Perhaps the most obvious effect of dams follows directly from their function of storing and controlling water. Dams divide free-flowing, continuous habitats into distinct, discontinuous fragments and create impounded waters. Riverine ecosystems rely upon basin-scale storage and transport of resources (Vannote et al., 1980) and the proliferation of damming has disrupted these processes by altering flow regimes and restricting the movement of aquatic fauna (Ward and Stanford, 1987). Habitat fragmentation may lead to local extirpation (Kiffney et al., 2009; Carvajal-Quintero et al., 2017), population declines (Limburg and Waldman, 2009), or extinction (Ricciardi and Rasmussen, 1999; Carvajal-Quintero et al., 2017).

Many of the Penobscot River’s migratory species are among those most vulnerable to the effects of damming globally, including alosines, lampreys, and eels (Liermann et al., 2012). Dams are migration barriers that exclude these species from important upstream habitats and caused some populations (e.g., Atlantic salmon, and American shad) to decline into single-digit abundances (Opperman et al., 2011; DMR, 2022). The depletion of Maine’s diadromous community mirrors the trend for these fishes across North America and globally (Waldman and Quinn, 2022). Three Penobscot River species have garnered federal protection (Atlantic sturgeon, shortnose sturgeon, and Atlantic salmon) under the Endangered Species Act (USFWS, 1967; NMFS, 2010; USFWS and NMFS, 2018) with rainbow smelt having been listed as a federal Species of Concern (Enterline et al., 2012). Dams are cited as a primary threat to these protected species within the Penobscot River (Atlantic and shortnose sturgeons; Fernandes et al., 2010; rainbow smelt, Enterline et al., 2012; Atlantic salmon, Rubenstein et al., 2023).

The removals of main-stem dams as part of the PRRP has revealed the direct relation between dams and restricted access to habitat. The PRRP and associated efforts have dramatically increased the abundance of diadromous fishes within the Penobscot River (Scherelis et al., 2020; DMR, 2022; Whittum et al., 2023). Before dam removal, the American shad population was of unknown size and limited to habitat downstream of Veazie Dam. Only 16 adults had passed through the fishway from 1978 to 2012 (Grote et al., 2014a; Grote et al., 2014b). Annual counts at Milford Dam have now surpassed 10,000 (in 2022) and the population has supported a growing recreational fishery (DMR, 2022). From a combination of passage improvements and adult stocking into spawning lakes, the 2023 river herring run has approached 6 million individuals, increased from tens of thousands of fish before dam removals (DMR, 2022; Figure 3). Shortnose sturgeon have been tracked moving upstream of the former Veazie Dam (Johnston et al., 2019) and several have been encountered at the Milford Dam fish lift. Atlantic salmon permeability through the region with the two dam remnants was greatly improved from critically poor passage (Holbrook et al., 2009) to that of an open river (Izzo et al., 2016). At a smaller scale, dam removal on a tributary of the Penobscot River, the Sedgeunkedunk Stream led to a rapid recolonization (within one year) by sea lamprey (Hogg et al., 2013) and other migratory fish (Hogg et al., 2015).

However, many species have not recovered to target abundances (Opperman et al., 2011). Penobscot River runs of Atlantic salmon remain low (only 589 individuals were counted in 2021). River herring runs, although greatly improved, are likely still an order of magnitude less than runs pre-1600s (e.g., Hall et al., 2012). Despite dam removals, the Penobscot River remains a heavily impounded system. The two main-stem dam removals, while significant, had the limited influence of opening 15 km of river. Aspirational projections for Atlantic salmon (12,000), and American shad (2 million) are fully dependent on restoring connectivity between the ocean and important habitats in northern headwaters (Opperman et al., 2011). Fishways are often used to allow access to habitat that is otherwise constrained (Waldman and Quinn, 2022), and this is the case in the Penobscot River.

Most hydropower dams in the Penobscot River now have at least one form of fish passage that serves the general fish community, and several have also installed eel-specific bypasses (Opperman et al., 2011; Mensinger et al., 2021a; Molina-Moctezuma et al., 2021; Peterson, 2022). Overall upstream passage for adult Atlantic salmon at Milford Dam (now the lowermost dam), was relatively high (92%) over a span of six years (Izzo et al., 2016; Peterson, 2022), though with significant passage delays (days to weeks). Surviving downstream passage remains challenging for migrating juvenile Atlantic salmon (smolts) at Milford and other dams (Holbrook et al., 2011; Stich et al., 2014; Stich et al., 2015a). Downstream passage (for both juveniles and adults) is demonstrably critical for population persistence (e.g., American shad; Stich et al., 2019). In general, however, adult and non-salmonid juvenile downstream passage performance remains poorly characterized making such assessments important directions of future study.

Overall, however, fishway mediated access to upstream habitat falls far short of the capacity of unimpeded river systems (Zydlewski et al., 2021). Most fishways fail to effectively restore connectivity for all native species (Bunt et al., 2012; Noonan et al., 2012; Algera et al., 2020; Hershey, 2021). It is also common for fishways to be designed and evaluated for passage of salmonids (Noonan et al., 2012) and salmonids typically have the highest passage success through these structures (Noonan et al., 2012; Hershey, 2021). Importantly, migratory fish populations passing dams incur mortality, injury, and delay (Roscoe et al., 2011). Even the nature-like fishway at Howland Dam, while providing greatly improved passage (compared to when it was a functioning hydropower facility) remains distinguishable from free-flowing river reaches in terms of both passage delays and survival of Atlantic salmon smolts (Molina-Moctezuma et al., 2021).

