Retrospective phenology in western Mediterranean plants: revealing climate change patterns through herbarium specimens

Recent advances in digital technology, coupled with rapidly increasing interest in the creation and dissemination of digitized specimen data for use in broad-scale research by botanists and other organismal scientists, have encouraged the development of a variety of new research opportunities in the botanical sciences (e.g., Page et al., 2015; Soltis, 2017). It is now increasingly possible to collect, use, re-use, and share data more easily and effectively. With the advent of the U.S. National Science Foundation's Advancing Digitization of Biodiversity Collections initiative and the establishment of iDigBio (Integrated Digitized Biocollections; www.idigbio.org) as the national resource for specimen digitization and digital data mobilization, researchers now have access to ever larger and varied digital data sets for visualization, analysis, and modeling and have new opportunities for adopting "big data" strategies for facilitating discovery. The iDigBio portal alone now includes nearly 20 million botanical specimen records, a figure that is growing rapidly as new institutions share their data. In this special issue of Applications in Plant Sciences, which is based on symposium presentations at Botany 2017 (the annual meeting of the Botanical Society of America and affiliated societies) and the XIX International Botanical Congress, authors present a broad array of examples of the latest developments in botanical biodiversity research using digitized specimen data, including in the fields of genomics, conservation assessment, ecology, phenology, and taxonomic revisions. The papers present current trends in the proactive digitization of specimen data that occurs during the collecting and vouchering of specimens and field data; the tools, skills, and strategies needed for linking and visualizing botanical data; and innovative methods for digital discovery. This collection also highlights how digital data are being used in research that expands our understanding and conservation of plant diversity and the environment. Although the source data for the papers in this collection are herbarium specimens, the topics extend well beyond systematics. Broadly integrative plant biologists will be interested in new approaches to using and re-using specimen data—whether locality information for modeling or images for analysis of morphology and/or functional traits. More importantly, digitized herbarium data become even more valuable when linked to other data sources, such as environmental or genetic data. In fact, emerging cyberinfrastructure and new data sources provide unparalleled opportunities for mobilizing and integrating massive amounts of information from organismal biology, ecology, genetics, climatology, and other disciplines. Particularly powerful is the integration of phylogenies with specimen data, enabling analyses of phylogenetic diversity in a spatio-temporal context, the evolution of niche space, and more. Such data-driven synthetic analyses may generate unexpected patterns, yielding new hypotheses for further study. However, a major challenge is the heterogeneous nature of complex data, and new methods are needed to link these divergent data types. Ongoing efforts to link and analyze diverse data are yielding new perspectives on a range of ecological problems. Integration of plant phylogeny, distributions, traits, and ultimately genetics is permitting new perspectives on landscape-level patterns of biodiversity, with implications for conservation and management of natural resources. Although many specific hypotheses may be addressed through integrated analyses of biodiversity and environmental data, perhaps the greatest value of such data-enabled science will lie in the unanticipated patterns that emerge. The papers in this collection capture some of the diversity of the emerging themes that can be addressed via use of digitized herbarium specimens. The authors address the broad range of research that can be facilitated by analysis of digitized herbarium specimens; limitations and bias of digitized specimens for certain avenues of research; future digitization and training needs; the role of globally unique identifiers (GUIDs) in integrative research involving herbarium specimens and other sources of data; digitization workflows that incorporate field, museum, and data mobilization components; the use of deep learning in specimen identification from images; the development of a standardized workflow for scoring plant phenology from herbarium specimens; the use of aggregated digitized data for fungi in generating a comprehensive mycological flora (or mycoflora) for North America; the role of digital images in education and public outreach; the effective contributions of citizen scientists of all ages to hypothesis-driven research; and the need for effective, comprehensive, and accurate tracking of data use for understanding the impact of digitized collections. Noting the centuries of exploration that have yielded the global span of the world's herbaria, James et al. (2018) provide an overview of how open, digitized, aggregated botanical data can be used to document global change, predict future impacts, and drive biological and environmental remediation. Herbarium data—from the information in labels to data that can be extracted from images—have an increasing role to play in analyses of temporal and spatial change in community composition and structure. Moreover, patterns identified via analyses of herbarium specimens can form the foundation for conservation, rehabilitation, and restoration efforts of not only single species but entire communities. However, collections data—whether plants, animals, or fungi—may not always be research-ready. James et al. address the fitness for use of herbarium data in basic and applied research, noting that taxonomic, spatial, and temporal limitations may hamper the usefulness of herbarium data for specific questions. Fortunately, research efforts addressing