Marine Ecology. ISSN 0173-9565

ORIGINAL ARTICLE Trends in population dynamics and fishery of Parapenaeus longirostris and Nephrops norvegicus in the Tyrrhenian Sea (NW Mediterranean): the relative importance of fishery and environmental variables Alessandro Ligas1, Paolo Sartor1 & Francesco Colloca2

1 Centro Interuniversitario di Biologia Marina ed Ecologia Applicata ‘G. Bacci’, viale N. Sauro 4, Livorno, Italy 2 Dipartimento di Biologia Animale e dell’Uomo, University of Rome ‘La Sapienza’, viale dell’Universita` 32, Rome, Italy

Keywords Abstract Environmental variables; fishing effort; landings; series analysis; trawl survey. Temporal variation in the population abundance of the deep-water rose shrimp, Parapenaeus longirostris (Lucas, 1846) (Decapoda, Penaeidae), and the Correspondence Norway lobster, Nephrops norvegicus (Linnaeus, 1758) (Decapoda, Nephropi- A. Ligas, Centro Interuniversitario di Biologia dae), in the Tyrrhenian Sea (NW Mediterranean), were evaluated using Marina ed Ecologia Applicata ‘G. Bacci’, viale time-series data (1994–2008) from experimental trawl surveys and commercial N. Sauro 4, I-57128 Livorno, Italy. landings. The influence of several environmental variables (sea surface tempera- E-mail: [email protected] ture, wind-mixing index and NAO index) and fishing effort indices (number Accepted: 1 February 2011 of days at sea per month and mean engine power of the trawl fleet) were inves- tigated. The time series were analysed by means of min ⁄ max auto-correlation doi:10.1111/j.1439-0485.2011.00440.x factor analysis (MAFA) and dynamic factor analysis (DFA). The abundance of P. longirostris showed a clear increasing trend, significantly correlated with the fishing effort index (number of days at sea per month), the sea surface temper- ature and the wind-mixing index. The temporal variations in the stock of P. longirostris, which has a preference for warm waters, were positively corre- lated with the rise of the sea surface temperature and the decrease of wind cir- culation. For N. norvegicus, an increasing trend of landings per unit of effort and recruitment index contrasted with a decreasing trend of relative population abundance (biomass and density indices).

(e.g. higher growth rate, earlier age-at-maturity) (Fromen- Introduction tin & Fonteneau 2001), effects on the populations of Understanding the causes and mechanisms of change in non-target species (e.g. cetaceans, sea birds, sea turtles) the abundance of species over time is a crucial issue in resulting from by-catch (Kaiser & De Groot 2000), sus- marine ecology. Fishing exploitation is considered to be pension of superficial sediments (Smith et al. 2003), and one of the main factors determining the dynamics of a reduction of habitat complexity and alteration of ben- marine populations and ecosystems (Baum et al. 2003; thic community structure (Kaiser et al. 2000). Morato et al. 2006; Ciannelli et al. 2008; Pauly 2009). Large-scale changes in climate and oceanographic condi- These effects include: changes in predator–prey relation- tions are also known to influence the dynamics of marine ships that lead to shifts in food-web structure (Cartes populations (Gislason et al. 2000; Lloret et al. 2001; et al. 2001), effects on abundance and body-size distribu- Rothschild et al. 2005). For example, the influence of tions that can result in fauna dominated by small-size global warming in the 20th century on long-term changes individuals (Jennings et al. 2001), genetic selection for in phytoplankton concentration in the North Atlantic different physical characteristics and reproductive traits has been demonstrated (Reid et al. 2001). In the

Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH 25 Trends in P. congirostris and N. norvegicus Ligas, Sartor & Colloca

Mediterranean Sea, enrichment of nutrients in surface ing effort, mainly due to the EU Common Fishery Policy, waters has been shown to affect pelagic food-web dynamics which promotes the reduction of fishing effort through and fishery productivity (Molinero et al. 2008). This in incentives to decommission. turn can affect deep-sea necto-benthic communities which To analyse the effects of environmental and anthropo- are known to depend on the downward flux of organic genic factors on demersal communities, two species which matter from the surface layers (Company et al. 2008). display different life cycles and behavioural strategies were Additionally, changes in river discharge and surface selected: the deep-water rose shrimp, Parapenaeus longi- production can alter trophic webs and assemblage compo- rostris (Lucas, 1846) and the Norway lobster Nephrops sitions in deep Mediterranean waters (Cartes et al. 2009). norvegicus (Linnaeus, 1758). Parapenaeus longirostris is a Company et al. (2008) described how climate-driven fast-growing, short-lived species with thermophilic prefer- cascading dense shelf water influences the ecology of deep- ence (Abello` et al. 2002), inhabiting the water column sea populations on a decadal timescale. Bartolino et al. layers close to the seabed. Nephrops norvegicus is a long- (2008) linked the wind circulation to the recruitment of lived decapod, typical of temperate and cold waters, the European hake, Merluccius merluccius, one of the most which dwells in burrows and exerts territorial behaviour important demersal species in Mediterranean waters. Many (Aguzzi et al. 2003). It was hypothesised that, due to other studies have highlighted significant relationships these contrasting characteristics, the two species would between large-scale atmospheric variables (such as the show different behaviours in relation to changes in envi- North Atlantic Oscillation index, NAO, which is tradition- ronmental and anthropogenic factors. From the results, ally defined as the normalised pressure difference between we suggest a mechanism and cause–effect relationships Azores and Iceland) or local scale (surface temperature, linking the atmospheric and environmental variables with wind circulation, etc.) atmospheric variables and demersal changes in the abundance of both species. populations (Lloret et al. 2001; Farin˜a & Gonza´lez Herraiz 2003; Zuur et al. 2003a,b; Zuur & Pierce 2004; Erzini 2005; Study area Maynou 2008; Cartes et al. 2009; Gonza´lez Herraiz et al. 2009). The study area covered part of the continental shelf and the Since the 1950s, a warming process has occurred in the upper and middle slope off the western coasts of Italy (Cen- Western Mediterranean basin. This is demonstrated by tral-northern Tyrrhenian Sea, Fig. 1). The Tyrrhenian Sea both environmental changes (e.g. surface temperature is semi-enclosed between islands (Corsica, Sardinia and increase; see Vargas-Ya´nez et al. 2009) and biological Elba) and the mainland (Italy), and is separated from the changes (e.g. northward advance of thermophilic species; rest of the western basin by a channel of moderate depth. It see CIESM 2008). In addition to environmental changes, can therefore be considered a distinct entity within the Italian fishing grounds, similar to those of other Mediter- Central-western Mediterranean basin (Artale et al. 1994; ranean countries, have been affected by a decrease of fish- Gasparini et al. 2005). The circulation in the Tyrrhenian

