Leveraging deep learning semantic segmentation for imaging coral skeletons.

Ecological niche models (ENMs) are used to assess the abiotic preferences of species by linking their occurrences to the environmental conditions in which they live. We developed a fossil-informed ENM framework that integrates mid-Holocene and modern occurrences to test niche stability and reconstruct abiotic niche characteristics for four critical reef-building Caribbean coral species (elkhorn coral [Acropora palmata], staghorn coral [Acropora cervicornis], boulder brain coral [Colpophyllia natans], and mustard hill coral [Porites astreoides]). Given evidence of niche stability, we used fossil-improved niche estimates to predict area and location of habitat for future climate scenarios in 2050 and 2100. We built species distribution models with environmental predictors and compared models trained with modern-only versus combined fossil and modern occurrences to evaluate differences in niche breadth, model performance, and projected habitat distributions under future climate scenarios. Including mid-Holocene fossil data in ENMs broadened niche estimates, resulting in a larger area of predicted habitat than models based solely on modern data (up to 114,559km2 more in 2100). Although our models showed that suitable habitats existed for most corals in 2100, the amount declined dramatically (45-100% decrease in area from the present day), there was a significant restriction of lower latitude habitat suitability, and marine protected areas did not overlap the majority of predicted future suitable habitat (8-20% overlap by 2100). Fossil-informed models expanded niche estimates in environmental space and incorporated environmental conditions not represented in modern data, resulting in broader projections of future habitat. Our results suggest that actions to reduce emissions and expand protected areas in the northern Caribbean are imperative to prevent significant degradation and that using fossil occurrences in niche estimation can improve the reliability of conservation forecasting, an approach that is transferable across taxa and regions.

Reactive oxygen species (ROS), including superoxide, are central molecules in eukaryotic growth, function, and immunity. Some coral species, such as Porites sp., have been associated with high extracellular superoxide concentrations, yet we lack a clear understanding of the role of controls on superoxide production in corals. Here, we combine extracellular superoxide concentration and decay rate measurements with bioinformatics to better constrain the controls and mechanisms underlying ROS formation by Porites species. Consistent with previous studies, we find that extracellular superoxide concentrations are significantly higher for Porites species compared with other coral species. We further find that superoxide decay rates are not significantly different across species, indicating that changes in production rather than decay control steady-state concentrations. Extracellular superoxide is produced by Porites astreoides across life stages from larval to newly settled polyps to adult colonies. Our bioinformatic analysis reveals that Porites lobata has genes that encode for two types of NADPH oxidase (NOX), enzymes that produce exclusively extracellular superoxide. In fact, we find widespread presence of NOX genes within genomes across scleractinian species despite species-specific variation in superoxide production. Together, these findings indicate that corals regulate extracellular superoxide levels and point to an important role for extracellular superoxide in coral physiology, such as cell signaling, cell differentiation, and growth regulation. These findings add to a growing appreciation for the beneficial role of ROS in marine organisms and provide a foundation to investigate the role of ROS in coral physiology that may provide insight into coral health and subsequent approaches for coral restoration and preservation.

Caribbean reefs face increasingly frequent and intense bleaching events, adding to the numerous other threats impacting these ecosystems. Addressing these challenges requires global action to reduce climate drivers, along with local efforts like reef restoration. Active restoration using thermotolerant coral colonies offers a potential strategy to alleviate these impacts; however, gaps remain in identifying context-specific temperature thresholds to guide colony selection and standardize thermotolerance assessment methods. This study addressed these gaps in two phases. First, by determining practical thresholds to differentiate species responses to heat stress; and second, by developing a framework to identify and prioritize resilient colonies for restoration. In the first phase, 70 colonies of Acropora cervicornis, Diploria labyrinthiformis, Montastraea cavernosa, Orbicella annularis, O. faveolata, Porites astreoides, and P. porites were sampled from reefs in the southeastern Dominican Republic. Heat stress responses were assessed through 3-hour heat pulse assays above the local maximum monthly mean (MMM) temperature, combining visual bleaching ranks, pixel intensity as a proxy for chlorophyll loss, and pulse amplitude modulated (PAM) fluorometry. Species-specific T50 thresholds were identified as the temperatures where 50% of colonies showed signs of stress. In the second phase, intraspecific thermotolerance was further examined for D. labyrinthiformis, M. cavernosa, O. annularis, O. faveolata, and P. astreoides using 99 colonies from known parent sources. Heat pulse assays at control (MMM) and T50 temperatures were repeated four times to assign colony-specific thermal performance scores. This study integrates inter- and intraspecific thermotolerance data into a practical selection framework, offering valuable insights to guide restoration under climate change.

