Assessment of upstream and downstream passability for eel at dams

Abstract

The American eel (Anguilla rostrata) population has experienced a marked population decline. Habitat loss resulting from dam construction to improve the control and use of freshwater discharge is one of the factors involved. There are some 5600 dams in rivers draining to the St. Lawrence River in Quebec (Canada). Their passability to eels migrating upstream and downstream has been assessed using the Québec Dam Database. Eighteen percent of the dams are used for supplying water and 13% for hydroelectricity, but >50% are used for recreational purposes. Although the majority of the dams are <3 m in height and are made of concrete or earthfill, dams present a great variety of physical characteristics. Passability ranks were assigned to each category of dam based on three assessment criteria: the height of the dam, the materials used in its construction, and its use. Passability to upstream migrants was also assessed from photographs for a subset of dams. The two methods (statistical analysis and the use of photographs) may yield different results, but the two methods were consistent to identify the impassable dams. This analysis shows overall that the problem of passability is more significant for upstream passage than it is for downstream passage.

Introduction

For over 20 years, a decline in the American eel (Anguilla rostrata) population has been observed in the St. Lawrence drainage system in Canada. Many initiatives have been undertaken to restore the stock. One of them, developed by the Department of Fisheries and Oceans Canada (DFO), is a Geographic Information System (GIS) tool to locate and describe watersheds and obstacles to fish movements in the watershed. The tool assesses habitat gains that could result from the construction of a fish-way or canal or an escapement device as well as habitat losses that could result from additional obstacles to fish movements.

In freshwater, eel density decreases progressively according to distance from the ocean (Smogor et al., 1995; Ibbotson et al., 2002; Imbert et al., 2008). Colonization of upper reaches of watersheds is often disrupted by the presence of natural and man-made obstacles. Dams induce full blockages or successive delays in the migration of elvers, young eel, and yellow eel, which results in species rarefaction in the more upstream reaches of watersheds (Larinier et al., 2006). Downstream passage can be slowed, and spawners (silver eel) returning to the Sargasso Sea may be subjected to (in) direct mortality caused by passage through turbines or by the dams' hydraulic head. The same indirect impacts (predation, migratory delay) are expected to occur as during upstream passage (Steinbach, 2001).

For these reasons, dams are one of the primary factors in the disruption of the eel survival and migration success on which action can be taken on the scale of a watershed. Eel's capacity to surmount obstacles is linked to crawling movements out of water on wet substrates (Legault, 1988; Larinier and Travade, 1998; Gillis, 1998a, 1998b, 2000; Van Ginneken et al., 2005).

This study examines the feasibility of assigning temporarily upstream and downstream passability ranks based on the data available in the Québec Dam Database. Once available and incorporated into a geodatabase, the passability ranks will enable managers to locate man-made structures impeding American eel movements and to set a course of actions where access has been partially or entirely blocked. As future steps, it would be valuable to managers to request an economical study to evaluate the costs required to increase passability at dams blocking access in lower reaches of watershed, e.g. where habitat loss is greater.

Background: upstream passability

The knowledge base for upstream passability of European eel is derived from the work assessing the impacts of the ONEMA protocol developed by Pierre Steinbach in the Loire catchment area in France. The Steinbach scoring key is widely used in the assessment of European eel upstream passability. It has been used as a premise for all of the work dealing with eel upstream passage and passability of hydraulic structures (Larinier et al., 2006; Steinbach, 2006; Leprevost, 2007; Muchiut et al., 2007; Hoffmann, 2008; Lelièvre and Steinbach, 2008; Steinbach, 2009). Recently, ONEMA developed a diagnostic tool to assess upstream passability of fish (Baudoin et al., 2014). Nevertheless, the general principles of Steinbach's work still remain valuable in the case of eel upstream passability. Before establishing any criterion to assess eel passability potential using a scoring key, certain criteria should be met. According to Steinbach (2006), the passage route selected is determined by three orientation factors: the dominant flow, or the most active channel; upmost point of discharge, i.e. the end-point migrants aim to pass the obstacle and the proximity of the bank as eels can use the bank to crawl if the bank constitutes the most favourable route (Steinbach 2006).

