Extreme environmental conditions reduce coral reef fish biodiversity and productivity

The study addresses why certain species occur at specific locations and how resultant assemblages affect ecosystem processes under accelerating anthropogenic change. It focuses on how environmental drivers, particularly temperature, shape organismal energetics and traits, which in turn influence community assembly and ecosystem functioning. Coral reefs, vulnerable to warming, are experiencing coral declines with cascading effects on fish communities. Cryptobenthic fishes, small, short-lived ectotherms with high turnover, are expected to be sensitive sentinels of environmental change but may also exhibit rapid transgenerational acclimation. The authors investigate whether extreme thermal regimes in the southeastern Arabian Gulf versus more moderate conditions in the nearby Gulf of Oman filter cryptobenthic fish assemblages via organismal tolerances and energetics, and how this affects biomass production, transfer, and renewal.

Prior work shows global stressors alter abiotic conditions, niches, and biotic interactions, affecting species coexistence and ecosystem productivity. Reef fishes often display narrow thermal tolerances, experiencing sublethal physiological and behavioral impairments under warming, with limited in situ evidence of temperature-driven population declines. Transgenerational plasticity can enhance thermal performance in some reef fishes but with energetic costs. Cryptobenthic fishes comprise nearly half of reef fish species, are abundant and ubiquitous, and strongly influence trophodynamics through rapid growth and high mortality. Their minute size entails high mass-specific metabolism and limited gill surface area, suggesting sensitivity to temperature fluctuations and limited scope for migration. While capable of rapid generational turnover and potentially fast adaptation, cryptobenthics have shown community shifts with benthic changes. The Arabian Gulf, with extreme thermal regimes, provides a natural laboratory to test organismal limits and energetic consequences for reef fish communities.

Study system and environmental context: The southeastern Arabian Gulf (extreme summers up to 36.0 °C in situ; mean daily maximum 33.7 °C, minimum 17.3 °C) and the Gulf of Oman (summer maxima 34.8 °C; mean daily maximum 29.9 °C, minimum 21.5 °C) were compared using remotely sensed SST (MODIS Aqua, 2010–2018) and in situ HOBO loggers at 4–6 m depth. Days above 34 °C were far more frequent in the Arabian Gulf. Field sampling: In April–May 2018, six reefs (three per location; Arabian Gulf: Dhabiya, Ras Ghanada, Saadiyat; Gulf of Oman: Dibba Rock, Sharm Rock, Snoopy Rock) were sampled. At each site, three reef outcrops were sampled using enclosed clove-oil stations (fine-mesh bell net and tarpaulin; 2 L clove-oil:ethanol 1:5), covering mean areas of ~4.6–4.7 m², yielding 18 community samples total. Fishes were collected, euthanized, measured (TL to 0.1 mm), weighed (0.001 g), photographed, and preserved in 95% ethanol. Benthic structure was quantified with five haphazard 20×20 cm quadrat photos per outcrop and points were categorized into functional/taxonomic groups. Thermal tolerance trials: Roving diver collections at Dhabiya (AG) and Snoopy Rock (GoO) provided individuals for critical thermal maximum (CTmax) and minimum (CTmin) assays. Species: AG: Coryogalops anomolus, Ecsenius pulcher, Enneapterygius ventermaculus; GoO: the same plus Eviota guttata, Helcogramma fuscopinna, Hetereleotris vulgaris. After ≥48 h holding, individuals were tested (0.1 °C min−1 ramp) to endpoints (loss of equilibrium/uncontrolled swimming ≥2 s). Sample sizes: 60 individuals for CTmax and 62 for CTmin across six species (see Supplementary Table 5). Gut content DNA metabarcoding: A subset of 88 individuals across six species was processed. Metazoan prey was targeted with COI (m1COlintF, jgHCO2198) and algal prey with chloroplast 23S (p23SrV_fl, Diam23Sr1) using two-step PCR, indexing, cleanups, and Illumina HiSeq sequencing. COI data were processed with QIIME/UPARSE (OTUs at 99% similarity; taxonomy via BLASTn to GenBank with coverage ≥95% and identity ≥80%); 23S data were processed with JAMP/USEARCH/UNOISE (ESVs; taxonomy via SILVA 23S with strict matching). Self-hits and implausible assignments were removed. COI yielded 547 unique OTUs and 23S yielded 3009 unique ESVs. Statistical analyses: Community metrics (species richness, individual abundance, biomass per m²) were estimated per outcrop surface area (hemi-sphere assumption from curved surface length). Bayesian hierarchical models (Gaussian; log-transformed responses) tested Location (fixed) with Site (random) for richness, abundance, biomass; posterior predictions plotted. Community composition was analyzed via nMDS (Bray–Curtis on square-root transformed data) and PERMANOVA; SIMPER identified influential species. Benthic composition used similar nMDS/PERMANOVA and compared live coral cover with a Bayesian hierarchical model (logit-transformed, Location fixed, Site random). Thermal tolerances: Bayesian linear models assessed Population (species × location) effects on CTmax and CTmin; post-hoc contrasts examined pairwise differences; body size effects were explored and not meaningful. Length–weight and abundance: For C. anomolus, E. pulcher, and E. ventermaculus, Bayesian models regressed log Weight on log TL with Location effect; abundance modeled with negative binomial error (Location fixed, Site random). Diet networks and diversity: Presence–absence matrices of prey (excluding singletons) built bipartite networks; modularity (Newman’s Q) detected with Beckett’s algorithm (20 iterations). Modularity mosaics and network trees were plotted. Prey diversity was compared via interpolated/extrapolated rarefaction curves (iNEXT). Ecosystem functioning modeling: Individual growth and mortality were modeled to estimate produced biomass (g d−1 m−2), consumed biomass (g d−1 m−2), and turnover (% d−1). Growth used VBGM and species-specific Kmax, maximum lengths, and family-level length–weight relationships; mortality incorporated VBGM parameters, SST, and body size scaling to derive daily survival probabilities. Individual production and mortality losses were summed to community-level metrics; turnover combined production and consumption relative to standing biomass.

