FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4379600298> ?p ?o ?g. }

• W4379600298 endingPage "829" @default.
• W4379600298 startingPage "824" @default.
• W4379600298 abstract "Shifts in the timing of spring flowering events are recognized as one of the clearest and most sensitive ecological indicators of climate change (Pau et al., 2011). Numerous efforts are under way to track yearly variation in flowering phenology (Crimmins et al., 2017; Templ et al., 2018), but most ongoing studies began in the last few decades and so lack data on phenology before the onset of anthropogenic climate change. Such studies typically calculate phenological sensitivity as the linear relationship between flowering time and temperature, but there is emerging uncertainty as to whether plants may be reaching the limit of their ability to track temperature (Iler et al., 2013; Wolkovich et al., 2021). If they are reaching this limit, we would expect to see either nonlinearities (e.g. asymptotes) in phenological sensitivity (Pope et al., 2013) or increased variability (or a lack of predictability) in comparison with what were once tight, strong relationships (Pearse et al., 2017). Indeed, some have suggested that drivers such as winter chilling and photoperiod will become more prominent in cueing flowering as spring temperatures become less limiting (Cook et al., 2012). To gain an understanding of whether climate change is resulting in different patterns of phenological response, we need to compare recent shifts to historical baseline data from before the onset of rapid anthropogenic climate change. The Korean Meteorological Agency (KMA) has been recording data on spring flowering time (date of first flower) for seven woody plant species – pear (Pyrus communis), Yoshino cherry (Prunus yedoensis), Korean forsythia (Forsythia koreana), black locust (Robinia pseudoacacia), peach (Prunus persica), Korean azalea (Rhododendron mucronulatum), and Japanese apricot (Prunus mume) – for 100 yr as of 2022, making it, to our knowledge, the longest continuously running phenological monitoring effort in Asia (Ibáñez et al., 2010; Fig. 1). This centennial study offers a unique opportunity to compare phenological trends before and after the onset of rapid climate change. The extremity of recent phenology is stark: during this 10th decade of study, the record for the highest average annual temperature was broken six times and, in 2021 alone, species at nine of the 72 sites exhibited the earliest flowering phenology ever recorded. Overall, species flowered an average 4.1 d earlier for every 1°C of warming, and as much as 5.5 d earlier in the case of apricots (Fig. 2; but see also Supporting Information Figs S2, S3; model results given in Notes S1). Spring temperature explained the highest amount of variation in flowering, at least twice as much as any other factor, with significant contributions from latitude and whether the site was coastal or inland. Partitioning variation in temperature and rainfall across locations, on average through time across all locations, and across years within sites (sensu Daru et al., 2019), made no impact on these findings (see Notes S2 and the Materials and Methods section). However, these analyses did reveal that variation among sites was the dominant driver of variation in phenology, followed by variation through time across South Korea – leaving site-specific variation as the weakest driver of phenology. But this still leaves open the question as to whether spring temperature sensitivity has changed over time in response to accelerated warming. We evaluated these data with two criteria that could indicate shifting sensitivity to spring temperature: nonlinearity in trends and increased interannual variability around the mean trend. Regarding linearity, we found that while models were best fit by quadratic and cubic phenology-to-temperature functions for most species, their gains in predictive ability were small, suggesting a high degree of fidelity with linear functions. Averaged across all species, quadratic and cubic models explained < 2% more variation than linear models. At most, the error associated with the sensitivity prediction was reduced by only 0.03 d (for forsythia). We found no evidence that an exponential decay response curve was more appropriate than a simple linear fit for any species. Thus, we find little evidence to suggest that the capacity for flowering phenology to respond to spring temperature is saturating in this system (all models given in Notes S3). There are, of course, hard physiological limits to many aspects of plant physiology, and as such there is no reason to suppose that these species will be able to track shifting climate forever, merely that they have not reached their limit of plasticity yet. We do also note that most species show a greater magnitude of response to yearly temperature post-2010 when climate change has been more intense. These trends are so small as to have limited impact on overall model fit (mean, additive increase in r2 of 0.61% across all species; see Notes S4). This is notable given that we would expect species to respond more weakly to climate as they approached their response