Varying effects of climate change on the photosynthesis and calcification of crustose coralline algae: Implications for settlement of coral larvae

Increasingly frequent marine heatwaves devastate coral reefs around the world, so there is great interest in finding warm-adapted coral populations that could be used as sources for assisted gene flow and restoration. Here, we evaluated the relative power of various environmental factors to explain coral genetic variation, suggestive of differential local adaptation to these factors, across the Florida Keys Reef Tract. We applied a machine learning population genomic method (RDAforest) to two coral species-the mustard hill coral Porites astreoides and the lettuce coral Agaricia agaricites-sampled from 65 sites covering the whole reef tract. Both species comprised three genetically distinct lineages distributed across depths in a remarkably similar way. Within these lineages, there was additional genetic divergence explained by depth, but even more within-lineage variation was cumulatively explained by water chemistry parameters related to nitrogen, phosphorus, silicate, and salinity. Visualizing the predicted environment-associated genetic variation on a geographic map suggests that these associations reflect adaptation to certain aspects of the inshore-offshore environmental gradient, and, to a lesser extent, to difference of Middle and Lower Keys from the rest of the reef tract. Thermal parameters, most notably maximal monthly thermal anomaly, were also consistently identified as putative drivers of genetic divergence, but had a relatively low explanatory power compared to depth and water chemistry. Overall, our results indicate that temperature was not the most important driver of coral genetic divergence in the Florida Keys, and underscore depth and water chemistry as more important environmental factors from the corals' perspective. Our study emphasizes the need for considering a variety of environmental variables, rather than solely focusing on temperature, when predicting how corals may respond to transplantation.

Mitochondrial respiration (R) and gross photosynthesis (GP) are crucial components of the energy and carbon budgets of photosynthesizing organisms in coral reefs. This study investigates the diurnal and seasonal patterns of R in common reef algae and corals, examining the relationship between R and photosynthesis. Additionally, it evaluates discrepancies between daily R and GP calculations based on diurnal variations versus constant nighttime R, latter being the more traditional approach. We collected three coral species (Montastrea cavernosa, Porites astreoides, and Diploria strigosa) and three algal species (Caulerpa verticillata, Ceramium nitens, and Laurencia obtusa) from a Bermuda reef in fall (October) and measured their metabolic rates in a controlled outdoor mesocosm environment. Diurnal patterns of photosynthesis were measured under natural sunlight, and respiration was measured at different times by covering the incubations with a black sheet. Measurements were repeated with re-collected corals in spring (April) and summer (July). Our findings reveal pronounced diurnal patterns in R for both corals and algae, with peak R in the afternoon, lagging behind peak GP by 1-3 hours. Seasonal analysis showed the highest R in summer and the lowest in fall, correlating with temperature and light intensity variations. The study indicates that traditional models, assuming constant nighttime R throughout the day, underestimate daily R and GP rates by an average of 14% and 13%, respectively, and by 23% and 18% at a maximum. These results highlight the need to incorporate the dynamic nature of respiration into our understanding of energy and carbon fluxes in reef organisms. As metabolic energy availability is crucial for organism resilience, improved estimates of R and GP are essential for predicting organism survival in a changing environment.

Mitochondrial respiration (R) and gross photosynthesis (GP) are crucial components of the energy and carbon budgets of photosynthesizing organisms in coral reefs. This study investigates the diurnal and seasonal patterns of R in common reef algae and corals, examining the relationship between R and photosynthesis. Additionally, it evaluates discrepancies between daily R and GP calculations based on diurnal variations versus constant nighttime R, latter being the more traditional approach. We collected three coral species (Montastrea cavernosa, Porites astreoides, and Diploria strigosa) and three algal species (Caulerpa verticillata, Ceramium nitens, and Laurencia obtusa) from a Bermuda reef in fall (October) and measured their metabolic rates in a controlled outdoor mesocosm environment. Diurnal patterns of photosynthesis were measured under natural sunlight, and respiration was measured at different times by covering the incubations with a black sheet. Measurements were repeated with re-collected corals in spring (April) and summer (July). Our findings reveal pronounced diurnal patterns in R for both corals and algae, with peak R in the afternoon, lagging behind peak GP by 1–3 hours. Seasonal analysis showed the highest R in summer and the lowest in fall, correlating with temperature and light intensity variations. The study indicates that traditional models, assuming constant nighttime R throughout the day, underestimate daily R and GP rates by an average of 14% and 13%, respectively, and by 23% and 18% at a maximum. These results highlight the need to incorporate the dynamic nature of respiration into our understanding of energy and carbon fluxes in reef organisms. As metabolic energy availability is crucial for organism resilience, improved estimates of R and GP are essential for predicting organism survival in a changing environment.