Environmental context: In situ summer maxima reached 36.0 °C in the Arabian Gulf (mean daily max 33.7 °C) vs. 34.8 °C (mean daily max 29.9 °C) in the Gulf of Oman; winter minima were 17.3 °C vs. 21.5 °C, respectively. Days >34 °C averaged 63–69 per year in the Arabian Gulf vs. 0–5 in the Gulf of Oman. Assemblage structure: The Arabian Gulf harbored less than half the cryptobenthic species richness (1.62 ± 0.01 SE species m−2) compared to the Gulf of Oman (3.40 ± 0.26 SE species m−2; Bayesian estimate for GoO: β = 0.73, 95% CI [0.44, 1.01]) and less than a quarter of the abundance (6.12 ± 1.09 vs. 31.94 ± 1.49 individuals m−2; GoO β = 1.77, 95% CI [1.03, 2.58]). Standing biomass was comparable (GoO β = 0.63, 95% CI [−0.54, 1.71]). Community composition differed strongly (PERMANOVA: Location F = 13.58, P = 0.001, R² = 0.46). Benthic community structure and live coral cover did not differ significantly between locations. Thermal tolerance: Mean CTmax for all six tested species equaled or exceeded Arabian Gulf maximum temperatures (36.0 °C). Lowest CTmax: Helcogramma fuscopinna at 36.0 ± 0.11 °C; highest: Coryogalops anomolus at 38.4 ± 0.06 °C. Ecsenius pulcher showed slightly higher CTmax in the Arabian Gulf than in the Gulf of Oman (37.9 ± 0.05 °C vs. 37.3 ± 0.06 °C; +0.6 °C). CTmin of all species tolerated Arabian Gulf winter minima (17.3 °C); E. ventermaculus had lower CTmin in the Arabian Gulf (12.3 ± 0.06 °C) than in the Gulf of Oman (13.3 ± 0.10 °C), indicating intraspecific differences. Diet and prey ingestion: Gut metabarcoding revealed strong location-based dietary segregation. COI networks (modularity = 0.472) formed five modules with 92.3% of Arabian Gulf individuals assigned to two modules; 23S networks (modularity = 0.359) showed even stronger regional separation with nearly all Arabian Gulf individuals in a single module. Gulf of Oman populations of E. pulcher and E. ventermaculus consumed higher prey diversity (both metazoan and algal) than Arabian Gulf populations. C. anomolus showed weaker inter-location dietary differences. Body condition: Length–weight analyses showed Gulf of Oman populations were heavier at a given length for all three co-occurring species. At mean TL, individuals were heavier in the Gulf of Oman by 67.2% (E. ventermaculus), 62.2% (E. pulcher), and 10.0% (C. anomolus). Largest C. anomolus in the Arabian Gulf fell below model fits, indicating poorer condition. Productivity and turnover: Modeled produced biomass was ~6× lower and nearly an order of magnitude lower in the Arabian Gulf: 0.038 ± 0.014 g d−1 m−2 (AG) vs. 0.231 ± 0.025 g d−1 m−2 (GoO). Consumed biomass: 0.007 ± 0.001 vs. 0.039 ± 0.015 g d−1 m−2 (more than fivefold lower in AG). Turnover: 0.006 ± 0.005% d−1 (AG) vs. 0.017 ± 0.005% d−1 (GoO).

