ICES Journal of Marine Science

ICES Journal of Marine Science Advance Access published March 11, 2014 ICES Journal of Marine Science ICES Journal of Marine Science; doi:10.1093/icesjms/fsu029 Settlement length and temporal settlement patterns of juvenile cod (Gadus morhua), haddock (Melanogrammus aeglefinus), and whiting (Merlangius merlangus) in a northern North Sea coastal nursery area Dorota K. Bastrikin 1 *, Alejandro Gallego 1, Colin P. Millar 1, Imants G. Priede 2, and Emma G. Jones 1 1 Marine Scotland Science, Marine Laboratory, PO Box 101, 375 Victoria Road, Aberdeen AB11 9DB, UK 2 University of Aberdeen, Oceanlab, Main Street, Newburgh AB41 6AA, UK *Corresponding author: tel: +44 1224 876544; fax: +44 1224 295511; e-mail: dorota.bastrikin@gmail.com Current address: BMT Cordah Ltd, Broadfold House, Broadfold Road, Bridge of Don, Aberdeen AB23 8EE, UK. Current address: National Institute of Water and Atmospheric Research Ltd (NIWA), PO Box 109-695, 269 Khyber Pass Road, Newmarket, Auckland 1149, New Zealand. Bastrikin, D. K., Gallego, A., Millar, C. P., Priede, I. G., and Jones, E. G. Settlement length and temporal settlement patterns of juvenile cod (Gadus morhua), haddock (Melanogrammus aeglefinus), and whiting (Merlangius merlangus) in a northern North Sea coastal nursery area ICES Journal of Marine Science, doi: 10.1093/icesjms/fsu029. Received 27 June 2013; accepted 31 January 2014. Knowledge of settlement timing and duration, which has been identified as an important milestone for demersal fish, is critical to understanding population connectivity, relevant to the development of spatially and temporally resolved conservation measures, and recruitment variability, as important density-dependent dynamics may take place at this stage. To study the settlement ecology of cod haddock, and whiting, sampling was conducted over spring and summer 2004 2006 at the northern North Sea nursery area. Over 4000 0-group juveniles were collected. Settlement was associated with clear and progressive changes in the prey composition of these juveniles. The size of fish that could be considered settled was estimated as 49 (+3) mm for cod, 78 (+4) mm for haddock, and 85 (+6) mm for whiting. Clear differences in temporal settlement patterns were also apparent. Cod settled in a single pulse lasting about a month (mid-may to mid-june) and initially occupied shallower, inshore waters, whereas haddock settled in one pulse, lasting 2 weeks (second half of May), favouring deeper, farther offshore locations. Whiting settled much later in the season and over a more protracted period (early June to early August), and their depth preferences also changed over time and with increasing length. Keywords: 0-group cod, 0-group haddock, 0-group whiting, settlement length. Introduction The timing of juvenile settlement on nursery grounds is determined by several factors. The time and duration of spawning by adult fish (Brander, 1994, 2005), water temperature-dependent development of eggs and larvae (Iversen and Danielssen, 1984; Pepin et al., 1997; Brander, 2000; Fox et al., 2003; Geffen et al., 2006), hatch date (Lapolla and Buckley, 2005), ocean currents and tides (Walford, 1938; Kingsford et al., 1991; Shenker et al., 1993; Brander and Mohn, 2004), and distance from the spawning ground (Jenkins et al., 1996), all influence when (and whether) juvenile fish will reach their nursery areas and make the transition to the demersal habitat. According to Hislop (1984) and Coull et al. (1998), the spawning period of cod (Gadus morhua) in the northern North Sea falls between January and April, haddock (Melanogrammus aeglefinus) between February/March and May, and whiting (Merlangius merlangus) between March and June/July or February and June. Within a species, spawning time and duration are influenced by maternal characteristics such as size and age (Wright and Trippel, 2009). Prey availability (Bailey and Houde, 1989; Heath, 1992; Brander et al., 2001; Beaugrand and Reid, 2003; Platt et al., 2003), growth (Campana, 1996; Gallego and # Crown copyright 2014.

Page 2 of 13 D. K. Bastrikin et al. Heath, 1997; Shima and Findlay, 2002), microhabitat selection (Lough et al., 1989; Tupper and Boutilier, 1995a; Cote et al., 2001), and inter- and intraspecific competition for food and space (Grant and Brown, 1998) can also potentially influence settlement characteristics and subsequent recruitment success. A significant correlation between indices of larval or juvenile fish abundance and subsequent recruitment was demonstrated by several studies (Helle et al., 2000; Begg and Marteinsdottir, 2002; Jonasson et al., 2009). There is no doubt that processes acting on the pelagic stages of gadoids are important controllers of recruitment success (Hjort, 1914; Walford, 1938; Cushing, 1975; Bailey and Houde, 1989; Sundby et al., 1989). However, the importance of subsequent settlement was also recognized by Campana et al. (1989), Tupper and Boutilier (1995a, b, 1997), Campana (1996), and Hüssy et al. (1997), who postulated that this transitional stage is crucial for future recruitment and that year-class strength is established at this point. The aim of the present study was to quantify (i) settlement length of cod, haddock, and whiting; (ii) the timing of settlement; and (iii) the duration of this process at the population and individual levels for these three species. Material and methods Data collection 0-group cod, haddock, and whiting were collected in Stonehaven Bay (56858 N02811 W), an inshore North Sea nursery area off the Scottish east coast (Figure 1). Stonehaven Bay has a largely sandy seabed, except three natural rocky reefs in the north, perpendicular to the shore, that rise a few metres from the bottom and consist of rough-shaped rocks of various sizes, covered with soft corals and macroalgae. The three reefs are located on the extension of a geological fault running across Scotland known as the Highland Boundary Fault (Craig, 1991). Fish were sampled around these three reefs (sites 1, 2, and 3) characterized by increasing distance offshore (800, 1600, and 2400 m, respectively) and average depths (22, 32, and 42 m, respectively). There were no overlapping depth ranges between sites for the demersal samples collected in 2004. However, there was a depth gradient within each site due to bathymetry, tide, and water level variability. Each site consisted of two different types of substrate: soft sand ( soft substrate), where demersal trawls and traps were deployed, and rocky reefs ( hard substrate), where only traps were deployed. The most intensive sampling took place in 2004 and provided weekly data on changes in the abundance and size frequency in the pelagic and demersal zone at the time of settlement for 0-group cod, haddock, and whiting. Information about spawning time, duration of the pelagic phase (Miller et al., 1963), and time and duration of settlement published previously in the literature determined the initial choice of a sampling period that extended from the end of April until the beginning of September. The limited sampling that was