chesapeake bay fisheries: prospects for multispecies ... · multispecies fisheries approaches in...

October 30, 2006

CHESAPEAKE BAY FISHERIES: PROSPECTS FOR MULTISPECIES MANAGEMENT AND

SUSTAINABILITY

Submitted to: STACthrough CRC

Submitted by: Thomas J. MillerEdward D. HoudeElizabeth J. Watkins

Chesapeake Biological LaboratoryCenter for Environmental andEstuarine StudiesUniversity of Maryland SystemSolomons, MD 20688-0038

ii

Chesapeake Bay Fisheries: Prospects for Multispecies Management andSustainability

EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

I. BACKGROUND. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1A. Why a Multispecies Approach? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1B. The Chesapeake Bay System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6C. Chesapeake Bay Commercial and Recreational Fisheries. . . . . . . . . . . . . . . . . . . . . . . . 6D. Fisheries Management in the Chesapeake Bay. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9E. Species Interactions in Fisheries Models; The Development of Multispecies

Approaches. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13F. Objectives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

II. PATTERNS IN FISHERIES HARVESTS AND ABUNDANCES. . . . . . . . . . . . . . . . . 17A. Data Sources and Preparation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17B. Multispecies Patterns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1. Graphical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182. Multivariate analyses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

III. SPECIES INTERACTIONS WITHIN CHESAPEAKE BAY. . . . . . . . . . . . . . . . . . . . 38A. Overview of Food Webs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38B. The Chesapeake Bay Food Web. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42C. Multispecies Interactions in Chesapeake Bay. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

IV. MULTISPECIES MANAGEMENT: CONCEPTS AND APPROACHES. . . . . . . . . . 47A. The Need for a Multispecies Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

1. Biological interactions: yield, stability and resilience. . . . . . . . . . . . . . . . . . . . . . 472. Technical interactions: bycatch. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

B. Multispecies Approaches. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491. Descriptive multivariate approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492. Dynamical multivariate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533. Multivariate system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554. Integral Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

C. Practical Considerations For Multispecies Approaches. . . . . . . . . . . . . . . . . . . . . . . . . 57V. THE POTENTIAL FOR MULTISPECIES APPROACHES IN CHESAPEAKE BAY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59VI. ACKNOWLEDGMENTS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 VII. LITERATURE CITED. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

iii

EXECUTIVE SUMMARY

Fishery resources in the Chesapeake Bay are currently managed as individual species. Inthis framework the potential effect of the harvest of a species on the ecosystem generally isignored. However, the Chesapeake Bay supports a multispecies fishery that annually landsfinfish and shellfish worth in excess of $100 Million. Over the past 25 years the average annualcommercial landing has been approximately 250,000 metric tonnes. Although menhaden andblue crabs represent ninety-five percent, by weight, of the commercial catch in the Bay, statisticsshow that 59 other species are also caught. The recreational sector also accounts for a large anddiverse catch. Furthermore, there have been significant changes in the nature of the fishery. Over the last one hundred years landings of oysters have diminished greatly, and in their place,landings of blue crab have risen dramatically. Over the same time period landings ofanadromous fishes, such as American shad have declined. In contrast, landings of menhadenhave risen so that its fishery now accounts for over 80%, by weight, of the total catch.

The multispecies nature of the combined fisheries arises for both technical and biologicalreasons. Technical interactions, which arise when a fishery targets on one species but catchesother species incidently as bycatch, are present in the Chesapeake Bay. For example, 45 speciesare taken in poundnet fisheries and 53 species are taken in gillnet fisheries. Technicalinteractions are important considerations in fisheries management as they may limit the ability toregulate overall rates of fishing mortality. Biological interactions, which arise when a targetedspecies is an important link in a food web, also occur in the Chesapeake Bay. For example theremoval of top predators (striped bass, bluefish and weakfish) may have significant impacts onthe dynamics of the planktivore species, and thus the plankton community itself. Additionally,harvests of blue crab, spot and croaker have the potential to influence energy and nutrientexchanges between the benthic and pelagic food webs. To address these multispeciesinteractions several new approaches to fisheries management have been developed. Theseapproaches implicitly account for intra-specific interactions. Ultimately, these approaches maybe more compatible philosophically with the ecosystem-level management of the ChesapeakeBay’s other natural resources.

We explored the need for and potential of multispecies approaches to the management offisheries resources in the bay. The evidence suggests that adopting a multispecies approachwould be advantageous. Many of the forces that lead to the adoption of multispeciesmanagement in other ecosystems are present in the Chesapeake, including concerns overextensive bycatches, and the presence of coupled population dynamics for several components ofthe ecosystem.

We reviewed multispecies approaches employed elsewhere in the U.S. and worldwide. We identified several broad classes of approach. The most direct approaches were descriptiveinvolving graphical or multivariate statistical approaches such as principal components analysisand state-space time series analysis. These approaches are suitable to identify the extent andimportance of the multispecies character of a fishery, but may have limited utility formanagement. Other approaches are more mechanistic. Examples include closed-form, andsimulation models of interacting species, and more holistic models of the entire system. Wesuggest that multispecies models addressing technical interactions, and those involvingdescriptive rather than mechanistic approaches are most likely to be successful in the near-term. However, several approaches such as simulation modeling and multispecies virtual population

Multispecies Fisheries Approaches in the Chesapeake

iv

analysis seem unsuitable management tools for the Chesapeake Bay due to their high demand fordata that is not currently available. New research, data collection and database development tocorrect these shortcomings are strongly recommended.

Our review indicates that several factors currently preclude adopting a multispeciesapproach in the Chesapeake Bay. Specific areas that must be addressed, which would improvecurrent single-species management and develop the capability to explore the application ofmultispecies approaches include the need for:< systematic information on catch and effort for exploited stocks,< fishery-independent estimates of abundance for principal species in the bay,< basic life history information,< detailed knowledge of species interactions (especially predator - prey relationships),< effects of habitat alteration,< detailed understanding of multispecies models.

Adopting multispecies approaches to management would be a major shift away from traditionalsingle species management and a major step toward fulfilling the ecosystem management goal ofthe Bay Program. We are not ready for this step today, but addressing the identified deficiencieswill prepare us for multispecies management in the future.

Miller, Houde and Watkins

v

1

CHESAPEAKE BAY FISHERIES: PROSPECTS FORMULTISPECIES MANAGEMENT AND SUSTAINABILITY

Thomas J. Miller, Edward D. Houde and Elizabeth J. Watkins

Chesapeake Biological LaboratoryCenter for Environmental and Estuarine StudiesUniversity of Maryland SystemSolomons, MD 20688-0038

I. BACKGROUND.

A. Why a Multispecies Approach?

Fisheries in Chesapeake Bay contributesignificantly to U.S. catches at the nationaland regional levels. Recent National MarineFisheries Service (NMFS) statistics indicatethat between 250,000 - 350,000 metrictonnes (t) of fish and shellfish are harvestedannually from Chesapeake Bay waters(Fig.1), with a dockside value of more than$100 Million. Maintaining the health of thisfishery is an important but difficult taskgiven the considerable interannualvariability in catches of component species(Fig. 2). Moreover, individual speciesexhibit different, and sometimes oppositetrends, further complicating analysis andunderstanding. The variability and trendsillustrated in Figures 1 and 2 are not uniqueto Chesapeake Bay, but characterizefisheries in general (Hilborn and Walters1992). Recent research has sought toinclude effects of species interactions andoverall community structure intomanagement practices by developingmultispecies management models, therebyincorporating the complexity of multispeciesdynamics into the regulatory process (Kerrand Ryder 1989; Appollonio 1994).

Multispecies approaches to fisheriesmanagement do not represent a single,unified methodology. However, all of the

approaches share a common attribute; allwere formulated to account for the effects ofremoval of a single species by a fishery onother species in the system. As an example,when individuals of one species are removedby a fishery from an ecosystem, its predatorspresumably have less food, and its preyexperience reduced mortality. Thesechanges can alter the population dynamics ofindividual species and overall communitystructure. Moreover, fisheries are unlikelyto exploit a single species. Thus, increasedfishing on one species, may lead to increasedincidental mortality, or bycatch, in otherspecies that also are vulnerable to theparticular fishing method. A variety oftechniques can account for such interspecificinteractions. Some are direct extensions ofexisting single-species approaches, such asthe multispecies surplus production models(Sainsbury 1988, 1991). However, there arealso techniques unique to the multispeciescase, such as network analysis (Christensenand Pauly 1992).

The Chesapeake Bay is a complexecosystem that supports living resources atseveral trophic levels and in diverse habitats(Baird and Ulanowicz 1989). Increases innutrient loading during the recent past havebeen implicated in or have lead to dramatic,system-wide changes (Horton and Eichbaum1991). The loss of submerged aquatic


2

0

50

100

150

200

250

300

350

400

1850 1900 1950 2000

Year

Land

ings

(Ton

nes

x 10

3 )

Figure 1. Time series of total Chesapeake Bay commercial fishery landings (thousands ofmetric tonnes) from 1880 - 1992. Prior to 1956, Atlantic coast catches landed at Chesapeakeports were included in this total. Data for 1993-1994 are preliminary and are shown by the


3

0

1

2

3

4

1860 1880 1900 1920 1940 1960 1980 2000

Year

0

20

40

60

1880 1900 1920 1940 1960 1980 2000

0

100

200

300

400

0

20

40

60

0

2

4

6

8

Land

ings

(ton

nes

x 10

3 )A) Hard-Shell Blue

B) Menhaden

C) Oyster

D) American Shad

E) Striped Bass

Figure 2. Time series of commercial landings (thousands of metric tonnes) for selectedindividual species taken from the Chesapeake Bay. Prior to 1956, Atlantic coast catcheslanded at Chesapeake ports were included in this total. In combination the 5 speciesillustrated have represented between 95-99% of the total the Chesapeake Bay commercialcatch. Data for 1993-1994 are preliminary and are shown by the dotted lines.


4

vegetation is one example. A secondexample is the increase in phytoplanktonblooms and related incidence of anoxicwaters in the Bay, caused by elevated levelsof nitrogen and phosphorus (Boicourt 1992). Overall, the system-wide changes hadimportant consequences for the distributionand productivities of living resources(Boynton et al. 1995). Currently, as nutrientcontrols are implemented throughout theChesapeake watershed, system-levelresponses are likely to re-occur. Suchsystem-level responses infer that livingresources in the Chesapeake Bay are tightlylinked and dependent upon one another,implying that a multispecies approach,which recognizes such dependencies, mayhave merit for fisheries management.

As with all ecosystems, multispeciesinteractions within the Chesapeake Bay maybe driven both by trophic relationshipswithin the ecosystem, and by forcing fromoutside, for example by nutrient loading(Carpenter et al. 1985). Research on thedynamics of the trophic network in theChesapeake Bay is well advanced and isrevealing a complex pattern of temporallyand spatially explicit dependencies amongpredators and prey (Baird and Ulanowicz1989; Brandt et al. 1995). This complexityresults, in part, from substantial seasonalvariation in community assemblages. Forexample, Baird and Ulanowicz (1989)indicated that control over the planktoncommunity is exercised by crustaceanzooplankton in the spring, but by the seanettle (Chrysaora quinquecirrha) in thesummer months. Moreover, strong seasonalmigrations within the fish communityproduce clear seasonal differences inpotential levels of piscivory and piscivoregrowth (Hartman and Brandt 1995a, b). Inaddition to the seasonal patterns in predator-

prey interactions, the linkage of benthic andpelagic food chains creates the potential forlarge seasonal shifts in nutrient cycling(Baird and Ulanowicz 1989; Baird et al.1995; Bartleson and Kemp 1990; Kemp andBartleson 1990).

