Ecology of Freshwater Fish 2013 Ó 2013 The Authors. Ecology of Freshwater Fish published by John Wiley & Sons Ltd. ECOLOGY OF FRESHWATER FISH
Morphology and life history of the Great Slave Lake ciscoes (Salmoniformes: Coregonidae)
Andrew M. Muir1,2, Paul Vecsei3, Michael Power4, Charles C. Krueger2, James D. Reist5 1Department of Fisheries and Wildlife, Michigan State University, East Lansing, MI USA 2Great Lakes Fishery Commission, Ann Arbor, MI USA 3Golder Associates, Yellowknife, Northwest Territories, Canada 4Department of Biology, University of Waterloo, Waterloo, Ontario, Canada 5Fisheries and Oceans Canada, Winnipeg, Manitoba, Canada
Accepted for publication August 15, 2013
Abstract – The taxonomy of the North American ciscoes (Salmoniformes: Coregonidae) remains unresolved. We provide the first comprehensive description of the Great Slave Lake ciscoes. Our analysis supports the hypothesis that the Great Slave Lake cisco complex includes at least two nominate species (Coregonus artedi and Coregonus sardinella) and an adfluvial C. artedi morph that is distinct from its lacustrine conspecific in terms of life history, morphology, age, growth and mortality. Coregonus sardinella has previously been identified from Great Slave Lake, but we provide the first comprehensive description of this species in the lake and confirm a significant range extension for the species. The lacustrine C. artedi differs little from descriptions throughout its range. In addition to these three ciscoes, linear phenotypic traits, gillraker number and morphology, and growth data support the possible occurrence of two other, less-distinct morphs, the big-eye cisco and a shortjaw-like morph Coregonus zenithicus. Although the big-eye morph was not identified by body shape and linear phenotypic measures, it was visually identified on the basis of differences in traditional phenotypic proportions, such as orbital length, paired fin lengths, head and gillraker morphology expressed as thousands of standard length and showed different age and growth structure compared with the other lacustrine cisco morphs. Coregonus zenithicus was distinguished visually and by a statistical model of linear phenotypic traits as well as by gillraker number and morphology. Identifying, characterising and managing locally adapted cisco morphs that reflect important ecological and bioenergetic linkages are critical to conserving the ecological integrity of northern ecosystems.
Key words: taxonomy; phenotype; morphotypes; geometric morphometrics; species at risk
been studied (Bernatchez et al. 1991; Turgeon et al. Introduction 1999; Turgeon & Bernatchez 2001a,b, 2003; Turgeon The ciscoes (Salmoniformes: Coregonidae) have radi- 2002), the phenotypic and life-history divergence of ated into complexes of closely related species, life- sympatric ciscoes in of North American lakes is less history types and ecological variants throughout well known. Similar challenges with resolving coreg- their holarctic distribution (McPhail & Lindsey 1970; onine taxonomy and evolutionary history occur in Sv€ardson 1979; Bernatchez 2004; Hudson et al. Europe (Kahilainen et al. 2005, 2007; Ohlberger 2007). Weak genetic differentiation, incomplete et al. 2008; Helland et al. 2009) and Asia (Politov reproductive isolation and strong plastic responses to et al. 2004; Sukhanova et al. 2012). environmental gradients make taxonomic classifica- A further complication to resolving cisco taxon- tion of coregonines challenging. Although the evolu- omy is that commercial exploitation, local extirpation tionary history of the North American ciscoes has (Phillips & Ehlinger 1995), and hybridisation and
Correspondence: Andrew Muir, Great Lakes Fishery Commission, 2100 Commonwealth Blvd., Suite 100, Ann Arbor, Michigan 48105, USA. E-mail: [email protected]
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. doi: 10.1111/eff.12098 1 Muir et al. introgression (Todd & Stedman 1989; Todd & Smith thicus and (5) Coregonus sardinella. The two 1992) have acted to reduce diversity in this group of lacustrine C. artedi (groups 1 and 2 above) were fishes. With the exception of Great Bear and Great shown to be synonymous; young lacustrine C. artedi Slave lakes in the Northwest Territories, Canada, underwent ontogenetic shifts in gillraker and body cisco diversity has been altered or supplemented and morphology and habitat use as they aged and grew even replaced by non-native planktivores elsewhere, (Muir et al. 2013). This analysis reduced the number such as the lower Laurentian Great Lakes (i.e., lakes of putative morphs and species to four. The lacustrine Erie and Ontario). Great Slave Lake (61°47′N; C. artedi group was subsequently subdivided into the 113°43′W; Fig. 1) contains an intact assemblage of typical lacustrine C. artedi and a deepwater variant – postglacial fishes, including ciscoes, and offers the the big-eye cisco (Muir et al. 2011), bringing the opportunity to study this diversity in a relatively number of morphs or species to five. Strong pheno- unperturbed ecosystem. In this sense, Great Slave typic variation between the lacustrine and adfluvial Lake serves as a model to understand the historical C. artedi morphs was shown to reflect adaptations to postglacial food web structure in the Laurentian Great their contrasting life histories and habitats (Blackie Lakes and may provide insights and expectations to et al. 2012). In addition, the identification of help guide restoration efforts in such systems (Zimm- C. sardinella in Great Slave Lake by Vecsei et al. erman & Krueger 2009). (2012) was recently confirmed using amplified frag- Our ultimate goal is to understand patterns of cisco ment length polymorphisms (AFLP; J. Turgeon, Uni- diversity within and among North American large versit e Laval, unpublished data). On the basis of lakes and the biotic and abiotic processes shaping these previous studies, five cisco forms were thought that diversity. A first step towards that goal is to to occur in Great Slave Lake including the following: describe the patterns of cisco diversity in Great Slave three nominate species (i.e., C. artedi, C. sardinella Lake (Fig. 1). A preliminary survey (Vecsei et al. and C. zenithicus) and within C. artedi, three morphs 2012) identified five putative ciscoes including (1) a (i.e., typical lacustrine C. artedi, adfluvial C. artedi large morph of Coregonus artedi with a lacustrine and the deepwater big-eye cisco variant). life history, (2) a ‘dwarf’ morph of C. artedi with a The purpose of this study is to provide an updated lacustrine life history, (3) a ‘dwarf’ morph of C. arte- and more complete description of Great Slave Lake di with an adfluvial life history, (4) Coregonus zeni- ciscoes and their life history. Our specific objectives
Fig. 1. Cisco sampling locations and sample sizes (n) in Yellowknife Bay and the East Arm, Great Slave Lake, NT.
2 Cisco diversity in Great Slave Lake, NT, Canada were to compare the following among all five cisco times of the year. On the basis of a visual examina- species and morphs: (1) gross body morphology and tion of the gonads, 70% of fish sampled were mature, phenotypic traits, (2) life-history dynamics and (3) but had not yet spawned; therefore, state of maturity physical resource use. Finally, objective (4) was to did not likely influence the morphological analyses. assess the validity of the species and morphs within the Fish were caught in the lake using <24 h, bottom set, context of the most recent taxonomy (Scott & Cross- multimesh gillnets. Nets were 200 m long 9 1.8 m man 1973). Achieving these objectives will facilitate deep and composed of eight 25-m panels of 12.7-, follow-on comparative studies among perturbed and 25.4-, 38.1-, 50.8-, 63.5-, 76.2-, 88.9- and 101.6-mm unperturbed lakes to test predictions about the processes stretch mesh (i.e., 6.4-, 12.7-, 19.1-, 25.4-, 31.8-, generating and maintaining ecological diversity and 38.1-, 44.5-, 50.8-mm bar length). Fish were caught studies of the functional diversity among cisco morphs. in the two rivers using a dipnet. Morphs were visu- ally identified in the field using characteristics given by Muir et al. (2011), and these groups were used in Methods subsequent statistical analyses. Study site Statistical methods Four lacustrine sites and one riverine site in Yellow- knife Bay and two lacustrine sites and one riverine Statistical analyses (significance level was set at site in the East Arm of Great Slave Lake (61°N, a = 0.05) were conducted using R (2.15.1; R Core Team 113°W) were sampled between 2008 and 2009 2012), SigmaPlot 11 (Systat Software Inc., San Jose, (Fig. 1; Table 1). Sites were sampled between 1 and CA, USA), the Thin-Plate Spline software suite (TPS; 5 times over the 2-year period in an attempt to span State University of New York at Stony Brook; http://life. the range of available cisco habitat (i.e., 0–29 m, 30– bio.sunysb.edu/morp) and MCLUST V.3 implemented 59 m, 60–89 m and >90 m). in R (University of Washington; http://cran.r-project. org/web/packages/mclust/index.html). Nonparamet- ric tests were used when data could not be transformed Fish collections to meet the assumptions of parametric methods. Fish were sampled during autumn because adfluvial cisco can be accessed during September when it Morphology moves into the rivers prior to spawning, and the other morphs are distributed throughout the lake at that Gross body shape of each individual was quantified time. The lake’s inclement climate and remote loca- using geometric morphometric methods (Bookstein tion also limited our ability to sample during other 1989; Rohlf & Bookstein 2003; Zelditch et al. 2004).