Steinbach (2006) assessed upstream passability based on six classes (Table 1), ranging from eliminated obstacle (Class 0) to impassable dam (Class 5). The assessment involves the weighting of different variables, using a scoring key that penalizes (+) or facilitates (−) an eel's upstream passage surmounting a dam. A review of all the physical factors listed and which are considered as having a positive or negative impact on passability will help managers to achieve an accurate classification of dams with intermediate upstream passability. The waterfall height or the difference between the upstream water level and the downstream water level (Muchiut et al., 2007), accurately represents the impact of an obstacle for eel during downstream migration (Steinbach, 2009). The type of material used to build a dam, or roughness (Steinbach, 2006, 2009), also influences the structure's degree of passability for eel. Thus, the rougher the surface and material of the face to be surmounted, the easier it will be for eel to crawl up the face (Steinbach, 2006; Muchiut et al., 2007). Steinbach (2006, 2009) also selected the profile of the structure, i.e. its slope in terms of its height–length ratio in the watercourse, as a decisive assessment criterion for the passability of an obstacle by eel. Finally, the presence of favourable laterally sloped banks can help the species surmount obstacles (Table 1).

Table 1.

Description of criteria used to assess upstream passability according to Steinbach (2009, modified from Steinbach, 2006).

Assessment criterion	Criterion description	Score
Waterfall height (m)	<0.5 m	1
	0.5–1 m	2
	1–2 m	3
	>2 m	4
Structure profile	Verticality of downstream dam face (slope >5H/1L)a and/or high sharp slope rupture	1
	Highly inclined part (between 5H/1L and 3H/2L)a and/or sharp slope rupture	0.5
	Downstream face inclined (slope between 3H/2L and 1H/5L)a	−0.5
	Downstream face with gentle slope (slope ≤1H/5L)a	−1
Roughness	Impervious and smooth materials	1
	Downstream face rough (joints, grooves, and moss)	−0.5
	Downstream face highly rough (rocky, heterogenous, and vegetated)	−1
Bank effect	Bank with favorable slope (lateral slope of transition areas near the bank)	−0.5
Diversity	Presence of an easier passage route	−0.5
	Presence of a much easier passage route	−1
Classes of passability	Assessment of Upstream Passability
0	Free passage
1	Passable without apparent difficulty
2	Passable at times
3	Passable with some difficulty
4	Nearly impassable
5	Impassable

^aSlope is calculated as the height (H) to length (L) ratio of the dam.

Table 2.

Grouping of material type and dam use for dams listed in the Québec dam database.

Variable	Type according to the database	Category	Database correspondence
Materials	Concrete-gravity, filled concrete-gravity, arch (vault) dam, concrete skin plate, concrete buttresses	Concrete	1
	Timber cribs (earthfill, rockfill)	Timber crib	2
	Sheet pile skin plate, sheet pile cribs, steel sheet pile	Sheet pile	3
	Wood buttresses	Wood buttresses	4
	Free weir (concrete)	Concrete free weir	5
	Rockfill free weir	Rockfill free weir	6
	Rockfill	Rockfill	7
	Rockfill (concrete flow retarding facing), rockfill (earthfill flow retarding facing), rockfill (skin plate), rockfill (core)	Impervious rockfill	8
	Earthfill	Earthfill	9
Use	Agriculture, water intake, fire reserve, old river driving	Supplying water	1
	Environmental purposes, historic site, sewage pond, pisciculture, others or unknown	Others	2
	Wildlife	Wildlife	3
	Flood control	Flood protection	4
	Hydroelectricity	Hydroelectricity	5
	Recreation and vacational facilities	Recreational	6