Dams prevent access of some migrants to upstream habitat, but those that attempt to reach that habitat may face risk of injury or mortality. Upstream migrants must navigate fishways (or other paths) to move upstream. Passage attempts may lead to sub-lethal injuries (Castro-Santos et al., 2009) or mortality (Roscoe et al., 2011) due to engineered conditions or operational failures. While the run of American shad remains low in the Penobscot River, dozens to hundreds of dead adult American shad are removed from the fishway structure annually. Similar incidental losses for river herring and other migrants occur (Jason Valliere, Maine Department of Marine Resources, personal communication, August 31, 2023). Delays in passage may increase injury risk (McLaughlin et al., 2013) however it is difficult to assess what occurs within the fishway versus during searches for the fishway entrance.

Because dams are not freely permeable to movements in both directions, fish passage risks may be compounded by extensive searching periods. Migrating Atlantic salmon adults (Izzo et al., 2016; Maynard et al., 2017) and American shad (Grote et al., 2014b; Peterson, 2022) in the Penobscot River make wide-ranging upstream and downstream directional movements. Alewife are also known to “oscillate” in this way in other systems (McCartin et al., 2019). Such behaviors may put an individual at a great disadvantage even after successful upstream dam passage. These fish may suffer high mortality while moving back downstream (Castro-Santos and Letcher, 2010) being caught in an ecological trap that is confounded by the energetic cost of delay (Rubenstein et al., 2023). For iteroparous species, whether searching or not, fish that make it successfully past a dam must survive at least one downstream passage event.

Atlantic salmon are known to experience high mortality after completing spawning, but if successful in navigating dams moving downstream, they return to the sea (Maynard et al., 2018) either in the fall or following spring (Ruggles, 1980; Maynard et al., 2017). The population cost of losing these larger, multi-year fish may be significant due to their increased reproductive potential. Female Atlantic salmon may invest up to 25% of their body mass into egg production (Fleming, 1996). The loss of “big old fat female fish” may have significant population effects (sensu Hixon et al., 2014). Though Atlantic salmon restoration has focused on upstream passage more than multiyear spawning (USFWS and NMFS, 2018), the biological importance of these fish is clear (Fleming, 1996; Hixon et al., 2014).

Recruitment of juvenile anadromous fish spawned upstream of dams is entirely dependent upon the successful emigration downstream. Downstream migrating Atlantic salmon smolts have been extensively studied in the Penobscot River and river sections with dams are consistently identified as areas of high mortality (Holbrook et al., 2011; Stich et al., 2014; Stich et al., 2015a; Molina-Moctezuma et al., 2021; Molina-Moctezuma et al., 2022). As we explore later, sub-lethal injuries may partly explain why Atlantic salmon smolt mortality in the estuary is elevated both through delay and predation (Stich et al., 2015b; Stich et al., 2015c; Molina-Moctezuma et al., 2022).

For American eel, downstream migration is the culmination of up to decades of growth in fresh water before initiating fall migration. Migrants must locate a passage route and some individuals spend days searching, drawing on energy stores (Carr and Whoriskey, 2008; Piper et al., 2015; Eyler et al., 2016). Like other downstream migrating fish, adults risk impingement and impact-related injuries, as well as lethal and sub-lethal strikes by turbine blades in power generating stations (Piper et al., 2015; Eyler et al., 2016). In the Penobscot River, eel mortality rates are elevated at the two extant main-stem dams (West Enfield and Milford) but is indistinguishable from background mortality in reaches where Veazie and Great Works Dams were removed (Mensinger et al., 2021a). Injuries consistent with turbine blade strikes are commonly observed downstream of dams (Figure 4) supporting the assumption that entrainment through the turbines occurs – with negative outcomes for survival.

Challenges of downstream passage are a partial driver of the decision to not pass sturgeon upstream of Milford Dam. The few shortnose sturgeon that entering the Milford fish lift annually are moved back downstream based on uncertainty in their historical range (assumed to be at the Milford Falls, although unclear [see Knight, 1985; Petersen and Sanger, 1986]) and resulting need to also move back downstream as adults or juveniles if spawning occurred (Jeff Murphy, NOAA’s National Marine Fisheries Service, personal communication). There mortality and injury have been demonstrated during downstream movements of sturgeon encountering dams during in other rivers systems (McDougall et al., 2013; McDougall et al., 2014; Jones and Cotel, 2023).

Changing conditions may also affect passage and survival at dams. At Milford Dam, water attraction conditions on the west shore of the Penobscot River tend to attract upstream migrants that are frequently left stranded in pools as river and operational conditions change (Jason Valliere, Maine Department of Marine Resources, personal communication, August 31, 2023). When noticed, this has led to concerted efforts to capture and move upstream migrating Atlantic salmon (Figure 5). Other species without federal protection, however, are not prioritized for such interventions. During the fall migration it is common to see stranded alosines that succumbed to low oxygen conditions (Figure 5).

Rivers are the highways for migratory fish, and they are critical corridors for rapid movement. As we have discussed, dams restrict access to habitat. Engineering solutions (fishways) may facilitate movements and partly mitigate habitat fragmentation. Success, however, depends on three steps: attraction to the fishway entrance, entry, and successful transit. In the Penobscot River, the speed of movements and fish passage for both upstream and downstream migrating animals are slowed by dams in the system. This influence is obviated by the aggregations of migrating fish that may be observed below dams (Figure 6), but telemetry studies performed in the Penobscot River have provided more quantitative estimates of delay.