Table 1. Location, data (mm/dd/yyyy), depth stratum sampled (m), latitude and longitude (degrees° minutes′ and seconds″), and numbers (n) of cisco collected from Great Slave Lake, Northwest Territories.
Site Basin Date Depth Stratum Latitude (N) Longitude (W) n
Mackenzie Channel Yellowknife Bay 8/24/2009 0–29 62°20′629″ 114°19′187″ 111 Tartan Rapids 9/27/2008 0–29 62°33′530″ 114°13′100″ 50 10/1/2008 0–29 62°33′530″ 114°13′100″ 100 Sub Islands North 10/4/2008 30–59 62°21′961″ 114°21′116″ 13 10/4/2008 30–59 62°21′961″ 114°21′116″ 14 10/17/2008 30–59 62°21′961″ 114°21′116″ 31 Sub Islands South 10/4/2008 30–59 62°21′419″ 114°21′957″ 64 10/9/2008 30–59 62°21′419″ 114°21′957″ 10 8/24/2009 30–59 62°21′419″ 114°21′957″ 50 9/25/2009 30–59 62°21′419″ 114°21′957″ 7 Negus Point 10/17/2008 0–29 62°25′475″ 114°21′030″ 7 10/18/2008 0–29 62°25′475″ 114°21′030″ 24 Beaulieu River East Arm 10/14/2008 0–29 62°20′792″ 113°11′487″ 75 Christie Bay 10/16/2008 30–59 62°29′053″ 111°11′790″ 12 10/16/2008 60–89 62°29′053″ 111°11′790″ 11 10/16/2008 >90 62°29′053″ 111°11′790″ 24 10/17/2008 60–89 62°29′053″ 111°11′790″ 23 10/17/2008 30–59 62°29′053″ 111°11′790″ 26 Red Cliff Bluffs 10/15/2008 30–59 62°21′388″ 111°40′776″ 24 10/15/2008 >90 62°21′388″ 111°40′776″ 18 Outer East Arm 10/14/2008 >90 61°59′448″ 113°92′881″ 0
3 Muir et al.
A calibrated digital image of the left side of each sidered strong if DBICi >150 (Posada & Buckley individual was captured on fresh fish in the field 2004). according to Muir et al. (2012). Sixteen homologous Traditional phenotypic traits were also quantified and four semi-sliding landmarks were digitised on for each individual fish. Twenty-three linear mor- images using TPS and following Muir et al. (2013). phometric measures (Table 2), and nine meristic Briefly, landmark data were used to scale each indi- characteristics (Table 3) were quantified on thawed vidual relative to a consensus form using TPSrelw. specimens according to Koelz (1929), Scott (1960) The 36 partial warp scores (i.e., size-independent and Vuorinen et al. (1993). Linear measurements shape) were entered into an ordination, and the first were made on the left side of the fish using a digital two principal components (PCs) were retained as calliper ( 0.01 mm). Linear measurements were new shape variates and used in subsequent analyses expressed as thousandths of Ls (i.e., [x/Ls]*1000, (see Zimmerman et al. 2009). where x is the linear measurement for a variable and Body shape was compared among morphs using Ls is the standard length). The first left gill arch was a Bayesian cluster analysis, which does not require extracted, fixed in 5% formalin and transferred to a priori assignments of individuals to groups (i.e., 95% ethanol prior to gillraker enumeration. MCLUST; Fraley & Raftery 2009). Two MCLUST Linear phenotypic measures were treated in a pro- models (EII and VII, see Fraley & Raftery 2009 cess that paralleled the analysis of geometric shape for model descriptions) were fit to the body shape data. Due to strong correlations with body size, all data (i.e., PCs 1–4 from an ordination of partial linear measures were size-adjusted using common- warps), and the ‘best’ model representing the most within-group residuals (Reist 1985, 1987). A subset likely number of groups was identified using of morphometric characters (i.e., OOL, PSL, IOW, Bayesian information criterion (BIC). The ‘best’ or GRL and CPL; see Tables 2 and 3) that are known maximum-BIC model assigned individuals to to vary among cisco morphs (see data herein as well groups and quantified uncertainty in group mem- as data in Koelz 1929 and Clarke 1973) were bership. Alternative candidate models were evaluated selected and used to discriminate cisco morphological by BIC differences – the difference for the ith model groups. The first four PCs from an ordination of the was calculated as DBICi = BICi BICmin, where size-corrected, linear phenotypic measures were BICmin was the smallest BIC value among all candi- retained as new variates and used in subsequent anal- date models (Burnham & Anderson 2002; Posada & yses. Scores on the first 2–4 PCs were analysed by Buckley 2004). Evidence supporting model i was con- MCLUST according to the methods described above
Table 2. Linear morphometric variables measured on ciscoes from Great Slave Lake, NT; modified from Koelz (1929); Scott (1960) and Vuorinen et al. (1993); see Muir et al. (2013) for schematic depicting the variables.
Character Acronym Definition
Adipose length ADL Distance from the point where skin and scales meet at the anterior end of the fin to the free posterior margin of the fin Anal length ANL Distance along the horizontal body axis between the origin and the posterior edge of the fin Body depth BDD Vertical distance from the dorsal origin to the ventral surface of the body Caudal peduncle depth CPD The least vertical depth of the caudal peduncle Caudal peduncle length CPL Distance along the horizontal axis of the body between the posterior of the anal fin and the caudal flexure Dorsal height DOH Origin of dorsal fin to the tip of the longest ray Dorsal length DOL Origin of dorsal fin to the posterior edge of the fin beyond the terminal ray Fork Length FRL Tip of premaxilla to the caudal fork with the fin open Head depth HDD Vertical distance through the pupil of the eye from the dorsal surface of the cranium to the ventral edge of the gular region Head length HLL Sum of preorbital, orbital and postorbital lengths Interorbital width IOW Shortest distance of bone between the upper rim of the orbits Lower arch length LAL Length from the start of the lower arch to the base of the middle gillraker Lumbar length LUL Distance along the horizontal body axis between the end of the dorsal fin and the origin of the anal fin Maxillary length MXL Anterior point of premaxilla to posterior end of the maxilla Maxillary width MXW Greatest width along the maxillary Middle gillraker length GRL Length of the gillraker on the ceratobranchial–epibranchial joint on the first arch Orbital length OOL Distance between anterior and posterior fleshy margins of the orbit Pectoral fin length PCL Extreme base of outermost ray to farthest tip of fin Pelvic fin length PVL Extreme base of outermost ray to farthest tip of fin Postorbital length PSL Posterior fleshy margin of the orbit to posterior bony margin of the operculum Premaxillary angle PMA Angle between the horizontal axis of the head and the premaxilla Preorbital length POL Tip of the premaxilla to the anterior fleshy margin of the orbit Standard length STL Tip of the premaxilla to the caudal flexure Trunk length TTL Distance along the horizontal body axis between the posterior margin of the operculum and the origin of the dorsal fin
4 Cisco diversity in Great Slave Lake, NT, Canada
Table 3. Meristic variables enumerated on cisco collected from Great Slave Lake, NT; modified from Koelz (1929) and Vuorinen et al. (1993); see Muir et al. (2013) for schematic depicting the variables.
