<<

Web Interactions and Modeling

Introduction changes in the productive capacity of The Chesapeake Bay has the environment, the of experienced significant historical planktonic and benthic , and the alterations due to overfishing, structure of food webs that support anthropogenic stress, and natural fisheries. disturbances. Although conventional Researchers have a good understand- single- management 2 approaches do not typically address ing of some food web relationships in predator–prey dynamics, these the Chesapeake Bay. Atlantic menha- dynamics form the heart of den Brevoortia tyrannus, Atlantic interactions among species affecting croaker Micropogonias undulatus, bay abundance and production. Such anchovy Anchoa mitchilli, blue interactions have dramatic and Callinectes sapidus, and spot substantial effects on Leiostomus xanthurus are integral structure, ultimately affecting fisheries links between and within benthic and yields in the Bay, and must be planktonic components of the Bay considered when developing or food web. Heavily exploited, predatory amending -based fishery consume forage species, such as management plans (FMPs). menhaden and bay anchovy; these predators may also rely on juvenile mortality—an important blue as part of their diet and may fraction of total mortality for most ultimately affect the abundance of exploited species—represents recruiting crabs. on fishery resources. Multispecies We must expand our understanding of incorporates not only fishing mortal- food web interactions, quantify their ity information but also key predator– effects, develop new food web models, prey linkages and their contributions and implement existing models to to natural mortality. Understanding provide the requisite information that such food web dynamics allows quanti- will permit managers to define sus- fication of the and tainable catch levels and estimate transfers in the food web that dictate fishing mortality rates of species in the sustainable levels of fishery exploita- webs. In this fisheries ecosystem plan tion. Food web relationships are not (FEP), we have included preliminary independent of and water diagrammed food webs of managed quality issues; they may vary with species, indicating strong and weak element

Food Web Interactions and Modeling 103 interactions between predator and Food Web Dynamics prey. These webs can guide managers as they explore policy options to Importance develop ecosystem-based regula- Sustainable use of exploited species tions—allowing high yields of pis- will depend, at least in part, upon civorous fish, for example, while inclusion of multispecies fisheries conserving resources and management approaches based largely important predator–prey relation- on food web dynamics (Christensen ships. Managers can now use funda- 1996; Daan 1997; Christensen and mental knowledge of food web Pauly 1998; Pauly et al. 1998). Manag- structure and relationships in a ers have not yet applied a multispecies precautionary manner, but major approach to fisheries management in research is needed to ensure effective the Chesapeake, despite its potential utility (Houde et al. 1998) as well as multispecies fisheries management in the availability of a food web model the Bay. for the mesohaline (middle) portion of This FEP element addresses the the Bay (Baird and Ulanowicz 1989), importance and limitations of which could provide a framework for developing food web models for the additional modeling focused on man- Chesapeake, considers the degree of agement needs. connectivity between particular Fishing affects by remov- species (or trophic groups) and their ing biomass from the complex of predators and prey, and describes species that feed upon each other in subwebs of the Bay’s economically the web (Pauly et al. 2000). It also valuable species. In addition, the shifts the relative abundance of element describes and discusses the exploited species at different trophic utility of several recognized levels. Such changes—from fishing, multispecies and ecosystem models other anthropogenic stresses (e.g., that managers could adopt for habitat alteration and ), or ecosystem-based fisheries environmental change—may lead to management in the Bay. shifts in the and sustain- able yields of species. These shifts, in turn, may affect the value of fisheries, species , or the structural integrity of the ecosystem (Winemiller and Polis 1996; Pauly et A food web is defined as a “network of - al. 2000; Jackson et al. 2001; Link interactions among a group of organisms, 2002a). For instance, researchers have populations, or aggregate trophic units” (Winemiller postulated that changes in the abun- and Polis 1996). dance of key fishery species, such as the and blue crab, may have altered community structure and pathways of production in the Bay (Jackson et al. 2001; Silliman and

104 Fisheries Ecosystem Planning for Chesapeake Bay Bertness 2002). Massive fishery- management. induced reductions in the abundance of eastern oyster Crassostrea virginica, Predation is key in determining the a suspension feeder on phytoplank- abundance and size structure of populations, as well as the ton, have contributed to abnormally organization and functioning of high production, in the Chesapeake and , and seasonal other ecosystems (Lipcius and Hines that reduce secondary production and 1986; Hines et al. 1990; Seitz et al. (Jackson et al. 2001). 2001). Predation affects all stages In coastal salt marshes, declines in of marine organisms and constitutes blue crab abundance (Lipcius and the primary source of natural Stockhausen 2002), due partly to mortality for fish in well-studied heavy fishing pressure, may have marine ecosystems (Bax 1991, 1998), allowed marsh periwinkle Littoraria even for those species with high irrorata to become more abundant fishing mortality during their and overconsume grasses— exploitable life stages. a process which ultimately could lead to the destruction of salt marshes The relative importance of predation important for blue crab production and fishing mortality varies among (Silliman and Bertness 2002). species, but is typically skewed towards predation for younger (and smaller) An ecosystem’s , individuals and towards fishing production potential, and total mortality for older individuals. For sustainable yield to fisheries cannot instance, predation largely accounts for simply be calculated as the sums of the mortality of young juvenile blue yields for individual component crabs whereas fishing becomes species (Link 2002a) using traditional, responsible for 80% of the mortality of single-species stock assessment older juveniles and adults. Predation techniques. Rather, fisheries may also a major role in production of an ecosystem depends controlling food web dynamics in significantly on food web dynamics marine ecosystems, altering the effects (Pauly et al. 2000; Link 2002a). To of reductions or increases in fishing evaluate the impact of a species’ mortality of species (Andersen and fishing mortality upon food web Ursin 1977; Laevastu and Favorite interactions and ecosystem processes, 1988; Bax 1991, 1998; Christensen therefore, the ecosystem’s chief food 1996; Trites et al. 1999; Pauly et al. web interactions must be defined and 2000; Link 2002a). quantified (Pauly et al. 2000). Similarly, researchers must consider With a heavily fished population at the effects of other controlling low abundance, predation may limit factors, such as habitat quality and population recovery despite potentially environmental conditions (see high of incoming year Habitat Requirements and classes (Sissenwine 1984; Bax 1991, Externalities elements), within the 1998; Christensen 1996; Link 2002a). context of ecosystem-based In such cases, the predator may have

Food Web Interactions and Modeling 105 remained at high population levels or or by blue crab predation; either of it may be a fished species that has these forces could prevent oyster recovered after management-induced recovery given that overfishing, reductions in fishing mortality. For habitat degradation, and disease have example, Lipcius and Stockhausen driven the population to extremely (2002) hypothesized that predation low levels (Rothschild et al. 1994). pressure by Atlantic croaker (at high abundance in Chesapeake Bay for For some species, natural mortality nearly a decade) or by striped bass through predation—especially on Morone saxatilis (which dramatically young stages—may prove more signifi- resurged during the last decade cant in controlling population abun- following rigorous management dance than fishing mortality on measures) may be responsible for the recruited stages. Such species may be lack of recovery of the depressed blue subject to little or no fishing pressure, crab population in the Bay. Similarly, but serve as forage species for a spec- restoration of the Bay’s native oyster trum of natural predators (Overholtz population may be hampered by et al. 2000). Historically, watermen disease in older juveniles and adults have fished some forage species (e.g. Atlantic menhaden), which form a major component of the Chesapeake a fisheries ecosystem. During the past 50 years, when overfishing has caused Fishery Top Predator declines of top predator species, fisheries have increasingly targeted species at lower trophic levels (Pauly et al. 1998). Significant reductions in Target Species the abundance of these species may cause fundamental changes in commu- nity structure and the ecosystem b (Pauly et al. 1998; Jackson et al. 2001). This process, referred to as Fishery Top Predator “fishing down food webs,” may disrupt natural predator–prey relationships. The effects of such fishing include Intermediate shifts in trophic-level structure, along Predator with changes in population abundance and age structure of target species. Selective fishing on forage species can Target Species precipitate indirect impacts on other species in the food web as predators transfer their emphasis to alternative Figure 1. Hypothetical simple (a) and moderately complex (b) prey. food webs with top and intermediate predators that are not fished, a target fishery species, and human fishing at a equivalent to that of the top predator (adapted from Yodzis In some cases, predation may not play 2001). a major role in controlling abundance,

106 Fisheries Ecosystem Planning for Chesapeake Bay as in shoaling pelagic species indirect effects confounds the ability (Overholtz et al. 1991, 2000; Jennings to determine either the direction of and Kaiser 1998), which fluctuate in the system response to the removal of response to variable ocean conditions. the top predator or its magnitude Such variability in controlling mecha- (Yodzis 2001). In this scenario, the nisms accentuates the need to define reduced top predator population will the major food web interactions in an eat less of the target species, which ecosystem before we can understand the relative impacts of natural and fishing mortality upon food web Stability is not limited to pristine systems; it is also a dynamics and ecosystem processes. feature of disturbed systems (Sheffer et al. 2001; Carpenter 2002) and contributes to the difficulty in Limitations restoring disturbed ecosystems such as Chesapeake Despite the growing awareness that Bay. fisheries management requires a multispecies approach, considerable debate remains over the reliability of should result in an increase in its predictions of changes in target abundance. The top predator will also species abundance derived from eat fewer of the intermediate preda- multispecies approaches. Yodzis (2001) tors, however, which should decrease provides an illuminating example of target species abundance. In this the problems associated with predic- circumstance, the net result of reduc- tions of fishery-induced alterations in ing the top predator upon the target food webs, examining a situation in fishery species shown in this relatively which fisheries cull a top predator to simple food web remains uncertain. increase production of a target species Ultimately, the abundance of the by reducing the natural predation on target species might increase, decrease, this species. In this case, fisheries or be unaffected, depending on the catch a target species consumed by the strengths of the various predator–prey top predator (Figure 1a). If the cull links (Punt and Butterworth 1995; significantly reduces the population of Abrams et al. 1996; Yodzis 2001). the top predator, then the abundance and yield of the target fishery species Another complication arising from the should increase. complexity of food web dynamics is the possibility that ecosystems have This simple view of food web dynam- alternative stable states (Sheffer et al., ics is based on the assumption that 2001; Carpenter 2002). Given that the removing a top predator from the Chesapeake Bay food web has system will increase the abundance of undergone dramatic, historical prey it would have consumed, which alterations due to anthropogenic then becomes available to the fishery. changes—such as overfishing (Jackson If, however, the addition of an inter- et al. 2001) and eutrophication mediate predator complicates the food (Boesch 2000)—and natural web (Figure 1b), the potential for disturbances (R. N. Lipcius and R. D.