limits; we see, if anything, an increase in responsiveness. We next quantified the degree to which interannual phenological variability changed over time. If climate change were disrupting the ability of species to track environmental variability through phenological plasticity, we might expect to see increases in interannual variability. This could occur if flowering were to be decoupled from spring temperature through sensitivity saturation, with other factors becoming the dominant drivers (Meng et al., 2021). We thus explicitly modeled changes in phenological variance over time as a function of spring temperatures to test for changes in predictability. All but one species (cherry) tended to become, if anything, slightly less variable in their interannual phenology over time. Of these, four (forsythia, apricot, black locust, and rhododendron) showed strong evidence (Bayesian P > 95% for estimated slope of σ2) that this reduction in variance over time was statistically significant, though we note that these variance reductions are very small when compared to the magnitudes of mean shifts (the greatest shift is in plums, which also have the greatest variance overall; see Notes S5). Further, five of the seven species (excepting cherry and pear) showed extremely modest (albeit statistically significant) evidence of reductions in variance in warmer years (model results given in Notes S6), suggesting that the reductions in variance over time are not simply a product of shifting temperature variance. These findings indicate that interannual phenological variation has not become more variable and any increase is so small as to be biologically irrelevant. This is in keeping with studies with greater taxonomic and geographic effects on phenology but more limited temporal extent (Stemkovski et al., 2023). Combined, these two lines of evidence suggest that phenological sensitivity of woody plants in Asia has not significantly changed over the past century. Indeed, it suggests that had scientists in 1971 used the first 50 yr of data to predict phenology in 2021, their predictions would have been highly accurate. To quantify this, we modeled the degree to which predictability has changed since the onset of climate change by comparing the predictive accuracy of models trained with the first 50 yr and the most recent 50 yr of data. A simple linear model containing only one term – average spring temperature – when fit to data from 1921 to 1971, predicts the observed phenology from 1972 to 2021 with an average error of 7.3 d across all species (lowest RMSE is 4.4 d for cherries and highest 13.1 d for apricots). On the contrary, the same model, when fit using the most recent 50 yr, had only slightly higher accuracy for all species, with an average error of 6.6 d: a 6.4% improvement. Thus, scientists in 1971 could have made very good predictions of shifts in phenology that would have been accurate for the next 50 yr (full model results given in Notes S7). Of course, we must temper this observation by remembering that this dataset, while almost unprecedented in temporal scale, is of only seven species – all of which are trees and shrubs – and so we caution against generalizing our findings to other taxa. We note, however, that other studies of more taxa (790 plants, 168 birds, and 79 insects; Stemkovski et al., 2023) and that extends back further in time (1852–2007, albeit not measured each year; Willis et al., 2008) find similar patterns. Datasets exist with longer time periods that could be used for additional tests but, we caution, they are not as standardized, making it difficult to accurately test constancy of responses through time (Christidis et al., 2022). These results provide three critical insights for modern climate change research. The first is confirmatory but sometimes overlooked: historic records form an important baseline against which contemporary change must be measured to avoid the problem of shifting baselines (Pearse et al., 2017). Second, while climate is shifting in unprecedented ways that are challenging to model and forecast, the biological responses to these changes may still be linear and predictable based on existing data. Third, we stress that even these linear phenological responses may still have complex, nonlinear, and multifaceted ecological consequences. All species adjust their phenology in response to temperature variation, but their sensitivities differ from one another (note the variation in slopes in Fig. 2). For every 1°C of warming, black locust flowering advances by 3.3 d, but apricot advances by 5.5 d. Even subtle, linear differences in species sensitivities have the potential to fundamentally reorder phenological relationships and vulnerability to frost and drought and profoundly affect ecological interactions and function. This may occur long before species reach hard physiological limits on flowering plasticity. Still, our results suggest that currently-available data hold the key to predicting the future consequences of shifts in flowering phenology. We analyzed changes in predictability of flowering time in seven species from a 100-yr dataset collected by the Korea Meteorological Administration. We