Introduction

Despite the coexistence of similar species across many taxa, the basic principles of niche theory suggest that complete niche overlap is not evolutionarily possible (Gause 1934; Hutchinson 1957; Hardin 1960). Niche partitioning (Levins 1968; MacArthur 1972) (also termed niche differentiation or niche segregation), the process by which competing species evolve different forms of resource use is a fundamental process in community ecology and has been widely used to explain the coexistence of similar species (Schoener 1974; Giller 1984; Ross 1986). Coexistence may arise from the segregation of specific resources (classical resource partitioning) or from differences in when (temporal resource partitioning) and where (spatial resource partitioning) resources are utilized (Pianka 1969; Schoener 1974; Ross 1986). In fish assemblages, partitioning of food resources is often the principal mechanism of niche segregation (Gascon and Leggett 1977; Gerking 1994).

In coastal Newfoundland, the closely related gadids Greenland cod (Gadus ogac) and Atlantic cod (Gadus morhua) are opportunistic predators with overlapping geographic distributions (Scott and Scott 1988). Juveniles of both species are common nearshore inhabitants and found intermixed in most bays (Rose 2007) and there is a long-standing view that competition is likely between the two species (Cohen et al. 1990). Since the early 1990s, G. morhua stocks around Newfoundland have been in a depleted state (for much longer further north off Labrador) (Rose 2007). There is little data to assess changes in G. ogac stocks, but a priori inference would suggest less or no change, as unlike G. morhua, G. ogac is a cold water species (Kearley 2012) and would not have been negatively influenced by the cold conditions of the early 1990s. In addition, G. ogac were never commercially harvested. Local knowledge of fishermen along the Newfoundland coast tends to support this inference (pers. comms.). Feeding competition between these species in inshore waters, where most juvenile G. morhua (Lear et al. 1980; Dalley and Anderson 1997; Methven and Schneider 1998) and all G. ogac reside (Scott and Scott 1988; Mikhail and Welch 1989), could help explain the slow rebuilding of depleted G. morhua stocks over the past decades, but few data existed to test this hypothesis.

Despite its historic commercial importance, the feeding ecology of older G. morhua juveniles (ages-2-4) in coastal areas of Newfoundland is poorly known. Clark and Green (1990) examined their diel activity patterns in Conception Bay using sonic telemetry and inferred that the higher activity rates observed were related to feeding, but provided no information on prey selection. Studies on age 1–2 juveniles in Conception Bay yielded differing results: Keats et al. (1987) found small (<12.5 cm) juveniles fed on pelagic prey and larger (16–23.5 cm) juveniles fed on benthic organisms whereas Keats and Steele (1992) reported that all juveniles (<23.5 cm) consumed mainly pelagic crustaceans.

Previous studies of diet overlap between G. morhua and G. ogac further north have yielded conflicting results. Feeding patterns from two inshore locations in southern Labrador suggested that the two species had dissimilar diets (Chaput 1981). In contrast, substantial overlap in diets was reported from West Greenland by Nielsen and Andersen (2001). In coastal Newfoundland, no comparisons of diet overlap have been made.

Studies of dietary resource partitioning in co-occurring or closely related fish have typically used stomach content analysis to examine dietary overlap (e.g., Grossman 1986; Garrison 2000; Corrêa et al. 2009). This method offers several benefits: stomachs samples are relatively easy to collect and prey items can be identified often to species and life stage. However, stomach analyses provide only a “snapshot” of dietary habits, often with many empty stomachs, and may also show bias toward prey items with lower digestion rates (Hyslop 1980). In contrast, stable isotope (δ13C and δ15N) signatures reflect biologically integrated nutrients in the diet over a long time period—up to several months for muscle tissue (Peterson and Fry 1987; Lorrain et al. 2002). Hence, isotope analysis identifies the longer term feeding habits of an individual, no matter their last meal. Used in conjunction, these methods provide a more complete representation of an organism’s dietary habits.

