FISHERIES OCEANOGRAPHY

Fish. Oceanogr. 9, 99±113, 2000

Modelling the transport of lobster (Homarus americanus) larvae and postlarvae in the Gulf of Maine

LEWIS S. INCZE1,* AND CHRISTOPHER E. NAIMIE2 1

Bigelow Laboratory for Ocean Sciences, West Boothbay Harbor, ME 04575, USA 2 Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA

ABSTRACT We used a coupled physical±biological model to examine potential distances between hatching and settlement locations for lobsters in the Gulf of Maine. The physical model is based on a ®nite-element mesh and climatological averages of the seasonally evolving temperature and density ®elds. Larval trajectories from coastal and offshore hatching sites (21±224 m deep) were calculated for early, middle and late-season hatching by coupling temperature-dependent development rates and depth (the biological model) to the circulation. Model results showed large spatial differences in larval development times (from 18 to 38 days) and distances transported (19±280 km) for the early hatch. Development time and transport decreased markedly by mid-season at most sites, but strong spatial differences persisted. The eastern Maine coast appears to experience stronger removal and less resupply of larvae than other regions, consistent with observed lower recruitment. Inverse solutions of the model for larvae arriving in mid-coastal Maine indicate that they originate from a broad section of the eastern coast `upstream', with those nearest the shoreline generally travelling the shortest distances. The postlarval stage is neustonic (living near the surface), and a simple inverse model demonstrates that a diurnal coastal sea breeze can contribute substantially to inshore movement during this ®nal planktonic stage. Thus, offshore reproduction may be linked to inshore recruitment. Key words: Gulf of Maine, Homarus americanus, larvae, lobster, numerical models, postlarvae, sea 1 breeze, transport *Correspondence. e-mail: [email protected] Received 7 March 1999 Accepted 3 May 1999 Ó 2000 Blackwell Science Ltd.

INTRODUCTION Most marine organisms produce planktonic larvae that require from several days to several weeks to develop 2 (Mileikovsky, 1971). Consequently, propagules may be transported considerable distances away from spawning or hatching locations (Scheltema, 1986). Local populations may be maintained in several ways. In some cases, planktonic stages may be retained through physical processes and biological behaviours, despite their potential for dispersal (Pennington and Emlet, 1986; Boehlert and Mundy, 1988; Sammarco and Andrews, 1988). A population also could be maintained by contranatant migrations of later stages, a pattern which may exist in concert with the use of specialized ``downstream'' nursery habitats for settlement and early 3 development (Harden-Jones, 1968; Bailey et al., 1997; Booth, 1997). Finally, local populations may be sustained by larval supply from the outside, which may be of mixed origin (Roughgarden and Iwasa, 1986; Botsford et al., 1994). All of these patterns are affected by transport processes and rates which can vary over time. As a result, the spatial and temporal patterns of distribution and production in populations with longlived propagules are shaped both by the general characteristics of circulation within a region and by its variability. Both must be understood in order to gain a better fundamental understanding of marine populations and to better manage human interactions with them, including patterns and intensity of harvests. Our goal is to understand the major spatial relationships between reproduction and recruitment in lobsters (Homarus americanus) around the Gulf of Maine and Georges Bank (Fig. 1). The Gulf is a semienclosed marginal sea with several deep basins, strong tidal currents and a generally cyclonic circulation. Scotian Shelf water enters along the south coast of Nova Scotia and exits primarily along the northern edge of Georges Bank and secondarily through the Great South Channel (Brooks, 1985). There is a seasonally intensi®ed westward ¯ow along the Maine coast (Maine Coastal Current), especially east of 69°W Longitude (Brooks and Townsend, 1989; Bisagni et al., 1996b). Anticyclonic ¯ow exists around Georges Bank, with a strong jet along the northern 99

