Date of Award

Spring 5-10-2019

Level of Access

Open-Access Thesis

Degree Name

Doctor of Philosophy (PhD)


Wildlife Ecology and Wildlife Conservation


Joseph Zydlewski

Second Committee Member

Erik Blomberg

Third Committee Member

Michael Kinnison

Additional Committee Members

Joan Trial
Gayle Zydlewski


The construction of industrial dams across major rivers in New England began in the early 1800s with textile mills in Massachusetts. Because of its legacy of mill dams and log driving dams, New England has the highest density of dams anywhere in North America, averaging one dam for every 44 km2 of drainage area. By the early 1900s, these dams drastically limited migrations by diadromous fishes, resulting in declines in populations of migratory fishes, including Atlantic salmon Salmo salar and American shad Alosa sapidissima. Since that time, different fishway designs and river management plans have been tried around the region in an attempt to balance hydropower generation, water storage, flood control, and fish passage.

The Penobscot River, in Maine, USA, is the second largest river in New England, with a drainage area of 22,000 km2, and presents an ideal opportunity for understanding fish passage and migratory movements in a heavily-modified river system. Historically, the river contained runs of eleven species of diadromous fishes including alewife Alosa pseudoharengus, American shad, blueback herring A. aestivalis, Atlantic salmon, sea lamprey Petromyzon marinus, American eel Anguilla rostrata, rainbow smelt Osmerus mordax, Atlantic tomcod Microgadus tomcod, striped bass Morone saxatilis, brook trout Salvelinus fontinalis, Atlantic sturgeon Acipenser oxyrinchus, and Shortnose Sturgeon A. brevirostrum, many of which supported subsistence and commercial fisheries prior to dam construction. The Penobscot River Restoration project (initiated in 2004), decommissioned three hydroelectric dams, upgraded fish passage at four dams, and increased generation capacity at three dams, theoretically opening much of the watershed to migratory fishes. We used a combination of telemetry methods, historic data, and modeling to assess the potential for fishways to select for certain traits with enough power to cause evolutionary change in migratory fish populations and evaluate the ability of migratory fish to use newly available habitat in the Penobscot River watershed.

The primary fishway design used in New England for many years was the Denil fishway, which uses angled baffles to reduce water velocity while maintaining high volume and thus, high attraction flow. In order to reach their spawning grounds in the headwaters of the Penobscot River watershed, in-migrating Atlantic salmon have had to navigate between five and seven (mostly Denil) fishways at different dams. Six years of PIT telemetry data were used to evaluate passage success at fishways on the second, third, and fourth dams in the system (Great Works Dam, Milford Dam, and either Howland Dam or West Enfield Dam, depending on path choice) as a function of water temperature, flow, migratory timing, and fish length. At the lower two fishways (Great Works and Milford Dams), fish length was a significant predictor of passage success, with a 91‐cm salmon 21%–27% and 12%–16% less likely to pass than a 45‐cm salmon, at Great Works and Milford Dams, respectively. Additionally, we analyzed thirty-four years of escapement data and found that the slow‐maturing and iteroparous individuals that represented the largest salmon size classes were essentially lost from the population during that time, and that Penobscot River fish have shorter fork lengths at maturity (45 – 91 cm) than Atlantic salmon in undammed systems (45 – 110 cm). Both of these results indicate that selective pressure towards smaller sizes at maturity (exerted by fishways) may be driving evolutionary responses in Atlantic salmon in the Penobscot River.

Size at maturity and age at maturity are heritable in salmonids and have ramifications for population stability over time, as body size is an important predictor of an individual salmon’s egg quality and quantity. To test whether the selection against large-bodied fish exerted by fishways was substantial enough to elicit an evolutionary response from a population of Atlantic salmon over many years, we developed a computer simulation of an Atlantic salmon population, which was primarily informed by research done in the Penobscot River as well as other rivers in Maine and rivers in Atlantic Canada. Using an individual based model, we allowed a population of Atlantic salmon to evolve over a 100-year period of exposure to between zero and five dams, using different narrow-sense heritability values for age at maturity and size at maturity. This necessitated 150 different combinations of heritability values and dams, each of which was simulated 1000 times. Populations without dams present never went extinct, but 7.2% of populations exposed to one dam, 63.2% of populations exposed to two dams, and > 85% of populations exposed to three or more dams went extinct more than 30 years before the end of the simulation. Coefficient of Variation of size at maturity decreased over time when dams were present, indicating that dams may be a source of stabilizing or directional selection.

Little is known about the survival of hatchery-spawned kelts that are released into the natural environment. As the Penobscot River population of Atlantic salmon is primarily captured and spawned out in captivity, managers are faced with a decision of whether to release the post-spawn adults (kelts) in the estuary (downstream of the dams) or upstream of the dams in freshwater habitat. We released radio-tagged kelts at two sites (upstream of Milford Dam and in the Penobscot River estuary) in the late fall to evaluate overwinter survival and successful outmigration from each release location. Unexpectedly, 71% of tagged kelts made upstream movements after release, but were blocked by dams with fishways that had been shuttered for the season. Overwinter survival was low (23.6%) and neither survival nor behavior differed between release locations. Low overwinter survival may have contributed to the precipitous decline in iteroparous spawners in this population over the past four decades.

Another potentially iteroparous species in the Penobscot River assemblage is the American shad. Thought to have been extirpated from the system until fairly recently, a spawning population was recently documented near the head of tide. However, with the removal of the first two dams in the system and passage improvements at upstream dams, American shad should have access to up to 93% of its historic spawning habitat. To assess habitat use by these fish, we gastrically tagged 265 American shad at the head of tide between 2015-2017 and monitored their movements from the time of tagging until the end of summer (mid-September) each year. Additionally, we collected scales and used them to estimate the age and spawning history of each tagged fish. Few tagged fish moved upstream beyond the previously documented spawning grounds. Those that did were all virgin spawners, indicating that repeat spawners may be less likely to overshoot known spawning areas. Additionally, none of the tagged fish that approached the dam successfully used the new fishway there. However, several thousand untagged fish used the fishway each year during our study.