Both exploited and unexploited speciesare involved in the extensive web of trophicinteractions in Chesapeake Bay (Fig. 3). Among the exploited species in the Bay,these interactions may affect productivepotentials. For example, bluefish(Pomatomus saltatrix) may compete withweakfish (Cynoscion regalis) and stripedbass (Morone saxatilis) because all rely onbay anchovy (Anchoa mitchilli) and Atlanticmenhaden (Brevoortia tyrannus) as prey(Hartman and Brandt 1995a). It is possibleto trace many similar dependencies in Figure3. Thus, because removal of individuals ofone species by fishing has effects on othercomponents of the ecosystem, a fishery mayexercise broad control over the structure andproductive capacity of the ecosystem. Forexample, Newell (1988) suggested that thereduction in oyster (Crassostrea virginica)populations in the Bay by fishing and habitatdegradation (Rothschild et al. 1994) mayhave amplified the effects of increasednutrient loadings which, combined with lossof oysters, has led to profound changes inthe living resources of Chesapeake Bay. Thus, management actions that targetparticular components of the ecosystem andfail to recognize interactions ordependencies may result in broad scalechanges in ecosystem structure andproductivity.

Overall, evidence suggests thatfisheries, including those in ChesapeakeBay, may benefit from management at thesystem level, rather than at the individual


5

Figure 3. Energy flow network in the mesohaline area of the Chesapeake Bay duringsummer (from Baird and Ulanowicz 1989). Values are as follows: numbers associated witharrows are carbon flows in mgAm-2Asummer-1, numbers in the tops of boxes are identifiers,numbers in the bottoms of boxes are the biomass of carbon mgAm-2.


6

species level. If true, then the traditionalfisheries approach of managing singlespecies must be reassessed. Traditionalsingle-species approaches and models (e.g.Beverton and Holt 1957) implicitly treat theexploited species in isolation from thesystem in which it lives. In the past 15years, a multispecies approach to fisheriesmanagement has emerged (Kerr and Ryder1989; Appollonio 1994). This evolvingmultispecies framework for fisheriesmanagement is philosophically compatiblewith the ecosystem-level focus of managingdischarges, runoff, and nutrient loading intothe Chesapeake Bay (Chesapeake BayProgram 1995).

B. The Chesapeake Bay System.

The Chesapeake Bay, located on themid-Atlantic coast of North America fromlatitude 36/50’ to 39/40’ N, is the largestestuarine system in the United States. It isapproximately 320 km long, up to 50 kmwide, and on average 6.4 m deep. Its 16Million km2 watershed spans six states (DE,MD, NY, PA, VA, WV) and includes fourmajor cities (Baltimore, MD; Norfolk, VA;Richmond, VA; and Washington, DC)(White 1984). Nineteen principal rivers andmore than 400 creeks and streams feed intoits 7,400 km of tidal shoreline. Eightypercent of the freshwater entering theChesapeake comes from the Susquehanna,James and Potomac rivers (Lippson andLippson 1984), with the Susquehanna alonecontributing almost 50% of the total freshwater flow. The average flushing time ofthe Bay is about 42 days (Baird andUlanowicz 1989). A semidiurnal tidalpattern occurs throughout the Chesapeake,ranging in magnitude from one meter in thesouthern Bay, to 30 centimeters nearAnnapolis, and to 60 centimeters at the head

of the estuary (Lippson and Lippson 1984). The salinity gradient within the

Chesapeake Bay ranges from freshwater atthe head of the Bay to nearly 30 PSUseawater at its mouth. There are three majorecotones within the Bay’s well-definedsalinity gradient: oligohaline (0-6 PSU),mesohaline (6-18 PSU), and polyhaline (>18PSU). Thus, organisms adapted to bothmarine and freshwater environments residein or visit the Bay seasonally, providingdiverse opportunities for fishery harvests.The mesohaline region spans 48% of thebay’s surface area, and encompasses 47% ofthe volume of the estuary (Baird andUlanowicz 1989). Longitudinal salinitygradients may change by as much a 5 PSU inthe course of a year due to seasonaldifferences in freshwater flow.

Because of its size and location, the Bayis important to the ecology of the entireAtlantic seaboard of North America. Itprovides wintering grounds and breedinggrounds for both shorebirds and neotropicalmigrants. It is an important nursery formany economically important fish species. Both commercial and recreational fishingoccur extensively throughout theChesapeake Bay. From an economicviewpoint the Bay provides direct benefitsfrom the animals harvested from its waters,the passage of shipping through itswaterways, and from expenditures derivedfrom the wide variety of recreational activitypursued in its watershed.

C. Chesapeake Bay Commercial andRecreational Fisheries.

Fisheries have always been important inthe Chesapeake Bay ( Fairbanks and Hamill1932; Wharton 1957). Many species havebeen and are exploited for consumption, forindustrial products such as fish oil and for


7

recreation. The patterns of fishing havebeen diverse and include examples ofspecialist, targeted fisheries such as themenhaden purse-seine fishery and moregeneral, opportunistic efforts such as thepoundnet fishery. Importantly, declines inabundance of fish within the Bay are notonly a recent concern. The effects of damson anadromous fish species has been aconcern for over 200 years (Wharton 1957;Loesch and Atran 1994). Overfishing is nota recent problem either. Rothschild et al.(1994) present compelling evidence that theoyster was overfished from the late 1800'sonwards. Atlantic sturgeon (Acipenseroxyrhynchus) abundances had been greatlyreduced early this century by the directeffects of fishing, and now are almost absentin the Chesapeake system (Horton andEichbaum 1991; Secor 1995).

The increase in demand for finfish andshellfish at the beginning of this century leadto increases in fishing effort in the Bay. In1880, more than 100,000 t (.250 Millionlbs) of fish and shellfish were landed fromthe Chesapeake and adjoining waters (Fig.1), with a total value of over $8 Million(Fairbanks and Hamill 1932). Between 1880and 1930, landings varied, peaking in 1920,when possibly 250,000 t (. 600 Million lbs)was harvested from Chesapeake Bay waters(Fairbanks and Hamill 1932). During thisperiod, approximately 60% of the annualharvest came from finfish, principallymenhaden (Fig. 4). Shellfish, principallyhard-shelled blue crab (Callinectes sapidus)and oysters, represented the remaining 40%(Fig. 4). Principal gears used were purseseines for menhaden, poundnets for shad,trotlines for crabs and tongs for oysters(Chowning 1990). Since the 1930's, totalcommercial landings have approximatelydoubled from almost 125,000 to 250,000 t

(. 300 - 600 Million lbs) in 1992 (Fig. 1). Dockside values of commercial harvests inthe 1990's are valued at more than $100Million annually.

Commercial landings, althoughvariable, clearly have risen steadily since1930. In fact, commercial landingsincreased quite rapidly until the mid-1980's(approx. 2.9% yr-1) before stabilizing (Fig.1). The most significant change in thecommercial fisheries since 1930 is theincreased contribution of menhaden to thetotal landings. The proportion of the totalcatch contirbuted by menhaden has doubledfrom approximately 40% in 1930 to over80% today (Figs. 2B, 4).1 A secondsignificant change is the decline in oysterharvests (Figs. 2C, 4), which began in thelate 19th century, but which have fallenfrom 15% of the total in 1930 to less than1% today (Rothschild et al. 1994). Thepicture is even more dramatic if oneconsiders value, rather than weight, of thetwo species. In the 1930's the values ofharvested menhaden and oysters were lessthan 10% and more than 40% of the totalvalue of Bay landings, respectively. In the1990's these figures were approximately20% and 5%, respectively. These dramaticchanges reflect both changes in theecosystem resulting from the precipitousdeclines of oyster stocks in the Bay and alarge and relatively stable stock ofmenhaden on the Atlantic coast since the1970's.

Recreational fisheries are of major

1Menhaden landings for recent years are notreported directly in Chesapeake Bay fisheriesstatistics. For this report, approximate landings werederived from an analysis of total menhaden landingsand Chesapeake Bay menhaden landings. See page18 for explanation.


8

16

11

9

7

13 0000000000

45

A

42

15

12

11

6

13 0000000000B

56

15

10

8

28C

60

17

7

5 2 3D

77

11

4 2 000000000E

82

132

F

Menhaden Crabs, Blue, Hard Oysters

Crab, Blue, Soft Spot Other

Figure 4. Average percent composition of Chesapeake Bay catch for the most abundantspecies for the decades A) 1930's, B) 1940's, C) 1950's, D) 1960's, E) 1970's and F) 1980's. Percent values are shown next to each segment.


9

importance in the Bay (Fig. 5). Since 1981, the National Marine Fisheries Service hasannually surveyed the recreational fishery. A diverse array of marine, anadromous andfreshwater species is caught, includingstriped bass, bluefish, weakfish, summerflounder (Paralichthys dentatus), croaker(Micropogonias undulatus), spot(Leiostomus xanthurus), white perch(Morone americana), and channel catfish(Ictalurus punctatus). NMFS data suggestthat for the principal species, recreationallandings are of the same magnitude orexceed the commercial harvest (Fig. 5A). Acatch exceeding 2,500 tons has been landedin recent years. However, this is far lowerthan the catches taken in the early to mid1980s. The sharp decline in bluefishlandings in the late 1980's reflects a coast-wide decline, rather than one specific to theChesapeake Bay. Additionally, effort in therecreational fishery is likely to increase. Inthe 15 years of the NMFS recreationalsurvey the estimated number of annualfishing trips on the Bay has more thandoubled from 2 - 4.5 million (Fig. 5B). Moreover, the catch and impact of the verylarge recreational fishery for blue crab areessentially unknown. Recreational landingsof blue crab might be 10-25% of those in thecommercial fishery (Chesapeake BayProgram 1996). D. Fisheries Management in theChesapeake Bay.

Both federal and state agencies haveresponsibility for managing fisheries withinthe Chesapeake Bay. Maryland and Virginiaare responsible for regulations of fisherieswithin their territorial boundaries. However,the majority of stocks of individual speciesin the Bay span both jurisdictions. The

efforts of both states are now coordinated viathe Chesapeake Bay Program, whichoversees development of FisheryManagement Plans (FMP) as specified in the1987 Chesapeake Bay Agreement. Furthermore, if a species is migratory andspends some of its life in coastal or oceanicwaters, it is subject to the jurisdiction offederally appointed bodies. In these cases,management is coordinated eitherindividually or jointly by the Atlantic StatesMarine Fisheries Commission (coastalspecies within 3 miles of the coast), and theMid-Atlantic Fishery Management Council(3 to 200 mile offshore zone).

All fisheries within the Bay aremanaged on an individual species basis. Richkus et al. (1992) provide an excellentsummary of the approaches used. It isimportant to note that all managementapproaches identified by Richkus et al.(1992) rely on similar conceptual models. Inall of the models applied, fluctuations inabundance of individual species result fromchanges in rates of four processes:recruitment, growth and natural and fishingmortalities (Fig. 6). Increases in the rates ofrecruitment and growth lead to increases inabundance and biomass. In contrast,increases in the rates of natural and fishingmortalities cause decreases in abundance andbiomass. Elaborations of the conceptual model(Fig. 6) have produced management models,such as the surplus production, spawner-recruit analyses and yield per recruit models,that were originally formulated by Schaefer(1954), Ricker (1954) and Beverton and Holt(1957), respectively. Recent efforts haveexpanded these themes to include age andspatial structure, stochastic processes andenvironmental covariates in the basicapproaches (Gulland 1988). Principal


10

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

1981 1986 1991

Year

Tota

l Num

ber o

f Fis

hing

Trip

s (M

illio

n)

0

1000

2000

3000

4000

5000

6000

1981 1986 1991

Rec

reat

iona

l Cat

ch (t

) CroakerSpotBluefishStriped BassWeakfish

A

B

Figure 5. The importance of the recreational catch. A) time series of the estimated totalrecreational catch for five species. B) time series of the estimated total number of fishingtrips. Data are the summed values from the NMFS marine recreational fishery survey for theState territorial waters of Maryland and Virginia. These figures include catches from allestuaries, enclosed bays and other marine waters not directly offshore.


