Otolith biochronology as an indicator of marine fish responses to hydroclimatic conditions and ecosystem regime shifts

Ecological Indicators 79 (2017) 286–294

Contents lists available at ScienceDirect

Ecological Indicators journal homepage: www.elsevier.com/locate/ecolind

Original Articles

Otolith biochronology as an indicator of marine fish responses to hydroclimatic conditions and ecosystem regime shifts

MARK

⁎

Szymon Smoliński , Zuzanna Mirny Department of Fisheries Resources, National Marine Fisheries Research Institute, Kołłątaja 1, 81-332 Gdynia, Poland

A R T I C L E I N F O

A B S T R A C T

Keywords: Climate Otolith Baltic Sea European flounder Platichthys flesus Fish growth Regime shift Ecological modeling

Sclerochronological studies based on hard structures of marine organisms are valuable tools—both for reconstructing past climate conditions and for predicting future impacts of environmental changes on marine resources. Existing archives, which house millions of fish otoliths (ear stones) constitute an excellent basis for such research; but, they remain underutilized. The objective of this project was to identify the factors that influence the annual growth patterns of the European flounder (Platichthys flesus) based on an analysis of otolith increments. We applied linear mixed models to develop a 74-year long chronology that reflects the inter-annual variations in flounder growth rates using otolith samples collected from 1957 to 2016 in the southern part of the Baltic Sea, which is considered to be highly vulnerable to global climate change. By analyzing the widths of otolith increments we revealed the existence of common environmental factors that influence fish growth. Using a mixed modeling framework, we incorporated a recent method to identify the optimal time window for climatic factors and showed that the most significant effect of the mean Baltic Sea Index occurs during August–December, while mean sea surface temperature is most significant from April–June. Change point analysis on the developed chronology identified major alterations occurred in flounder growth in 1988, 1992 and 2006. This result is in accord with published studies on regime shifts in the Baltic Sea ecosystem. This paper reports information concerning the response of the commercially important European flounder to the changing environment that may support future ecosystem-based management of fish stocks. Moreover, the results also highlight the potential for applying biochronological techniques to identify rapid regime shifts in marine ecosystems.

1. Introduction

(Mackenzie et al., 2007). Because current practices and fish stock assessment models may be inadequate for projected environmental conditions (Hoegh-Guldberg and Bruno, 2010), the management strategies of these marine resources will need to be modified (Brander, 2007). The form of such modifications will depend on the sensitivity of each species to ecological changes (Rowland et al., 2011). Such sensitivity can be measured by net growth response, which reflects the direct and indirect impacts of environmental parameters on population over time in an integrated manner (Rountrey et al., 2014). For these purposes, biochronologies based on the annual increments of otoliths (which are hard structures located in the inner ears of fish) may be used as proxies of somatic growth changes over time (e.g., Black et al., 2013; Rountrey et al., 2014; Izzo et al., 2016). In fisheries science, hundreds of thousands of otoliths are used annually to estimate the age of individual specimens (Campana and Thorrold, 2001). These methods have the potential to provide information concerning past ecological responses at unprecedented temporal and spatial scales (Morrongiello et al., 2012).

Ocean temperatures have increased substantially in recent decades and are projected to rise further during the 21st century under all investigated scenarios (IPCC, 2014). It is assumed that warming will lead to various alterations in ocean hydrology, such as increase in sea level, greater stratification of the water column or more intense storm systems (Hoegh-Guldberg and Bruno, 2010). The future consequences of these changes for the functions and structures of ecosystems are still highly uncertain—especially for marine systems, where long-term biological data are scarce (Poloczanska et al., 2013). Our understanding of how organisms respond to changes in climatic conditions could be extended through sclerochronological studies based on the hard structures of marine organisms (Morrongiello et al., 2012). This technique is a valuable tool both for reconstructing past climate conditions (Reynolds et al., 2016) and for predicting future environmental impacts on marine resources (Rountrey et al., 2014). Fish population dynamics are often driven by climatic variations

⁎

Corresponding author. E-mail address: [email protected] (S. Smoliński).

http://dx.doi.org/10.1016/j.ecolind.2017.04.028 Received 25 January 2017; Received in revised form 4 April 2017; Accepted 9 April 2017 1470-160X/ © 2017 Elsevier Ltd. All rights reserved.

Ecological Indicators 79 (2017) 286–294

S. Smoliński, Z. Mirny

2.2. Study species

Biochronological approaches have been successfully applied to a wide range of species and regions to deliver long-term data on fish responses to environmental factors. These studies have primarily used dendrochronological methods (e.g., Matta et al., 2010; Gillanders et al., 2012; Black et al., 2013; Ong et al., 2016), mixed models (Izzo et al., 2016; Morrongiello and Thresher, 2015; Morrongiello et al., 2011; Weisberg et al., 2010) and Bayesian techniques (Helser et al., 2012) to reveal a variety of relationships between fish growth and external drivers, quantifying sources of year-to-year variability or long-term trends. However, investigations of developed biochronologies in the context of their potential to detect periods of rapid changes in environmental conditions, such as ecosystem regime shifts, are still rare. Currently, term regime shifts is typically defined as infrequent and abrupt alterations in the whole ecosystem structure and function, occurring at multiple trophic levels and on large geographic scales (Möllmann et al., 2009). Regime shifts can cause losses of ecological and economic resources with important management implications (Möllmann et al., 2009). Such reorganization have been described in many marine ecosystems, including North Pacific (e.g. Hare and Mantua, 2000), Black Sea (e.g. Daskalov, 2002) or North Sea (e.g. Alheit et al., 2005) and have been explained mainly as a results of climatic forcing, anthropogenic impacts or combination of both causes (Möllmann et al., 2009). Development of robust methodologies for detecting previous regime shifts is considered the first step for prediction of this phenomena in the future and for adaptation of management policies to possible consequences of these rapid reorganizations of ecosystems (Kraberg et al., 2011). To our knowledge, such detection of past ecosystem alterations, using formal statistical analyses of fish otolith biochronology, have thus far been conducted only for Pacific ocean perch (Sebastes alutus) in the Bering Sea region. This study presented the significant influence of the 1976–1977 regime shift on fish growth (van der Sleen et al., 2016). Faced with a developing network of marine biochronologies (Reynolds et al., 2016), and because often other ecological time series are insufficient, the suggested approach may provide hints for future research regarding both species' sensitivity to rapid ecosystem alterations (Izzo et al., 2016) and identification of past abrupt ecosystem regime shifts. The objective of this study was, first, to develop a multidecadal biochronology for the commercially important European flounder (Platichthys flesus) in the southern Baltic Sea based on archived otoliths (using samples collected from 1957 to 2016) and, subsequently, to investigate relationships between fish growth patterns and selected hydroclimatic factors. To identify the best predictor and optimal time window for climatic signals, we applied a state-of-the-art exploratory “sliding window” approach. Moreover, we conducted statistically-based analyses of fish growth phase transitions to investigate whether the developed otolith biochronology was suitable for detecting discontinuities in ecosystem states. We compared the results with information from other studies concerning regime shifts in the function and structure of the Baltic ecosystem to validate the change points defined using the biochronology.