100

L. S. Incze and C. E. Naimie Figure 1. The Gulf of Maine and Georges Bank study area in the northwest Atlantic Ocean.

edge during all seasons and recirculation around the crest during warm months of the year (Butman et al., 1987; Limeburner and Beardsley, 1996). Other quasipermanent features of the circulation include a predominantly anticyclonic ¯ow around Browns Bank (Smith, 1989) and cyclonic ¯ow around Jordan Basin (Fig. 1; Brooks and Townsend, 1989). Complex patterns of lateral exchange of surface waters are common in the region between Browns Bank, Georges Bank and Jordan Basin (Bisagni et al., 1996a; Bisagni and Smith, 1998). In addition to gradients in transport, there are strong variations in sea surface temperature in the region, with the coldest spring and summer temperatures typically found along the south coast of Nova Scotia and the eastern coast of Maine. Lobsters are found throughout most of the Gulf of Maine and Georges Bank (Pezzack, 1992), where they are an important part of the benthic community (Ojeda and Dearborn, 1991) and support an extremely lucrative ®shery (Miller, 1995). Coastal regions inside the 50 m isobath are the most productive in terms of ®shery landings (Skud and Perkins, 1969; Krouse, 1981; Pezzack et al., 1992). Although some remarkable migrations have been documented (Campbell and Stasko, 1986), the coastal population of lobsters as a whole may be relatively nonmigratory (Krouse, 1981, reviewed by Lawton and Lavalli, 1995) and subjected to a high exploitation rate which may reach 90% or more of the individuals newly recruited (moulted) to legal size (Fogarty, 1995). Despite the intense ®shery,

the population has ¯ourished in recent years (Drinkwater et al., 1996). `Offshore' lobsters, which have only recently been exposed to heavy exploitation (Fogarty, 1995), have been considered by some to constitute an important reproductive source for the inshore population (Lawton and Lavalli, 1995). Larval transport is one possible mechanism for linking the two populations. Fogarty (1997) calculated that a modest amount of larval input from offshore could add resiliency to a heavily ®shed inshore stock. Source±sink relationships are not easily demonstrated, however (Stasko and Gordon, 1983; Harding and Trites, 1988, 1989; Pezzack, 1989), and most planktonic studies have focused on smaller geographical areas (Harding et al., 1987) or other aspects of recruitment (Incze et al., 1997, 2000). Inshore and offshore populations also might be linked by seasonal migrations. In some locations, offshore lobsters are known to undertake seasonal migrations to shallower waters during warmer months. The extent of these migrations is variable within and among areas where it has been documented: Georges Bank, Browns Bank and around Grand Manan Island in the south-western Bay of Fundy (Cooper and Uzmann, 1971; Uzmann et al., 1977; Campbell and Pezzack, 1986; Pezzack, 1992). For most of the Gulf of Maine, the size structure, seasonal movements and proportion of the population undertaking these migrations remain poorly known. Eggs are extruded in summer and autumn and remain attached to pleopods on the female for an Ó 2000 Blackwell Science Ltd., Fish. Oceanogr., 9, 99±113.