11

Stock

Recruitment NaturalMortality

FishingMortalityGrowth

+

+ -

-

Stock

Recruitment NaturalMortality

FishingMortalityGrowth

+

+ -

-

Figure 6. Conceptual representation of processes regulating stock size. Signs on each flowarrow indicate whether processes act to increase or decrease stock size. Recruitmentis defined as the number of young individuals entering the fishable stock annually. Natural mortality is the number of deaths from all causes other than fishing.


12

No N1 N2 N3 N4

D0 D1 D2 D3 D4

C0C1 C2

C3 C4

Virtual Population

No N1 N2 N3 N4

D0 D1 D2 D3 D4

C0C1 C2

C3 C4

Virtual Population

Figure 7. Conceptual representation of single species VPA (after Sparre 1991). Boxeslabeled N0 - N4 represent the abundance of individuals in age classes 0 - 4. Boxes labeled Ciand Di represent the associated losses due to fishing and natural mortality, respectively.


13

among these are the age-structured, cohort-based methods derived by Gulland (1965)and Pope (1972) that lead to virtualpopulation analysis (VPA), which underliesmuch of modern marine fisheriesmanagement. The VPA is an elaborateaccounting procedure. Essentially, VPA isan application of the basic conceptual modelpresented in Figure 6 to every age (or stage)of a fish’s life history. We illustrate thisconceptually in Figure 7. The “virtualpopulation” is the summed catches andmortalities over all ages. If the naturalmortality schedule is known, then the age-specific abundances and fishing mortalityrates required to produce the observedcatches can be hindcasted from the oldest age class to the youngest (Hilborn andWalters 1992). The application of a VPA orother model is dictated by both the biologyof the species concerned and the availabilityof biological and fishery data toparameterize the models. Richkus et al.(1992) identify deficiencies in data as aprincipal factor limiting application of themore sophisticated models, such as VPA, tothe Chesapeake Bay’s natural resources.

E. Species Interactions in FisheriesModels; The Development of MultispeciesApproaches.

Despite increases in sophistication,current fisheries modeling and managementapproaches, clearly show the heritage ofwork conducted during the first half of thiscentury. Most importantly, practicescontinue to be dominated by a single-speciesapproach which treat the harvested resourceas largely isolated from the ecologicalcommunity within which it lives. Effectively, management has focussedprimarily on the interaction between stocksize and fishing mortality and recruitment;

ignoring growth and natural mortality (Fig.6). However, all exploited species exist incomplex ecosystems, in which individualspecies interact with one another. Theseinteractions exert influence on the harvestedspecies through the processes of naturalmortality, growth and recruitment illustratedin Figure 8. The single-species approach,which discounts, a priori, several potentialcontrollers of stock size, may have seriouslimitations, especially when applied totightly coupled ecosystems. Moreover,single-species approaches also ignore thefact that the harvest in many, even perhapsmost, fisheries is not restricted to a singlespecies. Many non-targeted species aretaken as bycatch (i.e. a technicalinteraction), and on occasion, for example intropical shrimp fisheries, can represent themajor fraction of the total harvest (Alversonet al. 1994). Such biological and technicalconcerns are motivating the development ofmultispecies approaches to fisheriesmanagement.

Early attempts to incorporateinteractions between species into fisheriesmanagement models can be found inLarkin’s (1963) and Riffenburgh’s (1969)research. However, the first clear statementof the need for a multispecies approach camein 1977, when the U.N. Food andAgriculture Organization convened a panelof experts to examine the need for a broadermultispecies approach to management (FAO1978). This report highlights the potentialimportance of multispecies managementapproaches for both biological and technicalreasons. We summarize the features of afishery that have motivated the developmentof multispecies approaches in Table 1.

Subsequent to the FAO (1978) report,fishery scientists began to explore the


14

Table 1. Characteristics of fisheries motivating a multispecies approach

Category Examples

Technical Incidental bycatch of non-target species in targeted fisheriesUse of non-specific fishing gearExploitation of mixed species schools

Biological Extensive predation mortalityCompetition among species for foodSimilar responses to environmental forcing

importance of multispecies approaches(Sissenwine and Daan 1991). Researchpromoted by the International Council forthe Exploration of the Sea (ICES) wascentral in these developments. In 1977,Andersen and Ursin developed a detailedecosystem model of the North Sea that, forthe first time, permitted estimation of theintegrated effects of predators andcompetitors on harvested species. Simultaneously, Daan (1973, 1975)recognized the potential importance ofincluding interspecific interactions inunderstanding the dynamics of cod (Gadusmorhua) in the North Sea. The combinationof these two developments and the FAOreport lead to the international “year of thestomach” in 1981, which sought to providedetailed spatial and temporal data on the trophic interactions within the North Seathrough intensive stomach sampling(Sissenwine and Daan 1991).

Based upon the concepts in Andersenand Ursin’s (1977) simulation model, fourgroups independently developedmanagement approaches that have come tobe known as Multispecies Virtual PopulationAnalysis (MSVPA). Conceptually, MSVPAextends the single-species approach bymaking the natural mortality and growth

components of Figure 6 functions of theabundances of predators and competitors,respectively. This gives rise to a linkedsystem in which the effects of changes inabundance of a single species can be mappedonto changes in abundance of other speciesthrough its effects on its prey andcompetitors (Fig. 8). The only additionalcomplexities are that each stock is dividedinto a series of age classes and many speciesof minor importance are combined into asingle category.

Much of the early research on MSVPAtechniques was applied to the North Seagroundfish fishery (Sissenwine and Daan1991). ICES again played a prominent role. The wide application of MSVPA techniqueshas been limited because they are dataintensive, and because they often producecounter-intuitive predictions of populationbehavior under heavy exploitation (e.g.Sparholt 1994). Moreover, because MSVPAmodels are complex, their mathematicalbehavior remains poorly characterized orunderstood (Magnusson 1995). Consequently, MSVPA models are often used heuristically to investigate probablebehavior of fished populations underdifferent exploitation strategies. To date, managers have been reluctant to accept them


15

S1

R M

FG

S2

R M

FG

S3

R M

FG

By-catch

Pred

atio

n

S1

R M

FG

S2

R M

FG

S3

R M

FG

By-catch

Pred

atio

n

Figure 8. Conceptual examples of biological and technical stock interactions for three stocksS1, S2 and S3. Stocks S1 and S2 are linked by a predator-prey relationship (biologicalinteraction) in which adults of S1 eat juveniles of S2. Thus increased predation by S1 leads toincreases in its growth rate and decreases in recruitment to S2. Stocks S1 and S3 are linked bya by-catch issue (technical interaction) in which juveniles of S3 are a bycatch in a targetedfishery for S1. Increased F1 leads to a reduced recruitment to S3. R = recruitment; G =growth in weight; M = natural mortality; F = fishing mortality.


16

as a primary management tool (Brugge andHolden 1991). However, MSVPA modelshave been applied to several temperatefisheries (see examples in Pope and Macer1991; Sparholt 1991).

F. Objectives.

In this literature synthesis we explorethe background of and justification formultispecies management. We consider anddiscuss possible advantages of the approachto ensure sustainable harvests of ChesapeakeBay’s living resources. We review theapproaches that have been adopted in otherecosystems to determine which multispeciesapproach might be appropriate inChesapeake Bay and, in doing so, weidentify gaps in our understanding that mustbe filled before multispecies approaches canbe fully evaluated or adopted. Specifically,our objectives are to:1. Analyze fishery-dependent and fishery-

independent datasets to explore thedegree to which species interactions canbe identified and to determine if there issignificant evidence of correlatedpatterns in abundance of importantcomponents of the Bay’s fisheriesresources.

2. Review patterns of interaction amongspecies within the Chesapeake Bayecosystem. We focus particularly onidentifying resident species andseasonal visitors to the ecosystem thatinteract closely. We then explore thepotential impact of these stronginteractors on the fisheries within theBay.

3. Review the concepts and approachesthat underpin multispecies management. The review is conducted within aclassification framework ofmultispecies approaches that recognizesdescriptive, dynamical, multivariate andintegral categories. We discuss theassumptions and data requirements ofeach approach.

4. Address the potential application ofmultispecies management in theChesapeake Bay. Specifically, weassess whether the conditions that leadto multispecies approaches in othersystems are present in the Bay and itsfisheries. We identify which, if any, ofthese approaches is most appropriate forChesapeake Bay, and identify gaps inour knowledge that currently precludeeither the evaluation or application ofmultispecies management in the Bay.


17

II. PATTERNS IN FISHERIES HARVESTS AND ABUNDANCES.

In this chapter we analyze the principlefisheries within the Chesapeake Bay toidentify possible species interactions. As apreliminary step to assess the utility of amultispecies approach to Chesapeake Bayfisheries management, we explore thepatterns in landings to identify the extent towhich species landings are correlated.

A. Data Sources and Preparation.

We examined landings data provided toNMFS by the Virginia Marine ResourcesCommission and Maryland Department ofNatural Resources for the period 1880 -1992. Landings data for 1993 -1994 arepreliminary, due in part to statisticalreporting changes in Virginia. Thus, whilewe have analyzed only data up to 1992, weillustrate landings for 1993-1994. We wereunable to standardize catch to fishing effortbecause effort data are unsystematicallycollected and recorded for most species. Rothschild et al. (1981) identified the sameproblem 15 years ago. As a result, there issome uncertainty whether trends in the datareflect changes in overall populationabundance or changes in fishing effort. Togain insight on trends in populationabundances, we also analyzed survey datafrom the Virginia trawl survey seriesprovided by the Virginia Institute of MarineScience. These research-survey data arestandardized to effort and changes in catch-per-unit-effort reflect changes in underlyingpopulation abundances of some species.

Historically, landings records date to1880 but are inconsistent until 1929. Changes in reporting and recording methodscomplicate analysis and interpretation oflandings trends. Prior to 1956, landings

reported for Maryland and Virginia wereaggregated Chesapeake Bay and Atlanticcatches landed at ports in the two states. After 1956, the landings from the Bay andAtlantic sources were separated, and thus,from 1957 onwards harvest data are specificto the waters of the Bay. For some species,such as butterfish (Peprilus triacanthus) andscup (Stenotomus chrysops) which arecaught almost exclusively in Atlantic waters,the distinction is not problematic since allharvests prior to 1956 represent ocean-sidelandings. However, for species which arecaught both within the Bay and in theAtlantic, no simple partitioning is possible. A correlation analysis for principal Bay-specific and ocean-specific landings from1957-1990 was attempted to partitionlandings prior to 1957. However, only thecorrelation for weakfish was significant (r =0.76, n = 21). Thus, we were unable toremove this source of bias from our analysis.