Character Acronym Definition
Anal rays ARC All rays in the anal fin including rudimentary rays Dorsal rays DRC All rays in the dorsal fin including rudimentary rays; anterior fin rays were excluded from counts unless they were 2/3 the length of the longest ray; when the shortest ray was split at the base, it was counted as a single ray (for all ray counts) Lateral line scales LLS First pored scale touching the pectoral girdle to last scale of the body on lateral line. If scales were missing pockets from scales were counted Lower gillrakers LGR Number of gillrakers, including all rudiments, on the first, left ceratobranchial Pectoral rays PRC All rays in the left pectoral fin Pelvic rays VRC All rays in the left pelvic fin Scales above the lateral line ULS A single column of scales enumerated from the origin of the dorsal fin to the lateral line Suprapelvic scales SPS A single column of scales enumerated from the axillary process to the lateral line Upper gillrakers UGR Number of gillrakers, including all rudiments, on the first, left epibranchial including the raker on the ceratobranchial–epibranchial joint
for the body shape data. The best model was selected instantaneous rate at which Ls approaches L∞), plus by BIC and used to assign individuals to groups. additive error (e) (von Bertalanffy 1938). Age-0 and With the exception of gillraker number, most age-1 juveniles were underrepresented in the samples; meristic characters were minimally informative in therefore, L0 was fixed to the measured mean length characterising groups (i.e., low among group varia- at hatch (i.e., 10.69 mm) for laboratory-reared Lake tion). Therefore, these data were omitted from further Superior C. artedi (Oyadomari & Auer 2007). analyses and used for descriptive purposes only. Gill- Annual growth rate (x) was estimated by L∞*K raker distribution was compared among morphs (mm year 1; Gallucci & Quinn 1979). because gillraker number is traditionally used as a Four nested models were fit to length-at-age data primary trait for differentiating among coregonine to compare growth among morphs: (1) General: sep- taxa (Scott & Crossman 1973; Smith & Todd 1992; arate parameter estimates for each morph; (2) Com- Todd & Smith 1992). A Kruskal–Wallis one-way mon L∞: constant L∞ among morphs, but varying K; analysis on ranks tested for differences in gillraker (3) Common K: constant K, but varying L∞; and (4) distribution medians among cisco morphs, and Common: the same parameter estimates for all Dunn’s test was used for all pairwise comparisons morphs combined. The ‘best’ model was selected by due to unequal sample sizes. Akaike information criterion (AIC) (Akaike 1973), and a likelihood ratio (Bates & Watts 1998) tested the best model against the Common model (i.e., one Life history growth curve for all morphs combined). Describing seasonal variation in life history was Annual survival (S) was calculated for the adfluvial beyond the scope of this study as sampling was lim- and lacustrine C. artedi morphs, and for C. zenithi- ited to the autumn. Fish age was estimated using an cus as: e Z, where the instantaneous total mortality otolith crack-and-burn method (Muir et al. 2013). A (Z) was estimated for each morph by the slope of a Kruskal–Wallis one-way analysis on ranks tested for linear regression through the descending limb of an differences in age distribution medians among the otolith-based catch curve (Ricker 1975). Small sam- five putative cisco groups, and Dunn’s test was used ple sizes within many of the older age classes pre- for all pairwise comparisons due to unequal sample vented the calculation of S for the big-eye cisco and sizes. C. sardinella.AnF-test for the equality of regression Growth of cisco morphs was quantified by fitting a slopes was used to detect differences in S among series of von Bertalanffy models to length-at-age data morphs (Zar 1999). (Ogle 2012). The original model parameterisation was used: Physical resource use Kt E½Lsjt ¼L1 ðL1 L0Þe þ e; Minimum and maximum water depths (m) for each net set were recorded using an echo sounder. A Krus- where the expected standard length (Ls) at time (t)is kal–Wallis one-way analysis on ranks tested for dif- a function of the asymptotic mean length (L∞; theo- ferences in median depth of capture between cisco retical mean length to which a fish would grow if groups. Dunn’s test was used for all pairwise compar- time permitted), the mean length at time zero (L0; isons. Catch per-unit-effort (CPUE) was standardised i.e., hatch) and the growth coefficient (K year 1; i.e., to kg of fish·net 1 24 h 1 and considered a relative
5 Muir et al. index of abundance among depth strata. A Kruskal– Four of the five next best models also identified three Wallis one-way analysis with Dunn’s comparisons groups (243 ≤ DBIC ≤ 254), indicating strong sup- tested for differences in CPUE among morphs within port for three geometric body shape groups. Two of depth strata (0–29; 30–59; 60–89; >90 m). the three groups identified by the model conformed Buoyancy was used as one physiological indicator to our field taxonomic assignments as well as groups of habitat and prey use. Per cent buoyancy accounts previously identified by Vecsei et al. (2012). One of for differences in density and relative quantity of soft these groups consisted of 99% of fish identified as versus hard tissues that affect the specific gravity of adfluvial C. artedi in the field, and the second group the fish (Alexander 1972) and is positively correlated consisted of 93% of fish identified as C. sardinella with body lipid content and depth of capture in many in the field (Fig. 2). The third group identified by fishes (Zimmerman et al. 2006). High body lipid con- body shape contained the greatest number of speci- tent is a more energy-efficient adaptation than the mens and the most variation along both PC1 and swim bladder for facilitating vertical migration in PC2. This group contained specimens that were iden- deepwater fishes (Alexander 1972, 1993). Many tified in the field as lacustrine C. artedi, big-eye Great Lakes ciscoes undergo diel vertical migration cisco and C. zenithicus. The body shape group struc- (DVM) to prey on Mysis diluviana (hereafter Mysis); ture identified by the model was consistent with our high per cent buoyancy is indicative of this behaviour expectation based on field and laboratory observa- (Eshenroder et al. 1998; Clemens & Stevens 2003). tions that adfluvial C. artedi and C. sardinella could Per cent buoyancy (B) was calculated by the follow- easily be identified on the basis of body shape alone. ing equation: In contrast to the body shape model, four groups were identified on the basis of the selected linear B ¼ððWa WwÞ=WaÞ 100; phenotypic measures (MCLUST; DBIC = 1.60; Fig. 3). The next best model, which was virtually where Wa is weight of fish measured in air and Ww is identical to the first, also identified four groups. All the weight of the fish measured in water. Weights four groups were consistent with groups previously were measured ( 1 g) using a Pesola spring scale identified by Vecsei et al. (2012). Groups 1–4 con- (Jennings 1989). To measure Ww, an incision was sisted of 80%, 65%, 77% and 97% of fish identified made on the left side of the fish, just below the lat- in the field as adfluvial C. artedi, C. zenithicus, eral line, and extending from the anterior to the pos- lacustrine C. artedi and C. sardinella, respectively. terior of the gut cavity. The swimbladder was The selected linear morphometric measures were punctured; care was taken not to damage the internal insufficient to separate the big-eye C. artedi morph – organs. Once the fish was suspended in water, the 97% of big-eye cisco were grouped with the adfluvial remaining air was forced out of the swimbladder and C. artedi contrary to our expectation that they would gut cavity so that it did not bias the measurement. group with lacustrine C. artedi. An ANCOVA with standard length (Ls) as the covari- Median gillraker number, a character typically used ate tested for differences in buoyancy among cisco to differentiate coregonines, differed among the mor- morphs. This analysis was restricted to fish with a phological groups (H = 133.05; d.f. = 4; P < 0.001; weight in water >2 g due to inaccuracies in measur- Fig. 4). The C. zenithicus group had fewer gillrakers ing weights <2 g with spring scales. The interaction than all other cisco groups (all Dunn’s Q > 5.05; all term was included to test for homogeneity of slopes. P < 0.05). Lacustrine C. artedi had more gillrakers A lack of homogeneity of slopes necessitated a size than the adfluvial C. artedi (Dunn’s Q = 3.69; correction; so, per cent buoyancies were adjusted P < 0.05). using a regression technique where residuals from the relationship between per cent buoyancy and L for s Life history each of the five morphological groups were retained as new variates (Reist 1985). Residual per cent buoy- Age structure differed among the five putative cisco ancies were compared among groups using a Krus- groups (H = 178.25; d.f. = 4; P < 0.001). Adfluvial kal–Wallis one-way analysis on ranks with Dunn’s C. artedi was most divergent among the morphs method of multiple comparisons. having a narrow age distribution (range = 2–9 years; median = 4 years). By comparison, the other morphs had broad age distributions and were long-lived: lacus- Results trine C. artedi (range = 1–33 years; median = 6 years); C. zenithicus (range = 2–22 years; median = 7 Morphology years); big-eye cisco (range = 3–20 years; median = Three groups were identified on the basis of geomet- 8 years); and C. sardinella (range = 3–26 years; ric body shape (Fig. 2; MCLUST; DBIC = 259). median = 10.5 years). The median age of adfluvial
6 Cisco diversity in Great Slave Lake, NT, Canada
Fig. 2. Ordination of principal component scores (n = 508) for 36 partial warps defining geometric body shape for coregonine ciscoes in Great Slave Lake, NT. Three groups were identified on the basis of size-independent body shape with no a priori group assignments. The dot and line illustrations show the thin-plate spline deformation plots representing the shape variation along the two axes.
Fig. 3. Ordination of principal component scores (n = 616) for size-corrected orbital length, postorbital length, interorbital width, middle gillraker length and caudal peduncle length for ciscoes in Great Slave Lake, NT. Four groups were identified on the basis of the linear phe- notypic measures with no a priori group assignments.
7 Muir et al.
(a)
(b)
Fig. 4. Gillraker frequency distributions for the five cisco morphs from Great Slave Lake, NT; (a) differs from adfluvial Coregonus artedi; (b) differs from adfluvial C. artedi, lacustrine C. artedi, big-eye C. artedi and Coregonus sardinella (Dunn’s pairwise multiple compari- sons; all P < 0.05). Images: A.M.M.