Fig. 1. Study area; the main isobaths are shown, as well as the sampling stations investigated during the experimental trawl survey Medits 2008. The black triangles show the three points at which satellite data were collected (4230¢ N, 1100¢ E; 4200¢ N, 1200¢ E; and 4100¢ N, 1300¢ E).

26 Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH Ligas, Sartor & Colloca Trends in P. congirostris and N. norvegicus

Sea is organised in a series of cyclonic (anti-clockwise) and recruits per square kilometre, no. recruitsÆkm–2) was also anticyclonic (clockwise) gyres determined by the wind (Ar- computed. Following Mori et al. (2000) and Orsi Relini tale et al. 1994). Three main cold water gyres, two cyclonic et al. (1998), specimens under the size of 20 mm CL (car- and one anticyclonic, have been detected. They undergo apace length) were considered recruits. The indices of significant seasonal change, particularly the central anticy- N. norvegicus were computed taking into account only clonic gyre that spreads over most of the basin in spring the hauls carried out in the 200–800 m depth stratum. and summer and almost disappears in autumn and winter. To investigate the effect of hydrological conditions, The intermediate (LIW) and deep waters have a constant mean monthly values of satellite-derived (1994–2008) sea ) temperature (12.8–13.0 C). Mixing of surface and deep surface temperature (SST, C) and wind speed (W, mÆs 1) layers by wind-driven turbulence enriches the upper layer were gathered from the Physical Oceanography Distrib- with nutrients (Nezlin et al. 2004), giving the Tyrrhenian uted Active Archive Centre (PO.DAAC: http://pod- Sea a relatively high concentration of nutrients within the aac.jpl.nasa.gov/index.html). Data taken from three Mediterranean basin. locations (4230¢ N–1100¢ E; 4200¢ N–1200¢ E and 4100¢ N–1300¢ E; see Fig. 1) in the Tyrrhenian Sea were used to compute a mean monthly value. A wind-mixing Material and methods index was calculated as the cube of the wind speed From 1994 to 2008, landing data were collected monthly according to Bartolino et al. (2008). Monthly data of the over 3–5 days of observation at the auction of Porto NAO from 1994 to 2008 were obtained from the Pacific Santo Stefano, one of the most important fishing har- Fisheries Environmental Laboratory (PFEL: http://las. bours of the area. The exploitation of Parapenaeus longi- pfeg.noaa.gov/). rostris takes place in the fishing grounds between 200 and The time series were explored by means of auto- and 400 m depth, while catches of Nephrops norvegicus are cross-correlation functions. The auto-correlation function obtained from a greater depth range (200–600 m) (Sbrana gives an indication of the amount of association between et al. 2003). The number of trawlers habitually targeting variables Yt and Yt–k, where the time lag k takes the val- the two species decreased during the investigated period: ues 1, 2, 3, etc. (Zuur et al. 2007). Thus it is used to from 30 vessels in 1994 to 12 in 2008 (Sbrana et al. highlight the presence of cyclic patterns in time series. 2006). Data on specific composition of the landing (total Formulated differently, the auto-correlation with a time weight by species or commercial category) and fishing lag of k years represents the overall association between effort (number of fishing days) were collected for each values that are separated by k time points. vessel. The landing rates (landing per unit of effort, The cross-correlation function shows the relationship

LPUE) were calculated by taking into account the fishing between Yt and Xt–k. Therefore this tool can be used to day as a unit of effort (kg per day per vessel). In addition, explore whether there is a (linear) relationship between two indices of fishing activity and capacity were com- two variables (Zuur et al. 2007, 2009). In time series anal- puted: (i) the total number of days at sea performed by ysis, the use of significantly cross-correlated variables the fleet per month, and (ii) the mean engine power should be avoided. The confidence intervals of the auto- (kW) of the fleet per month. correlation were obtained from ±2 ⁄n, where n is the During the investigated period (1994–2008), two exper- length of the time series. imental trawl surveys per year were carried out under the To analyse the long-term changes of the variables, framework of the International bottom trawl survey in cyclical patterns were removed from the data obtained by the Mediterranean (Medits) and the Italian demersal the seasonal decomposition by Loess smoothing (Zuur resources program (Grund). According to the sampling et al. 2007). The data were then analysed by means of protocols (see Relini 1998 and Bertrand et al. 2002), the multivariate time series analysis techniques: min ⁄ max Medits survey was performed in spring and the Grund auto-correlation factor analysis (MAFA) and dynamic fac- survey in autumn. The two surveys were carried out tor analysis (DFA) to estimate common trends. These according to a depth-stratified sampling design with ran- tools were used to estimate common underlying trends domly allocated hauls within each stratum. In addition, from the multiple time series dataset, and to evaluate the the number of hauls in each stratum was proportional to correlations with species abundance and environmental the surface of the stratum itself. The haul position of the and fishery factors. For this purpose, the time series of Medits trawl survey 2008 is shown in Fig. 1. LPUE and of biomass and density indices were used as ) ) Mean biomass (kgÆkm 2) and abundance (NÆkm 2) response variables and the environmental and fishing indices were calculated for both species to obtain time effort factors as explanatory variables. All analyses were series composed of two observations per year, for a total performed using the software BRODGAR 2.6.6 (http:// of 30 observations. A recruitment index (number of www.brodgar.com).

Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH 27 Trends in P. congirostris and N. norvegicus Ligas, Sartor & Colloca

MAFA can be described in various ways: a type of were significantly correlated. In the case of Parapenaeus principal component analysis especially for (short) time longirostris, LPUE time series were positively correlated to series; a method for extracting trends from multiple experimental trawl survey time series (biomass and den- time series; a method for estimating index functions from sity indices) (Table 1a), suggesting a good match between time series; a smoothing method; or a signal extraction fishery-dependent and fishery-independent data. The procedure. The underlying idea is that a trend is associ- recruitment index was significantly correlated to the ated with high auto-correlation at time lag 1. Therefore, biomass and density indices, and the maximum cross- the first MAFA axis represents the trend, or the main correlations were at time lags 0 and 2, respectively underlying pattern in the data. This axis can also be seen (Table 1b). This means that a peak of recruitment was as an index function or smoothing curve. Cross-correla- directly reflected in a peak of density, while the peak of tions (canonical correlations) between the variables (both biomass followed with a time lag 2, which corresponds to response and explanatory variables) and the trends were 1 year. The correlations between the landing and survey computed to evaluate the significance of the relationship time series of Nephrops norvegicus were significant, but between the variables and the trends (Erzini et al. 2005; negative, suggesting an inverse relationship between the Zuur et al. 2007). The mathematics behind MAFA are two variables. However, the recruitment index was posi- described in Solow (1994). The underlying formula is tively correlated to the LPUE, with the maximum correla- similar to principal component analysis. The MAFA cal- tion corresponding to a time lag 3. This suggests that a culations involve a principal component analysis on cen- peak in recruitment was followed by a peak in LPUE after tred data, followed by a first-differencing on the principal a time lag of more than 1 year. components, and a second principal component analysis The analysis of cross-correlations among the explana- on these differenced components. As a result, the MAFA tory variables also provided significant results (Table 2). axes are mutually uncorrelated with unit variance, and the MAFA axes have decreasing auto-correlation with time lag 1 (Zuur et al. 2007). Table 1. Response variables. (a) Cross-correlations. L = landing per The DFA is a method to estimate common trends, unit of effort (kg per day per vessel); B = biomass index (kgÆkm)2); D effects of explanatory variables and interactions between = density index (nÆkm)2); R = recruitment index (no. recruitsÆkm)2). (b) the response variables in a multivariate time series data- Maximum cross-correlations: the time lags corresponding to the maxi- set. Statistical details and applications of DFA are given mum cross-correlations are shown in the grey part of the table. in Zuur et al. (2003a,b) and Zuur & Pierce (2004). DFA Parapenaeus longirostris Nephrops norvegicus applies a dimension reduction to the N time series. The dynamic factor model, in words, is given by N Time LBDRLBDR series = linear combination of M common trends + (a) explanatory variables + noise. P. longirostris DFA models with one common trend and a symmetric, L 1.00 non-diagonal covariance matrix were used to analyse the B 0.54 1.00 datasets. A series of models were fitted, ranging from the D 0.49 0.95 1.00 simplest, with only one explanatory variable, to the most R 0.12 0.46 0.69 1.00 N. norvegicus complex, with all the explanatory variables. Akaike’s L 1.00 information criterion (AIC) was used as a measure of B )0.55 1.00 goodness-of-fit and to compare models (Zuur et al. D )0.54 0.94 1.00 2003b), with the best model having the smallest AIC. Fac- R 0.42 0.38 0.52 1.00 tor loadings were used to make inferences regarding the (b) importance of particular trends, representing underlying P. longirostris common patterns over time, both to specific response L 0.79 0.82 0.54 B )1 0.95 0.53 variables and to different groups of response variables D1 0 0.69 (Erzini 2005; Erzini et al. 2005; Zuur et al. 2007). R120 N. norvegicus L 0.66 0.66 0.54 Results B )1 0.94 )0.47 The analysis of the time series dataset by means of D10)0.55 ) cross-correlation functions allows us to identify significant R311 relationships between the response variables. For both Significance level for correlations: ± 0.37; significant correlations are species, time series of landings and trawl surveys data highlighted in bold.

28 Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH Ligas, Sartor & Colloca Trends in P. congirostris and N. norvegicus

Table 2. Explanatory variables: cross-correlations. 1.0 0.8 3 SST W NAO Days kW 0.6 0.4 SST 1.00 0.2 W3 )0.45 1.00

Correlation 0.0 NAO 0.19 )0.12 1.00 –0.2 Days 0.30 )0.02 )0.02 1.00 –0.4 kW 0.18 0.14 0.00 0.41 1.00 0 2 4 6 8 10121416182022242628303234363840 Time lag SST = Sea Surface Temperature (C); W3 = wind-mixing index (m3Æs)3); NAO = North Atlantic Oscillation index; Days = days at sea Fig. 3. Auto-correlation plot of the landing per unit of effort (LPUE) per month; kW = mean engine power (kW). Significance level for time series of Parapenaeus longirostris and Nephrops norvegicus (thick correlations: ± 0.37. Significant correlations are highlighted in bold. line). Dotted lines: confidence interval limits (± 0.15).