issues of data quality, uncertainty, and bias are providing guidance for assessing limitations for specific uses and for ameliorating the effects. Given the enormous potential of herbarium data for research in systematics, ecology, conservation, and global change, the authors cite the need for greater global advocacy for collections, from curation of physical specimens to digitization to online publishing of digitized data. Future work to enhance digital herbarium collections through digitization of other resources, such as field notes, libraries, etc., and to develop tools for discovery, visualization, analysis, and communication is needed. Key to innovative and effective use of digitized herbarium data will be skills training for the next generation of botanical researchers. The assignment of GUIDs to facilitate the tracking, linking, and discovery of biodiversity specimens across the internet has been a hotly debated subject. Although the majority of biodiversity informaticians agree that the use of unique identifiers is essential, controversy remains about which types of identifiers are best, the most appropriate Darwin Core field in which they should be published, strategies for resolving identifiers to physical specimens across the internet, and effective implementation strategies for the wide variation in biodiversity collections storage, management, and digitization. Nelson et al. (2018) narrow the scope of this debate to the implementation of GUID assignments to the digitization and mobilization of herbarium specimen data. They review the types of GUIDs in current use and strongly recommend that GUID values be associated with all specimens and included in all digital records of those specimens. They address the lack of a universal, community-supported resolver for GUID values and offer guidelines and recommended practices for minting, managing, and sharing GUIDs for herbarium specimens. Contreras (2018) brings a paleobotanical perspective to this special edition, highlighting the important role researchers can play in incorporating collection, digitization, analysis, curation, and data mobilization into an integrated research and digitization pipeline. Although she emphasizes that the workflow and pipeline presented may be especially useful in smaller institutions with limited staff or when images and other digital data are integral to the research project, the protocols she outlines may have broad applicability to researchers and other staff working in larger collections, as well as to those in non-paleontological collections, including herbaria. Her workflow incorporates three components—field, museum, and data mobilization—that are often temporally and spatially separated in current practice. As a result, the paper brings a clear museum perspective to the research process, with the museum phase serving as a transition during which specimens are organized, data are bridged from field to museum, and the preparation of a museum workspace designed to facilitate these steps. Contreras' paper offers an important viewpoint on the ways in which research, collections management, digitization, and curation can be linked to support the management of specimens in the museum. Given the rapid increase in the availability of high-quality specimen and field images of plants, the capacity to utilize computer vision and image mining techniques to make automated taxonomic identifications, extract traits, and produce phenological scorings provides the field of convolutional neural networks and deep learning tremendous opportunities for applications in botany. Botella et al. (2018) review previous work with these tools, pointing out that recent progress with deep learning techniques has shown impressive recognition performance and that, when combined with mobile applications such as Pl@ntNet (https://identify.plantnet-project.org/), these techniques may contribute significantly to species distribution modeling (SDM), biodiversity monitoring, and the inclusion of citizen science observations within each of these domains. Their paper explores the use of automated identification in the absence of human validation for SDM, particularly the impact of the degree of uncertainty when training the MAXENT niche modeling approach. They evaluated five invasive species against a training set of 332,000 human-validated plant images belonging to about 11,000 species. Their results suggest significant research challenges for using these types of data in SDMs, as well as for developing models for integrating citizen science observations into conservation management. Automated image mining is of continuing importance to botany and is a worthwhile avenue for further research. Plant phenology (seasonal events such as leaf out, flowering, and fruiting) has complex effects on multiple levels of biological organization from individuals to ecosystems, and Yost et al. (2018) discuss the potential of herbarium specimens for addressing basic and applied research on plant phenology. Phenological shifts are key indicators of global change, and temporal mismatches in phenology may have important, even catastrophic effects on natural communities and agricultural systems. For example, such mismatches between plants and pollinators can quickly cause local extinctions, drive rapid evolutionary shifts, and cause billions of dollars of agricultural losses. Herbarium specimens are an excellent source of data for documenting changes in plant phenology (see review by Willis et al., 2017), but despite millions of specimens that could contribute to an understanding of historical phenology, inter-year variation in phenology, and true shifts in phenology, the use of these data suffers from a lack of standardized scoring methods and definitions of phenological states. To date, phenological information has been captured in a herbarium specimen record in multiple ways, for example, in Darwin Core fields from 'reproductiveCondition' to 'occurrenceRemarks,' 'organismRemarks,' 'dynamicProperties,' or 'fieldNotes.' The lack of standardization in scoring and