Prior to the removal of Veazie and Great Works dams, upstream migrating Atlantic salmon adults were denied access to, or delayed in, reaching upstream habitat in the Penobscot River (Holbrook et al., 2009; Sigourney et al., 2015). For adult Atlantic salmon, completion of the dam removals as part of the PRRP allowed rapid passage through the reaches with the dam remnants. However, adults now experience substantial delays (often several weeks) at Milford Dam, a facility that operates with a fish elevator (Izzo et al., 2016) compared with the original Denil style fishway (a series of closely-spaced U-shaped baffles; Holbrook et al., 2009). Similarly, American shad adults approached Veazie Dam prior to the dam removals but did not pass in large numbers (Grote et al., 2014a). While passage of American shad at Milford Dam has increased (to more than 10,000 annually; DMR, 2022), telemetry assessment suggests that passage remains hampered by delays (Peterson, 2022). In contrast, sea lamprey adults now reach Milford Dam and are relatively successful in passing the dam (70–82%) with little delay (Peterson et al., 2023). While mean delay times for passage were low for successful fish (<48h), others experienced substantial delays (9–11 days) before abandoning upstream movements altogether.

Delays at dams are commonly observed in many impounded river systems. American shad (e.g., Castro-Santos et al., 2017; Weaver et al., 2019) and river herring (Haro et al., 1999; Noonan et al., 2012) have been demonstrated to have difficulty passing dams. Sea lamprey face delays and repeated failures to move through the fishways in the Connecticut River (Castro-Santos et al., 2017). Delays for salmonids have been documented in many river systems (Raymond, 1979; Wertheimer and Evans, 2005; Scruton et al., 2008; Keefer et al., 2012; Caudill et al., 2013; Nyqvist et al., 2017; Hagelin et al., 2021; Ohms et al., 2022). These delays may be biologically relevant (e.g., impacting ontogenic synchrony with nature) or even deadly. When fish are delayed by dams, they may be subjected to temperatures that reach lethal or near-lethal levels (Marschall et al., 2011). As discussed above, flows or water regulation decisions may also trap fish in areas where they are unable to escape or are susceptible to predation from natural or human predators (Figure 5). The longer they remain in the area, the more protracted the risk.

For adult upstream migrants, even modest delays may be energetically costly. Rubenstein et al. (2023) found that Atlantic salmon were delayed an average of 16–23 days at Milford Dam prior to passing and had lost 11–22% of initial fat reserves. These losses may be compounded by high water temperatures if thermal refugia are not available (Holbrook et al., 2009). Such delay-mediated energy losses are likely to be important during a migration (Thorstad et al., 2008). Returning adults cease consumption upon freshwater entry (Kadri et al., 1995) so that energy stores are the sole fuel for survival, migration, and spawning success.

For iteroparous species (e.g., Atlantic salmon, American shad, and river herring) protecting energy stores may contribute to post-spawn survival. Glebe and Leggett (1981) suggested that loss of more than 60% of energy reserves may serve as a constraint to iteroparity. For American shad, empirical evidence suggests that the threshold for post-spawn survival may be as low as 30–40% (Leonard and McCormick, 1999). Risks of both direct and indirect mortality through delay-mediated energy depletion are heightened when fish must pass multiple dams. This heightened mortality is consistent with the observed declines of American shad repeat spawners in the Connecticut River (New England) that fell from 49% (Walburg and Nichols, 1967; Carscadden and Leggett, 1975; Limburg et al., 2003), to 5% in 60 years after accessing habitat upstream of impoundments (Atlantic States Marine Fisheries Commission [ASMFC], 2020). This pattern of “forced semelparity” (Zydlewski et al., 2021) has obvious implications for population demographics. For semelparous species such as the sea lamprey, adult delays may likewise result in added energy losses that impair survival, migration, and spawning. For upstream migrating juvenile American eel, delays may functionally restrict their access to rearing habitat (Verdon and Desrochers, 2002) and may have differential individual effects based on variation in motivation (Mensinger et al., 2021b).

Delays are also observed at dams as fish migrate downstream. In the Penobscot River, Atlantic salmon smolts displayed slower movement rates in areas where dams were located (Spicer et al., 1995; Stich et al., 2015b). Studies also found that smolts arriving during the day experienced longer delays than those that arrived at night (Holbrook et al., 2011). Delays are directly associated with lower survival in the Penobscot River (Molina-Moctezuma et al., 2022) and elsewhere (Castro-Santos and Haro, 2003; Marschall et al., 2011; Nyqvist et al., 2017). This pattern is not held at all dams in the Penobscot River, however, as Browns Mill Dam (on the Piscataquis River, Maine, USA) had the lowest mortality (indistinguishable from in-river mortality) but the highest delays (up to 10 days; Molina-Moctezuma et al., 2021). Delays for downstream migrating smolts were reduced after construction of the nature-like fishway at Howland Dam in 2016, however, only about one third of the smolts used the bypass. Downstream delays from dams on the Penobscot River were also found to occur for adult American eels (Mensinger et al., 2021a), which may have a significant impact on spawning and population success as these are old, semelparous individuals.

Flow conditions strongly influenced the delays incurred by fish moving up or downstream at dams. Downstream migrating adult American eels were slowed at West Enfield and Milford dams, but this lag was erased individuals by higher flows during the migratory season (Mensinger et al., 2021a). Similarly, the Penobscot River experienced exceptionally high spring flows in 2017, 2018 and 2019. Atlantic salmon smolts tracked during these three years had greatly reduced delays and higher survival relative to lower flow years total cumulative survival of greater than 75% versus less than 50% in previous years (Molina-Moctezuma et al., 2022).