Food Web Interactions and Modeling 107 Seitz, unpublished manuscript), a productive scallop fishery in the restoring the food web to its seaside for over 6 decades (R. “pristine” state may prove N. Lipcius and R. D. Seitz, unpublished impossible. Even if managers agree manuscript). on the preferred food web, its composition, and its biomass Three basic approaches exist for the analysis of food webs (Paine 1966; structure, such a food web may be Winemiller and Polis 1996). One is unattainable due to the stability of topological, providing a static descrip- the degraded ecosystem characterized tion of predator and prey links be- by the distorted food web (Sheffer et tween species or trophic groups. In al. 2001; Carpenter 2002; Peterson the following section, we offer a basic and Lipcius 2003). Stability refers to a topological analysis of the connectivity situation in which a disturbed or of species and trophic groups in the degraded ecosystem is in an Chesapeake Bay food web. A second “” (Sheffer et approach uses quantitative analysis of al. 2001; Carpenter 2002), which will energy and matter flow through the not easily shift back to the food web via predation (e.g., undisturbed state due to feedback with Ecosim, Pauly et al. 2000), a mechanisms maintaining the modeling approach that researchers structure of the disturbed stable have started using in the Bay. The state. Stability is not limited to final approach is functional, identify- pristine systems; it is also a feature of ing the species and trophic links that disturbed systems (Sheffer et al. determine community structure. The 2001; Carpenter 2002) and functional approach typically depends contributes to the difficulty in on field experiments that deal with restoring disturbed ecosystems such specific links between important as Chesapeake Bay. consumers (predators) and resources (prey) in the food web (see Silliman Management, therefore, should and Bertness 2002). consider the possibility that some desired food web configurations may Topological and analyses not be achievable (Peterson and are instructive (Pauly et al. 2000; Link Lipcius 2003), at least in the short 2002b), but not always capable of term, without massive intervention explaining the dynamics of (Carpenter 2002). For instance, the populations and communities. The seaside lagoons of the Eastern Shore dynamic influence of a particular harbored extensive beds species or trophic group is not that supported a lucrative Bay scallop necessarily proportional to the energy fishery until the Storm King flow between trophic links (see review hurricane of 1933 devastated the of the three analysis types in ecosystem. The resultant turbid Winemiller and Polis 1996). For conditions not only precluded instance, may initiate restoration of the seagrass beds, but trophic cascades (i.e., significant also hindered the reestablishment of effects of changes in one species upon

108 Fisheries Ecosystem Planning for Chesapeake Bay 43229.2 22774 14639.2 2.1 4.9 30. 7.5 20. 0.3 25. Bluefish 5. 6. 7. Fish Croaker Free Hetero. Microzoo- 1. larvae 2.5 Phytoplankton Microflag.

3439

7555 579.8 6157 14 50.3 34. 21. 31. 10716.2 26. DOC 7. Alewife/Blue 14 Hogchoker Weakfish Zoo- 154.8 447 22.3 plankton 9. 10. Ctenophore Sea Suspended nettle 2293.1 292.2 2. 105.4 35. Bac- POC teria 21986 97 22. 2 27. 32. 0.9 Bay Spot 251.9 0.9 Summer anchovy 55 flounder

11. 12. 13. Other susp. Mya Oyster feeders 33. 23. 59 78.8 2.4 28. Striped Menhaden 24062.9 19.8 White bass perch 23528 3. 137437 36. Bac- 15. 16. POC 14. teria Other Nereis Macoma polychaetes spp.

107.5 22.6 5013 286.9 24. 29. 113.8 43.1 293 Shad 30209 48.1 4. 19. Benthic 17. 18. 352 Blue crab Meiofauna Crust. dep. feeders 967

137437 4046 23176 1711 23721 219847 30209 194.7 60.3 198 510.5 10.8

Figure 2. Food web components of middle (mesohaline) Chesapeake Bay, indicating composite cycling of carbon. This web is generally representative of the major food web components used in previous network (energy flow) analyses of the Chesapeake food web (Baird and Ulanowicz 1989; Monaco and Ulanowicz 1997; Hagy 2002) (adapted from Baird and Ulanowicz 1989). others in the food web without direct may strongly influence seagrass links), forming a key force in production and community structure structuring marine, aquatic, and through consumption of seagrass terrestrial communities. Yet, their grazers (e.g., amphipods and isopods), influence often appears which increase seagrass productivity by disproportionately high relative to upon seagrass epiphytes (M. their biomass (Power et al. 1996). Harris, E. Duffy, and R. N. Lipcius, unpublished manuscript). The Recent research suggests that the blue influence of these complex crab is a keystone species in the mechanisms on community structure Chesapeake. First, the blue crab in Bay indicate that an enhances salt marsh grass production experimental, functional approach to and the associated marsh community food web analysis is needed to identify by feeding upon marsh periwinkles, the major controlling factors of food which at high densities can reduce salt web dynamics. Unfortunately, the marsh productivity (Silliman and experimental field manipulations Bertness 2002). Second, the blue crab typically required to evaluate the

Food Web Interactions and Modeling 109 dynamics and functional roles of , particularly in the middle species in a food web often prove Bay (Monaco and Ulanowicz 1997; logistically intractable at spatial and Hagy 2002). temporal scales that capture the full Although the ratio of primary produc- dynamics of an ecosystem. In such tion in the to that in cases, topological analyses or the is 6:1, production in the modeling may become the only Bay relies heavily on inputs from options. A balanced inclusion of the detritus and the (i.e., three approaches to food web cycles through bacteria analysis may best address the food to protozoan consumers with subse- web dynamics of large ecosystems. quent grazing by microzooplankton) Moreover, collective uncertainties in to fuel secondary production at higher food web investigations demand trophic levels, including most fishery caution in the application of food species (Monaco and Ulanowicz 1997). web analyses to fisheries Production of in the management. Bay may depend significantly on benthic deposit feeders, detritus, and Chesapeake Bay Food Web the microbial loop, in addition to pelagic (Baird and General Features Ulanowicz 1989; Monaco and The Chesapeake Bay food web (Figure Ulanowicz 1997; Hagy 2002), as in 2) contains several features typical of many marine ecosystems characterized most estuarine food webs: by high bacterial biomass (Pomeroy 1) Predominance of generalist feed- 2001). Benthic suspension feeders use ers, both benthic and pelagic, that phytoplankton production, typically consume prey in propor- allochthonous inputs (e.g., external tion to their availability (Baird and sources from freshwater Ulanowicz 1989; Monaco and inflows), and benthic production Ulanowicz 1997; Hagy 2002); (Monaco and Ulanowicz 1997). Conse- 2) Moderately interconnected trophic quently, benthic suspension feeders pathways between predators and and deposit feeders form critical prey (Monaco and Ulanowicz conduits between the pelagic and 1997; Dunne et al. 2002); benthic components of the food web, since deposit and suspension feeders 3) Modified food web structure due (such as worms and clams) are eventu- largely to anthropogenic alter- ally consumed by predatory demersal ations (Monaco and Ulanowicz 1997; Jackson et al. 2001; Hagy, fish and benthic . These 2002); predators subsequently become prey for larger, pelagic, predatory fish. 4) High fisheries production (Nixon 1982); and The species and trophic groups com- 5) High phytoplankton primary prising the Bay’s food web have been production, much of which is not assigned to specific trophic levels consumed and is transformed to (Table 1) and their importance identi-

110 Fisheries Ecosystem Planning for Chesapeake Bay Table 1. Trophic Monaco & Trophic Group Hagy Baird & Ulanowicz levels of Ulanowicz Chesapeake Bay D0OC, POC 10. 10. 1. food web components after P0hytoplankton 10. 10. 1. Hagy (2002), P0icoplankton 10. 10. 1. Baird and Ulanowicz (1989), S0AV 10. 10. 1. and Monaco and Ulanowicz (1997). M0icrophytobenthos 10. 10. 1. Numbers refer to B0acteria 20. 20. 2. the average trophic level of S1uspension feeders 21. 22. 2. each group, standardized to E1astern oyster 21. 22. 2. 1.0 for primary A1tlantic menhaden 28. 27. 2. producers and sources of organic M3eiobenthos 27. 2-. carbon (DOC and POC), 2.0 for R4otifers 22. 2-. primary D4eposit feeders 20. 38. 2. consumers, 3.0 for secondary M-2icrozooplankton 26. 2. consumers, and M5esozooplankton 22. 22. 2. so on.

C6iliates 28. 2-.

H0eteroflagellates 30. 3-.

B1lue crab 35. 32. 3.

S1pot 30. 4-.

A1tlantic croaker 30. 4-.

H1ogchoker 39. 3-.

A1merican 3-. -

C2atfish (Bay species) 30. 4-.

W4hite perch 30. 4-.

B5ay anchovy 38. 27. 2.

L5obate ctenophore 30. 32. 3.

A6lewife/Blueback herring 32. 3-.

S6hads 32. 3-.

S6triped bass 39. 38. 3.

W0eakfish 48. 38. 3.

B1luefish 46. 48. 3.

S6ea nettle 44. 3-.

S-0ummer flounder 48. 3.

Food Web Interactions and Modeling 111 fied through network analysis of standing of ontogenetic diet shifts. energy flow (Baird and Ulanowicz Consequently, the assumed trophic 1989; Monaco and Ulanowicz 1997; position of individual species should Hagy 2002). Baird and Ulanowicz be examined carefully during FMP (1989) and Monaco and Ulanowicz development or in food web modeling (1997) categorized the principal and investigations. consumers and trophic groups in terms of energy flow and cycling in Food Web Generalities for the Bay. the Bay food web. More recently, The following list cites some of the Hagy (2002) extended the earlier generalities that apply to the Chesa- analyses and reached fundamentally peake Bay food web. similar conclusions. Although the 1) Of the pelagic consumers, bay relative importance of a particular anchovy and menhaden transfer species or trophic group as a con- the most production from plankton sumer may differ based on occurrence to predatory fish and have the in the upper, middle, or lower Bay highest secondary fish production. (Hagy 2002), generalities do exist in 2) The lobate ctenophore (e.g., comb the Bay’s food web. The following jelly) is a major consumer of conclusions are summarized from the mesozooplankton (larger zooplank- extensive investigations of Baird and ton such as and cladocer- Ulanowicz (1989), Monaco and ans) and microzooplankton (smaller Ulanowicz (1997), and Hagy (2002). such as nauplii and Some species may be categorized ) particularly in the middle poorly, however, due to incomplete Bay. This consumption diverts diet data or an inadequate under- production from forage fish and,

12 Highly connected predators 10

8

6

as Predator 4 Number of Links

2

0

Birds Spot. SAV Ducks Croaker Catfish Oyster. Flatfish Seatrout Soft clam Hard clam Hogchoker Bluefish ad. BluefishBlack juv. drum Sea nettles Forage fish Blue crabBlue ad. crab juv. Weakfish ad. Cownose ray Weakfish juv. AmericanBay anchovy. eel Yellow perch Seatrout juv. Gizzard shadCtenophores Benthic Sandbar shark MenhadenMenhaden ad. juv. Alewife/herring Phytoplankton White perch ad. White perch juv. American shad Striped bass ad. Infauna/epifauna Striped bass juv. MesozooplanktonHorseshoe crab. Reef demersal fish Bad phytoplanktonMicrozooplankton Suspension feeders Trophic Group

Figure 3. Number of links from the species or trophic group listed on the X-axis to prey of that species or trophic group. For example, striped bass adults prey on 11 species or trophic groups; blue crab adults prey on nine species or trophic groups, including blue crab juveniles.

112 Fisheries Ecosystem Planning for Chesapeake Bay 25 Highly connected prey 20

15

as Prey 10 Number of Links 5

0

SAVSpot Oyster Ducks Croaker Catfish Flatfish Turtles Soft clam Forage fish Hard clam Sea nettles Hogchoker Bay anchovy BluefishBluefish ad.Black Seatroujuv. drum ad.t Blue catfish Blue crabBenthic juv. algae Ctenophores Gizzard shad Seatrout juv.YellowDemersal perch Blue fish crab ad. WeakfishWeakfish ad. juv. American eelCownose ray PhytoplanktonMenhaden juv. Menhaden ad. Alewife/herring Sandbar shark White perch juv. Striped bass juv. American shad Horseshoe crab White perch ad. Infauna/epifauna Summer flounder Striped bass ad. MesozooplanktonMicrozooplankton Reef demersal fish Bad phytoplankton Suspension feeders Trophic Group Figure 4. Number of links from the species or trophic group listed on the X-axis to predators of that species or trophic group. For example, 23 species or trophic groups prey on mesoplankton and 11 on bay anchovy. Data are derived from the diet matrix (August 2002) of the Chesapeake Bay Ecopath model.

