Global climate models predict substantial climatic changes throughout the 21st century, forcing changes in ecosystem and species distributions. Ecosystem management needs to shift from maintaining baseline states to managing ecosystem change trajectories driven by climate change. However, the strength and direction of these trajectories remain uncertain, hindering targeted management actions. Emerging novel climates and disappearing climates pose additional challenges, as novel ecosystems' functioning is unknown, increasing management failure risk. Previous risk assessments focused on climate change exposure, quantifying the distance between ambient and future climates or climate change velocity. However, these indices oversimplify ecological responses. Physiological responses are often nonlinear and co-limited by multiple climatic factors, and co-limitation hierarchies reorganize as climate changes. Existing exposure studies fail to account for these complexities. Different species and growth forms exhibit varying climatic preferences, leading to ecological responses that may not perfectly map to climate change exposure indices. Process-rich ecosystem models offer a more detailed approach, but are hindered by uncertainties. This study aims to bridge this gap by creating a phytoclimatic transformation, expressing climate in terms of its suitability for different plant growth forms that characterize terrestrial ecosystems.

Existing literature highlights the challenges of predicting and managing climate-driven ecological change. Studies have explored climate change exposure indices such as Euclidean distance between ambient and future climates and climate change velocity. However, these lack the ecological realism needed to accurately assess risks. There's a need to move beyond simply mapping climate changes to understanding the ecological consequences. Several studies have addressed the emergence of novel climates and disappearing climates and their potential impacts on species and ecosystems. However, a comprehensive approach integrating ecophysiological processes remains scarce. Previous work on ecosystem simulation models also highlighted the trade-off between prediction certainty and ecological interpretability, necessitating a novel approach that combines process-based models with aggregate metrics.

This study used a phytoclimatic transform of ambient (1979-2013) and future (2061-2080) climatologies. The process involved: (1) parameterizing an ecophysiological plant growth model (TTR-SDM) for 135,153 vascular plant species using the BIEN database; (2) using the fitted models to determine climatically suitable grid cells for each species; (3) calculating the proportion of species of each growth form suitable for each grid cell, creating a growth-form suitability index; (4) defining the vector of 14 growth-form suitabilities as the phytoclimate. The TTR-SDM describes plant carbon and nitrogen uptake and allocation, co-limited by temperature, soil moisture, solar radiation, and atmospheric CO2. Species distribution data were used to fit the model. Growth-form suitability surfaces were summarized using unsupervised classification, revealing phytoclimatic zones. The transformation was applied to future climatologies (RCP 2.6 and RCP 8.5 from five GCMs). Three multivariate Euclidean distances assessed ecological risk: (1) local phytoclimatic change; (2) novelty of future phytoclimates; (3) disappearance of ambient phytoclimates. The 5th percentile of inter-centroid distances between phytoclimatic zones served as a threshold for ecologically significant risk. Cells further than this threshold from ambient phytoclimatic states were designated as novel.

The analysis predicts substantial phytoclimatic changes by 2070. Under RCP 2.6, 33% of the terrestrial surface will experience significant change, rising to 68% under RCP 8.5. Most pronounced changes are expected in mountainous regions of South China, the Himalayas, northwestern Russia, Scandinavia, the eastern US, central Mexico, the tropical Andes, southeastern South America, southeastern Australia, and northern New Zealand. These changes translate to widespread shifts in phytoclimatic zones, particularly affecting temperate, boreal, and polar regions. Under RCP 8.5, cool-temperate and hemiboreal zones will shift poleward, while the tundra zone will shrink considerably. Tropical zones are less affected, but some regions (e.g., eastern Amazon) may transition to savanna-like phytoclimates. Only 0.3-2.2% of land cells will have novel phytoclimates (RCP 2.6-RCP 8.5), mainly in mesic subtropical climates, tropical and subtropical mountains. These novel climates are not functionally novel in the tropics; existing phytoclimates will merely shift location. 0.1-1.3% of ambient phytoclimates are projected to disappear, primarily in mesic subtropical regions. The spatial resolution (25 km) might mask finer-scale patterns. Uncertainty analysis considered RCPs and GCMs, but not species distribution model uncertainty. Sensitivity analysis showed robust estimates require hundreds of species distribution models, suggesting process-based models are superior. The hotspots of phytoclimatic change, novel and disappearing phytoclimates identified differ from those found in previous climate change studies that used untransformed climate variables, highlighting the limitations of the latter.

The substantial predicted changes in phytoclimates, especially under RCP 8.5, will significantly impact ecosystem structure, functioning, and dynamics. Ecosystem management needs to adapt accordingly, moving from preservation to managing change trajectories. The study's projections of phytoclimatic zone shifts can guide managers, providing insight into future climatic conditions and potential management strategies based on existing experience elsewhere. Other factors (disturbances, dispersal limitation, priority effects) influence ecosystem responses to climate change, potentially leading to disequilibrium and increased risk of sudden transitions. Proactive management (assisted migration, increased connectivity, rewilding) is needed to mitigate these risks. Novel phytoclimates increase uncertainty for biodiversity managers, as experience from other regions may be inapplicable. Regions with disappearing phytoclimates are high-risk areas for biodiversity loss. The study's findings highlight the importance of using ecophysiological models to assess climate change risks for ecosystems, providing a more ecologically relevant perspective compared to traditional climate exposure indices.

This study projects substantial transformation of the biosphere under current emission trends (RCP 8.5), necessitating significant adaptation measures for biodiversity conservation. Milder changes are predicted under RCP 2.6, underscoring the importance of emission reductions. Future work could investigate the interactions between climate change, other drivers of ecological change (disturbances, land use), and ecosystem responses at finer spatial scales.

The study's resolution (25 km) limits the detection of fine-scale phytoclimatic patterns. The analysis didn't incorporate uncertainty related to the species distribution model itself, focusing on RCPs and GCMs. Other ecological factors influencing ecosystem responses (dispersal limitation, biotic interactions) were not explicitly modeled. The projections are based on current growth forms and may not fully capture potential evolutionary adaptations.