Spatial structure of body size of European flounder (Platichthys flesus L.) in the Baltic Sea

Fisheries Research 189 (2017) 1–9

Contents lists available at ScienceDirect

Fisheries Research journal homepage: www.elsevier.com/locate/fishres

Spatial structure of body size of European flounder (Platichthys flesus L.) in the Baltic Sea Johan Erlandsson, Örjan Östman, Ann-Britt Florin, Zeynep Pekcan-Hekim ∗ Swedish University of Agricultural Sciences, Department of Aquatic Resources, Institute of Coastal Research, Skolgatan 6, 742 22 Öregrund, Sweden

a r t i c l e

i n f o

Article history: Received 5 August 2016 Received in revised form 30 December 2016 Accepted 2 January 2017 Handled by George A. Rose Keywords: Age-length distributions Fish management Maturation Spatial structure Stock identification

a b s t r a c t The spatial structure of fish species is important for stock identification and management. The European flounder (Platichthys flesus L.) shows morphological differences across the Baltic Sea Proper. However, it is not known whether flounders cluster into several distinct areas based on morphological characters, indicating discrete sub-populations, or whether they show continuous morphological variation along space indicating a more continuous population structure. Here, we study the spatial structure of body length and length-at-age distributions of the European flounder (Platichthys flesus L.) across the Baltic Sea Proper (International Council for the Exploration of the Sea (ICES) subdivisions 25–28) using high spatial resolution data (ICES rectangles) from fishery independent surveys 2008–2014. Our results are in agreement with genetic data suggesting a continuous gradient of decreasing body length from southwest to north-east. Further, we observed distance decay in the spatial synchrony of temporal changes in the length distributions, such that the temporal trends were correlated among adjacent ICES rectangles but independent across the whole study area. Length-at-age and maturity patterns that were calculated for each subdivision also showed a consistent spatial difference where SD 28 was significantly different from SD 25 and 26. Our results indicate that the European flounder in the Baltic Sea consists of several loosely defined sub-populations, which may warrant a reconsideration of assessment models, management targets and regulations across subdivisions. © 2017 Elsevier B.V. All rights reserved.

1. Introduction In the Baltic Sea Proper, the area between the Danish straits and the island of Åland (Fig. 1), the deeper (and anoxic) parts outside Gotland and Gdansk may be the only physical dispersal barriers for marine fish. However, the Baltic Sea has a salinity gradient, from 25 psu in Öresund to 3–4 psu in the Bothnian Bay (HELCOM, 1996; Olsson et al., 2012) that challenges both marine and freshwater adapted organisms affecting their distribution (Olsson et al., 2012). Temporal variation in salt water inflows, oxygen levels, eutrophication, climate change, and fisheries (Olsson et al., 2012, 2013; Niiranen et al., 2013) have also altered the Baltic Sea food-web structure over time (Casini et al., 2009; Olsson et al., 2013; Östman et al., 2016a). Spatial environmental variation may result in spatially structured sub-populations that are important to consider for management (Laikre et al., 2005; Spies et al., 2015; Östman et al., 2016b). The spatial structure of species could be formed by distinct sub-populations with specific characters (e.g. adapta-

∗ Corresponding author. E-mail address: [email protected] (Z. Pekcan-Hekim). http://dx.doi.org/10.1016/j.fishres.2017.01.001 0165-7836/© 2017 Elsevier B.V. All rights reserved.

tions, morphology, life-history traits). Alternatively, populations may show continuous change in traits along geographical or environmental gradients. For exploited species, the failure to identify relevant spatial structures may result in overexploitation of some stocks whereas other stocks are not efficiently used (Palsbøll et al., 2007; Allendorf et al., 2008; Spies et al., 2015). Spatial structure of a species is also important for preserving stocks with unique biological properties such as a specific morph or life-history trait or local adaptation (Allendorf et al., 2008). Units for management should preferably be based on the spatial structure of different sources of biological data, i.e. ‘Integrated Stock Definition’ (Welch et al., 2015; Hawkins et al., 2016). These different biological data sources include for example spatial genetic population structure, life history traits, demography, morphology and stable isotopes or otolith chemistry which indicate where individuals have been during their lives (Begg and Waldman, 1999; Laikre et al., 2005; Cadrin et al., 2014; Östman et al., 2016b). The European flounder (Platichthys flesus) inhabits most parts of the Baltic Sea. There are two genetically distinct ecotypes of flounder in the Baltic, coastal spawners with demersal eggs and offshore spawners with pelagic eggs (Hemmer-Hansen et al., 2007a; Florin and Höglund, 2008), which require different salinities for success-

2

J. Erlandsson et al. / Fisheries Research 189 (2017) 1–9

Fig. 1. Map of the studied area with L95 (shaded cells) for 2014 for each ICES rectangle (thin lines). Thick lines show subdivisions (SD). The broken lines show the three transects for the semivariogram analyses.

