Using the Long-Term Bottom-Trawl Survey of the Southern Gulf of St. Lawrence to Understand Marine Fish Populations and Community Change Hugues Benoît 1, Doug Swain1, and Ghislain Chouinard  2

1 Gulf Fisheries Centre, Box 5030, Moncton, NB, E1C 9B6 Canadian Science Advisory Secretariat, 200 Kent St., Ottawa, ON, K1A 0E6 [email protected]

2

Sommaire Un relevé au chalut de fond est entrepris chaque année au mois de septembre dans le sud du golfe du Saint-Laurent (sGSL). Ce relevé fournit d’importantes données sur l’abondance, la distribution et la répartition des tailles d’un grand nombre d’espèces de poissons et d’invertébrés. De plus, un échantillonnage océanographique synoptique à grande échelle est entrepris lors du relevé. Dans cet article nous résumons quelques aspects de ce relevé. En particulier nous évaluons à quel point il peut fournir un portrait fiable (exacte et précis) des dynamiques de la communauté de poissons du sGSL. Nous discutons ensuite de l’utilité des données récoltées lors du relevé. Pour ce faire, nous présentons en premier lieu une bibliographie qui a récemment été complétée et qui résume les diverses publications produites depuis 1971 à partir des données du relevé du sGSL. Cette bibliographie met en valeur la grande diversité des travaux scientifiques réalisés. Ensuite nous nous concentrons plus précisément sur deux avenues de recherche, à titre d’exemple : les espèces menacées d’extinction et les études sur la dynamique des communautés de poissons marins.

Introduction Bottom-trawl research vessel (RV) surveys constitute some of the longest, ongoing, annual biomonitoring on Canada’s Atlantic coast. They provide data on the abundance, distribution, and demographic composition (size and sometimes age) of fish populations. Increasingly, data for macroinvertebrates are also being collected. These surveys are an important platform for broad-scale synoptic oceanographic data collection. The history of RV surveys in Atlantic Canada was reviewed in a previous issue of the AZMP bulletin (Chadwick et al. 2007). In the present article, we focus on the RV survey of the southern Gulf of St. Lawrence (sGSL; Fig. 1). First, we begin by providing some brief background information about this survey and the data that are recorded from bottom-trawl sam-

49

48

The sGSL September RV Survey

Laure n chena tian Chan ne l Laur entien l /

200

The RV survey of the sGSL has been conducted annually each September since 1971 and is one of the few bottomtrawl surveys in Atlantic Canada with such a long uninterrupted history. With the addition of three small inshore strata in 1984, the survey covers nearly the entire sGSL with the exception of very shallow waters (80% in total abundance since 1971 (Benoît et al. 2003, Swain et al. 2006). This led COSEWIC to designate it as endangered, i.e., 24

The collapse and lack of recovery of cod in the sGSL and NW Atlantic in general is well known. Lack of recovery appears to result mainly from elevated natural mortality, but also from fishing mortality and generally lackluster recruitment (Shelton et al. 2006). Estimating natural mortality (M; mortality from all sources except accounted fishery removals) is only possible with a standardized fishery-independent survey A

4

Total mortality / Mortalité totale

Species at risk

facing imminent extinction. Using annual survey numbers at length and prior information on size at maturity and growth rate, Swain et al. (2009) developed a stage-structured model for the population. Model fits to the survey data suggest that population viability has been most affected by the mortality of adults, which increased considerably from the mid 1970s to present (Fig. 5A). Fishery-induced mortality stemming mainly from incidental capture in commercial fisheries directed at other species appears to contribute only a little to adult mortality and actually declined as total mortality increased (Benoît 2006, Swain et al. 2009). The cause of non-fishery-related mortality in adults is not well known, though as we discuss later, there is evidence for an effect of predation by top predators (Benoît and Swain 2008, Swain et al. 2009). Impairment of recruitment did not appear to be a contributing factor to the population status because as adult abundance declined during the 1980s and 1990s, the rate of young skate recruitment was high (Fig. 5A, left panel; Swain et al. 2006, 2009).

Juveniles / juvéniles

3 2

0.8

0.4

1

Adults / adultes

0.0

0 1970s

1980s

1990+

1970s

1980s

1990+

B Natural Mortality / Mortalité naturelle

an important component, of the assessment of ten exploited sGSL groundfish species (e.g., Atlantic cod, American plaice). Data collected are also routinely used in the assessments of pelagic species such as herring, capelin (Mallotus villosus), and shortfin squid (Illex illecebrosus). Given the nature of the forums in which stock status and assessment reports were produced in the past, it is very likely that the 283 reports identified are only a subset of the actual number written. The literature search also uncovered an additional 32 reports and 15 peer-reviewed journal articles that document scientific studies undertaken in support of stock assessments. These include studies on stock identification, assessment methodology, and the estimation of vital rates such as recruitment and mortality.

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1970

1975

1980

1985

1990

1995

2000

2005

Fig. 5 A) Boxplots of estimates (by decade) of total mortality in juvenile and adult winter skate (see Swain et al. 2009). B) Estimates of natural mortality (M; mean ± 2 SE) in adult cod in 5-year blocks, indicated by the horizontal lines (see Chouinard et al. 2006).

A) Tracé en rectangle et moustaches des estimations de la mortalité totale par décennie chez les juvéniles et les adultes de raie tachetée (voir Swain et al. 2009). B) Estimations de la mortalité naturelle (M; moyenne et deux erreurs-types) par périodes de cinq ans chez les adultes de morue (voir Chouinard et al. 2006). Les périodes de cinq ans sont représentées par les lignes horizontales.

