Abundance and Distribution of American Eel in a Heavily Dammed Urban River

The American Eel Anguilla rostrata is the only catadromous fish species that occurs in rivers along the North American Atlantic coast. American Eels spawn from the Sargasso Sea to the Mid-Atlantic Ridge (Chang et al. 2020), with larvae being passively transported from there by the Gulf Stream before dispersing to watersheds from northeastern South America to Labrador and Greenland (Vélez-Espino and Koops 2010). The larvae morph into transparent glass eels as they approach estuaries and rivers but then quickly transform to pigmented “elvers.” American Eels appear to undergo environmental sex determination, with males dominating in more saline waters where population densities are higher and females in fresher environments where population densities are lower (Krueger and Oliveira 1999). Once having entered a river, “yellow” eels may reside in those waters for up to 30 years (Jessop et al. 2004) before transforming to the mature “silver” phase, exiting freshwater habitat, and migrating to the ocean to spawn once and die (Béguer-Pon et al. 2015).

American Eels have declined greatly across their range (Vélez-Espino and Koops 2010; ASMFC 2017). At one time, American Eels accounted for as much as 25% of total fish biomass in Atlantic coastal rivers (Ogden 1970) and supported valuable commercial fisheries. Today they persist with overall recruitment at less than 1% of pre-1980 levels (Busch and Braun 2014). Overfishing, contamination, and infestation of a nonnative parasitic nematode Anguillicoloides crassus have contributed to the decline of American Eels (Ashley et al. 2003; Morrison and Secor 2003; Palstra et al. 2007). Also, existing dams have been major contributors to lowered abundances by causing habitat fragmentation, restricted upstream and downstream movements, and, in the case of hydrodams, downstream mortality from turbine blades (Carr and Whoriskey 2008; Eyler et al. 2016; Woods and McGarvey 2018). American Eels are now hindered from access to more than 84% of their historic inland waters along the U.S. Atlantic coast (Haro et al. 2000). Long term, the trend in their abundance may be for as much as a 50% population decline (NatureServe 2012). Conservation concerns prompted petitions for listing this species under the Endangered Species Act in 2004 and again in 2010; however, neither was successful, and the species is currently considered as stable by the U.S. Fish and Wildlife Service and as depleted at or near historically low levels by the Atlantic Marine Fisheries Commission (ASMFC 2012, 2017).

Upriver migration, with consequent expansion of habitat and trophic resources, is integral to sustaining a robust population of this panmictic species (Machut et al. 2007). Whereas blockage by dams is believed to be a large contributor to American Eel declines (Woods and McGarvey 2018), American Eels show the ability to pass some dams, even where there are no engineered fishways suitable for them (Machut et al. 2007). However, patterns of upriver migration vary with spatial scale (Smogor et al. 1995). At large spatial scales (e.g., states, physiographic provinces), American Eel distribution and abundance were consistent with a diffusion model based on distance from the ocean and not on more localized habitat features (Smogor et al. 1995). At smaller spatial scales, they found that patterns of American Eel occurrence were related to some habitat features, with dams possibly contributing to backups of upriver movements in some rivers. Southeastern New York’s Bronx River is an especially suitable location to assess the effects of barriers on the inland migration of American Eels. The river has nine significant dams of various forms and dimensions (1.2–5.5 m in height). To evaluate the effects of these barriers on the distribution and movements of the river’s American Eels, we used electrofishing, mopping, and mark–recapture techniques over 3 years.

METHODS

Study site

Originating north of New York City in Valhalla, New York, the Bronx River flows south for 37 km through central Westchester and Bronx counties before emptying into the tidal strait known as the East River (Figure 1), which is part of the broader Hudson River estuary. The Bronx River has been dammed since the late 17th century (De Kadt 2011) when the first dam was erected at what is now 182nd Street in the Bronx. This dam still stands as the lowest of nine dams (none hydroelectric) on the river (Table 1). The Bronx River is tidal nearly to the 182nd Street Dam, approximately 4.2 km upstream from the river’s mouth.

TABLE 1. List of the dams on the Bronx River and their characteristics.