Sea surface temperature (SST) and wind-mixing index 600 A 3 (W ) were negatively correlated. SST showed an increas- 500 3 ing trend, whereas the W series followed a decreasing 400 pattern. For the fishing efforts parameters, a positive cor- 300 relation was found between days at sea and mean engine 200

power. Days per month 100 According to the results obtained by means of cross- 0 correlations, it was decided to use the LPUE of P. longi- 30 B rostris and N. norvegicus only as response variables for the 26 analyses of time series. Among the explanatory variables, 22 the wind-mixing index, the NAO index and the days at 18 sea per month were used. SST (°C) The LPUE time series of P. longirostris and N. norvegi- 14 cus were characterized by wide fluctuations, making it 10 impossible to identify any clear trend (Fig. 2). Both time 12 C series showed peaks in late spring (April–June), when it is 10 known that the catch of the two species is higher. The 8 presence of a seasonal pattern was confirmed by the auto- –1 6 m·s correlation function: significant correlations were identi- 4 fied at time lags of 12 and 24 months (Fig. 3). Some 2 examples of the time series of the explanatory variables 0 Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 are shown in Fig. 4. The time series of days at sea per Months month was characterised by a clear decreasing trend (Fig. 4A). Figure 4B and C show fluctuations over time of Fig. 4. Time series plot of monthly data of explanatory variables from sea surface temperature and wind speed. Sea surface tem- January 1994 to December 2008. (A) Number of days at sea per month (computed from the Porto Santo Stefano trawl fleet). (B) Sea perature peaked in summer, when the wind speed was ) surface temperature (SST, C). (C) Wind speed (mÆs 1).

70 lower. These trends explain the significant, but negative, 60 cross-correlation between the two variables. To remove 50 the seasonal patterns, the explanatory variables were also 40 smoothed by means of the seasonal decomposition by per vessel 30 –1 Loess smoothing (see Fig. 5). 20 The results obtained by means of MAFA described a 10 kg·day clear scenario in the case of P. longirostris. The estimated 0 Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 trend showed an increasing pattern, although character- Months ized by fluctuations (Fig. 6). High and positive correla- Fig. 2. Landing per unit of effort (LPUE) time series of Parapenaeus tions between the trend and the LPUE time series of the longirostris and Nephrops norvegicus (thick line) from the Porto Santo two species were identified (Table 3). While for P. longi- Stefano trawl fleet. rostris all the response variables considered (LPUE,

Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH 29 Trends in P. congirostris and N. norvegicus Ligas, Sartor & Colloca

0.3 According to the factor loadings, only P. longirostris

0.2 was correlated to the trend computed by means of DFA (0.223 for P. longirostris, and 0.006 for N. norvegicus). 0.1 The estimated trend (Fig. 7) was similar to that obtained

0.0 using MAFA, with a general increasing pattern and two main peaks in 2001 and 2006. However, this model, –0.1 SST (deseasonalised) which was characterized by the lowest AIC value –0.2 (Table 4), suggested no significant relationship with fish- 1.0 ing effort, as it was correlated only to the monthly time 3 0.5 series of wind-mixing index (W ) and the NAO index. In fact, the estimated t-values for the regressions for 0.0 individual species with W3 and NAO were relatively large

–0.5 (deseasonalised) Wind mixing index –1.0 Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Months 10 Fig. 5. Sea surface temperature (SST) and wind-mixing index time series obtained by means of seasonal decomposition by Loess smooth- ing.

0

0.3

0.2

0.1

Score 0.0 50 100 150 Time –0.1

–0.2 Fig. 7. Common trend computed by means of DFA from the LPUE Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 time series of Parapenaeus longirostris and Nephrops norvegicus. Months

Fig. 6. Common trend computed by means of MAFA from the LPUE Table 4. Values of Akaike’s information criterion (AIC) for DFA mod- time series of Parapenaeus longirostris and Nephrops norvegicus. els with one common trend and different sets of explanatory variables (W3 = wind-mixing index; NAO = North Atlantic Oscillation index; Days = number of days at sea per month), based on diagonal and Table 3. Canonical correlations between the common trend obtained symmetric matrices. through MAFA and, respectively, the response variables (landing per unit of effort time series of Parapenaeus longirostris and Nephrops Model Matrix Explanatory variables AIC norvegicus), and the explanatory variables (W3 = wind-mixing index; NAO = North Atlantic Oscillation index; Days = number of days at sea 1 Diagonal W3, NAO, Days 319.7 per month, days). 2 Diagonal W3, NAO 298.4 3 Diagonal W3, Days 307.5 Response variables Explanatory variables 4 Diagonal NAO, Days 178.2 5 Diagonal W3 291.5 P. longirostris 0.97 W3 )0.45 6 Diagonal NAO 388.3 N. norvegicus 0.35 NAO )0.09 7 Diagonal Days 248.5 Days )0.31 8 Diagonal – 258.0 3 Significance level for correlations: ± 0.15. Significant values are high- 9 Symmetric W , NAO, Days 296.3 3 lighted in bold. 10 Symmetric W , NAO 159.0 11 Symmetric W3, Days 348.5 12 Symmetric NAO, Days 279.8 biomass, density and recruitment indices) were correlated 13 Symmetric W3 379.9 with the estimated trend, in the case of N. norvegicus, two 14 Symmetric NAO 406.9 contrasting scenarios emerged from the results: an 15 Symmetric Days 369.7 16 Symmetric – 372.1 increase in terms of LPUE and recruitment indices, and a decrease in terms of density and biomass indices. The lowest AIC value is highlighted in bold.