Larval lobster transport

incubation period of 9±10 months (Talbot and Hulley, 1995). Hatching occurs over a period of two months or more (Ennis, 1995) beginning in early to late June in the central and southern coastal Gulf of Maine (Fogarty, 1983 and references therein) and about a month later in southern Nova Scotia and Browns Bank (Stasko and Gordon, 1983; Harding and Trites, 1988; Tremblay and Sharp, 1989). Peak hatching may occur about a month after the initial hatch (Sherman and Lewis, 1967; Campbell, 1986; Harding and Trites, 1988), although this probably is variable. Hatching time can vary between years (Ennis, 1995) owing to temperature effects on moulting, mating (Waddy and Aiken, 1992) and embryonic development (Perkins, 1972). Geographical patterns may be complicated further by factors such as migrations of egg-bearing females (Campbell, 1986) and local environmental effects. Indeed, hatching occurs over a wide range of temperatures even within geographical subareas (Ennis, 1995). Larvae develop through three larval stages which are thought to reside mostly in the upper 2±3 m of the water column, and a subsequent postlarval stage which is neustonic, mostly in the upper 0.5±0.8 m (Scarratt, 1973; Harding et al., 1982; Stasko and Gordon, 1983; Hudon et al., 1986). Harding et al. (1987) reported much deeper larval distributions at a 60 m site on Browns Bank. The postlarva is the ®nal pelagic stage (Cobb et al., 1989). Development rates of all stages are positively correlated with temperature over the range 10±22°C (MacKenzie, 1988). In a seven-year study along the central coast of Maine, Incze et al. (1997) found that most postlarvae collected nearshore were in middle or later stages of development, which the authors ascertained from moult-cycle stages. They concluded that many of these postlarvae were coming from offshore, probably under the in¯uence of prevailing onshore breezes during summer months (Wahle and Incze, 1997), but the distances from which they were transported were not known. Postlarvae have been reported from samples taken more than 50 km offshore (Stasko and Gordon, 1983; Locke and Corey, 1988, Incze et al., in press; Harding et al., unpublished data) and from Browns Bank (Stasko and Gordon, 1983; Harding et al., 1987; Watson and Miller, 1991) and Georges Bank (Watson and Miller, 1991; Harding et al., unpublished data). Thus, the ®nal planktonic stage appears to be present over large areas of the offshore environment; the extent to which they settle there, are lost to the population, or ultimately are advected toward shore is unknown. The densest known settlement along the Maine coast occurs in cobble habitat shallower than 10 m below mean low water Ó 2000 Blackwell Science Ltd., Fish. Oceanogr., 9, 99±113.

101

4 (Wahle and Steneck, 1992; Incze et al., 1997; Wahle and Incze, 1997; Wilson, 1999). Small lobsters also are found on Georges Bank, however (200 m using seasonally evolving solutions from a prognostic 3-D ®nite-element circulation 5 model (Lynch et al., 1996, 1997) coupled with temperature-dependent development of larvae and postlarvae. We evaluate the spatial and temporal scales of larval transport predicted by the model for various points around the Gulf of Maine and Georges Bank, and demonstrate the potential importance of the diurnal sea breeze in promoting onshore transport of the postlarval stage. METHODS Our study area encompasses the Gulf of Maine from Cape Sable, Nova Scotia to Cape Cod, Massachusetts, including Browns Bank and Georges Bank, east to the continental slope and south to Nantucket Shoals (Fig. 1). Female lobster data were obtained from a research trawl survey database at the National Marine Fisheries Service (NMFS) Laboratory, Woods Hole, Massachusetts, and covered fall surveys from 1987 to 1997. Surveys were conducted from north to south between mid September and early November. The database includes federal US and Canadian waters, mostly where water depths exceeded 40 m, as well as state surveys within Massachusetts waters. We used female maturity ogives for the Gulf of Maine and Georges Bank provided by NMFS and plotted the locations of all stations where female lobsters met or exceeded the size (carapace length) where 50% of individuals are sexually mature. We used these distributions to indicate the potential distribution of hatching offshore. Possible differences in female lobster distributions between autumn (surveys) and spring (hatching) are discussed later in this paper. We computed the predicted trajectories of larval and postlarval lobsters in the Gulf of Maine using a coupled physical±biological, individual-based model (IBM). The physical environment was approximated by interpolation between the 3-dimensional 6 bimonthly (i.e. two-monthly) circulation ®elds presented by Naimie (1995, 1996) and the associated temperature ®elds from Lynch et al. (1997), as computed using the Dartmouth Circulation Model

102

L. S. Incze and C. E. Naimie

Table 1. Temperature-dependent rates of larval and postlarval development in lobsters. All data and the larval equations are from MacKenzie (1988); the postlarval equation is from Incze et al. (1997) based on a subset of Mac Kenzie's data relevant to the temperature range in this study of the Gulf of Maine. D = duration in days, T = temperature in °C. Stage I Stage II Stage III Postlarva