recording phenological data has limited large-scale use of specimens for phenological study. Yost et al. propose a standardized methodology for scoring phenological characters from herbarium specimens that can be applied by researchers across herbaria, research groups, and means of data collection, including via citizen science, satellite imagery, and stationary cameras. Herbaria for centuries have typically housed collections of not only plants, but also fungi. Despite current knowledge that fungi represent the sister group to animals and are not closely related to plants, many of the curatorial practices for fungi are similar to those for plants, and this similarity extends to digitization as well. Thiers and Halling (2018) describe the Macrofungi Collections Consortium (MaCC) and the development of MyCoPortal (http://mycoportal.org/portal/index.php) for serving digitized specimen information. MaCC digitized data from ~1.25 million specimens; including data contributed by the Microfungi Collections Consortium (http://www.microfungi.org/), the MyCoPortal database currently houses nearly 3.5 million specimens, as well as descriptions, illustrations, and observational records. The driving force behind development of MyCoPortal was production of a database to provide baseline data on the extent and distribution of macrofungal diversity, and the aggregated data have certainly accomplished this goal. Moreover, MyCoPortal has attracted the amateur mycological community from the United States, which comprises 80 clubs and 10,200 members. Together, professional and amateur mycologists, with the foundational data from MyCoPortal, are poised to produce a comprehensive mycoflora of North America, complete with DNA sequences, phenotypic descriptions, and images. Data from MyCoPortal have been used in taxonomic treatments, large-scale phylogenetic analyses, ecological studies, and analyses of native versus invasive species and set the stage for a broad range of uses into the future. Herbaria are reservoirs of both well-documented specimens and undescribed diversity. New species are described each year from specimens that have been housed in collections for decades, if not centuries. However, the pace of such discovery is slow, especially for non-angiosperms, and accelerating the process of discovery is expensive. von Konrat et al. (2018) explore the role of digitization in increasing accessibility to specimens, particularly in combination with citizen science efforts and online technology for uses beyond label transcription. The authors connect natural history collections to education and outreach through a citizen science tool based on the online Zooniverse platform. Their project, MicroPlants (http://microplants.fieldmuseum.org/), uses images of the liverwort genus Frullania and both a web-based platform and an interactive touchscreen version to capture large data sets for taxonomic analysis, engage a diverse participant group in research, and expose the public to novel analytical approaches and the scientific process. MicroPlants has been used in informal science settings at the Field Museum and in formal educational venues in middle schools, high schools, and colleges and universities. The project has provided valuable data on both morphological variation in Frullania and the educational effectiveness of this citizen science platform. Noteworthy is the fact that preliminary analyses indicate that data provided by non-experts were comparable to those generated by experts, supporting a role for citizen scientists in addressing authentic hypothesis-driven research. Data aggregators and publishers benefit significantly from knowing how their collections data are being used and attributed, the number of records and data sets being downloaded, the types of individuals who are finding these data useful, and the impact of projects for which the data are used. Usage metrics, in particular, help herbaria document value to institutional administrators as well as potential funders, and assist herbarium directors seek out and target underserved or expanding audiences. Cantrill (2018) summarizes tracked usage of nearly 900,000 records from the Royal Botanic Gardens Victoria, served through the Australasian Virtual Herbarium, and details trends in data usage since 2009. Queries were tracked in three broad categories, including general use, non-research use, and scientific research use, with histories of how these categories and their subcategories have become more refined over the past decade. Cantrill points out that even with more highly resolved classifications, about one third of all queries still remain unclassified. He further notes that although the data give a glimpse of data use as reported by users, they do not provide a metric for understanding the impact of the projects for which they were downloaded. Future research must assess this issue if we are to understand and report the full impact of our collections. This collection of papers provides a current snapshot of some of the issues surrounding the aggregation and use of digitized herbarium data and of some of the many possible uses of these data in research and education. However, the field is changing rapidly, with new tools for data mining, image analysis, and data tracking coming available at a rapid pace. The application of innovative analytic, algorithmic, and informatics approaches to centuries-old specimens is revolutionizing the role of herbaria and other museum collections in modern biology. Integrated Digitized Biocollections (iDigBio) is funded by grants from the U.S. National Science Foundation's Advancing Digitization of Biodiversity Collections program (Co-operative Agreements EF-1115210 and DBI-1547229). The authors thank the contributors to this special issue of Applications in Plant Sciences for their contributions to two symposia on Green Digitization (at Botany 2017 and the XIX International Botanical Congress), and we thank the Botanical Society of America and the International Botanical Congress Organizing Committee for supporting these symposia.