30 Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH Ligas, Sartor & Colloca Trends in P. congirostris and N. norvegicus for N. norvegicus, indicating strong relationships involved in linking upper layers and benthic habitats have (Table 5). However, a large diagonal element of the error been made (Company et al. 2008; Cartes et al. 2009). covariance matrix (R > 0.74) was obtained for N. norvegi- Many other environmental and oceanographic variables, cus, confirming that these variables did not fit well to the such as primary production, chlorophyll and nutrient model. concentrations, salinity, upwelling, currents, and river dis- charge, have been shown to be influential in the life cycles and dynamics of marine ecosystems (Bahamo´n & Cru- Discussion zado 2003; Erzini 2005; Erzini et al. 2005; Rothschild The present study aimed to understand the change over et al. 2005; Company et al. 2008; Cury et al. 2008; Cartes time of two demersal stocks in relation to environmental et al. 2009; Sarda` et al. 2009). In the Tyrrhenian Sea, and anthropogenic factors, using analysis of a relatively oceanographic parameters are poorly and irregularly con- long and complete time series of stock abundance data. sidered. The value of using SST, wind circulation and the In the Atlantic Ocean and North Sea, data have been col- NAO index was the availability of an extensive and com- lected since at least the 1950s (e.g. for many fish stocks in plete dataset covering the time span of the available fish- the North Atlantic; Rijnsdorp et al. 2006). In contrast, in eries data. This enabled a thorough investigation to be the Mediterranean Sea, time series of fisheries data usually made into their influence over the course of 15 years. only cover the last few decades. Therefore, the availability Among the environmental variables used, only SST and of 15 years of fisheries data from two of the most abun- the wind-mixing index (W3) were clearly related to the dant decapods of the demersal communities of Mediterra- trend showed by P. longirostris and N. norvegicus, whereas nean waters, both important target species (Aguzzi et al. the NAO index was not significantly associated with 2004; Sobrino et al. 2005; Guijarro et al. 2009; Morello either. Parapenaeus longirostris is considered to be a spe- et al. 2009), should be regarded as an unique opportu- cies with a preference for warm waters, being more abun- nity. dant in the Southeastern Mediterranean than in the The results clearly showed an increasing trend in the Northwestern basin (Abello` et al. 2002). The current abundance of the stock of P. longirostris and a more het- warming of the upper and intermediate water layers of erogeneous scenario for N. norvegicus in the Tyrrhenian the Western Mediterranean (Vargas-Ya´nez et al. 2009), Sea during the investigated period. In addition, our reflected in the observed increase in SST and decrease in results strongly suggest that temporal variations in the wind circulation (W3), could have had a positive effect abundance of the two species were correlated with both on the life cycle and abundance of this species in the Tyr- environmental and fishing activity variables. rhenian Sea. A possible explanation for this phenomenon is provided by Cartes et al. (2009). They hypothesised an association between high temperatures, low rainfall Environmental effects regimes and river discharges and a reduction in the flux Atmospheric and surface water environmental variables of organic matter that maintain deep-water benthic com- were investigated in this study. Although these variables munities off the Catalonian coasts. These environmental have been used in several studies which highlighted signi- conditions resulted in a higher abundance of zooplankton ficant correlations with demersal and deep-sea communi- and increased production of suprabenthos (Cartes et al. ties (Lloret et al. 2001; Farin˜a & Gonza´lez Herraiz 2003; 2009). Although P. longirostris displays a wide range of Zuur et al. 2003a,b; Zuur & Pierce 2004; Erzini 2005; prey items, its diet is mainly based on suprabenthic crus- Maynou 2008; Cartes et al. 2009; Gonza´lez Herraiz et al. taceans, such as mysids (especially Lophogaster typicus) 2009), very few attempts to explain the mechanisms (Sobrino et al. 2005). Therefore, the warming phase observed in recent years could have favoured the P. longi- rostris population in the Tyrrhenian Sea. In addition to Table 5. Estimated t-values for regressions between the response this, Bartolino et al. (2008) found a positive correlation variables (LPUE time series of Parapenaeus longirostris and Nephrops norvegicus) and the explanatory variables (W3 = wind-mixing index; between the recruitment of the European hake, Merluccius NAO = North Atlantic Oscillation index). merluccius, and wind circulation in the Tyrrhenian Sea: high recruitment rates were associated with strong water t-value and wind circulation. The recruitment of the two species, W3 NAO M. merluccius and P. longirostris, takes place in the same area, at a bottom depth of 100–150 m (Colloca et al. P. longirostris 1.26 1.23 2004). European hake juveniles are known to prey on N. norvegicus )4.82 )3.11 crustaceans, such as the juveniles of P. longirostris (Carp- Significant values are highlighted in bold. entieri et al. 2005b). The environmental conditions which

Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH 31 Trends in P. congirostris and N. norvegicus Ligas, Sartor & Colloca positively affect P. longirostris, such as high temperatures graphic evaluation by bottom trawl sampling. Commer- and low wind circulation, are the same as those that neg- cial fishing often operates on a 24-h basis, whereas atively affect the recruitment of M. merluccius (Bartolino experimental trawl surveys are usually carried out in the et al. 2008). The resulting lower predation pressure could daytime. Trawl fleets exploiting the Tyrrhenian Sea often have further enhanced the recruitment success of the perform 2–3 days of fishing operations, carrying out hauls deep-water rose shrimp. during both the day and night (Sbrana et al. 2003). In contrast, N. norvegicus, which showed a negative Although diurnal versus nocturnal bias in sampling has trend in terms of density and biomass indices, was nega- been found to be moderate when trawl catches are per- tively correlated with the NAO and the wind-mixing formed on fishing grounds on the continental slope (Agu- index. Again this may be related to mechanisms linking zzi & Bahamo´n 2009; Bahamo´ n et al. 2009), the fact that the productivity in the upper layers with the structure of N. norvegicus is a predominantly nocturnal species (Agu- demersal communities, as proposed by Cartes et al. zzi & Sarda´ 2008) may help explain the differences (2009). While phases of warmer and dryer atmospheric observed between commercial and experimental survey conditions favour planktonic ⁄ suprabenthic feeders, ben- data. Nephrops norvegicus spend most of the time within thic feeders and predators such as N. norvegicus (Aguzzi or at the entrance of their burrows and are caught by et al. 2009) are disadvantaged by the reduction of the trawling only when they emerge. Emergence varies with organic matter flux resulting from the decreased rainfall time of day, season, animal size, food presence, hunger and river discharge. Similar responses to atmospheric state, sex and reproductive status. Thus the fisheries warming processes by N. norvegicus have been observed exploit the population selectively and in a different man- in other areas. In the context of only minor changes in ner with respect to males and females (Aguzzi et al. 2003; fishing pressure, Farin˜a & Gonza´lez Herraiz (2003) and Aguzzi & Sarda´ 2008; Aguzzi & Bahamo´n 2009). In par- Gonza´lez Herraiz et al. (2009) showed a decline in the ticular, egg-bearing females spend most of their time in population abundance of N. norvegicus in the Atlantic. their burrows during the entire egg-incubation period, This decline was associated with the positive phase of the which lasts for 4–6 months in the Mediterranean Sea (Ag- NAO index, which determines warmer temperatures in uzzi et al. 2003). Furthermore, juveniles rarely leave their Northern Europe (Halliday & Pinhorn 2009). burrows. These factors, related to the biology of the spe- cies, therefore, strongly contribute to protection of the N. norvegicus life-stages that are perceived as sensitive to Fishing activity effects trawling exploitation. The fishing effort is a complex variable that is difficult to Parapenaeus longirostris also shows variations in density quantify because it is influenced by many different fac- and depth distribution according to daytime rhythms and tors, such as increasing catch efficiency and changes in photoperiod length. Carpentieri et al. (2005a) observed fleet characteristics. Increasing catch efficiency of the fleet higher catch rates of P. longirostris at night in the shelf- (also known as ‘technological creep’) is usually related break in the Tyrrhenian Sea. In addition, they found the positively to an increase in skipper skills, investments in highest density index during late winter–spring, which auxiliary equipment, more efficient gear and materials, corresponds to the spawning peak (Ardizzone et al. replacement of old vessels with new ones and, to a lesser 1990). As most larvae occur around the 100-m isobath, extent, upgraded engines (Rijnsdorp et al. 2006). During adults could displace during the spawning period to shal- the investigated period, a decrease in the number of ves- lower depths (Sobrino et al. 2005). sels occurred: the fleets of Porto Santo Stefano and adja- It is worth highlighting that the trend of P. longirostris cent ports decreased by about 50%, producing an almost was characterized by huge interannual fluctuations. Apart proportional decrease in fishing effort. from environmental conditions and fishing activity, this In the case of N. norvegicus the data showed two con- interannual variability was probably related to the short trasting trends: an increase of landings per unit of effort life-span and fast growth rates of this species (Abello` and recruitment index, and a decrease of relative popula- et al. 2002). A similar pattern, characterized by a biomass tion abundance. The daily activity of N. norvegicus could peak in 2001, was observed in other areas of the Western help explain these divergent trends. Light intensity influ- Mediterranean, such as in the Balearic sub-basin (Guijar- ences how organisms perceive their environment, modu- ro et al. 2009). lating their inter- and intra-specific interactions. Demersal communities exposed to light intensity variations are Conclusions expected to react to them, producing changes in species composition and density. Therefore, the diurnal activity Long-term changes in the abundance of two important cycles of demersal species may consistently bias demo- demersal species in the Tyrrhenian Sea, the deep-water