The European flounder (Platichthys flesus) is the most common and commercially important flatfish in the Baltic Sea, with yearly international landings of up to 15,000 tons. It is distributed throughout the Baltic Sea except in the eastern portion of the Gulf of Finland and in the Gulf of Bothnia (Nissling and Dahlman, 2010). Baltic flounder can be divided into two types: demersal (“bank flounder”) and pelagic (offshore) spawners. The demersal spawners produce small, heavy eggs that develop along the bottoms of shallow banks and coastal areas in the northern part of the Baltic. The pelagic spawners are distributed in the southern and the deeper eastern parts of the Baltic Sea and spawn at a depth of 40–80 m (Nissling and Dahlman, 2010). This study focused on flounders caught in the southern part of the Baltic Sea; consequently, most were assumed to be pelagic spawners. However, there is no clear geographic boundary between the demersal and pelagic types. During spawning, the demersal and pelagic types travel into shallower and deeper areas, respectively, but they probably mix during the feeding season (Florin and Höglund, 2008). Flounders from the southern Baltic spawn from March to May/June (Aro, 1989). After spawning, they migrate toward feeding grounds in the shallow coastal waters. Flounder can live up to 20 years (the oldest fish in this study), but most live to no more than 10 years. 2.3. Otolith measurements Studies that use annual increments of otoliths to reconstruct or predict fish growth assume that there is a relationship between fish length and otolith radius (Campana, 1990). This assumption was tested for European flounder within our study area using a group of 755 individuals caught between 1977 and 1992 (Draganik and Kuczyński, 1993). Because there was a high reported correlation between fish length and otolith radius (R = 0.98), we were able to assume that there was a strong relationship between fish somatic growth and otolith growth. We used otoliths from flounders sampled in the southern Baltic Sea from 1957 to 2016. We usually chose symmetric otoliths, but when symmetric otolith was not available or it was broken, an asymmetric one was used (20 samples). The otoliths were placed in horizontal rows concave side down, anterior end uppermost and embedded in epoxy resin. We sliced them into 0.4-mm-thick transverse sections through the primordium and along the dorsolateral plane using an automatic precision cut-off machine (Struers Accutom-50). The otolith slices were etched with 1% hydrochloric acid and stained with a solution of neutral red. We viewed all samples using reflected light at 32 × magnification. Otolith images were acquired using a stereo microscope (Leica M 205C) with a digital camera (Leica DFC 450). We analyzed a total of 518 otolith images using ObjectJ (an ImageJ plugin). Increments were marked along the polynomial axis perpendicular to the annual rings from the nucleus to the dorsal edge of each otolith and the distances between them were measured (4725 measurements in total). Measurements of the first and last ring widths were excluded because they showed incomplete growth (Fig. 2).

2. Materials and methods 2.4. Environmental predictors 2.1. Study area Environmental predictors were derived from Baltic Sea physics reanalysis data of the Swedish Meteorological and Hydrological Institute (SMHI) available at the Copernicus-Marine Environment Monitoring Service (CMEMS, 2016) and from Extended Reconstructed Sea Surface Temperature (ERSST v4) model data (Huang et al., 2015). The reanalysis products for the physical conditions in the Baltic Sea were combined by SMHI using the 3D Ensemble Variational scheme in the High-Resolution Operational Model for the Baltic (HIROMB) circulation model. The product provides data at a 3-nautical-mile grid (5.5 km) scale. The ERSST v4 is a global monthly sea surface tempera-

The Baltic Sea, located in Northern Europe, is one of the largest brackish-water basins in the world (415 200 km2). The functioning of this semi-enclosed sea is highly impacted by climate variability (Mackenzie et al., 2007; Möllmann et al., 2009). Because the Baltic Sea's only connection with the Atlantic Ocean is through the Danish Straits, the hydrological situation within the Baltic depends mainly on water mass exchanges with the North Sea (Lehmann et al., 2002). The area of this study was located in southern Baltic Sea—mainly the Bornholm Basin, Eastern Gotland Basin and the Gdansk Deep (Fig. 1). 287

Ecological Indicators 79 (2017) 286–294

S. Smoliński, Z. Mirny

Fig. 1. Approximate European flounder sampling area (dark gray). The main basins are indicated within the study area (1) the Bornholm Basin, (2) the Eastern Gotland Basin, and (3) the Gdansk Deep. The Baltic Sea Index (BSI) is the difference in normalized sea level pressures between Oslo in Norway and Szczecin in Poland (the two black dots).

ture dataset derived from the International Comprehensive OceanAtmosphere Dataset (ICOADS). It is produced on a 2° × 2° grid and is spatially complete. We gathered information on water temperatures from both mentioned sources for the approximate area of fish sampling (14°E–20°E, 54°N–56°N) (Fig. 1). Sea surface temperature (SST) from the ERSST model and bottom temperature (SMHI data) were used in the investigation. The SMHI data were limited to the period between February 1989 and November 2014, while the ERSST data covered from 1854 to the present. We used linear regression to test the relationship between mean monthly SST and mean monthly bottom temperature (Fig. S1, Table S1). The fitted regression models showed that mean monthly SST was a sufficient proxy for the thermal conditions of bottom waters only for the months from October to May (Table S1), which was helpful for further interpretation of the results. In addition to the thermal information we used the Baltic Sea Index (BSI) as a proxy of water mass circulation within the study area. BSI is highly related to the North Atlantic Oscillation (NAO), but reflects local hydrological conditions over the Central Baltic Sea (Lehmann et al., 2002). It is defined as the normalized sea level pressure difference between Oslo, Norway (53°13′N, 14°13′E) and Szczecin, Poland (59°30′N, 10°30′E) (Fig. 1).