Delays may influence migrants through a disassociation of ontogenic processes with environmental windows of opportunity. Many migratory species develop physiological characteristics associated with anticipated shifts in habitat at the time of migration (Zydlewski and Wilkie, 2012). Smoltification in salmonids is an adaptive developmental stage that synchronizes the physiological capacity to osmoregulate in seawater with migratory behavior (Zydlewski and Wilkie, 2012; Stich et al., 2015c) and has been described as a window of opportunity to match physiological capacity with environmental conditions (McCormick et al., 2009). Delays may disrupt the match between migration and development. Consequently, fish may enter the ocean in suboptimal conditions (McCormick et al., 1998; Zydlewski et al., 2005; Marschall et al., 2011). All other factors being equal, salmonids migrating later in the season likely face greater physiological challenges than early migrants (McCormick et al., 1999). Overall, this can create greater physiological challenges, affect sensitivity to starvation, and adversely influence ocean survival (McCormick et al., 1999; Zydlewski et al., 2003; Wilson et al., 2022).

In contrast to the ontogenic development of seawater tolerance in migratory salmon, American shad develop tolerance to full strength seawater about three months in advance of their downstream migration (Zydlewski and McCormick, 1997a). However, juveniles lose their ability to regulate ions in fresh water, a change influenced by declining autumnal temperatures (Zydlewski and McCormick, 1997b). These developmental changes mean that entry into seawater late in the migratory season is physiologically challenging (Zydlewski et al., 2003) which may ultimately reduce survival of juvenile shad (Shrimpton et al., 2001). Whether migrating as a juvenile or as an adult, delays at dams consume critical energy for migration, cause a mismatch between developmental stages and the environment, and expose fish to unfavorable environmental conditions. We have shown how delays may lead to diminished survival and reduce biological fitness. Delays are also intimately intertwined with other risks, such as predation.

Dam and fishway construction have been shown to create habitat suitable for opportunistic or ambush predators (Rieman and Beasmesderfer, 1991; Pasha et al., 1997). Fish that are aggregated near dams, delayed in passage, or disoriented by flow and turbulence, are increasingly vulnerable to predation (Ruggerone, 1986; Rieman and Beasmesderfer, 1991; Isaak and Bjornn, 1996; Blackwell and Juanes, 1998; Agostinho et al., 2012). Predator vulnerability of fish delayed near dams is obviated by seasonal diet shifts of predators incorporating more migrant prey species in their diets (Blackwell and Krohn, 1997; Fritts and Pearsons, 2006). This dietary shift has been observed in the Penobscot River where river herring have become an important seasonal prey item of smallmouth bass (Micropterus dolomieu) in the lower river (Watson et al., 2019).

Bald eagles (Haliaeetus leucocephalus), osprey (Pandion haliaetus) and double-crested cormorants (Phalacrocorax auritus) are known predators of diadromous species near impoundments (Ross and Follen, 1988; Call, 2015). In the Penobscot River, cormorants have been known to select seasonal foraging areas adjacent to dams to feed on migrating Atlantic salmon smolts in addition to other anadromous species such as rainbow smelt or river herring (Blackwell and Krohn, 1997). Prior to the PRRP, most avian predator diets in the upper Penobscot River were freshwater in source, although bald eagles were likely foraging on stocked Atlantic salmon smolts (Call, 2015). Presumably, avian predators now exploit the increased alosine forage base upstream of Milford Dam. Even large upstream migrants are vulnerable to avian predation. One of three adult salmon that passed Browns Mill Dam in 2020 was captured by a bald eagle while delayed in the 1 km head pond between dams Browns Mill and Moosehead Dams on the Piscataquis River (Peterson, 2022).

The role of in river predation for Atlantic salmon smolts has been of growing conservation concern. While marine mortality has been identified as being a critical source of loss for the species, freshwater and estuarine mortality may exceed coastal mortality for Atlantic salmon smolts (Kocik et al., 2009; Hawkes et al., 2019). In the Penobscot River, mortality rates for smolts are relatively low in free-flowing stretches of river but elevated near dams (Holbrook et al., 2011; Stich et al., 2015a; Molina-Moctezuma et al., 2022). Therefore, resolving the causal agent of mortality at dams is important for exploring possible mitigative actions. The development of acoustic predation tags provides a new tool for determining the disposition of tagged fish (Halfyard et al., 2017; Schultz et al., 2017). Recent acoustic telemetry data with predation sensors in the Penobscot River suggests that predation risk is 5-fold greater through impounded reaches (Mensinger et al., in press).

Atlantic salmon smolt mortalities in the estuary are also linked to their dam passage experiences (Stich et al., 2015a) likely because of increased delays (Molina-Moctezuma et al., 2022). Observed delays in transit rates at dams (Holbrook et al., 2011; Norrgård et al., 2013; Stich et al., 2015b) may result in loss of physiological smolt characteristics (McCormick et al., 1999; Budy et al., 2002; Ferguson, 2006) thereby reducing performance (Handeland et al., 1996). Thus, the asynchrony between the development of osmotic tolerance and timing of arrival in the estuary may contribute to mortality in the Penobscot estuary (Stich et al., 2015c). Such changes in salinity tolerance are exacerbated by descaling injuries as might occur at dams (Zydlewski et al., 2010) and may increase susceptibility to predation. Observations that predation risk in the estuary is nearly twice that of impounded areas (and 9-fold greater than in free-flowing river; Mensinger et al., in press) suggest a causal relation between dam delays and predation mortalities.