In this study, our objective was to compare the feeding habits of G. morhua and G. ogac in coastal Newfoundland and quantify dietary overlap using both stomach content and stable isotope analyses. The degree of overlap in dietary resources was expected to reflect the amount of interspecific feeding competition between these co-occurring species. Working null hypotheses were that G. morhua and G. ogac would not differ in: 1) diet, 2) pelagic and benthic oriented feeding, and 3) trophic position within the coastal ecosystem.

Methods

Collection of samples

Forty-seven mostly juvenile Gadus morhua and 42 Gadus ogac of comparable sizes (17–63 cm) were caught by hook and line over several (2–7) days in July of 2009 and 2010 from a small research vessel (RV Gecho II) within an area of approximately 2.5 ha near Petley Beach in Smith Sound, Trinity Bay, Newfoundland (Figs. 1 and 2). Forty-one fish (20 G. ogac and 21 G. morhua) were collected in 2009 and 48 fish (22 G. ogac and 26 G. morhua) were collected in 2010. Water depths at the site varied from <1 m to >40 m; most fish were caught within 1–2 m of the bottom. All fish were put on ice onboard the vessel and later sampled for total length, weight, sex, and reproductive stage. Stomachs were removed, weighed and frozen for later analysis and a small sample (1–2 cm2) of dorsal muscle tissue posterior to the head was removed and frozen for stable isotope analysis.

Fig. 1
figure 1

Map of the Eastern portion Newfoundland showing location of sampling area (black star) within Smith Sound. Inset shows position of enlarged map relative to the Island of Newfoundland and NAFO Divisions 2 J, 3 K and 3 L

Full size image
Fig. 2
figure 2

Size frequency distribution of sampled fish. Black bars = G. ogac; open bars = G. morhua

Full size image

Stomach contents analysis

Stomachs contents were sorted and identified to species or nearest taxonomic level, with weights recorded to the nearest 0.01 g. Cumulative prey curves were used to judge if n was sufficient to effectively describe diet compositions (Hoffman 1979; Cailliet et al. 1986; Cortés 1997). The order in which stomachs were analyzed was randomized 10 times and the mean number of new prey items found consecutively in the stomachs plotted against the number of stomachs that contained prey. Linear regressions were then performed on the last four points of the curve to assess if an asymptote had been reached (sensu Bizzarro et al. 2007). If the slope did not differ significantly from 0 (i.e., p > 0.05), the curve was considered to have reached an asymptote with n adequate to describe diet.

The relative quantity of stomach contents and relative importance of individual prey types were assessed using the following indices: 1) relative frequency of occurrence (FO%) = number of stomachs with prey item, i, as a percentage of the total number of stomachs, 2) relative gravimetric abundance (W%) = total weight of prey item i, as a percentage of the weight of total stomach contents summed for all fish, 3) mean total fullness index \( \left(\mathrm{TFI}\right)=\frac{1}{n}{\displaystyle \sum_{f=1}^n\left(\mathrm{weight}\kern0.15em \mathrm{of}\kern0.15em \mathrm{stomach}\kern0.15em \mathrm{contents}\kern0.15em \mathrm{of}\kern0.15em \mathrm{fish}\kern0.15em f/{\left(\mathrm{length}\ \mathrm{of}\ \mathrm{fish}\ f\right)}^3\right)\times {10}^4} \), and 4) mean partial fullness index \( \left(\mathrm{PFI}\right)=\frac{1}{n}{\displaystyle \sum_{f=1}^n\left(\mathrm{weight}\ \mathrm{of}\ \mathrm{prey}\ \mathrm{item},i,\mathrm{in}\ \mathrm{fish}\ f/{\left(\mathrm{length}\ \mathrm{of}\ \mathrm{fish}\ f\right)}^3\right)\times {10}^4}, \) where n is the number of stomachs examined, weight is in 0.1 g and fish length is in cm. Niche breadths for each species were estimated using Levins’ standardized index (Levins 1968; Hurlbert 1978; Krebs 1989): \( B=1/\left(n{\displaystyle \sum }{p}_{x{i}^2}\right) \), where p xi is the proportion of species x using prey item i, and n is the number of prey items available. Prey items available included all prey species identified in the study and availability was assumed to be the same for both species and size classes. B ranges from 1/n (use of a single resource) to 1 (equal usage of resources). Dietary niche overlap between species was assessed with Schoener’s (1970) overlap index: C = 1–0·5(∑|p ix –p iy |), where p ix and p iy are the proportions by weight of prey item i in the diets of species x and species y, respectively. Index values range from 0 to 1, with 0 representing no overlap and 1 representing complete overlap and values ≥0.6 generally considered biologically significant (Wallace 1981).