Table 1 List of predictors used in the mixed modeling of European flounder growth. Predictor

Description

Random effects FishID Year Cohort

Unique identifier of fish individual Calendar year of otolith ring formation Group of fish from the same spawning season

Fixed effects Age Sex AAC SST BSI

Age of fish in which growth increment was formed Sex of fish individual (female, male, unknown) Final age of fish at time of capture Mean sea surface temperature for a particular period Mean Baltic Sea Index for a particular period

Sex and Age-at-capture as fixed intrinsic effects, allowing for the interaction between Age and Sex. Age-at-capture term tests for potential bias associated with growth rate-based selectivity and corrects estimates of inter-annual growth variation taking into account influences of certain phenotypes (i.e. short-lived individuals) (Doubleday et al., 2015; Morrongiello and Thresher, 2015). As random effects, we had intercepts for individual fish (FishID), Year of otolith ring formation and fish Cohort, as well as by-FishID, by-Year and by-Cohort random slopes for the effect of Age. The inclusion of random Age intercept and slope allowed for fitting individual age-dependent growth trajectories for each sample (Izzo et al., 2016) and minimized the variations caused by the potential inconsistencies in the course of the axes used for otoliths measurements. We developed a series of mixed models to test different intrinsic and

2.5. Data analysis Prior to the analysis, otolith Increment width, Age and Age-atcapture data were log transformed and mean centered (Table 1). We treated Increment width as a response variable and included fish Age,

Fig. 2. Otolith transverse section from 14-year-old individual of Platichthys flesus. Measurement axis (white line) and annual rings (black and gray points) are shown. The first and last years of growth were excluded from the analysis because they showed incomplete growth.

288

Ecological Indicators 79 (2017) 286–294

S. Smoliński, Z. Mirny

Fig. 3. European flounder growth variations and their synchrony with environmental conditions. Predicted conditional mode of Year (best linear unbiased predictor, width in mm of annual increments ± SE) is indicated with (a) regime shifts detected by STARS analysis (vertical dotted lines), mean growth values for each regime (horizontal dashed lines) and the number of increments measured in each year (the size of the dots); (b) mean Baltic Sea Index (BSI) for the period August–December; (c) mean sea surface temperatures (SST) for the period April–June (open circles and dotted lines).

environmental sources of variations in fish growth (Morrongiello and Thresher, 2015; Weisberg et al., 2010). We compared the models of growing complexity with the Akaike Information Criterion corrected for the small sample sizes (AICc; Burnham and Anderson, 2004) to select the optimal base model describing fish growth. First, we selected the optimal random model with a full fixed-effect structure. The models were fitted using restricted maximum likelihood (REML). Second, we selected the optimal fixed effects with previously selected random terms (Morrongiello and Thresher, 2015; Zuur et al., 2009). For fixed effects optimization, models were fitted using maximum likelihood and the best-ranked model was then refitted with REML, allowing unbiased parameter estimates (Zuur et al., 2009). We calculated the conditional R2 metric for each model to estimate the variance in fish growth explained by the combined fixed and random effects (Nakagawa and Schielzeth, 2013). We assessed the synchrony of fish growth among individuals for the Year and Cohort terms using interclass correlation

coefficients (ICC) for a random intercept-only model (Goldstein et al., 2002; Morrongiello and Thresher, 2015). We extracted the best linear unbiased predictor (BLUP) of the base model random effect of Year to visualize the temporal patterns of fish growth variability and to investigate the occurrence of significant step changes (regime shifts) in historical fish growth. We also applied sequential t-test analysis (STARS) to the BLUP time series (Rodionov, 2006, 2004). STARS method allows to determine statistically significant shifts in the mean level and the magnitude of fluctuations in time series (Lindegren et al., 2010). It provides a probability level for the detected year of regime shift, based on the modified two-sided Student’s t-test. Regime Shift Index (RSI) calculated by the algorithm represents a cumulative sum of normalized anomalies (Rodionov, 2004). To make our results comparable with other studies on regime shifts in the Baltic Sea ecosystem, we set the cut-off length (l) to 5 years, the significance level (P) to 0.05 and implemented “prewhitening” on the time series, 289

Ecological Indicators 79 (2017) 286–294

S. Smoliński, Z. Mirny

similar to Möllmann et al. (2009) and Tomczak et al. (2013). “Prewhitening” removes the red noise component from the time series prior to an application of a STARS technique (Rodionov, 2006). To test the effects of the extrinsic environmental predictors (SST and BSI) we added them as additional fixed terms. In this step, the data were limited to the years in which both environmental variables were available (1948–2015). We applied an exploratory “sliding window” approach (van de Pol et al., 2016) considering mean aggregate statistics of variables from the current year according to growth increment formation, the absolute time window and the linear relationships between environmental variables and fish growth. We compared with AICc set of models that take into account different variables (SST or BSI) and their time windows. Based on these comparisons we identified the best predictor (i.e. one of the variables with its optimal time window). Then, we repeated the entire process of sliding window analysis using a model refitted with the added best predictor (identified in the first step) and found the optimal time window for the second environmental variable. We checked for possible collinearity of the identified predictors using a correlation test. We tested and satisfied the assumptions of homoscedasticity and normality by performing visual inspections of the final model residuals. Moreover, we performed spatial correlation tests between the extracted BLUP of the Year conditional mode and the water temperature for the region of North Atlantic Ocean. Spatial data on water temperature were extracted from the Royal Netherlands Meteorological Institute Climate Explorer website (http://climexp.knmi.nl). All data explorations, analyses and graphics generation were conducted using the R scientific computing language (R Development Core Team, 2011). Linear mixed effects modeling were performed using the lme4 (Bates et al., 2016) and effects (Fox, 2003) packages, while further identification of the optimal time windows for environmental variables were conducted using the climwin package (van de Pol et al., 2016).