describe the data processing and collection briefly due to its long record period and because it is given in more detail elsewhere, and describe the analyses in full detail. All reproducible analysis codes are released in the Notes S8; the data were not collected by us and are housed with the Korea Meteorological Agency (KMA) but are also available in Notes S9. We used data gathered by the Weather Service of the Republic of Korea on the first flowering date of Forsythia koreana (Korean forsythia), Prunus mume (Japanese apricot), Pyrus communis (pear), Prunus × yedoensis (Yoshino cherry), Prunus persica (peach), Robinia pseudoacacia (black locust), and Rhododendron mucronulatum (Korean rhododendron) at 74 weather stations distributed across the country, with some weather stations starting in 1921 (data available from Korea Meteorological Administration, https://data.kma.go.kr/data/seasonObs/seasonObsDataList.do?pgmNo=648; Fig. 1; Ibáñez et al., 2010). Monitoring at other weather stations started later, and not all stations have operated continuously. Temperatures have risen across the country over the study period, with greater temperature increases in areas with larger human populations (Primack et al., 2009). These data have been extensively used to investigate how species are changing their phenology over time and in response to annual variation in temperature (Lee et al., 2011; Ellwood et al., 2012; Kim et al., 2021). However, we emphasize that while our dataset spans a large number of sites and years, it is of only seven species and all of those species are trees and shrubs (and not, for example, understory plants, or even animals). This taxonomic and functional-grouping limitation is important and should be borne in mind when considering the broader implications of this study. Observations were made on plants growing in phenological gardens on the grounds of each weather station. All plants used in this study were obtained from stock maintained and distributed by the Weather Service of the Republic of Korea to maintain genetic uniformity. When plants died or were not healthy, they were replaced by new plants. To minimize the effects of plant aging, the plants in the phenological garden have been replaced at regular intervals of between 15 and 25 yr throughout the study. Surface air temperature data have been available continuously from as early as October 1907 (city of Seoul), and records at some stations were missing during the Korean War of 1950–1953. The phenological observations were made by weather station employees according to precise written instructions that have remained constant for the entire study. First flowering was recorded when at least three flowers were open on a plant. WDP and the Pearse lab are funded by NSF Grant ABI-1759965 and UKRI NE/X00547X/1 and NE/X013022/1. SDL is supported by MOE-2020002990006, 2021003360002, 2022003640003 and NRF-2021R1A2C1011213. We are grateful to the Editor and three anonymous reviewers for their feedback, which greatly improved this manuscript. RBP is funded by NSF Grant DEB 1950447. None declared. WDP, RBP, and SDL planned research. WDP, MS, and BRL carried out statistical analysis. All authors contributed to the writing of the manuscript. All reproducible analysis codes are released in the Supporting Information; the data are housed with the Korean Meteorological Agency and are also released in the Supporting Information. Fig. S1 Standardized effect sizes of the drivers of interannual variation in flowering phenology. Fig. S2 One hundred years of first flowering dates in seven species across the Korean Peninsula and islands. Fig. S3 One hundred years of mean temperature and precipitation data across the sites within the Korean Peninsula and islands. Notes S1 Outputs from models of all species against all explanatory variables, both scaled (to estimate the relative importance of variables) and unscaled. Notes S2 Outputs from models partitioning variation in temperature and precipitation among sites, time, and within-site variation. Notes S3 Outputs from linear and nonlinear models to assess support for saturation/nonlinearity of species' functional responses to temperature. Notes S4 Outputs from models testing for differences in species' responses post-2010. Notes S5 Outputs from Bayesian hierarchical models with variance terms accounting for correlated changes in variation with year. Notes S6 Outputs from Bayesian hierarchical models with variance terms allowing for correlated changes in variation with temperature. Notes S7 Forecast and hindcast predictions from models, along with within-sample model results. Notes S8 Compressed archive of all analysis codes. Notes S9 Raw data provided to us by the Korean Meteorological Agency. Notes S10 Information about sites. Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article." @default.
• W4379600298 created "2023-06-08" @default.
• W4379600298 creator A5029975385 @default.
• W4379600298 creator A5030862260 @default.
• W4379600298 creator A5039318656 @default.
• W4379600298 creator A5080142251 @default.
• W4379600298 creator A5082050029 @default.
• W4379600298 date "2023-06-07" @default.
• W4379600298 modified "2023-10-17" @default.