carried out in the 2005 and 2006 sampling seasons provided the basis for comparison of annual abundance fluctuations at the time of settlement and suggested that the 2005 sampling season was a year of higher abundance for all three species. Samples were collected in the 2004 sampling season by pelagic and demersal trawling. Pelagic sampling started on 26 April and continued weekly from 12 May to 16 August. Demersal sampling commenced on 12 May and continued in 1 2 weekly intervals (mainly weekly) until 31 August. Sampling in 2005 consisted of four demersal and two pelagic sampling events, and in 2006 of Figure 1. Location of the study site. two demersal and two pelagic sampling events. Additionally, traps were deployed on five occasions in 2005 and on three occasions in 2006. Pelagic sampling was carried out with a Methot Isaacs Kidd trawl and demersal sampling with a fine-mesh demersal trawl specifically designed to catch juvenile fish. Details of gear used in field sampling are given in Demain et al. (2011). The dataset from the 2004 sampling season was the most comprehensive and was used to investigate changes in the patterns of catch per unit effort (cpue; hereafter referred to as density ) of 0-group cod, haddock, and whiting in the pelagic and demersal zones over the settlement season. Data from the demersal trawls from 2005 and 2006 were only used for comparison with the patterns observed in 2004. Samples were subjected to taxonomic, morphometric, and stomach contents analyses, where relevant; see Demain et al. (2011) for further detail. All fish caught were identified to species level and measured. Juvenile cod, haddock, and whiting (from now on referred to as juvenile fish, unless otherwise stated) were preserved for further analyses, and the remaining fish were discarded.

Settlement length and temporal settlement Page 3 of 13 Fish total length (L T ) was measured to the nearest 0.1 mm using digital callipers, and juveniles were subsequently grouped into 5-mm size classes. Dietary analysis In total, 2391 stomachs were analysed, of which 1866 were sampled in 2004 and 525 in 2005 (Table 1). Prey items were identified to the lowest possible taxonomic level using the relevant sources (Naylor, 1972; Russell, 1976; Smaldon, 1979; Lowry and Springthrope, 2001; Meland, 2002; Pan, 2004). The resulting taxonomic resolution was considered adequate to provide a general quantitative description of the diet. Highly digested jelly like matter was recorded as unidentified and was weighed/counted as a single prey item. Empty stomachs formed an additional category. To provide a quantitative description of the diet, two methods (gravimetric and numerical) were used to analyse 2346 and 1850 fish stomachs, respectively. A comprehensive analysis of the diet was presented in Demain et al. (2011). To determine the length at settlement of cod, haddock, and whiting, prey were categorized as pelagic or benthic, and changes in their relative abundance with increasing size were examined. The relative importance of the pelagic and benthic prey categories was analysed in 5-mm size classes for juvenile fish sampled in all years and by all gear types. Prey types were classed as pelagic or benthic depending on the habitat of the prey item at the time samples were collected, based on information from the literature. For fish species with ontogenetic changes in habitat (e.g. Ammodytes spp.), this may introduce some uncertainty, which will be addressed in the Discussion section. Prey items of uncertain origin and those impossible to categorize ( unidentified ) were grouped into the category other. Two indices were used to measure changes in relative importance of the pelagic and benthic prey. The first, average weight percentage of each prey type (pelagic, benthic, or other), was calculated as: W% xy = ny j=1 W% xj n y, (1) where W% xj is the weight percentage of prey type x consumed by fish j. The second, the average number percentage of each prey type (pelagic, benthic, or other), was calculated as above, but by substituting weights by counts. The settlement threshold was set at the fish length where more than 50% of prey consumed was benthic by weight or number (Bowman, 1981). A binomial logistic regression was fitted to the dietary data, with fish length as a covariate. Logistic regression fits a logistic S-shaped curve to the proportion of prey consumed: ( ) P log = a + b length, (2) 1 P where a is the regression intercept, b a slope, and P the proportion of prey consumed. From the above model, the length at which the proportion of benthic or pelagic prey consumed equals 50% (l 50, corresponding to P ¼ 0.5) was calculated. The confidence intervals were estimated from the model and calculated as +2 s.e. The same process was followed to estimate l 50 for pelagic and benthic prey. The settlement interval is presented as the size range between l 50 pelagic and l 50 benthic. Table 1. Summary of the number of analysed juvenile fish stomachs collected by two different methods, sampled in 2004 and 2005 (see text for a description of numerical and gravimetric analyses). Type of analysis Year Cod Haddock Whiting Numerical 2004 367 396 1 103 2005 223 96 208 Gravimetric 2004 361 322 1 023 2005 128 43 20 Temporal patterns of density and length distribution To investigating the temporal settlement patterns of 0-group fish, density data from pelagic and demersal samples were analysed. T-tests were carried out to test for differences between sampling seasons. For the statistical analysis of the temporal density patterns, only those demersal trawl data collected in the same period in 2004 and 2005 were considered (i.e. between 2 June and 29 July in both years). The 2006 data were not used as they were collected over a very short sampling period and were insufficient for statistical analysis. The explanatory variables tested in density models were site, depth, time (day-of-year), and type of substrates. The appropriate tool for modelling population density data based on counts which have been collected with a variable sampling effort is a generalized mixed (linear or additive) models (Wood, 2006) using the appropriate probability distribution (Poisson with or without overdispersion and negative binomial) with a log link function and treating the log of the sampling effort as an offset variable. If we denote the underlying density at a given site and time by D site,time, then we can calculate density in terms of the observed counts N site,time and sampling effort E site,time and, therefore: D site,time = N site,time E site,time, (3) log N site,time = log D site,time + log E site,time. (4) The generalized mixed model fitted was: logn site + s(time)+ loge, (5) where density has been modelled as a function of site and a smooth function of time, s(time), represented here by day-of-year. The pelagic dataset consisted of single observations at each site over a series of weeks. Due to the scarcity of data, observations were grouped into two periods (period 1 before 15 June; period 2 after 15 June), where 15 June was the midpoint of the sampling season, and period was treated as a factor. The demersal trawl dataset consisted of weekly observations when sampling took place twice at the same site in close temporal proximity. This is known as cluster sampling and leads to correlation within the pair of samples. The random component of this model allowed the constant and depth effect to differ for each site and sampling week (time). The best predictive model was selected by forward and backward selection. Significance was assessed by the Akaike Information Criteria (AIC), as follows. The model chosen was the one with the fewest parameters within 2 units of the lowest observed AIC for