Table 2 Variance components and estimates of random (a) and fixed (b, c) effects of the optimal growth model. The Random Age slopes for each FishID, Year and Cohort are denoted by “|”. a) Random effects FishID Age|FishID Year Age|Year Cohort Age|Cohort Residual

Variance ( ± SD) 0.027 0.023 0.002 0.004 0.005 0.004 0.066

(0.165) (0.150) (0.045) (0.063) (0.069) (0.066) (0.257)

Correlation

0.56 0.56 0.60

b) Intrinsic effects

Estimate (SE)

t-value

Intercept Age Age-at-capture Sex (male) Sex (female) Age: sex (male) Age: sex (female)

−2.435 −1.278 −0.108 −0.115 0.047 −0.119 −0.146

−20.07 −31.84 −3.41 −2.43 1.29 −2.21 −3.55

(0.081) (0.040) (0.031) (0.047) (0.037) (0.054) (0.041)

c) Environmental effects

Estimate (SE)

t-value

Mean BSI (August–December) Mean SST (April–June)

0.099 (0.043) 0.019 (0.009)

2.29 2.18

in the effect of Cohort was also apparent, indicating that some groups of fish from the same spawning season experienced persistently higher or lower growth through their lifetimes (Fig. S7). The STARS analysis was conducted on the best linear unbiased predictor of the Year effect to detect the timing of potential changes in the growth of fish. Strong alterations (p < 0.001) were found in the years 1988, 1992, 2006 (RSI = 0.34, −2.19 and 0.20, respectively). Comparing the STARS results with visual inspections of the developed otolith biochronology time series, the years 1988–1992 were considered as a short transition period between two separate regimes. The inclusion of environmental variables significantly improved the model and were supported by the AICc comparisons (Table S4). The investigated population reacted to changes in mean BSI and mean SST (Fig. 4). The first step of this systematic approach for identifying climatic signals revealed that the strongest positive relationships occurred between fish growth and mean BSI for the period August–December (Fig. 4a). Including this predictor significantly improved the model according to AICc (Table S4). In the second step of testing the environmental variables (using models refitted with the effect of BSI) the results showed that optimal time window for SST was April–June (Fig. 4b). The influence of thermal conditions during this period was supported by AICc (Table S4). Incorporating both environmental predictors in the model was possible due to the low correlation between them (R = 0.03). A synchrony of otolith biochronology and the investigated hydroclimatic conditions were noticed on the plots (Fig. 3b and c). Within the range of the mean BSI experienced by the population in the years 1948–2015 (from −0.42 to 0.34), European flounder growth has changed by 7.20% (Table 3, Fig. S8a), while the predicted effect of the mean SST within the range experienced by fish in this period (from 6.17 °C to 10.19 °C) was 7.27% (Table 3, Fig. S8b). When compared using the gridded SST data, the highest correlations occurred in the area surrounding the British Isles (Fig. 5). However, the developed biochronology was positively correlated with SST in the broader region of the Northeast Atlantic, including the Baltic Sea where

3. Results We developed a 74-year (1942–2015) otolith biochronology of European flounder growth (Figs. 3 and S2). The age of the individual specimens used in this study ranged between 4 and 20 years (Fig. S3). Based on AICc the optimal intrinsic effect model structure contained a random Age intercept and slopes for FishID, Year and Cohort (Table S2). Comparisons of the models with AICc indicated that inclusion of Age, Sex and Age-at-capture terms, as well as interactions between Age and Sex as fixed effects is supported for the investigated population (Table S3). According to the calculated conditional R2, the selected model explained 91.1% of the variance in fish growth. Analysis of variance of the components used for the random structure showed that a relatively high portion of the inter-annual variability in otolith increments was explained by the FishID and Age|FishID effects; however, a high level of unexplained variance was observed as residuals (Table 2). According to the parameter estimates of the fixed intrinsic effects, Age had the highest influence on European flounder growth among tested variables (Table 2, Fig. S4). Considerable sex-specific differences were also found, showing that females grow faster than males. A significant interaction between Age and Sex revealed that these differences are more prominent in the younger age classes (Table 2, Fig. S5). The model identified a negative relationship between fish growth and Age-at-capture, suggesting that faster-growing individuals are subject to capture at younger ages than are slower-growing fish (Table 2, Fig. S6). The calculated interclass correlation coefficients for Year (ICC = 0.02) and Cohort (ICC = 0.02) indicated low growth synchrony among individual fish. Nonetheless, there was considerable interannual variation in fish growth. Weak growth rates were noticeable in the years 1962, 1976 and 1996, while peaks in fish growth were observed in the years 1954, 1990 and 2012 (Fig. 3). A high variability 290

Ecological Indicators 79 (2017) 286–294

S. Smoliński, Z. Mirny

Fig. 4. Results of optimal time window identification for the environmental variables. the months in which the window was open or closed are shown on the axes. ΔAICc (indicated by the gray gradient) are the differences in AICc between the alternative models: (a) the base model with an added effect of BSI vs the model without BSI; and (b) the base model with effects of the optimal BSI signal (identified in the first step) with added SST vs the same model without SST (b).

mental conditions, including extreme events and regime shifts. We showed that the Baltic population of flounder may be sensitive to alterations driven by large-scale climate variations (i.e., BSI, which is highly correlated with the NAO Index). A systematic approach for identifying the optimal time window for these climatic drivers (van de Pol et al., 2016) revealed that the mean BSI for the period August–December was the strongest environmental signal affecting European flounder growth rate. In particular, wintertime NAO (and BSI) contributes to the weather and climate in the Baltic Sea (HELCOM, 2013). High winter NAO values promote primary productivity and, thus, influence the growth of higher trophic organisms. Similarly, a positive relationship between fish growth and NAO was found by Thresher et al. (2014) for the orange roughy in the Irish Slope region (NE Atlantic Ocean). Thermal regimes, which are often strongly linked to large-scale climatic indices, have also been identified in sclerochronological studies as a potential driver of marine fish growth (e.g., Gillanders

Table 3 Predicted effect (% change in growth) of the selected environmental variables on the growth of European flounder. Environmental variable

Predictor range

Predicted effect (%)

BSI (mean August–December) SST (mean April–June)

−0.42 to 0.34 6.17 to 10.19 °C

7.20 7.27

the otoliths were sampled.