The Greater Himalayas hold the largest mass of ice outside polar regions and are the source of the 10 largest rivers in Asia. Rapid reduction in the volume of Himalayan glaciers due to climate change is occurring. The cascading effects of rising temperatures and loss of ice and snow in the region are affecting, for example, water availability (amounts, seasonality), biodiversity (endemic species, predator-prey relations), ecosystem boundary shifts (tree-line movements, high-elevation ecosystem changes), and global feedbacks (monsoonal shifts, loss of soil carbon). Climate change will also have environmental and social impacts that will likely increase uncertainty in water supplies and agricultural production for human populations across Asia. A common understanding of climate change needs to be developed through regional and local-scale research so that mitigation and adaptation strategies can be identified and implemented. The challenges brought about by climate change in the Greater Himalayas can only be addressed through increased regional collaboration in scientific research and policy making.

Generating ecological insights from historical data

Gaining a better understanding of global environmental change is an important challenge for conserving biodiversity. Shifts in phenology are an important consequence of environmental change. Measuring phenology of different taxa simultaneously at the same spatial and temporal scale is necessary to study the effects of changes in phenology on ecosystems. Camera traps that take both time‐lapse as well as motion‐triggered images are increasingly used to study wildlife populations. The by‐catch data of these networks of camera traps provide a potential alternative for measuring several climatic and phenological variables. Here, we tested this ability of camera traps, and quantified climatic variables as well as the timing of changes in plant and animal phenology. We obtained data from 193 camera‐unit deployments during a year of camera trapping on a peninsula in northern Sweden aimed at studying wildlife. We estimated daily temperature at noon and snow cover using recordings provided by cameras. Estimates of snow cover were accurate, but temperature estimates were higher compared with a local weather station. Furthermore, we were able to identify the timing of leaf emergence and senescence for birches (Betula sp.) and the presence of bilberry berries (Vaccinium myrtillus), as important food sources for herbivores. These were linked to the timing of the growth of antlers and the presence of new‐born young for three ungulate species as well as the presence of migratory Eurasian cranes (Grus grus). We also identified the timing of spring and autumn moulting of mountain hares (Lepus timidus) in relation to snow cover. In this novel study, we show the potential of (by‐catch) data from camera traps to study phenology across a broad range of taxa, suggesting that a global network of camera traps has great potential to simultaneously track wildlife populations and the phenology of interactions between animals and plants.

Studies in plant phenology have provided some of the best evidence for large-scale responses to recent climate change. Over the last decade, more than thirty studies have used herbarium specimens to analyze changes in flowering phenology over time. In this review, we summarize the approaches and applications used to date. Reproductive plant phenology has primarily been analyzed using two summary statistics, the mean flowering day of year and first flowering day of year, but mean flowering day has proven to be a more robust statistic. Three types of regression models have been applied to test for changes in phenology; flowering day regressed on year, flowering day regressed on temperature, and temperature regressed on year. Most studies analyzed the effect of temperature by averaging temperatures from three months prior to the date of flowering, but other approaches may be suitable in some cases. On average, published studies have used 55 herbarium specimens per species to characterize changes in phenology over time, but in many cases fewer specimens were used. Geospatial grid data is increasingly being used for determining average temperatures at herbarium specimen collection locations, allowing testing for finer scale correspondence between phenology and climate. Multiple studies have shown that inferences from herbarium specimen data are comparable to findings from systematically collected field observations. Herbarium specimens are expected to become an increasingly important resource for analyzing plant responses to climate change. As temperatures continue to rise globally, there is a need to understand phenological rates of change in response to warming and implications of these changes, especially in tropical environments where phenological studies are thus far generally lacking.

Long-term trend in vegetation gross primary production, phenology and their relationships inferred from the FLUXNET data

Climate-change-related shifts in annual phenology of a temperate snake during the last 20 years

Back to the future for plant phenology research.