30 Highly connected trophic groups 25

20

15

10 Number of Links

as Predator and Prey 5

0

Birds Spot SAV Ducks Turtles Oyster Croaker Catfish Flatfish Soft clam Hard clam Black drum Hogchoker Forage fish Bluefish ad. Blue catfishWeakfishSeatrout ad. ad. Bluefish juv. Sea nettles Blue crab Bayjuv. anchovy Blue crab ad. Benthic algae Ctenophores CownoseGizzardDemersal ray shad fish Yellow perch WeakfishSeatroutAmerican juv. juv. eel PhytoplanktonMenhaden juv. Menhaden ad. Alewife/herring Sandbar shark White perch juv. White perch ad. AmericanHorseshoe shad crab Infauna/epifauna Striped bass ad. Summer flounder Striped bass juv. MesozooplanktonMicrozooplankton Reef demersal fish Bad phytoplankton Suspension feeders Trophic Group Figure 5. Number of links from the species or trophic group listed on the X-axis to predators and prey of that species or trophic group. For example, 25 links occur from mesoplankton to species or trophic groups that are either predators or prey of mesoplankton along with 17 links from blue crab juveniles to either predators or prey. Data are derived from the diet matrix (August 2002) of the Chesapeake Bay Ecopath model.

ultimately, fisheries yield. Cteno- medusa Chrysaora quinquecirrha) phores also prey on bay anchovy predation on ctenophores appears eggs and larvae, reducing the to compensate for the negative potential for fish production. In effect of ctenophore predation on network analyses, sea nettle (the bay anchovy. Sea nettles can be

Food Web Interactions and Modeling 113 viewed positively in terms of Connectivity of energy flow to fish production Predators and Prey within the web, therefore, despite Topological analysis is a useful their moderate predation on bay instrument to determine the anchovy. Other forage fish species structure of particular trophic groups that form important links from in the food web, illustrating the the plankton to the benthos and connectivity between a particular predatory fish include the species or trophic group and its alosines—alewife Alosa pseudoharengus, blueback herring predators and prey (Winemiller and A. aestivalis, and American and Polis 1996). The connectivity (i.e., hickory shads A. sapidissima and number of linkages between trophic A. mediocris. groups) of the Bay’s food web is moderate relative to other terrestrial, 3) Benthic suspension feeders and aquatic, and marine food webs (Dunne deposit feeders, particularly et al. 2002; Link 2002b), suggesting bivalves (e.g., Atlantic rangia that it is reasonably resilient to Rangia cuneata, Baltic macoma modest perturbations such as the loss Macoma balthica, northern of a few species (Dunne et al. 2002). quahog Mercenaria mercenaria) and various polychaetes (e.g., Such a loss to the integrity of the food deep-burrowing Chaetopterus web becomes most pronounced when variopedatus), consume much of highly connected species (i.e., species the Bay’s detrital, planktonic, and possessing multiple linkages to other microbial loop output. In the predators and prey) are removed from middle Bay, seasonal hypoxia the web (Dunne et al. 2002) and may causes low benthic production have severe impacts on food web (Hagy 2002). Network analyses, integrity, carbon cycling, and resilience however, indicate that production to environmental perturbations remains sufficient to satisfy the (Dunne et al. 2002). Consequently, demands of demersal (i.e., scientists and fishery managers must epibenthic) and benthic (i.e., recognize those predators and prey infaunal) predators. Demersal fish having the highest degree of (hogchoker Trinectes maculatus, connectivity with other trophic groups spot, Atlantic croaker) and blue in the food web. crab are among the chief consumers of the benthos, To evaluate the degree of connectivity collectively consuming nearly 90% of the various species and trophic of benthic production. The most groups of the Bay food web, the NOAA productive of the Chesapeake Bay Office (NCBO) include weakfish Cynoscion regalis, EcoPath Working Group used the striped bass, bluefish Pomatomus August 2002 diet matrix of the Chesa- saltatrix, channel catfish Ictalurus peake Bay Ecopath model to define furcatus, Atlantic croaker, and the number of links between species spot, in no order of importance. and trophic groups (Figures 3–5). The

114 Fisheries Ecosystem Planning for Chesapeake Bay most highly connected predators (top 12%; 6 of 50) were piscivorous birds (e.g., American ), striped bass adults, Atlantic croaker, blue crab adults, bluefish adults, and blue crab Cownose ray Black seabass juveniles (Figure 3). Each had links to 8 to 13 species or trophic groups upon which they prey. Of the prey, the Croaker Blue crab (ad.) most highly connected trophic groups (top 12%; 6 of 50) were those near the base of the food web, including benthic deposit and suspension Disease Blue crab (juv.) Rapa whelk feeders, grazers, mesozooplankton, microzooplankton, littoral forage fish (e.g., silversides), Softshell clam Oyster Hard clam and bay anchovy (Figure 4). Each of these groups had 10 to 25 links to predators. When considering all trophic links, both to predators and Microzooplankton prey, the most connected trophic groups (top 12%; 6 of 50) were again those near the base of the food web, Phytoplankton specifically infaunal and epifaunal deposit and suspension feeders, Figure 6. Subweb of the hard clam, soft-shell clam, and native invertebrate grazers, oyster. mesozooplankton, microzooplankton, blue crab juveniles, littoral forage fish, aggregation of species as trophic and bay anchovy (Figure 5). These groups serving as their prey. If one groups had 14 to 31 links to predators considers ontogeny and increase in size and prey. at progressive life stages of species such as menhaden and lobate cteno- The results bear strong similarity to phores, then their connectivity may those of energy flow analyses (Figure increase. For example, larvae and early 2), indicating that relatively few juveniles of menhaden primarily species and trophic groups drive consume zooplankton, while cteno- energy flow and connectivity in the phores in the larval stage may have a Chesapeake food web. Some species broader diet than larger individuals. that may be critically important in Such patterns may hold for many food webs were not highly connected species, but the connectivity analyses (e.g., menhaden, ctenophores) due to may not fully account for them. narrow dietary preferences. The low to moderate connectivity of plank- Managers should value all species tonic consumers, such as menhaden categorized as important in either the and ctenophores, may result from the energy flow or the connectivity analy-

Food Web Interactions and Modeling 115 Humans Birds alterations, due in large part to overfishing (Jackson et al. 2001) (see Externalities Element). Most likely, Blue catfish Yellow perch the current major role of detritus in trophic dynamics results from considerable degradation of the Bay Alewife/ food web and the accompanying shift Herring from an ecosystem that functioned largely through benthic algal and seagrass production to one heavily Mesozooplankton dependent on the microbial loop, phytoplankton production, and detritus (Jackson et al. 2001). Microzooplankton Recognizing that the Bay’s food web has endured dramatic change due to Figure 7. Subweb of the alewife/herring complex. anthropogenic stress, we must now accept the possibility that certain components of the food web may not ses. Additionally, the connectivity be easily restored. between certain species might not be clear due to reduced abundance of the Subwebs of Fishery Species species. The most notable example— the native oyster—historically played Subwebs of each of the economically a dominant role as a consumer of valuable species in the Bay can guide phytoplankton. the identification of the species’ important predators and prey. The The Chesapeake Bay food web has subwebs do not portray the relative undergone substantial historical importance of links between predators

Humans Birds

Black seabass, Cobia, Bluefish (ad.), Sandbar Shark Striped Bass (ad.)

Atlantic croaker

Blue crab (juv.) White perch (juv.), Forage fish, Anchovy

Benthos Bivalves Softshell clam Mesozooplankton

Figure 8a. Subweb of the Atlantic croaker.

116 Fisheries Ecosystem Planning for Chesapeake Bay Humans

Striped Bass (ad.) Cobia Black seabass Sandbar Shark

American eel Summer flounder Croaker Red drum Spotted seatrout (ad.) Yellow perch

Blue crab (ad.)

Blue crab (juv.)

Benthos Bivalves Softshell clam Hard clam Oyster

Figure 8b. Subweb of the blue crab.

Humans Birds

Striped Bass (ad.) Cobia Black seabass Sandbar Shark

Summer flounder Bluefish (ad.) Spotted seatrout (ad.)

Reef fish

Spot

Benthos Mesozooplankton

Figure 8c. Subweb of the spot.

Food Web Interactions and Modeling 117 Humans Humans

Blue catfish Birds American eel

American shad Blue crab (juv.)

Mesozooplankton Benthos Bivalves

Microzooplankton

Figure 9a. Subweb of the American shad. Figure 9b. Subweb of the American eel.

Humans

Black Drum

Benthos Hard clam Bivalves Softshell clam

Figure 9c. Subweb of the black drum.

Humans

Bluefish (juv.) Bluefish (ad.)

Forage fish Anchovy Spot Menhaden (juv.) Menhaden (ad.)

Benthos Mesozooplankton

Microzooplankton

Figure 9d. Subweb of the bluefish.

118 Fisheries Ecosystem Planning for Chesapeake Bay Humans

Spotted seatrout (ad.)

Anchovy Spotted Spot Forage fish seatrout (juv.) Menhaden (juv.)

Blue crab (juv.) Mesozooplankton

Microzooplankton

Figure 9e. Subweb of the spotted seatrout.

Humans

Striped Bass (ad.)

Croaker Striped Bass (juv.)

White Anchovy Spot Forage fish Menhaden (ad.) perch (juv.) Menhaden (juv.)

Gizzard shad Blue crab (juv.) Mesozooplankton

Benthos Microzooplankton

Figure 9f. Subweb of the striped bass.

Food Web Interactions and Modeling 119 Humans

Summer Flounder

Spot Anchovy Forage fish Blue crab (juv.) Menhaden (juv.)

Benthos

Figure 9g. Subweb of the summer flounder.

Humans

Weakfish (ad.)

Weakfish (juv.) Anchovy Forage fish Menhaden (juv.) Menhaden (ad.)

Mesozooplankton

Benthos Microzooplankton

Figure 9h. Subweb of the weakfish.

120 Fisheries Ecosystem Planning for Chesapeake Bay Humans

Striped bass (ad.) Blue catfish Birds

Croaker Channel catfish White perch (ad.) Black seabass

Forage fish Anchovy Striped bass (juv.) White perch (juv.)

Benthos Mesozooplankton

Microzooplankton

Figure 9i. Subweb of the white perch.

Humans Birds

Yellow perch

Blue crab (juv.) Alewife/herring

Benthos

Figure 9j. Subweb of the yellow perch.