Page 4 of 13 D. K. Bastrikin et al. the dataset in question. This use of AIC for model selection has been discussed by Jones (1993). Significance of individual factors was tested using a Wald test (Agresti, 1990), a multivariate version of z-test, which tests for a multivariate difference from zero. Samples collected by demersal trawls in 2004 were used to investigate changes in the length distribution of 0-group fish over time (day-of-year) on the three sites. These data were analysed using linear regression and analysis of variance (ANOVA). The appropriate distribution for length data is Gaussian (Sokal and Rohlf, 1995; Zuur et al., 2007). The assumptions of ANOVAwere tested to ensure that independence, normality, and homogeneity criteria were satisfied. The best predictive model was chosen by backward selection from the most complex model, and forward selection from the model including only the global mean. The selected model was the one with fewest parameters within 2 units of the lowest AIC (Jones, 1993). Results Foraging habitat and length of settlement Cod Cod smaller than 30 mm preyed exclusively on pelagic food items that consisted entirely of copepods, mainly Temora longicornis and Calanus finmarchicus. From a length of 30 mm, juvenile cod began to feed on benthic prey items, such as Crangon crangon, megalopa larvae of crabs, fish (Ammodytes spp.), and euphausiids. However, pelagic food still constituted over 75% of the diet weight (84% of prey numbers). Between 30 and 45 mm, the importance of pelagic food decreased; by 50 mm, over 60% of consumed prey weight consisted of benthic animals. At this length, the range of benthic prey consumed expanded to include prawns, shrimps, molluscs, cumaceans, crab larvae (megalopa), amphipods (mainly caprellids and gammarids), and plaice (Pleuronectes platessa). The average pelagic food fraction by weight dropped below 1% by the 70-mm length class, and benthic prey constituted.80% of prey weight. The average abundance of pelagic prey by numbers in the 70-mm length class dropped to,10%. Benthic prey numbers were maintained at near 70% or more. Generally, cod over 85 mm did not feed on pelagic prey. The exceptions were observed only in length classes with small numbers of stomachs analysed (80, 95, 110, 115, and 135 mm), but even then, the decrease in the content of benthic components coincided with increases in unidentified highly digested matter, not pelagic prey. The changes in relative weight and numbers of pelagic and benthic prey indicate that cod started feeding on benthic prey at 30 mm and by.85 mm, the diet was exclusively benthic. Cod settlement was a gradual process that took place in the length interval 34 (+2) 49 (+3) mm; by 49 (+3) mm (Figures 2a and 3a), as calculated from the logistic regression model, cod juveniles were considered settled. Haddock Juvenile haddock in length classes,20 mm fed exclusively on pelagic prey (100% of prey by weight and numbers), mainly Limacina spp., small copepods, T. longicornis, C. finmarchicus, and ostracods. In the 20 30 mm interval, the range of crustaceans consumed increased and included amphipods (hyperids), euphausiids (larvae calyptopis and furcila), eggs of invertebrates, zoea larvae of crabs, increasing numbers of copepods, particularly C. finmarchicus, T. longicornis, and Pseudocalanus spp. Insects were also occasionally Figure 2. Proportion of benthic prey items by weight in the diet of 0-group cod (a), haddock (b), and whiting (c). Data points represent the percentage of benthic prey in individual fish. Curved solid lines are the logistic regression model fit. Vertical solid lines indicate the estimated l 50. Dotted lines indicate +2 s.e. confidence intervals. observed. In this length interval, prey items classed as benthic were encountered in stomachs for the first time. In all instances in this size range, the benthic prey components were Ammodytes spp. (in the 30-mm length class, they constituted over 60% of consumed weight) and were found among juvenile haddock sampled by pelagic tows on 24 May 2004. From the 35-mm and above length class, megalopa larvae of crabs, polychaetes, adult euphausiids, and benthic fish (Pleuronectiformes) began to be consumed. In the 50 55 mm length range, there was an abrupt drop in the proportion of pelagic prey consumed. The average pelagic prey by weight dropped from 29.6% in the 50-mm length class to 1.9% at 55 mm and was,3% for all subsequent length classes. The average pelagic prey numbers fell from 50.1% at 45 mm to 25% at 55 mm, to 7.7% by 80 mm, and fluctuated between 0 and 3% at the 100-mm length class. Despite the decrease in the consumption of pelagic prey, an increase in the proportion of benthic prey was not clearly apparent, mainly due to the increase in unidentified stomach content. Particularly in the 55 85 mm length classes, unidentifiable highly digested matter constituted between 50 and 85%

Settlement length and temporal settlement Page 5 of 13 of total prey weight. The majority of haddock with highly digested stomach content were sampled on 15 June 2004, 15 July 2004, and 23 June 2005. The benthic components constituted over 50% of the diet at 78 (+4) mm (by numbers; Figure 3b), as calculated from the logistic regression model, the length haddock were considered settled. The proportion of pelagic food in stomachs of haddock decreased below 50% at 29 (+6) mm (by weight). Therefore, haddock settlement was also a gradual process, which took place in the 29 (+6) 78 (+4) mm length interval. Whiting Juvenile whiting up to 25 mm fed exclusively on a pelagic diet consisting of T. longicornis and small copepods (mainly C1 C4 stages of C. finmarchicus). From 20 mm, adult stages of C. finmarchicus and Limacina spp. were also present in the stomachs. Whiting as small as 25 mm also consumed Ammodytes spp. and occasional benthic prey items (polychaetes). From 35 mm, amphipods were also included in the diet and, from 45 mm, crab megalopa. Small copepods and T. longicornis, along with larger species such as C. finmarchicus and Candacia armata, were the most important items in the diet of smaller whiting (up to 45 mm), but they were still present in significant numbers in larger length classes. Pelagic food items constituted over 50% of the diet up to a fish length of 30 mm by weight and 40 mm by number. The one exception, in stomachs analysed by the gravimetric approach, was in the 30-mm length class, where benthic prey constituted 83.3% by