There are difficulties in determiningmenhaden catches, the most abundantspecies landed by weight. Since 1977 onlytwo firms have fished for menhaden, thuspreventing NMFS from reporting the catch,because reporting it would allow each of thecompanies to calculate the catch of the other. Consequently, from 1977, there is anenormous increase in the landings of“unclassified finfish” in the data set,increasing by almost two orders ofmagnitude from under 3,000 to 200,000 tbetween 1976 and 1977. Obviously, theincrease represents menhaden, and we haveused data for “unclassified finfish” as asurrogate for menhaden from 1977 onwards. However, approximately 1% of this catchthat we ascribe to menhaden is unclassifiedfinfish other than menhaden and cannot be


18

(2)

separated in the statistics. Our analyses arenot compromised by this small error. Asecond problem with menhaden landingsrecords is a complete lack of data formenhaden in 1980-1983. To estimate themissing landings data for these four years wecorrelated menhaden landings for theChesapeake region with the total U.S.menhaden catch for the period 1963-1977 (Fig. 9A). Menhaden landings fromChesapeake Bay were significantlycorrelated with U.S. menhaden landings:

Bay Menh. = 0.59@ U.S. Menh. - 11.4 x 103, r2 = 0.57, n = 13 (1)

Overall, there was a satisfactory pattern inthe residuals of the relationship betweenChesapeake Bay menhaden and total U.S.menhaden landings (Fig. 9B). Consequently, we derived an estimate ofChesapeake Bay menhaden landings for thefour years 1980 -1983 using U.S. menhadenlandings in Equation 1.

No fisheries landings data wereavailable for 1943. To fill this gap, weestimated landings for each species for thisyear as:

where the subscripts refer to year. The final data set, developed as

described above, was analyzed to determinethe presence and importance of multispeciespatterns in the Chesapeake Bay. Anecessary step, though not a direct objectiveof this literature synthesis, was an analysisof trends in landings of individual species. For brevity, we will only summarize them

here. There is a strong increasing trend intotal landings. However, when oneconsiders fluctuations around this trend, totallandings in the current year are stronglyrelated only to landings in the previous year. Such time series are said to beautoregressive. This form of autoregressiverelationship, with a clear relationship only tothe year immediately prior, describes a timeseries that exhibits abrupt changes from yearto year. The landings time series formenhaden, oysters, soft-shell blue crab andalosids (Alosa spp.) show similar patterns. In contrast, striped bass and hard-shell bluecrab exhibit a different pattern. Landings foreach of the latter two species show anautoregressive pattern in which currentlandings are best predicted from landingsover the last three years. This form ofautoregressive relationship describes alandings time series in which there are moregradual fluctuations, with distinct periods ofabove average, and below average landings.

In the following sections we present theanalysis of multispecies patterns. Simplegraphical techniques and multivariatestatistical approaches were used to determinethe extent of multispecies interactions. Allstatistical tests were conducted using SASv6.10. Details of the statistical approachesare provided at the beginning of eachsection.

B. Multispecies Patterns.

1. Graphical analysis.To detect multispecies trends in the fish

landings in the Chesapeake Bay, wedeveloped plots of rank order of abundance in catches (Fig. 4) and time series ofmultispecies groupings (Figs. 10-17). No


19

0

50

100

150

200

250

300

0 50 100 150 200 250 300 350 400 450

Che

sape

ake

regi

on la

ndin

gs (t

onne

s x

103 )

Total U.S. menhaden landings (tonnes x 103)

-150

-100

-50

0

50

100

1960 1965 1970 1975 1980 1985 1990 1995

Year

Res

idua

l

B

A

Figure 9. Relationships between landings data used to estimate menhaden landings in 1980-1983. A) Relationship between Chesapeake Bay menhaden and total U.S. coastwidemenhaden landings for 1963 - 1977. B) Annual residuals of regression relationships betweenChesapeake Bay menhaden and total coastwide menhaden landings for 1963-1977.


20

formal statistical analysis was conducted oneither the rank abundances or on the timeseries. The length of most time seriesprecluded application of formal time seriesapproaches. However, visual inspection wassufficient to detect the presence of generaltrends or striking coherence among timeseries.

The rank order of commercial landings,which may index abundance in some cases(e.g. oyster), has remained stable since the1930's (Fig. 4). Menhaden, blue crab,oyster, spot and alewife (Alosapseudoharengus) have dominated thelandings. However, the evenness of thecatch has declined. In the 1930's, thelandings of menhaden, blue crab, croaker,oyster, alewife, weakfish, and Americanshad combined to produce 95% of the totallandings. By the end of the 1980s,menhaden and hard-shell blue crabs aloneproduced 95% of the landings. This patternis driven by the increases in crab andmenhaden catches (Figs. 2A, B) andconcomitant declines in oyster landings (Fig.2C).

It is important to note that the historicallandings pattern likely does not reflectunderlying changes in the abundance ofsome species, but is a consequence ofshifting effort that reflects changes in themarketplace. To assess the extent to whichlandings data reflected underlyingabundances we compared the landings timeseries with equivalent fishery-independenttime series collected by the Virginia Instituteof Marine Sciences. To aid presentation, wepresent representative comparisons asanomaly plots in which the time series areplotted as percent deviations from the longterm mean (Fig 10A-D). Direct comparisonof these plots is difficult because the VIMSsurvey focuses on younger, pre-recruitfishes. Thus, for some species, the two time

series should be lagged to reflect age atrecruitment. Moreover, the period of lagwill be species-specific, preventing use of ageneral correction factor. However, formost of the principal species the grosspattern of the two time series is similar. Weinterpret this to mean that while the landingsdata are not equivalent to abundance, theydo reflect underlying changes in abundanceand are consequently, an adequate surrogatein our analyses.

To explore multispecies patterns in thedata set, we grouped species that werereported in commercial landings into fourbroad categories. Some species wereassigned to more than one category. Thefour categories were taxonomic, trophic,habitat, and habit. Groups were designatedbased upon two taxonomic categories: 1)families and 2) finfish vs. shellfish. Trophicgroup membership was determined by thepredominant items in the diet. We definedthree trophic groups (planktivore, piscivore,benthivore). Habitat groupings were basedupon whether an animal was pelagic ordemersal and with respect to the salinityrange within which it occurs. Finally, threehabit groups (resident, seasonal resident,occasional) were defined. Unless noted,data for menhaden were excluded from theseanalyses because they dominate most trends. Details of the classification are presented inTable 2.

Landings of non-menhaden finfish andshellfish have varied considerably (Fig. 11). Several features of this figure arenoteworthy. Between 1920-1960 non-menhaden finfish and shellfish landingswere of similar magnitude. Moreover, theydemonstrate a high degree of coherence (r =-0.61 at lag = 4 yrs. See Table 4 for details). However, since 1970, the two time serieshave diverged as the non-menhaden finfish


21

-200.00

-100.00

0.00

100.00

200.00

1957 1967 1977 1987

Year

Rel

ativ

e A

nom

lay

C. Alewife

-200.00

-100.00

0.00

100.00

200.00

1957 1967 1977 1987

Rel

ativ

e A

nom

lay

-200.00

-100.00

0.00

100.00

200.00

1957 1967 1977 1987

Year

D. Striped Bass

-200.00

-100.00

0.00

100.00

200.00

1957 1967 1977 1987

B. Blue Crab

Figure 10. Anomaly plots for commercial landings time series (solid line) and VIMS trawlsurvey data CPUE (dotted line) for four species.


22

Table 2. Classification of common Chesapeake Bay species present in the commercial catchinto taxonomic, trophic, and life history groups. Taxonomic groups were shellfish (S) andfinfish (F), based upon taxonomic order. Trophic criteria were planktivore (P), benthivore (B),piscivore (C) and detritivore (D) based upon the primary dietary items. Life history groupswere resident (R), seasonal (S) and occasional (O). Resident species were those whichremained in the Bay for the entire year, and completed the bulk of their life history within theBay. Blue crab were included in this group even though they do spend a brief period as larvaeoutside of the Bay. Seasonal species were ones that have an obligate estuarine phase, andprolonged periods outside of the Bay. Spot and croaker were included in this group as adultsleave the bay system to spawn. The final group were occasional visitors, which were defined asones that are caught in Bay waters, but are not obligate estuarine species

Common Name Genera Species Fish/Shellfish

TrophicStatus Habit BAY

spawnerWindowpane Flounder Scopthalmus aquosus F B O nSturgeon Acipenser oxyrhyncus F B S yBowfin Amia calva F C R yEels, Common Anguilla rostrata F C S nSummer Flounder Paralichthys dentatus F C S nCrevalle Caranx crysos F C O yCrappie Pomoxis spp. F C O yThread herring Opisthonema oglinum F P O nGizzard Shad Dorosoma cepedianum F P/D R yHickory Shad Alosa mediocris F C S yMenhaden Brevoortia tyrannus F P S nAlewives Alosa psuedoharengus F P S yAm. shad Alosa sapidissima F P S yDolphinfish Coryphaena hipprus F C O nCarp Cyprinus carpio F C/D R yPikes Or Pickerels Esox F C R yRed Hake Urophycis chuss F B O nPigfish Orthopristis chrysoptera F B O nCatfish Various F C R yTautog Tautoga onitis F B O nGar Lepisosteus osseus F C R yTripletail Lobotes surinamensis F C O nSnapper Various F C O nTilefish Lopholatilus chamaeleonticep

sF B O n

Mullet, Striped Mugil cephalus F P O nWhite perch Morone americana F C R yStriped Bass Morone saxatilis F C S yYellow perch Perca flacescens F C R yWinter flounder Psuedopleuronectes americanus F B S y


23

American plaice Hippoglossoides platessoides F B O nBluefish Pomatomus saltatrix F C O nDrum, Black Pogonias cromis F B O yDrum, Red Sciaenops ocellatus F B O nSea Trout, Spotted Cynoscion nebulosus F C O nCroaker,Atantic Micropogonias undulatus F B S nSpot Leiostomus xanthurus F B S nWeakfish Cynoscion regalis F C S nSea Basses Centropristis striata F C O nHogchoker Trinectes maculatus F B R ySheepshead, At Archosargus probatocephalus F B O nSea Robins Prionotus carolinus F B O nClams, Razor Ensis directus S P R yClams,Hard Mercenaria mercenaria S P R yClams,soft Mya arenaria S P R yClams, Surf Spisula solidissima S P O nOyster Crassostrea virginica S P R yHorseshoe Crab Limulus polyphemus S B R yCrab, At, Rock Cancer irroratus S B R yCrabs, Blue Callinectes sapidus S B R yLobster, American Homarus americanus S B O nShellfish, Other S B R yTurtle, Green (Sea) Chelonia mydas R B O nTurtle, Loggerhead (Sea) Caretta caretta R B O nTurtles, Snapper Chelydra serpentina R B R yTurtles, Unc R B R y


24

0

10

20

30

40

50

60

1860 1880 1900 1920 1940 1960 1980 2000

Year

Non

-men

hade

n fin

fish

and

shel

lfish

land

ings

(t x

103 )

Figure 11. Time series of Chesapeake Bay commercial landings by broad taxonomiccategories of finfish (heavy line) or shellfish (light line). Note that data for menhaden havebeen removed from this time series. See Table 2 for taxonomic affiliations of the speciesused in the analysis.


25

catch has declined. Landings of non-menhaden finfish have fallen since the late1940's from almost 60,000 t to less than10,000 t currently (Fig. 11). This is dueprincipally to decline in alewife landings. Incontrast, total shellfish landings, whilevariable, remained relatively constantthrough the 1980's.

Within the broad finfish taxonomicgroup considerable variability exists (Fig.12). For example, while alewife catchesdeclined dramatically in the mid 1970's,catches of the other anadromous shads andriver herrings (blueback herring, Alosaaestivalis, American shad, A. sapidissima,and hickory shad, A. mediocris) declinedmore gradually over the last one hundredyears. Moreover, catches of a freshwaterclupeid, the gizzard shad (Dorosomacepedianum), have increased dramatically inrecent years, although its landings are smallrelative to the historical alosid catches (Fig.12).

Average landings of shellfish haveremained approximately constant since1920, at approximately 40,000 t (Fig. 11). However, the time series of landings ofindividual species in the group differ widely(Fig. 13). It is clear from Figure 13 that theoverall stability of shellfish landings iscaused by a replacement of oyster, whichdeclined dramatically during this century, byhard-shell blue crab in the total shellfishlandings. It is also evident from this figurethat there is coherence between blue craband both hard and soft clam landings. Forexample, the time series of both speciesshow unusually low landings in 1963, 1968and 1978.