The slow rate of travel faced by downstream migrants in impoundments exposes migrants to fish, avian and mammalian predators. Weldon Dam’s impoundment is approximately 5 km long and is a reach of exceptionally high mortality risk for Atlantic salmon smolts (as high as 25%; Stich et al., 2015a; Molina-Moctezuma et al., 2022). It is notable that the gauntlet of predators in the impounded regions differs from the natural river. The changed lentic community favors predatory species including smallmouth bass, largemouth bass (Micropterus salmoides) and chain pickerel (Esox niger). While all these predators may be found throughout the Penobscot River, smallmouth bass are exceptionally widespread (Kramer, 2006; Kiraly et al., 2014). Smallmouth bass are a generalist and piscivorous fish that has been widely introduced (Loppnow et al., 2013) and implicated in declines of salmonids and other taxa (Magoulick, 2004; Fritts and Pearsons, 2006; Middaugh et al., 2016). As these predatory species are abundant in impoundments (Whittum et al., 2023), dam removal may notably reduce predation risk to juveniles by both reducing predator density and increasing speed of migration (Pasha et al., 1997).

The presence of predation in and near river impoundments is widely observed. In Pacific Northwest, migrating salmon (Oncorhynchus spp) are a key prey item for smallmouth bass near dams (Tabor et al., 1993; Fritts and Pearsons, 2004). Adult Atlantic salmon are preyed upon by the European catfish (Silurus glanis) near fishways in France (Boulêtreau et al., 2018) and in Brazil, redeye piranha (Serrasalmus rhombeus) prey on neotropical fish near fishway entrances (Agostinho et al., 2012). We note the presence of humans as another predator that are drawn to the aggregations of fish below dams (Jackson and Davies, 1988; Carey et al., 2011). Migratory fish in the Penobscot River are vulnerable enough near dams to warrant closure to “That area within 150 feet of any part of the Medway, West Enfield and Milford Dams, including fishways” (MDIFW, 2023). Milford Dam is highlighted as a place to fish for striped bass as their prey are congregated (Holyoke, 2021). Such angling pressure for striped bass and American shad results in the hooking of endangered adult Atlantic salmon (Jason Valliere, Maine Department of Marine Resources, personal observation, August 31, 2023).

Dams fundamentally change local biophysical conditions. Impoundments warm quickly, and shift from a lotic to lentic habitat. These changed conditions may disadvantage native species, providing a permissive environment for non-native species (Baxter, 1977; Ward and Stanford, 1987; Pess et al., 2008). Non-native fishes may find themselves well-positioned to outcompete or prey upon native salmonids and other fishes adapted to cold, free-flowing habitats. Impoundments may also tend to favor non-native minnow species (e.g., Whittum et al., 2023), which often arrive as bait-bucket introductions (Ludwig and Leitch, 1996). Dam-influenced fish assemblages are therefore not necessarily less diverse than those assemblages in free-flowing rivers (e.g., Burroughs et al., 2010; Hogg et al., 2015), but they are generally depleted in native species and enriched with non-native species. The Penobscot River is predominantly inhabited by macrohabitat generalist species (e.g., smallmouth bass) and riverine species (e.g., white sucker), and these species dominate the biomass in the main-stem Penobscot River.

Dam removals have been observed to result in rapid and often profound changes to riverine fish communities (Catalano et al., 2007; Burroughs et al., 2010; Hitt et al., 2012; Poulos et al., 2014). These changes have been demonstrated in the Sedgeunkedunk Stream, a tributary of the Penobscot River where persistent changes were observed within weeks of dam removal (Gardner et al., 2013; Hogg et al., 2015). In coastal systems, recolonization of diadromous fishes in newly available habitat represents a major shift (Hitt et al., 2012; Weigel et al., 2013; Hogg et al., 2015). Similarly, the fish community in the Penobscot River was substantially modified by the increased presence of migratory fish, but this influence diminishes with the number of dams that must be passed (Kiraly et al., 2014; Watson et al., 2018; Whittum et al., 2023).

The Penobscot River remains a heavily impounded system with improved, but demonstrably imperfect fish passage. Although upgraded passage has allowed migratory species to attain greater upstream ranges, much of the community structure above the lowermost dam has remained similar to pre-dam removal conditions (Whittum et al., 2023). Upriver, the East Branch of the Penobscot River currently has had little to no presence of alosines, suggesting poor combined passage through Milford, West Enfield, and Weldon dams. Weldon Dam is a significant barrier (13.7 m; USACE, 2023) compared to downstream dams. Fish passage at Weldon Dam (pool and weir) and West Enfield Dam (vertical slot) are likely more selective than the fish elevator system at Milford Dam (Bunt et al., 2012).

In the Piscataquis River, the Howland nature-like fishway allows for fish passage (to and from the Penobscot River) without changing the impoundment. The solution of a nature-like fishway is desirable when there is high social or cultural value attached to the impoundment. Like other fishways, this structure leaves the habitat upstream of Howland Dam with little change after construction (Kiraly et al., 2014; Watson et al., 2018; Whittum et al., 2023). As might be anticipated, the extant impounded riverine reaches within the Penobscot River continue to favor cyprinid species and higher relative abundance of top predators, such as chain pickerel, smallmouth bass, and largemouth bass. As a result, any juvenile migrants above these dams may encounter an enhanced local density of piscivores when moving downriver. Migration delays at impounded locations may provide additional opportunity for predation from these reservoir species (Molina-Moctezuma et al., 2021). The risk is multiplied if passing several dams.