32 Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH Ligas, Sartor & Colloca Trends in P. congirostris and N. norvegicus rose shrimp (Parapenaeus longirostris) and the Norway cean Nephrops norvegicus on the activity rhythms of continen- lobster (Nephrops norvegicus), were found to correlate sig- tal margin prey decapods. Marine Ecology, 30, 366–375. nificantly with identified environmental and anthropo- Aguzzi J., Sarda´ F. (2008) A history of recent advancements on genic factors. While the increasing abundance of Nephrops norvegicus behavioral and physiological rhythms. P. longirostris was correlated to a rise of sea surface tem- Reviews in Fish Biology and Fisheries, 18, 235–248. perature, a corresponding decrease of wind circulation Aguzzi J., Sarda` F., Abello` P., Company J.B., Rotlant G. (2003) and to the reduction of fishing effort, a corresponding Diel and seasonal patterns of Nephrops norvegicus (Deca- trend for N. norvegicus was not evident. On one hand, poda: Nephropidae) catchability in the western Mediterra- the population abundance of N. norvegicus was negatively nean. Marine Ecology Progress Series, 258, 201–211. Aguzzi J., Sarda` F., Allue´ R. (2004) Seasonal dynamics in correlated with environmental variations, while on the Nephrops norvegicus (Decapoda: Nephropidae) catches off other hand, it did not show any association with the gen- the Catalan coast (Western Mediterranean). Fisheries eral decrease of fishing effort in the area. However, the Research, 69, 293–300. recruitment index, as calculated in the study, could be Ardizzone G.D., Gravina M.F., Belluscio A., Schintu P. (1990) used as a proxy for change in stock abundance. Depth-size distribution pattern of Parapenaeus longirostris Some mechanisms have been proposed to link atmo- (Lucas, 1846) (Decapoda) in the central Mediterranean Sea. spheric conditions (sea surface temperature, wind circula- Journal of Crustacean Biology, 10, 139–147. tion and the NAO) to the trophic webs and community Artale V., Astraldi M., Buffoni G., Gasparini G.P. (1994) Sea- structure in the deep-water benthic habitats. However, sonal variability of gyre-scale circulation in the northern these models need to be improved to achieve a deeper Tyrrhenian Sea. Journal of Geophysical Research C, 99, and more accurate understanding of the mechanisms 14127–14137. linking these ecosystems. Further analyses are required to Bahamo´n N., Cruzado A. (2003) Modelling nitrogen fluxes in better understand the relationships between variations in oligotrophic environments: NW Mediterranean and NE the abundance of demersal species and environmental Atlantic. Ecological Modelling, 163, 223–244. and anthropogenic factors. The availability of suitable Bahamo´n N., Sarda´ F., Aguzzi J. (2009) Fuzzy diel patterns in information on environmental characteristics, such as catchability of deep-water species on the continental margin. appropriate sea-floor topography, sediment composition, ICES Journal of Marine Science, 66, 2211–2218. hydrographical characteristics, and trophic webs (prey Bartolino V., Colloca F., Sartor P., Ardizzone G.D. (2008) availability, presence and abundance of predator species) Modelling recruitment dynamics of hake Merluccius merluc- is necessary to better understand the temporal change in cius, in the central Mediterranean in relation to key environ- species abundance, distribution and biology. mental variables. Fisheries Research, 93, 277–288. Baum J.K., Myers R.A., Kehler D.G., Worm B., Harley S.J., Doherty P.A. (2003) Collapse and conservation of shark Acknowledgements populations in the Northwest Atlantic. Science, 299, 389– 392. The authors would like to thank the Editor and the Ref- Bertrand J.A., Gil de Sola L., Papaconstantinou C., Relini G., erees for useful comments and revisions that significantly Souplet A. (2002) The general specifications of the MEDITS improved the structure and the quality of this paper. We survey. Scientia Marina, 66, 9–17. acknowledge also Dr Jessica Craig, University of Aber- Carpentieri P., Colloca F., Ardizzone G.D. (2005a) Day–night deen, Anna Davies, Andrew Silva and Letizia Simeni for variations in the demersal nekton assemblage on the Medi- proof-reading and English revision. terranean shelf-break. Estuarine and Coastal Shelf Science, 63, 577–588. Carpentieri P., Colloca F., Cardinale M., Belluscio A., Ardizz- References one G.D. (2005b) Feeding habits of European hake (Merluc- Abello` P., Abella A., Adamidou A., Jukic-Peladic S., Maiorano cius merluccius) in the central Mediterranean Sea. Fishery P., Spedicato M.T. (2002) Geographical patterns in abundance Bulletin, 103, 411–416. and population structure of Nephrops norvegicus and Parape- Cartes J.E., Elizalde M., Sorbe J.C. (2001) Contrasting life his- naeus longirostris (Crustacea: Decapoda) along the European tories, secondary production, and trophic structure of Perac- Mediterranean coasts. Scientia Marina, 66, 125–141. arid assemblages of the bathyal suprabenthos from the Bay Aguzzi J., Bahamo´n N. (2009) Modeled day–night biases in of Biscay (NE Atlantic) and the Catalan Sea (NW Mediter- decapod assessment by bottom trawling survey. Fisheries ranean). Deep-Sea Research I, 48, 2209–2232. Research, 100, 274–280. Cartes J.E., Maynou F., Fanelli E., Papiol V., Lloris D. (2009) Aguzzi J., Bahamo´n N., Marotta L. (2009) The influence of light Long-term changes in the composition and diversity of availability and predatory behaviour of the decapod crusta- deep-slope megabenthos and trophic webs off Catalonia

Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH 33 Trends in P. congirostris and N. norvegicus Ligas, Sartor & Colloca

(western Mediterranean): are trends related to climatic oscil- groundfish populations in the early 1990s. Fisheries Research, lations? Progress in Oceanography, 82, 32–46. 97, 163–182. Ciannelli L., Fauchald P., Chan K.S., Agostini V.N., Dingsør Jennings S., Dimore T.A., Duplisea D.D., Warr K.J., Lancas- G.E. (2008) Spatial fisheries ecology: recent progress and ter J.E. (2001) Trawling disturbance can modify benthic future prospects. Journal of Marine Systems, 71, 223–236. production processes. Journal of Animal Ecology, 70, 459– CIESM (2008) Climate warming and related changes in Medi- 475. terranean marine biota. CIESM Workshop Monographs, 35, Kaiser M.J., De Groot S.J. (2000) Effects of Fishing on Non- 152. Target Species and Habitats: Biological, Conservation, Socio- Colloca F., Carpentieri P., Balestri E., Ardizzone G.D. (2004) A Economic Issues. Blackwell Science, Oxford: 399 pp. critical habitat for Mediterranean fish resources: shelf-break Kaiser M.J., Ramsay K., Richardson C.A., Spence F.E., Brand areas with Leptometra phalangium (Echinodermata: Crinoi- A.R. (2000) Chronic fishing disturbance has changed shelf dea). Marine Biology, 145, 1129–1142. sea benthic community structure. Journal of Animal Ecololo- Company J.B., Puig P., Sarda` F., Palanques A., Latasa M., gy, 69, 494–503. Scharek R. (2008) Climate influence on deep sea popula- Lloret J., Lleonart J., Sole´ I., Fromentin J.M. (2001) Fluctua- tions. PLoS ONE, 3, e1431. doi:10.1371/journal.pone. tions of landings and environmental conditions in the 0001431 north-western Mediterranean Sea. Fisheries Oceanography, Cury P.M., Shin Y.J., Planque B., Durant J.M., Fromentin 10, 33–50. J.M., Kramer-Schadt S., Stenseth N.C., Travers M., Grimm Maynou F. (2008) Environmental causes of the fluctuations of V. (2008) Ecosystem oceanography for global change in fish- red shrimp (Aristeus antennatus) landings in the Catalan eries. Trends in Ecology and Evolution, 23, 338–346. Sea. Journal of Marine Systems, 71, 294–302. Erzini K. (2005) Trends in NE Atlantic landings (southern Molinero J.C., Casini M., Buecher E. (2008) The influence of Portugal): identifying the relative importance of fisheries the Atlantic and regional climate variability on the long- and environmental variables. Fisheries Oceanography, 14, term changes in gelatinous carnivore populations in the 195–209. northwestern Mediterranean. Limnology and Oceanography, Erzini K., Inejih C.A.O., Stobberup K.A. (2005) An application 53, 1456–1467. of two techniques for the analysis of short, multivariate Morato T., Watson R., Pitcher T.J., Pauly D. (2006) Fishing non-stationary time-series of Mauritanian trawl survey data. down the deep. Fish and Fisheries, 7, 24–34. ICES Journal of Marine Science, 62, 353–359. Morello E.B., Antolini B., Gramitto M.E., Atkinson R.J.A., Farin˜a A.C., Gonza´lez Herraiz I. (2003) Trends in catch-per- Froglia C. (2009) The fishery for Nephrops norvegicus unit-effort, stock biomass and recruitment in the North and (Linnaeus, 1758) in the central Adriatic Sea (Italy): preli- Northwest Iberian Atlantic Nephrops stocks. Fisheries minary observations comparing bottom trawl and baited Research, 65, 351–360. creels. Fisheries Research, 95, 325–331. Fromentin J.M., Fonteneau A. (2001) Fishing effects and life Mori M., Sbrana M., De Ranieri S. (2000) Reproductive biol- history traits: a case study comparing tropical versus tem- ogy of female Parapenaeus longirostris (Crustacea, Decapoda, perate tunas. Fisheries Research, 53, 133–150. Penaeidae) of the north Tyrrhenian Sea (western Mediterra- Gasparini G.P., Ortona A., Budillon G., Astraldi M., Sansone nean). Atti della Societa` Toscana di Scienze. Naturali Memo- E. (2005) The effect of the Eastern Mediterranean Transient rie Serie B, 107, 1–6. on the hydrographic characteristics in the Strait of Sicily Nezlin N.P., Lacroix G., Kostianoy A.G., Djenidi S. (2004) and in the Tyrrhenian Sea. Deep-Sea Research Part I, 52, Remotely sensed seasonal dynamic of phytoplankton in the 915–935. Ligurian Sea in 1997–1999. Journal of Geophysical Research, Gislason H., Sinclair M., Sainsbury K., O’Boyle R. (2000) Sym- 109, C07013, doi:10.1029/2000JC000628. posium overview: incorporating ecosystem objectives within Orsi Relini L., Zamboni A., Fiorentino F., Massi D. fisheries management. ICES Journal of Marine Science, 57, (1998) Reproductive patterns in Norway lobster Nephrops 468–475. norvegicus (L.) (Crustacea Decapoda Nephropidae) of Gonza´lez Herraiz I., Torres M.A., Farin˜a A.C., Freire J., Cance- different Mediterranean areas. Scientia Marina, 62, lo J.R. (2009) The NAO index and the long-term variability 25–41. of Nephrops norvegicus population and fishery off West of Pauly D. (2009) Beyond duplicity and ignorance in global Ireland. Fisheries Research, 98, 1–7. fisheries. Scientia Marina, 73, 215–224. Guijarro B., Massutı´ E., Moranta J., Cartes J.E. (2009) Short Reid P.C., Borges M., Svendsen E. (2001) A regime shift in the spatio-temporal variations in the population dynamics and North Sea circa 1988 linked to changes in the North Sea biology of the deep-water rose shrimp Parapenaeus longiros- horse mackerel fishery. Fisheries Research, 50, 163–171. tris (Decapoda: Crustacea) in the western Mediterranean. Relini G. (1998) Demersal trawl surveys in Italian seas: a short Scientia Marina, 73, 183–197. review. IFREMER, Actes de Colloques, 26, 46–75. Halliday R.G., Pinhorn A.T. (2009) The roles of fishing and Rijnsdorp A.D., Daan N., Dekker W. (2006) Partial fishing environmental change in the decline of Northwest Atlantic mortality per fishing trip: a useful indicator of effective