We explored patterns in the time seriesof broad trophic groupings. With menhaden removed, the landings of planktivorousspecies declined steadily and quite rapidlyover the period 1960-1992 (Fig. 14). This

reflects the continuing decline ofanadromous clupeoid fishes and oysters. Landings of piscivorous species declinedslowly. However, considering that acomplete fishing moratorium was imposedon striped bass in Maryland from 1985-1989, the overall trend indicates relativestability in levels of total piscivore landings. Yet, the time series of landings of individualpredator species is highly variable,especially for bluefish and weakfish (Fig.15). Peak years of striped bass landings(1950s to 1970s) coincided with a period oflow weakfish and bluefish catches. Moreover, commercial catches of bluefishonly peaked in the mid to late 1970's whenstriped bass catches were declining rapidlyand weakfish catches were at low levels. We do not suggest that there is necessarilyany causal relationship in these patterns,because they are not simple replacements. We do, however, suggest that they may bedriven by the same underlying mechanisms. Shifts in the dominance of piscivore speciesalmost certainly do reflect a combination ofvarying natural abundances and the behaviorof the fishery, which can shift its effort inaccord with both availability of fish andmarketing opportunities.

In contrast to the planktivores andpiscivores, the probable abundances ofbenthivorous fishes, as indexed by landings,have fluctuated widely around a long-termmean, with no clear long-term trend (Fig.14). The fluctuations represent variation inthe harvests of component species, includingspot, croaker and blue crab. With theexception of blue crab, landings of whichhave increased (Fig. 13), due most likely toincreased effort, responses by other speciesof the benthivore group are more similar(Fig. 16). Catches of benthivore species,primarily croaker and spot, peaked in the1940s and 1950s at levels exceeding 30,000


26

t and then declined. Landings of channelcatfish have increased recently, although it isuncertain if this reflects an increase in theirabundance.

The species are grouped by life historyhabits in Figure 17. The dramatic decline incatches of the seasonal, anadromousclupeoid fishes is clearly evident since the1960's. Landings of resident species,primarily blue crab and oyster, haveremained surprisingly stable, centered on along term mean, although they have variedwidely. As previously mentioned, thestability reflects the replacement of oyster byblue crab. Variations in resident specieslandings mostly reflect variability in thelandings of hard-shell blue crab. Landingsof the occasional species, representedprimarily by scombrids have changedrelatively little since 1963.

Overall, there is no direct or clearevidence in these summary plots of dramaticspecies replacements in the Bay, nor ofcomplementary patterns in abundance. However, there is ample evidence ofcomplex patterns of covariation in thelandings time series of groups of species. The presence of covariation and its potentialto affect harvests and overall well being ofresources and fisheries should be consideredby resource managers. To explore furtherthe degree of covariation in time series oflandings, more sophisticated methods arerequired.

2. Multivariate analyses.A principle components analysis (PCA)

was applied to investigate patterns in thelandings of commercial species (Anderson1984). This approach has been usedpreviously for fisheries data sets to suggestsignificant groupings of species (Pepin 1990;Rothschild 1991). PCA is a statistical

technique that uses the correlation in thedata observed for all species to generate asmall number of component groupings thataccount for most of the variability in theoriginal data set. The components areweighted, linear combinations of the originaldata. PCA calculates the correlation orloading of each species on each component. Species with high absolute loadings on anindividual component contribute most to thevariation expressed by that component. Furthermore, each component is associatedwith a percentage of the total variation in thedataset. In this instance, we analyzed thelandings anomalies, i.e. the % deviation ofannual landings from the mean 1930-1992landings. This transformation scales all timeseries equally, removing the effects ofabsolute amounts of fish landed. Consequently, the analysis will identifythose groups that exhibit similar dynamics intheir time series. The results aresummarized in Table 3 and Figure 18.

The first principle component (PCA1)explained 52% of the variation, and wasweighted heavily and positively towardhard-shell blue crabs and negatively towardcroaker (Fig. 18). These two speciesexhibited patterns in their landings timeseries quite unlike any other speciesconsidered. Hard-shell crabs were the onlyspecies in the analysis that exhibited agenerally increasing trend. The croaker timeseries is unique in that, except for a peakperiod in the mid 1940's, croaker landingshave been consistently below the long-termmean. Most of the other species exhibitedloadings close to 0 on PCA1. One group offishes, bluefish, striped bass and whiteperch, that share similar patterns in theirlandings time series, all exhibit similarloadings on


27

0

5

10

15

20

25

30

35

1860 1880 1900 1920 1940 1960 1980 2000

Year

Ale

wife

and

Sha

d L

andi

ngs (

t x103 )

0

0.1

0.2

0.3

0.4

0.5

Gizzard Shad L

andings (t x 10 3)

Figure 12. Time series of Chesapeake Bay commercial landings for alewives (solid line)other Alosa species (dashed line), and gizzard shad (dotted line).


28

0

10

20

30

40

50

60

1860 1880 1900 1920 1940 1960 1980 2000

Year

Cra

b an

d O

yste

r L

andi

ngs (

t x103 )

0

1

2

3

4

Clam

Landings (t x 10 3)

Crab, Blue, SoftCrabs, Blue, HardOystersHard ClamsSoft Clams

Figure 13. Time series of commercial landings for four components of the shellfish harvestfrom Chesapeake Bay (hard blue crab, soft blue crab, oyster, hard clam).


29

0

10

20

30

40

50

60

1960 1965 1970 1975 1980 1985 1990 1995

Year

Non

-men

hade

n la

ndin

gs (t

x10

3 )

Figure 14. Time series of commercial landings by trophic category of planktivore (heavyline), benthivore (dashed line) or piscivore (dotted line) for both finfish and shellfish. Notedata for menhaden have been removed from this time series. See Table 2 for details of


30

0

2

4

6

8

10

12

1860 1880 1900 1920 1940 1960 1980 2000

Year

Wea

kfish

and

Str

iped

Bas

s Lan

ding

s (t x

103 )

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Bluefish L

andings (t x10 3)

Figure 15. Time series of Chesapeake Bay commercial landings for three species within thepiscivore guild, striped bass (heavy solid line), weakfish (solid line), and bluefish (dottedline).


31

0

5

10

15

20

25

30

1860 1880 1900 1920 1940 1960 1980 2000

Year

Spot

and

Cro

aker

Lan

ding

s (t x

103 )

0

1

2

3

4

5

Catfish and C

arp Landigns (t x 10 3)

SpotCatfish,totalCarpCroaker

Figure 16. Time series of Chesapeake Bay commercial landings for four components of thebenthivore guild; croaker, spot, carp, and catfish.


32

0

10

20

30

40

50

60

70

1960 1965 1970 1975 1980 1985 1990 1995

Year

Non

-men

hade

n la

ndin

gs (t

x103 )

Figure 17. Time series of commercial landings by life history category of resident (solidline), seasonal (dashed line), or occasional (dotted line) species for both finfish and shellfish. Note that data for menhaden have been removed from this time series. See Table 2 for detailsof life history affiliations for species used in the analysis.


33

oyst

stbs

w eak

blueyprchw prch

alw v

shadsbsscrabs

spot

crab

crkr

-8

-6

-4

-2

0

2

4

6

8

10

12

-15 -10 -5 0 5 10 15Prin 1

Prin 2

sbss shad

crabsspot

w prchyprch

stbs

blue

oyst

w eak

alw v

crkr

crabs

-6

-4

-2

0

2

4

6

8

10

12

-15 -10 -5 0 5 10 15

Prin 2

Prin 3

Figure 18. Results of Principal Components Analysis of landings anomalies for 13 species inthe Chesapeake Bay commercial fishery during the period 1930-1992. Scatter plots ofspecies loadings on the first three principal components axes. Species abbreviations are:alewife (alwv), bluefish (blue), hard-shell crab blue crab (crab), soft-shell blue crab (crabs),croaker (crkr), oyster (oyst), combined non-alewife alosids (shad), sea bass (sbss), spot(spot), striped bass (stbs), weakfish (weak), white perch (wprch), and yellow perch (yprch).


34

-250.00

-150.00

-50.00

50.00

150.00

250.00

350.00

-100.00 -50.00 0.00 50.00 100.00

Menhaden Anomaly

Ale

wife

Ano

mal

y

-250.00

-200.00

-150.00

-100.00

-50.00

0.00

50.00

100.00

150.00

200.00

250.00

-150.00 -100.00 -50.00 0.00 50.00 100.00

Striped bass anomaly

Whi

te p

erch

ano

mal

y

-250.00

-200.00

-150.00

-100.00

-50.00

0.00

50.00

100.00

150.00

200.00

250.00

-100.00 -50.00 0.00 50.00 100.00

Hard shell blue crab anomaly

Oys

ter a

nom

aly

-250.00

-200.00

-150.00

-100.00

-50.00

0.00

50.00

100.00

150.00

200.00

250.00

-500.00 -300.00 -100.00 100.00 300.00 500.00

Croaker anomaly

Wea

kfis

h an

omal

y

Figure 19. Examples of correlation structure between species pairs identified in the timeseries cross-correlation analysis. Data are plotted as the anomalies from the overall meanlanding for each species.


35

Table 3. Summary of PCA analysis of time series for 13 principal species present incommercial landings from Chesapeake Bay.

A) Explained variation

PCA Axis Eigenvalue % variation explained Cumulative variationexplained

PCA1 33.22 51.9 51.9

PCA2 14.97 23.3 75.3

PCA3 9.44 14.7 90.0

PCA4 2.87 4.4 94.5

PCA5 1.91 2.98 97.5

B) Species Loadings

Species PCA1 PCA2 PCA3

Alewife -2.45 10.65 -3.58

Bluefish 1.75 -1.65 3.32

Soft Shell Crab 0.58 0.45 1.02

Hard Shell Crab 14.48 -1.85 -6.11

Croaker -10.99 -5.95 -5.11

Oyster -5.95 3.37 0.19

Black Sea Bass 0.60 0.17 1.12

Shad -0.25 0.29 1.13

Spot 1.16 -0.11 1.80

Striped Bass 1.93 0.54 2.80

Weakfish -3.69 -3.31 -1.08

White Perch 1.50 -0.65 2.22

Yellow Perch 1.32 -1.04 2.25


36

Tabl

e 4.

Cro

ss-c

orre

latio

n co

effic

ient

s for

land

ing

time

serie

s. D

ata

are

not l

agge

d, a

nd c

ross

-cor

rela

tions

may

be

high

er fo

rla

gged

val

ues.

Val

ues a

ssoc

iate

d w

ith p

roba

bilit

y le

vels

> 0

.05

are

show

n as

ns.

All

othe

r val

ues a

re si

gnifi

cant

at p#

0.0

5.

To il

lust

rate

the

natu

re o

f the

stro

nger

cor

rela

tions

, sel

ecte

d sp

ecie

s pai

rings

are

plo

tted

in F

igur

e 19

.

Ale

wiv

esB

luef

ish

Blu

e C

rab,

Soft

Blu

e C

rab,

Har

dC

roak

erM

enha

den

Oys

ters

Wea

kfis

hSh

adSp

otSt

ripe

dba

ssW

hite

Perc

hY

ello

wPe

rch

Ale

wiv

es1

Blu

efis

h-0

.591

1B

lue

Cra

b, S

oft

0.50

9-0

.466

1B

lue

Cra

b, H

ard

nsns

ns1

Cro

aker

ns-0

.367

0.42

3-0

.605

1M

enha

den

-0.7

060.

595

-0.6

150.

602

-0.6

421

O

yste

rs0.

592

-0.3

60.

497

-0.6

510.