Food items were also classified into pelagic, suprabenthic and benthic categories based on studies of the prey taxa and previous cod diet studies (e.g., Scott and Scott 1988; Parrish et al. 2009). To compare the relative importance of prey categories, the gravimetric abundance (W%) for each prey category was calculated for all individuals and tested statistically for differences between species using a Kruskal-Wallis test.

A one-way analysis of similarity (ANOSIM) (Clarke 1993) of gravimetric abundance (W%) and frequency of occurrence (FO%) of prey items for each individual was used to assess dietary differences between species. The proportion by mass of each prey item in the stomach contents of each individual was used to calculate gravimetric abundance (W%) while the presence or absence of each prey item was used to determine frequency of occurrence (FO%). Prior to analysis, data were square-root transformed and used to construct a Bray-Curtis similarity matrix. Similarity percentages (SIMPER) analysis was used to identify which prey categories contributed most to dissimilarities between species (Clarke 1993). Both ANOSIM and SIMPER were performed using PRIMER 6 software (Clarke and Gorley 2006).

Stable isotope analysis

Dorsal muscle tissue samples were thawed, dried to constant weight (48 h at ~80 ° C in a drying oven), crushed to a fine powder using a mortar and pestle and sent to the CREAIT Network Stable Isotope Lab Facility at Memorial University of Newfoundland. Stable carbon and nitrogen isotope ratios and elemental determinations for each sample were determined by analysis of CO2 and N2, respectively, produced by combustion using a Carlo Erba NA1500 Series II Elemental Analyser followed by gas chromatograph separation and online analysis by continuous-flow mass spectrometer. Stable carbon and nitrogen ratios were expressed in delta (δ) notation, defined as the parts per thousand (‰) differences from a standard material: δX = [(R sample /R standard ) − 1] × 103, where δ = the measure of heavy to light isotope in the sample, X = 13C or 15 N and R = the corresponding ratio (13C/12C or 15 N/14 N). International Standard references are Vienna Pee Dee Belemnite (VPDB) for carbon, and atmospheric N2 for nitrogen.

To estimate trophic niche breadth and structure, quantitative metrics based on the position of individuals in trophic niche space developed by Layman et al. (2007) and described by Jackson et al. (2011) were applied at the population level using individuals as measurement units. Metrics were calculated using the Stable Isotope Analysis in R (SIAR) package (Parnell et al. 2008) for R statistical computing package (R Development Core Team 2007) and are briefly defined as follows: 1) δ 15N Range (NR): a measure of degree of trophic diversity calculated as the distance between the most enriched and most depleted δ 15N values for a given species or group; 2) δ 13C Range (CR): distance between the highest and lowest δ 13C which indicates the variability of food sources consumed; 3) Standard Ellipse Area (SEA): a measure of the total trophic niche breadth for a given species or group; 4) Mean distance to centroid (CD): average Euclidean distance of each individual to the mean δ 13C and δ 15N value which provides a measure of the average degree of trophic diversity within a species or group; 5) Mean nearest neighbour distance (MNND): mean of the Euclidean distances to each species’ nearest neighbour in bi-plot space which provides a measure of the overall density of species packing (i.e., a group comprised of many individuals with similar trophic ecologies would show a smaller MNND than a group in which individuals are more varied in terms of their trophic niche); 6) Standard deviation of nearest neighbour distance (SDNND): a measure of the evenness of species packing in bi-plot space with lower SDNND values suggesting a more even distribution of trophic niches.

