# Phylogenetic Diversity Hotspots as Climate Refugia for Marine Fish: A Quantitative Transferability Test

## Metadata

- **SPORE ID**: SPR-2026-47F4
- **Domaines**: Botany and Plant Ecology Studies x Marine and fisheries research
- **Date de generation**: 2026-08-21
- **Panel consensus score**: 6.1/10
- **Novelty score**: 0.75/1.0
- **Panel verdict**: publish_brief

## Abstract

If the spatial distribution of phylogenetic diversity (PD) in marine fish communities is projected under climate scenarios, then the top 10% PD hotspots will retain at least 15% more PD than the top 10% species richness (SR) hotspots, because PD captures evolutionary distinctiveness that buffers against functional loss under warming.

The proposed mechanism involves 6 causal steps: (1) Construct a molecular phylogeny for the target fish species pool (e.g., 200-400  -> (2) Compute PD and PE for each 0.25° grid cell using occurrence data from NOAA trawl -> (3) Estimate thermal tolerance (CTmax in °C) for each species from published physiol -> (4) Project species distributions under RCP 4.5 and 8.5 (2050 and 2100) using specie.

Literature grounding on 16 verified references yields a novelty score of 0.75 (novel). A 3-phase experimental protocol (budget: €20k-80k, timeline: 8-14 months) is proposed, starting with in silico validation. A panel of 5 expert reviewers reached a consensus score of 6.1/10.

## 1. Hypothese et mecanisme propose

### 1.1 Formulation formelle

If the spatial distribution of phylogenetic diversity (PD) in marine fish communities is projected under climate scenarios, then the top 10% PD hotspots will retain at least 15% more PD than the top 10% species richness (SR) hotspots, because PD captures evolutionary distinctiveness that buffers against functional loss under warming.

### 1.2 Variables

**Variables independantes :**

| Variable | Type | Plage | Unite |
|----------|------|-------|-------|
| Climate scenario (RCP) | categorical | RCP 4.5 vs. RCP 8.5 | dimensionless |
| Climate model ensemble member | categorical | 3 models (e.g., GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR) | dimensionless |
| Spatial grid cell | continuous | 0.25° x 0.25° (approx. 500-2000 cells in study region) | degrees |

**Variables dependantes :**

| Variable | Type | Direction attendue | Unite |
|----------|------|-------------------|-------|
| Phylogenetic Diversity (PD) retention | continuous | increase | million years (My) |
| Phylogenetic Endemism (PE) retention | continuous | increase | My |
| PD loss per unit area protected | continuous | decrease | My/km² |

### 1.3 Chaine causale

1. Step 1: Construct a molecular phylogeny for the target fish species pool (e.g., 200-400 species) using existing time-calibrated trees (e.g., from FishTreeOfLife), with branch lengths in millions of years.
1. Step 2: Compute PD and PE for each 0.25° grid cell using occurrence data from NOAA trawl surveys (presence/absence), with a minimum of 500 cells containing ≥5 species.
1. Step 3: Estimate thermal tolerance (CTmax in °C) for each species from published physiological data or trait databases; calculate Blomberg's K for CTmax across the phylogeny.
1. Step 4: Project species distributions under RCP 4.5 and 8.5 (2050 and 2100) using species distribution models (SDMs) with dispersal limitation parameterized as a variable (e.g., 10-50 km/decade).
1. Step 5: Recalculate PD and PE under projected distributions; identify top 10% hotspots for PD, PE, and SR under each scenario.
1. Step 6: Compare PD retention (PD_projected / PD_current) between PD hotspots and SR hotspots; also compare PD loss per unit area protected for each hotspot type.

**Hypotheses cles :**

- Assumption 1: The molecular phylogeny is well-resolved and branch lengths are proportional to evolutionary time (ultrametric).
- Assumption 2: Trawl survey data have sufficient spatial coverage and detection probability > 0.3 for all species; otherwise, occupancy-detection models must be applied.
- Assumption 3: Thermal tolerance (CTmax) is phylogenetically conserved with Blomberg's K ≥ 0.5; if K < 0.5, the framework is invalid.
- Assumption 4: Dispersal limitation is a variable parameter, not a fixed constant, and its effect on PD retention is monotonic within the tested range.

**Inconnues identifiees :**

- We do not know whether PD hotspots will remain stable under climate change or shift to new locations, which affects the definition of 'refugia'.
- We do not know whether high PD confers resilience to warming or merely reflects historical biogeographic patterns unrelated to thermal tolerance.

### 1.4 Conditions aux limites

- **Blomberg's K for CTmax must be ≥ 0.5 in the target species pool.** — If K < 0.5, thermal tolerance is not phylogenetically conserved, and PD cannot serve as a proxy for climate resilience.
- **The study region must contain at least 500 grid cells with ≥5 species each.** — Power analysis indicates that 500 cells are required to detect a 15% PD retention difference with 80% power at alpha = 0.05, assuming spatial autocorrelation of 0.5.
- **Climate projections must be limited to RCP 4.5 and 8.5, with 3 ensemble members.** — Sensitivity analysis across scenarios and models is necessary to ensure robustness; RCP 8.5 represents the high-emission boundary condition.

### 1.5 Cadre theorique

Phylogenetic conservation prioritization (Faith's PD framework) integrated with climate refugia theory (evolutionary refugia concept).

## 2. Etat de l'art et positionnement

### 2.1 Travaux les plus proches

- **[2022] Disentangling the components of coastal fish biodiversity in southern Brittany by applying an environmental
 DNA
 approach** — [10.1002/edn3.305](https://doi.org/10.1002/edn3.305)
  - Similarite: related
  - Difference cle: This paper applies phylogenetic diversity concepts to marine fish but focuses on eDNA-based monitoring rather than explicitly linking PD hotspots to climate refugia or conservation prioritization, and does not use trawl survey data.
- **[2011] Phylogenetic Analysis of Conservation Priorities for Aquatic Mammals and Their Terrestrial Relatives, with a Comparison of Methods** — [10.1371/journal.pone.0022562](https://doi.org/10.1371/journal.pone.0022562)
  - Similarite: related
  - Difference cle: This paper applies phylogenetic conservation prioritization to aquatic mammals, not fish, and does not focus on spatial hotspot mapping or climate refugia identification.
- **[2021] Plant Endemism Centres and Biodiversity Hotspots in Greece** — [10.3390/biology10020072](https://doi.org/10.3390/biology10020072)
  - Similarite: analogous
  - Difference cle: This is a botany paper demonstrating the exact methodology (phylogenetically informed hotspot identification) but applied to plants, not marine fish, and does not address climate refugia in a marine context.