weight. Such a large proportion of benthic prey in this length category was mainly due to Ammodytes spp., which had been classified as benthic. All 25 30 mm juvenile whiting that preyed on Ammodytes spp. originated from a single pelagic tow on 24 May. As whiting length increased further, so did the variety of prey consumed. Larger, heavier prey items such as megalopa larvae of crabs, juvenile crabs, prawns, shrimps, amphipods, cumaceans, and fish made up most of the prey weight. From the 35-mm length class and upwards, the consumption of benthic prey gradually increased to exceed the 50% threshold at 85 (+6) mm (by weight), as calculated from the logistic regression model. Benthic prey constituted the majority of food of 0-group whiting, by weight and number, from these size classes onwards. An exception was noted at 150 mm, with three fish analysed numerically only, where benthic food content fell to 19.4%. However, this coincided with an increase in unidentified matter, not an increase in pelagic prey. Pelagic prey consumption by whiting juveniles dropped below the 50% level at 29 (+6) mm (by weight). Settlement of whiting was a gradual process that took place in the 29 (+6) 85 (+6) mm size interval; by 85 (+6) mm (Figures 2c and 3c), whiting could be considered settled. Pelagic prey numbers constituted on average over 20% of juvenile whiting prey up to the 115-mm length class, and even in larger length classes, they accounted for a significant amount of prey (only twice below 9% at 130 and 160 mm). In stomachs analysed gravimetrically, the average proportion of pelagic prey by weight fell below 10% by 60 mm (with an exception at 110 mm) and below 1% by 130 mm. This indicates that pelagic prey were many contributors to the juvenile whiting diet through the entire length spectrum. No stomachs were analysed gravimetrically in the 10 20 and 150 170 mm length classes. Temporal patterns of density and length distribution Cod Except 2005, cod were the least abundant demersal juvenile gadoids of the three species of interest. During 2004 2006, 715 cod (15 160 mm) were sampled (Table 2). Cod were first caught in the pelagic samples (Figure 4a) on 17 May (day 138) and last on 14 June (day 166). On the basis of the mixed model analysis, it was concluded that there was a significant time effect on pelagic Figure 3. Proportion of benthic prey items by numbers in the diet of 0-group cod (a), haddock (b), and whiting (c). Data points represent the percentage of benthic prey in individual fish. Curved solid lines are the logistic regression model fit. Vertical solid lines indicate the estimated l 50. Dotted lines indicate +2 s.e. confidence intervals. Table 2. Summary of the total catches of cod, haddock, and whiting over the 2004 2006 sampling seasons, sampled by all types of gear. Type of sampler Year Cod Haddock Whiting Pelagic trawl 2004 10 152 96 2005 2 0 2 2006 0 0 0 Demersal trawl 2004 455 822 2 248 2005 190 95 209 2006 1 30 200 Trap 2004 N/A N/A N/A 2005 30 0 1 2006 27 0 0

Page 6 of 13 D. K. Bastrikin et al. cod density (Wald test, p, 0.05; Table 3). The density of juvenile cod in the pelagic zone was significantly higher (t-test, p, 0.05) before mid-june (first period) than thereafter (second period). Their occurrence in the pelagic zone in 2005/2006 was consistent with the patterns observed in 2004 (Figure 4a). When sampling in 2005 and 2006 took place, no fish were expected to be present in the pelagic zone. This was generally the case, except 27 July 2005, when two 40-mm cod juveniles were captured. Cod were found in the demersal catches from the beginning of sampling (12 May; day 133), a week earlier than in pelagic samples. The density of demersal juvenile cod changed significantly over time (Wald test, p, 0.05; Table 4) on all three sites (Wald test, p, 0.05; Figure 5a). The smallest cod was 15 mm long. The length of cod juveniles increased through the season. The maximum length of cod juveniles in the pelagic samples was 40 mm, a length recorded for the first time on 24 May (day 145; Figure 6a). In 2004, cod were the smallest gadoids recorded on the seabed, at a length of 20 mm (Figure 7a). The occurrence of cod in the 20 30-mm length range was quite common in the demersal zone through May and until mid-june. In 2004, in terms of maximum length, cod were the smallest of the three species. From mid-june (day 166), an increase in the minimum and median size was recorded. This coincided with the disappearance of cod from the pelagic catches. Cod reached a maximum size of 100 mm in June, 125 mm in July, and 135 mm in August (over all seasons and all types of sampling gear). The lengths of demersal 0-group cod in 2005 and 2006 were consistent with the patterns observed in 2004, although in 2005, a significantly higher abundance of juvenile cod was recorded (t-test; p, 0.05). The length distribution data from 2005 and 2006 obtained by demersal trawling and traps provide additional information to complement the 2004 dataset. Data from 2005 were obtained by trawls and traps, and for 2006 mainly from traps, as only one cod was caught in the demersal trawls (27 July; day 208). The juvenile cod length data for 2005 and 2006, from mid-june (day 174) until the end of sampling (day 251), showed a pattern of gradual increase in minimum and median length (Figure 7a). Statistical analysis of the combined demersal cod dataset showed that juvenile size was positively correlated with time (day-of-year; ANOVA, p, 0.0001; Table 5), and it also indicated that there was a gear selectivity effect (traps caught bigger fish than trawls), and a year effect (fish caught in 2005 and 2006 were bigger than in 2004; ANOVA, p, 0.0001). Therefore, the combination of all data sources in Figure 7a is just for illustrative purposes. Most cod settled at the shallowest depth, closest to the shore (site 1) and at the intermediate depth and distance offshore (Site 2) at the beginning of the sampling season, and subsequently the numbers of cod declined on both sites. At site 3, the deepest and farthest offshore, the density of fish also declined over time. The density of cod at this site was significantly lower (t-test, p, 0.05) than at the two other sites. There were significant (F-test, p, 0.001) changes in fish length over time and significant differences (F-test, p, 0.001) between sites. Cod had a larger mean length at deeper sites (site 2, t-test, p, 0.001; and site 3, t-test, p, 0.05) than at site 1. The length Figure 4. Catches of cod, haddock, and whiting in pelagic trawls in 2004 2006. Data points indicate square root transformed mean number of fish, standardized to number per hour effort, on the sampling day. Error bars indicate +2 s.e. Table 3. Model choice summary for pelagic abundance data for cod, haddock, and whiting. AIC value Model terms Cod Haddock Whiting Whiting* Constant 49.9 127.5 144.1 136.9 Site 51.4 130.0 143.6 139.6 Period 42.1 109.9 144.3 138.3 Period + site 43.8 112.1 142.5 140.6 Period site 47.8 116.1 140.9 138.9 The best fitting models are indicated in bold. Whiting* was an alternative model for whiting (see text).