D D D D

= = = =

851 ´ (T ) 0.84))1.91 200 ´ (T ) 4.88))1.47 252 ´ (T ) 5.30))1.45 703.5 ´ T)1.26

(a prognostic, 3-dimensional, ®nite-element circulation model: Lynch et al., 1996, 1997). The horizontal discretization of the ®nite-element grid is adjusted for greater resolution in regions of steep bathymetric change and irregular shoreline topography such that the node spacing within 50 km of shore and over the offshore banks is »2±5 km. Over the deep basins the spacing is 10±15 km. The biological component of our model consists of temperature-dependent development rates based on polynomial equations of MacKenzie (1988) for larval stages and the equation given by Incze et al. (1997) for postlarvae (Table 1). `Numerical lobster' trajectories were computed within the hydrodynamic environment using the IBM presented by Blanton (1995) and described by Werner et al. (1993, 1996). A number of other studies have used a similar modelling strategy within these circulation ®elds to address other questions (Hannah et al., 1998; Naimie et al., 1999; Page et al., in press).

We used tidal and residual transport calculations from 5 m depth to avoid a bias in the near-surface trajectories (Ekman layer) which is imposed by the unidirectional nature of the climate-averaged seasonal winds (0.0136 Pa toward 49.4°True during May±June, 0.0138 Pa toward 51.0°T during July±August, and 0.0186 Pa toward 145.6°T during September±October). We used temperatures from 1 m for larval and postlarval development, although there is negligible temperature change in the upper few metres of the model. We selected 10 transects for larval releases (Fig. 2). Each transect consisted of 10 release points beginning from 5 to 27 km from shore in 21±61 m of water. Deep ends of the transects were located 52± 116 km offshore in 150±220 m (Table 2). We used three sets of geographically adjusted dates of hatching to simulate transport and development from early, middle and late hatching times at each location. For transects 1±6, dates of hatching were 10 June, 10 July and 10 August, respectively. The corresponding dates for transect 7 were delayed 15 days, i.e. to the 25th of each month. An additional delay of 15 days was applied to transects 8±10 so that hatching occurred 10 July, 10 August and 10 September. In addition to the forward runs of the model (above), we conducted inverse runs for a region in the mid-coast of Maine for which we have seven years of data on the timing and abundance of postlarvae (Incze et al., 1997). Our region of interest was the 54 km section of coast between Penobscot Bay and Casco Bay (Fig. 1) out to the 100 m isobath, »20 km offshore. To represent some Figure 2. Transects used for larval releases are numbered 1±10. Each transect has 10 release points denoted by asterisks. Shaded regions correspond to depth intervals (m) shown in the scale at right and are the same throughout this paper.

Ó 2000 Blackwell Science Ltd., Fish. Oceanogr., 9, 99±113.

Larval lobster transport

103

Table 2. Physical characteristics of the transects used in this study. Temperatures are shown in Fig. 6. Transect 1

2

Depth (m) at ends of transects Inshore 39 33 Offshore 220 213

3

4

5

6

7

8

9

10

45 199

61 158

21 154

24 175

43 194

40 200

28 188

30 144

15 59

5 59

14 52

14 54

15 104

27 116

18 113

Distance (km) of release points from the coast Inshore NAa NA 27 Offshore NA NA 69 a

NA, not applicable.