Abstract The ongoing global‐scale reassembly of modern coral reefs is unprecedented compared with the observed stability of most late Quaternary reef assemblages. One notable exception is the marine isotope stage (MIS) 5e (ca 130–116 thousand years ago [ka]) reefs in the Florida Keys, where the ubiquitous shallow‐water coral, Acropora palmata, was near absent. Little is known, however, about reefs that grew during MIS5d–a (ca 116–74 ka), between MIS5e and the Holocene. It is therefore unclear whether Florida's unique MIS5e coral assemblages represent a geologically brief anomaly or a more persistent departure from the western Atlantic coral‐reef archetype. We addressed that question by reconstructing the composition of MIS5d–a reefs within 29 coral‐reef cores collected throughout the Florida Keys. We then compared the relative composition of corals during MIS5d–a to existing datasets from MIS5e, Holocene and modern (1996 and 2022) reefs to evaluate how far today's reef assemblages have diverged from geological baselines. We show that although the proportion of reef frameworks built by corals was remarkably consistent (ca 38%), species composition changed significantly through time. Acropora palmata was rare throughout MIS5, which we hypothesise was due to greater cold‐temperature stress in Florida's subtropical reefs compared with the more climatically stable tropics. In contrast, the massive reef‐building coral, Orbicella spp., was regionally dominant throughout the late Quaternary, but has become increasingly rare on modern reefs. By 2022, reefs in the Florida Keys were characterised by a truly novel coral assemblage dominated by Porites astreoides and Siderastrea siderea. In many ways, Florida's reefs defy the concept of a natural baseline; instead, their most persistent characteristic since the Late Pleistocene is their uniqueness. Yet, as reefs are increasingly subjected to unprecedented levels of environmental change, the exceptions to what was normal in the past could, paradoxically, provide the best geological analogues for the future.

Abstract The characteristics of stony corals and their environmental conditions play a critical role in shaping the abundance and composition of their associated faunal symbionts. Studies have consistently shown a positive correlation between symbiont community assemblages, number of symbionts and the size of branching coral colonies, as well as habitat depth. However, symbiont assemblages associated with non-branching corals, such as plating or massive species, remain underexplored. To address this gap, we conducted visual surveys of coral-dwelling faunal symbionts on three non-branching coral species—Agaricia agaricites, Porites astreoides, and Siderastrea siderea—at two depths (6 and 15 m) along the leeward coast of Curaçao. Symbiont species and abundance were recorded for 585 coral colonies, yielding a total of 4,969 records representing over 15 species. Observed symbiont species included barnacles, polychaete worms, gall crabs, boring bivalves, and blennies. We examined the relationship between overall symbiont abundance (irrespective of species), colony size, and depth, as well as the relative proportion of colonies hosting varying numbers of symbiont taxa. For all three coral species, the number of individual symbionts was uncorrelated with colony size. Depth significantly influenced symbiont abundance in the massive corals P. astreoides and S. siderea, but not in A. agaricites. Across all species, most coral colonies hosted one or two symbiont taxa, and this reduced symbiont composition became more prevalent at greater depths. These findings underscore the complexity of coral-symbiont assemblages and highlight the important role of environmental factors, such as depth, in structuring these communities.