Food Web Interactions and Modeling 121 Life Cycle Diagram The diet matrix of the NOAA of the Oyster Chesapeake Bay Office EcoPath Working Group provided the information to create the subwebs. The first subweb, shown in Figure 6, consists of benthic suspension-feeding bivalves near the base of the food web Larvae Juveniles Adults (Table 1) that support major fisheries, including the northern quahog, softshell (also known as softshell Figure 10. Simplified life cycle diagram of the Eastern oyster. clam) Mya arenaria, as well as the highlighting its key life stages. eastern oyster. Predator-prey Interactions The next subweb (Figure 7) includes and Sources of Mortality finfish near the base of the food web, specifically the alewife-blueback Fishing herring complex. Following is the suite of subwebs that includes most of Disease Blue Crab Adults the intermediate trophic links, and those with a high degree of connectiv- ity, such as Atlantic croaker, blue crab, Adult and spot (Figures 8a, 8b, and 8c). Finally, we detail the food webs of higher-level predators, including most of the top predatory , specifically American shad, American eel Anguilla Zooplankton Substrate rostrata, black drum Pogonias cromis, Phytoplankton bluefish, spotted seatrout Cynoscion nebulosus, striped bass, summer flounder Paralichthys dentatus, weak- Figure 11. Adult stage of the Eastern oyster with major sources fish, white perch Morone americana, of mortality (disease, fishing, blue crab predation), prey (plankton), and habitat needs (substrate). and yellow perch Perca flavescens (Figures 9a–9j). and prey. Additional factors and processes, external to traditional For all subwebs illustrated above, the predator–prey relationships, may NCBO EcoPath Working Group is prove critical in the dynamics of the modifying the predator and prey species, such as disease in the case of linkages as new information is gath- the native Eastern oyster. These ered and incorporated into the current factors should be considered in the Ecopath model. Updated subwebs may comprehensive analysis of all sources be downloaded at of mortality affecting a species and in http:noaa.chesapeakebay.net/ development of ecosystem-based fepworkshop/netfep.htm. Further management plans. details on the prey and predators are

122 Fisheries Ecosystem Planning for Chesapeake Bay available on the web site. the analysis of the food web and its predator–prey relationships Incorporation of Food Web (Caswell 2000). Dynamics into FMPs 2) Identify the major predator-prey The following list provides general interactions and sources of guidelines for incorporating food web mortality for each life stage by dynamics into FMPS. expansion of the life cycle diagram (Figure 11). 1) Develop and define the life cycle diagram of the target species. 3) Identify critical habitat relation- ships for each life stage. The life cycle diagram explicitly recognizes the critical life stages Expansion of the life cycle and food and sources of mortality. For web diagrams should consider example, Figure 10 diagrams a habitat needs (see Habitat Require- simple life cycle diagram for the ments Element), including possible Eastern oyster. Although simplistic, use of protected or closed areas in elaboration of such diagrams in management and conservation of terms of habitat requirements and exploited species. Figure 12 shows sources of mortality allows one to an example of stage-specific preda- discern if key processes or sources tor-prey interactions with probable of mortality have been ignored in modification in protected areas for

Strong effect Fishery

Weak effect Exploitation

Striped Bass Striped Bass Croaker, Red Drum Croaker, Red Drum

1+ 0+ 0+ 1+ Blue Crabs Blue Crabs Blue Crabs Blue Crabs

Protected SAV Beds

1+ 0+ 0+ 1+ Oysters, Clams Oysters, Clams Oysters, Clams Oysters, Clams

Protected Oyster Reefs

Figure 12. Potential food web interactions in no-take protected (shaded) and open-to- fishing (clear) habitats for a benthic component of the Chesapeake Bay food web that emphasize predator-prey relationships of striped bass/blue crab and other demersal fishes. This example, while hypothetical, indicates the kind of process that fisheries managers should follow in developing multispecies fisheries management plans that account for predator–prey interactions in complex ecosystems such as the Chesapeake.

Food Web Interactions and Modeling 123 modeling. Natural mortality Recruitment This exercise entails a thorough and comprehensive comparison of the findings from food web analyses (i.e., topological, energy flow, functional) Stock 1 with those from modeling efforts, as described below.

Growth Fishing mortality Multispecies Modeling Approaches Predation Bycatch Historically, fisheries management has relied on single-species models that ignored the effects of biological and Growth Recruitment technical interactions on population abundance (Figure 13). Traditional fisheries models focus on the interplay Stock 2 between exploitation level and sustainability, generally not consider- ing in detail the and of Natural Fishing the managed species. In recent years, mortality mortality however, researchers have started to overcome this deficiency by consider-

Figure 13. A conceptual description of two fish stocks (S1 and ing the feasibility of ecosystem-based S ) linked through both biological and technical interactions. A 2 approaches to fisheries management. predator-prey relationship is shown in which S preys upon the 1 One important element of ecosystem- juveniles of S2. Increased predation by S1, therefore, leads to an increase in its growth rate and a decrease in the rate at based management is development

which the juveniles of S2 are captured incidentally by the fishery and incorporation of multispecies that targets recruitment rate for S . Thus, an increase in the 2 models into management programs. fishing will again lead to a decrease in the recruitment rate for S . Adapted from Miller et al. (1996). Like single-species models, 2 multispecies models yield information about sustainability but are structured striped bass/blue crab/bivalve to do so by more accurately reflecting linkages. biological and ecological reality. Although such detailed linkages Over the past several years, the are difficult to evaluate and fully number and types of multispecies understand, they do reflect eco- models that provide insight on fisher- logical reality; therefore, consider- ies issues have grown significantly ing subsets of the food web re- (Hollowed et al. 2000). This growth mains important in developing has been fueled by the need to better and implementing ecosystem- inform fisheries policymakers and based fisheries management. managers; however, recent concerns 4) Compare and contrast analyses of about fishing effects on the structure food web dynamics with food web of ecosystems has also prompted

124 Fisheries Ecosystem Planning for Chesapeake Bay research on multispecies modeling and Theory of implied predator-prey relationships. Multispecies Modeling From a theoretical perspective, basing fisheries stock assessments on multispecies (rather than single- species) models appears more appro- Technical interactions Biological interactions priate, since multispecies approaches allow explicit modeling of more of the MS surplus production processes that govern population Ralston and Polovina (1982) Koslow et al. (1994) abundance. This increased realism, however, requires additional param- MSYPR eters (particularly for models that Murawski (1984) assess the impact of biological interac- tions), which in turn creates the need for more types of data. In the absence of these additional data, or if unreli- MSVPA Size-spectra Sparre (1991) Pope et al. (1988) able data are used to meet the require- Magnusson (1995) Duplisea and Bravington (1999) ments of multispecies stock assess- ments, more uncertainty in manage- ment outcomes will undoubtedly arise Ecosystem models Aggregate system: Baird and Unlanwicz (1989) Christensen and Pauly (1992); Bax (1985, 1991) compared to single-species stock Dynamic system: Walters et al. (1997) assessment methods. Consequently, multispecies models are not replace- Figure 14. General overview of multispecies fisheries models. ments for single-species models, but Adapted from Hollowed et al. (2000). rather that provide additional types of stock assessment insight when used in concert with single- Species 1: Species 2: species models (National Research Prey Predator Council 1999). Age 1 Age 1 In recent years, interest has grown in F multispecies fisheries management in F M1 M1 the Chesapeake region, as evidenced M2 by the development of fisheries Age 2 Age 2 steering groups, the convening of F multispecies technical workshops F M M1 (Miller et al. 1996; Houde et al. 1998), 1 and the requirement for development Age 3 Age 3 and implementation of multispecies fisheries management plans by the Chesapeake 2000 agreement (CBP Figure 15. General schematic of two- 2000). In this section, we describe and species MSVPA with arrows indicating losses due to fishing (F), predation (M ), evaluate some multispecies models 2 and residual (M1) mortality (Jennings et al. commonly applied to understand the 2001).

Food Web Interactions and Modeling 125 Inputs Model Outputs multispecies surplus production models (Figure 14). MSVPA-species

Catch at age in numbers Terminal F s Biological Interactions Residual mortality Predator consumption rates Multispecies Virtual Population Body weight at age Analysis Predator stomach contents Single-species virtual population analysis (VPA), as developed by Fry Fishing mortality rates Other Predators Retrospective stock sizes (1949) and Gulland (1965), is a stock Suitability coefficients Minumum abundances in numbers assessment technique that uses Predation mortality rates Consumption rates commercial-catch-at-age data to Body weight Stomach contents calculate retrospective stock sizes and fishing mortality rates (F) of recruited, age-based cohorts. This approach Other Prey often is referred to as cohort analysis. Minumum abundances in numbers For a given cohort, the number alive Consumption rates in the previous year is calculated by Body weight adding the number caught by the fishery in the current year to the Figure 16. Types of data needed to perform an MSVPA based estimated number that died from on the role each species assumes in the analysis. “MSVPA- natural causes during that same time species” are those predators and prey for which retrospective stock sizes are reconstructed using the MSVPA model and period. Inherent in this technique are “other predators” and “other prey” represent species for which two important features: each cohort is VPA-type results are not desired, but the researchers know or treated separately (i.e., the variables surmise that these species significantly influence the trophic associated with a cohort are calculated dynamics of the food web under study. From Latour et al. 2003. independently of those from other cohorts); and an estimate of the natural mortality rate (M) is required effects of both biological and techni- as input for the model. When M is not cal interactions; these models have known, the traditional approach is to potential for development and estimate it roughly from life history application in the Bay. parameters or to use an educated guess. This overview should inform fisher- ies managers and policymakers about The dependence of VPA on a the capabilities of more commonly reasonable estimate of M has applied multispecies modeling tech- motivated researchers to focus on niques. For biological interactions, we natural mortality estimations. review multispecies virtual popula- Although natural mortality occurs tion analysis (MSVPA), size spectrum from various causes, predation is analysis, and ecosystem models generally believed to be the dominant (Ecopath with Ecosim). For technical source of mortality. This belief, along interactions, we review multispecies with preliminary quantitative work on yield per recruit (MSYPR) and feeding and food consumption of

126 Fisheries Ecosystem Planning for Chesapeake Bay North Sea cod Gadus morhua (Daan “MSVPA-species,” the data require- 1973, 1975), provided the foundation ments include catch-at-age in num- for the development of models that bers, fishing mortality rates in the accounted for species interactions. terminal year and for the oldest age Anderson and Ursin (1977) developed class, residual mortality rates, predator an that gave a consumption rates, body weights at conceptual framework for modeling age, and predator stomach contents. predator-prey interactions. Although this complex model could not adapt to Largest real-world management applications, it ultimately facilitated the extension Fish of VPA to multispecies virtual Body Size population analysis (MSVPA). Zooplankton

Helgason and Gislason (1979) and Phytoplankton Pope (1979) independently combined Smallest the theoretical predation relationships Abundance of the Anderson and Ursin (1977) model with the VPA methodology of Figure 17a. Trophic pyramid relating the abundance of general Gulland (1965) to develop MSVPA. species groups within an . The width of the pyramid is proportional to abundance; the height is proportional The primary feature of the method is to body size (Jennings et al. 2001). With biomass used as a the split of the natural mortality rate metric rather than abundance, the pyramid would be greatly into two components. That is, compressed with little difference among trophic levels (Sheldon et al. 1972; Kerr and Dickie 2001).

M = M1 + M2 (1) in which M2 represents the predation mortality between and within the Phytoplankton exploited species in the ecosystem—as determined by suitability parameters that reflect predator preference for a particular prey species—and M1 Zooplankton represents the mortality due to all factors not explicitly included in the model (Figure 15). For a review of the MSVPA approach, see Sparre (1991), Abundance (log) Magnusson (1995), and Jennings et Fish al. (2001).

