Experimental warming differentially affects vegetative and reproductive phenology of tundra plants

High latitudes and elevations are warming much faster than the global average, with Arctic models suggesting significant warming in spring and autumn by the end of the century. A major consequence of this warming is altered plant phenology, affecting both the initiation and duration of vegetative and reproductive phases. Changes in tundra plant phenology have substantial implications for plant-pollinator interactions, herbivory, productivity, and carbon and energy balances. Despite this, there's a limited understanding of how warming affects the timing of multiple plant phenophases, particularly later in the season, and whether responses will be uniform or divergent across phenophases. This knowledge gap is especially pronounced in the tundra biome, which is among the least studied for plant phenology responses to climate change. Several scenarios might explain distinct phenophase responses to warming. Reproductive and vegetative phenophases may differ due to physiological mechanisms and evolutionary drivers. For instance, leaves and flowers use different mechanisms to prevent frost damage and respond to spring temperature cues. Differing evolutionary pressures, such as co-evolution with pollinators for flowering and herbivore pressure for leaf development, also play a role. Early and late-season phenophases may respond differently if co-limited by factors like snowmelt and photoperiod. While remote sensing suggests early-season advancement with warming, late-season patterns (e.g., leaf senescence) remain unclear due to conflicting evidence and a lack of studies. Plant sensitivity to warming also varies across spatial and temporal gradients. Warming can dry out soils and accelerate snowmelt, creating abiotic stress and influencing plant growth. At colder sites, phenological responses to temperature change may be stronger. Initial responses may differ from long-term responses, as many tundra plants use stored resources, potentially delaying immediate reactions. Experimental warming manipulations, particularly using open-top chambers (OTCs), isolate the effects of temperature on plant phenology. However, in situ studies are limited in species coverage, spatial extent, and the number of phenophases monitored. This study addresses these limitations by synthesizing data from the International Tundra Experiment (ITEX), encompassing 18 sites, 46 OTC warming experiments, and observations spanning 1 to 20 years. The study investigates six plant phenophases: green up, flowering, end of flowering, fruiting, seed dispersal, and leaf senescence. The key research questions explore the magnitude and direction of phenological shifts, differential effects on reproductive versus vegetative phenology, the impact of warming on the duration of growth periods, spatial and temporal variations in plant responses, and the persistence of responses over time. The hypothesis posits that warming will differentially affect reproductive and vegetative phenophases, shift both the timing and duration of growth periods, have stronger effects at higher latitudes and in colder years, and show enhanced responses over time due to initial lag effects.

The introduction extensively reviews existing literature on tundra plant phenology and its response to climate change. It highlights the well-established advancement of green-up and flowering but emphasizes the uncertainty surrounding the response of later-season phenophases and the potential for divergent responses across different phenological events. The review cites several studies showing the impacts of warming on various aspects of tundra ecosystems, including plant-pollinator interactions, herbivory, and carbon cycling. Furthermore, it discusses various potential scenarios for how and why plant phenophases might respond distinctly to a warming climate, considering factors such as plant physiology, co-evolutionary drivers, and the influence of non-temperature variables like snowmelt and photoperiod. The literature review also acknowledges the limitations of previous studies, such as the limited spatial and temporal scales, species coverage, and inclusion of multiple phenophases. The authors highlight the importance of experimental approaches to disentangle the multiple interacting drivers of plant phenology and improve predictions of future responses to climate change.