Stable isotope ratios (δ 13C and δ 15N) and metric data (CD, MNND, and SDNND) were tested for normal (Gaussian) distribution using probability plots and frequency distributions and non-normal data were transformed using the Johnson transformation tool in Minitab 16. Between species differences in metrics and effect of body size on δ 13C and δ 15N values were evaluated using t-tests and regression analysis, respectively. Trophic niche overlap was estimated as the percent of overlapping SEA between species.

Results

Stomachs of G. morhua and G. ogac contained a substantial variety of prey items (Table 1). Cumulative prey curve regressions on the last four measures for both species had slopes that did not differ from 0 (regression, t = 3.66, p = 0.07 for G. morhua and t = 2.54, p = 0.13 for G. ogac) (Fig. 3).

Table 1 Importance of all prey items based on frequency of occurrence (FO%), relative weight (W%), and mean partial fullness (PFI) for G. ogac and G. morhua. Prey types: B = benthic; SB = suprabenthic; P = pelagic. Bold values indicate maximums for each index
Full size table
Fig. 3
figure 3

Cumulative prey curves for G. ogac (filled circle) and G. morhua (filled diamond). Symbols show the mean cumulative number of prey items per stomach sampled and error bars indicate SD

Full size image

Stomach contents

Indices of relative importance (FO%, W%, and PFI) for all prey items (Table 1) indicated that for G. ogac, polychaetes and mysids/euphausiids were the dominant prey items, occurring in 28.6 % and 50 % of stomachs and making up 36.5 % and 13.3 % of the total diet by weight. Polychaetes and mysids/euphausiids also had the highest PFI values at 0.797 and 0.262, respectively. Unidentified bony fish was the next most important prey item by frequency of occurrence (9.5 %) while G. morhua had the next highest relative weight (20 %) and PFI value (0.107).

In the stomachs of G. morhua, mysids/euphausiids and hyperiids had the highest frequency of occurrence (FO%) (mysids/euphausiids = 66.7 %; hyperiids = 62.5 %) and PFI values (mysids/euphausiids = 0.182; hyperiids = 0.406) while hyperiids and polychaetes had the highest relative weights at 38.7 % and 19.4 %, respectively. Annelids (12.5 %) and polychaetes (12.5 %) had the next highest frequency of occurrence while G. morhua had the next highest relative weight (16.5 %) and PFI (0.104) values.

G. morhua had significantly higher proportions by weight of pelagic prey items in their diet than G. ogac (Kruskal-Wallis, H = 28.8, p < 0.01) (Fig. 4). This was attributed almost entirely to the high relative weight of hyperiids in the stomachs of G. morhua (Table 1). In comparison, G. ogac had a higher relative abundance of benthic prey items (Kruskal-Wallis, H = 5.25, p = 0.02) (Fig. 4). No significant difference between species was found for the suprabenthic prey category (Kruskal Wallis, H = 1.27, p = 0.26) (Fig. 4).

Fig. 4
figure 4

Plots of relative gravimetric abundance (W%) of pelagic, suprabenthic and benthic prey categories by fish length (cm) for G. ogac and G. morhua

Full size image

Total fullness (TFI) values were somewhat higher for G. ogac than for G. morhua but did not differ significantly between species (Kruskal Wallis, H = 0.13, p = 0.72) (Table 2). G. ogac had a significantly lower niche breadth index than G. morhua (Kruskal Wallis, H = 11.25, p < 0.01) and low (C = 0.28) overlap in diet was found between species (Table 2).

Table 2 Mean total fullness index (TFI), dietary niche breadth (B) (Levins 1968) and dietary overlap (Schoener 1970) for G. ogac and G. morhua
Full size table

ANOSIMs showed significant differences in diet composition between G. ogac and G. morhua by both gravimetric abundance (W%) (R = 0.243 p < 0.01) and frequency of occurrence (FO%) (R = 0.227, p | < 0.01) (Table 3). Results from SIMPER analyses revealed high average dissimilarity between species (W%: 93.3 % dissimilarity; FO%: 90.3 %) with hyperiids, euphausiids, mysids/eupausiids and polychaetes contributing most to the dissimilarity for both dietary indices (Table 3).