ful reproduction and show different fecundity and growth (Nissling et al., 2002; Nissling and Dahlman, 2010). These two ecotypes are divided into separate stocks for management (ICES, 2014a, 2015). Despite that the tagging studies of European flounders have indicated short dispersal range in the Baltic and the deeper parts can act as barriers for dispersal (Aro, 1989, 2002), the coastal spawning flounders show no spatial genetic structure and is treated as one stock in ICES Subdivisions SD 27, 29–32. In contrast, the offshore spawning flounders show a genetic isolation-by-distance pattern (Florin and Höglund, 2008) and is currently assessed as three separate stocks by ICES: one unit in the straits (ICES Subdivisions, SD, 22–23), one western unit in SD 24–25 and one eastern unit in SD 26 and 28 (ICES, 2014a; ICES, 2015; see Fig. 1 for ICES subdivisions). Earlier studies of European flounder show spatial variation in demographic, morphologic and life-history traits across the Baltic Sea. Mean length-at-age was found to decrease gradually along the coast from the Great Belt to the Gulf of Finland in (quarter 4) 1994 (Drevs et al., 1999). Up until year 2007, flounders in the central Baltic Sea Proper (SD 28) were on average smaller than in the southern Baltic Sea (SD 24–26), while small differences in length was observed within the southern Baltic Sea (Gårdmark and Florin, 2007). Higher salinity improves reproduction and somatic growth of offshore spawning flounder (Nissling et al., 2002) while high population density may have negative effects on body growth. For example, Florin et al. (2013) found that length-at-age of European flounder was lower in a high density no-take zone compared to a fished area in SD 28, likely reflecting negative density dependent body growth. Thus, temporal variation in abiotic and biotic factors may affect the length distribution and spatial structure of European Flounder over time. In this study, we extend on previous studies of spatial differences based on genetics, eggs and sperm characteristics, length-at-age and tagging studies (ICES, 2014a) by studying the

spatial structure of length distributions of the European flounder across the Baltic Sea Proper using more detailed data. First we use high spatial resolution data (ICES rectangles) on length distributions from the fishery independent Baltic International Trawl Survey 2008–2014 to study the spatial structure of European flounder body length distributions. Specifically, we study whether the European flounder displays distinct and significant spatial differences in length distributions between areas indicating sharp transition zones or whether there are more continuous (i.e. gradual) differences that increase with distance between rectangles. Next, we investigate the temporal stability of spatial differences in length-at-age and age-and size specific maturity at the spatial scale of ICES subdivisions. The overall aim is to study if body size shows consistent gradual changes or distinct spatial patterns over time, and if so to identify eventual boundaries of areas with similar length distributions and life-history traits that could indicate subpopulations, enabling a more integrated stock definition of offshore European flounder in the Baltic Sea.

2. Materials and methods 2.1. Data extraction Data were compiled from the ICES data base (DATRAS, https:// datras.ices.dk/Data products/Download/Download Data public. aspx) from the Baltic International Trawl Survey (BITS) in Quarter 1 and 4 including Latvian, Polish and Swedish data for ICES subdivisions (SD) 25–28 (excluding Gulf of Riga) over the time period 2008–2014 (Fig. 1). During this time period the same age determination method (using sectioned and stained otoliths) has been used by all three countries. For a detailed description of the trawl survey see ICES (2014b). European flounders spawn in March-May and the spatial structure in Quarter 1 is therefore more likely to reflect the spatial structure of spawning individ-

J. Erlandsson et al. / Fisheries Research 189 (2017) 1–9

uals, whereas flounder in Quarter 4 is more representative of the non-spawning population. In Quarter 4 length distributions were calculated for both sexes merged by summing up catch per unit effort for centimeter length classes of all survey trawl hauls for each ICES statistical rectangle (RECT). For Quarter 1 we could calculate sex-specific length distributions for each rectangle from the data of age, length, weight, sex and gonad status of individual flounders, referred to as the ALK (the age-length-key) data in DATRAS. However, as individual flounders determined to sex, age and maturation in the ALK-data are only available at an ICES subdivision (SD) level from a length-stratified sample of the total catch in the trawl surveys (a maximum of 20 individuals per length group and SD), we converted the length distribution in the stratified individual (ALK) sample to be identical to the length distribution in each rectangle. This was done by weighting individual flounders in different length classes (Wl,r ) in the ALK differently: Wl,r = Pl,r /Sl , where Pl,r is the proportion of length class l in the survey hauls of rectangle r each year, and Sl is the proportion of length class l in the stratified sample. From these length corrected individual ALK data we calculated the upper 95 percentiles of length (L95) for each sex and rectangle in Quarter 1 (but too few flounders were available in the ALK data for a similar calculation of sex specific length distributions in Quarter 4). By using Pl for the proportion of flounders in length class l in the whole subdivision each year we could also estimate the yearly mean length-at-age for each sex and SD. 2.2. Analyses of differences among subdivisions and over time We used generalized linear models to determine if length distributions of European flounders differed between subdivisions over time using length at the 95th percentile (L95) with subdivision and year as fixed factors. To account for the eventual temporal dependence among years an AR(1)-covariance structure between years was applied. The 95th percentiles were analyzed because we were interested in the distribution of larger and older individuals, and wanted to reduce the effects of variation due to recruitment (Piet and Jennings, 2004; Hixon et al., 2014). It should be pointed out that L95 depends on multiple factors like mortality and body growth. We did separate analyses for L95 of females and males in Quarter 1 but both sexes were combined for L95 in Quarter 4. We calculated an average growth in length by age over the whole study period 2008–2014 for each sex separately. A twofactor ANOVA of mean length-at-age 5 and 4 for females and males, respectively, with subdivision and year as fixed factors were conducted to study the temporal consistency of differences in mean length-at-age of these age-classes. We used age 5 and 4 for females and males, respectively, because most European flounders reach maturity at this age (and will show slow growth; Drevs et al., 1999) but are still common in the ALK data for confident statistical analysis. Due to the fact that immature age classes were rare in the samples and lead to the risk of size selective sampling, an estimate of pre-mature variation in body growth could not be calculated. Subdivision 27 was excluded from the analyses when comparing different SDs due to insufficient amount of data. Year 2009 for L95 were excluded for both sexes and 2014 for only females due to the low number of samples in those years. To study whether flounder also differed in body mass in relation to body length we calculated the differences in length-specific body mass between rectangles using the standardized residuals from a single body weight-length regression for each sex, i.e. loge [weight (mg)] − loge [length (mm)], for all measured individual flounders. Individuals with positive residual weigh more than the average individual of that length, and vice versa. A two-factor ANOVA with residuals as dependent variable and subdivision and year as fixed factors was tested for each sex. In addition, we also calculated the

