological models can produce the illusion of declining sensitivities with warming using current methods. We suggest simple alternative approaches that can better identify when and how warming will reshape biological processes.

show abstract

Section: Main Textmentioning

confidence: 93%

Section: Main Textmentioning

confidence: 99%

A simple explanation for declining temperature sensitivity with warming

Wolkovich

Auerbach

Chamberlain

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We applied two simplifications to Chuine's unified model, because the high number of parameters can complicate the optimization, often impairing the accuracy of the estimate [62][63][64]. Moreover, there is a high probability that the underlying biological content disappears during optimization [7,65].…”

Section: The Unified Modelmentioning

confidence: 99%

Estimation of Blooming Start with the Adaptation of the Unified Model for Three Apricot Cultivars (Prunus armeniaca L.) Based on Long-Term Observations in Hungary (1994–2020)

et al. 2022

View full text Add to dashboard Cite

The aim of our research was to adapt Chuine’s unified model to estimate the beginning of blooming of three apricot cultivars (‘Ceglédi bíborkajszi’, ‘Gönci magyar kajszi’, and ‘Rózsakajszi C.1406’) in Hungary in the time period 1994–2020. The unified model is based on the collection of chilling and forcing units. The complexity of the model lies in the high number of parameters necessary to run it. Following the work of other researchers, we reduced the number of relevant model parameters (MP) to six. In order to estimate the six MPs, we used a simulated annealing optimization method (known for being effective in avoiding getting stuck in local minima). From the results, we determined the local optimum of six MPs, and the global optimum parameter vector for three apricot cultivars. With these global optimum parameter vectors, the beginning of blooming could be estimated with a root-mean-square error (RMSE) of less than 2.5 days, using the knowledge of the daily mean temperature in the time period 1994–2020.

show abstract

“…Pruning is another well documented algorithm used to limit the effect of model overfitting and is commonly applied to RFs (Singh et al 2013;Csépe et al 2014). Alternative methods to correct overfitting include bootstrap aggregation/bagging (Navares & Aznarte 2020;Soundiran et al 2019), gradient boosting (Dai et al 2019) andBayesian Regularization (Balram et al 2019;Zanotti et al 2019).…”

Section: )mentioning

confidence: 99%