Background: downstream passability

The downstream passability of an obstacle for fish depends on the swimming capacity and the behaviour of the targeted species as well as on the hydraulic and physical conditions of water intakes (Larinier, 2000). Passage through a turbine causes significant (in) direct mortalities (Larinier and Dartiguelongue, 1989). The works of Larinier et al. (2006), Muchiut et al. (2007), and Lelièvre and Steinbach (2008) deal with the downstream passability of structures. Many others worked on mortality, focusing on turbine passage rather than on survival of passing through a waterfall/weir. However, Lelièvre and Steinbach (2008) assessed five defined passability classes based on mortality rate during downstream passage at obstacles (micro-hydropower dam, and sill or dam) on the Sioule River: (i) <3%; (ii) 3–10%; (iii) 10–25%; (iv) 25–50%, and (v) >50% of mortality rate. The mortality rate takes into account various elements, i.e. the portion of the spill flow from structures during downstream migration periods, the average size of migrating fish, the space between the bars of the screens upstream of the turbines, the presence of outlets located upstream of screens, and the characteristics of the turbines (nominal head, throughput, type of turbine, runner diameter, rotational speed, number of blades, and number of turbines). The presence of systems for safer passage (e.g. fine screens, “fish-friendly” intakes) and turbines throughput at dams do affect the downstream passability. Nevertheless, given this information was not systematically available in the Québec Dam Database, the presence of turbines was not selected among the criteria for establishing downstream passability ranks in this study. Thus, the Québec Dam Database was conceived by the Centre d'expertise hydrique du Québec to manage and regulate the Québec water regime by monitoring dam security and integrity. During downstream passage at dams, the waterfall height affects the mortality rate of fish. Experiments have shown that significant damage occurs (with injuries to gills, eyes, and internal organs) when the impact velocity of the fish on the water surface in the downstream pool exceeds 16 m s⁻¹, whatever its size (Bell and Delacy, 1972 cited in Larinier, 2000). In a situation of free fall (e.g. free from the water column), a fish over 60 cm in length reaches this critical velocity (16 m s⁻¹) from a waterfall height of 13 m (Larinier and Travade, 1999,, 2002). Beyond this limit injuries may become significant and mortality will increase rapidly in proportion to the drop (100% mortality for a drop of 50–60 m). Mortality in silver eels is generally higher than other species given their length (Larinier, 2000). Given it is also safer for a fish to migrate through a dam where depth at the foot is adequate, the presence of a pool of considerable depth downstream of the dams would minimize damage during downstream passage (Travade, 2005).

Methods

The Québec Dam Database contains both numerical and categorical variables. The variables selected for statistical analyses are dam height, dam storage depth, structure length, impounding capacity, calculated flow, material type (dam type variable; Table 2), and dam use (use variable; Table 2). Of the numerical values, only the dam height was available for each of the dams being studied.

The flow was calculated based on the area drained upstream of the structure using the following equation where MAF is the mean annual flow (Caissie and Robichaud, 2009) and DA the drainage area. The coefficients a and b are derived from a linear relation between these two log-transformed variables (log–log). Since the values of a and b vary according to the hydrological region to which the considered river belongs, they are applied according to the hydrological region to which the dam belongs.

Relationships among the physical characteristics of the dams and dam classification

To properly identify the relationships existing among the physical characteristics of the dams, a principal component analysis (PCA) was conducted. This method makes it possible to determine which variables form a group that is independent of another group of variables (Tabachnick and Fidell, 2001). The PCA was conducted only on numerical values to determine the principal components. The data were log-transformed prior to analysis (log₁₀). Afterwards, a hierarchical cluster analysis was conducted using Euclidean distance as a measure of dissimilarity and the average junction algorithm as a clustering method (Quinn and Keough, 2002; Garson, 2010). Groups of dams whose characteristics present a degree of similarity were defined in order to assign to each group an upstream and downstream passability rank for eels.

Given the large number of dams listed in the database, it became difficult to visually determine the number of clusters obtained solely through the examination of graphs. Groups were selected according to validation indices and various cluster trials. Three indices were selected for the validation, based on the recommendations described in Systat 12 (2007): pseudo F (CHF), pseudo T-square (PTS), and root-mean-square standard deviation (RMSSTD).

Establishment of upstream passability ranks

The criteria used for European eel to assess upstream passability were used to establish the passability grid for the upstream passage of American eel. Nevertheless, when assessing the feasibility of using Steinbach's criteria (2006, 2009), certain of these criteria were modified and their scores adjusted to make them representative of the characteristics of obstacles encountered in the watersheds under this study. Moreover, it should be noted that the presence of passage facilities (ladders, fish passes, or others) was not selected among the criteria for establishing the upstream passability ranks in this study, since this information was lacking in the database.

In this study, the upstream passability ranks are semi-quantitative, do not take into account the presence or absence of passage facilities, and contain four levels of difficulty for upstream passage: (i) Rank 1: obstacle passable without apparent difficulty (free passage ensured at all flow levels); (ii) Rank 2: obstacle passable at times or with some difficulty, but significant delays encountered or only certain stages surmount it (elvers and young eels); (iii) Rank 3: obstacle nearly impassable for all stages (significant impact in average runoff conditions; passage possible only under exceptional conditions); and (iv) Rank 4: obstacle impassable (passage impossible, even under exceptional conditions).

