Kelp Forests Modulate Fish Community Dynamics and Responses to Ocean Warming

Study Overview and Geographic Setting

• Ecological Context & Background:

  • Foundation habitats like kelp forests, coral reefs, and seagrass meadows create structurally complex, highly biodiverse marine ecosystems that underpin coastal functioning and resilience.
  • Kelp forests are among the most productive marine biogenic structures, dampening wave energy, slowing currents, and providing critical three-dimensional living space for juvenile and adult fishes, macroinvertebrates, and epiphytes.
  • Exported kelp detritus (particulate and dissolved organic matter) subsidizes adjacent soft-bottom food webs, extending ecological influence beyond standing biomass.
  • Global threats including ocean warming, marine heatwaves, eutrophication, and intensified herbivory drive canopy collapses, turf algae proliferation, and poleward expansion of warm-affinity species (tropicalisation).
  • The Atlantic-Iberian coastline sits within a major eastern-boundary upwelling system. Despite upwelling buffering, models project weakened upwelling, a $1\text{--}2^\text{o}\text{C}$ sea surface temperature (SST) increase by mid-century, and heightened marine heatwave frequencies.
  • Compressed within a latitudinal span of $< 7^\text{o}$, Atlantic Iberia serves as a natural climate transition zone to evaluate how kelp species identity and canopy abundance mediate fish community dynamics.

• Geographic Locations & Coordinates:

• Surveys spanned $\text{∼}600\,\text{km}$ of coastline across four key regions in the Southern European Atlantic Ecoregion:

  • Coruña, Spain: $43.4690^\text{o}\,\text{N}, 8.2470^\text{o}\,\text{W}$

  • Vigo, Spain: $42.2220^\text{o}\,\text{N}, 8.7996^\text{o}\,\text{W}$

  • Viana, Portugal: $41.6340^\text{o}\,\text{N}, 8.8343^\text{o}\,\text{W}$

  • Peniche, Portugal: $39.3780^\text{o}\,\text{N}, 9.3740^\text{o}\,\text{W}$

  • Sampling Design & Field Protocols:

• Temporal scope: 5 consecutive years ($2020\text{--}2024$), sampled annually during mid-summer ($\text{July}\text{--}\text{August}$).

• Spatial structure: 5 permanent rocky reefs per region, separated by $2\text{--}5\,\text{km}$, positioned within the mid-shelf kelp depth zone ($7\text{--}18\,\text{m}$).

• Replication: 5 replicate $25\,\text{m}$ belt transects per reef ($n = 450$ total transects across the multi-year study, with slightly lower replication in Coruña due to logistics).

• Fish visual censuses:

  • Cryptic/benthic fishes: Counted within a $1\,\text{m}$ swatch on either side of the transect tape ($2\,\text{m}$ total width).
  • Pelagic/benthopelagic fishes: Counted within a $2\,\text{m}$ width on either side ($4\,\text{m}$ total width).
  • Size estimation: Visual total length estimated to the nearest $5\,\text{cm}$ and categorized into small (s), medium (m), and large (l) size classes based on species-specific parameters from FishBase.

• Kelp censuses: All kelp individuals $> 25\,\text{cm}$ height counted to species level within $1\,\text{m}$ on each side of the tape ($50\,\text{m}^2$ area per transect).

• Diver-Operated Video (DOV): Complementary video recording using a SeaGIS camera system equipped with GoPro7 cameras, analyzed using SeaGIS EventMeasure software to validate underwater visual census (UVC) counts.

• Environmental Sea Surface Temperature Data: Extracted daily from March to August ($2020\text{--}2024$) from Copernicus Marine Services Mediterranean Sea High-Resolution L4 SST Reprocessed product (MED-REP-L4, $0.05^\text{o}$ spatial resolution) using Python xarray at the nearest grid point for each site.

Kelp Forest Community Structure and Biogeography

• Sea Surface Temperature (SST) Spatial & Interannual Trends:

  • Significant spatial variation ($p < 0.0001$) and temporal variation ($p < 0.0001$) in average March–August SST.
  • Regional SST ranking (warmest to coolest):
  • Peniche: $16.1 \pm 0.42\,^\text{o}\text{C}$
  • Vigo: $15.4 \pm 0.44\,^\text{o}\text{C}$
  • Viana: $15.1 \pm 0.40\,^\text{o}\text{C}$
  • Coruña: $15.0 \pm 0.42\,^\text{o}\text{C}$ (Viana and Coruña did not differ significantly, $p = 0.26$).
  • Temporal SST trajectory: Rose by $+0.96\,^\text{o}\text{C}$ between 2020 and 2023 ($p < 0.001$), followed by a minor dip in 2024 (2024 did not differ significantly from 2020, $p = 0.57$).