Table 4. Results from four nested von Bertalanffy models fit to length-at- C. artedi was lower than the median age for all other age data for ciscoes from Great Slave Lake, NT: (1) General: separate > < parameter estimates for each morph; (2) Common L∞: constant L∞ among ciscoes (all Q 4.50; all P 0.05). morphs, but varying K; (3) Common K: constant K, but varying L∞; and (4) Separate growth models for each morph best fit Common: the same parameter estimates for all morphs combined. L∞ is the length-at-age data (AIC = 4825; Table 4); the theoretical length to which the fish would grow if permitted to grow infinitely old; K is the instantaneous rate at which length approaches L∞. ∞ other models (i.e., Common, Common L and Com- The ‘best’ model was selected by Akaike information criterion (AIC); ΔAIC mon K) were not supported (all ΔAIC >>>2; Burn- is the AIC difference between a candidate model and the general model. ham & Anderson 2002). von Bertalanffy growth RSS is the residual sum of squares; d.f. is the degrees of freedom; MS is model parameters varied among the five morphs the mean square error; and the F-statistic (F) and associated significance value (P) are given for each model. (F8,499 = 57.29; P < 0.001; Fig. 5). The two mor- phs that grew the fastest – adfluvial C. artedi (x = Model RSS d.f. MS FP AIC ΔAIC 112.4 mm year 1) and big-eye cisco (x = 54.3 1 mm year ) – had the lowest average asymptotic General 373,651 499 748 57.29 <0.001 4825 – < size among the morphs (170.3 and 164.7 mm, Common L∞ 483,505 503 961 60.70 0.001 4948 123 Common K 508,530 503 1011 51.52 <0.001 4974 149 respectively). Lacustrine C. artedi grew the slowest Common 716,880 507 1414 30.34 <0.001 5140 315 to the largest average asymptotic mean size. Coreg-
8 Cisco diversity in Great Slave Lake, NT, Canada
Fig. 5. Growth curves for cisco morphs from Great Slave, NT generated by fitting von Bertalanffy length-age models to standard-length- 1 at-otolith-age for the five cisco morphs. L∞ is the asymptotic average length (mm) and K is the growth coefficient (year ). Annual growth 1 rate (x) was estimated by L∞*K (mm year ; Gallucci & Quinn 1979). onus zenithicus expressed a faster growth rate than Physical resource use lacustrine C. artedi, its closest morphological vari- ant. Median depth of capture differed among the five puta- Instantaneous total mortality (Z) differed among tive morphs (H = 455.65; P < 0.001) with adfluvial adfluvial and lacustrine C. artedi morphs, and C. artedi occupying the shallowest water (i.e., caught C. zenithicus (F = 41.67; P < 0.001; Fig. 6). Annual in rivers during their spawning migration). Lacustrine survival was low for adfluvial C. artedi (35%) and C. artedi (median depth of capture = 54 m), big-eye high for lacustrine C. artedi (95%) and C. zenithicus cisco (median depth of capture = 61 m) and C. zeni- (86%). Model fit was good for adfluvial C. artedi thicus (median depth of capture = 48 m) were all (r2 = 0.92), moderate for C. zenithicus (r2 = 0.62) caught in waters deeper than C. sardinella and poor for lacustrine C. artedi (r2 = 0.19). (median = 24 m; all P < 0.05). The CPUE differed
9 Muir et al.
6 C. artedi 5 C. artedi (a)Adfluvial Z = 1.04 (b) Lacustrine Z = 0.05 S = 35 % S = 95 % 5 4
4 3 3 2 2 1 1
0 0 y = –1.04x + 8.20; r2 = 0.92 y = –0.05x + 2.07; r2 = 0.19
012345678 0 51015202530 Otolith age (y) 3.0 (c) C. zenithicus Z = 0.16 S = 86 %
Ln (Number of fish +1) 2.5
2.0
1.5
1.0
0.5
0.0 y = –0.16x + 3.39; r2 = 0.62 0 5 10 15 20 25 Otolith age (y) Fig. 6. Otolith catch curves for cisco from Great Slave Lake, NT. Linear regressions were fit to age at recruitment + 1 (white circles). Z is the instantaneous total mortality rate; S is the annual survival rate; r2 is the coefficient of determination. among morphs within the 30–59 and 60–89 m depth to small sample sizes, limited spatial coverage of strata (all P < 0.04), but not in the shallow stratum investigation, poor condition of preserved collections (0–29 m) or the deep stratum (>90 m). Lacustrine and a limited scope of analysis (Dymond 1943; C. artedi and C. sardinella had higher CPUE in the Clarke 1973; Murray 2006). Taxonomic distinctions 30–59 m depth stratum and C. zenithicus had higher were also probably confounded by the considerable CPUE in the 60–89 m depth stratum. variation and plasticity in character traits observed Per cent buoyancy (B) varied among the five mor- among Great Slave Lake ciscoes. phs (H = 119.74; P < 0.001) with adfluvial C. artedi A multivariate approach to analysing levels of vari- (B = 92.7%) weighing proportionally more in water ation in morphology, meristics, age, growth, life his- due to lower lipid content than the other morphs tory and habitat use, allowed us to describe except C. sardinella (B = 92.7; all P < 0.05). ecological groups of ciscoes that generally reflected C. sardinella also weighed proportionally more in existing coregonine taxonomy (Scott & Crossman water than C. zenithicus (B = 93.9), but lacustrine 1973). At a minimum, our analysis supports the C. artedi (B = 93.7) big-eye cisco (B = 93.4), and hypothesis that the Great Slave Lake ciscoes include C. zenithicus were equally buoyant. two strongly differentiated species (C. artedi and C. sardinella) and an adfluvial C. artedi morph that is distinct from its lacustrine conspecific in terms of Discussion life history, morphology, age, growth and mortality Previous efforts to resolve coregonine cisco taxon- (see also Blackie et al. 2012). Coregonus sardinella omy in Great Slave Lake have been inconclusive due has previously been identified from Great Slave Lake
10 Cisco diversity in Great Slave Lake, NT, Canada
(Turgeon & Bernatchez 2003), but herein, we provide migrated from the lake for spawning. It is probable the first comprehensive description of this species in that other adfluvial cisco populations occurred histor- the lake and confirm a significant range extension for ically in the Laurentian Great Lakes (Lawrie & Rah- the species (McPhail & Lindsey 1970). The lacustrine rer 1973; Christie 1974). Accumulations of lumber C. artedi differs little from descriptions throughout milling wastes and dredged shipping channels have its range. In addition to these three ciscoes, linear been implicated in the destruction of C. artedi and phenotypic traits, gillraker number and morphology, Coregonus clupeaformis spawning grounds in many and growth data support the possible occurrence of Great Lakes tributaries (Christie 1974), demonstrating two other, less-distinct morphs, the big-eye cisco the susceptibility of adfluvial ciscoes to habitat dis- C. artedi and C. zenithicus. Although the big-eye turbances. morph was not identified by the statistical models, it In Great Slave Lake, spawning migrations of adflu- could be discriminated visually on the basis of differ- vial cisco are subject to intense subsistence and com- ences in linear phenotypic traits, such as orbital mercial dipnet fisheries in some rivers, including the length, paired fin lengths, head and gillraker mor- Yellowknife River. The effects of these fisheries on phology (expressed as thousandths of standard population dynamics are unknown. Data presented length; see Table 5). In addition, the big-eye morph herein show that mortality is relatively high and sur- showed different age and growth structure compared vival is only 35%, possibly due to the fishery. In with the other lacustrine cisco morphs. The C. zeni- addition, mining operations and hydroelectric devel- thicus morph was distinguished visually and by the opments (e.g., Bluefish Hydro Dam replacement; statistical model of linear phenotypic traits as well as Northwest Territories Power Corporation 2010) have by gillraker number and morphology (see Table 6), the potential for adverse effects on rivers that support which were within the range for the species across its adfluvial cisco populations. The adfluvial cisco is a distribution (Scott & Crossman 1973). key horizontal vector of energy transfer between Great Slave Lake, its inflowing rivers, and its connecting inland lakes. Conserving adfluvial cisco Adfluvial cisco populations is a high priority of Fisheries and Oceans The adfluvial cisco is strikingly uniform in appear- Canada (D. Leonard, Fisheries and Oceans Canada, ance with little variation among individuals. This Yellowknife, Northwest Territories, personal commu- morph is characterised by its small size (max = nication, 2011). 192 mm), young maximum age (9 years), young age at maturity, rapid growth, high mortality, relatively Lacustrine cisco short fins with a yellow hue, long caudal peduncle, relatively short snout and a terminal jaw (Muir et al. The lacustrine C. artedi of Great Slave Lake is 2011; Blackie et al. 2012). similar in gross morphology, ecology and biology to Adfluvial life-history types occur in Lake Baikal, populations throughout their North American range. Russia [e.g., Coregonus autumnalis migratorius The ranges of morphometric and meristic traits for (Georgi)], throughout Siberia (e.g., Coregonus peled; lacustrine C. artedi fell within those reported for the Berg 1948), and are common in Scandinavia (e.g., taxon, but were the widest among the Great Slave Coregonus albula and Coregonus lavaretus; Næsje Lake ciscoes. The literature on C. artedi is extensive et al. 1986; Sandlund et al. 2012). Amphidromous and need not be repeated here. For data on the spe- (i.e., travelling between fresh and saltwater for feeding cies, see Koelz (1929), Dymond & Pritchard (1930), rather than breeding) C. artedi occur throughout Pritchard (1931), Dymond (1943), McPhail & Arctic Canada (Morin et al. 1981; Bernatchez & Lindsey (1970), Scott & Crossman (1973). This spe- Dodson 1990), but few adfluvial ciscoes have been cies is an important prey fish and has supported described in North America (except see Blackie et al. subsistence and commercial fisheries throughout its 2012). range. A large population of C. artedi spawns in the St. Marys River, the connecting channel between lakes Big-eye cisco Superior and Huron. However, the extent these fish use lake habitat is currently unknown (M.P. Ebener, The big-eye cisco was not discriminated by statistical Chippewa Ottawa Resource Authority, Sault Ste. models, but it was visually distinct and characterised Marie, Michigan, personal communication, 2011). by a small body size (maximum = 204 mm; Fielder (2000) reported evidence from the St. Marys Table 6), rapid growth, large eye, long, narrow dorsal River of an apparent upriver progression in female fin and darkly pigmented scales (Muir et al. 2011). CPUE by week concurrent with an increase in go- The taxonomic affinity of the big-eye cisco remains nadosomatic index suggesting that these fish had uncertain, but it could either be a fast-growing deep-
11 12 al. et Muir
Table 5. Mean proportionate measurements of body parts for ciscoes from Great Slave Lake, NT expressed in thousandths of the standard length (STL; mm) standard error; raw data are given in parentheses. Variable definitions are given in Tables 2 and 3.