The results indicate that species-specific lethal thermal limits do not explain the depauperate cryptobenthic assemblages of the southeastern Arabian Gulf. All examined species from both regions tolerated the Gulf’s thermal extremes in CTmax/CTmin assays, and some exhibited intraspecific plasticity. Instead, energetic constraints linked to extreme and variable thermal regimes, and concomitant differences in prey resources, appear to filter species. Arabian Gulf populations showed poorer body condition (substantially lower mass at a given length) and narrower prey spectra, implying higher maintenance costs and potentially reduced digestive efficiency or prey quality. These energetic deficits cascade to ecosystem functioning: cryptobenthic produced and consumed biomass and turnover were markedly reduced in the Arabian Gulf, suggesting impaired energy transfer from benthic productivity through small fish to higher trophic levels. The goby Coryogalops anomolus, with evolutionary ties to extreme, fluctuating habitats and potentially lower energetic demands, fared better, highlighting phylogenetic and life-history influences on persistence in harsh environments. Overall, extreme environmental conditions forecast for future reefs may restructure cryptobenthic communities via energetic filtering, undermining a key pathway of heterotrophic productivity on coral reefs.

On the world’s hottest coral reefs, cryptobenthic fishes are less diverse, far less abundant, exhibit poorer body condition, and drive substantially lower biomass production, consumption, and turnover than on nearby, more moderate reefs. These patterns are not explained by lethal thermal limits but by energetic challenges associated with extreme environmental regimes and altered prey resources. The study bridges organismal physiology, diet, community assembly, and ecosystem functioning, showing that energetic constraints can decouple fish assemblages from benthic structure and live coral cover. Future research should quantify prey quality and energetic density across regions, assess temporal dynamics and seasonality, evaluate additional abiotic stressors (e.g., salinity, hypoxia), investigate genomic and transgenerational mechanisms underlying thermal plasticity and their energetic costs in the wild, and expand productivity modeling to incorporate seasonal growth and condition trajectories.

Sampling was conducted at a single time period (April–May 2018) without temporal replication, limiting inference on interannual variability. Environmental factors co-vary with temperature in the Arabian Gulf (e.g., salinity, productivity, geomorphology), potentially contributing to observed patterns. Diet analyses did not quantify prey quality (nutrients/energy density) or prey availability, and some network filtering steps excluded singleton prey occurrences. Differences in larger-scale reef structure and recent bleaching histories could influence communities. Productivity modeling did not account for seasonal growth constraints or reduced individual mass gain per unit length, potentially rendering Arabian Gulf productivity estimates optimistic. Gulf of Oman reefs may be especially productive due to upwelling, potentially accentuating contrasts.