Dams are a major contributor to the historic decline and current low abundance of diadromous, or sea-run fish (species that migrate between freshwater and the ocean). Dams affect diadromous species through various mechanisms, such as preventing or impeding fish passage, degrading the productive capacity of habitats upstream, and killing and injuring fish on both upstream and downstream migrations. These effects have lead to the implementation of the Penobscot River Restoration Project, which includes removing Great Works (2012) and Veazie (2013) dams and decommissioning and building a bypass around Howland Dam (expected completion in autumn of 2015). In addition to large-scale habitat improvements like this one, the recovery program for Gulf of Maine Atlantic salmon also includes increasing abundance through hatchery supplementation.
Julie Nieland, Timothy Sheehan, and Rory Saunders, researchers with NOAA, decided to assess the interactions of Atlantic salmon with 15 Federal Energy Regulatory Commission (FERC)-licensed hydroelectric dams in the system. Their objective was to estimate the relative change in Atlantic salmon abundance and distribution throughout the Penobscot River system due to varying levels of fish passage at these 15 hydroelectric dams. They also assessed the change in the abundance of this population under varying levels of indirect latent mortality, increased marine and freshwater survival rates, and no hatchery supplementation.
Methods
The researchers modelled the relative effects of 15 FERC-licensed hydroelectric dams in the watershed on Atlantic salmon. Using these dams, they divided the Penobscot River into 15 sections, called production units. This scheme helped isolate the locations where salmon and dams interacted within the model.
To evaluate the effects of dams on various life stages, the team divided the Atlantic salmon life cycle into five discrete stages:
- Female spawners
- Produced eggs
- Eggs survived to the juvenile smolt stage
- Surviving smolts transitioned to the marine phase as post-smolts
- Fish returned to the coast as mature adults after two winters at sea (referred to as two sea-winter or 2SW fish). The 2SW adults migrated upstream to spawning grounds, completing the cycle.
The researchers used a simple age distribution for the smolt and adult life stages in their model based on known characteristics of this population.
Building the Model
The model for the Atlantic salmon life cycle has several inputs, including:
- Spawning to Pre-migration Smolt Abundance: Using data from previous studies and databases, the researchers seeded the model with 2SW females, wild-hatched smolts, and stocked hatchery smolts.
- Downstream Migration Dynamics: For this input, the researchers incorporated mortality rates for smolts, both natural in-river mortality rates and dam-related mortality, using data extrapolated from previous studies and models to predict survival rates at each dam.
- Post-smolt Abundance: For this input, the scientists made sure to incorporate what is called indirect latent mortality, which is an additional dam-related mortality that occurs in the early marine phase of the salmon’s life history and is due to passing over one or multiple dams. They accounted for this factor by applying an indirect latent mortality rate of 5% per dam to fish that survived to the marine environment based on the number of dams that the fish passed.
- Marine Survival: The researchers used Penobscot River smolt stocking data, smolt survival data from telemetry studies, and estimates of 2SW female returns to build a submodel to estimate a more accurate 2SW female marine survival distribution.
- Upstream Migration Dynamics: The researchers assumed a 100% return rate for all salmon migrating back to the Penobscot River to spawn. However, they did account for the fish straying from their original hatching or stocking site within the river system by using within-river straying rates from previous field studies and expert knowledge, as well as dam-passage efficiency estimates from previous field studies and modelling efforts.
"We wanted to build a stochastic model rather than a deterministic model, and we were able to easily incorporate the variation in our model inputs by making a random draw from input distributions for each iteration. @RISK also enabled us to quickly and efficiently run thousands of iterations and get both summary statistics and individual iteration data from our model runs. Finally, as an add-on to Microsoft Excel, @RISK allowed us to have the familiarity of the Excel user-interface, which was an important feature for demonstrating our model to colleagues."
Julie Nieland
Researcher, National Oceanic and Atmospheric Administration
Running the Different Model Scenarios
The research team next ran six model scenarios to evaluate the effects of dams and dam location. The different scenarios were:
- All dams on;
- All dams off;
- Mainstem river dams on, with tributary dams off;
- Mainstem river dams off, tributary dams on;
- Penobscot River Restoration Project scenario (three mainstem dams off, with all others on);
- Species Protection Plan scenario (Penobscot River Restoration Project scenario plus improved salmon passage at 4 other dams).
They ran additional model scenarios to examine the recovery potential of Atlantic salmon under varying indirect latent mortality rates and increased marine and freshwater survival rates and to evaluate the viability of the population without hatchery supplementation.
Results
After running the models, the researchers found that adult abundance and distribution to upper areas of the Penobscot River watershed increased when fewer dams, particularly mainstem dams, were in the watershed or passage efficiency at these dams was increased. The proportion and the number of wild-origin adults in upper areas of the watershed also increased with fewer mainstem dams and increased passage efficiency.
“Our results show that [conservation measures] could result in an increase in adult abundance similar to what would occur if all mainstem dams were removed,” the authors wrote in their journal paper. “Dam removal and bypass had a large effect on salmon abundance and distribution to upper reaches of the watershed, but improved passage efficiency also contributed to increases in abundance and distribution and should be considered a useful management tool in achieving restoration goals.”
Benefits of @RISK
“@RISK was beneficial for our project in many ways,” explains researcher Julie Nieland. “For example, we wanted to build a stochastic model rather than a deterministic model, and we were able to easily incorporate the variation in our model inputs by making a random draw from input distributions for each iteration. @RISK also enabled us to quickly and efficiently run thousands of iterations and get both summary statistics and individual iteration data from our model runs. Finally, as an add-on to Microsoft Excel, @RISK allowed us to have the familiarity of the Excel user-interface, which was an important feature for demonstrating our model to colleagues.”