The data requirements for an MSVPA Body Size (log weight) vary according to the role each species assumes in the model and the pre- Figure 17b. Plot of normalized biomass (i.e., number density) ferred model output (Figure 16). If as a function of body size. The slope of the line is a qualitative species’ stock sizes are reconstructed representation of the structure of an aquatic ecosystem (Jennings et al. 2001). As with Figure 17a, using biomass using the MSVPA model, with these instead of abundance results in a “flat” slope, although a small species referred to as negative slope sometimes occurs.

Food Web Interactions and Modeling 127 With MSVPA, modeling species as to support its ecological role as a “other predators” or “other prey” is prey species and the goals of the possible in cases for which the stan- directed fishery; and dard VPA results are not desired, but 4) The biological reference points for the researchers know or surmise that menhaden recommended for these species significantly influence management derived from single- the trophic dynamics of the food web species assessments along with under study. For “other predators,” determination of whether data requirements include minimum adjustments are necessary with abundances in numbers, body predation included in the weights, consumption rates, and assessment. stomach contents; for “other prey” only minimum abundances in num- Two major extensions to the base bers and body weights are typically model also are being developed. First, required. a stochastic feeding model is being incorporated into the MSVPA frame- MSVPA in Chesapeake Bay work to account for the effects of changes in menhaden population To date, MSVPA has not been used abundance on the diets and consump- for stock assessments of fish popula- tion patterns of predators. This model tions in Chesapeake Bay, although it will require additional input data on is recognized as a possible approach (Miller et al. 1996; Houde et al. the relative abundance of alternative 1998). An expanded MSVPA model prey species. Second, the growth and for assessment of the Atlantic men- of predators haden stock in the coastal waters of (striped bass, bluefish, and weakfish) the eastern United States is currently will be modeled more explicitly to under development (Garrison and explore and evaluate the effects, if Link 2002) to supplement existing any, of prey quality. This extension single-species assessments of the will incorporate the effects of prey Atlantic menhaden stock and to availability, diet composition, and allow fisheries managers to evaluate feeding rates. possible alternatives to the current In recent years, fishing effort on management scenario. The MSVPA menhaden has shifted from northerly model addresses four major topics waters to southern areas; as a result, through evaluation of the Chesapeake Bay has become a 1) The nature and magnitude of center of menhaden fishing. Although linkages among menhaden and its recent coastwide stock assessments key predators; have characterized the menhaden 2) The current use of menhaden as a stock as healthy, concern exists that directed fishery, its ecological role this characterization does not apply to as a forage fish, and sustainability menhaden in the Bay. Recruitment of of the stock; Bay menhaden has declined since the 3) The possible optimal size (or age) 1990s and young-of-year abundance composition of Atlantic menhaden has been low. Low menhaden abun-

128 Fisheries Ecosystem Planning for Chesapeake Bay dance may cause nutritional stress in plot results that linearly relates log predators, such as striped bass. Recent numbers (or production) to log body studies suggest that striped bass in size (Figure 17b). Based on the the Bay suffer from poor nutrition, aforementioned law of conservation, evidenced by an increase in the num- perturbations to the ecosystem via ber of diseased fish exhibiting lesions removals will cause a change in the from mycobacteriosis. The Maryland line slope. In theory, therefore, it is Department of Natural Resources possible to detect and interpret (MD DNR) Pound Net Survey (2002) changes in the structure of an revealed that 17% of the striped bass exploited ecosystem by comparing had lesions or sores. slopes of biomass or abundance (i.e., normalized biomass) in relation to The MSVPA analysis of menhaden body size. This approach is formally represents one of the first known as size-spectrum analysis. multispecies modeling efforts of a fish species indigenous to Chesapeake Bay Size–spectrum analysis has since been (discussed later is another modeling adapted and applied in ecological effort—the Chesapeake Bay Ecopath studies ranging from characterization with Ecosim model). The MSVPA will of marine benthic invertebrate document quantitatively the assemblages (Schwinghammer 1981, simultaneous effects of predation and 1983; Saiz-Salinas and Ramos 1999) to fishing on the menhaden stock. harvesting strategies and community Although best interpreted on a structure of fish populations (Pope et coastwide scale, these results should al. 1988; Macpherson and Gordoa provide information to better 1996; Duplisea and Bravington 1999; evaluate the role of menhaden as a Kerr and Dickie 2001). Researchers forage fish in the Bay. An MSVPA have also used the results of a size- analysis reflecting Bay-specific input spectrum model (e.g., quadratic data for menhaden, striped bass, regression equations depicting the bluefish, and weakfish would provide major biomass domes in a particular additional insight into management ecosystem [Thiebaux and Dickie 1993]) of these species. to develop estimates of annual production for the taxonomic groups Size-spectrum Analysis represented by these domes. The study A fundamental characteristic of by Sprules and Goyke (1994) aquatic food webs is conservation of represents a specific example of this mass through energy conservation via application, in which the researchers production, respiration, growth, and computed an estimate of annual predation (Jennings et al. 2001). Body production for zooplankton in Lake size determines these processes, Ontario. Sprules et al. (1991) show leading to trophic pyramids with the that examining the complete biomass smallest species at the bottom and size spectrum is possible; they the largest species on top (Figure 17a). described the pelagic biomass size spectrum including phytoplankton, By turning this pyramid on its side, a zooplankton, planktivorous fish, and

Food Web Interactions and Modeling 129 piscivorous fish from nine major Size–spectra in Chesapeake Bay regions of Lake Michigan in both Researchers have not yet used size– spring and summer and also spectra analyses and models to de- estimated the potential annual velop management strategies for production of several trophic groups. Chesapeake fish. Recently, however, Within fisheries, the use of spectral Jung (2002) conducted biomass size– methods has rarely been applied spectrum analyses using midwater when developing management trawl catch data to estimate biomass, strategies. One exception is the production, contribution to predators, recent study by Duplisea and and recruitment numbers (young of Bravington (1999) in which they year) for forage fish (primarily bay used spectra models to explore the anchovy). Jung also included analyses implications of different harvesting of higher trophic-level pelagic and strategies on total yields and commu- benthopelagic fishes, some of which nity stability and persistence in are piscivores (e.g., weakfish) in the marine ecosystems. Their analysis Bay. Jung’s analysis indicated that indicated potential for application of annual, seasonal, and regional differ- the method, but counterintuitive ences in size spectra occur in response fishing strategies emerged for some to changing environmental conditions harvesting questions in a few in- and these environmental conditions stances. For example, one might primarily affected recruitment and expect that the best strategy for young of year biomass production. The maximizing total yield would specify results may prove useful in develop- the harvest of fish at lower trophic ment of biomass spectrum models levels (i.e., remove biomass from the that address fishery management system by fishing before it is lost to issues. predation up the ). Multispecies fisheries management is Duplisea and Bravington (1999) designed, by definition, to incorporate showed, however, that total yield ecosystem processes into management would be maximized if larger fish plans. Knowledge of the magnitude of were exploited preferentially (i.e., predation on bay anchovy and other intentionally fishing the larger fish in forage fish is important, therefore, if the ecosystem), since this strategy multispecies plans for these predator reduces predation on smaller fish species (e.g., bluefish, weakfish, and, therefore, increases their pro- striped bass) are developed. duction. Although other researchers within the ICES community have explored the size–spectrum approach Ecosystem Models as an option for fisheries and ecosys- Ecosystem models form another tem management (Rice and Gislason approach to characterize biological 1996; Gislason and Rice 1998; interactions in multispecies fisheries. Jennings et al. 2002), additional In effect, these models are mathemati- research is needed to fully character- cal representations of whole ecosys- ize its potential. tems, typically used to elucidate the

130 Fisheries Ecosystem Planning for Chesapeake Bay Table 2. A subset of the potential policy questions motivated by the Chesapeake 2000 agreement prioritized into the present (P), the near future (NF), and the longer-term future (LTF) based upon the Chesapeake Bay EwE model’s ability to address each issue.

Policy Issue P NF LTF

1. How can we bring back oysters tenfold and what are the cXonsequences?

2. How can we increase populations of crabs through predator X manipulation or fishery reductions?

3. How many striped bass can Chesapeake Bay support; is game X fish restoration prudent given low stocks of forage fish?

4. What are the effects of forage fish in Chesapeake Bay X ecosystem dynamics?

5. What defines a healthy Chesapeake Bay or what are trophic X limits to production?

6. How might changes in primary productivity affect upper trophic X levels?

7. How could SAV be restored and what trophic effects might X occur?

8. Can we assess the effectiveness of closed/protected areas? X

9. How might land management practices affect the estuarine X food web?

10. Can we manipulate freshwater input to increase oyster X survival?

11. How should we fish menhaden for optimal ecosystem X function? effects of fishing pressure on the depends on the availability of demo- system. Although many of these graphic data for ecologically valuable models are extremely quantitative species (as opposed to species that and employ sophisticated math- are commercially or recreationally ematical theory, researchers have valuable); such data are often un- used the models primarily for policy available. This limitation is impor- exploration and the models have tant. Nevertheless, in recent years yielded results best interpreted ecosystem models have received qualitatively. As such, ecosystem increasing attention from fisheries models can serve a useful purpose in researchers and managers who used developing management strategies. them successfully to summarize Using these models for tactical knowledge and determine properties applications is difficult, however, related to structure and function of because accurate parameterization aquatic ecosystems worldwide.

Food Web Interactions and Modeling 131 single species or a group of species 40 that represent an ecological . MSY3 35 Researchers can also split them into 6 30 ontogenetic age or size categories (juvenile, subadult, adult, etc.) if 25 necessary. The data requirements for

20 MSY2 Ecopath are fairly simple and often Species 3 15 obtainable from traditional single-

Catch (lbs x 10 ) species analytical stock assessment MSY 10 1 techniques (e.g., VPA). The 5 Species 2 parameterization of an Ecopath model Species 1 rests on the satisfaction of two master Fishing Effort equations. Equation 2 (below) Figure 18. A hypothetical graph showing a three-species describes how the biomass production fishery in which all species are caught in the same fishing gear. for each group is allocated within the The productivity of each species differs, leading to different ecosystem over an arbitrary time levels of maximum sustainable yield (MSY). The fishing effort period term. Using the principle of (i.e., levels of fishing mortality, F) that yields the respective MSYs also differs for the three species. To illustrate the conservation of matter within a importance of technical interactions, suppose that the group, equation 3 (below) balances the multispecies fishery exerts a fishing effort that leads to a energy flows of a biomass pool. mortality rate of F3. Fishing at F3 will cause the abundance of species 1 and 2 to decline to a level at which substantial risk of Production = catch + predation + net stock collapse exists. From a fisheries management migration + biomass accumulation+ perspective, reducing fishing effort to either F1 or F2 to lower the risk of collapse is desirable. This management stratetgy is fairly other mortality (2) obvious given the graph above; however, independent examination of catch/effort data for each species might not Consumption = production + yield as certain a conclusion. Adapted from Houde et al. (1998). respiration + unassimilated food (3)

One widely used class of ecosystem In general, an Ecopath model requires models is that which packages input of three of the following four Ecopath with Ecosim (EwE) and parameters: biomass, total mortality, Ecospace (Christensen et al. 2000). consumption and biomass ratio, and The development of this modeling ecotrophic efficiency for each species technique stems from early work on or biomass pool in a model. The the ecosystem dynamics of a coral ecotrophic efficiency represents the reef in Hawaii (Polovina 1984). proportion of the production used in Application of the EwE approach the ecosystem, incorporating all begins with the construction of an production terms apart from “other Ecopath model (Christensen and mortality” (for more details on Pauly 1992; Pauly et al. 2000), which Ecopath, including the equations, see creates a mass-balanced snapshot of Christensen and Pauly 1992; Walters the resources and interactions in an et al. 1997; Pauly et al. 2000). ecosystem represented by trophically linked biomass pools. The biomass Although Ecopath can describe an pools generally consist of either a ecosystem, it cannot project the