### 2.2 Base de preuves

- **[2021] Plant Endemism Centres and Biodiversity Hotspots in Greece** — [10.3390/biology10020072](https://doi.org/10.3390/biology10020072)
  - Type: analogous | Citations: 76
  - Demonstrates the methodological framework of combining phylogenetic diversity metrics with species distribution data to identify conservation hotspots, which is the core transferable method proposed in the hypothesis.
- **[2017] Hotspot analyses indicate significant conservation gaps for evergreen broadleaved woody plants in China** — [10.1038/s41598-017-02098-0](https://doi.org/10.1038/s41598-017-02098-0)
  - Type: analogous | Citations: 65
  - Provides a concrete example of using hotspot and gap analyses in botany to identify conservation priorities, supporting the transferability of these spatial-phylogenetic methods to other taxa.
- **[2021] Prioritizing conservation of biodiversity in an alpine region: Distribution pattern and conservation status of seed plants in the Qinghai-Tibetan Plateau** — [10.1016/J.GECCO.2021.E01885](https://doi.org/10.1016/J.GECCO.2021.E01885)
  - Type: analogous | Citations: 32
  - Shows the integration of species richness, complementarity, and spatial phylogenetics to identify hotspots, directly supporting the methodological transfer proposed in the hypothesis.
- **[2021] Spatial phylogenetics of the native woody plant species in Hainan, China** — [10.1002/ece3.7180](https://doi.org/10.1002/ece3.7180)
  - Type: analogous | Citations: 20
  - Provides another example of spatial phylogenetics applied to identify biodiversity hotspots, reinforcing the methodological framework that the hypothesis proposes to transfer to marine fish.
- **[2022] Biodiversity hotspots and conservation efficiency of a large drainage basin: Distribution patterns of species richness and conservation gaps analysis in the Yangtze River Basin, China** — [10.1111/csp2.12653](https://doi.org/10.1111/csp2.12653)
  - Type: analogous | Citations: 20
  - Demonstrates the use of phylogenetic reconstruction and hotspot analysis for conservation planning in a large basin, supporting the transferability of these methods to marine systems.
- **[2020] Are we ready to track climate‐driven shifts in marine species across international boundaries? ‐ A global survey of scientific bottom trawl data** — [10.1111/gcb.15404](https://doi.org/10.1111/gcb.15404)
  - Type: indirect | Citations: 67
  - Provides evidence that bottom trawl survey data are available globally and are used to track climate-driven shifts in marine species, supporting the feasibility of using such data for PD mapping.
- **[2020] Are we ready to track climate-driven shifts in marine species across international boundaries? - A global survey of scientific bottom trawl data** — [10.1101/2020.06.18.125930](https://doi.org/10.1101/2020.06.18.125930)
  - Type: indirect | Citations: 21
  - Preprint version of the above paper, confirming the availability and use of bottom trawl data for tracking climate-driven shifts in marine species.
- **[2022] Disentangling the components of coastal fish biodiversity in southern Brittany by applying an environmental
 DNA
 approach** — [10.1002/edn3.305](https://doi.org/10.1002/edn3.305)
  - Type: indirect | Citations: 20
  - Shows that phylogenetic diversity is being considered in marine fish biodiversity assessments, indicating a growing interest and feasibility of applying PD concepts to fish.
- **[2011] Phylogenetic Analysis of Conservation Priorities for Aquatic Mammals and Their Terrestrial Relatives, with a Comparison of Methods** — [10.1371/journal.pone.0022562](https://doi.org/10.1371/journal.pone.0022562)
  - Type: indirect | Citations: 30
  - Demonstrates the application of phylogenetic conservation prioritization to aquatic species, supporting the feasibility of using PD metrics for marine taxa.
- **[2013] Evolutionary refugia and ecological refuges: key concepts for conserving Australian arid zone freshwater biodiversity under climate change** — [10.1111/gcb.12203](https://doi.org/10.1111/gcb.12203)
  - Type: indirect | Citations: 238
  - Provides a conceptual framework for evolutionary refugia, which is central to the hypothesis that PD hotspots may serve as climate refugia.
- **[2019] Climate resilience in marine protected areas and the ‘Protection Paradox’** — [10.1016/J.BIOCON.2019.05.005](https://doi.org/10.1016/J.BIOCON.2019.05.005)
  - Type: indirect | Citations: 188
  - Discusses climate resilience in marine protected areas, providing context for the hypothesis that PD hotspots may predict resilience to climate change.
- **[2018] Genetic diversity in fishes is influenced by habitat type and life‐history variation** — [10.1002/ece3.4661](https://doi.org/10.1002/ece3.4661)
  - Type: indirect | Citations: 124
  - Provides evidence that genetic diversity in fishes varies with habitat and life-history traits, supporting the idea that evolutionary information is relevant for fish conservation.
- **[2023] eDNA metabarcoding enriches traditional trawl survey data for monitoring biodiversity in the marine environment** — [10.1093/icesjms/fsad083](https://doi.org/10.1093/icesjms/fsad083)
  - Type: indirect | Citations: 53
  - Shows that trawl survey data can be enriched with eDNA data for biodiversity monitoring, suggesting that comprehensive species occurrence data for PD mapping are becoming more available.
- **[2021] Linking Taxonomic, Phylogenetic and Functional Plant Diversity with Ecosystem Services of Cliffs and Screes in Greece** — [10.3390/plants10050992](https://doi.org/10.3390/plants10050992)
  - Type: analogous | Citations: 26
  - Demonstrates the integration of phylogenetic diversity with ecosystem services in plants, supporting the idea that PD can be linked to conservation value.
- **[2021] The biogeographic history of eelpouts and related fishes: linking phylogeny, environmental change, and patterns of dispersal in a globally distributed fish group** — [10.1101/2021.01.11.426281](https://doi.org/10.1101/2021.01.11.426281)
  - Type: indirect | Citations: 18
  - Provides an example of a fish phylogeny being constructed and used to understand biogeographic history, supporting the feasibility of building fish phylogenies for PD calculations.

### 2.3 Contre-preuves et limitations connues

- **[addressable] Climate resilience in marine protected areas and the ‘Protection Paradox’**
  - The paper suggests that protection from local pressures does not always lead to increased resilience to climate events, which could challenge the assumption that PD hotspots (as conservation priorities) will necessarily buffer against climate change.
- **[?] The Rapid Population Collapse of a Key Marine Predator in the Northern Antarctic Peninsula Endangers Genetic Diversity and Resilience to Climate Change**
  - 

### 2.4 Evaluation de nouveaute

- **Score**: 0.75/1.0
- **Verdict**: novel

## 3. Predictions falsifiables

| # | Prediction | Borne quantitative | Methode | H0 | Test statistique |
|---|-----------|-------------------|---------|-----|-----------------|
| 1 | The top 10% PD hotspots will retain at least 15% more PD than the top 10% SR hotspots under RCP 8.5 by 2100. | PD retention difference = (PD_retention_PD_hotspots - PD_retention_SR_hotspots) ≥ 15% (absolute percentage points) | Projected PD calculated from SDM outputs and phylogeny; hotspots defined as top 10% of grid cells by PD or SR. | H0: The mean PD retention difference between PD and SR hotspots is ≤ 0%. | Paired t-test (or Wilcoxon signed-rank if non-normal) on cell-wise PD retention differences, with spatial autocorrelation accounted for via spatial generalized least squares (GLS) with a spherical correlation structure; alpha = 0.05. |
| 2 | Blomberg's K for CTmax will be ≥ 0.5 across the target species pool. | K ≥ 0.5 (dimensionless), with 95% confidence interval excluding values < 0.4 | Phylogenetic signal analysis using the 'phytools' R package on CTmax data (in °C) from published sources. | H0: K = 0 (no phylogenetic signal). | Randomization test (999 permutations) comparing observed K to null distribution; alpha = 0.05. |
| 3 | PD loss per unit area protected will be at least 20% lower for PD hotspots compared to SR hotspots. | PD_loss_per_area_PD_hotspots ≤ 0.8 × PD_loss_per_area_SR_hotspots | Calculate PD loss (My) per km² for each hotspot type from projected PD and area of protected cells. | H0: PD loss per unit area is equal between PD and SR hotspots. | Bootstrap 95% confidence intervals for the ratio of PD loss per area; reject H0 if CI excludes 1.0. |

## 4. Protocole experimental

**Timeline globale**: 8-14 months
**Budget global**: €20k-80k

### 4.1 Phase 1 — In Silico Validation

**Objectif**: To rapidly falsify the core assumption (Blomberg's K ≥ 0.5 for CTmax) and to provide a preliminary, low-cost test of the central prediction (PD hotspots retain ≥15% more PD than SR hotspots) using existing data and simplified projections.