Settlement length and temporal settlement Page 7 of 13 Table 4. Model choice summary for abundance of cod, haddock and whiting in demersal trawls. AIC value Model terms Cod Haddock Whiting Constant 198.2 283.4 295.5 Site 196.3 265.0 298.9 Depth 196.8 272.2 297.2 Time 171.4 285.1 251.8 s(time) N/A N/A 248.3 Depth + time 168.0 N/A 252.2 Depth + s(time) N/A N/A 248.0 Site + time 164.0 266.0 252.8 Site + s(time) N/A 257.5 247.1 Site:s(time) N/A N/A 243.0 Site + site:s(time) N/A 265.1 245.8 Site + depth 196.1 267.0 N/A Site + depth + time 163.7 N/A N/A Site + s(time) + depth N/A 259.5 N/A Depth site + s(time) 162.8 N/A 247.7 The best fitting models are indicated in bold. distribution, combined with abundance patterns in pelagic and demersal zones, indicates that cod probably settled in a single pulse lasting from mid-may to mid-june. Haddock During 2004 2006, 1099 haddock (10 155 mm) were sampled (Table 2). Haddock were the most abundant gadoid juveniles in the 2004 pelagic catch and were sampled for the first time on 26 April (day 117), the earliest of all three gadoid species, at the beginning of the sampling period. Numbers of 0-group haddock subsequently increased, reaching a peak of pelagic abundance on 17 May (day 138). Subsequently, the abundance of haddock declined, and they became absent from pelagic samples by 7 June (day 159; Figure 4b). Time was the only significant effect (Wald test, p, 0.05; Table 3), with juveniles significantly more abundant in the first period (t-test, p, 0.05). From the patterns of abundance of pelagic haddock observed in 2004, it was not expected to find any haddock juveniles in the pelagic zone when sampling was carried out in 2005 and 2006, as it was indeed the case. Haddock were present in the demersal catches from mid-may, at the beginning of demersal sampling. The increase in the numbers of juvenile haddock in the demersal catches coincided with their decrease in the pelagic catches. The density of haddock changed over time (Wald test, p, 0.05; Table 4), and there were significant differences between sites (Wald test, p, 0.05; Table 4). Time was fitted as a smooth function, as the density changes over time were continuous, but non-linear. The plot (Figure 5b) of the predicted changes in the density of 0-group haddock with time at different sites (depths) shows that there was a sudden increase in the abundance of juveniles, particularly in deeper waters (sites 2 and 3), from the beginning of June (day 154) until the peak in abundance at the beginning of July (day 188). It was followed by a decline in the numbers of haddock to an almost constant level from mid-july (day 197). Intermediate depths (site 2) and, to a lesser degree, deeper waters (site 3) seemed to be particularly favoured. The density of haddock at site 1 was significantly lower (t-test, p, 0.05) than at the other sites. At the beginning of the sampling season, fish were present at all three sites, but after 6 July (day 188), haddock were caught only at sites 2 and 3. In the 2005 sampling season, a Figure 5. Mixed models outputs of the temporal changes in demersal juvenile cod (a), haddock (b), and whiting (c) density (solid lines) on different sites (site 1, red lines; site 2, black; site 3, blue). Density is expressed as numbers of juvenile fish per hour of sampling effort. Dashed lines indicate +2 s.e. confidence intervals. Dots represent individual data. significantly higher abundance of demersal juvenile haddock was observed than in 2004 (t-test; p, 0.05; Figure 4b), but overall, results in 2005 and 2006 were consistent with the patterns of abundance of demersal haddock in 2004. There were significant differences in demersal haddock lengths between different sites (F-test, p, 0.001; Table 5). There was also a significant interaction term between site and time (F-test, p, 0.001; Table 5), indicating that the pattern of temporal changes in length distribution differed between sites. From the beginning of the sampling season until 6 July (day 188), there was a similar length-change pattern at all sites (Figure 5b). After this

Page 8 of 13 D. K. Bastrikin et al. Figure 6. Length distribution of 0-group cod (a), haddock (b), and whiting (c) caught in the pelagic zone in the 2004 sampling season. Boxes indicate the 25th and 75th percentiles of all sizes measured. The upper bars indicate the 10th and the lower bars the 90th percentiles. The thick line indicates the median size. Dots indicate the outliers. date, fish of increasing length were sampled at sites 2 and 3, but were not present at site 1. Fish at site 1 were significantly smaller than at the other two sites (t-test, p, 0.001). Fish with the largest mean length were found initially at intermediate site 2 then, towards the end of sampling season, at site 3, the deepest and farthest away from shore. The smallest pelagic haddock (10 mm) were present in the samples on 26 April (day 117). The length of pelagic juveniles subsequently increased, reaching a maximum of 45 mm on 17 May (day 138; Figure 6b). From the beginning of sampling until 25 May (day 146), the smallest demersal juveniles (30 45 mm) were present in the catch (Figure 7b). From 2 June (day 154), there was an increase in the minimum size of haddock, which continued until the end of sampling. This, combined with the fact that the last pelagic haddock was sampled on 1 June (day 153), suggested that they settled in one pulse that lasted until late May early June. The length distribution of haddock juveniles in 2005 and 2006 was similar to the patterns observed in 2004 (Figure 7b). Whiting Whiting was the most abundant of the three species of interest in all 3 years of study. During 2004 2006, 2756 whiting (10 170 mm) were sampled (Table 2). They were first caught in the pelagic zone on 12 May 2004 (day 133), and the peak in abundance was observed on 17 May (day 138). The last pelagic whiting were caught on 27 July 2004 (day 208; Figure 4c). Exploratory models had indicated that there may be significant time and site effects on juvenile density (Table 3). Small differences between AIC values in all models tested led to further analysis. New model fitting was carried out after removing the most extreme observation (whiting* model in Table 3). This analysis led to the conclusion that there were no significant time or site effects and that pelagic whiting density was constant throughout the area during the sampling period. In the 2005 sampling season, the occurrence of juvenile whiting in the pelagic zone was consistent with the pattern observed in 2004. Whiting were found in the 2004 demersal samples from the beginning of June, 3 weeks after they were first detected in the pelagic zone. The density of 0-group whiting increased throughout the sampling season (Figure 5c). The patterns of density over time were different among sites (indicated by the interaction between site