The increased frequency and intensity of marine heatwaves (MHWs) induced by continued global warming are the greatest threat to tropical coral reefs, causing mass bleaching events and widespread mortality of reef building corals. In 2023, the isolated and well-protected reefs around Little Cayman experienced a MHW of > 17 Degree Heating Weeks (DHW), far exceeding any DHW measure previously captured. During the peak of the heatwave, ~ 80% of all corals were either bleached or showing signs of mortality. On the final survey date ~54% of all corals surveyed were recorded as dead. However, we identified significant differences in bleaching susceptibility and mortality across taxonomic groups, related to different life history strategies. Notably, weedy coral taxa such as Agaricia spp., Porites astreoides, and Porites porites, experienced high bleaching and suffered extensive mortality. Meanwhile, stress-tolerant reef building taxa such as Orbicella spp., experienced bleaching, but suffered low mortality. Given Little Cayman reefs have not been exposed to previous thermal stress events, the highly sensitive weedy taxa disproportionately contributed to coral abundance. Thus, the occurrence of a high magnitude - long duration heatwave resulted in catastrophic mortality of corals in Little Cayman, despite ~57% of the coastal environment being classified as no-take Marine Protected Areas. These findings underscore that the global stressor of global climate change, which drives MHWs, cannot be mitigated by local protection and isolation, thus highlighting the need to directly tackle the cause of coral decline (i.e., global climate change).

Coastal deoxygenation poses a critical threat to tropical coral reefs. Dissolved oxygen (DO) depletion can cause hypoxia-induced stress and mortality in scleractinian corals. Coral hypoxic responses are species-specific and likely modulated by the duration and severity of low-DO conditions, although the physiological mechanisms driving hypoxia tolerance are not fully understood. In this study, the Caribbean corals Acropora cervicornis, Porites astreoides and Siderastrea siderea were exposed to either severe (1.5 mg l-1 DO) or moderate (3.5 mg l-1 DO) deoxygenation or a control treatment (6 mg l-1 DO). All corals survived 2 weeks of deoxygenation but exhibited sublethal changes to coral metabolism after 1- and 2-week exposures, compared with controls. Maximum quantum yield (Fv/Fm) was suppressed after 1 week in both deoxygenation treatments in A. cervicornis, and after 2 weeks in S. siderea and P. astreoides exposed to severe or moderate treatments, respectively. Respiration rates were lower than controls in A. cervicornis and S. siderea after 1 and 2 weeks of severe deoxygenation. The reduced respiration of P. astreoides after 1 week of moderate deoxygenation returned to control levels in week 2. Overall coral metabolic budgets, assessed by ratios of gross photosynthesis to respiration (Pg:R), were more autotrophic, or photosynthesis-dominant, after 1 week of severe deoxygenation in S. siderea and P. astreoides, whereas Pg:R was not significantly different in A. cervicornis between treatments. These results reveal that some corals shift their metabolism to tolerate low-oxygen conditions and avoid bleaching or mortality, indicating that metabolic plasticity is an important aspect of coral resistance to deoxygenation.

Abstract Seawater hypoxia is increasing globally and can drive declines in organismal performance across a wide range of marine taxa. However, the effects of hypoxia on early life stages (e.g., larvae and juveniles) are largely unknown, and it is unclear how evolutionary and life histories may influence these outcomes. Here, we addressed this question by comparing hypoxia responses across early life stages of three cnidarian species representing a range of life histories: the reef‐building coral Galaxea fascicularis, a broadcast spawner with horizontal transmission of endosymbiotic algae (family Symbiodiniaceae); the reef‐building coral Porites astreoides, a brooder with vertical endosymbiont transmission; and the estuarine sea anemone Nematostella vectensis, a non‐symbiotic broadcast spawner. Transient exposure of larvae to hypoxia (dissolved oxygen < 2 mg L−1 for 6 h) led to decreased larval swimming and growth for all three species, which resulted in impaired settlement for the corals. Coral‐specific responses also included larval swelling, depressed respiration rates, and decreases in symbiont densities and function. These results indicate both immediate and latent negative effects of hypoxia on cnidarian physiology and coral–algal mutualisms specifically. In addition, G. fascicularis and P. astreoides were sensitized to heat stress following hypoxia exposure, suggesting that the combinatorial nature of climate stressors will lead to declining performance for corals. However, sensitization to heat stress was not observed in N. vectensis exposed to hypoxia, suggesting that this species may be more resilient to combined stressors. Overall, these results emphasize the importance of reducing anthropogenic carbon emissions to limit further ocean deoxygenation and warming.