Dam number (from lowermost)	Dam name	Dam height (m)	Distance from mouth (km)
1	182nd Street Dam	5.5	4.8
2	Twin Dams	3	6.3
3	Stone Mill Dam	2.1	7.2
4	Bronxville Dam	1.5	17.7
5	Old Stone Mill Dam	1.2	19
6	Hodgman Dam	1.2	19.3
7	Strathmore Dam	1.5	23.5
8	Popham Road Dam	3	25.4
9	Brook Lane Dam	2.1	29.6

Because American Eels typically aggregate below dams (Machut et al. 2007; Woods and McGarvey 2018), sampling sites were chosen as near to the base of each dam as was safely possible. For electrofishing, 30-m-long sections of the river were blocked off using two 15.2- × 1.2-m 3/8-in delta knotless seine nets (Memphis Net and Twine, Memphis, Tennessee). At dam sites where electrofishing was possible up to the base of a dam, only one net was set 30 m downstream from the dam itself. To estimate the area of each sample site, river width at the time of sampling was measured every 10 m with a Bushnell Tour V3 laser range finder. The GPS locations of the two nets (or one net and dam location) were recorded. Area was measured using the Measure Area tool in ArcGIS. Each site was measured three times and the mean used as the site area. In total, we sampled sites immediately below nine dams (Table 1; Figure 1). To assess whether movements occurred between dam and nondam sites, we also sampled at two sites between dams (i.e., between dams 2 and 3 and between dams 3 and 4) once each in 2015, 2016, and 2017.

American Eel capture and tagging

American Eels were captured using depletion electrofishing (Peterson et al. 2004) with a backpack electrofisher (Halltech Aquatic Research, Guelph, Ontario; 100 volts at 40 Hz) in summer and autumn from June to November over 2014 to 2016 (Table 1). Three netters accompanied the shocker to collect stunned American Eels. To keep capture probability constant, we used the same three netters for all passes at each site. Sites were sampled multiple times per year to detect movement of tagged American Eels.

After each pass, captured American Eels were anesthetized in a bath of clove oil (12 mL) and river water (3 gal). Once American Eels were sedated, their total length was measured to the nearest millimeter. Yellow eels measuring greater than or equal to 250 mm were first scanned with a passive integrated transponder (PIT) tag reader (Oregon RFID, Portland, Oregon). If no tag was present, we implanted an 8-mm FDX PIT tag (Oregon RFID, Portland, Oregon) subcutaneously, just below the origin of the dorsal fin using an injection needle. Tag retention at this location has been found to be 94–100% (Verdon et. al. 2003; Zimmerman 2008). This location is also less likely to puncture internal organs and cause mortality, as opposed to visceral cavity placement (Zimmerman 2008). Furthermore, we chose to allow wounds to close naturally as healing times are faster than using an adhesive and mortality is lower than using sutures (Baras and Jeandrain 1998). American Eels deemed too thin or in poor physical condition were not tagged. Any American Eels with comparatively large eyes and a silvery coloration were identified as “silver” (maturing prior to migrating out to sea) and were noted. All captured American Eels were then placed in a bin with fresh river water to recover until all passes were completed. Once recovered, American Eels were released at their capture site.

To complement electrofishing and to determine upstream penetration and timing of ingress by smaller American Eels of glass, elver, and yellow phases to the Bronx River, we used passive traps called “eel mops.” These are made from unraveled rope attached to a weighted disc that serves as attractive artificial habitat (Silberschneider et al. 2001; Sullivan et al. 2009). Mops were deployed in sets of three below each of the first six dams on the Bronx River, as close to the base of the dam as safely possible, from April to October 2015 and 2016 and below the first two dams from April to September 2017. Each mop was sampled twice per week (flow conditions permitting) using methods from Sullivan et al. (2009). After each mop was sampled, American Eels were counted, identified to life stage (glass phase = unpigmented, elver phase = uniform yellow coloration, yellow phase = brown dorsal coloration), and transferred to a holding bucket until all mops were sampled. American Eels were then released at the capture site. These American Eels were not measured or tagged. Mop data was summarized by site, year, and life stage. Qualitative comparisons were based on differences in catch per unit effort (CPUE) among sampled sites. One unit of effort was considered one individual mop sampled.