and time in the model; Wald test, p, 0.05; Table 4) and showed a rapid increase in the number of settling juveniles from 8 June (day 160). Until 27 June (day 181), they were present exclusively at site 1 then were caught at all sites. The density of whiting at site 1 from 15 July (day 197) reached a plateau at a level much lower than at the two other sites. At site 2, after 6 July (day 188), numbers of fish increased rapidly, decreased, then levelled off at lower densities than at site 3. Although the increase in the density of whiting at site 3 was delayed relative to site 2, their numbers increased in time to higher levels than at site 2. Overall, juvenile whiting at sites 2 and 3 were significantly (t-test, p, 0.05) more abundant than at site 1. The abundance patterns of juvenile whiting were consistent in 2004 2006 demersal catches. In 2005, whiting abundance was significantly (t-test; p, 0.05) higher than in 2004. Pelagic juvenile whiting ranged from 10 to 60 mm in size in 2004. Between the beginning of sampling and 14 June (day 166), a gradual increase in the maximum size of juveniles was recorded. Throughout the sampling season, small whiting were consistently present in the pelagic samples (Figure 6c). The size of juveniles in demersal samples ranged between 30 and 145 mm. There were significant differences in fish length between sites (F-test, p, 0.001; Table 5) and a significant interaction term between site and time (F-test, p, 0.001; Table 5), indicating that length distributions changed with time with a different pattern between sites. Although before 27 June (day 181), small whiting were caught only at site 1, fish caught at the deeper sites 2 and 3 were overall significantly smaller than at site 1 (t-test, p, 0.05 and,0.001, respectively). The size distribution of whiting, particularly the continuous presence of small juveniles until the end of July, combined with abundance patterns in the pelagic and demersal habitats indicate a protracted population settlement pattern lasting from the beginning of June until the beginning of August. The length distribution patterns of juvenile whiting in the demersal catches in 2005 and 2006 were consistent with the patterns observed in 2004. Discussion Similar to other developmental processes, settlement does not occur at a fixed length, but happens gradually and can be influenced by a range of factors such as resource availability and body reserve. In this

Settlement length and temporal settlement Page 9 of 13 Figure 7. Length distribution of 0-group cod (a), haddock (b), and whiting (c) caught in the demersal zone (combined data for 2004 2006 catches from demersal trawls and traps). Boxes indicate the 25th and 75th percentiles of all sizes measured. The upper bars indicate the 10th and the lower bars the 90th percentiles. The thick line indicates the median. Dots indicate outliers. Black colour indicates juveniles caught in 2004, blue in 2005, and red in 2006. On day 208 in 2006, only one cod was sampled by a demersal trawl. Table 5. Model choice summary for the length distribution of cod, haddock, and whiting caught by demersal trawls in 2004. AIC value Model terms Cod Haddock Whiting Constant 1 174.2 3 382.8 8 022.4 Site 1 159.9 3 310.8 7 959.4 Time 1 061.7 2 243.1 7 512.9 Site + time 1 031.8 2 239.3 7 513.9 Site:time + time 1 031.2 2 238.8 7 513.2 Site:time + site 1 033.2 2 235.9 7 469.3 Site time 1 033.2 2 235.9 7 469.3 The best fitting models are indicated in bold. study, probabilistic estimates of size at settlement were obtained using a logistic regression technique. Juvenile cod occurred in the pelagic zone between 17 May and 14 June 2004, but were found in demersal catches from 12 May, suggesting that pelagic cod must have also been present before that date. The very small numbers of 0-group cod caught in the pelagic zone and the absence of juveniles in catches before 17 May were most likely a result of the dispersion of juveniles and their low abundance. The peak in the abundance of pelagic cod was in mid-may, followed by their final disappearance from the pelagic zone by mid-june. Cod were found in the demersal samples throughout the whole sampling period in 2004, with a declining pattern of abundance over time. The drop in abundance may be due to a potential switch to rocky reef habitat by juvenile cod and/or mortality of young settlers. A preference for structured habitat that provides cover from predation was previously reported by Grant and Brown (1998), Lindholm et al. (1999 2001), Laurel et al. (2003), and Juanes (2007), among others. Attempts at trap sampling on different bottom substrates during the 2004 sampling season were unsuccessful at capturing any juveniles, so no further information is available. In this study, haddock was the first species to appear (on 26 April), reaching peak of abundance in the pelagic zone on 17 May and finally disappearing from the pelagic zone by the first week of June. This short period of pelagic presence suggests that haddock settled in one single pulse. Coinciding with the decline in pelagic haddock numbers, there was an increase in the demersal zone. This reverse pattern of abundance between the two habitats suggests that settlement in the study area occurred during 1 2 weeks and was completed by the beginning of June. It is not possible to be absolutely certain that there were no haddock present on the seabed before 12 May 2004. An earlier presence of demersal juveniles would imply that settlement had started earlier. However, there was only a single 10-mm specimen sampled in the pelagic zone on 26

Page 10 of 13 D. K. Bastrikin et al. April and two demersal haddock sampled on 12 May, suggesting that any significant earlier settlement was unlikely. After a period of high abundance between mid-june and the beginning of August, the numbers of haddock in the demersal catches dropped. This decline could be due to the movement of growing juveniles into deeper waters (Fulton, 1890; McIntosh, 1897; Klein-MacPhee, 2002), as haddock are generally associated with deeper waters (Klein-MacPhee, 2002), and/or mortality of newly settled juveniles. Whiting had the most protracted settlement period among all three species investigated. Juveniles were sampled in the pelagic zone from 12 May to 27 July 2004. This was consistent with the general knowledge of the biology of the species, which is known to have an extended spawning season lasting until June July (Hislop, 1984). Whiting appeared in the demersal catches from 8 June 2004 until the end of the sampling season. The patterns of demersal abundance, pelagic abundance, and length distribution patterns all point towards a very extended period of settlement lasting from the beginning of June until the beginning of August. The length distribution of 0-group cod, haddock, and whiting in 2004 confirmed the general patterns derived from the