596

-0.7

391

Wea

kfis

hns

ns0.

495

-0.4

850.

879

-0.5

660.

554

1Sh

ad0.

766

-0.5

170.

651

-0.4

620.

452

-0.7

870.

762

0.56

31

Spot

0.41

8-0

.365

nsns

nsns

0.42

3ns

0.36

11

Strip

ed b

ass

0.63

nsns

nsns

-0.3

3ns

ns0.

321

ns1

Whi

te P

erch

0.64

7-0

.38

nsns

nsns

nsns

ns0.

310.

741

Yel

low

Per

ch0.

6-0

.618

0.55

5-0

.32

0.67

7-0

.74

0.51

0.61

50.

651

nsns

ns1


37

PCA1. All three of these species exhibited apeak in landings in the mid to late 70's untilthe early 80's, and then declined rapidlythereafter. No other distinct groupings couldbe identified on PCA1.

The second component (PCA2)explained an additional 23% of the variation(Table 3). Alewife exhibited a high positiveloading on PCA2, whereas croaker exhibiteda strong negative loading on PCA2. Thethird component (PCA3) explained anadditional 15% of the variation in thedataset. A group consisting of bluefish,striped bass, white perch and yellow perchall exhibited positive loadings on PCA3,while croaker and hard shell blue crab hadnegative loadings on this axis. In general,with the exception of the bluefish-stripedbass-white perch-yellow perch complex,PCA did not suggest any strongcomplementary patterns in the landing timeseries.

To explore the correlation patternbetween individual pairs of species we usedtime series techniques to calculate the cross-correlation between pairs over time. Cross-correlation is the time series equivalent ofthe correlation coefficient in linear models. Trends in time series have to be removed

prior to cross-correlation analysis. This wasachieved by using differencing or “pre-whitening.” A summary of the results ispresented in Table 4. The analysis clearlyidentify species whose landings havegenerally declined and those whose landingshave generally increased. For example, thisis particularly clear in the highly significant,negative correlations between alewives andmenhaden (Fig. 19A), and between hard-shell blue crab and oysters (Fig. 19B). However, there are some significantcorrelations among species whose landingsdo not show a clear trend in the time series.For example, the correlations between whiteperch and striped bass (Fig. 19C) andbetween weakfish and croaker (Fig. 19D)probably represent examples of biologicalinteractions. Considering the top piscivores,i.e. bluefish, striped bass, and weakfish,there was no compelling evidence of acorrelation structure that would suggestcompetition for similar prey resources(Table 4). The influence of heavyexploitation and major declines in landingsof several species, and of the long-termincrease in menhaden landings, complicatedthe detection of fine-scale patterns in crosscorrelations of the landings time series.


38

Table 5. Assessment of the presence of factors motivating multispecies approaches in theChesapeake Bay.

Category Motivation Example in Chesapeake Bay

Technical Incidental bycatch of non-targetspecies in targeted fisheries

Migrating alosids in coastalintercept fishery

Use of non-specific fishing gear Poundnet fishery

Exploitation of mixed speciesschools

Gillnet fishery

Biological Extensive predation mortality Striped bass, bluefish andweakfish are all top predators

Competition among species forfood

Control of zooplanktonabundances by gelatinouszooplankton

Similar responses toenvironmental and biologicalforcing

Spot, croaker and crabrecruitment patterns

III. SPECIES INTERACTIONS WITHIN CHESAPEAKE BAY.

Individual species perform only a fewdirect roles in a food web: species may becompetitors, predators or prey. However,this simplified view focuses attention onspecies within particular trophic levels andon those immediately adjacent levels. In thissection, we review current hypotheses oncontrol and regulation of aquatic food webs,including a discussion of complex indirecttrophic effects, to identify those species inChesapeake Bay for which a multispeciesapproach may be warranted. We alsoidentify biological factors, in addition totechnical factors already identified, that havemotivated the development of multispeciesapproaches in other ecosystems and whichare present in the Chesapeake Bay (Table 5).

A. Overview of Food Webs.

Since the turn of the century ecologists havebeen depicting the flow of energy inecosystems with diagrams of food chainsand food webs. For example, in its simplestform a food chain between top predators(highest trophic level) and primaryproducers (lowest trophic level) shows thedependence of top predators on primaryproduction, and explains changes inabundance through predator-preyinteractions (Fig. 20). A classic example inthe Chesapeake Bay is the pelagic foodchain linking striped bass and otherpiscivorous fishes to phytoplankton throughtheir predation on bay anchovy which feed


39

upon copepods, the predators ofphytoplankton (Baird and Ulanowicz 1989).In such chains, species one trophic linkbelow a specified species are its prey, andthose one link above are its predators. Thus,in Figure 20, striped bass and weakfish arepredators on bay anchovy. Species thatshare a common resource occupy the sametrophic level and may affect each other’sabundance through competition for resource. Consequently, striped bass and weakfishalso are potential competitors. Incompetitive interactions, changes in speciesabundances can be attributed to one species’ability to exploit the resource moreefficiently (due to feeding efficiency,reproductive rate, spatial overlap, etc.). Ifthe resource is limited, a decrease in theweaker competitor’s growth rate orabundance is expected. Competition mayoccur over different life stages, adding to asystem’s complexity.

This classical view of trophicinteractions, involving purely predation andcompetition, has been challenged by therecognition of additional classes ofinteractions termed intra-guild predation andindirect effects. A guild is a group ofspecies that exploits the same resource in asimilar way. Therefore, intra-guildpredation is the “the killing and eating ofspecies that use similar, often limiting,resources and are thus potential competitors”(Polis et al. 1989). Such complexinteractions provide benefits to an organismdirectly through predation and indirectly byreducing competition on a similar resource. Most intra-guild predation occurs bygeneralist predators in a size-structuredcommunity. For this reason intra-guildpredation probably is particularly importantin aquatic systems in which size-dependentpredation predominates. An intra-guildpredator is usually larger than its prey and

has a broad diet, which may includejuveniles of its own species. An excellentexample of intra-guild predation in theChesapeake Bay occurs between the twodominant jellyfishes, sea nettle and thectenophore Mnemiopsis leidyi. While boththe sea nettle and ctenophore feed uponzooplankton, the sea nettle also feedsvoraciously upon the ctenophore(Feigenbaum and Kelly 1984; Purcell andCowan 1995: see Fig. 3 in Baird andUlanowicz 1989). The dynamics of thesetwo gelatinous organisms are regulated bytheir competitive abilities and their relativeabundances. Thus, intra-guild predationactually may insure the coexistence of thesecompeting species via the differential effectsof prey switching by the dominant species. A complex pattern of competitiveinteractions among species, such as amonggelatinous zooplankton in Chesapeake Bay,has been used as justification for amultispecies approach in other systems(Table 5).

In traditional models of food webssuch feedback is not considered and foodweb structure is controlled only by theamount of energy entering the system and bythe efficiency of its transfer between trophiclevels (Fig. 20). Indirect effects, which arehighly likely in aquatic ecosystems,complicate the traditional view of control offood web structure in which the controllingprocess alternates between predation andcompetition at successive trophic levels(Hairston et al. 1960). Complex, indirectinteractions allow feedback between non-adjacent trophic levels. For example, onepredator may cause shifts in the relativeabundances of species at lower trophiclevels. In turn, these changes at lowertrophic levels may feed back to alter therelative abundances of predators. A specificexample of the importance of indirect effects


40

in the Chesapeake Bay ecosystem involves


41

Bay anchovy

Zooplankton

Phytoplankton3400

5667

60

652

85

8.1

StripedBass

10

NominalTrophic Level

IVTop predator

III2E Consumer

II1E Consumer

IProducer

Weak-fish

8

30

Bay anchovy

Zooplankton

Phytoplankton3400

5667

60

652

85

8.1

StripedBass

10

StripedBass

10


IVTop predator

III2E Consumer

II1E Consumer

IProducer


IVTop predator

III2E Consumer

II1E Consumer

IProducer

Weak-fish

8

30

Figure 20. An example of a Chesapeake Bay food chain. Values associated with arrows arerates of carbon flow between components (mg.m-2.summer-1) and values in each componentbox are biomasses of carbon (mg.m-2). Data are taken from Baird and Ulanowicz (1989).


42

the high summer rate of predation onzooplankton by bay anchovy, ctenophores,and sea nettles, which precipitates a decreasein the standing stock of zooplankton. However, the three zooplanktivores alsointeract, with the gelatinous zooplanktersfeeding on the egg and larval stages of bayanchovy (Govoni and Olney 1990; Cowanand Houde 1992, 1993; Purcell et al. 1994). These interactions produce complex,dynamic patterns of planktivore abundancesduring the summer. Moreover, the lowzooplankton biomass is coincident with theannual summer peak in phytoplanktonproduction and a peak in bacterial biomassand production (Malone et al. 1986; Harding1994; Shiah and Ducklow 1994). Baird andUlanowicz (1989) suggest that the lowabundance of zooplankton frees thephytoplankton from predator-mediatedcontrol and makes available morephytoplankton production to supportprotozoan and bacterial production. Theyestimate that as much as 50% of netphytoplankton production is channeled andcycled through a microbial loop in thesummer (as compared to 30% in fall, 28% inwinter, and 25% in spring), rather thanchanneled through zooplankton andultimately to fish.

Food web diagrams are useful tools togain an understanding of organisminteractions, but often do not represent thecomplexities of natural ecosystems fully. Food web diagrams generally do not depictthe relative strength of species interactionsor the spatial and temporal components offood webs (Paine 1989). For example,although striped bass in Chesapeake Bayfeed mainly on menhaden and bay anchovyin the summer, their diet shifts to juvenilespot and Atlantic croaker in the winter(Setzler-Hamilton and Hall 1991). Furthermore, during the summer, differences

in the strength of striped bass predation onmenhaden and bay anchovy may be relatedto their spatial overlap in the estuary, thetiming of prey production and abundance, orpredator preference (Hartman and Brandt1995a, b).

Consequences of increased nutrientloading provide another example of thecomplexity of ecological interactions in theChesapeake Bay, which are not accountedfor in food webs. Excess nutrients lead toincreased phytoplankton abundance andepiphytic growth on submerged aquaticvegetation (SAV), causing a drastic declinein SAV acreage during the 1960s and 1970s. This decline directly and indirectlyinfluenced the productivity of manyorganisms that rely on SAV beds for shelter,nursery areas, or on the foraging potentialassociated with greater fish density andorganism diversity in SAV beds (Funderburket al. 1991). Increased nutrient loading alsohas lead to increased duration and extent ofanoxic conditions of water below thepycnocline in the mainstem of the Chesapeake Bay. This causes potentiallycomplex effects upon the biota and speciesinteractions in the Bay (Breitburg 1992;Smith et al. 1992; Breitburg et al. 1994). For example, anoxic waters, combined withtemperature preferences, limit thedistribution of striped bass and its prey, anddecrease habitat area where potential forgrowth is positive (Brandt et al. 1992;Brandt and Kirsch 1993). Yet, food webdiagrams do not account for such temporal,spatial and behavioral complexities, eventhough these features may be importantcontrolling factors (Pimm and Kitching1988).