In August 2015, the New York City Department of Parks and Recreation installed an eel ladder at the first dam. The ladder did not reach the top of the dam but emptied into a collection bin allowing “passed” eels to be counted, identified to life stage, and then transported manually over the dam. The ladder was operational from August to October in 2015, though no American Eels were transported over the dam. In 2016 and 2017 the pass was open between March and October, and American Eels were transported over the dam.

Statistical analysis

To estimate abundance at each of the sites where American Eels were found, we used the robust design parameterization of the Huggins closed-capture model within the program MARK (White 2008). In all cases, we fixed the recapture probability to zero to account for removal of captured American Eels during each depletion pass. During each depletion event, which encompassed all depletion passes on a given date, we assumed that capture probability was constant among depletion passes. In most cases, we utilized three-pass depletion; however, we also utilized two-pass depletion to allow us to sample additional sites during a given date. Separate abundance estimates were calculated for American Eels greater than or equal to 250 mm TL and for those less than 250 mm TL.

The Huggins parameterization of the robust design (Huggins 1989) allowed us to set capture probability as equal among all depletion passes for a given site. We were then able to estimate abundance in situations where depletion estimates could not be calculated with an individual model for a given depletion event. We further fixed survival to zero and used the no-movement parameterization for gamma prime and gamma double prime, which fixes probabilities to one and zero, respectively (Williams et al. 2002). Capture histories for each American Eel were adjusted so that if an eel was captured during two or more depletion events, the individual capture history was separated into a separate capture history for each depletion event.

Mean abundance at each site was calculated as the mean of all of the point estimates for the two size-groups at a given site. Mean length was calculated for each size-group. Weight for each American Eel was estimated based on the composite length–weight formula for American Eel available on Fishbase ( www.fishbase.org; November 2018), which incorporated 10 separate length–weight relationships and a and b parameters of 0.0010 and 3.17, respectively. From these estimates, mean mass per eel for each site and depletion event was calculated for American Eels greater than or equal to 250 mm TL and those less than 250 mm TL. Mean mass per eel per site and depletion event was then multiplied by the associated abundance estimate for each size-class at each site and averaged for each size-group to calculate mean biomass. Densities (in eels/m² [numerical] and in g/m² [biomass]) were then calculated by dividing mean abundance and mean biomass at a given site by the area of the site.

Our initial plan was to use mark–recapture data to explore site-to-site transitions and apparent survival within the Bronx River using the Robust Design approach within MARK. Because of a limited number of site-to-site recaptures, we were not able to pursue this method. Further, exploratory analysis with a simple Cormack–Jolly–Seber model (Williams et al. 2002) showed that we were unable to obtain usable estimates (i.e., estimates with confidence intervals that effectively spanned from 0 to 1) for apparent survival for each of our sites. Therefore, we summarized recapture data and calculated a measure of minimum annual apparent survival by dividing the number of American Eels recaptured at a given site that were marked the previous year by the number of eels captured and released at that site the year prior. American Eels that were recaptured would have thus had to have survived, been subsequently captured, retained their tag, and remained at the original capture location. Because American Eels would be likely to move between sampling events, estimates calculated through this method will likely be underestimated as the number of eels available for capture would likely be lower than expected. For the first three dam sites, we also calculated an estimate of minimum biennial survival by dividing the number of American Eels recaptured that were marked 2 years prior.

RESULTS

A total of 671 American Eels were captured using electrofishing at the seven lowermost sites surveyed; no eels were captured at dams 8 and 9 (Table 2; Figure 1). Of these, 199 were PIT tagged. The overall median size of captured American Eels was 250 mm TL. Median size increased progressing upstream, from a low of 225 mm TL at the 182nd Street Dam (dam 1) to a high of 721 mm TL at the Strathmore Dam (dam 7) (Figure 2). Median size exceeded 500 mm TL at all locations upstream of the Twin Dams (dam 2).