**Methodologie**: 1. **Phylogeny Assembly**: Extract the target species list (200-400 species) from NOAA trawl survey data. Use the 'fishphylomaker' R package to prune a time-calibrated tree from the FishTreeOfLife database. 2. **CTmax Data Compilation**: Compile CTmax data from published sources (e.g., FishBase, primary literature) for all species. If data are missing for >30% of species, use phylogenetic imputation with the 'Rphylopars' R package. 3. **Phylogenetic Signal Test**: Calculate Blomberg's K for CTmax using the 'phytools' R package. Perform a randomization test (999 permutations) to assess significance. 4. **Preliminary PD/SR Projection**: Use a simplified 'climate velocity' approach. For each grid cell, calculate current PD and SR. Project future distributions by shifting species ranges poleward based on their thermal niche (CTmax) and the projected warming velocity under RCP 8.5 for 2100 (using data from the 'sdmpredictors' R package). This is a coarse, correlative projection, not a full SDM. 5. **Hotspot Comparison**: Identify top 10% PD and SR hotspots in current and projected distributions. Calculate the mean PD retention difference between PD and SR hotspots.

- Cout: €500-2000
- Duree: 4-6 weeks
- Equipement: Standard workstation (8+ cores, 32GB RAM)
- Logiciels: R (v4.2+), RStudio, R packages: phytools, fishphylomaker, Rphylopars, raster, sf, dplyr, ggplot2

**Criteres de succes :**

- Blomberg's K for CTmax: K ≥ 0.5, with 95% CI excluding values < 0.4
- Preliminary PD retention difference: PD retention difference ≥ 5% (preliminary, lower bar than final 15% due to simplified projections)

- **GO**: Both success criteria are met (K ≥ 0.5 and preliminary PD retention difference ≥ 5%).
- **NO-GO**: Blomberg's K < 0.5, or the preliminary PD retention difference is ≤ 0%.
- **PIVOT**: If K ≥ 0.5 but the preliminary PD retention difference is between 0% and 5%, refine the simplified projection (e.g., use a more detailed dispersal model) before deciding. If data for CTmax are too sparse, pivot to a targeted literature search or physiological experiment (Phase 2) to fill critical gaps.

### 4.2 Phase 2 — Minimal Experimental Validation

**Objectif**: To empirically validate the mechanistic link between PD and thermal tolerance (CTmax) for a subset of species, and to refine the SDM framework with a small, targeted dataset to confirm the PD retention signal under a more realistic projection.

**Methodologie**: 1. **Physiological Validation (Sub-experiment A)**: Select 10-15 species representing a range of phylogenetic distinctiveness (high and low PD contribution). Measure CTmax using a standardized thermal ramping protocol (e.g., 0.3°C/min) in a temperature-controlled water bath. This provides direct, high-quality data to validate the imputed CTmax values from Phase 1 and confirm the phylogenetic signal. 2. **Refined SDM & Projection (Sub-experiment B)**: Build a more robust SDM for all species using the 'biomod2' R package. Use environmental predictors (e.g., bottom temperature, salinity, depth) from public databases (Bio-ORACLE). Calibrate models with current occurrence data, then project to 2050 and 2100 under RCP 4.5 and 8.5. Use a more realistic dispersal scenario (e.g., 50 km/decade) compared to Phase 1. 3. **Hotspot Analysis & Comparison**: Recalculate PD and PE under the new projections. Identify top 10% hotspots for PD, PE, and SR. Calculate the primary prediction: PD retention difference between PD and SR hotspots.

- Cout: €5k-15k
- Duree: 2-3 months
- Equipement: Temperature-controlled water baths (2-3 units), Aquaria and aeration systems, Digital thermometers/loggers, Standard wet lab consumables
- Logiciels: R (v4.2+), R packages: biomod2, raster, sf, phytools, ggplot2

**Criteres de succes :**

- PD retention difference (primary prediction): PD retention difference ≥ 10% (interim threshold, lower than final 15% due to smaller scale)
- Phylogenetic signal (K) with empirical data: K ≥ 0.5 with the new empirical CTmax data

- **GO**: Both success criteria are met (PD retention difference ≥ 10% and K ≥ 0.5).
- **NO-GO**: PD retention difference is < 5% or K < 0.5 with empirical data.
- **PIVOT**: If PD retention difference is between 5% and 10%, investigate whether the signal is weak due to specific model choices (e.g., dispersal rate) or spatial scale. Consider adjusting the hotspot definition (e.g., top 15%) or the study region. If K is borderline (0.4-0.5), consider a different phylogenetic metric or a trait-based approach.

### 4.3 Phase 3 — Full Experimental Protocol

**Objectif**: To conduct a rigorous, publication-ready validation of the hypothesis, including full uncertainty quantification, spatial autocorrelation analysis, and a comprehensive comparison of PD, PE, and SR hotspots under multiple climate scenarios and models.

**Methodologie**: 1. **Comprehensive Data & Phylogeny**: Finalize the species pool and phylogeny. Use the most recent and complete version of FishTreeOfLife. Ensure all branch lengths are ultrametric. 2. **Full SDM & Projection Ensemble**: Run a full ensemble of SDMs (e.g., 10 algorithms) for all species using 'biomod2'. Use 3 GCMs (GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR) and 2 RCPs (4.5, 8.5). Incorporate a mechanistic dispersal model (e.g., a simple cellular automaton) with a range of dispersal rates (10, 30, 50 km/decade). 3. **Phylogenetic Diversity Metrics**: Calculate PD, PE, and SR for each grid cell for current and all future projections. Use the 'phyloregion' R package for efficient computation. 4. **Hotspot Identification & Comparison**: Identify top 10% hotspots for PD, PE, and SR for each scenario. Calculate PD retention, PE retention, and PD loss per unit area for each hotspot type. 5. **Statistical Analysis**: Use spatial Generalized Least Squares (GLS) with a spherical correlation structure to test the primary prediction, accounting for spatial autocorrelation. Use bootstrap to calculate confidence intervals for the ratio of PD loss per area. 6. **Sensitivity Analysis**: Test the sensitivity of results to key assumptions (e.g., K threshold, hotspot definition, dispersal rate, GCM).

- Cout: €20k-80k+
- Duree: 6-12 months
- Equipement: High-performance computing cluster or cloud computing (e.g., AWS, Google Cloud) for large-scale SDM runs
- Logiciels: R (v4.2+), R packages: biomod2, phyloregion, phytools, raster, sf, spdep, nlme, boot, ggplot2, viridis, Python (optional, for custom dispersal models)

**Criteres de succes :**

- Primary prediction (PD retention difference): PD retention difference ≥ 15% (absolute percentage points) under RCP 8.5 by 2100, with the 95% CI of the GLS model excluding 0.
- Secondary prediction (PD loss per area): PD loss per unit area for PD hotspots is ≤ 0.8 × that of SR hotspots, with bootstrap 95% CI excluding 1.0.
- Robustness across scenarios: The primary prediction holds (≥15% difference) in at least 5 out of 6 scenario combinations (3 GCMs × 2 RCPs).

