Climate change, human exploitation, and land/sea-use changes are major drivers of environmental change, negatively impacting biodiversity by increasing extinction rates, altering species phenologies, and reshaping ecological communities. These impacts are especially pronounced in marine environments, which absorb approximately 80% of excess heat from greenhouse gas emissions. Increasing temperatures, changing ocean chemistry, and the frequency of extreme weather events pose significant threats to marine organisms, including foundation species such as seagrasses. Seagrasses are a crucial group of marine plants that provide a wide array of ecosystem services, including habitat provision, sediment stabilization, carbon sequestration, and support for diverse food webs. However, seagrass meadows are being lost at alarming rates due to both anthropogenic and climate change impacts. While there have been regional assessments of climate change effects on seagrasses, a global assessment has been lacking. This study addresses this gap by investigating how predicted range dynamics caused by climate change will influence seagrass α- and β-diversity.

Although there's a strong focus on predicting climate-related range shifts for selected seagrass species at regional scales, a global assessment of this phenomenon has been absent. This is partly due to the scarcity of georeferenced point records, coverage gaps, sampling biases, and a lack of appropriate analytical tools. This study overcomes these limitations by using range polygons from the IUCN, which integrate point records, expert knowledge, and literature data, to model species distributions. This approach has yielded robust results in other taxonomic groups.

This study uses species distribution modeling (SDM) to project future distributions of 66 seagrass species under four representative concentration pathways (RCPs: 2.6, 4.5, 6.0, and 8.5) at two time periods (T1: 2040-2050, T2: 2090-2100). α-diversity was measured using species richness, phylogenetic diversity, weighted endemism, and phylogenetic endemism. β-diversity, quantifying variation in species/phylogenetic composition between sites, was assessed using the Simpson dissimilarity index. MaxEnt was used for SDM, incorporating environmental variables such as sea temperature, salinity, and current velocity. Phylogenetic data was obtained from a Bayesian analysis of 3738 base pairs of DNA sequences. The study also assessed the effectiveness of the current network of Marine Protected Areas (MPAs) in safeguarding future seagrass diversity hotspots. IUCN polygons were used to model species distributions, and hotspots were identified using a 97.5th percentile threshold. The overlap between hotspots and MPAs was then calculated.

The models projected widespread range contractions for seagrass species, with reductions in climatically suitable areas ranging from -3.17% to -0.29% for T1 and -6.38% to -0.141% for T2. Approximately 31.82% of species are projected to experience range losses of >10%, while 28.79% will gain ranges of >10%. The magnitude of range change was weakly and negatively correlated with species' current range size. Analysis showed weak support for a correlation between evolutionary relatedness and changes in range size. In terms of α-diversity, most areas showed no notable changes in species richness and phylogenetic diversity between current and projected distributions by 2050 and 2100, except in the Tropical Eastern Pacific and Central Indo-Pacific. However, most areas are projected to experience increases in weighted endemism and phylogenetic endemism. Spatial correlations of current and future changes in α-diversity showed a weak to negligible relationship, suggesting that the drivers of current and future α-diversity may differ. Regarding β-diversity, models projected gains in species and phylogenetic β-diversity across most regions, leading to increased differentiation of seagrass communities. Conversely, the Eastern Indo-Pacific, Tropical Eastern Pacific, Western Indo-Pacific, and Temperate Southern Africa are projected to experience decreased β-diversity, resulting in homogenization. Finally, a significant proportion of future hotspots of α- and β-diversity are predicted to fall outside the current network of MPAs, indicating that these MPAs will be insufficient to conserve seagrasses in the future.

The findings highlight the vulnerability of seagrass communities to climate change, emphasizing the need for effective conservation strategies. The projected range contractions and increases in endemism indicate that some regions will experience a loss of seagrass diversity, while others will see an increase in unique species assemblages. The observed divergent shifts in β-diversity, with both homogenization and differentiation, underscore the complex impacts of climate change on community structure. The significant proportion of future hotspots outside MPAs emphasizes the limitations of current conservation efforts and necessitates the identification of new priority areas for protection. The consistency of these trends across various climate scenarios suggests that the response of marine primary producers to climate change is relatively predictable.

This study projects widespread range contractions for seagrasses and increases in endemism, highlighting the urgent need for enhanced conservation efforts. The significant portion of future diversity hotspots falling outside existing MPAs points to the insufficiency of current protection strategies. Future research should focus on refining species distribution models with more precise environmental data, incorporating biotic interactions, and developing targeted conservation plans that consider the projected shifts in both α- and β-diversity.

The study uses IUCN range polygons to represent species distributions, which may not capture the fine-scale details of species occurrences. Also, the study focuses primarily on climate change impacts, while other factors (e.g., pollution, habitat destruction) also influence seagrass distribution. The use of SDM introduces inherent uncertainties related to model selection and parameterization, which might affect the precision of predictions.