3

coefficients of the best linear fit between loge [weight (mg)] and loge [length (mm)] for each sex and subdivision separately. All statistical analyses were done with the ‘nlme’-function (Pinheiro et al., 2016) in R 3.2.5 (R Core Team, 2016). 2.3. Spatial structure of size distributions To investigate spatial variability and heterogeneity of the length distribution of European flounders (e.g. the presence of gradients or distinct transition zones) across different distances between ICES rectangles we combined the geostatistical tools semivariogramand fractal analyses (Dale, 2000). In the studied Baltic Sea region for each of three years (2008, 2011, 2014; 33–39 rectangles each year) and sex, we analysed the variability of L95 along three transects of rectangles from south-west to north-east in this stretch (11–15 rectangles per transect; 0–840 km): (1) along or close to the coast of Poland and the Baltic countries, (2) in the middle of open sea, and (3) along or close to the Swedish coast and Gotland (Fig. 1). By combining these three transects we were able to estimate the variation and structure of L95 by calculating semivariances (changes in the variability) as a function of scale in a semivariogram. Semivariograms examine the variability of a variable over a range of spatial lags, here distances between rectangles, thus estimating spatial dependence of a variable, and can indicate whether there is significant patchiness, gradients or stochasticity (Dale, 2000; Kostylev and Erlandsson, 2001; Erlandsson and McQuaid, 2004; Erlandsson et al., 2005) in the studied variable (here L95). Semivariance (Y(h) ) was calculated for different spatial lags across the studied Baltic Sea region using the equation: N(h)

Y(h)

1  = (Zi+h − Zi )2 2N (h)

(1)

i=1

where N(h) is the number of pairs of data points separated by the spatial lag h, and Z i+h and Zi are the values of the examined variables at points i and i + h (Burrough, 1983; Dale, 2000). Seven different lags (in the range 60–420 km) were included in the analyses. Spatial heterogeneity, defined as the change or structure in variance across lags (detected by the presence of a significant linear regression in the logarithmic semivariogram), was estimated using the fractal dimension ‘D’ (e.g. Burrough, 1983; Dale, 2000; Kostylev and Erlandsson, 2001), calculated as D = (4 − m)/2, where ‘m’ represents the absolute slope of the regression between the natural logarithm (ln) of the semivariance and the ‘ln’ of the spatial lag (Burrough, 1983; Dale, 2000). D varies from 1 to 2, and D values between ca 1.97 and 2.00 indicates the presence of a homogeneous or random pattern with the slope being close to 0 (Erlandsson et al., 2005). Spatial synchrony of temporal changes in L95 over the whole study area was calculated as the mean of all between rectangle Pearson correlation coefficients, and denoted Mean rp . Mean rp > 0 indicates positively correlated changes in the size-structure among rectangles, i.e. a spatial synchrony over the whole study area. Values of Mean rp close to zero, or negative, indicate spatially independent or negatively correlated changes, respectively, in size-structure between rectangles. To study the average synchrony among adjacent rectangles (side-to-side, i.e. 60 km apart), we calculated the mean Pearson correlation coefficients of all adjacent rectangles, r60 . Between rectangle correlations of annual L95 were used to calculate the Mantel correlation (rM), and the slope of the distance decay in synchrony (bS ), between the matrix of synchrony and the matrix of Euclidean distance between rectangles (e.g. Liebhold et al., 2004; Borcard and Legendre, 2012). rM indicates the strength of the relationship between synchrony and distance. A strong negative value of rM means that distance explains variation in synchrony among sites, analogous to the isolation-by-distance pattern in genetics.

J. Erlandsson et al. / Fisheries Research 189 (2017) 1–9

Year

Sex

Slope

R2

D

P

PBonferroni

Pattern

2008

Females Males Females Males Females Males

0.73 0.84 0.81 0.83 0.51 0.85

0.92 0.97 0.94 0.82 0.83 0.89

1.64 1.58 1.59 1.58 1.74 1.58

<0.001 <0.001 <0.001 0.005 0.005 0.001

0.002 <0.001 0.002 0.005 0.009 0.004

Gradient Gradient Gradient Gradient Gradient Gradient

2011 2014

Stronger negative values of b indicate a more rapid decay in synchrony with distance than lower values. To estimate the spatial scale at which changes in size-structures were independent we calculated the distance at which two rectangles were expected to have correlation coefficients r = 0 as −a/b, where a is the intercept and b the slope from the correlation-distance relationship.

400

Females Q1

(A) 25

380

L95 (mm)

Table 1 Regression components of the logarithmic semivariograms and fractal dimensions (D) for the length of the 95 percentile (L95) across different ICES rectangles (lags 60–420 km) 2008, 2011 and 2014 for females and males. Dependence between semivariance and spatial scale (i.e. lags or different distances between rectangles) is estimated from three combined transects from south-west to north-east. PBonferroni indicates the significance level after sequential Bonferroni corrections.