abundance data. There was an initial inflow of juvenile cod in the pelagic zone, with the largest cod sampled being 40 mm long. Specimens as small as 20 25 mm were regularly caught in the demersal zone until mid-june. There was an increase in the size of cod with time in the trawl samples, and the largest cod were found at site 2 at intermediate depths. A further increase in size at the deeper site 3 was not observed, but this could be due to the termination of sampling before any changes could be observed, especially since the increasing affinity of cod for deeper waters with age is well documented (Riley and Parnell, 1984; Tremblay and Sinclair, 1985; Keats et al., 1987; Sinclair, 1992; Hessen, 1993; Linehan et al., 2001). The presence of the smallest haddock length class (10 mm) was observed until mid-may in the pelagic zone and, from the second half of May, they were also sampled in the demersal zone (30 mm). The smallest fish were observed in the demersal zone only for the initial 2 weeks of sampling. After that period, a gradual increase in the size of haddock was observed which, together with the absence of juveniles in the pelagic zone, indicated that no new settlers were present. Small new whiting arrivals were recorded in the pelagic catches through most of the settling season, close to the time of their disappearance from the pelagic zone. The demersal fish size distribution showed the presence of small juveniles between the beginning of June and the end of July. This period of settlement agrees with that derived from the abundance data. The presence of 40 60 mm juveniles in the pelagic catches could be an indication that whiting in the area were also undertaking vertical movements after settlement. Whiting settlement took place over a relatively small size range. The dietary analysis indicated that settlement, defined by the switch from pelagic to benthic prey, was a gradual process that, for cod, occurred at the size range from 34 (+2) mm until final settlement at 49 (+3) mm, a smaller size than previously reported (Daan, 1973, 50 90 mm; Bowman, 1981,.90 mm; Hüssy et al., 1997, 50 70 mm; Lomond et al., 1998, 60 100 mm). For haddock, it occurred at the size range from 29 (+6) mm until settlement at 78 (+4) mm, and for whiting from 29 (+6) mm until it was completed by 85 (+6) mm, also at smaller sizes than previously reported (Heincke, 1905; Robb and Hislop, 1980). Juvenile haddock of the size range 25 30 mm that preyed on Ammodytes spp. originated from a single pelagic sample taken on 24 May 2004. This suggests that they were very likely feeding on pelagic forms of Ammodytes spp., found in the water column at that time of year (end of May) immediately after metamorphosis, before descending to the seabed (Peter Wright, Marine Scotland Science, pers. comm.; and unpublished data from MSS coastal environment monitoring station in Stonehaven). Therefore, the proportion of benthic prey in these size classes may have been an artefact of the classification of prey items, which did not distinguish between developmental stages of Ammodytes spp., as the appearance of pelagic post metamorphic and early demersal stages is very similar. To test for this effect, additional analyses excluding the juvenile size classes that preyed on these Ammodytes spp. sizes were carried out. The results showed that, for haddock, the pelagic prey consumption still dropped below 50% at 29 (+3) mm, but benthic prey consumption increased above 50% level at 88 (+3) mm. This range is similar to the original estimate, but suggests a potential underestimate of the maximum haddock settlement size in the original analysis (78 +4 vs. 88 +3 mm). An equivalent bias is unlikely to have taken place in the cases of cod and whiting, as the number of juveniles of the relevant size categories that fed on Ammodytes spp. was very small and did not affect the statistical outcome. All the evidence suggests that juvenile cod settle initially close to the shore in shallow waters and move deeper as size increases. This results in at least a partial size segregation, which can reduce cannibalism and increase the chances of survival of 0-group cod (Riley and Parnell, 1984). Although a pattern of increasing numbers in deeper waters coinciding with the observed decline in the shallowest depths was not evident in our data, this in itself does not exclude the possibility of the hypothesized offshore movement without being able to rule out other factors that may reduce numbers farther offshore, such as juvenile mortality. The size distribution pattern observed suggests that, in 2004, there may have been a single pulse of cod settlement lasting 1 month between mid-may and mid-june. This was consistent with the abundance data, where the presence of pelagic juveniles was recorded until mid-june. The subsequent absence of small fish indicates the increasing trend in length distribution was likely due to fish growth. Due to very low catches of cod in July and August, it is difficult to be absolutely certain that there were no more small juveniles settling. However, the absence of juvenile cod from the pelagic catches after mid-june supports the notion that the transition of cod to the seabed was completed by this time. The maximum size of pelagic juveniles, combined with the most common size of small cod in the demersal zone, also supports our conclusions. Juvenile haddock favoured deeper waters than cod. These depth differences were apparent from the very beginning of the settlement period, when cod favoured the shallower site 1, whereas haddock settled in greatest numbers at site 2 at intermediate depths. Haddock disappeared entirely from the shallowest site from the beginning of July. After an initial increase in abundance, there was a slight decline in the numbers of juvenile haddock towards the end of the season, especially at intermediate depths (site 2). This decline could be due to mortality, a shift to structured habitat, movement of growing individuals out of the study area into deeper waters, or a combination of the above. The size of haddock increased with depth, with the largest fish found at site 3, confirming that larger haddock show an affinity towards deeper waters, farther away from shore (Fulton, 1890; McIntosh, 1897; Klein-MacPhee, 2002). Throughout June, whiting were found only at the shallowest site. After that time, their numbers increased rapidly at the deeper sites. At the shallowest site, the density of whiting reached a plateau by