4. Discussion In this study, we developed a 74-year growth biochronology for the commercially important European flounder growth based on otolith increment measurements. As recommended by Matta et al. (2016) our research covered long-term biological responses to changing environ-

Fig. 5. Map of spatial correlation between developed otolith biochronology (best linear unbiased predictor for the years 1942–2015) and gridded mean sea surface temperature (SST) for the period April–June. Cells with significant relationships (p < 0.1) are marked. The gray gradient indicates the value of the correlation coefficient. A black rectangle with white line patterns depicts the otolith sampling location. The dashed arrows indicate the main ocean currents in the North Atlantic.

291

Ecological Indicators 79 (2017) 286–294

S. Smoliński, Z. Mirny

2005). Because fish growth rates depend on both a set of intrinsic (e.g., individual’s age, sex) and extrinsic drivers (e.g., thermal conditions, intra- or interspecific competition), modeling of such ecological phenomena requires advanced techniques of statistical analysis for the proper identification of different sources of variability (Weisberg et al., 2010). Linear mixed modeling allows the incorporation of main intrinsic and extrinsic factors with additional sources of random variation (Morrongiello and Thresher, 2015; Weisberg et al., 2010; Zuur et al., 2009). This study revealed that biochronological approaches have the potential ability to detect periods of rapid change in ecosystems, that affect organisms’ growth patterns. The STARS analysis revealed abrupt changes in fish growth rates in 1988, 1992, and 2006—results that agree with previous studies on Baltic Sea ecosystem regime shifts. A STARS analysis conducted using the results of an ecological network model for the Baltic Proper (see Tomczak et al., 2012, 2013 for detailed descriptions) suggested that three alternative regimes have occurred since 1974: (i) up to 1988, (ii) between 1994 and 2006, and (iii) from 2006 onwards. The period between 1989 and 1993 was interpreted as a transition phase (Tomczak et al., 2013). Similarly, the STARS analysis conducted by Möllmann et al. (2009) using time series (1974–2005) data on key abiotic and biotic ecosystem components revealed two major ecosystem states: (i) up to 1987, (ii) from 1994 onwards, with a transition period between 1988 and 1993. In the late 1980s (1987–1989) strong changes from a positive to a negative phase have been observed for NAO and BSI, causing abrupt alterations in salinity, oxygen and temperature conditions in the region (Lindegren et al., 2010). Alterations in the Baltic ecosystem were initiated mainly by these climate-induced changes in the abiotic environment associated with NAO transition (Möllmann et al., 2009). Whole ecosystem structure and functioning has been profoundly reorganized, including changes in the composition of phyto- and zooplankton communities. During this period dinoflagellate abundance increased and diatom abundance decreased. Also zooplankton species, which constitute main food resources for fish, experienced changes. Biomass of Pseudocalanus sp. decreased, while Temora longicornis and Acartia spp. increased. Further, the bottom-up mechanisms caused the reaction of fish populations, observed as low level of Baltic cod biomass and strong increase of sprat stock (Alheit et al., 2005). Fisheriesinduced feedback loops in the food web stabilized this new alternative state (Möllmann et al., 2009). The topology of Baltic Sea ecosystem shifted from a web-like structure to more linearized food-web with high primary production and high fishing pressure on lower trophic level species (Tomczak et al., 2013). The same drivers, both directly and indirectly, seem to affect European flounder growth. Although the analysis presented in this paper focuses on the consequences (fish growth), rather than underlying reasons for alterations in the state of the ecosystem, routine STARS tests based on developed marine biochronologies can be valuable for identifying the potential points at which ecosystem regime shifts occur or in revealing extreme past environmental events. To our knowledge, the only formal investigation of regime shift based on fish sclerochronological data was conducted recently for Pacific ocean perch, using otolith chronology (van der Sleen et al., 2016). This study, based on long-term data, showed that ecosystem reorganization in the Bering Sea during 1976–1977 dominated the growth variability of this species. Van der Sleen et al. (2016) revealed that fish responses to extreme low-frequency environmental forces are not synchronous across taxa and regions. Similar findings were presented by Izzo et al. (2016), who showed that different fish species have different sensitivity to rapid environmental alterations based on their ecological requirements (e.g., specialists vs generalists). Thus, robust ecological inference with the relationships between fish growth and climatic factors should be preceded by comparisons among different populations or species to identify the differences and similarities in regional or species-specific