Variable Adfluvial Coregonus artedi Lacustrine C. artedi Big-eye cisco C. artedi Coregonus sardinella Coregonus zenithicus
STL* 157.79 0.83 200.83 5.13 146.06 3.72 225.83 7.02 185.36 3.95 ADL 60.53 0.4 (9.55 0.08) 68.05 0.61 (13.90 0.42) 66.91 0.96 (9.76 0.28) 69.83 1.08 (15.70 0.53) 68.40 0.75 (12.75 0.32) ANL 104.82 0.64 (16.55 0.14) 108.39 0.9 (21.87 0.61) 105.10 1.66 (15.36 0.48) 122.68 1.28 (27.76 0.99) 107.40 0.88 (19.98 0.47) BDD 208.48 0.93 (32.91 0.23) 223.35 2.06 (46.76 1.7) 218.57 3.47 (32.05 1.13) 199.91 2.51 (45.19 1.6) 218.01 2.36 (40.91 1.15) CPD 73.14 0.25 (11.53 0.06) 72.82 0.44 (14.94 0.47) 71.77 0.83 (10.48 0.31) 78.51 0.82 (17.76 0.62) 73.05 0.66 (13.60 0.35) † CPL 131.98 0.9 (20.79 0.16) 117.27 0.9 (23.66 0.66) 115.53 2.07 (16.84 0.53) 121.68 1.86 (27.50 1.03) 122.27 1.19 (22.76 0.58) DOL 105.50 0.47 (16.66 0.12) 112.22 0.66 (22.97 0.71) 105.29 1.05 (15.40 0.46) 103.90 1.2 (23.26 0.69) 110.77 0.86 (20.65 0.5) DOH 175.55 0.76 (27.68 0.17) 191.64 1.37 (38.23 0.89) 200.59 2.39 (29.19 0.73) 201.97 2.14 (45.29 1.36) 191.73 1.47 (35.26 0.69) † GRL 40.99 0.23 (6.45 0.04) 39.51 0.54 (7.93 0.22) 41.74 1.01 (6.06 0.18) 34.80 0.67 (7.68 0.31) 34.79 0.43 (6.37 0.14) HDD 104.92 0.51 (16.52 0.09) 105.76 0.5 (21.33 0.59) 117.28 1.13 (17.09 0.42) 96.21 1.07 (21.46 0.57) 107.80 0.64 (19.91 0.43) HLL 228.97 0.72 (36.05 0.15) 233.12 1.02 (46.83 1.25) 250.73 1.91 (36.52 0.87) 211.01 2.3 (47.10 1.27) 234.88 1.19 (43.28 0.86) † IOW 53.69 0.33 (8.46 0.06) 54.58 0.49 (11.41 0.4) 53.94 0.95 (7.94 0.27) 44.51 0.54 (9.96 0.3) 54.18 0.59 (10.17 0.28) LAL 93.28 0.44 (14.68 0.08) 100.97 0.67 (20.49 0.59) 110.71 1.12 (16.18 0.51) 85.74 1.32 (18.80 0.63) 102.33 0.78 (18.83 0.43) LUL 184.55 1.15 (29.09 0.22) 181.05 1.3 (36.58 1.03) 170.93 3.21 (24.97 0.8) 193.39 2.27 (43.69 1.55) 177.87 1.34 (32.94 0.74) MXL 77.87 0.41 (12.25 0.07) 84.78 0.57 (16.98 0.45) 91.60 0.97 (13.32 0.31) 74.65 0.81 (16.73 0.5) 83.90 0.6 (15.41 0.3) MXW 23.58 0.14 (3.71 0.02) 22.81 0.21 (4.58 0.13) 23.50 0.39 (3.42 0.09) 21.78 0.51 (4.86 0.16) 22.34 0.22 (4.13 0.09) † OOL 65.19 0.36 (10.24 0.04) 60.64 0.53 (11.80 0.24) 75.22 0.85 (10.98 0.3) 64.99 0.97 (14.44 0.36) 63.41 0.6 (11.56 0.2) PCL 170.17 0.68 (26.78 0.12) 178.08 1.59 (35.30 0.85) 187.00 2.77 (27.16 0.64) 187.04 2 (41.88 1.23) 172.56 1.49 (31.85 0.69) POL 52.46 0.34 (8.26 0.06) 57.43 0.47 (11.66 0.36) 60.93 1.2 (8.82 0.2) 45.57 0.94 (10.13 0.29) 56.41 0.55 (10.38 0.22) † PSL 111.32 0.33 (17.55 0.09) 115.05 0.56 (23.37 0.68) 114.59 1.03 (16.72 0.45) 100.44 0.9 (22.53 0.67) 115.05 0.59 (21.34 0.47) PVL 165.77 0.65 (26.12 0.14) 174.74 1.27 (34.65 0.82) 185.99 2.38 (27.00 0.59) 185.32 1.91 (41.44 1.17) 171.52 1.33 (31.59 0.64) TTL 263.37 0.93 (41.60 0.29) 257.24 1.19 (51.56 1.33) 260.46 2.43 (38.06 1.06) 244.38 2.42 (55.12 1.87) 260.95 1.88 (48.55 1.19)
*Standard length is given as an untransformed measure. †Traits used in secondary analysis of morphological variation (see Methods). Cisco diversity in Great Slave Lake, NT, Canada
Table 6. Mean counts of calcified body parts of ciscoes standard error from Great Slave Lake, NT; ranges are given in parentheses.
Variable Adfluvial Coregonus artedi Lacustrine C. artedi Big-eye cisco C. artedi Coregonus sardinella Coregonus zenithicus
TGR* 42.25 0.12 (39–46) 43.93 0.28 (36–56) 42.5 0.37 (38–48) 42.84 0.28 (41–46) 39.94 0.2 (33–46) LLS 74.95 0.25 (67–86) 73.44 0.42 (64–84) 71.09 1.07 (61–82) 77.47 1.22 (63–83) 72.95 0.5 (65–83) SPS 7.52 0.04 (6–9) 7.53 0.07 (6–9) 7.26 0.1 (7–8) 7.21 0.19 (6–9) 7.48 0.08 (6–9) ULS 7.7 0.04 (7–9) 7.84 0.07 (6–9) 7.37 0.14 (7–9) 8 0.11 (7–9) 7.7 0.08 (6–9) DRC 11.06 0.04 (9–13) 10.97 0.06 (9–14) 10.65 0.17 (9–14) 10.2 0.11 (9–12) 10.76 0.08 (9–14) ARC 12.04 0.05 (9–14) 11.85 0.07 (10–15) 11.74 0.14 (10–14) 13.18 0.11 (12–14) 11.75 0.1 (9–15) PRC 16.11 0.06 (14–19) 16.05 0.09 (12–18) 16 0.17 (12–18) 14.81 0.09 (14–16) 16.23 0.11 (11–18) VRC 11.35 0.03 (10–13) 11.29 0.04 (10–15) 11.26 0.14 (9–13) 11.15 0.09 (10–12) 11.22 0.07 (9–16) PMA 45.75 0.52 (21–72) 43.39 0.64 (23–76) 34.29 1.43 (20–59) 24.03 1.12 (15–45) 50.49 0.92 (22–76)
*Total number of gillrakers (i.e., sum of UGR and LGR). water morphological variant of C. artedi, or alterna- the upper range of the Great Slave Lake C. sardinella tively, a commonly occurring hybrid. Turgeon (2000) specimens (Table 6), although anadromous corego- reported that ciscoes from Great Slave Lake with nines typically have more gillrakers than their fresh- 40–46 gillrakers had mtDNA and nuclear alleles water counterparts (Scott & Crossman 1973). The characteristic of C. sardinella, whereas those that had high number of anal rays in C. pusillus was consis- 49–59 gillrakers had mtDNA and nuclear alleles tent with our finding that C. sardinella never had characteristic of C. artedi. Two of 63 (i.e., 3%) indi- <12 rays, whereas the other morphs had as few as viduals with low gillraker numbers (41 and 46) pos- nine rays (minimum), and the anal fin base length sessed nuclear alleles unique to both C. artedi and (ANL) greatly exceeded that of the other Great Slave C. sardinella at three loci providing evidence of Lake ciscoes. The fins of C. pusillus were also hybrids between these species (Turgeon 2000). If reported as dark, especially towards the distal ends, big-eye is indeed a hybrid, it is not rare because it and the ventral fins were quite black. This fin pig- represented about 5% of our sample. mentation pattern is a key characteristic of C. sardi- nella in Great Slave Lake (Muir et al. 2011) and distinctly separated it from all other ciscoes in the Least cisco lake, which have weakly pigmented or immaculate Coregonus sardinella in Great Slave Lake attain a fins. In general, the morphometric data given for moderate size (maximum = 298-mm STL and C. pusillus by Dymond (1943) are consistent with 255 g) and can be easily identified by its appear- those for C. sardinella from Great Slave Lake (cur- ance. This species is characterised by a relatively rent study and Murray 2006). Moreover, genetic data large orbit (but not as big as the big-eye), which previously showed C. sardinella DNA in Great Slave often extends beyond the dorsal body margin, nar- Lake (Turgeon & Bernatchez 2003), and recent anal- row interorbital width, low premaxillary angle, yses confirm that the specimens described herein superior mouth orientation, extended lower jaw, belong to C. sardinella (AFLP; J. Turgeon, Univer- long pelvic and dorsal fins, but short dorsal base, sit e Laval, unpublished data). >12 anal fin rays (all other ciscoes had <12 anal The Siberian C. sardinella are typically fluvial, rays), a long anal fin base, and heavy black pig- inhabiting rivers, but some populations are amphidr- mentation, especially on the ventral fins (Muir et al. omous and concentrate in the freshened portions of 2011). Coregonus sardinella was also captured in the sea for feeding (Berg 1948). Migratory and non- greater numbers in shallower water than the other migratory C. sardinella occur across northern North morphs. America with amphidromous river- and lake-spawn- Uncertainties about the occurrence of C. sardinella ing life-history types (McPhail & Lindsey 1970; in Great Slave Lake date to the early descriptions of Brown et al. 2007). The North American amphidr- north-western coregonines. Bean (1889) described a omous form has a gillraker range of 48–53, whereas new species Coregonus pusillus from Alaska and the the freshwater form ranges from 41 to 47 (Scott & Mackenzie River delta. Berg (1932), however, Crossman 1973), consistent with the range for the described Siberian C. sardinella as nearly identical to specimens in the current study (41–46). Although we C. pusillus, and both Dymond (1943) and Berg did not confirm lacustrine spawning grounds, no (1948) considered the two species synonymous. The C. sardinella were caught in extensive sampling in C. pusillus type (USNM 38366) had 88 lateral-line the Yellowknife, Beaulieu or Stark rivers flowing into scales, 49 gillrakers and 14 rays in the anal fin. The Great Slave Lake (Blackie et al. 2012; A. Muir & number of gillrakers on the C. pusillus type exceeded P. Vecsei, personal observations).