34 Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH Ligas, Sartor & Colloca Trends in P. congirostris and N. norvegicus

fishing effort in mixed demersal fisheries. ICES Journal of Parapenaeus longirostris, in European Atlantic and Mediter- Marine Science, 63, 556–566. ranean waters (Decapoda, Dendrobranchiata, Penaeidae). Rothschild B.J., Chen C., Lough R.G. (2005) Managing fish Crustaceana, 78, 1153–1184. stocks under climate uncertainty. ICES Journal of Marine Solow A.R. (1994) Detecting changes in the composition of a Science, 62, 1531–1541. multispecies community. Biometrics, 50, 556–565. Sarda` F., Company J.B., Bahamo´n N., Rotllant G., Flexas M.M., Vargas-Ya´nez M., Moya F., Tel E., Garcı´a-Martı´nez M.C., Sa´nchez J.D., Zu´ n˜iga D., Coenjaerts J., Orellana D., Jorda` G., Guerber E., Bourgeon M. (2009) Warming and salting in Puigdefa´bregas J., Sa´nchez-Vidal A., Calafat A., Martı´n D., the western Mediterranean during the second half of the Espino M. (2009) Relationship between environment and the 20th century: inconsistencies, unknowns and the effect of occurrence of the deep-water rose shrimp Aristeus antennatus data processing. Scientia Marina, 73, 7–28. (Risso, 1816) in the Blanes submarine canyon (NW Mediter- Zuur A.F., Pierce G.J. (2004) Common trends in Northeast ranean). Progress in Oceanography, 82, 227–238. Atlantic squid time series. Journal of Sea Research, 52, Sbrana M., Sartor P., Belcari P. (2003) Analysis of the factors 57–72. affecting catch rates of crustacean trawl fishery of the north- Zuur A.F., Fryer R., Jolliffe I., Dekker R., Beukema J. (2003a) ern Tyrrhenian Sea (western Mediterranean). Fisheries Estimating common trends in multivariate time series using Research, 65, 271–284. dynamic factor analysis. Environmetrics, 14, 665–685. Sbrana M., Viva C., Belcari P. (2006) Observation on the fish- Zuur A.F., Tuck I.D., Bailey N. (2003b) Dynamic factor analy- ery of the deep water rose shrimp Parapenaeus longirostris sis to estimate common trends in fisheries time series. Cana- (Lucas, 1846) (Crustacea: Decapoda) in the northern Tyr- dian Journal of Fisheries and Aquatic Sciences, 60, 542–552. rhenian Sea (western Mediterranean). Hydrobiologia, 557, Zuur A.F., Ieno E.N., Smith G.M. (2007) Analysing Ecological 135–144. Data. Springer, London: 680 pp. Smith C.J., Rumohr H., Karakassis I., Papadopoulou K.N. Zuur A.F., Ieno E.N., Elphick C.S. (2009) A protocol for data (2003) Analysing the impact of bottom trawls on sedimen- exploration to avoid common statistical problems. Methods tary seabeds with sediment profile imagery. Journal of Exper- in Ecology and Evolution, 1, 3–14. doi: 10.1111/j.2041- imental Marine Biology and Ecology, 285–286, 479–496. 210X.2009.00001.x Sobrino I., Silva C., Sbrana M., Kapiris K. (2005) A review of the biology and fisheries of the deep water rose shrimp,

Marine Ecology 32 (Suppl. 1) (2011) 25–35 ª 2011 Blackwell Verlag GmbH 35