Studies in plant phenology have provided some of the best evidence for large-scale responses to recent climate change. Over the last decade, more than thirty studies have used herbarium specimens to analyze changes in flowering phenology over time, although studies from tropical environments are thus far generally lacking. In this review, we summarize the approaches and applications used to date. Reproductive plant phenology has primarily been analyzed using two summary statistics, the mean flowering day of year and first-flowering day of year, but mean flowering day has proven to be a more robust statistic. Two types of regression models have been applied to test for associations between flowering, temperature and time: flowering day regressed on year and flowering day regressed on temperature. Most studies analyzed the effect of temperature by averaging temperatures from three months prior to the date of flowering. On average, published studies have used 55 herbarium specimens per species to characterize changes in phenology over time, but in many cases fewer specimens were used. Geospatial grid data are increasingly being used for determining average temperatures at herbarium specimen collection locations, allowing testing for finer scale correspondence between phenology and climate. Multiple studies have shown that inferences from herbarium specimen data are comparable to findings from systematically collected field observations. Understanding phenological responses to climate change is a crucial step towards recognizing implications for higher trophic levels and large-scale ecosystem processes. As herbaria are increasingly being digitized worldwide, more data are becoming available for future studies. As temperatures continue to rise globally, herbarium specimens are expected to become an increasingly important resource for analyzing plant responses to climate change.

Human-induced changes to the climate and environment have precipitated dramatic declines in abundance and shifts in plant and animal phenologies. These changes have been especially pronounced for migratory species that rely on numerous geographic locations throughout the year. Migratory bird species are notable in the number of species that have experienced both declines in abundance and shifts in phenology over the past 50 years, although the magnitude and direction of changes vary considerably across species. The community-level impacts of species declines and phenological shifts have been explored in stationary communities, but we know little about the effects of these changes on species relationships during migration seasons when species may interact in ways that influence their route, timing, or success of the journey (e.g., through competition or access to information about resources). Therefore, we assessed the extent to which co-migrating bird communities have changed over time, and whether changes in species co-occurrence are associated with changes in abundance or shifts in migration timing. We used over 700,000 records of birds captured at five long-term migration monitoring stations in eastern North America and found that pairwise species co-occurrences have changed by as much as 40% over the past 50 years. Changes in co-occurrence were consistently associated with species-specific changes in phenology and sometimes associated with changes in abundance. Overall, stopover communities at three sites have significantly changed over the past few decades. Numerous and dramatic changes in co-occurrence could be affecting the types and frequencies of interspecific interactions like competition and the exchange of social information, transforming the journeys of migratory birds in innumerable ways that could be altering their timing, energy, and safety.

Phenological shifts in wild-growing plants and wild animal phenophases are well documented at many European sites. Less is known about phenological shifts in agricultural plants and how wild ecosystem phenology interacts with crop phenology. Here, we present long-term phenological observations (1961–2021) from the Czech Republic for wild plants and agricultural crops and how the timing of phenophases differs from each other. The phenology of wild-growing plants was observed at various experimental sites with no agriculture or forestry management within the Czech Hydrometeorological Institute observations. The phenological data of the crops were collected from small experimental plots at the Central Institute for Supervising and Testing in Agriculture. The data clearly show a tendency to shift to earlier times during the observation period. The data also show some asynchrony in phenological shifts. Compared with wild plants, agricultural crops showed more expressive shifts to the start of the season. Phenological trends for crop plants (Triticum aestivum) showed accelerated shifts of 4.1 and 5.1 days per decade at low and middle altitudes, respectively; on the other hand, the average phenological shift for wild plants showed smaller shifts of 2.7 and 2.9 days per decade at low and middle altitudes, respectively. The phenophase ´heading´ of T. aestivum showed the highest correlation with maximum temperatures (r = 0.9), followed by wild species (with r = 0.7–0.8) and two remaining phenophases of T. aestivum jointing and ripening (with r = 0.7 and 0.6). To better understand the impacts of climate on phenological changes, it is optimal to evaluate natural and unaffected plant responses in wild species since the phenology of field crops is most probably influenced not only by climate but also by agricultural management.