We would be remiss if we did not note that dam removal may also have unintended (and undesired) outcomes for the native fish community by facilitating the movement of non-native species. The expansion of some non-native fish such as the white catfish (Ameriuris catus) are likely the result of increased movement permeability in the system (Whittum, 2022). When contemplating the use of dam removal as a conservation approach, managers may benefit from considering how enhanced passage might influence non-native species (Cooper et al., 2021). Indeed, fears of the expansion of angler transported (and non-native) northern pike (Esox lucius) have prompted state proposed legislation to modify fishways on tributaries of the Penobscot River to exclude “invasives” (LD 1049, 131^st Maine Legislature, An Act to Protect Maine’s Inland Fisheries from Invasive Fish). In principle, fishway modification may allow for differential passage (and therefore connectivity) for different species based on species-specific fishway passage performance (Noonan et al., 2012). This approach would, however, make fish passage more difficult for all species.

Dams have the potential to cause demographic shifts in populations due to a suite of influences on distinct ontogenetic stages and variable life histories within species. These influences may include shifts that result from selective pressures such as size selection in fish passage and mismatches between upstream access and downstream survival through dams for migrants. At the population level, shifts may manifest as changes in size and age structure, reductions in rates of iteroparity, or loss of life-history complexity and variability. While demographic shifts are increasingly well documented, the degree to which selective pressures influence local adaptation is poorly understood for many species.

Size selection in fish passage may limit access to spawning and rearing habitat by individuals of specific sizes, which alters average size of individuals while reducing variability in size. For anadromous species, selectivity in fish passage directly imposes selection on the component of adult spawners that access spawning habitat. This selection may operate on physiology, anatomy, or behavior (Mensinger et al., 2021b). Size-selective passage of Atlantic salmon has been observed in the Penobscot River (Sigourney et al., 2015; Maynard et al., 2017), whereby larger fish were less successful in passing dams to reach spawning habitat. Exclusion of the largest females from spawning grounds may affect both underlying phenotypes of spawners and survival in other life stages for a population. Trucking spawning fish around fishways, combined with conservation hatchery practices, may partly alleviate this selective pressure (Sigourney et al., 2015; Maynard et al., 2017).

Mensinger et al., 2021b observed that larger juvenile (glass) eels climbed faster than smaller ones. For catadromous species this size selection in passage may lead to differences in individual growth opportunities in freshwater habitats, or even skew sex ratios based on physiological limitations of habitat downstream of dams (Mensinger et al., 2021b). Size selection at challenging passage structures (anthropogenic or natural) may impose energetic limitations on rates of growth, maturation, and migration that have not been extensively studied. It may simply make some upstream habitat functionally inaccessible (Verdon and Desrochers, 2002).

Dams may also cause shifts in size structure and life history traits through mortality during downstream passage. If downstream survival of adult and juvenile fish through dams is not sufficiently high, upstream fish passage may become an ecological trap (Ohms et al., 2022). Because life history traits such as size, age at maturity, and iteroparity are co-inherited (Aykanat et al., 2019), downstream survival through dams may influence multiple population demographics simultaneously. Low survival through dams during downstream migration by post-spawn adults was associated with reduced iteroparity of Atlantic salmon in the Penobscot River (Maynard et al., 2018) and steelhead trout (Oncorhynchus mykiss) in the Snake and Columbia rivers in the northwestern USA (Wertheimer and Evans, 2005; Keefer et al., 2008). When adult downstream survival rates through dams were less than perfect (i.e., 100%), American shad age structures were predicted to be truncated through loss of older fish and repeat spawning rates were predicted to be reduced with increasing upstream fish passage (Castro-Santos and Letcher, 2010; Stich et al., 2019). Additive mortality incurred by passing multiple dams is compounded at the watershed scale with respect to changes in demographics such as abundance, size structure, and iteroparity (Castro-Santos and Letcher, 2010; Stich et al., 2019; Zydlewski et al., 2021).

The degree to which demographic shifts are realized may vary with damming intensity and upstream fish passage, environmental conditions, and life history variation. Abundance of American shad, for example, varies as a function of upstream passage, number of dams, configuration of spawning and rearing habitat relative to dams, as well as latitudinal clines in growth, maturation, fecundity, and iteroparity (Zydlewski et al., 2021). In the Penobscot River, where only 16 adults passed the lowermost dam from 1978 through 2012, fish exhibited repeat spawning rates as high as 75–95% (Grote et al., 2014b), though the population persisted at low abundance (Grote et al., 2014a) prior to the removal of Veazie Dam. Fish reached smaller maximum sizes and reached older ages despite elevated natural mortality estimates in the Penobscot River compared to other rivers in the northeastern part of their range (Gilligan-Lunda et al., 2021). In the Connecticut River, maximum age and repeat spawning rates have been reduced since implementation of upstream fish passage and despite closure of commercial fisheries and stable spawner abundances (ASMFC, 2020). These reductions in maximum age and repeat spawning have occurred even though Connecticut River American shad reached larger sizes and experienced lower mortality rates than in the Penobscot River (Gilligan-Lunda et al., 2021).