Corals have been used as geochemical proxies since the 1970s, playing a prominent role in paleoceanography. However, it has not been well elucidated how aqueous ions sourced from seawater are transported and precipitated in coral skeletons. There are limited foundational methods to differentiate and quantify biogenic and abiogenic effects during skeletal formation. Especially, Mg in coral skeletons show individual variations suggesting large biogenic effects. Here, we evaluated biological complexity by investigating how coral genes evolved over geologic time scales. We focused on Mg transporter and analysed five species from genus Acropora and three species from genus Porites. Mg transporter of Acropora digitifera, Acropora hyacinthus, Acropora millepora and Porites australiensis showed higher similarity to Mg transporter of vertebrates and were reported to appear on Earth during the Pleistocene. On the other hand, Acropora palmata, Acropora tenuis and Porites astreoides showed lower or no similarity to vertebrates, and they were reported to appear on Earth before the Pleistocene. We suggest such evolutional records can be evidence to demonstrate biological complexity of Mg transport from seawater. This might explain that Mg transport is subject to evolution and why Mg incorporated in coral skeletons tends to show strong biogenic effects compared with other elements.

Coral exhibits substantial variation in pelagic larval duration, dispersal range, and population connectivity. In this study, we used reduced representation genotyping to compare the genetic structure of Caribbean reef-building species along the southeastern Dominican Republic coastline to assess connectivity within the likely dispersal kernel. Despite relatively small geographic distance between reefs, species-specific differences in genetic structure were observed. The broadcasting coral Orbicella faveolata had high levels of genetic connectivity. Between the two brooding species, Agaricia agaricites showed strong genetic subdivision, while Porites astreoides exhibited high levels of gene flow. These results suggest that multiple factors outside of life history characteristics influence genetic differentiation among populations, with species-level variability underscoring the importance of restoration and management strategies tailored to individual species, considering regional genetic and environmental variability.

Phenotypic variability is the ability of the same species to express different phenotypes under different environmental conditions. Several coral species that exist along a broad depth distribution have been shown to differ in skeletal morphology and nutrient acquisition at different depths, which has been attributed to variability in response to differing levels of light availability. This study examined the phenotypic variability of two common depth generalist corals, Montastraea cavernosa and Porites astreoides, along a gradient from 10 to 50 m in the Cayman Islands, by examining changes in skeletal morphology, photophysiology, symbiont cell density, and chlorophyll concentration. Skeletal features of M. cavernosa were found to increase in size from 10 to 30 m, but returned to smaller sizes from 30 to 50 m, while P. astreoides skeletal features continued to increase in size from 10 to 40 m. No differences were observed in either symbiont density or chlorophyll concentration across depths for either species. However, all photophysiological parameters exhibited significant depth-dependent variations in both species, revealing adaptive strategies to different light environments. These results suggest that both species have high variability in response to depth. Patterns of skeletal morphology and photophysiology, however, suggest that M. cavernosa may be more variable in regulating photosynthetic efficiency compared to P. astreoides, which likely facilitates the broader depth distribution of this species.

Re-evaluating Marine Isotope Stage 5a paleo-sea-level trends from across the Florida Keys reef tract

Geochemical proxies of sea surface temperature (SST) and seawater pH (pHsw) in scleractinian coral skeletons are valuable tools for reconstructing tropical climate variability. However, most coral skeletal SST and pHsw proxies are univariate methods that are limited in their capacity to circumvent non-climate-related variability. Here we present a novel multivariate method for reconstructing SST and pHsw from the geochemistry of coral skeletons. Our Scleractinian Multivariate Isotope and Trace Element (SMITE) method optimizes reconstruction skill by leveraging the covariance across an array of coral elemental and isotopic data with SST and pHsw. First, using a synthetic proxy experiment, we find that SMITE SST reconstruction statistics (correlation, accuracy, and precision) are insensitive to noise and variable calibration period lengths relative to Sr/Ca. While SMITE pHsw reconstruction statistics remain relative to δ11B throughout the same synthetic experiment, the magnitude of the long-term trend in pHsw is progressively lost under conditions of moderate-to-high analytical uncertainty. Next, we apply the SMITE method to an array of seven coral-based geochemical variables (B/Ca, δ11B, Li/Ca, Mg/Ca, Sr/Ca, U/Ca & Li/Mg) measured from two Bermudan Porites astreoides corals. Despite a <3.5 year calibration period, SMITE SST and pHsw estimates exhibit significantly better accuracy, precision, and correlation with their respective climate targets than the best single- and dual-proxy estimators. Furthermore, SMITE model parameters are highly reproducible between the two coral cores, indicating great potential for fossil applications (when preservation is high). The results shown here indicate that the SMITE method can outperform the most common coral-based SST and pHsw reconstructions methods to date, particularly in datasets with a large variety of geochemical variables. We therefore provide a list of recommendations and procedures for users to begin implementing the SMITE method as well as an open-source software package to facilitate dissemination of the SMITE method.