Capture probability per electrofishing pass varied from a low of 0.29 at the Old Stone Mill Dam (dam 5) site to a high of 0.73 at the Twin Dams site (Table 3). Capture probability per electrofishing pass at the other five sites ranged more narrowly, between 0.39 and 0.59.

American Eel density followed a similar trend (Table 4). Density peaked at the 182nd Street Dam at 0.098 eels/m², before declining to 0.029 eels/m² at the Twin Dams site. Above the Twin Dams site, American Eel density was consistently low, ranging from 0.006 to 0.012 eels/m². American Eel biomass density was relatively consistent among sampling sites, with the exception of the Strathmore site, which had a biomass density of 8.03 g/m²; biomass density elsewhere ranged from 1.83 to 4.56 g/m².

Only one American Eel was marked and subsequently recaptured at a different site. This individual (750 mm) was initially caught and tagged in June of 2015 at the Old Stone Mill Dam (dam 5) and recaptured in July of the same year at the next upstream site, Hodgman Dam (dam 6). Year-to-year recaptures within a site occurred at all sites (Table 5). The highest rate of year-to-year recapture occurred at Strathmore Dam at a minimum annual apparent survival of 0.29. American Eels were marked and recaptured 2 years later at the Twin Dams and the Stone Mill Dam site but not at the 182nd Street Dam despite having the largest overall number of tagged American Eels. Over the course of the study, we observed a total of six silver eels, all in October of 2015. Of these silver eels, five were captured at the 182nd Street Dam and one was captured at the Old Stone Mill Dam.

A total of 1,038 American Eels were caught in eel mops below the first six dams on the Bronx River. Effort was not evenly distributed among the dam sites. During the summer of 2015, citizen science groups monitored the eel mops at the Twin Dams and Stone Mill Dam only once per week, resulting in less effort at those two sites. Eel mop CPUE decreased progressing upstream (Table 6). Of all American Eels caught in eel mops, 92% were caught below the 182nd Street Dam and 99% caught below the Twin Dams. Glass eels were only present at the 182nd Street Dam site, with CPUE ranging from 0.08 in 2016 to 1.17 in 2015. Elvers were present at all sampled sites except for at Old Stone Mill and Hodgman dams, and yellow eels were present at all sites sampled. The CPUE varied considerably year over year, with 2016 catches notably lower than the other two years. Elver CPUE at the 182nd Street Dam ranged annually between 0.55 and 1.5. Elver CPUE was low (≤0.19) at all other sites and years. Yellow eels were captured in mops at all sites. Yellow eel CPUE ranged from 0.35 to 0.59 at the 182nd Street Dam but did not exceed 0.21 at any other site in any year. Due to insignificant CPUE at all other sites, only the 182nd Street Dam and Twin Dams were sampled in 2017. The CPUE at the second dam did not differ significantly between years when the pass was in operation (0.05 for elvers and 0.06 for yellow eels in 2016; 0.04 for elvers and 0.21 for yellow eels in 2017) and when it was not (0.19 for elvers and 0.18 for yellow eels in 2015).

DISCUSSION

Our results suggest that dams significantly impede the movements of American Eels upstream in the Bronx River, although some successful surmounting of dams does occur. This resulted in clear gradients in the abundance and size of American Eels at the first three dams, with decreasing numbers and increasing size moving upstream. This pattern was not evident at dam sites 4–9, in which abundances were low (or absent) and sizes high relative to dam sites 1–3. However, abundances and sizes were approximately equal across dam sites 4–7.

The minimum apparent survival estimates that we observed, which exceeded 0.20 at the Twin Dams and Strathmore Dam sites, indicated high levels of stationariness within the Bronx River. Our estimates of minimum apparent survival represent the number of tagged American Eels that survived from one year to the next and that stayed at their capture location, which was relatively small compared with individual river reaches and did not account for capture probability or potential tag loss. Thus, they represent an underestimate of true survival and stationariness as migration out of the study area would decrease the number of American Eels available for capture. These unavailable eels may actually be alive but out of the capture area. A high degree of stationariness is also indicated by several recaptures of American Eels at the sites where they were tagged and the absence of recaptures between dams. This agrees with the conclusion of Cairns et al. (2004) that individual American Eels tend to remain at sites in dammed rivers over years in contrast with their habits in undammed rivers, where they are able to make routine movements upstream and downstream and across salinity levels.