- **GO**: All three success criteria are met.
- **NO-GO**: The primary prediction fails (PD retention difference < 15%) or the secondary prediction fails (ratio > 0.8).
- **PIVOT**: If the primary prediction holds but the secondary fails, investigate the spatial configuration of hotspots. If PD hotspots are large and diffuse, they may protect more total PD but be less efficient per unit area. Consider alternative conservation metrics (e.g., PE). If results are highly sensitive to GCM, focus on the consensus across models and report this uncertainty prominently.

### 4.4 Quick start : comment demarrer aujourd'hui

- **Peut demarrer maintenant**: Oui
- **Premiere action**: Download the NOAA trawl survey occurrence data from OBIS and the FishTreeOfLife phylogeny. Install R and the required packages (phytools, fishphylomaker, raster, sf).
- **Outils**: R, RStudio, R packages: phytools, fishphylomaker, raster, sf
- **Donnees ouvertes**: OBIS (Ocean Biodiversity Information System), FishTreeOfLife (fish-tree-of-life.org), FishBase (for CTmax data), Bio-ORACLE (for environmental layers)

## 5. Analyse d'impact

### 5.1 Impact scientifique

Novelty score: 0.75/1.0 (novel)

### 5.2 Applications industrielles et marche

**Score industriel**: 6.5/10

**Forces :**
- Cible un besoin réglementaire et ESG émergent : les grands groupes de l'agroalimentaire et de la pêche (ex. Thai Union, Nomad Foods, Maruha Nichiro) et les assureurs (ex. AXA Climate) doivent justifier la résilience de leurs chaînes d'approvisionnement face au changement climatique. Un indice de 'refuges évolutifs' monétisable comme service de scoring de risque pourrait être intégré à des plateformes existantes (ex. Global Fishing Watch, Ocean Data Alliance) pour un abonnement B2B annuel de 50k€-200k€.
- Avantage différenciant fort : contrairement aux cartes de richesse spécifique (SR) déjà saturées par les ONG (ex. WWF, IUCN), ce produit combine phylogénie et refuges climatiques, offrant une couche de données unique pour la sélection de zones de 'no-take' ou de corridors. Cela permet de breveter un algorithme de scoring (méthode de calcul du ratio PD_loss/area sous scénarios climatiques) et de créer une barrière à l'entrée logicielle, même si les données sources sont publiques (OBIS, AquaMaps).

**Faiblesses :**
- Marché adressable principal limité aux gouvernements et grandes ONG de conservation (ex. Blue Nature Alliance, Pew Charitable Trusts) avec des cycles de vente longs (18-36 mois) et des budgets discrétionnaires. Le segment privé (pêche, assurance) est potentiel mais n'a pas encore démontré sa volonté de payer pour ce type de métrique 'évolutionnaire' plutôt que pour des données de biomasse ou de rendement halieutique, ce qui rend le ROI immédiat incertain.
- Barrière scientifique majeure : la dépendance à la valeur de Blomberg's K (≥0.5) pour CTmax est un point de fragilité. Si la Phase 1 échoue (K < 0.5), l'hypothèse s'effondre. De plus, la résolution spatiale des données de PD pour les poissons marins est souvent grossière (maillage 1°), ce qui limite la précision opérationnelle pour des zones côtières de 10-50 km², là où les gestionnaires ont besoin d'agir, réduisant la valeur perçue du produit.

**Recommandation**: Ne pas lancer un produit complet immédiatement. Financer uniquement la Phase 1 (in silico, <2k€) avec un partenaire académique (ex. IRD, CSIRO) pour valider le K. En parallèle, signer une lettre d'intention avec une ONG de conservation marine (ex. Blue Nature Alliance) pour co-développer un 'pilote' sur une écorégion (ex. Triangle de Corail) et tester la volonté de payer. Si la Phase 1 est positive, lever des fonds pour la Phase 3 et cibler la vente de données à des assureurs paramétriques (ex. AXA Climate) pour l'assurance de récifs, en utilisant la PD comme proxy de résilience, avec un prix de 150k€ par contrat.

### 5.3 Opportunites de financement

| Programme | Agence | Fit | Budget type | Taux succes |
|-----------|--------|-----|-------------|-------------|
| ANR Jeunes Chercheuses et Jeunes Chercheurs (JCJC) 2026 | Agence Nationale de la Recherche (ANR), France | 0.85 | €150,000 - €300,000 | ~15-20% (varies by panel) |
| ERC Starting Grant (StG) 2026 | European Research Council (ERC) | 0.75 | €1.5 million (for 5 years) | ~13-15% (for StG) |
| Biodiversity and Ecosystem Services Call (Cluster 6, Destination 'Biodiversity and Ecosystems') - Horizon Europe | European Commission (DG Research & Innovation) | 0.65 | €5-10 million (for collaborative projects, RIA) | ~10-15% (for RIA topics) |

- **ANR Jeunes Chercheuses et Jeunes Chercheurs (JCJC) 2026** (Agence Nationale de la Recherche (ANR), France): This is the ideal entry point. The JCJC program is designed for early-career researchers to lead their own project. The phased, low-cost protocol (especially Phase 1 & 2) fits perfectly within the budget. It allows the PI to generate the preliminary data needed to build a stronger, more competitive application for a larger European grant. The 'transferability test' aspect is a strong, novel concept that can shine in a national competition.
- **ERC Starting Grant (StG) 2026** (European Research Council (ERC)): The ERC StG is the perfect fit for the full, ambitious 3-phase vision. The 'bottom-up' nature of ERC is ideal for this hypothesis-driven research. The core concept—quantitatively testing if PD is a better refugia indicator than SR—is a frontier question in ecology. The phased approach demonstrates scientific maturity and risk management, which is highly valued. The PI would need to build a small consortium (e.g., a PhD student and a postdoc) and secure access to experimental facilities, but the project is perfectly sized for a StG.
- **Biodiversity and Ecosystem Services Call (Cluster 6, Destination 'Biodiversity and Ecosystems') - Horizon Europe** (European Commission (DG Research & Innovation)): This is a strategic alternative if the PI can integrate the hypothesis into a larger, policy-oriented consortium. The project's focus on climate refugia and marine biodiversity directly aligns with the EU's Biodiversity Strategy for 2030. The 'transferability test' could be a key work package in a larger project aimed at developing decision-support tools for MPA design. However, this requires a significant shift in scope and consortium building, moving from a single-PI project to a collaborative, multi-partner project.

**Recommandation financement**: Position this as a high-risk, high-reward 'proof-of-concept' project. The ideal strategy is to submit the Phase 1 and 2 components to the ANR JCJC or a similar national early-career grant to secure initial funding and generate the preliminary data. Simultaneously, use the results from Phase 1 to build a stronger consortium and submit the full 3-phase protocol to an ERC Starting Grant, framing it as a foundational step in predictive evolutionary conservation. The phased design is a perfect narrative for a 'StG' which values ambitious, high-risk research with a clear potential for breakthrough.