The size structure of stony coral populations can reveal underlying demographic barriers to population growth or recovery. Recent declines in coral cover from acute disturbances are well documented, but few studies have assessed size structure and the demographic processes that determine population growth. Vital rates, such as recruitment and survival, vary spatially and temporally in response to environmental conditions, in turn influencing assemblage composition. The Southeast Florida Coral Reef Ecosystem Conservation Area (Coral ECA) is a high-latitude reef system offshore of a heavily urbanized coastline. Consecutive heat stress events, stony coral tissue loss disease (SCTLD), and Hurricane Irma caused significant declines in stony coral cover and density from 2014 to 2018. The recovery potential of stony coral assemblages is influenced by their composition, the size structure of the remnant populations, and population growth during inter-disturbance periods. To assess the viability of the remaining stony coral assemblages in the Coral ECA, we quantified variation in stony coral recruit density, abundance, size structure, and assemblage composition across depth and latitude at permanent sites over 3 years (2019–2022) when no disturbances occurred. We found spatial decoupling in recruit density, adult colony density, and cover that maintains a preponderance of small colonies and skewed size structure. At sites close to shore where recruit density was higher, there was limited evidence of survival and growth of recruits, while at sites where large colonies were sampled or cover was relatively high, there was limited recruitment. The majority (&amp;gt;75%) of recruits sampled were Siderastrea siderea, but size frequency distributions were positively skewed and the coefficient of variation was high, suggesting high recruit/juvenile colony mortality and little growth into larger size classes. Porites astreoides size structure was generally lognormal and mesokurtic, particularly closer to shore, suggesting a transition between size classes. Skewness decreased moving offshore in Montastraea cavernosa and S. siderea, suggesting a transition between size classes. Recruit and adult diversity also increased moving offshore, but recruits of most species were uncommon throughout the study area. We suggest that low recruitment and high mortality, particularly in small colonies and inshore, even during inter-disturbance periods, limit the population growth of stony coral assemblages in southeast Florida.

Abstract AimGiven the rapid pace of climate change and its significant impact on species distribution and survival, understanding the dynamics of ecological niches over time becomes imperative. By employing ecological niche models and species distribution models, alongside analyses of historical occurrence records and palaeoclimatic data, we aimed to assess the extent of changes in the niche of Porites astreoides from the Holocene to the present. Specifically, we quantified the amount of: (1) niche stability, (2) niche expansion, (3) niche decline and (4) niche overlap in both the environmental and geographical space.LocationAtlantic Ocean.TaxonCnidarian, Scleractinia.Materials and MethodsThrough fossil and current records and environmental variables from the same periods, we use two ecological niche modelling approaches (overlap and maximum entropy) to analyse niche dynamics in environmental and geographical space, specifically the degree of expansion, stability and loss of niches over the last ~8000 years. We used a principal component analysis to build the models within two different calibration areas (minimum convex polygon and buffer).ResultsOur results were similar, except for the loss of habitat, across the calibration areas used to build the models. The PCA used to build the models accounts for more than 65% of the variation in the data. Most of the environmental niche remained stable with an increase of almost a third and a negligible loss. Models in geographical space showed that most of the suitable area remained stable, but the species expanded its range from the Caribbean to the coast of Brazil. The analysis also showed a high value of niche overlap in environmental and geographical spaces.Main ConclusionsThe species studied is one of the corals with the greatest capacity to adapt to new environments and locations, but it also has a high tendency to preserve niche traits. These characteristics helped the species to expand its environmental and geographical niche in a period of relative climatic stability. On the other hand, these same characteristics of niche conservatism can lead the species to population decline and habitat loss, if the speed of adaptation is lower than the rapid climate changes predicted for the middle and end of the 21st century.