In species or populations with variable life histories, dams may influence population demographics through elimination of life history complexity, thereby reducing evolutionary stability of populations in variable environments. For example, coastal cutthroat trout (Oncorhynchus clarkii; Trotter, 1989) and steelhead trout (Thorpe et al., 1998; Satterthwaite et al., 2009; Hodge et al., 2016) exhibit high diversity in anadromous and freshwater resident life histories, in addition to variability in iteroparity. This diversity of life histories presumably reduces risk to extirpation through a portfolio effect (Moore et al., 2014) and includes genotypic and plastic responses (Satterthwaite et al., 2009; Whiteley et al., 2010). Whereas population structuring has been observed in steelhead trout upstream and downstream of barriers in the Elwha River, this structuring rapidly degraded following dam removal. Data suggest that overall genetic diversity was preserved within isolated freshwater and anadromous populations, indicating strong potential for recovery (Fraik et al., 2021). Similarly, a large body of research has demonstrated rapid adaptation of freshwater life histories in alewife following dam construction and land locking independently among many populations in Connecticut, USA (Palkovacs et al., 2008). Both phenotypic and genotypic responses in alewife populations are postulated to create eco-evolutionary feedbacks that drive rapid changes in populations (Palkovacs and Post, 2008). Recent evidence suggests that many of the underlying genomic changes can also be reversed through introgression with anadromous individuals following implementation of fish passage or dam removal (Reid and Goodman, 2020).

Finally, American shad exhibit parity on a continuum across their native range, with semelparous populations in southern rivers (south of 35° latitude) and increasingly high rates of iteroparity in northern rivers (Leggett and Carscadden, 1978) that correlate to differences in maturation (ASMFC, 2020), growth, and longevity (Gilligan-Lunda et al., 2021) and regional population structuring (Hasselman et al., 2010). Within the central range (35–41° latitude) of American shad, intermediate rates of iteroparity prevail, with multiple life histories present in some rivers (ASMFC, 2020). It remains unknown to what degree these life histories vary longitudinally within rivers. Work in undammed rivers (e.g., Delaware River, mid-Atlantic region of the USA) may aid in differentiating between the influence of migratory distance and damming.

The six influences of dams discussed above are directly tied to interactions with migratory fishes. The ecological influences of dams, however, include changes in ecosystem function and human use that are indirectly linked to migrating fishes through ecosystem functions. We broadly identify these changes as “ecosystem services”. Inclusion of connections between migratory fishes and broader ecosystem function in this paper is consistent with shifts towards more holistic approaches to ecosystem and multispecies management (Larkin, 1996; Eriksson et al., 2011; Andersen et al., 2015). Ecosystem services have supported human well-being through history in both expected and unanticipated ways (Limburg and Waldman, 2009; Hall et al., 2012).

Dams have been the primary cause of migratory fish population declines across North America (Limburg and Waldman, 2009) and Europe (Wilson and Venerata, 2019). The direct loss of fisheries potential in the Penobscot watershed has been documented (Saunders et al., 2006). Hall et al. (2011) calculated that by 1850, river herring spawning habitat in Maine had been reduced to less than 5% of available habitat because of dam construction on small rivers, and to 1% of habitat by 1887 when the largest rivers were spanned by dams (Atkins, 1887). Impoundment of the Penobscot River has reduced American shad production potential nearly 90% (Stich et al., 2019), contributing to a dam-related coast-wide loss of 39% based on habitat access alone (Zydlewski et al., 2021). The loss of recreational angling for sea run fish also affects the region’s economy (Pinfold, 2011) and culture (Schmitt, 2016). The loss of fisheries has been particularly detrimental to indigenous communities.

The connection of Wabanaki people on the Penobscot River to once abundant sea run-fish species has been important for both sustenance and cultural connection (Speck, 1940; Harper and Ranco, 2009). The Penobscot Nation, penawahpkekeyak, are the people of the place of the white rocks, referring to a reach of river that bears their name. Similar connections are reinforced through stories and folklore (Kolodny, 2007). This federally recognized tribe has more than 2,400 enrolled members and is considered one of the oldest continuous governments in the world (Charlie Loring, Jr., Director, Department of Natural Resources, Penobscot Indian Nation, personal communication, August 31, 2023). Dam-mediated losses of sea run fish are viewed as a critical threat to the Penobscot Nation’s formative connection to the river, and the PRRP has been viewed as integral to the “reclamation of their cultural identity and sovereignty” (Frederick, 2006).

Dam-mediated losses of migratory fish runs may have effects beyond their local geography, either through trophic connections (as predator or prey) or range-wide population resilience. For the semelparous, panmictic American eel, adult downstream migrations are limited by access to upstream juvenile rearing habitat. Dam related loss of production from any river means that the entire population is diminished, particularly as adults from northern regions tend to be female (Wang and Tzeng, 1998; Jessop, 2010). Ames and Lichter (2013) assert that large, stable concentrations of young-of-the-year alosines influenced where resident and migrating gadid (cod) groups were located. The dam-associated loss of river herring and shad resulted in a loss of forage for ground fish such as Atlantic cod (Gadus morhua) haddock (Melanogrammus aeglefinus), pollock (Pollachius virens), and white hake (Urophyscus tenuis). These important northern coastal shelf fisheries collapsed coincident with the damming of rivers in the region (Lotze and Milewski, 2004). Striped bass also prey on blueback herring, alewife, and American shad, thereby benefiting from their abundance (Trent and Hassler, 1966; Nelson et al., 2003; Walter and Austin, 2003; Savoy and Crecco, 2004).

As dams alter fish assemblages, other interactions may directly or indirectly occur (Hanson and Curry, 2005; Kiffney et al., 2009). The presence of alewife and blueback herring in high numbers may benefit other species through substitution. As river herring populations rebound in the Penobscot River, it has been suggested that they may serve as a prey buffer for species such as the endangered Atlantic salmon (Saunders et al., 2006; Oke et al., 2020; Hare et al., 2021). These species are likely to become more important as a forage base as populations increase both in river and along the coast (Ames and Lichter, 2013). Mammalian predators such as the harbor seal (Phoca vitulina) and birds (e.g., bald eagles, osprey, double-crested cormorants) would also likely be limited by dam-mediated population losses (Able and Fahay, 2010; Call, 2015).