B. The Chesapeake Bay Food Web.


43

The diverse topography, extendedsalinity gradient, and dynamic tidal regimein the Chesapeake Bay provide a wide rangeof habitats that in turn support diversebiological communities. Because ofseasonal patterns of abundances and theimportance of seasonal migrants, there isextensive coupling between the differentcommunities (Baird and Ulanowicz 1989). Shifts in habitat use by different life stagesof fishes and invertebrates (e.g. striped bassand blue crabs) increase connections andlinkages among habitats and communities. Recently, declines in SAV and oysters, twoimportant components of the Chesapeakeecosystem, have had dramatic impacts. Forexample, at the turn of the century the oysterpopulation in the Bay may have had thepotential to filter more than half the carbonproduction on a daily basis but by the 1980'sthat figure had fallen to less than 1%(Newell 1988). Moreover, the decline of theoyster has lead to the general decline in theoyster reef community which it supported(Kennedy 1991). Despite these declines,Nixon (1988), in a comparison of marine andfreshwater systems, indicated that primaryproduction in the mid-Chesapeake Bay washigh, amounting to 335-780 g C m-2 yr-1. Thus, the Bay remains a highly productive,if altered, ecosystem (Harding 1994). Muchof its productivity, originally associated withbenthic communities, apparently has shiftedto the pelagic zone.

Strong seasonal patterns in migrationand productivity are important to theecology of the Chesapeake Bay. In general,biological activity in Chesapeake Bay risesin spring, peaks during summer, anddeclines through the fall and winter. Thistypical seasonal pattern is driven by nutrientinputs and increasing temperatures thattrigger biological activity in the spring(Harding 1994). Phytoplankton biomass

peaks in spring, although production ratesare highest in summer (Harding 1994). About 35% of net phytoplankton biomass isgrazed during spring and summer (comparedto 10% in fall and 14% in winter). As wehave discussed previously, phytoplanktonare grazed heavily by numerous herbivores. The herbivore community responds to theseasonal pulses in phytoplankton production,and itself exhibits seasonal pulses inabundance and complex internal dynamics.

Gelatinous organisms are believed tobe important consumers in the Bay (Cowanand Houde 1992, 1993; Purcell et al. 1994;Purcell and Cowan 1995). Although presentyear-round, abundance of the ctenophoreMnemiopsis leidyi peaks in the earlysummer and fall. It feeds mainly onzooplankton, along with microzooplanktonand suspended detrital matter. Medusae ofthe sea nettle appear in summer and feed onctenophores, zooplankton, fish eggs and fishlarvae. The sea nettle feeds higher on thefood chain than blue crab, oyster, anchovy,or American shad, and may control recyclingof carbon in the planktonic communityduring summer and early fall (Baird andUlanowicz 1989).

In the mesohaline region of theChesapeake Bay, the biomass andproductivity of suspension-feeding fishpeaks in summer. Seasonally abundantmenhaden are major consumers ofphytoplankton and resident bay anchovy aremajor consumers of zooplankton(Funderburk et al. 1991). In a recent studyof the relative abundance, biomass andproduction of bay anchovy Wang and Houde(1995) suggested that early life stages of thisspecies are major consumers of zooplanktonin Chesapeake Bay. They estimated thatwhen larval and juvenile anchovy areincluded, total annual production of anchovyreaches 233,000 t in the upper to mid-Bay


44

regions. Based upon projections of abioenergetics model, Luo and Brandt (1993)noted that during fall months bay anchovyalone may consume a substantial proportionof the total zooplankton production. Moreover, both striped bass and menhadenare important prey for piscivorous fish.

Benthic communities in ChesapeakeBay also show complex spatial and temporalvariation. The Chesapeake Bay has a welldeveloped benthic community dominated bydeposit feeders such as Nereis succinea,other polychaetes, tellinid bivalves,amphipod crustaceans, and other meiofauna(Baird and Ulanowicz 1989). These depositfeeders utilize a diverse flora of sediment-bound bacteria as a major food source. Several species of clams, including the softclam Mya arenaria, are important filterers ofphytoplankton and zooplankton, and hencelink the benthic and pelagic food chains. They are also an important prey item in bluecrabs and several fishes, furtherstrengthening their role as a bridge betweenpelagic and benthic communities.

As with the pelagic community, themacrobenthic community biomass andproductivity show distinct seasonal patterns,with declines by late fall due to predation byblue crabs and nekton as well as cumulativeeffects of physical stress (low temperaturesand low dissolved oxygen levels). Moreover, the extent to which benthic andpelagic systems exchange energy viapredator-prey interaction varies seasonallyand spatially. Predation on benthicinvertebrates by age 0+ striped bassrepresents one example of coupling. Strongyear classes of striped bass and otherbenthic-feeding fishes may transfer largeamounts of the energy flowing through thebenthic community to the pelagic realm,directly through their consumption ofbenthos and indirectly through predation

upon them by larger pelagic predators. Several environmental factors

influence the complex biotic interactionssummarized above. As the ecotones in theChesapeake Bay are principally defined bysalinity, the extent and timing of springrunoff can affect the spatial extent of theoligo-, meso-, and polyhaline zones. Yearswith above average runoff may becharacterized by more southerlydistributions of many seasonally-residentfish species. Moreover, the intensity andtiming of individual runoff events can beimportant. Secor and Houde (1995) suggestthat recruitment in striped bass may becontrolled by the pattern of runoff. Theextent of anoxia maybe an importantregulator of biological interactions(Breitburg 1990). Coutant and Benson(1990) argued that a spatial “squeeze” maybe suffered during summer by large predatorfish (e.g. striped bass) in the ChesapeakeBay as they try to balance feedingopportunities with thermal optima and avoidanoxia. Kemp and Boynton (1992)documented the importance of oxygendynamics in controlling the extent ofbenthic-pelagic interactions. They discussedboth the direct effect of anoxia on benthicorganisms and the indirect effects of thisreduced productivity on the pelagic system,presumably a result of eutrophication.

C. Multispecies Interactions inChesapeake Bay.

In detailed studies, Hartman andBrandt (1995a, b) examined the seasonal dietpatterns and the predatory demand of stripedbass, bluefish and weakfish. Bay anchovydominate the diets of age 0+ weakfish andbluefish, accounting for up to 90% of thedietary intake of this age class. Olderindividuals of both species utilize bay


45

anchovy less, relying primarily onmenhaden. Spot and croaker may also beimportant components of the diet. Incontrast, the diet of age 0+ striped bass isdominated by invertebrates. But, olderstriped bass shift to a menhaden-dominateddiet, such that menhaden represent fully 70%of the diet of age 6+ striped bass. Thus, withthe exception of age 0+ striped bass, thethree piscivores exploit the same preyspecies. Moreover, in comparing predictedpredatory demand and the concentration ofprey in the field, Based upon bioenergeticsmodels, Hartman and Brandt (1995a, b)concluded that potential demand was oftengreater than supply, with the differencebetween the two increasing over the courseof the year. Yet, detailed information on theactual biomasses of each predator and theextent of spatial overlap is required beforeHartman and Brandt’s (1995a, b)conclusions can be fully substantiated. However, the three top piscivores inChesapeake Bay likely compete for limitedfood resources, particularly as the seasonsprogress. Thus there is a clear potential forcomplex dynamics among the threepredators and two prey species. Theinteractions among these species representan example of a motivation for developingmultispecies fisheries in other systems(Table 5).

We offer the following hypotheticalexample of the potential for multispeciesinteractions in the Chesapeake Bay. In a“normal” year most energy flows into thethree top predators via two routes (Fig.21A). Striped bass rely principally on thephytoplankton - menhaden pathway,whereas weakfish and bluefish rely more onthe bay anchovy pathway. In addition,

weakfish and bluefish depend more uponspot and croaker and so are coupled to thebenthic food chain. However, whenmenhaden abundance is low the patternchanges (Fig. 21B). A lack of menhadenmay cause striped bass diets to shift toinclude more spot and croaker. This shiftwill increase competition with weakfish andbluefish. Hypothetically, in years of lowmenhaden abundance, the reduced grazingpressure on phytoplankton potentially mayincrease zooplankton abundance, leading toenhanced growth and survival of bayanchovy. In turn, increases in bay anchovymay lead to increased production ofweakfish and bluefish, thus furtherincreasing potential competition among thetop predators. Consequently, we suggestthat the major fishery for menhaden inChesapeake Bay, combined with naturalfluctuations menhaden abundance,potentially can alter routes of energy flowamong top predators in Chesapeake Bay andpossibly affect their relative productivities.

Intense seasonal effects and variationin recruitment of predators combine tocomplicate potential interactions. Bluefish,weakfish, and summer flounder, all areresident in the Bay during spring, summerand fall but are mostly absent in winter. Some striped bass are present year-round. Recruitments of bluefish and weakfish are dependent upon processes largely external tothe Bay. Thus, not only are the dynamics oftop predators potentially affected byprocesses in the Bay, but they also areheavily influenced by processes that occuroutside the Bay. Hence, the dynamics ofspecies and interactions among specieswithin the Bay reflect the outcome of


46

0123456

012

012

Bay Anchovy

ZooplanktonMenhaden

Phyotplankton

Spot &Croaker

StripedBass WeakfishBluefish

0123456

012

012

Bay Anchovy

ZooplanktonMenhaden

Phyotplankton

Spot &Croaker


A

B

0123456

012

012

012

012

Bay Anchovy

ZooplanktonMenhaden

Phyotplankton

Spot &CroakerSpot &Croaker


0123456

012

012

012

012

Bay Anchovy

ZooplanktonMenhaden

Phyotplankton

Spot &CroakerSpot &Croaker


A

B

Figure 21. Conceptualexample of complexinteractions in theChesapeake Bay piscivoreguild (striped bass,bluefish, weakfish) in A)normal years whensufficient energy flowsthrough the bay anchovyand menhaden pathwaysand B) years whenmenhaden abundance isreduced leading toincreased competition forspot and croaker by the toppiscivores. Numbers inboxes represent age classesof the predators. Theimportance of theinteraction with respect toenergy flow is indicated bythe weight of the arrow.


47

combined external and internal naturalforces as well as directed fishing effort onpredators and prey.

Although the suite of piscivores havebeen most intensively studied, they are notthe only example of multispeciesinteractions within the Bay’s ecosystem. Wehave already indicated potential feedbackamong the two gelatinous zooplankton andtheir prey. An additional example of aninteracting group is the spot, croaker,flounder and blue crab complex. All fourspecies principally eat clams andpolychaetes. Moreover, each species has anobligate offshore larval stage and all rely onspecific oceanographic conditions totransport larvae back into the Bay. Thus, allfour might be expected to show similartrends in abundance. Counteracting thistendency is the fact the species are alsopotential competitors (Baird and Ulanowicz1989) and a particularly strong year class of

one species, could be expected to negativelyimpact the other three. A recognition thatspecies may exhibit common responses toexternal forcing, such as with the fourspecies complex discussed here has beenused to justify multispecies approaches inother systems (Table 5).

This brief review of the ChesapeakeBay food web has emphasized theimportance of species interactions andindirect trophic effects in complexcommunities. Observed patterns are oftensurprising or counter-intuitive. If themanagement goal of the Chesapeake BayProgram is to insure sustainability of livingresources while maintaining viable fisheries,then an understanding of multispeciesinteractions that control and modify structureand productivity of biological communitiesbecomes an important requirement forsuccessful long-term fisheries management.


48

IV. MULTISPECIES MANAGEMENT: CONCEPTS AND APPROACHES.

We have shown that all of the factorsused to justify multispecies approaches inother systems are present in Chesapeake Bay(Table 5). In this section, we reviewevidence in support of multispeciesmanagement, briefly discuss the principalmethods to achieve it, and point out some ofthe problems associated with its application.

A. The Need for a MultispeciesApproach.

The justification for multispeciesapproaches falls into two broad categories:biological and technical (Table 6).