## 6. Panel Review Summary

| Reviewer | Score | Verdict | Point cle |
|----------|-------|---------|-----------|
| methodologist | 6.8/10 | weak_accept | The staged, phased approach (in silico → minimal experimental → full protocol) with explicit GO/NO-GO/PIVOT criteria at each phase is an excellent example of adaptive experimental design. This allows for early falsification of core assumptions (e.g., Blomberg's K) before committing significant resources, which is a hallmark of rigorous scientific methodology. |
| domain_expert | 6.5/10 | weak_accept | The hypothesis directly addresses a critical gap in marine conservation: the explicit integration of phylogenetic diversity (PD) with climate refugia theory. While spatial phylogenetics is well-established in terrestrial systems, its application to dynamic marine environments under climate change, particularly using the PD retention metric, is a timely and relevant extension. |
| contrarian | 3.5/10 | weak_reject | The explicit articulation of a falsifiable prediction (≥15% PD retention advantage) with a defined null hypothesis is methodologically sound and rare in this literature. |
| industrialist | 6.5/10 | weak_accept | Cible un besoin réglementaire et ESG émergent : les grands groupes de l'agroalimentaire et de la pêche (ex. Thai Union, Nomad Foods, Maruha Nichiro) et les assureurs (ex. AXA Climate) doivent justifier la résilience de leurs chaînes d'approvisionnement face au changement climatique. Un indice de 'refuges évolutifs' monétisable comme service de scoring de risque pourrait être intégré à des plateformes existantes (ex. Global Fishing Watch, Ocean Data Alliance) pour un abonnement B2B annuel de 50k€-200k€. |
| funding_strategist | 7.2/10 | accept | Originality and conceptual novelty: The hypothesis directly challenges the conventional use of species richness (SR) as a proxy for conservation priority by proposing a quantitative, transferable test of phylogenetic diversity (PD) as a climate refugia indicator. This addresses a recognized gap in macroecology and conservation planning, positioning the project at the forefront of evidence-based biodiversity management. |

**Consensus score**: 6.1/10
**Verdict final**: publish_brief

### 6.1 Consensus

- Le protocole en phases (in silico → expérimental → complet) avec des critères GO/NO-GO/PIVOT explicites est une force majeure, saluée par tous les reviewers pour sa rigueur méthodologique et sa gestion du risque.
- La formulation d'une prédiction quantitative et falsifiable (rétention de PD ≥ 15% dans les hotspots PD vs SR) avec une hypothèse nulle claire est unanimement considérée comme exemplaire et rare dans la littérature.
- L'utilisation stratégique de données existantes (FishTreeOfLife, NOAA trawl surveys) pour la Phase 1 permet une falsification rapide et à faible coût de l'hypothèse centrale, ce qui est un point fort pour le financement et la faisabilité.

### 6.2 Points de desaccord

- Le principal désaccord porte sur le mécanisme causal : le domain_expert et le contrarian estiment que la chaîne causale PD → résilience est sous-développée et confond PD avec diversité fonctionnelle, tandis que le methodologist et le funding_strategist considèrent que le protocole est suffisamment solide pour tester cette relation de manière empirique.
- Un désaccord existe sur la définition de 'refuges climatiques' : le domain_expert souligne que la mesure de 'rétention de PD' pourrait simplement identifier des zones à faible vélocité climatique plutôt que de véritables refuges, alors que le funding_strategist voit cette approche comme une avancée conceptuelle suffisante pour une première validation.
- Le contrarian rejette l'hypothèse en raison d'un biais potentiel d'échantillonnage spatial dans les données de chalutage NOAA et d'un manque de contrôle pour la configuration spatiale des hotspots, tandis que le methodologist propose des solutions (modèle nul spatial) sans considérer cela comme fatal.

### 6.3 Critical path

Le facteur le plus déterminant est la validation de l'hypothèse mécanistique reliant la diversité phylogénétique à la résilience thermique (CTmax). La Phase 1, qui teste si Blomberg's K ≥ 0.5 pour CTmax, est le point de bascule : si cette condition échoue, l'hypothèse s'effondre. De plus, la distinction entre un effet de la PD et un simple artefact de configuration spatiale ou de vélocité climatique doit être clarifiée pour que la prédiction de 15% soit interprétable.

**Recommandation finale**: Le panel recommande de publier le brief pour l'itération actuelle, en considérant que l'hypothèse est suffisamment prometteuse et le protocole suffisamment rigoureux pour justifier une validation empirique. Les critiques du contrarian, bien que sérieuses, ne sont pas fatales à ce stade : elles doivent être traitées comme des conditions de validation dans le protocole expérimental (ajout d'un modèle nul spatial, contrôle de la vélocité climatique, test de la couverture de l'espace des traits). Le brief doit être publié avec une mention explicite que la Phase 1 est un test de faisabilité critique, et que les résultats de cette phase détermineront si l'hypothèse mérite d'être poursuivie. Les recommandations du funding_strategist (cibler ANR JCJC pour les phases 1-2) sont intégrées comme stratégie de financement optimale.

## 7. Gap Manifest residuel

### 7.1 Data gaps

- The direct link between high PD and climate resilience in marine fish remains unproven, with at least one case (Antarctic fur seals) suggesting that high genetic diversity does not guarantee resilience.
- The specific algorithms for spatial-phylogenetic hotspot mapping (e.g., PD, PE) have not been explicitly applied to marine fish trawl data in the provided papers, requiring further methodological validation.

### 7.2 Competence gaps

- Phase 1: R programming
- Phase 1: Phylogenetic comparative methods
- Phase 1: Spatial data analysis (GIS)
- Phase 1: Basic ecological niche modeling
- Phase 2: Experimental physiology (thermal tolerance)
- Phase 2: Advanced SDM using biomod2
- Phase 2: Spatial statistics
- Phase 2: R programming
- Phase 3: Advanced ecological niche modeling
- Phase 3: Phylogenetic comparative methods
- Phase 3: Spatial statistics (GLS, autocorrelation)
- Phase 3: High-performance computing
- Phase 3: Scientific writing

### 7.3 Epistemic gaps

- We do not know whether PD hotspots will remain stable under climate change or shift to new locations, which affects the definition of 'refugia'.
- We do not know whether high PD confers resilience to warming or merely reflects historical biogeographic patterns unrelated to thermal tolerance.