132 Fisheries Ecosystem Planning for Chesapeake Bay effects of different management thoroughly describe them. Researchers strategies on the structure and have formulated several EwE models function of an ecosystem. Only and used them for policy exploration. Ecosim—a time-dynamic simulation Trites et al. (1999) developed an EwE module that facilitates policy model of the Bering Sea to examine exploration—can accomplish these possible explanations for the changes types of projections. Ecosim re- that occurred in the ecosystem expresses the static mass-balanced between the 1950s and 1980s. Kitchell equations inherent to Ecopath as a et al. (1999) used the EwE approach to system of coupled differential study the effects of fishing down top equations (Walters et al. 1997). This predators in the central Pacific. system of equations represents the Shannon et al. (2000) used EwE to spatially aggregated dynamics of entire compare the effects of fishing in the ecosystems and is combined with southern Benguela system delay-difference, age- and size- under different combinations of structured equations to represent bottom-up and top-down control. On populations with complex ontogenies a larger scale, Stevens et al. (2000) and selective harvesting of older summarized the direct effects of . Summarized, the important fishing on chondrichthyans by computational aspects of Ecosim are examining global information on the 1) Parameter estimation based on the responses of shark and ray populations mass-balance results from Ecopath; to fisheries. They developed Ecosim 2) Variable speed-splitting methods to models of three previously published simulate the dynamics of both fast Ecopath models to simulate changes in (e.g., phytoplankton) and slow (e.g., biomass of all groups in response to large predatory fish) biomass the rapid declines of shark species. groups; 3) Explicit incorporation of top-down EwE in Chesapeake Bay (i.e., predation) vs. bottom-up In 2001, the NOAA Chesapeake Bay control (i.e., food limitation); and Office (NCBO), in collaboration with 4) Flexibility to incorporate age- and the University of British Columbia size-structure of biomass groups. (UBC), initiated a workshop series to provide the foundation for con- One obvious deficiency of EwE is its struction of an EwE model of Chesa- assumption that the resources, peake Bay. Researchers are develop- interactions, and subsequent dynamics ing a large-scale model (~50 species/ of an ecosystem are spatially functional groups) to serve as a homogeneous. Ecospace—a dynamic “base” and “continuously living” spatial version of Ecopath that model for future fisheries policy includes all of the key features of exploration (Table 2). Ecosim—overcomes this deficiency. The details of Ecospace are not Technical Interactions presented here, but Walters et al. Although technical interactions are (1999) and Pauly et al. (2000) not inherent to food web dynamics,

Food Web Interactions and Modeling 133 they are important and relevant to varying rates of fishing mortality. the topic of multispecies fisheries management (Figure 18 provides a Fishing gear (e.g., trawls, gill nets) detailed description of technical used to exploit fish populations are interactions). somewhat indiscriminant. If several fish species occupy the same geographic location at a particular Multispecies time, these fish may be captured in Yield-per-Recruit proportion to their relative Managers can use single-species yield- abundance, subject to the selectivity per-recruit (SSYPR) models to of the gear. The overall fishing determine fishing mortality rates mortality rates of the different species that achieve optimal trade-off captured can then be interpreted as a between the size of the individuals function of gear selectivity and culling harvested and the number of practices. Typically, the highest F individuals available for capture. If values are associated with the target fishing mortality is too high, then species (and size classes); a gradient of yield will not be optimal since too F values that depends on selectivity many individuals will be harvested and post-capture survival (assuming it before having a chance to grow. is not uniform) is associated with Conversely, if fishing mortality is too bycatch or discarded species. low, the yield will also not be optimal because not enough individuals will Researchers use multispecies yield-per- be harvested (even though each recruit (MSYPR) models to study individual will be large when situations in which several stocks are captured). simultaneously exploited by a single fishing gear. The calculations Conducting an SSYPR analysis re- associated with the approach follow quires an estimate or assumed value those of single-species models except for the natural mortality rate, infor- that the results are summed over all mation on growth in weight, and species (see Murawski 1984 for a data on selectivity (i.e., age or size at detailed description of the equations recruitment). Such models generally associated with MSYPR). Given this assume that recruitment—and hence summation, researchers can simulate the age-structure of the population— various regulatory scenarios that are constant over time and that reflect different levels of total fishing fishing and natural mortality remain effort (and thus F values) and the constant once the fish become selectivity of different gear types. vulnerable to fishing gear. These assumptions are important and their Using MSYPR, it is also possible to violation may have adverse effects on accommodate a situation in which model performance; however, practi- several independent fisheries harvest cal application of an SSYPR model one or more species and stocks (e.g., usually characterizes the effects of exposure to multiple fisheries due to different ages at first capture and seasonal migrations; the use of differ-

134 Fisheries Ecosystem Planning for Chesapeake Bay ent gear types). In this instance, the requirements (catch and fishing ef- fishing mortality rates for each spe- fort), SSP models generally provide a cies/stock must be adjusted to reflect starting point for fisheries stock the relative contribution by each assessments. The general surplus fishery (see Murawski 1984 for de- production model in discrete time ). takes the form:

B = B + f(B )–Y (4) Multispecies Yield-per-recruit in t+1 t t t Chesapeake Bay in which Bt represents the exploitable To date, researchers have not used biomass of a particular species at time MSYPR models to study the effects of t;f (Bt) is a general function describing technical interactions in Chesapeake surplus production (often assumed to Bay. Several types of fishing gear have be the difference between production been used historically to harvest and natural mortality) as a function of several fish species in the Bay biomass at time t; and Yt is the yield to simultaneously—most notably pound the fishery at time t. nets and haul seines (Chittenden 1989). Species typically captured in Schaefer (1954) formulated the first pound nets and haul seines include widely used equation for f (B) from striped bass, Atlantic croaker, Atlantic earlier research by Graham (1935). An menhaden, and weakfish, making application of the classical logistic these species candidates for an equation for population growth, the MSYPR analysis. The results of any Schaefer model is relatively simple and MSYPR analysis should be interpreted yields a symmetric relationship cautiously, however, since a reasonable between surplus production (dB/dt) risk exists that the assumptions and biomass. Pella and Tomlinson inherent to the model will be violated. (1969) proposed a generalized As such, characterizing the potential extension of the Schaefer model to for and effects of assumption violation alleviate the inherent requirement of should become an important having a symmetrical relationship component in any MSYPR analysis of between surplus production and Chesapeake Bay fisheries. biomass. Studies of surplus production related to stock size have shown that Multispecies Surplus non-symmetrical relationships often exist, implying that use of this Production generalized model may prove desirable. Single-species surplus production (SSP) models are typically used to One of two general modeling strategies identify the rates of fishing mortality is possible when extending the that generate sustainable yields— production model approach to including the maximum sustainable estimate multiple stocks production in yield (MSY)—given a population’s rate a multispecies fishery. The first—a of growth in terms of changes in temporal multispecies production biomass over time. Due to their (TMP) model—evaluates production simplicity and relatively modest data by applying a single-species model

Food Web Interactions and Modeling 135 using the assumption that condition of the fisheries, the hetero- multispecies fisheries behave as a geneous mix of species both within single-species stock. The term and between Jamaica and Belize, the “temporal” is used to define this diversity of fisheries targeting various approach because the production spawning, sedentary, and migratory analysis is usually based on combined fish, and the possible differences in time series of catch-and-effort data productivity among sites as factors for all species under consideration contributing to the limited success of (i.e., the parameters of f (B) are the analysis. estimated from that time-series data). Ralston and Polovina (1982) Multispecies Surplus Production in used this approach to investigate the Chesapeake Bay total production of 13 demersal fish To date, TMP and SMP models have species from the Hawaiian not been used to develop archipelago. In general, they management strategies for Bay fish. concluded that the approach could The modest data requirements of prove useful for production analysis these models relative to other in a multispecies fishery. multispecies modeling approaches, however, imply some fairly immediate The second modeling strategy evalu- possibilities for development. Since ates production over space rather watermen have used pound nets and than time. A spatial multispecies haul seines to harvest several fish production (SMP) model treats the species in the Bay, development of various locations within the total both TMP and SMP models could fished area (e.g., islands, reefs) as utilize landings data from these gears. replicate fisheries and assumes the Importantly, the MSY of the species production from each location is the complex in a multispecies production same (i.e., the parameters of f (B) are model is not simply the sum of the estimated from spatial fisheries MSYs of the individual species. data). The assumption remains that the spatially explicit fisheries are in Modeling Summary equilibrium. As with the SSP and TMP models, the results of an SMP We have reviewed five valuable analysis become unreliable if this approaches for investigating various assumption is violated. multispecies fisheries questions. Specifically, we considered MSVPA, Koslow et al. (1994) used this ap- size-spectra, and EwE models to proach to study two Caribbean reef evaluate the effects of biological fisheries in southern Jamaica and interactions, as well as MSYPR and Belize, concluding that SMP models multispecies production models to should be used with caution in reef make inferences about technical fisheries management due to the interactions. In addition to the high probability of assumption techniques described here, other violation. Specifically, Koslow and modeling techniques have proved others noted the nonequilibrium useful in evaluating impacts on marine

136 Fisheries Ecosystem Planning for Chesapeake Bay communities. Hollowed et al. (2000) Major Findings reviewed a larger body of multispecies Food Web models and described their strengths and weaknesses (as compared to The Chesapeake Bay food web has single-species models) in determining undergone significant historical the causal mechanisms responsible for alterations, primarily due to shifts in anthropogenic influences such as production. A brief summary of their eutrophication, overfishing, and general conclusions follows. habitat degradation over the past three centuries. Eutrophication may Hollowed et al. (2000) concluded that drive bottom-up control of food web multispecies models have a distinct dynamics, whereas fishing upon top advantage over single-species models predators likely dominates top-down since they depict natural mortality and control. growth rates more realistically. An exception to this lies in the use of Food web interactions—the outcome single-species models for short-term of predator–prey relationships—can predictions of large fish species, as have dramatic and substantial effects trends in predation mortality are not on the ecosystem’s community often immediately obvious. With structure, including productivity of respect to their ability to generate species supporting important fisheries. reliable long-term predictions, The form and magnitude of the effects multispecies models are a work in from altering food web interactions progress primarily because of their are somewhat unpredictable, both in sometimes strong sensitivity to form and magnitude, due to the high parameter estimates and assumptions connectivity within and between the about recruitment. Additionally, benthic and planktonic components of multispecies models may have the the Chesapeake Bay food web. Connectivity among the food web potential to describe the indirect components must be better effects of fishing on individual species. understood to avoid individual and Until they are more fully tested and aggregate population collapses or validated, however, relying on general of Bay species. rather than specific model predictions (i.e., qualitative rather than Major links between and within quantitative results) seems more benthic and planktonic components of prudent. Undoubtedly, multispecies the Chesapeake food web include models that incorporate biological Atlantic menhaden, Atlantic croaker, interactions have improved our bay anchovy, blue crab, forage fish understanding of fish population (e.g., bay anchovy), spot, and Atlantic dynamics. In some cases, lessons croaker. Researchers have only learned from these models have even identified a few keystone (e.g., blue led to improvements in the single- crab) or dominant (e.g., oyster) species species models used to characterize the in Chesapeake Bay. Further research fishing impact on individual species. must identify such species as they may