Photo-interpretation analysis

The Web site of the Québec Dam Database (CEHQ, 2014) also provides a photograph of each dam. The photo was not initially taken to assess passability for a fish species nor was it representative of the average water conditions.

Photographs for a random subsample of dams were used to assess the validity of ranks based on the statistical analyses and to assess additional upstream passability assessment criteria. A total of 1078 photos (19.8% of the dams) were examined. For each of the dams subjected to photo-interpretation, a comparison was made with the rank assigned based on cluster analysis.

Two criteria were assessed during the photo-interpretation analysis. These were Slope (height–length ratio), and Bank effect, that is, whether or not the banks provide a slope and/or a material conducive to eel crawling. These have been assessed as far as possible according to the most attractive path of passage through the structure.

Establishment of downstream passability ranks

Most of the assessment criteria used to assess downstream passability for European eel are not currently documented in the Québec Dam Database (e.g. presence of turbines and type, passage facilities, pool with considerable depth downstream of the dams).

The downstream passability ranks in this study were semi-quantitative and did not take into account whether turbines were present (information not available). As for upstream passage, the downstream passability of dams was assessed under four levels of difficulty: (i) Rank 1: obstacle passable without apparent difficulty (free passage ensured at all flow levels); (ii) Rank 2: obstacle partially passable, but significant delays encountered; (iii) Rank 3: obstacle almost impassable because of significant delays encountered, and significant mortalities; and (iv) Rank 4: impassable obstacle (100% mortality).

Results and discussion

Dam distribution and characteristics

The number of obstacles listed in the Québec Dam Database was 5443 dams. Structures made of concrete, timber cribs and buttresses, concrete and rockfill free weirs were mainly used for recreational purposes (26–47%) and for supplying water (24–44%) (Figure 1). Rockfill—free weir—type dams were used for recreation (48%) and wildlife (29%). Impervious rockfill structures were mainly used for hydroelectricity (54%), but also for recreational (26%) purposes. Sheet pile structures were useful for hydroelectricity (48%) and wildlife (39%). Finally, earthfilled structures mainly had recreational uses (61%).

Figure 1.

Mean dam height according to material type and dam use.

Relationships among the physical characteristics of the dams and dam classification

The PCA indicated that the variables dam height, dam length, and dam storage depth were strongly correlated with axis 1, whereas the calculated flow and impounding capacity were strongly correlated on axis 2 (Figure 2). Therefore, dam height (“obstacle size” component) and impounding capacity (“obstacle attraction” component) were retained and used for subsequent cluster analyses.

Figure 2.

Principal component analysis on numerical values.

An initial cluster analysis of dams was conducted using the variables dam height, impounding capacity, and the dummy variables created for the types of material and uses. Validity graphs and tests of various cluster scenarios suggested three to nine clusters. A distance of ∼0.65 (Figure 3a) was retained that resulted in seven groups, including one group that included most of the dams (Figure 3a and b). Selecting seven clusters made it possible to isolate the dams that presented particular characteristics such as for clusters G1, G4, and G7, and made it possible to reveal group G2, which was not apparent at a lower level of similarity. Nevertheless, the cluster analysis did not provide satisfactory results for upstream passability assessment as two of the seven groups (groups 3 and 5) alone contain nearly all of the dams—5304 of a total of 5443 dams The groups obtained from hierarchical cluster analysis will be used in the assessment of downstream passability as downstream passage is particularly linked to the type of dam use (Figure 4).

Figure 3.

(a) Hierarchical cluster analysis of dams on 15 binary variables and two numerical variables. (b) Validation index of numbers of clusters according to three criteria. Indices: root-mean-square standard deviation (RMSSTD), pseudo F (CHF), PTS.

Figure 4.

(a) Distribution of dam use. (b) Material type as a function of groups established by the hierarchical cluster analysis.