Table 3 Percent contribution to average dissimilarity by prey item and dietary index for G. ogac and G. morhua
Full size table
Table 4 Summary of isotopic metrics by species. NR = δ 15N range; CR = δ 13C; SEA = standard ellipse area; CD = distance to centroid; MNND = mean nearest neighbour distance; SDNND = standard deviation of nearest neighbour distances
Full size table

Stable isotopes

Carbon isotope (δ 13C) levels in tissues of G. ogac were significantly more positive, or enriched in 13C, as compared to those of G. morhua, which were relatively depleted in 13C (t-test, t = 4.82, p < 0.01) (Table 4, Fig. 5). In contrast, no significant difference in mean δ 15N values was found between species (t-test, t = 1.59, p = 0.12) (Table 4). A significant positive relationship between body size (TL) and isotopic δ 13C values was evident in both species (regression, t = 3.52, p < 0.01 and t = 2.13, p = 0.04 for G. morhua and G. ogac, respectively) (Fig. 6). No significant relationship between body size and isotopic δ 15N values was observed for either species.

Fig. 5
figure 5

δ 15N - δ 13C bi-plots and group means (± 2 SE) of sampled G. ogac (filled circle) and G. morhua (empty circle). Enclosed areas represent the standard ellipse trophic niche area (SEA) occupied by each group (dashed line = G. ogac; solid line = G. morhua)

Full size image
Fig. 6
figure 6

Plots of δ 13C versus TL and regression slopes for a G. ogac and b G. morhua. Both regressions were significant at p < 0.05

Full size image

Neither mean distance to centroid (CD) or mean nearest neighbour distance (MNND) differed between species (CD: t-test, t = −1.16, p = 0.25; MNND: t = −1.30, p = 0.20). SDNND was lower for G. ogac than for G. morhua (Table 4). The percentages of SEA overlap between G. ogac and G. morhua were moderate at 43.3 % and 24.7 %, respectively (Table 4; Fig. 5).

Discussion

Results from both stable isotope and stomach analyses provide support for dietary niche partitioning between G. ogac and G. morhua and suggest only minor competition for food resources between species during the summer. This conclusion is based on rejection of the working null hypotheses on diet similarity and benthic-pelagic prey similarity. Stomach analyses indicated differing prey and a higher proportion of benthic items for G. ogac and pelagic items for G. morhua. In support of these conclusions, isotopic signatures for G. ogac were significantly more enriched in 13C, indicating more benthic feeding, whereas signatures for G. morhua were relatively depleted in 13C, indicative of more pelagic feeding (Davenport and Bax 2002; Hobson et al. 2002; Sherwood and Rose 2005).

Consistent with the present study, Chaput (1981) concluded that diets of G. morhua and G. ogac were dissimilar based on low correlation coefficients for major prey items identified from stomachs from specimens caught in shallow (<25 m) water at two nearshore sites in Labrador (NAFO Div 2 J). The authors attributed these differences to the high frequency of occurrence and contribution to total fullness of pelagic invertebrates for G. morhua and shrimp, fish and polychaetes for G. ogac. In contrast, Nielsen and Andersen (2001) found no difference in the diet of G. morhua and G. ogac in West Greenland and concluded the two species compete for food where their ranges overlap.