et al., 2012; Black et al., 2013; Morrongiello and Thresher, 2015; Izzo et al., 2016). The results obtained in our study are in line with these previous findings, showing a significant relationship between fish growth and water temperatures during a certain period of the year. By applying the “sliding window” technique we showed that the mean SST for the months April–June is the best predictor of fish somatic growth. Temperature conditions in April-June may be highly influential for European flounder somatic growth, as the investigated population begins to feed intensively after the spawning season during that period. However, our preliminary analysis revealed that only in a distinct period of the year (October–May) did mean monthly SST seem to be a good proxy of bottom temperatures (a crucial measure for demersal European flounder), a result that is partly due to thermal stratification in the Baltic Sea. Thus, we assume that there are also potential relationships between fish growth and thermal conditions during the remaining periods of the year that we were unable to detect due to insufficient hydrological data. Correlation tests between the developed biochronology and the gridded water temperatures conducted for the whole North Atlantic identified a clear environmental signal in the waters around the Europe. The relatively high positive correlations in this region were interpreted as additional evidence of the relationship between somatic growth and environmental thermal conditions. However, the highest correlation between fish growth rates and SST was observed in the region around the British Isles where the influence of the North Atlantic Current is more apparent than in the Baltic Sea basin. The hydrological conditions of bottom waters within the Baltic Sea depend heavily on either galeforced barotropic or baroclinic salt water inflows from the North Sea through the narrow and shallow Danish Straits (Lehmann et al., 2002; Omstedt et al., 2014). More saline and oxygen-rich waters from the North Sea inflow by intrusive layering under the Baltic permanent halocline due to higher density and refresh bottom waters in the stratified deeps (Omstedt et al., 2014). Thus, an indirect link between the hydrological situation in the North Sea (including SST of this region) and Baltic flounder growth may be reasonable. Physiological studies have determined that the optimal temperature for European flounder is 18 °C–22 °C (Fonds et al., 1992). Because the Baltic Sea is semi-enclosed and limits the dispersion of marine organisms, a projected increase of water temperatures could optimize the conditions for the growth of the investigated population. According to projections under the A2 and B2 emissions scenarios (IPCC, 2014), surface temperatures in the Baltic Sea will increase by approximately 2–3 °C over the 21st century (HELCOM, 2013). The increase will be even larger (a rise of 3–4 °C) in May and June (Meier et al., 2006), which are precisely the months in which the strongest climatic signal to predict European flounder growth was found. Our model showed that a change of 4 °C in the mean SST experienced by the investigated population over the years 1948–2015 equated to a 7% change in fish growth. Because individual growth rate is one of the crucial factors that affects the fish population productivity (Morrongiello et al., 2014), these results may be important for managing flatfish resources. However, future predictions of European flounder responses using the developed model seem to be uncertain at this time. According to Mackenzie et al. (2007) salinity and temperature are two abiotic factors that simultaneously affect and may have significant effects on Baltic fish populations. In this study we tested only the effects of BSI and SST, which were chosen a priori from a set of possible environmental drivers. Thus, more realistic predictions of the future European flounder growth require the incorporation of other factors, especially salinity, which is essential for the functioning of ecosystem and species biology. Difficulties in recognizing fish growth-hydroclimate relationships are often caused by the complex nature of these processes (Rountrey et al., 2014). Apart from direct effects on marine organisms (e.g., by physiologically enhancing growth rates), there could be a number of indirect consequences that could occur through a chain of physical and biological processes such as food resource restrictions (Alheit et al., 292