Predation pathways contribute important linkages among freshwater, marine, and terrestrial ecosystems by transference of energy and nutrients (e.g., Durbin et al., 1979; Petticrew et al., 2011; Willis et al., 2017; Barber et al., 2018). Anadromous species transfer freshwater-derived nutrients to marine environments during juvenile seaward migration (Willis et al., 2017; Barber et al., 2018). Then adults, which obtain a high proportion of their mass while feeding in marine environments, transfer “marine-derived nutrients” (hereafter MDN) to freshwater ecosystems during the spawning season through direct consumption, the release of gametes, excretion of metabolic wastes, and carcasses decay (e.g., Durbin et al., 1979; Twining et al., 2017; Barber et al., 2018). The pathways of MDN transference from anadromous fish to streams vary among their life history. While carcass decay is the principal input from semelparous species, excretion is the principal input from some iteroparous species such as alosines (Schindler et al., 2003; Post and Walters, 2009; Figure 7). Additionally, carcasses of anadromous fishes may be transferred to terrestrial ecosystems by water movement and terrestrial predators, thereby transferring MDN to riparian food webs (Hocking and Reynolds, 2011; Quinn et al., 2018). Catadromous species can generate the same transference of energy and nutrients in an opposite pathway (Saboret et al., 2021).

Pulsed subsidies of MDN from anadromous fish increase the primary productivity of freshwater ecosystems through the bottom-up pathway of nutrient incorporation. In temperate regions, where freshwater ecosystems are less productive than marine ecosystems, MDN from anadromous species transfers nutrients to freshwater ecosystems increasing their primary productivity (Wipfli et al., 2010; Weaver et al., 2016; Weaver et al., 2018a). In the Penobscot River basin, both bottom-up and top-down pathways of MDN incorporation in freshwater ecosystems have been studied experimentally through carcass addition and theoretically through modeling (Guyette et al., 2013; Guyette et al., 2014; Weaver et al., 2015; Weaver et al., 2016; Weaver et al., 2018b; Weaver et al., 2018c; Zydlewski et al., 2021). Both producers and consumers incorporate MDN from anadromous fish subsidies (Guyette et al., 2013; Guyette et al., 2014; Weaver et al., 2016; Weaver et al., 2018b). Direct or indirect assimilation of carcass material may increase growth of young Atlantic salmon thereby bolstering survival (Guyette et al., 2013). Sea lamprey larvae assimilated nutrients found from the carcasses of their post-spawn adult conspecifics, which may improve their growth and enhance earlier metamorphosis (Weaver et al., 2018c).

Consumption of MDN from anadromous fish subsidies may also transfer beneficial biomolecules to in-stream consumers. Anadromous fishes accumulate high contents of vital dietary biomolecules such as ω-3 highly unsaturated fatty acids (hereafter ω-3 HUFAs) in their tissues while feeding in marine food webs; therefore, they transfer them from ω-3 HUFAs-rich (marine) to ω-3 HUFAs-poor food webs (freshwater) through their subsidies (Fuiman et al., 2015; Figueroa-Muñoz et al., 2022; Závorka et al., 2023). Evidence of transference of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) to in-stream consumers following the consumption of anadromous fish subsidies (i.e., carcasses and eggs) has been documented both experimentally and in natural systems (Heintz et al., 2004; Landsman et al., 2018; Figueroa-Muñoz et al., 2022). Indeed, pulsed subsidies of MDN from anadromous fishes, especially eggs, can constitute an important source of ω-3 HUFAs to predators in the Penobscot River.

The absence of migratory fish from habitat in the Penobscot River has deprived the ecosystem of other services as well. Atlantic salmon and sea lamprey are considered “ecosystem engineers”: they physically change benthic habitats during their nest construction, thereby affecting other organisms in streams (Moore, 2006; Hogg et al., 2014). Spawning sea lamprey generate changes in the streambed during nest construction, making it more complex and benefitting benthic invertebrates (Hogg et al., 2014; Weaver et al., 2018b; Figure 7). Furthermore, nest construction may benefit drift-feeding species (e.g., brook trout and Atlantic salmon) by providing them with energetically profitable foraging habitats (Fausch and Northcote, 1992) or by improved spawning habitat through removal of fine sediments (Saunders et al., 2006; Hogg et al., 2014).

The interruption of migratory fishes by dams also interrupts the life cycle of a myriad of parasitic organisms. While impacts on some species may not initially be recognized as a lost opportunity cost for society (e.g., a host for freshwater copepods; Powell et al., 1999), other interactions are readily recognized as valuable from the human perspective. Freshwater mussels are one of the most imperiled groups of animals in North America (Master et al., 1998) with many species federally listed or extinct (Bogan, 1998). As a result, two species – the yellow lamp mussel (Lampsilis cariosa) and tidewater mucket (Leptodea ochracea) – were listed as Threatened under the Maine Endangered Species Act in 1997. Within watersheds, dams constrain fish movement and limit freshwater mussel dispersal through their fish-host dependent life history. Glochidia – the parasitic larval stage of freshwater mussels – generally require a vertebrate host to complete development (Kat, 1984; Kneeland and Rhymer, 2008). It is easy to see, therefore, how limitations in fish movement also may limit distribution patterns of mussels. Such examples demonstrate how the loss of migratory fish on one system may influence local and regional ecosystems in complex ways.