13 Muir et al.
do not undertake deep DVM, but move large quanti- Coregonus zenithicus ties of energy from offshore pelagia to nearshore Great Slave Lake shortjaw cisco can be characterised habitats during fall spawning migrations (Stockwell by an included lower jaw, high premaxillary angle, et al. 2009). Ciscoes are important forage for top downward projecting mouth and tan dorsum colour- predators such as lake trout and burbot Lota lota ation in fresh specimens. Coregonus zenithicus is (Scott & Crossman 1973; Ray et al. 2007; Sitar et al. known to show considerable morphological variation 2008; Gamble et al. 2011). across its range (Koelz 1929; Dymond 1943; Rawson The North American ciscoes do not easily fit into 1951; Todd & Steinhilber 2002), although the Great the current framework for resource management, Slave Lake specimens collected in the current study which is based on the concept of the ‘biological spe- fall within and close to the mean values reported for cies’. For this reason, understanding how ciscoes the species. The total range of gillraker number for evolve and their functional role in energy transfer C. zenithicus in the Laurentian Great Lakes was 32– within the food web comes to the forefront as a pre- 46 (Koelz 1929), almost identical to the range we requisite for determining appropriate means of cate- observed for the species in Great Slave Lake (33–46). gorising and conserving diversity within this group Dymond (1943), Rawson (1947), Clarke (1973) of fishes. The ciscoes typically form complexes that and Todd & Steinhilber (2002) considered C. zeni- show ecological and morphological variations that thicus to occur in Great Slave Lake, but recent col- can be both genetically- and environmentally based. lections from the main basin of the lake led Murray In an extensive mitochondrial and microsatellite & Reist (2003) to conclude that on the basis of survey of North American cisco phylogeography, gross morphology, those specimens were more simi- Turgeon & Bernatchez (2003) reported that the tax- lar to C. artedi than C. zenithicus. Dymond (1943) onomy of the C. artedi complex better reflected concluded that the type specimen from Great Slave geography than evolutionary history. These authors Lake described by Harper & Nichols (1919) as found that, based upon the genetic characters used, Leucichthys macrognathus was synonymous with C. zenithicus was genetically more similar to sympat- C. zenithicus. Dymond’s measurements of the type ric or nearby C. artedi than to C. zenithicus from specimen of L. macrognathus [see Dymond (1943) other drainages, indicative of multiple independent p. 216 for data] had morphological characteristics origins of morphs. Turgeon & Bernatchez (2003) consistent, and in some cases nearly identical to the argued that C. artedi should be recognised as the sole mean values for C. zenithicus in our collections legitimate taxon for North American ciscoes and that (Tables 5 and 6). unique ecomorphotypes be recognised as evolution- Although our statistical models of body shape did ary significant units (ESUs). not discriminate the shortjaw morph from the other Identifying, characterising and managing locally Great Slave Lake ciscoes, they were reasonably well adapted cisco morphs that reflect important ecologi- discriminated (~65% of individuals) by a model of cal and bioenergetic linkages are critical to conserv- linear phenotypic traits. Consistent with Laurentian ing the ecological integrity of northern food webs Great Lakes specimens, low numbers of short gillrak- and ecosystems. The latter approach has been ers and a slightly shorter lower jaw than upper jaw embraced by the Committee on the Status of were distinguishing traits. The shortjaw morph also Endangered Wildlife in Canada (COSEWIC). The had a faster growth rate and a shorter asymptotic Species at Risk Act (SARA) defines a ‘wildlife spe- average length compared with its closest morphologi- cies’ as a ‘species, subspecies, variety or geographi- cal variant – lacustrine C. artedi. cally or genetically distinct population of animal, plant or other organism, other than a bacterium or virus’ (Government of Canada 2002). This defini- Management implications tion explicitly recognises the importance of ecologi- Ciscoes play a key role in nutrient cycling by func- cally distinct phenotypes and, therefore, provides tioning as horizontal and vertical vectors of energy protection and status for cisco morphs in Canada. transfer from primary and secondary production to On the basis of the available evidence reviewed their predators (Stockwell et al. 2009; Gorman et al. herein, we concur with the summary conclusion of 2012). For example, Coregonus kiyi in the deep Turgeon & Bernatchez (2003) regarding the need to waters (i.e., >100 m) of Lake Superior undertake recognise the uniqueness of many lake-dwelling cis- DVM to prey on M. diluviana in the pelagia during coes and the need to develop management strategies night where they are vulnerable to predation by sisco- that focus on local circumstances so as to ensure wet lake trout Salvelinus namaycush (Hrabik et al. the protection of environments and the processes 2006; Sitar et al. 2008; Stockwell et al. 2010; giving rise to evolutionary divergence within the Ahrenstorff et al. 2011; Gorman et al. 2012). C. artedi species complex.