Ecological Indicators 79 (2017) 286–294

S. Smoliński, Z. Mirny

Fox, J., 2003. Effect displays in R for generalised linear models. J. Stat. Softw. 8, 1–27. http://dx.doi.org/10.2307/271037. Gillanders, B.M., Black, B.A., Meekan, M.G., Morrison, M.A., 2012. Climatic effects on the growth of a temperate reef fish from the Southern Hemisphere: a biochronological approach. Mar. Biol. 159, 1327–1333. http://dx.doi.org/10.1007/s00227-0121913-x. Goldstein, H., Browne, W., Rasbash, J., 2002. Partitioning variation in multilevel models. Underst. Stat. 1, 223–231. http://dx.doi.org/10.1207/S15328031US0104_02. HELCOM, 2013. Climate change in the Baltic Sea area, HELCOM thematic assessment in 2013. Balt. Sea Environ. Proc. 137, 54. Hare, S.R., Mantua, N.J., 2000. Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr. 47, 103–145. http://dx.doi.org/10.1016/S0079-6611(00) 00033-1. Helser, T.E., Lai, H., Black, B.A., 2012. Bayesian hierarchical modeling of Pacific geoduck growth increment data and climate indices. Ecol. Modell. 247, 210–220. http://dx. doi.org/10.1016/j.ecolmodel.2012.08.024. Hoegh-Guldberg, O., Bruno, J.F., 2010. The impact of climate change on the world’s marine ecosystems. Science 328, 1523–1528. Huang, B., Banzon, V.F., Freeman, E., Lawrimore, J., Liu, W., Peterson, T.C., Smith, T.M., Thorne, P.W., Woodruff, S.D., Zhang, H.M., 2015. Extended reconstructed sea surface temperature version 4 (ERSST.v4). Part I: upgrades and intercomparisons. J. Clim. 28, 911–930. http://dx.doi.org/10.1175/JCLI-D-14-00006.1. IPCC, 2014. Climate change 2014: synthesis report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva. http://dx.doi.org/10.1017/CBO9781107415324. Izzo, C., Doubleday, Z.A., Grammer, G.L., Barnes, T.C., Delean, S., Ferguson, G.J., Ye, Q., Gillanders, B.M., 2016. Multi-species response to rapid environmental change in a large estuary system: a biochronological approach. Ecol. Indic. 69, 739–748. http:// dx.doi.org/10.1016/j.ecolind.2016.05.019. Kraberg, A.C., Wasmund, N., Vanaverbeke, J., Schiedek, D., Wiltshire, K.H., Mieszkowska, N., 2011. Regime shifts in the marine environment: the scientific basis and political context. Mar. Pollut. Bull. 62, 7–20. http://dx.doi.org/10.1016/j.marpolbul.2010.09. 010. Lehmann, A., Krauss, W., Hinrichsen, H.H., 2002. Effects of remote and local atmospheric forcing on circulation and upwelling in the Baltic Sea. Tellus A 54, 299–316. http:// dx.doi.org/10.1034/j.1600-0870.2002.00289.x. Lindegren, M., Diekmann, R., Möllmann, C., 2010. Regime shifts, resilience and recovery of a cod stock. Mar. Ecol. Prog. Ser. 402, 239–253. http://dx.doi.org/10.3354/ meps08454. Möllmann, C., Diekmann, R., Müller-karulis, B., Kornilovs, G., Plikshs, M., Axe, P., 2009. Reorganization of a large marine ecosystem due to atmospheric and anthropogenic pressure: a discontinuous regime shift in the Central Baltic Sea. Glob. Change Biol. 15, 1377–1393. http://dx.doi.org/10.1111/j.1365-2486.2008.01814.x. Mackenzie, B.R., Gislason, H., Möllmann, C., Köster, F.W., 2007. Impact of 21st century climate change on the Baltic Sea fish community and fisheries. Glob. Change Biol. 13, 1348–1367. http://dx.doi.org/10.1111/j.1365-2486.2007.01369.x. Matta, M.E., Black, B.A., Wilderbuer, T.K., 2010. Climate-driven synchrony in otolith growth-increment chronologies for three Bering Sea flatfish species. Mar. Ecol. Prog. Ser. 413, 137–145. http://dx.doi.org/10.3354/meps08689. Matta, M.E., Helser, T.E., Black, B.A., 2016. Otolith biochronologies reveal latitudinal differences in growth of Bering Sea yellowfin sole Limanda aspera. Pol. Biol. 39, 2427–2439. http://dx.doi.org/10.1007/s00300-016-1917-y. Meier, H.E.M., Feistel, R., Piechura, J., Arneborg, L., Burchard, H., Fiekas, V., Golenko, N., Kuzmina, N., Mohrholz, V., Nohr, C., Paka, V.T., Sellschopp, J., Stips, A., Zhurbas, V., 2006. Ventilation of the Baltic Sea deep water: a brief review of present knowledge from observations and models. Oceanologia 48, 133–164. http://dx.doi. org/10.1007/s00382-006-0124-x. Morrongiello, J., Thresher, R., 2015. A statistical framework to explore ontogenetic growth variation among individuals and populations: a marine fish example. Ecol. Monogr. 85, 93–115. Morrongiello, J.R., Crook, D.A., King, A.J., Ramsey, D.S.L., Brown, P., 2011. Impacts of drought and predicted effects of climate change on fish growth in temperate Australian lakes. Glob. Change Biol. 17, 745–755. http://dx.doi.org/10.1111/j.13652486.2010.02259.x. Morrongiello, J.R., Thresher, R.E., Smith, D.C., 2012. Aquatic biochronologies and climate change. Nat. Clim. Change 2, 849–857. http://dx.doi.org/10.1038/ nclimate1616. Morrongiello, J.R., Walsh, C.T., Gray, C.A., Stocks, J.R., Crook, D.A., 2014. Environmental change drives long-term recruitment and growth variation in an estuarine fish. Glob. Change Biol. 20, 1844–1860. http://dx.doi.org/10.1111/gcb. 12545. Nakagawa, S., Schielzeth, H., 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol. Evol. 4, 133–142. http://dx. doi.org/10.1111/j.2041-210x.2012.00261.x. 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. http://dx.doi.org/10.1016/j.seares.2010.02.001. Omstedt, A., Elken, J., Lehmann, A., Leppäranta, M., Meier, H.E.M., Myrberg, K., Rutgersson, A., 2014. Progress in physical oceanography of the Baltic Sea during the 2003–2014 period. Prog. Oceanogr. 128, 139–171. http://dx.doi.org/10.1016/j. pocean.2014.08.010. Ong, J.J.L., Rountrey, A.N., Zinke, J., Meeuwig, J.J., Grierson, P.F., O’Donnell, A.J., Newman, S.J., Lough, J.M., Trougan, M., Meekan, M.G., 2016. Evidence for climatedriven synchrony of marine and terrestrial ecosystems in northwest Australia. Glob. Change Biol. 22, 2776–2786. http://dx.doi.org/10.1111/gcb.13239. Poloczanska, E.S., Brown, C.J., Sydeman, W.J., Kiessling, W., Schoeman, D.S., Moore,

responses (Morrongiello et al., 2012). Age-reading routines have already been developed for many species; automatic measurements of otolith annual rings and easily accessible databases containing growth increment records in many labs have substantially reduced the effort required for such research. Assessments of different populations and species responses to past environmental changes using a sclerochronological approach may allow the adaptation of future management policies and the development of ecosystem-level indicators. We stress the urgent need for such studies—especially now, in a time of rapid global climate changes. Conflict of interest The authors declare no conflict of interests. Acknowledgments Otolith samples were acquired from the National Marine Fisheries Research Institute archive. A portion of the otoliths were collected during the National Programme for Fisheries Data Collection under the European Union Data Collection Framework. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. We would like to acknowledge all who have been involved over the years in systematically collecting European flounder otolith samples. The authors are also grateful to the anonymous reviewers for their valuable comments on the manuscript. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ecolind.2017.04.028. References Alheit, J., Möllmann, C., Dutz, J., Kornilovs, G., Loewe, P., Mohrholz, V., Wasmund, N., 2005. Synchronous ecological regime shifts in the central Baltic and the North Sea in the late 1980. ICES J. Mar. Sci. 62, 1205–1215. http://dx.doi.org/10.1016/j.icesjms. 2005.04.024. Aro, E., 1989. Review of fish migration patterns in the Baltic Sea. Rapp. Proce’s-Verbaux des Réunions. Cons. Int. pour L’Explor. Mer 190, 72–96. Bates, D., Maechler, M., Bolker, B., Walker, S., Christensen, R.H.B., Singmann, H., Dai, B., Grothendieck, G., 2016. Lme4: Linear Mixed-Effects Models Using Eigen and S4. R Package Version 1. pp. 1–12. http://dx.doi.org/10.18637/jss.v067.i01. Available at:. http://cran.r-project.org/web/packages/lme4/lme4.pdf. Black, B.A., Matta, M.E., Helser, T.E., Wilderbuer, T.K., 2013. Otolith biochronologies as multidecadal indicators of body size anomalies in yellowfin sole (Limanda aspera). Fish. Oceanogr. 22, 523–532. http://dx.doi.org/10.1111/fog.12036. Brander, K.M., 2007. Global fish production and climate change. Proc. Natl. Acad. Sci. U. S. A. 104, 19709–19714. http://dx.doi.org/10.1073/pnas.0702059104. Burnham, K.P., Anderson, R.P., 2004. Multimodel inference: understanding AIC and BIC in model selection. Sociol. Methods Res. 33, 261–304. http://dx.doi.org/10.1177/ 0049124104268644. CMEMS, 2016. Copernicus—Marine Environment Monitoring Service. Available at www. marine.copernicus.eu [WWW Document]. URL http://marine.copernicus.eu/ (Accessed 22 August 2016). Campana, S.E., Thorrold, S.R., 2001. Otoliths, increments, and elements: keys to a comprehensive understanding of fish populations? Can. J. Fish. Aquat. Sci. 58, 30–38. http://dx.doi.org/10.1139/f00-177. Campana, S.E., 1990. How reliable are growth back-calculations based on otoliths? Can. J. Fish. Aquat. Sci. 47, 2219–2227. Daskalov, G.M., 2002. Overfishing drives atrophic cascade in the Black sea. Mar. Ecol. Prog. Ser. 225, 53–63. http://dx.doi.org/10.3354/Meps225053. Doubleday, Z.A., Izzo, C., Haddy, J.A., Lyle, J.M., Ye, Q., Gillanders, B.M., 2015. Longterm patterns in estuarine fish growth across two climatically divergent regions supp. Oecologia 179, 1079–1090. http://dx.doi.org/10.1007/s00442-015-3411-6. Draganik, B., Kuczyński, J., 1993. A review of growth rate of the Baltic Flounder [Platichthys flesus (L.)] derived from otolith measurements. Bull. Sea Fish. Inst. 130, 21–36. Florin, A.-B., Höglund, J., 2008. Population structure of flounder (Platichthys flesus) in the Baltic Sea: differences among demersal and pelagic spawners. Heredity (Edinb.) 101, 27–38. http://dx.doi.org/10.1038/hdy.2008.22. Fonds, M., Cronie, R., Vethaak, A.D., Van Der Puyl, P., 1992. Metabolism, food consumption and growth of plaice (Pleuronectes platessa) and flounder (Platichthys flesus) in relation to fish size and temperature. Neth. J. Sea Res. 29, 127–143. http:// dx.doi.org/10.1016/0077-7579(92)90014-6.