This study compiled data from the International Tundra Experiment (ITEX), a network of long-term open-top chamber (OTC) warming experiments across Arctic, sub-Arctic, and alpine ecosystems. The dataset included observations from 18 sites and 46 experimental locations (subsites), spanning 1992-2019. The experiments measured six plant phenophases: green up, flowering (start and end), fruiting, seed dispersal, and leaf senescence. OTCs increased plot-level air temperature by 0.5-2.3°C. Each subsite was categorized into one of three soil moisture classes (dry, moist, wet). Snowmelt timing data was also collected for a subset of sites. The study used a two-step hierarchical modeling approach. The first step used interval-censored regression to estimate the mean day of year (DOY) and variation for each treatment (OTC or control) within species x subsite x year combinations. The second step used Bayesian hierarchical modeling with default priors to estimate the effects of OTC warming on plant phenology. This model included random effects for species, site, year, and subsite. Interactions between warming and spatiotemporal factors (years of warming, latitude, soil moisture, OTC deployment period, site mean temperature, and site-year temperature anomaly) were also assessed. The analysis determined whether warming altered the duration of phenological periods by calculating differences between posterior distributions of paired phenophases (e.g., green-up and senescence). The study used Bayesian credible intervals (90%, 95%) to determine whether modeled parameters indicated an effect. Climate data (daily mean air temperatures) were collected from weather stations and ERA5 reanalysis data was used to infill any gaps. All data and code are publicly available. The study used rigorous statistical methods to account for data censoring, missing data and variation among sites, years and species.

The study found significant effects of experimental warming on five of six phenophases: green up, flowering, end of flowering, seed dispersal, and leaf senescence. Reproductive phenophases (flowering, end of flowering, seed dispersal) advanced earlier and with greater magnitude than vegetative phenophases. Green up advanced, while leaf senescence was delayed, resulting in a lengthening of the growing season by approximately 1.5 days (approximately 3% of average growing season length). These patterns were largely consistent across sites, plant species, and over time. While there was considerable variation between species for reproductive phenophases, their response to warming was still consistent overall. Fruiting showed no consistent response to warming. Analysis revealed interactive effects of soil moisture, OTC deployment period, and site-level climate on flowering and seed dispersal. Flowering was more strongly affected by warming in dry sites and in year-round OTCs. Seed dispersal was more advanced in warmed plots for species whose dispersal periods coincide with warmer ambient temperatures. Contrary to expectations, latitude and inter-annual climate variability did not significantly influence the response to experimental warming. The response to experimental warming was consistent over time, showing no significant changes across the years of observation, even though a number of important processes could shift species' response on longer time-scales.

The findings demonstrate that experimental warming consistently affects multiple tundra plant phenophases, with significant advances in reproductive phases and a lengthening of the growing season. The divergent responses of reproductive and vegetative phenophases suggest that warming impacts are tissue-type specific. These findings highlight that the effects of warming are not uniform across all phenophases and will likely lead to altered timing of key ecological interactions. The observed lengthening of the growing season may have significant implications for ecosystem carbon cycling and productivity. The minor interactive effects of spatiotemporal factors suggest that experimental warming is a robust tool for assessing phenological responses to climate change. However, it is important to consider that actual Arctic warming may be considerably more dramatic than seen here, which will likely result in considerably more dramatic phenological responses. While results were consistent across most sites, years, and species, some species had more variable responses. This underscores the importance of studying the response at various levels of biological organization to more fully understand the effects of climate warming.

This large-scale synthesis of experimental warming effects on tundra plant phenology reveals consistent advancements in most phenophases, but a delay in leaf senescence, resulting in a longer growing season. Reproductive phenophases showed stronger responses than vegetative ones, suggesting tissue-type-specific impacts of warming. Minor interactions with environmental factors were observed. These findings, while based on a modest level of warming, indicate that significant changes in tundra plant phenology are likely under projected climate change scenarios. Future research should investigate the physiological mechanisms driving these differences, assess impacts on plant-pollinator interactions, and monitor phenology at the community level. Also, using a gradient of experimental warming will allow for better understanding of the effects of more severe warming and limits of linearity of responses to temperature. Finally, incorporating these refined phenological estimates into ecosystem models will significantly improve predictions of ecosystem-level responses to climate warming.

While this study represents the most extensive synthesis of experimental warming effects on tundra plant phenology to date, some limitations should be acknowledged. First, the level of warming in the OTC experiments is only a fraction of what is projected for the Arctic in the future. Second, although the study incorporated many species, not all species at a site were measured; therefore, the community-level response may differ from what the study has found here. Third, the analysis relied on existing ITEX data, which may have variations in measurement protocols or missing data that could impact the results. These potential biases were minimized with data processing and robust statistical methods.