Description of the two criteria selected for the assessment of upstream passability

In light of the results and in accordance with the literature review, two variables of the database were selected to establish an upstream passability rank, i.e. dam height and material type. The data for these two variables were available for all dams. These two variables correspond to two of the five assessment criteria in Steinbach's scoring key (2006, 2009), i.e. waterfall height and surface roughness. The two variables were first grouped into four categories, each category corresponding to a rank from lowest to highest (dam height; Table 3) and from most permeable to most impervious (type of material; Table 4).

Table 3.

Passability rank according to dam height classes and correspondence between dam storage depth and dam height for dams listed in the Québec Dam Database.

Assessment criterion	Dam height class (m)	Mean height (m)		Difference between dam height and storage depth (m)	Rank
		Dam height (m)	Storage depth (m)
Dam height	<1 m	0.90	0.80	+0.10	1
	≥1 and <2 m	1.51	1.09	+0.42	2
	≥2 and <3 m	2.40	1.73	+0.67	3
	≥3 m	7.36	4.80	+2.56	4
Passability rank		Assessment of upstream passability
1		Passable without apparent difficulty
2		Passable at times or with some difficulty
3		Nearly impassable
4		Impassable

Table 4.

Passability rank established for types of material used in dams listed in the Québec Dam Database.

Assessment criterion	Criterion description	Rank
Material type	Downstream face very rough and permeable (free weir rockfill dam, rockfill dam)	1
	Downstream face rough, vegetated and semi-permeable (timber cribs [earthfill, rockfill]; buttresses)	2
	Downstream face smooth with spillway or semi-impervious (free weir concrete, sheet pile skin plate, steel sheet pile)	3
	Impervious materials (rockfill core, concrete skin plate, concrete buttresses, concrete-gravity, filled concrete-gravity, arch (vault) dam, rockfill earthfill flow retarding facing, rockfill skin plate, rockfill concrete flow retarding facing, earthfill)	4
Passability rank	Assessment of upstream passability
1	Passable without apparent difficulty
2	Passable at times or with some difficulty
3	Nearly impassable
4	Impassable

Since waterfall height was not available for all the dams listed in the Québec Dam Database, the dam height was used to assign a passability value in this study. For grid categories determination, a correspondence had to be determined between the classes of waterfall height and dam height. The differences between the two variables were <0.5 m for dam heights under 2 m and a little over 0.6 m for dam heights between 2 and 3 m (Table 3). These differences justified the intervals considered for dam height. Thus, a rank 4 was assigned to dams with height exceeding 3 m.

The dams made of only impervious materials (including concrete, 46% of dams) were considered as those having the greatest impact on upstream passage (Table 4).

Description of dam groups based on the two criteria selected for upstream passability

The dams were then subjected to a hierarchical cluster analysis based on these two criteria, i.e. the height classes (Table 3) and the material classes (Table 4). Thus with regard to upstream passability for eel, six groups were formed (Figures 5 and 6 and Table 5). By associating upstream passability ranks for the classes of dam heights and for types of material with the six dam groups obtained from analysis, an overall upstream passability rank was determined.

Table 5.

Hierarchical cluster analysis results for upstream passability.

Group	Dam no.	Height variation (m)	Height rank	Material rank	Final rank for upstream passabiltya	No. of dams by rank
1	474	0.8–2.95 m	1, 2, 3	1, 2	1 or 2 for dams of less than 1 m (depends on material type)	1 (cote/rank 1)
					2, for dams between 1 and 2 m in height	301
					3, for dams of 2 m and more	172
2	274	2–155 m	3, 4	1	3, for dams of <3 m	93
					4, for dams of 3 m and more	181
3	217	3–15.2 m	4	2, 3	4	217
4	3640b	2–214 m	3, 4	4	4	3640
5	26	1.0–2.8 m	2, 3	3	3	26
6	812	0.6–1.98 m	1,2	4	4	812
Passability rank		Assessment of upstream passability
1		Passable without apparent difficulty
2		Passable at times or with some difficulty
3		Nearly impassable
4		Impassable

^aUpstream passability final rank (most severe from the height/material interaction).

^bGroup 4: 3640 dams including 1142 of <3 m in height.

Figure 5.

(a) Hierarchical cluster analysis of dams based on two criteria (height and material classes). (b) Validation index of numbers of clusters according to three criteria. Indices: Root-mean-square standard deviation (RMSSTD), Pseudo F (CHF), and PTS.

Figure 6.