26

28

360 340 320 300 280 2008

2009

2010

350

2011

Year

2012

2013

2014

2013

2014

2013

2014

Males Q1

(B)

25

26

28

330

L95 (mm)

4

310 290

2.4. Maturation analyses 270 250 2008

2009

370

2010

There were significant differences in L95 among subdivisions for both sexes in Quarter 1, with European flounders being smallest in SD 28 and largest in SD 25 (females: F2,152 = 110, p < 0.0001; males: F2,140 = 120, p < 0.0001; Figs. 1 and 2). There was no difference between years for females (F6,152 = 0.9, p = 0.5) (Fig. 2a), while for males there was a significant interaction between subdivision and year (F12,150 = 2.5, p < 0.001) because of lower L95 in SD 26 and 28 for the years 2009–2010 (Fig. 2b). Also in Quarter 4, flounders (both sexes combined) were larger in SD 25 than in SD 28 (F1,38 = 112, p < 0.001; Fig. 2c) and a non-significant difference of app. 3 cm in SD 25 between years (F6,38 = 2.1, p = 0.08; Fig. 2c). Also the body weight-length relationship differed between subdivisions for both sexes (females: F2,4826 = 130, p < 0.0001; males: F2,4097 = 83, p < 0.0001, Appendix Fig. A1) as well as between years (females: F6,4826 = 54, p < 0.0001; males: F6,4097 = 4, p < 0.0001, Appendix Fig. A1). The length specific body mass of flounders was highest in SD 25 (both sexes) and lowest for females in SD 28 (Appendix Table A1, Fig. A1). The trajectories in mean length with age (in quarter 1) indicated slower growth in SD 28 than in SD 25 and 26 (Fig. 3). The mean length-at-age of both females (age 5 in quarter 1 and age 4 in quarter 4) and males (age 4 in quarter 1 and age 3 in quarter 4)

Year

2012

Females+Males Q4

(C)

25

28

350 330 310 290 270 2008

2009

2010

3. Results 3.1. Length distributions

2011

L95

To study spatial differences in maturity patterns we estimated the age-specific length at which 50% of the individuals are sexually mature from the ALK data which includes body length, age and maturation status of individual European flounder. Gonad status is determined on all individuals in the ALK data and classified as immature, i.e. gonads so undeveloped that spawning is not likely within the nearest months, or mature according to the WGBIFS BITS Manual (2011). Because there were too few individuals with age and maturity data in each year in order to get robust results we grouped individuals into two periods: 2007–2010 and 2011–2014. We estimated the body length at which 50% of respective age class had matured for each period and age class including 3 and 4 for females and 2 and 3 for males (PROC PROBIT in SAS 9.2 (SAS Institute Inc., 2009)). In the analyses, maturity status was the dependent variable and body length the explanatory variable.

2011

Year

2012

Fig. 2. Average length of the 95 percentile (L95) in the ICES rectangles for each subdivision SD 25, SD 26 and SD 28 over the time period 2008–2014 divided per sex and sampling period (in quarter 4 both sexes are combined because of lower sample size). Error bars are standard deviations from the rectangles within each subdivision.

were consistently lower in SD 28 compared to SD 26 and 25 over the whole study period (females: F2,8 = 54, p < 0.0001; males: F2,11 = 92, p < 0.0001), and did not differ between years (Fig. 3). 3.2. Type of spatial structures Semivariogram analyses of L95 detected significant spatial dependence (i.e. the closer the rectangles are the less variation there is between rectangles) across geographic scales in the three years analyzed (2008, 2011 and 2014) for both females and males. Semivariance (measure of variation) increased with distance between ICES rectangles up to 420 km (Table 1, Figs. 1 and 4 A–B). The spatial structure detected was a gradient with decreasing L95 from south-west to north-east (low patchiness or spatial heterogeneity; D = 1.58–1.74; Table 1, Fig. 4A–B). The semivariance increased from 2008 to 2014 (ANCOVA: Females: F2,17 = 57.0, p = 0.0005; Males: F2,17 = 11.0, p = 0.001; see intercepts in Fig. 4A–B) but the heterogeneity (D) was similar between years (Table 1, Fig. 4A–B; slopes are not significantly different among years). It

J. Erlandsson et al. / Fisheries Research 189 (2017) 1–9

320

(A) Females Q1

350 300 250 SD25 SD26 SD28

200 150 100 0

5

15

300

200

SD26

2010

SD28

2012

2014

Year

(F) Males Quarter 1

220

SD25 SD26 SD28

150

0

5

10

15

Mena length at age 3 (mm)

250 200

SD25 SD26 SD28

150 100 0 350

5

(D) Males Q4

10

15

2014

270

220 SD25

320

300

2012

(G) Females Quarter 4

170 2008

20

Age

SD28

Year 320

300

SD26

2010

20

(C) Females Q4

350

SD25

170 2008

Age

400

Mean length (mm)

SD25 170 2008 320

270

250

100

Mean length (mm)

20

220

Mena length at age 4 (mm)

Mean length (mm)

(B) Males Q1

10

Age

270

Mena length at age 5 (mm)

350

(E) Females Quarter 1

Mena length at age 4 (mm)

Mean length (mm)

400

5

2010

2012

2014

Year

(H) Males Quarter 4

270

250 SD25 SD26 SD28

200 150

220

100 0

5

10

15

20

170 2008

SD25 2009

2010

Age

2011

SD28

2012

2013

2014

Year

Fig. 3. (A–D) Mean length-at-age of flounders per subdivision (SD) over the study period 2008–2014 divided by sex (females, males) and sampling period (Quarter 1 or 4). Error bars are standard errors. (E–H) Mean length of flounder at age over the study period 2008–2014 for the different subdivisions divided on sex and sampling period. Note that too few individuals were determined to age and sex in SD 26 in Quarter 4 to be meaningful.