Food Web Interactions and Modeling 137 exercise control over community in that they depict natural mortality structure and productivity out of and growth rates more realistically. proportion to their abundance and dominant species may be the major An exception lies in the use of single- contributors to energy flow and species models for short-term biomass production in aquatic predictions of growth and mortality in ecosystems. large fish species in which trends in predation mortality are not always Some food web interactions that are obvious. Regarding reliable long-term not normally regarded as predator– predictions, multispecies models prey or consumer–prey interactions should be considered works in may be consequential in the food web progress due to their strong sensitivity and population dynamics of key to parameter estimates and species (e.g., disease in oyster, assumptions about recruitment. In bycatch mortality for endangered or addition, multispecies models may threatened sea turtles and birds of have the potential to describe the prey). indirect effects of fishing on individual species quantitatively. Until these Modeling models are more fully tested and The five modeling approaches validated, however, relying on reviewed in this element may prove qualitative rather than specific model useful in addressing many predictions remains prudent. multispecies fisheries questions. Specifically, MSVPA, size–spectra, Multispecies models that incorporate and EwE models can evaluate the biological interactions have improved effects of biological interactions; our understanding of fish population MSYPR and multispecies production dynamics. In some cases, the lessons models can make inferences about learned from these models have led to technical interactions. improvements in the single-species models used to characterize the In addition to models described here, impact of fishing on individual species. other modeling techniques could be used to evaluate the impacts on Panel Recommendations marine communities. Hollowed et al. (2000) reviewed a larger body of Management multispecies models and described 1) Develop life cycle diagrams and food their strengths and weaknesses (as webs for target species; use them to compared to single-species models) define important food web linkages for determining the causal and validate food web models. mechanisms that induce production Identify the major predator–prey shifts in marine ecosystems. interactions, including all significant sources of food and mortality without Hollowed et al. (2000) concluded that ignoring atypical sources (e.g., disease, multispecies models have a distinct bycatch) and noncommercial species. advantage over single-species models

138 Fisheries Ecosystem Planning for Chesapeake Bay Document beneficial aspects of food Ecosim) when developing FMPs to web interactions, such as the potential understand linkages between food benefits of bycatch as food for webs, fish habitat, environmental endangered, threatened, and changes, and fisheries production. To overexploited species (e.g., sea detect robust responses to changes in turtles). Consider vital prey species target species abundance, develop that potentially are affected by several models within each class of increases in the abundance of the food web model (e.g., multispecies target species. In examining food web stock assessments, ecosystem models), interactions of the target species, when possible, to facilitate model explore the likelihood that food web comparison. (energy transfer) approaches may fail to identify dynamically important Needed Research and Development linkages (e.g., trophic cascades and 4) Conduct field investigations to keystone species). determine and quantify major predator–prey interactions and 2) Distinguish between anthropogenic significant sources of food and and natural processes that affect mortality. water and habitat quality and thus trophic interactions, such as pollu- Predator–prey dynamics are at the tion effects on water quality, heart of interactions among species watershed influences, hydrodynam- that affect abundance and production. ics, and variation in environmental Such interactions have dramatic and processes. substantial effects on community Both anthropogenic and natural structure and may influence the yields causes—or some combination of the of Bay fisheries. At present, managers two—can alter food web dynamics. It can use the fundamental knowledge of is important to recognize the causes food web structure in a precautionary of variability before undertaking manner, but major research must management actions intended to proceed to ensure that multispecies either shift the balance among preda- fisheries management in the Bay is tor and prey species or promote the confidently implemented. productivity of prey resources by appropriate controls of fishing on 5) Conduct modeling and field investigations to determine food target species. web (energy transfer) approaches to 3) Use multiple models and varying identify and quantify the effects of data sources to explore the Chesa- dynamically important linkages peake Bay ecosystem and the ways (e.g., trophic cascades and keystone in which fishing may affect the species). food web dynamics and production Do trophic cascades and keystone of target species. species exist in Chesapeake Bay? Each modeling approach may serve a Evaluate existing Bay-area fishery- unique purpose. Run alternative food independent and fishery-dependent web models (e.g., Ecopath with databases as sources for inputs to

Food Web Interactions and Modeling 139 multispecies models. Integrate these causes of differences in model results databases to reflect a baywide scale. remain important. Thorough comparative modeling research will 6) Investigate the connectivity lead to rigorous and robust model among components of the food applications for evaluating food web web to better understand and relationships. avoid the risk of individual or aggregate extirpation of Chesa- peake Bay species. References The form and magnitude of effects of Abrams, P. 1996. The role of indirect effects in food webs. Pages 371–395 in G. A. Polis alterations in food web interactions and K. O. Winemiller, editors. Food Webs, are somewhat unpredictable due to Integration of Pattern and Dynamics. the high degree of connectivity Chapman and Hall, New York. within and between the benthic and Anderson, K. P. and E. Ursin. 1977. A multispecies extension to the Beverton and planktonic components of the Chesa- Holt theory of fishing, with accounts for peake Bay food web. phosphorus circulation and primary production. Meddelelser Fra Danmarks 7) Investigate anthropogenic (e.g., Fiskeri-Og Havundersogelser 7: 310–435. hypoxia) and natural (e.g., climate Baird, D. and R. E. Ulanowicz. 1989. The and weather) processes that affect seasonal dynamics of the Chesapeake Bay ecosystem. Ecological Monographs 59: water or habitat quality and, 329–364. therefore, trophic interactions. Bax, N. J. 1985. Application of multi- and univariate techniques of sensitivity analysis Understanding how natural processes to SKEBUB, a biomass-based fisheries control or destabilize food web ecosystem model parameterized to relationships (in addition to fishing Georges Bank. Ecological Modelling 29: effects) is important. Research on 353–382. Bax, N. J. 1991. A comparison of fish biomass variability in food consumption by flow to fish, fisheries, and in six target species, in relation to environ- marine ecosystems. ICES Marine Science mental factors and under varying Symposium 193: 217–227. environmental conditions, will help Bax, N. J. 1998. The significance and prediction of predation in marine fisheries. ICES address this issue. Journal of Marine Science 55: 997–1030. Boesch, D. F. 2000. Measuring the health of the 8) Compare and contrast the vari- Chesapeake Bay: toward integration and ous modeling approaches of food prediction. Environmental Research 82: web dynamics. Several modeling 134–142. approaches (e.g., EcoPath and Carpenter, S. R. 2002. Ecological futures: EcoSpace, multispecies virtual building an ecology of the long now. Ecology 83: 2069–2083. population analysis) are available Caswell, H. 2000. Matrix population models. to address food web dynamics. Sinauer Associates, Inc., Sunderland, Researchers should apply and Massachusetts. evaluate more than one model CBP (Chesapeake Bay Program). 2000. Chesapeake 2000 Agreement. Chesapeake concurrently. Bay Program. www.chesapeakebay.net/ Testing and comparing modeling agreement.htm. (June 2003). approaches to assess model Chittenden, M. E. 1989. Sources in variation in Chesapeake Bay pound-net and haul-seine performance and to understand the catch compositions. North American

140 Fisheries Ecosystem Planning for Chesapeake Bay Journal of Fisheries Management 9: 367– ICES Journal of Marine Science 55: 362- 371. 370. Christensen, V. 1996. Managing fisheries Graham, M. 1935. Modern theory of exploiting including top predator and prey compo- a fishery and application to North Sea nents. Reviews in Fish Biology and Fisher- trawling. Journal Conseil International Pour ies 6: 417–442. l’Exploration de la Mer 10: 264–274. Christensen, V. and D. Pauly. 1992. ECOPATH Gulland, J. A. 1965. Estimation of mortality II – a software for balancing steady-state rates. Annex to Arctic Fisheries Working ecosystem models and calculating network Group Report. ICES CM/1965/D:3. characteristics. Ecological Modelling 61: (mimeo), Copenhagen. 169–185. Hagy, J. D. 2002. Eutrophication, hypoxia, and Christensen, V. and D. Pauly. 1998. Changes in trophic transfer efficiency in Chesapeake models of aquatic ecosystems approaching Bay. Ph.D. dissertation. University of carrying capacity. Ecological Applications Maryland, College Park. 8(1) Supplement: S104–S109. Helgason, T. H. and H. Gislason. 1979. VPA- Christensen, V., C. J. Walters, and D. Pauly. analysis with species interaction due to 2000. Ecopath with Ecosim Version 4, Help predation. International Council for the system. University of British Columbia, Exploration of the Sea Committee Meeting Vancouver. 1979/G: 52, Copenhangen. Daan, N. 1973. A quantitative analysis of the Hines, A. H., A. M. Haddon, and L. A. food intake of North Sea cod, Gadus Wiechert. 1990. Guild structure and morhua. Netherlands Journal of Sea impact of blue crabs and Research 6: 497–517. epibenthic fish in a subestuary of Chesa- Daan, N. 1975. Consumption and production peake Bay. Marine Ecology Progress Series in North Sea cod, Gadus morhua: an 67: 105–126. assessment of the ecological status of the Hollowed, A. B., N. Bax, R. Beamish, J. Collie, stock. Netherlands Journal of Sea Research M. Forgarty, P. Livingston, J. Pope, and J. C. 9: 24–55. Rice. 2000. Are mulitspecies models an Daan, N. 1997. Multispecies assessment issues improvement on single-species models for in the North Sea. American Fisheries measuring fishing impacts on marine Society, Symposium 20: 126–133. ecosystems? ICES Journal of Marine Science Dunne, J. A., R. J. Williams, and N. D. 57: 707–719. Martinez. 2002. Network structure and Houde, E. D., M. J. Fogarty and T. J. Miller in food webs: robustness (Convenors). 1998. STAC Workshop increases with connectance. Ecology Report: Prospects for multispecies fisheries Letters 5: 558–567. management in Chesapeake Bay. Chesa- Duplisea, D. E. and M. V. Bravington. 1999. peake Bay Program, Scientific Technical Harvesting a size-structured ecosystem. Advisory Committee, Annapolis, Maryland. International Council for the Exploration Jackson J. B. C., M. X. Kirby, W.H. Berger, K. A. of the Sea Committee Meeting 1999/Z: 01, Bjorndal, L. W. Botsford, B. J. Bourque, R. Copenhagen. H. Bradbury, R. Cooke, J. Erlandson, J. A. Eggleston, D. B. and D. A. Armstrong. 1995. Estes, T. P. Hughes, S. Kidwell, C. B. Lange, Pre- and post-settlement determinants of H. S. Lenihan, J. M. Pandolfi, C. H. Peterson, estuarine Dungeness crab recruitment. R. S. Steneck, M. J. Tegner, and R. R. Ecological Monographs 65: 193–216. Warner. 2001. Historical overfishing and Fry, F. E. J. 1949. Statistics of a lake trout the recent collapse of coastal ecosystems. fishery. Biometrics 5: 27–67. Science 293: 629–638. Garrison, L. P., and J. S. Link. 2002. A Jennings, S. and M. J. Kaiser. 1998. The effects Multispecies Modeling Approach for of fishing on marine ecosystems. Advances Assessment of the Atlantic Menhaden in 34: 201–352. Stock. Chesapeake Bay Stock Assessment Jennings, S., M.J. Kaiser, and J.D. Reynolds. Committee Workshop Report – 2001, 2001. Marine Fisheries Ecology. Blackwell Washington, D. C. Science, UK. Gislason, H. and J. Rice. 1998. Modelling the Jennings, S., K. J. Warr, and S. Mackinson. response of size and diversity spectra of 2002. Use of size-based production and fish assemblages to changes in exploitation. stable isotope analyses to predict trophic