14 Cisco diversity in Great Slave Lake, NT, Canada
Acknowledgements revealed by mitochondrial DNA restriction analysis. Journal of Fish Biology 39(Suppl. A): 283–290. This research was supported by Canada’s Interdepartmental von Bertalanffy, L. 1938. A quantitative theory of organic Recovery Fund and Species at Risk programme and the Great growth. Human Biology 10: 181–213. Lakes Fishery Commission. Fisheries and Oceans Canada also Blackie, C.T., Vecsei, P. & Cott, P.A. 2012. Contrasting phe- provided financial and in-kind support. The paper was pre- notypic variation among river and lake caught cisco from sented at the 11th Annual Coregonid Symposium in Mondsee Great Slave Lake: evidence for dwarf and large morphs. Austria, September 25–30, 2011. Thanks to two anonymous Journal of Great Lakes Research 38: 798–805. reviewers for their thoughtful insights, which helped improve Bookstein, F.L. 1989. Principal warps, thin plate splines the manuscript. We thank B. Arquilla, K. Bourassa, R. Bour- and the decomposition of deformations. IEEE Transac- ke, E. Coe, J. Crowe, J. Galloway, P. Smith, N. Sweetman tions on Pattern Analysis and Machine Intelligence 11: and C. Wood for their assistance with sampling and data col- 567–585. lection. Special thanks to C. Sawatzky for coordinating and Brown, R.J., Bickford, N. & Severin, K. 2007. Otolith trace organising all administrative activities associated with the pro- element chemistry as an indicator of anadromy in Yukon ject and to N. Mochnacz and C. Myskiw for supplying gear to River drainage coregonine fishes. Transactions of the Ameri- support our efforts. can Fisheries Society 136: 678–690. Burnham, K.P. & Anderson, D.R. 2002. Model selection and References multimodel inference: a practical information-theoretic approach, 2nd edn. New York: Springer-Verlag. 488 pp. Ahrenstorff, T.D., Hrabik, T.R., Stockwell, J.D., Yule, D.L. Christie, W.J. 1974. Changes in fish species composition of & Sass, G.G. 2011. Seasonally dynamic diel vertical the Great Lakes. Journal of the Fisheries Research Board of migrations of Mysis diluviana, coregonine fishes, and si- Canada 31: 827–854. scowet lake trout in the pelagia of western Lake Supe- Clarke, R.M. 1973. The systematics of ciscoes (Coregonidae) rior. Transactions of the American Fisheries Society 140: in central Canada. Winnipeg: Doctoral dissertation, Univer- 1504–1520. sity of Manitoba. 243 pp. Akaike, H. 1973. Information theory as an extension of the max- Clemens, B.J. & Stevens, E.D. 2003. Buoyancy range, gas imum likelihood principle. In: Petrov, B.N. & Csaki, F., eds. bladder volume, and lipid content of adult bloater, Coreg- Proceedings of the second international symposium on infor- onus hoyi Gill, in the Laurentian Great Lakes. Environmen- mation theory. Budapest: Akademiai Kiado, pp. 267–281. tal Biology of Fishes 68: 175–182. Alexander, R.M. 1972. The energetics of vertical migration by Dymond, J.R. 1943. The coregonine fishes of northwestern fishes. In: Sleigh, M.A. & MacDonald, A.G., eds. The Canada. Transactions of the Royal Canadian Institute 24: effects of pressure on living organisms. Symposia of the 171–232. Society for Experimental Biology. New York: Academic Dymond, J.R. & Pritchard, A.L. 1930. Some ciscoes or lake Press, pp. 273–294. herrings of western Canada. Contributions to Canadian Biol- Alexander, R.M. 1993. Buoyancy. In: Evans, D.H., ed. The ogy and Fisheries 5: 469–474. physiology of fishes, 2nd edn. Boca Raton, Florida: CRC Eshenroder, R.L., Argyle, R.L. & TeWinkel, L.M. 1998. Evi- Press, pp. 75–79. dence for buoyancy regulation as a speciation mechanism in Bates, D.M. & Watts, D.G. 1988. Nonlinear regression analysis Great Lakes ciscoes. Archive fur Hydrobiologie 50: 207– and its applications. New York: John Wiley & Sons. 394 pp. 217. Bean, T.H. 1889. Description of Coregonus pusillus, a new Fielder, D.G. 2000. Lake herring spawning grounds of the St. species of whitefish from Alaska. Proceedings of the United Marys River with some potential implications for early States National Museum 11: 526. spring navigation. North American Journal of Fisheries Berg, L.S. 1932. Les poissons des eaux douces de l’U.R.S.S. Management 20: 552–561. et des pays limitrophes. Leningrad: U.S.S.R. 745 pp. Fraley, C. & Raftery, A.E. 2009. MCLUST Version 3 for R: Berg, L.S. 1948. Freshwater fishes of the U.S.S.R. and adja- normal mixuture modeling and model-based clustering. Seat- cent countries. Washington, DC: Academy of Sciences of tle. Washington: Department of Statistics, University of the U.S.S.R Zoological Institue [Translated from Russian Washington, p. 56. by the Israel Program for Scientific Translations Jerusalem Gallucci, V.F. & Quinn, T.J. 1979. Reparameterizing, fitting, 1962 and published for the National Science Foundation]. and testing a simple growth model. Transactions of the 504 pp. American Fisheries Society 108: 14–25. Bernatchez, L. 2004. Ecological theory of adaptive radiation. Gamble, A.E., Hrabik, T.R., Stockwell, J.D. & Yule, D.L. An empirical assessment from Coregonine fishes (Salmoni- 2011. Trophic connections in Lake Superior Part I: the off- formes). In: Hendry, A.P., ed. Evolution illuminated, salmon shore fish community. Journal of Great Lakes Research 37: and their relatives. New York: Oxford University Press, pp. 541–549. 175–207. Gorman, O.T., Yule, D.L. & Stockwell, J.D. 2012. Habitat Bernatchez, L. & Dodson, J.J. 1990. Mitochondrial DNA vari- use by fishes of Lake Superior. II. Consequences of diel ation among anadromous populations of cisco (Coregonus habitat use for habitat linkages and habitat coupling in artedii) as revealed by restriction analysis. Canadian Journal nearshore and offshore waters. Aquatic Ecosystem Health & of Fisheries and Aquatic Sciences 47: 533–543. Management 15: 355–368. Bernatchez, L., Colombani, F. & Dodson, J.J. 1991. Phyloge- Government of Canada. 2002. Species at Risk Act (c. 29). netic relationships among the subfamily Coregoninae as Available at: http://canlii.ca/t/j104.
15 Muir et al.
Harper, G. & Nichols, J.T. 1919. Six new fishes from north- Canadian Manuscript Report of Fisheries and Aquatic Sci- western Canada. Bulletin of the American Museum of Natu- ences 2638: vii + 56 p. ral History 41: 263–271. Næsje, T.F., Jonsson, B. & Sandlund, O.T. 1986. Drift of Helland, I.P., Vøllestad, L.A., Freyhof, J. & Mehner, T. 2009. cisco and whitefish larvae in a Norwegian river. Transac- Morphological differences between two ecologically similar tions of the American Fisheries Society 115: 89–93. sympatric fishes. Journal of Fish Biology 75: 2756–2767. Northwest Territories Power Corporation 2010. Project infor- Hrabik, T.R., Jensen, O.P., Martell, S.J.D., Walters, C.J. & mation: Bluefish Lake Dam Replacement. Yellowknife, Kitchell, J.F. 2006. Diel vertical migration in the Lake Northwest Territories: Northwest Territories Power Corpora- Superior pelagic community. I. Changes in vertical migra- tion. 16 pp. tion of coregonids in response to varying perdation risk. Ogle, D. 2012. FishR Vignette – von Bertalanffy growth mod- Canadian Journal of Fisheries and Aquatic Sciences 63: els. Available at: http://www.ncfaculty.net/dogle/fishR/ 2286–2295. index.html. Hudson, A.G., Vonlanthen, P., Muller,€ R. & Seehausen, O. Ohlberger, J., Mehner, T., Staaks, G. & Holker,€ F. 2008. Tem- 2007. Review: the geography of speciation and adaptive perature-related physiological adaptations promote ecological radiation in Coregonines. Archiv fur€ Hydrobiologie Special divergence in a sympatric species pair of temperate freshwa- Issues in Advanced Limnology 60: 111–146. ter fish, Coregonus spp. Functional Ecology 22: 501–508. Jennings, M.R. 1989. Use of spring scales for weighing live Oyadomari, J.K. & Auer, N.N. 2007. Influence of rearing tem- fish in the field. North American Journal of Fisheries Man- perature and feeding regime on otolith increment deposition agement 9: 509–511. in larval ciscoes. Transactions of the American Fisheries Kahilainen, K., Alajarvi, E. & Lehtonen, H. 2005. Planktivory Society 136: 766–777. and diet-overlap of densely rakered whitefish (Coregonus Phillips, R.B. & Ehlinger, T.J. 1995. Evolutionary and ecolog- lavaretus (L.)) in a subarctic lake. Ecology of Freshwater ical considerations in the reestablishment of Great Lakes Fish 14: 50–58. coregonid fishes. American Fisheries Society Symposium Kahilainen, K.K., Malinen, T., Tuomaala, A., Alaj€arvi, E., 17: 133–144. Tolonen, A. & Lehtonen, H. 2007. Empirical evaluation of Politov, D.V., Bickham, J.W. & Patton, J.C. 2004. Molecular phenotype–environment correlation and trait utility with phylogeography of palearctic and nearctic ciscoes. Annales allopatric and sympatric whitefish, Coregonus lavaretus Zoologici Fennici 41: 13–23. (L.), populations in subarctic lakes. Biological Journal of the Posada, D. & Buckley, T.R. 2004. Model selection and model Linnean Society 92: 561–572. averaging in phylogenetics: advantages of Akaike informa- Koelz, W. 1929. Coregonid fishes of the Great Lakes. U.S. tion criterion and bayesian approaches over likelihood ratio Bureau Fisheries Bulletin 43: 297–643. tests. Systematic Biology 53: 793–808. Lawrie, A.H. & Rahrer, J.F. 1973. Lake Superior: a case his- Pritchard, A.L. 1931. Taxonomic and life history studies of tory of the lake and its fisheries. Ann Arbor, Michigan: the ciscoes of Lake Ontario. Publications of the Ontario Great Lakes Fishery Commission, Technical Report Series Fisheries Research Laboratory No. 41. Toronto, Ontario: No. 19. 69 pp. University of Toronto. 78 pp. McPhail, J.D. & Lindsey, C.C. 1970. Freshwater fishes of R Core Team 2012. R: a language and environment for statis- Northwestern Canada and Alaska. Ottawa, Ontario: Fisheries tical computing. Vienna, Austria: R Foundation for Statisti- Research Board of Canada Bulletin 173. 381 pp. cal Computing. Morin, R., Dodson, J.J. & Power, G. 1981. The migrations of Rawson, D.S. 1947. Great Slave Lake and Lake Athabasca: anadromous cisco (Coregonus artedii) and lake whitefish Northwest Canadian fish surveys in 1944–45. Bulletin of the (C. clupeaformis) in estuaries of eastern James Bay. Cana- Fisheries Research Board of Canada 72: 45–94. dian Journal of Zoology 59: 1600–1607. Rawson, D.S. 1951. Studies of the fish of Great Slave Lake. Muir, A.M., Vecsei, P. & Reist, J.D. 2011. A field guide to the Journal of the Fisheries Research Board of Canada 8: 207– taxonomy of ciscoes in Great Slave Lake, Northwest Territo- 240. ries, Canada. Ann Arbor, Michigan: Great Lakes Fishery Ray, B.A., Hrabik, T.R., Ebener, M.P., Gorman, O.T., Schre- Commission Miscellaneous Publication 2011–02. Available iner, D.R., Schram, S.T., Sitar, S.P., Mattes, W.P. & Bronte, from: http://www.glfc.org/pubs/SpecialPubs/2011-02.pdf. C.R. 2007. Diet and prey selection by Lake Superior lake Muir, A.M., Vecsei, P. & Krueger, C.C. 2012. A perspective trout during spring 1986–2001. Journal of Great Lakes on perspectives: a method toward reducing variation in digi- Research 33: 104–113. tal shape analysis. Transactions of the American Fisheries Reist, J.D. 1985. An empirical evaluation of several univariate Society 141: 1161–1170. methods that adjust for size variation in morphometric data. Muir, A.M., Vecsei, P., Pratt, T.C., Krueger, C.K., Power, M. Canadian Journal of Zoology 63: 1429–1439. & Reist, J.D. 2013. Ontogenetic shifts in morphology and Reist, J.D. 1987. Comparative morphometry and phenetics of resource use of cisco C. artedi from Great Slave Lake, N.T., the genera of esocoid fishes (Salmoniformes). Zoological Canada. Journal of Fish Biology 82: 600–617. Journal of the Linnean Society 89: 275–294. Murray, L. 2006. A morphological examination of sympatric Ricker, W.J. 1975. Computation and interpretation of biologi- cisco forms in four lakes with specific reference to the occur- cal statistics of fish populations. Ottawa, Ontario: Fisheries rence of shortjaw cisco (Coregonus zenithicus) in Manitoba. Research Board of Canada Bulletin 191, 382. Winnipeg: Master’s thesis, University of Manitoba. 297 pp. Rohlf, F.J. & Bookstein, F.L. 2003. Computing the uniform Murray, L. & Reist, J.D. 2003. Status report on the shortjaw component of shape variation. Systematic Biology 52: 66– cisco Coregonus zenithicus in central and western Canada. 69.