Proportion of dams by material category (from the most permeable to the most impervious) in groups established by the cluster analysis.

The final passability rank was set at that of the highest rank. Thus, the final rank recorded in the database for upstream passability is the higher rank of the two assessment criteria considered, i.e. of the passability rank for the dam height and that for type of material. According to this analysis, only one dam would be passable without apparent difficulty (rank 1), 5.5% would be partially passable (rank 2), 5.3% would be almost impassable, whereas the majority, i.e. 89.1% of the dams, would be entirely impassable.

Validation of upstream passability ranks using photo-interpretation

According to photo-interpretation, dams were classified into rank 2 (partially passable; 30.8%), rank 3 (almost impassable; 26.0%), and rank 4 (impassable; 29.8%) obstacles. Thus, rank assignment using only two variables (height and type of material) did not match consistently those determined from photographs. In fact, except for the dams ranked impassable, the correspondence was low, especially in the case of ranks 1 and 3. This stems from the fact that, besides dam height, the slope of the structures (steep vs. gentle facing) and the nature of the banks (alternative route) greatly affect the assessment of upstream passability. For instance, rockfill free weir dams present a gentle slope and a rough substrate, both favourable to upstream passage even if the structure is >3 m. Although there are disparities in the results between the two methods (cluster analysis, photo-interpretation), the passable dams (rank 1) are still under-represented in both cases. Both methods also made it possible to properly identify impassable dams, with a concordance of 88.5%. Thus, dam height and type of material, or the final passability rank, were useful for identifying impassable dams.

Photos of earthfill dams could not be used to determine passability. For example, one could not determine whether earthfill dams (embankment dam) had a discharge system (e.g. spillway, low-level outlet) allowing the passage of fish. Nevertheless, for the largest hydropower complex dams, earthfill structures were properly identified as dikes. In this case, no upstream passage is possible and an upstream passability rank of 4, i.e. impassable, was assigned to the obstacle. Dams with cribs, buttresses, or dead shores were also problematic for photo-interpretation and thus for the assessment of upstream passability. In fact, these offer a range of passage possibilities, depending on their state and their imperviousness. Certain structures are entirely impervious and thus do not allow any upstream passage, especially when the structure is very high (ranks 3 and 4), whereas others are entirely submerged or eliminated and no longer constitute an obstacle to upstream passage (rank 1). Nevertheless, their classification as passable obstacles with delay, based on height and type of material, appears justified in most cases in the field, even though photos may suggest otherwise.

Thus, photos from the Québec Dam Database Web site did not necessarily include key elements required for the present study, such as photo date (vs. flow conditions), main water intake, upstream view, downstream view, a view clearly illustrating the characteristics of banks, presence of a by-pass route, etc.

Description of the two criteria selected for the assessment of downstream passability

In light of the results and in accordance with the literature review, two variables of the database were selected to establish a downstream passability rank, i.e. dam height and dam use.

As already stated, the limit for the waterfall height for a silver eel was assessed at 13 m. The correspondence with the dam height was established according to dams presenting both height (2.56 m on average for dams with a height exceeding 3 m). Since the storage depth is akin to waterfall height, dams of a height of over 15.5 m were classed impassable (rank 4). Dams whose dam height is less than 15.5 m can therefore be assigned a rank ranging from 1 to 3, depending on the other factors affecting downstream passability. Dams less than 15.5 m in height were assigned rank 2.

Two classes of dam use type are proposed, one hydroelectric and the other including all other uses. Thus, eel passability likely can vary widely among hydro projects depending on project configuration, intake bar rack spacing, presence of bypasses, spillway design, and turbine passage mortality varies with turbine design and operation. Nevertheless, consideration of these two categories is based on the fact that only hydroelectric dams can be equipped with turbines and thereby cause significant eel mortality. A cautionary yet conservative approach was considered appropriate for determining passability ranks as a function of dam usage. Thus no dam was attributed rank 1 based on the type of use. Rank 2 was assigned to dams serving all purposes but hydroelectricity. While turbines cause direct mortalities to downstream migrants, no information specific to each dam was available for this study. Thus, a rank 3 (and not a rank 4) was assigned to hydroelectric dams. Rank 4 would be attributed only where 100% mortality occurs. These ranks will have to be reviewed when pertinent information becomes available, such as the presence of spillway passage, etc.