Although there were clear trends in an upstream direction in abundances and size, these two factors combined to show somewhat equivalent biomass across the seven dams at which we collected American Eels. This pattern could be a result of a diffusion effect, as proposed by Smogor et al. (1995). It also is suggestive of a carrying capacity effect, in which American Eels distribute at dam sites in relation to biomass. However, in order to attain a leveling of biomass among dam sites, movements among dam reaches, though possibly difficult, must be achievable. In 3 years of tagging, we only recovered one American Eel that had moved among segments between dams. This individual, who moved from dam site 5 to dam site 6 in summer 2015 made the expected upriver movement of a yellow eel. The near absence of movements upriver to new dam sites may attest to the difficulty in making such movements given the sizes of dams in this river system. However, the longevity of American Eels may allow more opportunities for such, albeit rare, upriver movements to occur. We would anticipate that silver-stage eels, preparing to enter the sea, would be recaptured downriver from where they were tagged. Although we did observe six apparently silvering eels, only one large individual was marked in its yellow stage and recaptured silvering later the same year and at the same site. We did not recapture any previously marked silvering eels at different sites.

The lack of glass eels and near absence of elvers above dam 1, as well as the sharp decline in density of American Eels <250 mm TL at sites farther upstream, suggests that colonization rates are inherently low in upstream areas. However, the presence of small American Eels (<250 mm TL) at upstream sites does suggest that some pass but not as glass eels.

We did not directly address the mechanisms of upriver American Eel movements past dams in the Bronx River. However, we speculate that some glass eels and elvers may ascend the face of the dam adhering via surface tension. This is supported by the observation of common grackle Quiscalus quiscula apparently feeding on the face of dam 1; common grackles are known to hunt small ascending eels (S. Gephard, Connecticut Department of Energy and Environmental Protection, personal communication). Another possibility is passage through cracks in the base of the dam, where they exist.

These avenues for upriver passage would not be available for larger American Eels. We speculate that the relatively low height of some of these dams might allow upriver movement under high flow conditions. The Bronx River is heavily urbanized and has 21% impervious surface in its watershed (Bronx River Alliance 2010); consequently, it behaves as a spate river. It may be that during high runoff American Eels can move upriver through almost laminar flows over the shorter of these dams or through vortexes at their ends. If this is true, then the effect of urbanization in generating high punctuated flows may actually assist upriver movements of yellow eels. Thus, it appears that upriver movements of American Eels past dams may occur more frequently in the elver phase and only rarely in the yellow phase.

American Eel densities in the Bronx River were low in comparison with those estimated by Machut et al. (2007) for Hudson River tributaries. Our estimate for below the first dam, the 182nd Street Dam site, was 0.098 eels/m². This contrasts with the first barrier sites in the study by Machut et al. (2007) that had a high of 1.55 eels/m² at Hannacroix Creek and a low of 0.12 eels/m² at Black Creek. For second dams, the 0.029 eels/m² at the Twin Dams site is near the middle of the range found by Machut et al. (2007) of 0.42 eels/m² at Minisceongo Creek to 0.05 eels/m² at Peekskill Hollow Brook. Third dams and dams farther upriver showed high variability in the study by Machut et al. (2007), with some zero values but ranging to 0.27 eels/m². However, American Eel densities in the Bronx River were far greater than estimates for the main-stem Hudson River at 0.003–0.024 eels/m² (Morrison and Secor 2004).

In their study, Machut et al. (2007) found that American Eel demographics were most strongly influenced by barriers and, secondarily, by urbanization. We did not address the overall effects of urbanization. However, we did identify a negative effect of serial dams on the ingress of American Eels manifested by abundance, biomass, and life stage differences. It is clear that American Eels in the Bronx River, like many other diadromous and resident lacustrine fish species in the Bronx River and elsewhere, would benefit from ease of passage via either dam removals or the implementation of effective fish ladders.