The River Continuum Concept

The Serial Discontinuity Concept

The impacts of dams on riverine ecosystems

How these impacts spread throughout the river-sea continuum

A meta-ecosystem perspective

How multiple stressors interact

• Stressors can interact with one another such that they can multiply or cancel their individual impacts (Jackson et al. 2021)
• The interaction can be synergistic or antagonistic (Orr et al. 2024)
• Temperature \(\times\) nutrients
  • Expected in theory to be antagonistic (Binzer et al. 2012)
  • But not supported by empirical data (Bonnaffé et al. 2024)
• Dam \(\times\) nutrients
  • Decrease in connectivity which favor accumulation and so eutrophication (Maavara et al. 2020)
• Dam \(\times\) dam
  • Dam have cumulative impact on migratory species (Dean et al. 2023)

In brief

• Dams have been reported to have an impact on riverine ecosystem
• But studies are mostly local and focused on species richness
• Furthermore the interaction of dams with other stressors has not been clearly assessed
• Lastly, dam impact could spread to the sea but this propagation has not been investigated
  • Separation between freshwater and marine ecology
  • Different survey and dataset
  • Different interests, terminology, etc.
• We aim to link these fields, to answer the following question…

What we have

What we plan to add

For each local community:

• Distance to the sea/source
• Dams location
• Salinity
• Temperature
• Nutrient concentration
• … (any suggestion?)

Measuring the impact of dams

• Distance to the closest dam
• Height/reservoir volume of the closest dam
• Accumulated heights of upstream/downstream dams (Dean et al. 2023)
• At what scale dam impacts operate?
  • Use the literature, but difficult because of varied impacts on species
  • Make the statistical model infer the scale
  • \(\text{Dam}_\text{down}(L) = \sum_{i \in \text{dam}} h_i e^{-\frac{d_i}{L}}\)
  • \(L\) define the scale at which dam operates
  • Try different values and compare WAIC (that is, model performance)

How we reconstruct food webs

The base

How we reconstruct food webs

Diet and piscivory

• Extract fish diet with fishbase
• Diet changes with ontogeny (juvenile vs adults)
• Piscivorous interaction infered with predation window (e.g. 5-45% body size)
• Build a global metaweb with all possible interactions
• Infer local food web by subsampling
• Quantify structure with connectance and trophic length (a metric weighted by biomass?)

What we want

How dams reshape food web along the river-sea continuum by disrupting spatial fluxes?

• Quantify the impact of dams on food web structure
• In space and time
• Along the river-sea continuum
• Quantify interaction of dam with other stressors
• Bonus (liste au Père Noël):
  • Use spatial correlation to assess flux between ecosystem (meta-ecosystem view)
  • Predict the impact of dam removal (may be possible with bayesian model)
  • Sélune project: tracking the impact of dam removal in Britany

The statistical model

Causal graph (DAG)

The statistical model

Equation

\[ \begin{align} \text{TL}_i &= \text{Normal}(\mu_i, \sigma_i) \\ \text{logit} (\mu_i) &= \beta_0 + \beta_1 D_i + \beta_2 \text{Dam} + ... ~ \text{[fixed effects]} \\ &+ \text{interactions} \\ &+ \text{random effects} \end{align} \]

• Account for survey random effects
• Account for spatial and temporal auto-correlation
• Use a spatiotemporal model (R-INLA)
• Possible limitations
  • Can we distengle survey bias from environment effect?
  • Until where the concept of continuum holds?
  • Do we have high variability in dams density between hydrographic bassin?

Roadmap

• Build food webs in the coast/estuary
• Update food webs in rivers
• Collect and infer if necessary environmental data
• First perform “coarse-grained” statistical analysis
  • At the bassin scale: Number of dam vs. Trends [TL vs Distance to the source]
  • PCA to assess the main environmental drivers of food web structure
• Next, refine statistical analysis (R-INLA)…