Table 2 Spatial synchrony of L95 among rectangles for males and females. Mean rp is the average Pearson correlation between all pairwise rectangles, rM is the Mantel correlation coefficients of spatial correlation, r60 is the average correlation between adjacent rectangles (60 km between center of rectangles), and log b is the slope between the Pearson correlation coefficient and log distance between rectangles. r = 0 is the estimated distance between pairwise rectangles with an average Pearson r = 0. Variable

Females

Males

Average rp rM r60 log b r = 0 (km)

0.10 −0.18** 0.13* −0.13 350

0.025 −0.18** 0.23*** −0.15 240

should be noted, however, that the slopes of the semivariance curves tend to increase (a scaling region of steeper increase in variance) at distances of ca 300–360 km among rectangles. Analyses of spatial synchrony indicated that the correlation of L95 was more similar among nearby rectangles than by chance (in the order r60 = 0.1–0.3; Table 2), and that the degree of synchrony in L95 over time decreased with distance between rectangles (Table 2; Fig. 4C–D). For females, rectangles more than 350 km apart showed independent temporal development, i.e. no synchrony, whereas

Table 3 Maturity ogive of flounders in subdivisions (SD) 25, 26 and 28 at ages 2–4 divided by sex at two different periods. Age

Females 25

Males 26

28

25

26

28

2008–2010 6% 2 52% 3 77% 4

5% 66% 71%

0% 28% 74%

55% 69% 80%

15% 65% 63%

33% 59% 82%

2011–2014 3% 2 38% 3 76% 4

1% 40% 93%

4% 29% 79%

47% 63% 80%

13% 82% 97%

28% 70% 85%

corresponding distance was around 240 km for male flounder (Table 2). 3.3. Maturation A lower proportion of European flounders at age 3 had matured in SD 28 than in SD 25 and SD 26 (Table 3). This was not just due to a lower age specific length in SD 28, but the age specific length at which 50% are mature also differed between the SDs (␹2 (2) = 61,

6

J. Erlandsson et al. / Fisheries Research 189 (2017) 1–9 0.2

A) Females

C) Females

0.15

1000

Semivariance(log scale)

0.1

rM

0.05

2008 females

0 -0.05

0

100

200

300

400

500

600

2011 females

-0.1

2014 females

100 50

500

Spaal lag (km, log scale)

R² = 0.4762

-0.15 -0.2 Distance (km) 0.25

B) Males

D) Males

0.2

500

Semivariance(log scale)

0.15

rM

0.1 2008 males 2011 males

0.05 0

2014 males

0

100

200

300

400

500

600

-0.05

50 50

Spaal lag (km, log scale)

500

-0.1

R² = 0.3161

-0.15 Distance (km)

Fig. 4. (A–B) Semivariograms of length at the 95 percentile (L95) in 2008, 2011 and 2014 at different scales (i.e. different distances between rectangles) along 3 different and combined transects from south-west to north-east in the Baltic Sea Proper divided between A) Females and B) Males. C–D) Mantel correlograms of spatial synchrony of changes, rM, in L95 between ICES rectangles over the period 2008–2014 depending on distance between rectangles. rM is the Mantel correlation coefficients of spatial correlation. Positive values indicate a positive correlation over time whereas values close to zero or negative indicate no correlation or negative correlation, respectively, in L95 over time between rectangles.

While previous studies have shown spatial differences in length distribution and length-at-age of European flounder (Drevs et al. 1999; Gårdmark and Florin, 2007) we here show that the shorter mean length and smaller length-at-age and maturation towards the northern SD 28 have been stable over time despite the changes in environmental conditions such as decline in salinity. Moreover, our study on length distributions was done on a finer spatial scale and suggests a spatial gradient in the size structure (L95) of European flounder also within subdivisions in the Baltic Sea Proper with gradually decreasing L95 towards the north-east of the study area. This spatial gradient was observed both for spawning (Quarter 1) as well as before spawning individuals (Quarter 4). The spatial gradient in the length distribution of European flounder (L95 in both sexes) is reflecting the genetic isolationby-distance pattern of offshore spawning flounder in the Baltic Sea (Hemmer-Hansen et al., 2007a,b; Florin and Höglund, 2008). Hemmer-Hansen et al. (2007a,b) suggested that the genetic isolation-by-distance pattern in offshore spawning European flounder is a genetic adaptation towards the salinity gradient in the Baltic Sea. This may indicate a genetic component in addition to plastic responses to the environment for the consistent spatial differences in length distributions, and the spatial differences in mean length-at-age and maturity. The decrease in salinity from 8 psu in SD 25 to 6–6.5 psu in SD 28 influences reproduction by reducing fecundity and in higher salinity areas faster somatic growth

Esmated length (mm) at 50% maturity ogives

4. Discussion

280

(A) Females

260 240 220

25

200

26 28

180 160 140 3 220

Esmated length (mm) at 50% maturity ogives

p < 0.001, Fig. 5). Thus, despite a larger proportion of flounder had matured at a given size in SD 28 than in the two more southern SDs, flounder in SD 28 has a lower maturity ogive at age 3 than in the other SDs (Table 3). There was also a difference in age specific length at which 50% are mature between the two time periods (␹2 (1) = 38), especially females in SD 26 and 28 are matured at a smaller size in the latter period.

Age

4

(B) Males 200 180 25 160 26 28

140 120 100 2

3 Age

4

Fig. 5. Estimated age-specific length at 50% probability of maturity, ␮, of flounder at two different time periods, 2007–2010 (hatched lines) and 2011–2014 (solid lines) in three different ICES subdivisions. Lines are for illustrative purposes only.