• Kelp Abundance and Spatial Density:

  • Average kelp density ($50\,\text{m}^{-2}$ transect) varied by region ($p = 0.0418$) and year ($p < 0.0001$):
  • Coruña recorded highest density: $323 \pm 85$
  • Peniche recorded lowest density: $226 \pm 59$
  • Temporal density trajectory:
  • 2020: $271 \pm 77$
  • 2021 (Peak year): $427 \pm 115$
  • 2024 (Minimum year): $173 \pm 49$

• Kelp Species Identity and Richness:

• Four canopy species identified across the regional network:
  1. Laminaria hyperborea: Boreal perennial kelp; restricted strictly to cooler northern sites (Coruña and Viana).
  2. Laminaria ochroleuca: Lusitanian perennial kelp; dominant canopy former in warmer sites (Vigo and Peniche), co-occurring in northern forests.
  3. Saccorhiza polyschides: Opportunistic annual kelp; ubiquitous across all four regions.
  4. Undaria pinnatifida: Non-native invasive annual kelp; present exclusively in northern Spanish sites (Coruña and Vigo) with marked interannual fluctuations.
• Species Richness: Significant regional divergence ($p < 0.0001$), stable across years ($p = 0.9995$):
  • Coruña: $3.19 \pm 0.23$ species
  • Vigo: $3.06 \pm 0.16$ species
  • Peniche: $2.39 \pm 0.14$ species
  • Viana: $2.21 \pm 0.14$ species
• Collinearity test: Generalized Variance-Inflation Factors (GVIF) among SST, kelp abundance, and kelp richness were negligible ($\text{GVIF} < 1.04$).

Reef-Fish Diversity, Abundance, and Community Dynamics

• Fish Species Diversity & Richness:

  • Total fish species recorded: 53 unique species across all surveys.
  • Regional total species richness counts:
  • Peniche: 33 species
  • Coruña: 28 species
  • Viana: 24 species
  • Vigo: 23 species
  • Shannon–Wiener Diversity Index ($H'$) generalized linear mixed models showed significant effects for region ($p = 0.00014$), year ($p < 0.0001$), and their interaction ($p = 0.00018$):
  • Peniche: $1.03 \pm 0.50$
  • Viana: $0.98 \pm 0.49$
  • Coruña: $0.89 \pm 0.50$
  • Vigo: $0.80 \pm 0.51$
  • Temporal anomaly: A sharp drop in $H'$ occurred in 2021 at Coruña ($H' = 0.66$) and Vigo, followed by multi-year rebounds.

• Fish Abundance Dynamics:

  • Total fish counts per transect differed by region ($p < 0.0001$) and year ($p = 0.0146$).
  • Positive relationship with kelp abundance ($p = 0.037$), whereas direct effect of SST was non-significant ($p = 0.851$).
  • Regional mean fish abundance ($\text{ind}\cdot\text{transect}^{-1}$):
  • Peniche: $22.2 \pm 2.07$
  • Vigo: $22.1 \pm 1.77$
  • Coruña: $14.1 \pm 2.07$
  • Viana: $8.2 \pm 0.86$
  • Annual mean fish abundance peak occurred in 2023 ($20.3 \pm 1.93\,\text{ind}\cdot\text{transect}^{-1}$), followed by a steep collapse in 2024 ($10.5 \pm 1.44\,\text{ind}\cdot\text{transect}^{-1}$), tracking the drop in kelp density.
  • Quantitative effect size: Predictive models demonstrate a $1.04$–fold increase in total fish abundance for every $1000$–individual increase in kelp count.

Fish Community Size Structure and Recruitment Dynamics

• Spatial Variation in Size Classes:

  • Size-class proportions varied significantly by region ($p = 0.00017$) and year ($p < 0.0001$).
  • Vigo exhibited a persistent enrichment of small (juvenile) fish ($p < 0.001$), with standardized Pearson residuals exceeding $+4$. This highlights Vigo's key functional role as a coastal nursery ground driven by lower wave energy and larval retention in the Rías.
  • Coruña, Viana, and Peniche had lower relative proportions of small fish, showing no significant differences among each other ($p \ge 0.27$ for small; $p \ge 0.46$ for medium).
  • Large adult fish densities did not differ significantly across regions ($p > 0.05$).

• Temporal Recruitment Shifts:

  • Coast-wide collapse in juvenile (small) and large fish proportions occurred in 2021 ($p < 0.0001$) across all four regions.
  • Partial size-class recovery took place during 2022 and 2023 before dropping again in 2024.
  • Medium-sized fish exhibited an inverse temporal trend, surging in relative proportion during 2021.