of the range of possible sources for larvae entering this region, we started the inverse calculations at various points around its perimeter. For these calculations, we `started' at the end of Stage III larval development (just before the moult to postlarva), and back-calculated the larval trajectories to their predicted hatching locations according to the inverse solution of the coupled circulation/development model. We also conducted inverse calculations for postlarvae in the same mid-coast region of Maine. Instead of establishing `end-points' on the outer perimeter of the region as we did for larvae, we used the original transect 6 (in the centre of the larval box used above) and an along-isobath transect near shore (20 m isobath). We did this because of our speci®c interest in the source of postlarvae arriving at coastal recruitment habitats. Tidal and residual ¯ows at 5 m depth were used, as with the larval stages, but we superimposed a sea breeze on the tidal residual output using a transfer ef®ciency of 3% at 45 degrees to the right of the wind. Characteristics of the sea breeze came from Simpson (1994), local observations (Wahle and Incze, 1997) and data from NOAA Weather Buoy 44007 (10 km offshore of southern Casco Bay, Fig. 1). We present results for a `weak' sea breeze, which we de®ned as blowing onshore at 5 m s±1 for 3 h day±1 with 1 h spin-up and decay. The offshore extent of the sea breeze was speci®ed by the 60 m or 100 m isobaths for convenience, these depths corresponding closely to our target distances of 10 and 20 km from land (see Simpson, 1994). We applied the sea breeze each day during postlarval development, and we started the inverse solution with postlarvae near the end of their development. Data and observations indicate that a nocturnal land breeze is not a regular, signi®cant feature for this location and time of year, and none was built into our calculations. Ó 2000 Blackwell Science Ltd., Fish. Oceanogr., 9, 99±113.

We also conducted a full inverse simulation by: (1) `initializing' the model with arrival locations and dates for late moult-cycle postlarvae in mid-coast Maine; (2) modelling the postlarval stage as above; and (3) modelling the larval stages as above. Results from both the larval and the full (larval + postlarval) inverse runs suggested that surface temperatures in the model might be too cold (see later details). This motivated us to compare the near-surface temperature predicted by the model with an AVHRR climatology compiled by D. 8 Ullman (Ullman and Cornillon, 1998 and unpublished data). We conducted additional simulations with AVHRR corrections to the background temperature ®eld. We also compared temperature predictions of the model with 10-year averages (1989±1998) of the daily sea-surface temperatures recorded by US NOAA buoys at two locations. One buoy was in the central Gulf of Maine (NOAA Buoy 44005, located at 42.9°N, 68.94°W, sensor depth at 1 m) and the other was moored in 19 m of water »10 km off Cape Elizabeth in the south-western Gulf (43.53°N, 70.14°W, sensor depth at 0.6 m). RESULTS Sexually mature female lobsters were found by Autumn trawl surveys at virtually all depths in the Gulf of Maine, although positive tows were relatively infrequent in the central Gulf. Mature females were relatively common in samples out to the 100 m isobath along all of the western Gulf of Maine and to 200 m in the northern Gulf off Nova Scotia (Fig. 3). The surveys did not cover Browns Bank, where a signi®cant population of reproductive females also exists, particularly on the western side of the bank (Campbell and Pezzack, 1986). Autumn survey results were used because catch ef®ciency for lobsters in the spring is

104

L. S. Incze and C. E. Naimie Figure 3. Locations (open symbols) where sexually mature female lobsters were found by bottom trawl surveys in the Gulf of Maine and Georges Bank, 1987±1997. Except for Browns Bank and Nantucket Shoals, which are not included, the Gulf is sampled extensively (stations are not shown if they did not yield female lobsters meeting the size criteria). Depths and shading as in Fig. 2.

low throughout the region (J. Idoine, NMFS, Woods Hole, MA, pers. comm.). The climatologically averaged model circulation at 5 m depth shows a residual cyclonic circulation around the Gulf of Maine; the coastal current shows strongest residual velocities of 10±15 cm s±1 south of Nova Scotia and along the eastern Maine coast. The average ¯ows weaken slightly from May±June to July± August and then increase slightly in the fall (Fig. 4a±c). Vertical shear in transport down to 10 m does not produce large differences in transport length scales or directions. Well-developed anticyclonic ¯ows exist around Browns Bank and Georges Bank and become stronger from spring to summer. Strong horizontal temperature gradients exist throughout the period (Fig. 4d±f). Vertical temperature differences were always