Food Web Interactions and Modeling 141 transfer efficiencies and predator-prey Reviews in Fish Biology and Fisheries 5: body mass ratios in food webs. Marine 195–212. Ecology Progress Series 240: 11–20. Miller, T. J., E. D. Houde, and E. J. Watkins. Jung, S. 2002. Fish community structure and 1996. STAC Workshop Report: perspectives the spatial and temporal variability in on Chesapeake Bay fisheries: prospects for recruitment and biomass production in multispecies management and Chesapeake Bay. Ph.D dissertation. sustainability. Chesapeake Bay Program, University of Maryland, College Park. Scientific and Technical Advisory Commit- Kerr, S. R. and L. M. Dickie. 2001. The tee, Annapolis, Maryland. biomass spectrum: a predator-prey Monaco, M. E. and R. E. Ulanowicz. 1997. theory of aquatic production. Columbia Comparative ecosystem trophic structure University Press, New York. of three U.S. mid-Atlantic . Marine Kitchell, J.F ., C. H. Boggs, X. He, and C.J. Ecology Progress Series 161: 239–254. Walters. 1999. Keystone predators in the Murawski, S. A. 1984. Mixed-species yield-pre- central Pacific. Pages 665–689 in Ecosys- recruitment analysis accounting for tem Approaches for Fisheries Manage- technical interactions. Canadian Journal of ment. University of Alaska Sea Grant, AK- Fisheries and Aquatic Sciences 41: 897– SG-99-01, Fairbanks. 916. Koslow, J. A., K. Aiken, S. Auil, and A. NRC (National Research Council). 1999. Clementson. 1994. Catch and effort Sustaining marine fisheries. Commission on analysis of the reef fisheries of Jamaica Geosciences, Environment and Resources, and Belize. Fisheries Bulletin 92: 737–747. Ocean Studies Board, Washington, D.C. Laevastu, T. and F. Favorite. 1988. Fishing Nixon, S. W. 1982. Nutrient dynamics, primary and stock fluctuations. Fishing News production and fisheries yields of lagoons. Books Ltd., Surry, England. Oceanologica Acta SP: 357–371. Latour, R. J., M. J. Brush, and C. F. Bonzek. Overholtz, W. J., J. S. Link, and L. E. Suslowicz. 2003. Toward ecosystem-based fisheries 2000. The impact and implications of fish management: strategies for multispecies predation on and on the modeling and associated data require- eastern USA shelf. ICES Journal of Marine ments. Fisheries 28: 10–22. Science 57: 1147–1159. Lindquist, N. and M. E. Hay. 1995. Can small Overholtz, W. J., S. A. Murawski, and K. L. rare prey be chemically defended? The Foster. 1991. Impact of predatory fish, case for marine larvae. Ecology 76: 1347– marine mammals, and on the 1358. pelagic fish ecosystem of the northeastern Link, J. S. 2002a. Ecological considerations in USA. ICES Marine Science Symposium 193: fisheries management: when does it 198–208. matter? Fisheries 27: 10–17. Paine, R. T. 1966. Food web complexity and Link, J. S. 2002b. Does food web theory work species diversity. American Naturalist 100: for marine ecosystems? Marine Ecology 65–75. Progress Series 230: 1–9. Pauly, D., V. Christensen, A. Dalsgaard, R. Lipcius, R. N., and A. H. Hines. 1986. Variable Froese, and J. Torres. 1998. Fishing down functional responses of a marine predator marine food webs. Science 279: 860–863. in dissimilar homogeneous microhabitats. Pauly, D., V. Christensen, and C. J. Walters. Ecology 67: 1361–1371. 2000. Ecopath, ecosim, and ecospace as Lipcius, R. N. and W. T. Stockhausen. 2002. tools for evaluating ecosystem impact of Concurrent decline of the spawning stock, fisheries. ICES Journal of Marine Science recruitment, larval abundance, and size of 57: 697–706. the blue crab Callinectes sapidus in Pella, J. J. and P. K. Tomlinson. 1969. A Chesapeake Bay. Marine Ecology Progress generalized stock production model. Inter- Series 226: 45–61. American Tropical Commission Macpherson. E. and A. Gordoa. 1996. Bulletin 13: 419–496. Biomass spectra in benthic fish assem- Peterson, C. H. and R. N. Lipcius. 2003. blages in the Benguela system. Marine Conceptual progress towards predicting Ecology Progress Series 138: 27–32. quantitative ecosystem benefits of ecologi- Magnusson, K.G. 1995. An overview of cal restorations. Marine Ecology Progress multispecies VPA-theory and applications. Series 264: 297–307.

142 Fisheries Ecosystem Planning for Chesapeake Bay Polovina, J. J. 1984. Model of a fisheries. Inter-American Tropical Tuna ecosystem: the ECOPATH model and its Commission Bulletin 2: 247–285. application to French Frigate Shoals. Coral Scheffer, M., S. Carpenter, J. A. Foley, and B. Reefs 3: 1–11. Walker. 2001. Catastrophic shifts in Pomeroy, L.R. 2001. Caught in the food web: ecosystems. Nature 413: 591–596. complexity made simple? Scientia Marina Schwinghamer, P. 1981. Characteristic size 65 (Suppl. 2): 31–40. distributions of integral benthic communi- Pope, J. G. 1979. A modified cohort analysis in ties. Canadian Journal of Fisheries and which constant natural mortality is Aquatic Sciences 38: 1255–1263. replaced by estimates of predation levels. Schwinghamer, P. 1983. Generating ecological International Council for the Exploration hypotheses from biomass spectra using of the Sea Committee Meeting 1979/H: causal analysis: a benthic example. Marine 16, Copenhagen. Ecology Progress Series 31: 131–142. Pope, J. G., T. K. Stokes, S. A. Murawski, and S. Seitz, R. D., R. N. Lipcius, A. H. Hines, and D. B. I. Iodine. 1988. A comparison of fish size- Eggleston. 2001. Density-dependent composition in the North Sea and on predation, habitat variation, and the Georges Bank. Pages 146–152 in W. Wolff, persistence of marine bivalve prey. Ecology C. J. Soeder, and F. R. Drepper, editors. 82: 2435–2451. Ecodynamics: contributions to theoretical Shannon, L. J., P. M. Cury, and A. Jarre. 2000. ecology. Springer-Verlag, Berlin. Modeling effects of fishing in the southern Power, M. E., D. Tilman, J. A. Estes, B. A. Benguela ecosystem. ICES Journal of Menge, W. J. Bond, L. S. Mills, G. Daily, J. C. Marine Science 57:720–722. Castilla, J. Lubchenco, and R. T. Paine. Sheldon, R. W., A. Prakash, and W. H. Sutcliffe, 1996. Challenges in the quest for key- Jr. 1972. The size distribution of particles stones. BioScience 46: 609–620. in the ocean. and Oceanography Punt, A. and D. Butterworth. 1995. The effects 17: 327–340. of future consumption by the Cape fur seal Silliman, B. R. and M. D. Bertness. 2002. A on catches and catch rates of the Cape regulates salt marsh hakes. 4. Modelling the biological interac- primary production. Ecology 99: 10500– tion between Cape fur seals Arctocephalus 10505. pusillus pusillus and the Cape hakes Sissenwine, M. P. 1984. Why do fish popula- Merluccius capensis and M. paradoxis. tions vary? Pages 59–95 in R.M. May, South African Journal of Marine Science editor. Exploitation of Marine Communi- 16: 255–285. ties. Springer-Verlag, New York. Ralston, S. and J. J. Polovina. 1982. A Sparre, P. 1991. Introduction to multispecies multispecies analysis of the commercial virtual population analysis. ICES Marine deep-sea handline fishery in Hawaii. Science Symposium 193: 12–23. Fisheries Bulletin 80: 435–448. Sprules, W. G., S. B. Brandt, D. J. Stewart, M. Rice, J. and H. Gislason. 1996. Patterns of Munawar, E. H. Jin, and J. Love. 1991. change in the size spectra of numbers and Biomass size spectrum of the Lake Michi- diversity of the North Sea fish assemblage, gan pelagic food web. Canadian Journal of as reflected in surveys and models. ICES Fisheries and Aquatic Sciences 48: 105– Journal of Marine Science 53:1214–1225. 115. Rothschild, B. J., J. S. Ault, P. Goulletquer, and Sprules, W. G. and A. P. Goyke. 1994. Size- M. Heral. 1994. Decline of the Chesapeake based structure and production in the Bay oyster population: a century of habitat pelagia of Lakes Ontario and Michigan. destruction and over fishing. Marine Canadian Journal of Fisheries and Aquatic Ecology Progress Series 111: 29–39. Sciences 51: 2603–2611. Saiz-Salinas, J. I. and A. Ramos. 1999. Biomass Stevens, J. D., R. Bonfil, N. K. Dulvy, and P. A. size-spectra of macrobenthic assemblages Walker. 2000. The effects of fishing on along water depth in Antarctica. Marine sharks, rays, chimaeras (chondrichthyans), Ecology Progress Series 178: 221–227. and the implications for marine ecosystems. Schaefer, M. B. 1954. Some aspects of the ICES Journal of Marine Science 57:476– dynamics of populations important to the 494. management of commercial marine Thiebaux, M. L. and L. M. Dickie. 1993.

Food Web Interactions and Modeling 143 Structure of the body-size spectrum of exploited ecosystems from trophic mass- the biomass in aquatic ecosystems: a balance assessements. Reviews in Fish consequence of in predatory- Biology and Fisheries 7: 139-172. prey interactions. Canadian Journal of Walters, C. J., V. Christensen, and D. Pauly. Fisheries and Aquatic Sciences 50: 1308– 1999. Ecospace: prediction of mesoscale 1317. spatial patterns in trophic relationships of Trites, A. W., P. A. Livingston, M. C. exploited ecosystems, with emphasis on Vasconcellos, S. Mackinson, A. M. the impacts of marine protected areas. Springer, and D. Pauly. 1999. Ecosystem Ecosystems 2: 539–554. considerations and the limitations of Wilson, W. H. 1991. and ecosystem models in fisheries manage- predation in marine soft-sediment ment: insights from the Bering Sea. Pages communities. Annual Review of Ecology 609–620 in Ecosystem approaches for and Systematics 21: 221–241. fisheries management. University of Winemiller, K. O. and G. A. Polis. 1996. Food Alaska Sea Grant, AK-SG-99-01, webs: what can they tell us about the Fairbanks. world? Pages 1–22 in G. A. Pollis and K. O. Vermeij, G. J. 1987. and escala- Winemiller, editors. Food webs: integra- tion: an ecological . tion of patterns and dynamics. Chapman Princeton University Press, Princeton, and Hall, New York. New Jersey. Yodzis, P. 2001. Must top predators be culled Walters, C. J., V. Christensen, and D. Pauly. for the sake of fisheries? Trends in Ecology 1997. Structuring dynamic models of and Evolution 16: 78–84.

144 Fisheries Ecosystem Planning for Chesapeake Bay