Climate‐driven changes in phenology have widespread effects on ecological interactions and species' abundances. Most predictions of changes in phenology and the consequences for ecology and conservation are based on research in temperate systems. Climate‐driven changes in phenology are largely undocumented in subtropical regions, which host much of the world's biodiversity. Factors important to regulating phenology in temperate systems (e.g. winter chilling requirements) are likely weak or absent in subtropical ecosystems; plant phenology in these regions could respond to climate differently than in the temperate zone. Here we examine flowering phenology data for 105 plant species based on herbarium specimens and photographs from 1911 to 2015 in the southern subtropical Nanling region in south China. Temperatures in this region warmed 0.3°C over the 105‐year study period, and most plant species flowered earlier over time, although species varied substantially in the magnitude of phenological response to warming temperatures. Spring flowering times advanced in response to warming temperatures in late summer and early autumn and in early spring, with late summer and early autumn temperatures having almost twice as strong an effect on spring flowering times as early spring temperatures (−4.7 versus −2.5 days °C−1). This strong effect of late summer and early autumn temperatures is very different from temperate systems and has substantial implications for anticipating future changes in phenology. The temperatures in late summer and early autumn may affect spring phenology by accelerating bud formation or initial growth. Warming January temperature delayed summer flowering and advanced winter flowering. Increases in precipitation during April to June also tended to delay summer flowering. Autumn flowering species showed inconsistent responses to warming. These results highlight important differences between climate‐driven changes in phenology in temperate and subtropical areas. Understanding these differences will be important in understanding the effects of climate change on vegetation phenology and ecosystem processes.

Plants respond to climate change through shifts in traits such as height, leaf width, and flowering time. However, little is known about how grass species in semiarid ecosystems are responding. In this study, we tested three hypotheses: (1) grass species are experiencing shifts in their vegetative and reproductive organs through time, (2) precipitation is the primary driver of these morphological shifts, and (3) the reproductive period of annual grasses changes through years in the Brazilian semiarid region. We analyzed morphological and phenological data from 590 herbarium specimens of four annual grass species collected between 1859 and 2022, along with climate data from 1960 onwards. Using simple and multiple linear regressions, we assessed relationships between morphological, phenological, climatic, and temporal variables. We tested changes in phenological synchronicity related to two periods of land use alterations. Our results revealed morphological changes throughout 1859-2022: three species showed reductions in plant height, two species exhibited shorter leaves and inflorescences, and one species presented smaller spikelets. Phenological times were delayed with increasing temperatures, although no consistent directional change in reproductive phenology was observed over the last 163 years. We also found a reduction in phenological synchronicity correlated with increasing land use shift. These findings contribute to understanding morphological and phenological shifts of grasses from semiarid ecosystems in parallel to climate and land use changes.

Snow cover is an important water source for vegetation growth in arid and semi-arid areas, and grassland phenology provides valuable information on the response of terrestrial ecosystems to climate change. The Mongolian Plateau features both abundant snow cover resources and typical grassland ecosystems. In recent years, with the intensification of global climate change, the snow cover on the Mongolian Plateau has changed correspondingly, with resulting effects on vegetation growth. In this study, using MOD10A1 snow cover data and MOD13A1 Normalized Difference Vegetation Index (NDVI) data combined with remote sensing (RS) and geographic information system (GIS) techniques, we analyzed the spatiotemporal changes in snow cover and grassland phenology on the Mongolian Plateau from 2001 to 2018. The correlation analysis and grey relation analysis were used to determine the influence of snow cover parameters (snow cover fraction (SCF), snow cover duration (SCD), snow cover onset date (SCOD), and snow cover end date (SCED)) on different types of grassland vegetation. The results showed wide snow cover areas, an early start time, a late end time, and a long duration of snow cover over the northern Mongolian Plateau. Additionally, a late start, an early end, and a short duration were observed for grassland phenology, but the southern area showed the opposite trend. The SCF decreased at an annual rate of 0.33%. The SCD was shortened at an annual rate of 0.57 d. The SCOD and SCED in more than half of the study area advanced at annual rates of 5.33 and 5.74 DOY (day of year), respectively. For grassland phenology, the start of the growing season (SOS) advanced at an annual rate of 0.03 DOY, the end of the growing season (EOS) was delayed at an annual rate of 0.14 DOY, and the length of the growing season (LOS) was prolonged at an annual rate of 0.17 d. The SCF, SCD, and SCED in the snow season were significantly positively correlated with the SOS and negatively correlated with the EOS and LOS. The SCOD was significantly negatively correlated with the SOS and positively correlated with the EOS and LOS. The SCD and SCF can directly affect the SOS of grassland vegetation, while the EOS and LOS were obviously influenced by the SCOD and SCED. This study provides a scientific basis for exploring the response trends of alpine vegetation to global climate change.

Plants meet machines: Prospects in machine learning for plant biology