J. Erlandsson et al. / Fisheries Research 189 (2017) 1–9

is observed (Nissling et al., 2002; Nissling and Dahlman, 2010). Also the oxygen levels vary within the Baltic Sea Proper, especially regarding hypoxia and anoxia events at the bottom. Higher frequency (80–100%) and larger areal extent of hypoxia occurs in SD 28 compared to SD 25 and 26 (Conley et al., 2009; Hansson and Andersson, 2013; Lehmann et al., 2014). Hypoxia affects somatic growth negatively in juveniles of related flounder species (Stierhoff et al., 2006). The spatial structure of continuous change in population genetics and size distribution of European flounder contrasts with the spatial structure of synchrony in population abundances. Changes in population abundances are almost independent between the ICES rectangles (outside the spawning season). This would indicate that flounders are either stationary within rectangles, which is unlikely based on tagging studies (Florin 2005), or a significant movement (asymmetric redistribution between rectangles) of offshore flounder takes place between rectangles in the Baltic Sea Proper (Östman et al., 2016b). All together this suggests a population connectivity among nearby rectangles and gene flow between local spawning aggregations. However, increasing reproductive isolation and reduced population connectivity occur with greater spatial distance or different local abiotic conditions but lacking strong dispersal barriers in the Baltic Sea Proper. The age specific size at which 50% of the flounders were mature was considerably smaller for flounder in SD 28 than in SD 25 and 26. But fewer flounder had matured at age 3 in SD 28 compared to the other two subdivisions because of the lower length-atage. This could indicate the maturation at a smaller size in SD 28 as a “counter-adaptation” to the local environmental conditions (Conover and Schultz, 1995). Otherwise flounders in SD 28 would mature at even higher ages and few may reach maturity. These spatial differences in maturity patterns may be important to consider in management and assessments of flounder in the Baltic Sea, for example minimum mesh sizes to protect immature individuals or conducting assessment models for different subdivisions. Since SD 28 is known to harbour both ecotypes of flounder and the demersal spawning flounder is known to grow slower (Nissling and Dalman, 2010), it is possible that the decrease in L95 towards the north-east could be influenced by a shift in dominance between the two ecotypes. There are more spatially refined genetic studies on the distribution of demersal and pelagic flounder currently in progress, thus the dominance of the two ecotypes can be clarified in the near future. However, the demersal spawning flounder is assumed to be less abundant in this area and caught to a lesser extent in the BITS survey (ICES, 2014a). Furthermore, we speculate that since the data in Quarter 1 is from or close to spawning time, the presence and influence of this ecotype in our results during this period is expected to be low. Also, the coefficient of variation for length-at-age is similar or lower in SD 28 (not shown) and not higher as would have been expected if we had sampled two populations with different length distributions. Gårdmark and Florin (2007) also found the variation in the length distributions generally did not differ between subdivisions in the Baltic Sea. Spatial variation in size-selective mortality and density dependence may be two factors contributing to the observed spatial patterns in length distributions, length-at-age and maturity. Estimates of mortalities from cohorts are uncertain because the sampling effort has been variable between years, but using the age distributions from each years sample do not indicate higher mortality in SD 28 (Appendix Fig. A2). There are some indications that body growth in European flounder is density dependent (Florin et al., 2013) but if that can explain spatial variation in length distributions and life-history traits is currently unknown. We can here not pinpoint the mechanisms for the observed patterns but this spatial variation may be important to consider in the management of European flounder in the Baltic Sea.

7

Integrated Stock Definition has been advocated to integrate different processes affecting the spatial structure of exploited species (Welch et al., 2015; Hawkins et al., 2016). For sustainable yield it is important to know whether a population consists of a few or several distinct populations, or several more or less connected populations (‘stepping-stone’) causing an isolation by distance pattern in the region (Laikre et al., 2005; Spies et al., 2015). Our results indicate, in congruence with population genetic data, continuous change in flounder length distributions across the Baltic Sea Proper region. There is no fixed management protocol for populations with the isolation by distance structure (Spies et al., 2015), but in a population with a continuous structure, such as in the present study, local management actions can often lead to higher sustainable yields compared to treating all subpopulations identically (Spies et al., 2015). European flounder in the Baltic Sea lacks an analytical assessment and management is currently based on minimum size, mesh size and protection during spawning time that differ between subdivisions. Our results imply that a larger minimum size should be considered in SD 26 compared to SD 28, as flounders grow faster and mature at larger sizes in SD 26. Our results also imply that different age-length key might be applied when calculating biomass indices, which should be relatively straightforward. Finally, when inferring processes that drive temporal variation in demographic or life-history traits our results suggest this should be done on a relatively small spatial scale, i.e. ICES subdivision or smaller. Acknowledgements Pontus Mattsson (SLU Aqua) provided us great help producing the maps. We are grateful to two anonymous reviewers that improved the earlier version of this manuscript. Appendix A. See Table A2 .

Table A1 Relationships between loge [weight (mg)] and loge [length (mm)] for flounder in each subdivision and sex over the time period 2008–2014. Sex

Subdivison

Intercept

Slope

r2

N

Females

SD 25 SD 26 SD 28 SD 25 SD 26 SD 28

−11.4 −11.5 −10.7 −10.3 −10.0 −9.54

3.02 3.03 2.88 2.80 2.75 2.65

0.97 0.97 0.97 0.95 0.95 0.96

1985 1356 1507 1294 1219 1635

Males

Table A2 Mean length-at-age, standard deviation and Coefficient of variation (CV) for age 4 females and age 5 male European flounders in the ICES subdivisions 25, 26, 28 over the time period 2008–2014. Sex

Subdivision

Mean (mm)

Standard deviation

CV

N

Females (age 4)

SD 25 SD 26 SD 28 SD 25 SD 26 SD 28

273 257 219 260 245 204

42 36 31 29 29 23

15.2 14.1 14.3 11.1 11.7 11.1

541 322 156 181 265 167

Males (age 5)