## References

[1] Konstantinos Kougioumoutzis, Ioannis P. Kokkoris, et al. (2021). *Plant Endemism Centres and Biodiversity Hotspots in Greece*. DOI: [10.3390/biology10020072](https://doi.org/10.3390/biology10020072)
[2] Yue Xu, Zehao Shen, et al. (2017). *Hotspot analyses indicate significant conservation gaps for evergreen broadleaved woody plants in China*. DOI: [10.1038/s41598-017-02098-0](https://doi.org/10.1038/s41598-017-02098-0)
[3] Tiantian Xue, S. Gadagkar, et al. (2021). *Prioritizing conservation of biodiversity in an alpine region: Distribution pattern and conservation status of seed plants in the Qinghai-Tibetan Plateau*. DOI: [10.1016/J.GECCO.2021.E01885](https://doi.org/10.1016/J.GECCO.2021.E01885)
[4] Zhi-Xin Zhu, AJ Harris, et al. (2021). *Spatial phylogenetics of the native woody plant species in Hainan, China*. DOI: [10.1002/ece3.7180](https://doi.org/10.1002/ece3.7180)
[5] Wendi Zhang, R. Bussmann, et al. (2022). *Biodiversity hotspots and conservation efficiency of a large drainage basin: Distribution patterns of species richness and conservation gaps analysis in the Yangtze River Basin, China*. DOI: [10.1111/csp2.12653](https://doi.org/10.1111/csp2.12653)
[6] Aurore Maureaud, R. Frelat, et al. (2020). *Are we ready to track climate‐driven shifts in marine species across international boundaries? ‐ A global survey of scientific bottom trawl data*. DOI: [10.1111/gcb.15404](https://doi.org/10.1111/gcb.15404)
[7] Aurore Maureaud, R. Frelat, et al. (2020). *Are we ready to track climate-driven shifts in marine species across international boundaries? - A global survey of scientific bottom trawl data*. DOI: [10.1101/2020.06.18.125930](https://doi.org/10.1101/2020.06.18.125930)
[8] R. Różański, V. Trenkel, et al. (2022). *Disentangling the components of coastal fish biodiversity in southern Brittany by applying an environmental
 DNA
 approach*. DOI: [10.1002/edn3.305](https://doi.org/10.1002/edn3.305)
[9] L. May‐Collado, I. Agnarsson (2011). *Phylogenetic Analysis of Conservation Priorities for Aquatic Mammals and Their Terrestrial Relatives, with a Comparison of Methods*. DOI: [10.1371/journal.pone.0022562](https://doi.org/10.1371/journal.pone.0022562)
[10] Jenny Davis, A. Pavlova, et al. (2013). *Evolutionary refugia and ecological refuges: key concepts for conserving Australian arid zone freshwater biodiversity under climate change*. DOI: [10.1111/gcb.12203](https://doi.org/10.1111/gcb.12203)
[11] A. Bates, Robert S. C. Cooke, et al. (2019). *Climate resilience in marine protected areas and the ‘Protection Paradox’*. DOI: [10.1016/J.BIOCON.2019.05.005](https://doi.org/10.1016/J.BIOCON.2019.05.005)
[12] Alexander S Martinez, J. Willoughby, et al. (2018). *Genetic diversity in fishes is influenced by habitat type and life‐history variation*. DOI: [10.1002/ece3.4661](https://doi.org/10.1002/ece3.4661)
[13] X. He, N. Jeffery, et al. (2023). *eDNA metabarcoding enriches traditional trawl survey data for monitoring biodiversity in the marine environment*. DOI: [10.1093/icesjms/fsad083](https://doi.org/10.1093/icesjms/fsad083)
[14] M. Panitsa, Ioannis P. Kokkoris, et al. (2021). *Linking Taxonomic, Phylogenetic and Functional Plant Diversity with Ecosystem Services of Cliffs and Screes in Greece*. DOI: [10.3390/plants10050992](https://doi.org/10.3390/plants10050992)
[15] S. Hotaling, M. Borowiec, et al. (2021). *The biogeographic history of eelpouts and related fishes: linking phylogeny, environmental change, and patterns of dispersal in a globally distributed fish group*. DOI: [10.1101/2021.01.11.426281](https://doi.org/10.1101/2021.01.11.426281)
[16] D. Krause, C. Bonin, et al. (2022). *The Rapid Population Collapse of a Key Marine Predator in the Northern Antarctic Peninsula Endangers Genetic Diversity and Resilience to Climate Change*. DOI: [10.3389/fmars.2021.796488](https://doi.org/10.3389/fmars.2021.796488)

## Annexes

### A. Detailed Reviewer Reports

#### Methodologist

- **Score**: 6.8/10 | **Verdict**: weak_accept | **Confidence**: 0.82
- **Strengths**: The staged, phased approach (in silico → minimal experimental → full protocol) with explicit GO/NO-GO/PIVOT criteria at each phase is an excellent example of adaptive experimental design. This allows for early falsification of core assumptions (e.g., Blomberg's K) before committing significant resources, which is a hallmark of rigorous scientific methodology.; The protocol demonstrates a strong awareness of the multiple dimensions of validity. It explicitly addresses phylogenetic signal (construct validity of the PD-buffer mechanism), uses a multi-model ensemble and multiple RCPs (external validity), and proposes spatial GLS to handle spatial autocorrelation (statistical conclusion validity). The inclusion of a sensitivity analysis for key assumptions (hotspot definition, dispersal rate) is particularly commendable.; The formulation of falsifiable predictions with explicit bounds and null hypotheses (e.g., PD retention difference ≥ 15% vs. H0: ≤ 0%) is exemplary. This moves beyond vague directional hypotheses and provides a clear, quantitative target for the statistical test, which is crucial for a decisive test of the hypothesis.
- **Weaknesses**: The protocol lacks a formal a priori power analysis. The number of spatial grid cells (500-2000) is mentioned, but the expected effect size, variance in PD retention, and spatial autocorrelation structure are not used to justify that this sample size is sufficient to detect a 15% difference with adequate power (e.g., 0.80). The staged design mitigates this, but a power analysis for the Phase 3 primary prediction is missing.; The 'Preliminary PD/SR Projection' in Phase 1 using a 'climate velocity' approach is a potential source of a significant confirmation bias. If this crude model is structurally biased towards showing a PD advantage (e.g., by not modeling competition or other biotic interactions that might affect PD hotspots differently), it could lead to a false GO decision for Phase 2. The protocol does not specify a validation step for this simplified model against observed range shifts.; The primary comparison between PD and SR hotspots is not a controlled experiment but an observational comparison of two different spatial configurations. The protocol does not address the critical issue of spatial confounding. For example, PD hotspots might be larger or more spatially clustered than SR hotspots, which would inherently affect their 'PD loss per unit area' and retention, independent of the ecological mechanism. A null model (e.g., comparing PD hotspots to randomly selected sets of cells with the same total area) is not proposed.
- **Questions**: What is the expected variance and spatial autocorrelation range for the PD retention metric, and how was the sample size (number of grid cells) determined to achieve 80% power to detect the 15% difference in Phase 3? Please provide the formal power analysis.; How will you control for spatial confounding in the comparison of PD and SR hotspots? For instance, if PD hotspots are more fragmented or larger than SR hotspots, how will you distinguish the effect of the metric (PD vs. SR) from the effect of the spatial configuration of the hotspots themselves?; The protocol relies on imputed CTmax data if >30% is missing. How will you assess the sensitivity of the Blomberg's K result to the imputation model, and what is the a priori threshold for the uncertainty around the K estimate that would trigger a pivot to Phase 2 earlier?
- **Recommendation**: This is a well-structured and thoughtful protocol that demonstrates a strong grasp of the scientific method. The phased, falsifiable approach is a major strength. However, the lack of a formal power analysis and the potential for spatial confounding in the core comparison are significant weaknesses that need to be addressed before this can be considered a fully rigorous test. I recommend proceeding with the Phase 1 in silico validation as a low-cost first step, but require the addition of a power analysis and a spatial null model to the Phase 3 design before full funding is allocated.