The differing results between the present and Labrador study and the Greenland study may relate to timing and differing prey fields. The present results indicate that larger G. ogac feed primarily on fish (capelin, Mallotus villosus, when available), crustaceans and polychaetes, which is consistent with previous reports (Jensen 1948; Chaput 1981; Mikhail and Welch 1989; Morin et al. 1991; Nielsen and Andersen 2001). The main prey items for G. morhua found in the present study were in accordance with results from numerous feeding studies from Newfoundland waters (e.g., Templeman 1965; Lilly et al. 1984; Paz et al. 1991; Sherwood et al. 2007; Krumsick and Rose 2012) with the exception that capelin was found in only a single stomach. Capelin are only available seasonally in coastal Newfoundland, and have had depressed stock levels since 1990 (DFO 2010). In the present study, very few capelin were observed during daily echosounding of the study area, and none spawned on the beach prior to or during the course of the study. When capelin are available, they are likely to be preyed on heavily by both species, and this predation is likely to temporarily influence the degree of diet overlap. Nonetheless, competition may still be limited, as a consequence of the typical high density of spawning capelin. Similar increases in resource sharing at times of very high prey abundance has been demonstrated across several taxa (reviewed by Schoener 1982) including among co-occurring gadid species in south-western Norway (Høines and Bergstad 1999). Furthermore, in West Greenland, in contrast to Newfoundland, capelin have a quasi-continuous distribution along the coast and undergo more limited spawning migrations (Friis-Rødel and Kanneworff 2002). West Greenland G. morhua and G. ogac both had capelin as their dominant prey (Nielsen and Andersen 2001) but it remains unclear if feeding competition exists there.

Large G. ogac did show evidence of feeding on juvenile G. morhua, as did larger G. morhua to a lesser extent. Although the present study did not attempt to evaluate predation as a potential impact on either species, large G. ogac were relatively rare in our study area, hence despite their predatory habits they may be too few to impact the overall abundance of G. morhua.

The present stable isotope results were consistent with the wider ranging records for G. morhua from northeast Newfoundland (NAFO Division 3KL) (Sherwood and Rose 2005; Sherwood et al. 2007) and represent the first records of stable isotope signatures (δ 15N and δ 13C) for G. ogac in Newfoundland waters. For both gadids, isotope values shifted from pelagic (more negative δ 13C values) to more benthic (more positive δ 13C values) with increasing body size. This shift is consistent with diet transitions from invertebrates to fish as gape size increases and young gadids are able to exploit the higher energy content of piscivorous prey (e.g., Høines and Bergstad 1999; Nielsen and Andersen 2001; Link and Garrison 2002; Sherwood et al. 2007).

Although no significant differences were found, mean total fullness indices for G. ogac were higher than for G. morhua, a pattern previously reported by Chaput (1981), who suggested that in the absence of capelin, G. ogac are more efficient predators (in terms of prey weight consumed per predator body weight). It is possible that slight differences in relative stomach fullness between species may be due to differences in preferred prey availability or from differential habitat utilization. It may also be that G. ogac has a more generalist (less discriminate) feeding approach than G. morhua. This theory is supported by dietary breadth indices that were more than twice as high for G. ogac as for G. morhua of the same size. However, it should be noted that in the absence of prey availability data indices of niche breadth must be interpreted with caution (Hurlbert 1978; Feinsinger et al. 1981; López et al. 2009).

Differences in diet between G. morhua and G. ogac could also reflect differences in pelagic habitat use that lead to differences in prey availability (e.g., Baker and Ross 1981; Shpigel and Fishelson 1989; Helland et al. 2008). Specifically, G. morhua could occupy a broader vertical distribution that encompasses both benthic and pelagic environments while G. ogac maintain a closer association with the bottom (Scott and Scott 1988). This hypothesis will be tested in a further study. In addition, the more slender body and lighter colouration of G. morhua, is suggestive of more pelagic behaviour, whereas the stouter form and darker colouration of G. ogac is consistent with more demersal habits.

In contrast to the competition hypotheses, the trophic position null hypothesis was not rejected, as 15N signatures were similar between species (Minagawa and Wada 1984; Post 2002). These results suggest that despite differing diets, G. morhua and G. ogac occupy similar trophic positions within the coastal Newfoundland ecosystem. However, it should be noted that in the absence of measured isotope values for specific prey items some degree of caution must be used in the interpretation of the results.

In conclusion, G. morhua and G. ogac, that co-occupy much of the coastal zone of Newfoundland and Labrador and other areas of the north Atlantic, appear to have similar trophic positions but limited diet competition. Our evidence provides little support for the notion that feeding competition could be limiting G. morhua recovery in these waters.