8

J. Erlandsson et al. / Fisheries Research 189 (2017) 1–9

Average standardized residuals

0.8 0.7

A) Females SD 25

0.6

SD 26

0.5

SD 28

0.4 0.3 0.2 0.1 0 -0.1 -0.2 2007

2008

2009

2010

2011

2012

2013

2014

2015

Year

Average standardized residuals

0.2 0.1

B) Males SD 25

0

SD 26

-0.1

SD 28

-0.2 -0.3 -0.4 -0.5 -0.6 -0.7 -0.8 2007

2008

2009

2010

2011

2012

2013

2014

2015

Year Fig. A1. Average standardized residuals of the relationship between loge [weight (mg)] and loge [length (mm)] in each subdivision over the time period 2008–2014 divided for A) Females, and B) Males. Error bars indicate standard errors, lines are only for illustrative purposes.

Fig. A2. Estimated total annual instantaneous mortality from the catch curve of European flounder in three different ICES subdivisions divided per sex. Mortality is estimated as the (negative) slope between loge (proportion of individuals in age class x) and age 4–14 each year from the individual data from the DATRAS ALK-dataset. Because the individual data come from a length stratified sampling we weighted individuals differently according to the length class (see Materials & Methods) to correct the age distributions for stratified sampling.

J. Erlandsson et al. / Fisheries Research 189 (2017) 1–9

References Allendorf, F.W., England, P.R., Luikart, G., Ritchie, P.A., Ryman, N., 2008. Genetic effects of harvest on wild animal populations. Trends Ecol. Evol. 23, 327–337. Aro, E., 1989. A review of fish migration patterns in the Baltic. Rapports et Procès-Verbaux des Réunions du Conseil international pour l’Exploration de la Mer, 190, 72–96. Aro, E., 2002. Fish migration studies in the Baltic Sea: a historical review. ICES Mar. Sci. Symp. 215, 361–370. Begg, G.A., Waldman, J.R., 1999. A holistic approach to fish stock identification. Fish. Res. 43, 35–44. Borcard, D., Legendre, P., 2012. Is the mantel correlogram powerful enough to be useful in ecological analysis? A simulation study. Ecology 93, 1473–1481. Burrough, P.A., 1983. Multiscale sources of spatial variation insoil: 1. The application of fractal concepts to nested levels of soil variation. J. Soil Sci. 34, 577–597. Cadrin, S.X., Kerr, L.A., Mariani, S., 2014. Stock Identification Methods: Applications in Fishery Science, 2nd ed. Academic Press, Amsterdam. Casini, M., Hjelm, J., Molinero, J.-C., Lövgren, J., Cardinale, M., Bartolino, V., Belgrano, A., Kornilovs, G., 2009. Trophic cascades promote threshold-like shifts in pelagic marine ecosystems. Proc. Natl. Acad. Sci. U. S. A. 106, 197–202. Conley, D.J., Björck, S., Bonsdorff, E., Carstensen, J., Destouni, G., Gustafsson, B.G., Hietanen, S., et al., 2009. Hypoxia-related processes in the Baltic sea. Environ. Sci. Technol. 43, 3412–3420. Conover, D., Schultz, E.T., 1995. Phenotypic similarity and the evolutionary significance of countergradient variation. Trends Ecol. Evol. 10, 248–252. Dale, M.R.T., 2000. Spatial Pattern Analysis in Plant Ecology. Cambridge University Press, Cambridge. Drevs, T., Kadakas, V., Lang, T., Mellergaard, S., 1999. Geographical variation in the age/length relationship in Baltic flounder (Platichthys flesus). ICES J. Mar. Sci. 56, 134–137. Erlandsson, J., McQuaid, C.D., 2004. Spatial structure of recruitment in the mussel Perna perna at local scales: effects of adults, algae and recruit size. Mar. Ecol. Prog. Ser. 267, 173–185. Erlandsson, J., McQuaid, C.D., Kostylev, V.E., 2005. Contrasting spatial heterogeneity of sessile organisms within mussel (Perna perna L.) beds in relation to topographic variability. J. Exp. Mar. Biol. Ecol. 314, 79–97. Florin, A.-B., Höglund, J., 2008. Population structure of flounder (Platichthys flesus) in the Baltic Sea: differences among demersal and pelagic spawners. Heredity 101, 27–38. Florin, A.-B., Bergström, U., Ustups, D., Lundström, K., Jonsson, P.R., 2013. Effects of a large northern European no-take area on flatfish populations. J. Fish Biol. 83, 939–962. Florin A.-B., 2005, Flatfishes in the Baltic Sea −a review of biology and fishery with a focus on Swedish conditions, FINFO, 2005, 14. Gårdmark, A., Florin, A.-B., 2007. Flounder (Platichtys flesus) growth and length distributions—an input to length based assessments. In: Working Document to the ICES Baltic Fisheries Assessment Working Group (WGBFAS), 17–26 April 2007, Copenhagen, Denmark. HELCOM, 1996. Third periodic assessment of the state of the marine environment of the Baltic Sea, 1989–1993. In: Background Document Baltic Sea Environment Proceedings, no 64B, Helsinki Commission, Helsinki, Finland. Hansson, M., Andersson, L., 2013. Oxygen survey in the Baltic Sea 2013–Extent of anoxia and hypoxia, 1960–2013. SMHI, Report Oceanography no 49, ISSN: 0283-1112. Hawkins, S.J., Bohn, K., Sims, D.W., Ribeiro, P., Faria, J., Presa, P., et al., 2016. Fisheries stocks from an ecological perspective: disentangling ecological connectivity from genetic interchange. Fish. Res. 179, 333–341. Hemmer-Hansen, J., Nielsen, E.E., Grönkjaer, P., Loeschcke, V., 2007a. Evolutionary mechanisms shaping the genetic population structure of marine fishes; lessons from the European flounder (Platichthys flesus L.). Mol. Ecol. 16, 3104–3118. Hemmer-Hansen, J., Nielsen, E.E., Frydenberg, J., Loeschcke, V., 2007b. Adaptive divergence in a high gene flow environment: Hsc70 variation in the European flounder (Platichthys flesus L.). Heredity 99, 592–600. Hixon, M.A., Johnson, D.W., Sogard, S.M., 2014. BOFFFFs: on the importance of conserving old-growth age structure in fishery populations. ICES J. Mar. Sc. 71, 2171–2185.

