The Penobscot River drains the largest watershed in Maine and once provided spawning and rearing habitats to 11 species of diadromous fishes. The construction of dams blocked migrations of these fishes and likely changed the structure and function of fish assemblages throughout the river. The proposed removal of two main‐stem dams, improved upstream fish passage at a third dam, and construction of a fish bypass on a dam obstructing a major tributary is anticipated to increase passage of and improve habitat connectivity for both diadromous and resident fishes. We captured 61 837 fish of 35 species in the Penobscot River and major tributaries, through 114 km of boat electrofishing. Patterns of fish assemblage structure did not change considerably during our sampling; relatively few species contributed to seasonal and annual variability within the main‐stem river, including smallmouth bass Micropterus dolomieu, white sucker Catostomus commersonii, pumpkinseed Lepomis gibbosus, and golden shiner Notemigonus crysoleucas. However, distinct fish assemblages were present among river sections bounded by dams. Many diadromous species were restricted to tidal waters downriver of the Veazie Dam; Fundulus species were also abundant within the tidal river section. Smallmouth bass and pumpkinseed were most prevalent within the Veazie Dam impoundment and the free‐flowing river section immediately upriver, suggesting the importance of both types of habitat that supports multiple life stages of these species. Further upriver, brown bullhead Ameiurus nebulosus, yellow perch Perca flavescens, chain pickerel Esox niger, and cyprinid species were more prevalent than within any other river section. Our findings describe baseline spatial patterns of fish assemblages in the Penobscot River in relation to dams with which to compare assessments after dam removal occurs. Copyright © 2014 John Wiley & Sons, Ltd.

Dams and their impoundments disrupt river habitat connectivity to the detriment of migratory fishes. Removal of dams improves riverine connectivity and lotic habitat, which benefits not only these fishes but also resident fluvial specialist species. Restoration efforts on the Penobscot River, Maine, are among the largest recently completed in the United States and include the removal of the two lowermost dams and improvements to fish passage at several remaining barriers. We assessed fish assemblages in the main‐stem river and several major tributaries before (2010–2012) and after (2014–2016) dam removal using boat electrofishing surveys and a stratified random sampling design. In total, we sampled 303 km of shoreline and captured 107,335 individual fish representing 39 species. Similarity indices and rarefaction curves indicated that significant changes in fish assemblage composition occurred in reaches that underwent both habitat and connectivity changes (i.e., directly above removed dams). The newly connected reaches became more similar in fish assemblage composition, as demonstrated by an average increase of 31% in similarity scores. The changes in similarity score in these reaches were driven by increasing access for anadromous fishes and decreasing abundances of slow‐water specialist species. For example, we observed a marked reduction in lacustrine species in former impoundments. These assemblage shifts were further illustrated by nonmetric multidimensional scaling in which sites directly above former dams exhibited the largest ordinal shifts immediately following dam removal. We also found all anadromous species in greatest abundance below the lowermost dam during each respective sampling period, though we did find some anadromous species above the lowermost dam during postremoval sampling. Our results demonstrate the potential for large dam removal projects to restore both fluvial and anadromous fish assemblages.

We compared the efficiency of stratified random and fixed-station sampling designs to characterize fish assemblages in anticipation of dam removal on the Penobscot River, the largest river in Maine. We used boat electrofishing methods in both sampling designs. Multiple 500-m transects were selected randomly and electrofished in each of nine strata within the stratified random sampling design. Within the fixed-station design, up to 11 transects (1,000 m) were electrofished, all of which had been sampled previously. In total, 88 km of shoreline were electrofished during summer and fall in 2010 and 2011, and 45,874 individuals of 34 fish species were captured. Species-accumulation and dissimilarity curve analyses indicated that all sampling effort, other than fall 2011 under the fixed-station design, provided repeatable estimates of total species richness and proportional abundances. Overall, our sampling designs were similar in precision and efficiency for sampling fish assemblages. The fixed-station design was negatively biased for estimating the abundance of species such as Common Shiner Luxilus cornutus and Fallfish Semotilus corporalis and was positively biased for estimating biomass for species such as White Sucker Catostomus commersonii and Atlantic Salmon Salmo salar. However, we found no significant differences between the designs for proportional catch and biomass per unit effort, except in fall 2011. The difference observed in fall 2011 was due to limitations on the number and location of fixed sites that could be sampled, rather than an inherent bias within the design. Given the results from sampling in the Penobscot River, application of the stratified random design is preferable to the fixed-station design due to less potential for bias caused by varying sampling effort, such as what occurred in the fall 2011 fixed-station sample or due to purposeful site selection.Characterizing fish assemblage structure is an important component of fisheries research and management. Some assessments are conducted within relatively large ecosystems over multiple seasons or years, and sampling effort required for researchers to provide repeatable estimates is often unknown. Assessments are particularly difficult in large rivers where longitudinal variation and impacts of dams on fish assemblage structure can be profound. Low levels of sampling effort yield imprecise data, which could yield ambiguous results and poorly informed recommendations for management. Researchers must balance precision with many other considerations, including potential bias within the sampling design and budget limitations

The Penobscot River Restoration Project in Maine was a large river rehabilitation project that culminated in the removal of the two lowermost dams and improvements to fish passage on several remaining dams. Fish assemblages were surveyed for 3 years prior to rehabilitation, 3 years after rehabilitation, and 8 years after rehabilitation. Approximately 475 km of shoreline were sampled via boat electrofishing, yielding 133,394 individual fish of 41 species. The greatest shifts in assemblage structure occurred immediately after dam removal in formerly impounded sections, with an increased prevalence of riverine and migratory species. Long‐term sampling documented changes within tributaries and tidally influenced river segments, where large schools of adult and young‐of‐the‐year alosines increased in abundance. Upstream of the lowermost dam, the river remains dominated by lacustrine species, while adult anadromous fishes continue to be most abundant immediately downstream of the lowermost dam. Our results provide increased evidence that dam removals result in altered fish assemblages, which are now dominated by riverine and anadromous species in previously impounded habitats. Alosines in the Penobscot River have exhibited the greatest long‐term response to river restoration efforts.