Settlement length and temporal settlement Page 11 of 13 mid-july at low densities. From mid-july, the highest densities of whiting were found at intermediate depths and, towards the end of the sampling season, on the deepest, most distant site from shore. Small whiting were present in the area from the beginning of June, several weeks after the first cod were caught, and for the first month were found only at the most inshore, shallow site. Throughout most of the sampling season, the largest fish were found at the shallowest site, closest to shore. The abundance of cod and haddock in 2005 was significantly higher than in 2004, although these results must be treated with caution, owing to the very limited number of samples taken in 2005. These findings were, however, consistent with data from the International Bottom Trawl Survey (IBTS), the Scottish Ground Fish Survey (SCOGFS), and the English Ground Fish Survey (ENGGFS; ICES 2007a, b). They all indicate that, from the estimates of 0-group cod and haddock recorded in 2005 and 1-group cod and haddock recorded in 2006, the 2005 year-class abundance was higher in the North Sea, particularly in the central and northern part, than recent low levels (for haddock about tenfold higher than the average for the 2001 2004 year classes). IBTS data showed the highest numbers of 0-group cod in quarter 3 (July September) in 2005 since 1998, and the highest numbers of 1-group cod in quarter 1 (January March) in 2006 since 2002. SCOGFS and ENGGFS data showed the highest numbers of 1-group cod in quarter 3 in 2006 since 2000 and 1998, respectively. According to the IBTS dataset, haddock had a moderate 0-group year class in quarter 3 in 2005 (compared with the high 0-group class abundance in 1999), comparable in size to the 2000 year class, and consistently large numbers of 1-group haddock were sampled in quarter 1 (January March) in 2006. No notable differences in abundance between the years were recorded in IBTS, SCOGFS, or ENGGFS data for whiting. Conclusions Differences in the timing of settlement among the three gadoid species investigated have been identified, with cod being the first species to commence settlement followed by haddock then whiting. There were also considerable differences in the duration of the process. Haddock settled in the Stonehaven area in one short pulse lasting 2 weeks, while cod settled over 1 month, and whiting over 2 months. The patterns of length distribution changes through time were different for all three species. Juvenile cod, haddock, and whiting in the Stonehaven area displayed patterns of distribution with depth and distance from shore that led to species and size segregation in time and space, minimizing the potential for competition. The analysis of the feeding ecology of these species showed major differences in dietary composition and little evidence of juveniles preying on each other (Demain et al., 2011), which further supports this conclusion. These factors, taken together, suggest a degree of niche separation which would facilitate the coexistence of these three species. Settlement of cod, haddock, and whiting appears to be a gradual process which may start with exploratory migrations towards the seabed, during which the juveniles start feeding on benthic prey; as they grow, they increasingly specialize in benthic feeding, although whiting continue to feed in the water column even at larger sizes. A number of authors referred to exploratory migration before actual settlement in the North Sea (Russell, 1922; Bromley and Kell, 1999). The presence in demersal samples of very small juveniles still feeding on pelagic prey is consistent with that exploratory behaviour. The settlement size range calculated for cod in this study points towards a transition to demersal life at smaller sizes than previously reported for North Sea cod (45 65 mm, Robb and Hislop, 1980; 40 80 mm, Bromley and Kell, 1999). Also, the settlement of haddock and whiting observed in this study took place at smaller sizes than those reported in the literature (Heincke, 1905; Robb and Hislop, 1980). Perhaps, a hypothetical tendency to smaller settlement sizes is driven by the same fishery-induced evolutionary process that causes demographic changes in heavily fished populations (Wright and Trippel, 2009). In exploited populations, fish display a large proportion of first-time spawners (Wigley, 1999; Morgan et al., 2003) and, as a consequence, have a narrower spawning season that can affect their reproductive success (Wright and Trippel, 2009) and mature at a younger age (Trippel, 1995). This earlier development may also be found during the early life stages and result in smaller settlement lengths as well. However, to test this hypothesis would require further investigations into the timing and size at settlement on a much wider geographical and temporal scale than in the present study. Alternatively, the differences between these settlement sizes and some of those reported in the literature may be the result of differences in the definition of settlement among authors, and/or more limited datasets. References Agresti, A. 1990. Categorical Data Analysis. John Wiley and Sons, New York. Bailey, K. M., and Houde, E. D. 1989. Predation on eggs and larvae of marine fishes and the recruitment problem. Advances in Marine Biology, 25: 1 83. Beaugrand, G., and Reid, P. C. 2003. Long-term changes in phytoplankton, zooplankton and salmon linked to climate change. Global Change Biology, 9: 801 817. Begg, G. A., and Marteinsdottir, G. 2002. Environmental and stock effects on spawning origins and recruitment of cod Gadus morhua. Marine Ecology Progress Series, 229: 263 277. Bowman, R. E. 1981. Food of 10 species of Northwest Atlantic juvenile groundfish. Fishery Bulletin US, 79: 200 206. Brander, K. M. 1994. Spawning and life history information for North Atlantic cod stocks. ICES Cooperative Research Report, 205. 150 pp. Brander, K. M. 2000. Effects of environmental variability on growth and recruitment in cod (Gadus morhua) using a comparative approach. Oceanologica Acta, 23: 485 496. Brander, K. M. 2005. Cod recruitment is strongly affected by climate when stock biomass is low. ICES Journal of Marine Science, 62: 339 343. Brander, K. M., Dickson, R. R., and Shepherd, J. G. 2001. Modelling the timing of plankton production and its effect on recruitment of cod (Gadus morhua). ICES Marine Science Symposia, 58: 962 966. Brander, K., and Mohn, R. 2004. Effect of North Atlantic Oscillation (NAO) on recruitment of Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences, 61: 1558 1564. Bromley, P. J., and Kell, L. T. 1999. Vertical migration and spatial distribution of pelagic 0-group gadoids (cod, haddock, whiting and Norway pout) prior to and during settlement in the North Sea. Acta Adriatica, 40: 7 17. Campana, S. E. 1996. Year-class strength and growth rate in young Atlantic cod Gadus morhua. Marine Ecology Progress Series, 135: 21 26. Campana, S. E., Frank, K. T., Hurley, P. C. F., Koeller, P. A., Page, F. H., and Smith, P. C. 1989. Survival and abundance of young Atlantic cod (Gadus morhua) and haddock (Melanogrammus aeglefinus) as indicators of year-class strength. Canadian Journal of Fisheries and Aquatic Sciences, 46: 171 182.