9

ICES, 2014. Report of the Benchmark Workshop on Baltic Flatfish Stocks (WKBALFLAT), 27–31 January 2014, Copenhagen, Denmark. ICES CM 2014/ACOM:39, 320 pp. ICES, 2014. Manual for the Baltic International Trawl Surveys (BITS). Series of ICES Survey Protocols SISP 7 – BITS. 71 pp. ICES, 2015. Report of the Baltic Fisheries Assessment Working Group (WGBFAS), 14–21 April 2015, ICES HQ, Copenhagen, Denmark. ICES CM 2015/ACOM:10, 806 pp. Kostylev, V., Erlandsson, J., 2001. A fractal approach for detecting spatial hierarchy and structure on mussel beds. Mar. Biol. 139, 497–506. Laikre, L., Palm, S., Ryman, N., 2005. Genetic population structure of fishes: implications for coastal zone management. Ambio 34, 111–119. Lehmann, A., Hinrichsen, H.-H., Getzlaff, K., Myrberg, K., 2014. Quantifying the heterogeneity of hypoxic and anoxic areas in the Baltic Sea by a simplified coupled hydrodynamic-oxygen consumption model approach. J. Mar. Syst. 134, 20–28. Liebhold, A., Koenig, W.D., Bjørnstad, O.N., 2004. Spatial synchrony in population dynamics. Annu. Rev. Ecol. Evol. Syst. 35, 467–490. Niiranen, S., Yletyinen, J., Tomczak, M.T., Blenckner, T., Hjerne, O., MacKenzie, B.R., Muller-Karulis, B., et al., 2013. Combined effects of global climate change and regional ecosystem drivers on an exploited marine food web. Glob. Change Biol. 19, 3327–3342. Nissling, A., Westin, L., Hjerne, O., 2002. Reproductive success in relation to salinity for three flatfish species, dab (Limanda limanda), plaice (Pleuronectes platessa), and flounder (Pleuronectes flesus) in the brackish water Baltic Sea. ICES J. Mar. Sci. 59, 93–108. Nissling, A., Dahlman, G., 2010. Fecundity of flounder, Pleuronectes flesus, in the Baltic Sea −reproductive strategies in two sympatric populations. J. Sea Res. 64, 190–198. Olsson, J., Bergström, L., Gårdmark, A., 2012. Abiotic drivers of coastal fish community change during four decades in the Baltic Sea. ICES J. Mar. Sci. 69, 961–970. Olsson, J., Bergström, L., Gårdmark, A., 2013. Top-down regulation, climate and multi-decadal changes in coastal zoobenthos communities in two Baltic Sea areas. PLoS One 8, e64767. Östman, Ö., Eklöf, J., Eriksson, B.K., Olsson, J., Moksnes, P.-O., Bergström, U., 2016a. Top-down control as important as nutrient enrichment for eutrophication effects in North Atlantic coastal ecosystems. J. Appl. Ecol. (in press). Östman, Ö., Olsson, J., Dannewitz, J., Palm, S., Florin, A.-B., 2016b. Inferring spatial structure from population genetics and spatial synchrony in demography of Baltic Sea fishes: implications for management. Fish Fish., http://dx.doi.org/10. 1111/faf.12182. Palsbøll, P.J., Bérubé, M., Allendorf, F.W., 2007. Identification of management units using population genetic data. Trends Ecol. Evol. 22, 11–16. Piet, G.J., Jennings, S., 2004. Response of potential fish community indicators to fishing. ICES J. Mar. Sci. 62, 214–225. Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., R Core Team 2016. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3. 1–125. URL: http:// CRAN.R-project.org/package=nlme. R Core Team, 2016. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria URL: https://www.R-project.org/. SAS Institute Inc, 2009. Base SAS 9.2 Procedures Guide. SAS Institute Inc., Cary, NC. Spies, I., Spencer, P.D., Punt, A.E., 2015. Where do we draw the line? A simulation approach for evaluating management of marine fish stocks with isolation-by-distance stock structure. Can. J. Fish. Aquat. Sci. 72, 968–982. Stierhoff, K.L., Targett, T.E., Miller, K., 2006. Ecophysiological responses of juvenile summer and winter flounder to hypoxia: experimental and modeling analyses of effects on estuarine nursery quality. Mar. Ecol. Prog. Ser. 325, 255–266. Welch, D.J., Newman, S.J., Buckworth, R.C., Ovenden, J.R., Broderick, D., Lester, R.J., et al., 2015. Integrating different approaches in the definition of biological stocks: a northern Australian multi-jurisdictional fisheries example using grey mackerel, Scomberomorus semifasciatus. Mar. Policy 55, 73–80. WGBIFS BITS Manual, 2011. Manual for the Baltic International Trawl Surveys. Available at: https://datras.ices.dk/Documents/Manuals/Addendum 1 WGBIFS BITS Manual 2011. pdf. [Visited 2016–12-21].