16 Cisco diversity in Great Slave Lake, NT, Canada
Sandlund, O.T., Diserud, O.H. & Næsje, T.F. 2012. Lessons Turgeon, J. 2000. Ecologie mol eculaire et phylog eographie to learn from 123 years of catch data from a small scale des ciscos d’am erique du nord (teleostei: Coregonus spp.). whitefish fishery. In: Tallman, R.F., Howland, K.L., Rennie, Quebec: Doctoral dissertation, Universit e Laval. 219 pp. M.D. & Mills, K., eds. 10th International Symposium on the Turgeon, J. 2002. Molecular ecology and phylogeography of Biology and Management of Coregonid Fishes. Proceedings North American ciscoes (Teleostei: Corgoninae: Coregonus of the 10th International Symposium on the Biology and spp.). Bulletin of the Canadian Society of Zoologists 33: Management of Coregonid Fishes. Winnipeg, Manitoba: 110. Special Issues in Advances in Limnology, Schweizerbart Turgeon, J. & Bernatchez, L. 2001a. Clinal variation at micro- Science Publishers, Stuttgart, 63: 371–382. satellite loci reveals historical secondary intergradation Scott, B.W. 1960. Summaries of current information on short- between glacial races of Coregonus artedi (Teleostei: Coreg- jaw cisco, shortnose cisco, and blackfin cisco. Maple, oninae). Evolution 55: 2274–2286. Ontario: Research Branch Ontario Department of Lands and Turgeon, J. & Bernatchez, L. 2001b. Mitochondrial DNA phy- Forests, Research Information Paper (Fisheries) No 9. 26 pp. logeography of lake cisco (Coregonus artedi): evidence sup- Scott, W.B. & Crossman, E.J. 1973. Freshwater fishes of Can- porting extensive secondary contacts between two glacial ada. Volume 184. Ottawa, Ontario: Fisheries Research races. Molecular Ecology 10: 987–1001. Board of Canada Bulletin. 966 pp. Turgeon, J. & Bernatchez, L. 2003. Reticulate evolution and Sitar, S.P., Morales, H.M., Mata, M.T., Bastar, B.B., Dupras, phenotypic diversity in North American ciscoes, Coregonus D.M., Kleaver, G.D. & Rathbun, K.D. 2008. Survey of si- spp. (Teleostei: Salmonidae): implications for the conserva- scowet lake trout at their maximum depth in Lake Superior. tion of an evolutionary legacy. Conservation Genetics 4: Journal of Great Lakes Research 34: 276–286. 67–81. Smith, G.R. & Todd, T.N. 1992. Morphological cladistic Turgeon, J., Estoup, A. & Bernatchez, L. 1999. Species flock study of coregonine fishes. In: Todd, T.N. & Luczynski, M., in the North American Great Lakes: molecular ecology of eds. Biology and management of coregonid fishes. Polskie Lake Nipigon ciscoes (Teleostei: Coregonidae: Coregonus). Archiwum Hydrobiologii, pp. 474–490. Evolution 53: 1857–1871. Stockwell, J.D., Ebener, M.P., Black, J.A., Gorman, O.T., Vecsei, P., Blackie, C.T., Muir, A.M., Machtans, H.M. & Re- Hrabik, T.R., Kinnunen, R.E., Mattes, W.P., Oyadomari, ist, J.D. 2012. A preliminary assessment of cisco (Coreg- J.K., Schram, S.T., Schreiner, D.R., Seider, M.J., Sitar, S.P. onus spp.) diversity in Yellowknife Bay, Great Slave Lake, & Yule, D.L. 2009. A synthesis of cisco recovery in Lake Northwest Territories. In: Tallman, R.F., Howland, K.L., Superior: implications for native fish rehabilitation in the Rennie, M.D. & Mills, K., eds. 10th International Sympo- Laurentian Great Lakes. North American Journal of Fisher- sium on the Biology and Management of Coregonid Fishes. ies Management 29: 626–652. Proceedings of the 10th International Symposium on the Stockwell, J.D., Hrabik, T.R., Jensen, O.P., Yule, D.L. & Bal- Biology and Management of Coregonid Fishes. Winnipeg, ge, M. 2010. Empirical evaluation of predator-driven diel Manitoba: Special Issues in Advances in Limnology, Schwe- vertical migration in Lake Superior. Canadian Journal of izerbart Science Publishers, Stuttgart, 63: 299–322. Fisheries and Aquatic Sciences 67: 473–485. Vuorinen, J.A., Bodaly, R.A., Reist, J.D., Bernatchez, L. & Sukhanova, L.V., Smirnov, V.V., Smirnova-Zalumi, N.S., Dodson, J.J. 1993. Genetic and morphological differentia- Belomestnykn, T.V. & Kirilchik, S.V. 2012. Molecular phy- tion between dwarf and normal size forms of lake whitefish logeography of Lake Baikal coregonid fishes. In: Tallman, (Coregonus clupeaformis) in Como Lake, Ontario. Canadian R.F., Howland, K.L., Rennie, M.D. & Mills, K., eds. 10th Journal of Fisheries and Aquatic Sciences 50: 210–216. International Symposium on the Biology and Management Zar, J.H. 1999. Biostatistical analysis. New Jersey: Prentice- of Coregonid Fishes. Proceedings of the 10th International Hall. 663 pp. Symposium on the Biology and Management of Coregonid Zelditch, M.L., Swiderski, D.L., Sheets, D.H. & Fink, W.L. Fishes. Winnipeg, Manitoba: Special Issues in Advances in 2004. Geometric morphometrics for biologists: a primer. Limnology, Schweizerbart Science Publishers, Stuttgart, 63: New York: Elsevier Academic Press. 443 pp. 261–283. Zimmerman, M.S. & Krueger, C.C. 2009. An ecosystem per- Sv€ardson, G. 1979. Speciation of Scandinavian Coregonus. spective on re-establishing native deepwater fishes in the Report of the Institute of Freshwater Research, Drottning- Laurentian Great Lakes. North American Journal of Fisher- holm 57: 1–95. ies Management 29: 1352–1371. Todd, T.N. & Smith, G.R. 1992. A review of differentiation Zimmerman, M.S., Krueger, C.C. & Eshenroder, R.L. 2006. in Great Lakes ciscoes. Polskie Archiwum Hydrobiologii Phenotypic diversity of lake trout in Great Slave Lake: dif- 39: 261–267. ferences in morphology, buoyancy, and habitat depth. Trans- Todd, T.N. & Stedman, R.M. 1989. Hybridization of ciscoes actions of the American Fisheries Society 135: 1056–1067. (Coregonus spp.) in Lake Huron. Canadian Journal of Zool- Zimmerman, M.S., Schmidt, S.N., Krueger, C.C., Vander Zan- ogy 67: 1679–1685. den, M.J. & Eshenroder, R.L. 2009. Ontogenetic niche shifts Todd, T.N. & Steinhilber, M. 2002. Diversity in shortjaw cisco and resource partitioning of lake trout morphotypes. Cana- (Coregonus zenithicus) in North America. Archive fur Hydro- dian Journal of Fisheries and Aquatic Sciences 66: 1007– biologie Special Issues in Advanced Limnology 57: 517–525. 1018.
17