Community Temperature Index and Canopy Climate Buffering

• Methodology & Mathematical Formulation:
  • Species Thermal Index (STI): Defined as the median ($50\text{th}$ percentile) annual mean SST across a species' global distribution, derived from the National Marine Fisheries Service Atlantic (NMFSA) database or AquaMaps midpoint ($10\text{th}\text{--}90\text{th}$ percentile midpoint).
  • Community Temperature Index (CTI): Calculated per transect as the abundance-weighted mean of constituent species' STIs:

$\text{CTI}_{r,y} = \frac{\sum_{i} N_{i,t,y} \times \text{STI}_i}{\sum_{i} N_{i,r,y}}$

• Where $N_{i,t,y}$ is the density of species $i$ at transect $t$ in year $y$.

  • Spatiotemporal CTI Patterns:

• Mixed models confirmed significant main effects for region ($p < 0.0001$), year ($p < 0.0001$), and region $\times$ year interaction ($p < 0.0001$).

• Five-year regional mean CTI values:

  • Peniche (warmest): $16.4 \pm 0.10\,^\text{o}\text{C}$
  • Coruña: $14.7 \pm 1.36\,^\text{o}\text{C}$
  • Vigo: $14.0 \pm 1.12\,^\text{o}\text{C}$
  • Viana: $13.9 \pm 1.56\,^\text{o}\text{C}$

• Temporal CTI dynamics: Stable between 2020 ($14.5 \pm 0.20\,^\text{o}\text{C}$) and 2021 ($14.4 \pm 0.19\,^\text{o}\text{C}$), rising significantly in 2023 ($15.2 \pm 0.15\,^\text{o}\text{C}$, $p = 0.020$ vs 2020; $p = 0.0015$ vs 2021).

  • Canopy Buffering Effect:

• Kelp abundance exerted a significant negative linear effect on fish CTI ($p = 0.00017$).
• High kelp canopy density buffers the fish community thermal signature, maintaining lower CTI values than predicted by ambient SST.
• Realized buffering magnitude: Dense kelp canopies delay realized CTI shifts by approximately $0.25\,^\text{o}\text{C}$ to $1.33\,^\text{o}\text{C}$ relative to ambient warming.
• At Peniche, low kelp abundance resulted in a loss of buffering capacity, causing CTI to exceed ambient SST and reflecting an advanced stage of tropicalisation.

Multivariate Community Patterns and Indicator Species

• PERMANOVA Multivariate Analysis:

  • Permutational analysis of variance (Bray–Curtis distance on $\log_{10}(x+1)$ transformed abundances, 999 permutations):
  • Region effect: $12.3\%$ of total variance ($p = 0.001$
  • Year effect: $5.5\%$ of total variance ($p = 0.001$
  • Region $\times$ Year interaction: $21.1\%$ of variance ($p = 0.001$
  • Kelp abundance (scaled covariate): $1.9\%$ of variance ($p = 0.001$

• Redundancy Analysis (RDA) & Species Vectors:

• Major explanatory vectors constraining community structure:
  • Pomatoschistus flavescens (Two-spotted goby): Vector length $= 1.02$
  • Diplodus vulgaris (Two-banded seabream): Vector length $= 0.54$
  • Coris melanura: Vector length $= 0.26$
  • Labrus bergylta (Ballan wrasse): Vector length $= 0.25$
• Regional indicator species groupings:
  • Coruña: Pomatoschistus flavescens, Serranus cabrilla, Diplodus vulgaris
  • Viana: Dicentrarchus labrax (European seabass), Parablennius gattorugine, Labrus bergylta
  • Peniche: Diplodus vulgaris, Coris melanura
  • Vigo: Pomatoschistus flavescens, Labrus bergylta
• Temporal species transitions: Diplodus vulgaris dominated community structure in 2020 and 2023, whereas Pomatoschistus flavescens dominated in 2021, 2022, and 2024.

Conceptual Framework and Conservation Implications

• Tropicalisation Sequence & Buffer Dynamics:

• Early tropicalisation indicator: An initial increase in localized fish species diversity ($H'$) paired with rising CTI acts as an early-warning pulse. Warm-affinity generalists colonize newly available space before cold-affinity specialists experience catastrophic population drops.

• Buffer Mechanism: Perennial canopy stands (Laminaria hyperborea and Laminaria ochroleuca) provide critical thermal and physical refugia, slowing CTI increases and stabilizing cold-affinity populations.

  • Management & Policy Recommendations:

• Habitat Conservation: Prioritize active preservation and restoration of perennial kelp stands (Laminaria hyperborea in northern zones, Laminaria ochroleuca in southern zones) to maintain climate refugia.

• Biomonitoring Indicator: Utilize Labrus bergylta density as a ecological proxy for kelp forest structural health and functional integrity across Iberian temperate reefs.

• Fisheries Assessment Models: Implement size-structured stock assessment models. Tropicalisation pulses can artificially inflate total fish abundance and richness while truncating overall body size, risking over-optimistic biomass estimates if individual length distributions are unmonitored.