#### Domain Expert

- **Score**: 6.5/10 | **Verdict**: weak_accept | **Confidence**: 0.88
- **Strengths**: The hypothesis directly addresses a critical gap in marine conservation: the explicit integration of phylogenetic diversity (PD) with climate refugia theory. While spatial phylogenetics is well-established in terrestrial systems, its application to dynamic marine environments under climate change, particularly using the PD retention metric, is a timely and relevant extension.; The proposed causal chain is methodologically sound and operational. It leverages existing, high-quality resources (FishTreeOfLife, NOAA trawl surveys) and standard tools (SDMs, Blomberg's K), making the study feasible. The explicit parameterization of dispersal limitation as a variable (Step 4) and the inclusion of a phylogenetic signal threshold (Assumption 3) demonstrate a sophisticated understanding of the underlying ecological and evolutionary processes.; The hypothesis is framed as a quantitative, testable prediction (top 10% PD hotspots retain ≥15% more PD than SR hotspots). This clear, falsifiable statement is a significant strength, as it moves beyond vague calls for 'evolutionary-informed conservation' to a specific, measurable outcome that can be directly compared across different hotspot types.
- **Weaknesses**: The core assumption that PD hotspots will act as climate refugia is mechanistically underdeveloped. The hypothesis states PD 'buffers against functional loss' but does not specify the causal pathway. High PD does not inherently confer resilience; it is the presence of specific traits (e.g., thermal tolerance, dispersal ability) within that phylogeny that matters. The framework risks conflating a descriptive metric (PD) with a predictive one (resilience) without a clear trait-based mechanism linking them.; The definition of 'refugia' is ambiguous. The hypothesis measures 'PD retention' in the same geographic cells (top 10% hotspots). However, true climate refugia are areas where species persist *in situ* due to buffered microclimates, not just areas where projected species loss is lowest. The proposed method identifies 'areas of low projected PD loss', which could simply be areas with low climate velocity or high species turnover, not true refugia. This is a critical semantic and conceptual distinction.; The evidence base is heavily skewed towards terrestrial botanical studies. While these are excellent analogues for the *methodology* of spatial phylogenetics, they do not provide direct evidence for the *mechanism* proposed (PD buffering thermal stress). The cited marine papers are either indirect (data availability) or preliminary (eDNA). The hypothesis lacks direct support from marine systems demonstrating a link between PD and climate resilience, making the foundational premise weaker than it appears.
- **Questions**: What is the specific, testable mechanism by which high phylogenetic diversity per se leads to higher PD retention under warming? Is it solely a probabilistic effect (more lineages = higher chance of including a pre-adapted one), or is there an emergent property of diverse communities (e.g., facilitation) that enhances resilience? The hypothesis needs to distinguish between these to be truly mechanistic.; How will you differentiate between a 'climate refugium' (a site with stable environmental conditions) and a 'macroclimatic buffer zone' (a site where species loss is projected to be low due to the regional climate scenario)? Your metric of PD retention will capture both, but they have fundamentally different implications for conservation planning.; Given that trawl survey data are presence/absence and often have variable detection probabilities, how will you ensure that the initial PD calculations are not biased by sampling effort? The assumption of detection probability >0.3 is a high bar for many rare or deep-sea species, and failure to meet it could severely compromise the baseline PD maps.
- **Recommendation**: The hypothesis is promising and addresses a relevant question, but it needs to be sharpened. I recommend refining the definition of 'refugia' to be more mechanistically explicit, perhaps by incorporating environmental stability layers (e.g., climate velocity) to identify true refugia and then testing if PD hotspots overlap with them. Furthermore, the authors should explicitly state and, if possible, test the assumption that PD is a proxy for functional diversity (e.g., by correlating PD with trait space volume). This would strengthen the causal chain from PD to resilience. The current framing is more about 'PD persistence' than 'refugia', and clarifying this would improve its theoretical contribution.

#### Contrarian

- **Score**: 3.5/10 | **Verdict**: weak_reject | **Confidence**: 0.85
- **Strengths**: The explicit articulation of a falsifiable prediction (≥15% PD retention advantage) with a defined null hypothesis is methodologically sound and rare in this literature.; The acknowledgment of dispersal limitation as a variable parameter rather than a fixed constant demonstrates awareness of a key source of uncertainty in range-shift projections.
- **Weaknesses**: FAIL REASON #1: The causal chain conflates phylogenetic diversity (PD) with functional/trait diversity. The claim that PD 'buffers against functional loss' requires that CTmax (or other relevant traits) exhibit strong phylogenetic signal. The authors set K ≥ 0.5 as a threshold, but even if met, this only guarantees that closely related species share similar CTmax values—it does not guarantee that PD hotspots contain the *range* of CTmax values needed to buffer community-level loss. High PD can be driven by deep, ancient lineages with similar thermal tolerances (e.g., all cold-water specialists), yielding high PD but zero functional buffering. The framework must explicitly test whether PD hotspots maximize trait space coverage, not just evolutionary branch length.; FAIL REASON #2: The trawl survey data (NOAA) are spatially biased toward the Northwest Atlantic and Gulf of Alaska. The minimum of 500 cells with ≥5 species is trivially met in these regions, but the 'hotspots' identified will be artifacts of sampling effort, not biological reality. More critically, the SDMs will be calibrated on this biased occurrence data, and projected distributions under RCP 8.5 will extrapolate to unsampled environmental space. The PD retention metric will then be computed on projections that are essentially unvalidated outside the sampled domain. The paper does not address how spatial sampling bias propagates through SDM projections to the PD retention estimate, and the 15% threshold could be entirely driven by this bias.; FAIL REASON #3: The 'Protection Paradox' paper (cited in the authors' own counter-evidence) demonstrates that protected areas often fail to buffer climate impacts because species shift out of them. The proposed framework assumes PD hotspots are static refugia, but the authors admit this is unknown. The prediction of ≥15% PD retention in PD hotspots vs. SR hotspots is only meaningful if hotspots are defined *a priori* on current distributions. If PD hotspots shift spatially under climate change, then the 'refugia' concept collapses. The analysis must include a dynamic hotspot definition (e.g., tracking where PD is projected to be high in 2100) and compare that to static current hotspots. Without this, the prediction is trivially true if PD hotspots happen to be in areas with low projected climate velocity—but this is a geographic coincidence, not a mechanism. The paper does not control for this.
- **Questions**: How do you distinguish between a PD hotspot that is a refugium because of high evolutionary distinctiveness versus one that is simply located in a region with low projected climate velocity? If you control for climate velocity (e.g., by matching hotspots on this variable), does the 15% PD retention advantage persist?; Given that CTmax data are available for only a fraction of the 200-400 species, how will you impute missing values? If imputation is done using phylogenetic ancestral state reconstruction, you will artificially inflate Blomberg's K. What is your plan to avoid this circularity?
- **Recommendation**: The authors must add a critical control: compute PD retention for hotspots selected on *trait diversity* (e.g., CTmax range) and compare to PD hotspots. If PD hotspots do not outperform trait-diversity hotspots, the mechanism is not evolutionary distinctiveness but trait coverage. Additionally, they must perform a spatial bias correction (e.g., rarefaction or inclusion of a sampling-effort covariate in SDMs) and report PD retention under a dynamic hotspot definition. Without these, the 15% prediction is uninterpretable.