(e.g., Sinclair 2001) or a good tagging program, which is difficult to achieve when fisheries and therefore potential tag returns are small. For the period of the 1993–1997 moratorium, almost all of the mortality in sGSL cod estimated from the survey was M. With this independent estimate and another one for an earlier period, it has been possible to estimate in the assessment model fit to the survey data the trends in adult cod M for other blocks of years (Chouinard et al. 2005, 2008). The result is an increasing trend since the early 1980s (Fig. 5B). In fact, recent estimates of M suggest that the sGSL cod population is no longer viable; if its current low productivity persists, the population is projected to become extinct in 40 years or fewer, depending on fishery quotas (Swain and Chouinard 2008). As with winter skate, the causes of elevated M are uncertain, though the strongest evidence again points to an effect of top predators (Chouinard et al. 2005; D. Swain, unpublished analyses). Community dynamics The direct effect of fishing (removal of individuals) has been one of the most prominent causes of population change for exploited sGSL fish species. However, interactions with other species and changes in the environment also need to be considered, particularly to understand why many of these species have failed to recover despite strong reductions in fishing (e.g., cod, white hake, American plaice). Furthermore, there is growing interest worldwide to better understand the effects of fishing on non-target species, which can be direct or indirect (i.e., via competitive or predatory release). RV surveys are an integral component of such studies, providing standardized time series of environmental conditions and of the abundance and distribution of a large number of species. Furthermore, these surveys are an important platform for collecting stomach content data to characterize trophic interactions between species. We again turn to examples from the sGSL. Recently, Benoît and Swain (2008) undertook an analysis of the dynamics of the sGSL marine fish community. Using RV survey time series of abundance spanning 1971–2005 for 52 species, they examined the relative influence on community dynamics of fishing, predation by a top predator (grey seals, Halichoerus grypus), changes in bottom-water temperature, and bottom-up effects of changes in prey abundance. A traitsbased approach was used, relating similarity among species in their abundance trends to similarities in their ecological traits. Four traits were selected based on a priori beliefs of how each should reflect susceptibility to changes in one of the different external factors potentially affecting the community. Benoît and Swain (2008) interpreted comparable abundance responses among species sharing a trait as an effect of the external factor implied by the trait. The statistical method used allowed for testing both the marginal (i.e., trait by trait) and conditional (i.e., after controlling for the effect of other traits) significance of individual traits. Benoît and Swain (2008) described dramatic shifts in the species composition of the marine fish community of the sGSL from 1971 to 2005. Some species declined sharply in abundance in the late 1970s and early 1980s, remaining at

low levels since then (e.g., American plaice, winter skate); others increased in abundance in the late 1970s or late 1980s but suffered rapid population collapses in the 1990s (e.g., Atlantic cod, redfish); yet others increased dramatically in abundance in the mid or late 1990s (e.g., capelin). Across species, these trends were independently related to three traits. With few exceptions, species that increased in abundance over the study period were characterized by one or more of the following: little susceptibility to fishing, little susceptibility to seal predation, or a large portion of their biogeographic range occurring in arctic/subarctic waters (i.e., indicative of possible cold-water tolerance) (Fig. 6). Conversely, those species showing an overall decline in abundance were almost all susceptible to both fishing and seal predation, and tended to have a more southerly biogeographic distribution. Species susceptible to fishing included those that were targeted by commercial fisheries as well as incidentally captured ones. Average species’ diet, the fourth trait examined and one that was chosen to reflect possible bottom-up effects of changes in prey abundance, was not related to abundance trends. The apparent trait-dependency of population dynamics reflects trends in the relevant external drivers (Benoît and Swain 2008). Mean bottom-water temperatures cooled throughout the 1980s, resulting in a period of recordlow temperatures in the early to mid 1990s (Gilbert and Pettigrew 1997, Drinkwater and Gilbert 2004). The abundance of cooler-water species, particularly arctic species, peaked during or shortly after the period of coldest water temperatures, declining somewhat subsequently as temperatures warmed to more average recent levels (Benoît and Swain 2008). Fishing effort directed towards demersal fish was elevated during much of 1970s and 1980s, increasing rapidly during the late 1980s and early 1990s, followed by a dramatic reduction that has persisted. Much of the decline in abundance of the species most impacted by fisheries occurred during the period of increasingly high fishing effort. The lack of recovery or continued decline of many of these species despite large subsequent reductions in fishing effort appears to correspond with dramatic increases in the abundance of grey seals in the ecosystem (Hammill 2005). Bioenergetic modelling confirms that seals may have indeed largely replaced fisheries in terms of fish removals (Savenkoff et al. 2008). Trends in both cod and winter skate M are consistent with an effect of seals (Chouinard et al. 2005, Swain et al. 2009). Analyses of RV survey data also suggest that there have been important indirect (i.e., trophic-mediated) effects of fishing on the sGSL fish community. All small-bodied (mean length 75%

Little change / Increase/Augmentation Peu de changement 15 - 75%

15 - 75%

Increase/Augmentation

> 75%

E P G L

Legend / Légende

E: Susceptibility to exploitation /

P: Susceptibility to seal predation /

Susceptibilité à l’exploitation

G: Biogeographic group

Susceptibilité à la prédation par les phoques

High / Élevée

Above average / Au-dessus la moyenne

Intermed. / Interméd.

Below average / Sous la moyenne

Low / Faible

None / Aucune

None / Aucune

(% range arctic/subarctic) / Groupe biogéographique (% arctique/subarctique)

L: Length class /

Classe de longueur

45+ cm

> 50 %

30-44 cm

1% to 50%

15-29 cm

0