Description of dam groups based on two criteria selected for downstream passability

The seven groups of dams obtained from the initial cluster analysis can be directly used to determine downstream passability ranks following the literature review and the ranks determined. Table 6 presents the final downstream passability ranks for each of the groups obtained through cluster analysis. According to this analysis, 86% of the dams would be partially passable obstacles causing significant delays (rank 2), and 9.6% of the dams would be nearly impassable because of significant delays and mortalities (rank 3). However, this analysis does not take into account whether or not there are facilities to support downstream passage. The dams would be impassable in only 4.4% of cases.

Table 6.

Results from the cluster analysis for downstream passability.

Group	Dam no.	Height variation (m)	Height rank	Use type	Use rank	Final rank for downstream passabilitya	No. of dams by rank
1	4	3.0–5.0	2	Hydro	3	<15.50 m and Hydro: 3	4
2	23	0.60–3.60	2	Hydro	3	<15.50 m and Hydro: 3	23
3	4560	0.70–13.00	2	Hydro (n = 45) Autres/Others (n = 4515)	2, 3	<15.50 m and Autres/Others: 2	4515
						<15.50 m and Hydro: 3	45
4	1	28.20	4	Hydro	3	≥15.50 m and Hydro : 4	1
5	744	2.00–53.00	2, 4	Hydro (n = 576) Autres/Others (n = 168)	2, 3	<15.50 m and Autres/Others: 2	168
						<15.50 m and Hydro: 3	448
						≥15.50 m and Hydro : 4	128
6	109	9.00–168.20	2, 4	Hydro	3	<15.50 m and Hydro: 3	3
						≥15.50 m and Hydro : 4	106
7	2	155.00–214.00	4	Hydro	3	≥15.50 m and Hydro : 4	2
Passability rank			Assessment of downstream passability
1			Passable without apparent difficulty (free passage ensured at all flow levels)
2			Partially passable, but significant delays encountered
3			Almost impassable because of significant delays encountered, and significant mortalities
4			Impassable (100% mortality)

^aDownstream passability final rank (more severe from the height × use interaction).

Apart from the waterfall/dam height, the other variables of interest to be taken into account when assessing downstream passability and which are not listed in the Québec Dam Database are the presence of passage facilities and the presence of turbines (type, diameter of runner, rotational speed, and throughput) at hydroelectric dams. Moreover, no data on hydrological conditions are available, either in the database or in the photographs. Hydraulic parameters such as main flow attraction and current velocities upstream and downstream of the structures would have been useful for the assessment of passability.

Conclusion

The dams listed in the Québec Dam Database present a great variety of physical characteristics and are used for all kinds of purposes. Although the Québec Dam Database was not designed to assess passability for aquatic wildlife, upstream and downstream passability ranks were temporarily assigned to these dams. Two variables were used to establish a chart showing passability during the period of both upstream and downstream migration for eel. These were dam height and type of material for upstream passage and waterfall height and type of use (hydroelectric production vs. other usages) for downstream passage.

For eel migration and survival, it appears that upstream passage is more problematic than downstream passage. Used in parallel with presence/absence accounts or density records of eels and incorporated into the management GIS tool (in progress), passability ranks will make it possible to identify dams that are problematic for eel movements, calculate the amount of habitat loss, and locate and establish intervention priorities for eel recovery.

This study demonstrated that a preliminary eel passability assessment at dams can be made using standard database variables not designed a priori for eel assessment, and this, without any cost related to field validation.

In terms of applied management and tangible solutions, the next steps would be to conduct an economical study to establish the costs that would be required to increase passability at dams blocking access in lower reaches of watershed, e.g. where habitat loss is greater.

Funding

The work was supported through SARCEP funding (DFO).

Acknowledgements

The authors thank the Québec Dam Database (e.g. Centre d'expertise hydrique du Québec, CEHQ) who provided access to the database of the dams; Gontrand Pouliot, Serge Proulx, and the staff of the Fish Habitat Management Information System (FHAMIS-DFO) who assisted with preliminary analyses; Yves Mailhot (MRNF), and Jean-Guy Jacques (MPO) for their comments on an earlier version of the text, as well as the three reviewers contacted under the ICES Journal of Marine Science process.

Author notes

Handling editor: Caroline Durif