#### Industrialist

- **Score**: 6.5/10 | **Verdict**: weak_accept | **Confidence**: 0.65
- **Strengths**: Cible un besoin réglementaire et ESG émergent : les grands groupes de l'agroalimentaire et de la pêche (ex. Thai Union, Nomad Foods, Maruha Nichiro) et les assureurs (ex. AXA Climate) doivent justifier la résilience de leurs chaînes d'approvisionnement face au changement climatique. Un indice de 'refuges évolutifs' monétisable comme service de scoring de risque pourrait être intégré à des plateformes existantes (ex. Global Fishing Watch, Ocean Data Alliance) pour un abonnement B2B annuel de 50k€-200k€.; Avantage différenciant fort : contrairement aux cartes de richesse spécifique (SR) déjà saturées par les ONG (ex. WWF, IUCN), ce produit combine phylogénie et refuges climatiques, offrant une couche de données unique pour la sélection de zones de 'no-take' ou de corridors. Cela permet de breveter un algorithme de scoring (méthode de calcul du ratio PD_loss/area sous scénarios climatiques) et de créer une barrière à l'entrée logicielle, même si les données sources sont publiques (OBIS, AquaMaps).
- **Weaknesses**: Marché adressable principal limité aux gouvernements et grandes ONG de conservation (ex. Blue Nature Alliance, Pew Charitable Trusts) avec des cycles de vente longs (18-36 mois) et des budgets discrétionnaires. Le segment privé (pêche, assurance) est potentiel mais n'a pas encore démontré sa volonté de payer pour ce type de métrique 'évolutionnaire' plutôt que pour des données de biomasse ou de rendement halieutique, ce qui rend le ROI immédiat incertain.; Barrière scientifique majeure : la dépendance à la valeur de Blomberg's K (≥0.5) pour CTmax est un point de fragilité. Si la Phase 1 échoue (K < 0.5), l'hypothèse s'effondre. De plus, la résolution spatiale des données de PD pour les poissons marins est souvent grossière (maillage 1°), ce qui limite la précision opérationnelle pour des zones côtières de 10-50 km², là où les gestionnaires ont besoin d'agir, réduisant la valeur perçue du produit.
- **Questions**: Quel est le coût d'acquisition client (CAC) réaliste pour vendre un 'indice de refuges phylogénétiques' à des ministères de l'environnement (ex. Indonésie, Mexique) vs. le prix de vente potentiel (licence annuelle de 100k€) ? Un partenariat avec une ONG comme Conservation International peut-il réduire ce CAC à <30% du prix de vente ?; Pouvons-nous obtenir un brevet (ou un logiciel enregistré) sur la méthode de 'transfert de PD' qui soit suffisamment large pour bloquer les concurrents (ex. NatureServe, Map of Life) qui pourraient intégrer cette métrique dans leurs outils existants dans les 2 ans ?; Quel est le plan de contingence si la Phase 2 (validation expérimentale) montre un K < 0.5 ? Existe-t-il une version 'allégée' du produit (basée uniquement sur la rareté des traits fonctionnels) qui pourrait être commercialisée pour ne pas perdre l'investissement initial ?
- **Recommendation**: Ne pas lancer un produit complet immédiatement. Financer uniquement la Phase 1 (in silico, <2k€) avec un partenaire académique (ex. IRD, CSIRO) pour valider le K. En parallèle, signer une lettre d'intention avec une ONG de conservation marine (ex. Blue Nature Alliance) pour co-développer un 'pilote' sur une écorégion (ex. Triangle de Corail) et tester la volonté de payer. Si la Phase 1 est positive, lever des fonds pour la Phase 3 et cibler la vente de données à des assureurs paramétriques (ex. AXA Climate) pour l'assurance de récifs, en utilisant la PD comme proxy de résilience, avec un prix de 150k€ par contrat.

#### Funding Strategist

- **Score**: 7.2/10 | **Verdict**: accept | **Confidence**: 0.82
- **Strengths**: Originality and conceptual novelty: The hypothesis directly challenges the conventional use of species richness (SR) as a proxy for conservation priority by proposing a quantitative, transferable test of phylogenetic diversity (PD) as a climate refugia indicator. This addresses a recognized gap in macroecology and conservation planning, positioning the project at the forefront of evidence-based biodiversity management.; Methodological rigor and falsifiability: The protocol is exceptionally well-structured with clear, pre-defined GO/NO-GO criteria (e.g., Blomberg's K ≥ 0.5, PD retention difference ≥ 15%). This staged approach (in silico → minimal experimental → full experimental) demonstrates a high level of scientific maturity and risk mitigation, which is highly attractive to evaluators looking for feasible and well-planned projects.; Strategic use of existing data and phased budget: The initial Phase 1 relies on existing datasets and simplified projections, allowing for a rapid, low-cost falsification of the core assumption. This efficient use of resources de-risks the larger investment in Phase 3, making it a perfect fit for seed-funding mechanisms and early-stage grant schemes.
- **Weaknesses**: Missing consortium and data access plan: The proposal lacks a named consortium or a clear plan for accessing the necessary data (e.g., OBIS, AquaMaps, FishBase) and the experimental facilities for Phase 2 and 3 (e.g., marine biology labs with thermal ramping equipment). This is a critical gap for a project that requires both computational and experimental expertise.; Limited mechanistic depth in the narrative: While the hypothesis is clear, the narrative does not deeply explore the physiological mechanisms linking PD to thermal tolerance (CTmax). A stronger link to evolutionary theory (e.g., phylogenetic niche conservatism) and potential confounding factors (e.g., dispersal capacity, local adaptation) would strengthen the scientific case for reviewers.; Unclear pathway to conservation impact: The project stops at validation of the hypothesis. It lacks a clear articulation of how the results will be translated into actionable conservation tools (e.g., dynamic ocean management, MPA network design) beyond a 'publication-ready' validation. This could be seen as lacking a broader impact strategy.
- **Questions**: What specific datasets and computational resources will be used for Phase 1, and what is the preliminary evidence that the spatial resolution of existing species distribution models is sufficient to capture the microhabitat dynamics relevant to thermal refugia?; Given the phased approach, which consortium partners (e.g., a computational ecology lab and an experimental physiology lab) have you already engaged in discussions to ensure seamless transition between the in silico and experimental phases?; How will you handle the inherent uncertainty in climate projections and species distribution models, and what is your plan for incorporating multiple emission scenarios and model ensembles in the final Phase 3 analysis to ensure the 'transferability' of your test?
- **Recommendation**: Position this as a high-risk, high-reward 'proof-of-concept' project. The ideal strategy is to submit the Phase 1 and 2 components to the ANR JCJC or a similar national early-career grant to secure initial funding and generate the preliminary data. Simultaneously, use the results from Phase 1 to build a stronger consortium and submit the full 3-phase protocol to an ERC Starting Grant, framing it as a foundational step in predictive evolutionary conservation. The phased design is a perfect narrative for a 'StG' which values ambitious, high-risk research with a clear potential for breakthrough.

### B. Semantic Scholar Search Queries

- [novelty] `phylogenetic diversity marine fish conservation` — Direct test of the core hypothesis: applying PD metrics to marine fish for conservation, checking if this exact transfer already exists.
- [novelty] `phylogenetic endemism hotspot fisheries management` — Searches for the specific application of phylogenetic hotspot algorithms (PE, PD) to fisheries, a key component of the proposed transfer.
- [novelty] `evolutionary distinctiveness climate refugia fish` — Tests whether evolutionary history data (ED) has already been linked to climate refugia identification in marine systems, the stated non-trivial insight.
- [novelty] `phylogenetic diversity climate resilience marine` — Checks for existing evidence linking PD to climate change resilience in marine environments, validating the proposed mechanism.
- [evidence] `phylogenetic diversity hotspot mapping algorithm` — Finds the foundational methodological papers on PD hotspot computation and spatial mapping from botany, providing the algorithmic basis for transfer.
- [evidence] `fish phylogeny evolutionary tree species` — Identifies existing molecular phylogenies for fish species, a prerequisite for computing PD in marine systems, as proposed in the hypothesis.
- [evidence] `trait-based diversity functional loss warming` — Searches for evidence that high diversity (including phylogenetic) buffers against functional loss under warming, supporting the resilience mechanism.
- [evidence] `NOAA trawl survey species occurrence data` — Locates the specific data source (NOAA trawl surveys) mentioned in the hypothesis, confirming data availability for PD mapping in marine fish.
- [cross_domain] `spatial phylogenetics plant conservation hotspots` — Establishes the botanical precedent for combining phylogenies with spatial data to identify conservation hotspots, the source domain for the transfer.
- [cross_domain] `phylogenetic diversity terrestrial aquatic comparison` — Searches for prior comparative studies between terrestrial and aquatic systems using PD, providing interdisciplinary precedents for the transfer.
- [cross_domain] `evolutionary history ecosystem management interdisciplinary` — Broadly looks for interdisciplinary applications of evolutionary history in ecosystem management, beyond botany, to support the cross-domain feasibility.

---

*Generated by SPORE (Systeme de Production d'Opportunites de Recherche par Exploration) on 2026-08-21*