293

Ecological Indicators 79 (2017) 286–294

S. Smoliński, Z. Mirny

Thresher, R., Morrongiello, J., Sloyan, B.M., Krusic-Golub, K., Shephard, S., Minto, C., Nolan, C.P., Cerna, F., Cid, L., 2014. Parallel decadal variability of inferred water temperatures for Northern and Southern Hemisphere intermediate water masses. Geophys. Res. Lett. 41, 1232–1237. http://dx.doi.org/10.1002/2013GL058638. Tomczak, M.T., Niiranen, S., Hjerne, O., Blenckner, T., 2012. Ecosystem flow dynamics in the Baltic proper-using a multi-trophic dataset as a basis for food-web modelling. Ecol. Modell. 230, 123–147. http://dx.doi.org/10.1016/j.ecolmodel.2011.12.014. Tomczak, M.T., Heymans, J.J., Yletyinen, J., Niiranen, S., Otto, S.A., Blenckner, T., 2013. Ecological network indicators of ecosystem status and change in the Baltic Sea. PLoS One 8, 1–11. http://dx.doi.org/10.1371/journal.pone.0075439. van de Pol, M., Bailey, L.D., McLean, N., Rijsdijk, L., Lawson, C.R., Brouwer, L., Gimenez, O., 2016. Identifying the best climatic predictors in ecology and evolution. Methods Ecol. Evol. 7, 1246–1257. http://dx.doi.org/10.1111/2041-210X.12590. van der Sleen, P., Dzaugis, M.P., Gentry, C., Hall, W.P., Hamilton, V., 2016. Long-term Bering Sea environmental variability revealed by a centennial- length biochronology of Pacific ocean perch (Sebastes alutus). Clim. Res. 71, 33–45. Weisberg, S., Spangler, G., Richmond, L.S., 2010. Mixed effects models for fish growth. Can. J. Fish. Aquat. Sci. 277, 269–277. http://dx.doi.org/10.1139/F09-181. Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., Smith, G.M., 2009. Mixed Effects Models and Extensions in Ecology with R. Springer, New York. http://dx.doi.org/10. 1017/CBO9781107415324.004. Journal of Chemical Information and Modeling.

P.J., Brander, K., Bruno, J.F., Buckley, L.B., Burrows, M.T., Duarte, C.M., Halpern, B.S., Holding, J., Kappel, C.V., O’Connor, M.I., Pandolfi, J.M., Parmesan, C., Schwing, F., Thompson, S.A., Richardson, A.J., 2013. Global imprint of climate change on marine life. Nat. Clim. Change 3, 919–925. http://dx.doi.org/10.1038/ Nclimate1958. R Development Core Team, 2011. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Reynolds, D.J., Richardson, C.A., Scourse, J.D., Butler, P.B., Hollyman, P., RománGonzález, A., Hall, I.R., 2016. Reconstructing North Atlantic marine climate variability using an absolutely-dated sclerochronological network. Palaeogeogr. Palaeoclimatol. Palaeoecol. 465, 333–346. http://dx.doi.org/10.1016/j.palaeo.2016. 08.006. Rodionov, S.N., 2004. A sequential algorithm for testing climate regime shifts. Geophys. Res. Lett. 31, 2–5. http://dx.doi.org/10.1029/2004GL019448. Rodionov, S.N., 2006. Use of prewhitening in climate regime shift detection. Geophys. Res. Lett. 33, 1–4. http://dx.doi.org/10.1029/2006GL025904. Rountrey, A., Coulson, P.G., Meeuwig, J.J., Meekan, M., 2014. Water temperature and fish growth: otoliths predict growth patterns of a marine fish in a changing climate. Glob. Change Biol. 20, 2450–2458. http://dx.doi.org/10.1111/gcb.12617. Rowland, E.L., Davison, J.E., Graumlich, L.J., 2011. Approaches to evaluating climate change impacts on species: a guide to initiating the adaptation planning process. Environ. Manage. 47, 322–337. http://dx.doi.org/10.1007/s00267-010-9608-x.

294