• W4379600298 title "Consistent, linear phenological shifts across a century of observations in South Korea" @default.
• W4379600298 cites W1975639236 @default.
• W4379600298 cites W1986057938 @default.
• W4379600298 cites W2042488221 @default.
• W4379600298 cites W2099737847 @default.
• W4379600298 cites W2115528890 @default.
• W4379600298 cites W2127970775 @default.
• W4379600298 cites W2154457698 @default.
• W4379600298 cites W2160381002 @default.
• W4379600298 cites W2163011788 @default.
• W4379600298 cites W2183338192 @default.
• W4379600298 cites W2747626639 @default.
• W4379600298 cites W2749069611 @default.
• W4379600298 cites W2765148896 @default.
• W4379600298 cites W2788091003 @default.
• W4379600298 cites W2922332037 @default.
• W4379600298 cites W3122236491 @default.
• W4379600298 cites W3134188190 @default.
• W4379600298 cites W3193043031 @default.
• W4379600298 cites W4249890794 @default.
• W4379600298 cites W4280587484 @default.
• W4379600298 cites W4290975932 @default.
• W4379600298 doi "https://doi.org/10.1111/nph.18938" @default.
• W4379600298 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37285120" @default.
• W4379600298 hasPublicationYear "2023" @default.
• W4379600298 type Work @default.
• W4379600298 citedByCount "0" @default.
• W4379600298 crossrefType "journal-article" @default.
• W4379600298 hasAuthorship W4379600298A5029975385 @default.
• W4379600298 hasAuthorship W4379600298A5030862260 @default.
• W4379600298 hasAuthorship W4379600298A5039318656 @default.
• W4379600298 hasAuthorship W4379600298A5080142251 @default.
• W4379600298 hasAuthorship W4379600298A5082050029 @default.
• W4379600298 hasBestOaLocation W43796002981 @default.
• W4379600298 hasConcept C100970517 @default.
• W4379600298 hasConcept C127313418 @default.
• W4379600298 hasConcept C18903297 @default.
• W4379600298 hasConcept C205649164 @default.
• W4379600298 hasConcept C39432304 @default.
• W4379600298 hasConcept C49204034 @default.
• W4379600298 hasConcept C51417038 @default.
• W4379600298 hasConcept C86803240 @default.
• W4379600298 hasConcept C91586092 @default.
• W4379600298 hasConceptScore W4379600298C100970517 @default.
• W4379600298 hasConceptScore W4379600298C127313418 @default.
• W4379600298 hasConceptScore W4379600298C18903297 @default.
• W4379600298 hasConceptScore W4379600298C205649164 @default.
• W4379600298 hasConceptScore W4379600298C39432304 @default.
• W4379600298 hasConceptScore W4379600298C49204034 @default.
• W4379600298 hasConceptScore W4379600298C51417038 @default.
• W4379600298 hasConceptScore W4379600298C86803240 @default.
• W4379600298 hasConceptScore W4379600298C91586092 @default.
• W4379600298 hasFunder F4320306076 @default.
• W4379600298 hasIssue "3" @default.
• W4379600298 hasLocation W43796002981 @default.
• W4379600298 hasLocation W43796002982 @default.
• W4379600298 hasLocation W43796002983 @default.
• W4379600298 hasOpenAccess W4379600298 @default.
• W4379600298 hasPrimaryLocation W43796002981 @default.
• W4379600298 hasRelatedWork W2032521526 @default.
• W4379600298 hasRelatedWork W2095287163 @default.
• W4379600298 hasRelatedWork W2109926766 @default.
• W4379600298 hasRelatedWork W2154195245 @default.
• W4379600298 hasRelatedWork W2162941789 @default.
• W4379600298 hasRelatedWork W2317731395 @default.
• W4379600298 hasRelatedWork W2899084033 @default.
• W4379600298 hasRelatedWork W3111121122 @default.
• W4379600298 hasRelatedWork W3111997059 @default.
• W4379600298 hasRelatedWork W4385396513 @default.
• W4379600298 hasVolume "239" @default.
• W4379600298 isParatext "false" @default.
• W4379600298 isRetracted "false" @default.
• W4379600298 workType "article" @default.