1. Biological interactions: yield, stabilityand resilience.

A common motivation formultispecies management is that theknowledge that sum of the maximumsustainable yields for individual species isless than the estimated maximum sustainablesurplus production from the community. The principal reason underlying this fact isthat the major predators of fish in the oceanare other fish (Sissenwine 1984, 1986; Bax1991). As the transfer of energy from onetrophic level to the next is not efficient, andon average, only about 8-10% of the energyis transferred, a substantial proportion of theestimated total biological productionsupports the metabolic costs and growth ofother fishes (Pauly and Christensen 1995). By accounting for these interactions in ourefforts to manage fisheries, we may be ableto increase the aggregate catch. Forexample, when we harvest a predator, wepresumably also may be able to harvest thebiomass of its prey that would havesupported the predators removed from thesystem. In an extreme example, if weoverharvest a predator to the point of

population collapse, a very much largertheoretical yield of its prey (perhaps 8 to 10times the maximum sustainable yield of thepredator) could be harvested from theecosystem (May et al. 1979). Single-speciesmanagement ignores the possibility of suchfeedbacks between the species. Thus,although single-species approaches areconservative, and limit the risk of collapse ofindividual species, they are unlikely tooptimize yields of component speciessimultaneously.

Multispecies approaches may alsohelp preserve the stability and resilience of asystem and hence its ability to sustain long-term exploitation. Ideally, an exploitedsystem should exhibit “global stability,”such that it will always tend to return to itspre-exploited state after perturbation byharvests (Beddington 1986). In all exploitedsystems, short-term fluctuations in theabundances of individual species necessarilylead to different rates of exploitation on thecomponent species. But, if the exploitedsystem is globally stable, such annualvariations will be damped out in the longterm to produce predictable and sustainablelong term harvests (Beddington 1986).

However, it is by no means certainthat ecosystems are or remain globally stableeven in an unexploited state (May 1972). For example, between 1940 and 1960 theestimated biomass of sardine (Sardinopssagax caerulea) off California declined bytwo orders of magnitude (MacCall 1990). Simultaneously, the biomass of northernanchovy (Engraulis mordax) increased by asimilar factor. This species replacementdramatically affected the fishing economyalong the California coast (MacCall 1990). It initially was believed that the speciesreplacement was driven by heavy


49

exploitation of the sardine which lead to itspopulation collapse, and that in the absenceof fishing pressure the dominance of sardinewould have continued. However, inanalyses of the deposition rates of fish scalesin anoxic sediments, the two species weredemonstrated to have undergone dramatictransitions such as the one observed in the1950's for at least 2,000 years (Soutar andIsaacs 1974; Baumgartner et al. 1992). Thefluctuation in the 1950's probably was not aresult solely of exploitation, but was morelikely a result of shifts in ocean circulationpatterns, perhaps abetted by heavy fishing onsardine. Other examples include fisheries inEurope and Asia. Historical landings ofherring in Norway have fluctuated widelyprior to the recent periods of extremeexploitation (Daan 1980) indicating naturalvariability. The Japanese sardine (Sardinopssagax melanosticta) yielded more than 1million tonnes annually to a fishery in the1930's before collapsing and essentiallyproviding no yield until its recovery duringthe 1970's and 1980's, when annual catchespeaked at more than 5 million tonnesannually (Kawasaki 1992, Kawasaki andOmori 1995; Watanabe et al. 1996). Furthermore, Crawford (1991) andKawasaki (1992) have shown thatfluctuations of sardine stocks (Sardinopsspp), and possibly associated pelagicspecies, are coherent globally and havehypothesized that major shifts in oceanclimate drive the observed changes incommunity structure and species dominance.

Even were stability the norm,exploitation clearly can shift communitiesaway from stable equilibria. For example,harvests of cod and haddock from the fishinggrounds in the northwest Atlantic onceseemed limitless. However, in the face ofprolonged over-exploitation the groundfish

stocks collapsed (Hutchings and Myers1994; Walters and Maguire 1996). OnGeorge’s Bank the commercially valuablegroundfish species have been replaced by asuite of commercially less desirable species,dominated by skates and dogfish (Rothschild1991). Similar shifts in community structurealso have been reported in the South Atlantic(Gulland and Garcia 1986). In the Gulf ofThailand, small, short-lived fishes andsquids came to dominate the communityafter large piscivore fishes wereoverexploited (Pauly 1988). Managingfisheries in the face of these complexpatterns of species replacements andinteractions clearly requires consideration ofmultispecies approaches.

2. Technical interactions: bycatch.In addition to the biological reasons

for adopting a multispecies approach,technical ones also must be considered(Table 6). The principal motivation forthese models is that even though a singlespecies may be the target of a fishery, otherspecies are inevitably caught in the process. This is termed the bycatch. Pauly (1995)noted that about one third of the annualworld marine landings are discarded asbycatch. Although fishing techniques haveimproved, non-target species inevitably arecaught as bycatch in targeted fisheries(Alverson et al. 1994; Alverson and Hughes1995). In some cases, such as the menhadenpurse-seine fishery in Chesapeake Bay andthe Mid-Atlantic Bight, the bycatch isminimal (Austin et al. 1994). In others, suchas the penaeid shrimp fisheries, it is asubstantial problem (Alverson et al. 1994). Often the species caught as bycatch arejuvenile stages of other commerciallyimportant species. For example, Goodyear(1995) has suggested that the red snapper(Lutjanus campechanus) fishery in the Gulf


50

of Mexico cannot recover from itsoverfished state until a substantial reductionin bycatch of snapper juvenilesby the shrimpfishery is achieved . Thus, increased effortdirected at one species may be detrimental tosustained harvests of other species (Fig. 8). Table 5 presents examples of technicalinteractions that may be important in themanagement of Chesapeake Bay livingresources.

The need to account for significantnon-target landings is obvious. Initialattempts to account for technical interactionsinvolved allocating effort and, hence,partitioning fishing mortality intocomponents for each species (Alverson andHughes 1995). However, multispeciesmanagement may provide better ways toaccount for such technical interactions. Onefurther important difference betweentechnical and biological motivation formultispecies approaches is that data requiredto evaluate and account for technicalinteractions typically is already collectedwhile data on biological interactions are not(Pikitch 1988). Consequently, multispeciesapproaches directed at problems arising fromtechnical interactions may seem moretractable than ones designed to account forbiological interactions.

B. Multispecies Approaches.

Just as the motivation for multispeciesapproaches can be separated into two broadcategories, so too can their application. Concern over both technical interactions andbiological interactions have motivatedmultispecies approaches. Approaches to theformer are similar in structure to single-species approaches. Laurec et al. (1991) andMurawski et al. (1991) provide examples ofmultispecies models involving technicalinteractions. The basic thrust of these

models is to partition allocation of effortonto both target and non-target fisheries. Once achieved, total effort can be regulatedto minimize the impact of the fishery onnon-target organisms. Alternatively, gearcan be modified to reduce the effective efforton non-target species. However, in spite ofcomplexity, multispecies models that dealwith technical interactions are of the samephilosophical lineage as single-speciesapproaches. This is not the case formultispecies models aimed at accounting forbiological interactions. While themotivation for multispecies approaches isoften the same, approaches that emphasizebiological interactions recognize thatharvests of individual species do not occur inan ecological vacuum. Consequently, theapproaches adopted can be widely different(Table 6).

In a review of multispecies analysesfor biological interactions, Kerr and Ryder(1989) suggested that the range ofapproaches adopted can be usefullycategorized into four groups: descriptivemultivariate, dynamical multivariate,multivariate systems, and integral systems(Table 6). We use these categories todocument the range of approaches inmultispecies management. Below wepresent an overview of each category.

1. Descriptive multivariate approachesIn this category Kerr and Ryder

(1989) include those techniques thatprincipally describe the pattern of speciesinteractions observed in empirical datasets. Hence, these approaches are not mechanisticdescriptions of species interactions: rather,they are statistical or graphic description of


51

Tabl

e 6.

Mot

ivat

ions

, adv

anta

ges a

nd d

isad

vant

ages

of t

he p

rinci

pal m

ultis

peci

es a

ppro

ache

s

Obj

ectiv

eC

ateg

ory

Mot

ivat

ion

Adv

anta

geD

isad

vant

age

Exa

mpl

e

Tech

nica

l Int

erac

tion

Effe

cts o

f cat

ch o

fin

cide

ntal

spec

ies i

nta

rget

ed fi

sher

y

1. M

ost e

asily

ada

pted

from

sing

le sp

ecie

s dat

a2.

Pro

duce

man

agem

ent

alte

rnat

ives

1. D

o no

t inc

lude

bio

logi

cal

inte

ract

ions

Laur

ec e

t al.

1991

; Mur

awsk

iet

al.

1991

Bio

logi

cal

Inte

ract

ions

Des

crip

tive

Mul

tivar

iate

Des

crip

tion

of d

ynam

ics

of sp

ecie

s in

mul

tispe

cies

asse

mbl

ages

1. E

mpi

rical

, usi

ng e

xist

ing

data

2. D

iver

sity

of s

impl

eap

proa

ches

3. C

ombi

ning

bio

logi

cal a

nden

viro

nmen

t var

iabl

es

1. L

imite

d fo

reca

stin

g ab

ility

2. D

ata

inte

nsiv

e3.

Not

read

ily a

dapt

able

tom

anag

emen

t

Tyle

r 197

1;Sa

ila a

nd E

rzin

i19

87;

Rot

hsch

ild19

91.

Dyn

amic

alM

ultiv

aria

tePr

edic

tion

of d

ynam

ics o

f m

ultis

peci

es a

ssem

blag

es

1. P

redi

ctiv

e2.

Ada

ptab

le to

man

agem

ent

use

3. E

xplo

ratio

n of

alte

rnat

ive

harv

estin

g st

rate

gies

4. A

sses

smen

t of s

yste

mst

abili

ty

1. D

ata

inte

nsiv

e, o

ften

requ

iring

non

-sta

ndar

d da

ta2.

Pro

paga

tion

of p

aram

eter

erro

r 3.

Cou

nter

-intu

itive

resu

lts4.

Poo

r abi

lity

to in

clud

een

viro

nmen

tal f

orci

ng

Daa

n 19

87;

Chr

iste

nsen

and

Paul

y 19

92

Mul

tivar

iate

Syst

emPr

edic

tion

of a

ggre

grat

epr

oper

ties o

f ass

embl

ages

1. P

redi

ctiv

e2.

Ext

ensi

on o

f exi

stin

gte

chni

ques

3. U

se a

ggre

gate

dat

e

1. In

sens

tive

to d

ynam

ics o

fin

divi

dual

spec

ies,

and

spec

ies

inte

ract

ions

2. L

imite

d ab

ility

to in

clud

een

viro

nmen

tal f

orci

ng

Ral

ston

and

Polo

vina

198

2;Sa

insb

ury

1988

,19

91;

Hig

htow

er19

90,

Bra

nder

197

7,19

91

Inte

gral

Syst

ems

Mec

hani

stic

unde

rsta

ndin

g of

asse

mbl

age

dyna

mic

s

1. D

erriv

ed fr

om fu

ndam

enta

lpr

oper

ties o

f eco

syst

ems

2. F

lexi

ble

1. L

imite

d m

anag

emen

t abi

lity

2. R

equi

re n

on-s

tand

ard

data

3. U

ntes

ted,

em

ergi

ngte

chni

que

Pope

et a

l.19

94; H

eath

1995

.


52

the observed dynamics (Table 6). This isoften the initial effort in the assessment ofwhether multispecies approaches arerequired in a particular system. Ourliterature synthesis has been largelydependent upon descriptive approaches.

Within the overall category, severalanalytical tools have been used. One maysimply plot abundance times series, andvisually inspect for species which showvariations in abundance that are consistentlyin- or out-of-phase. Rothschild et al.’s(1981) summary of trends in the abundanceof commercially important species withinthe Chesapeake Bay is a good example. Rothschild et al. (1981) plot species thatshow similar trends in abundance on thesame graph. Thus, 1) American shad (Alosasapidissima) and river herrings (A. aestivalisand A. pseudoharengus) , 2) menhaden andstriped bass, and 3) weakfish and Atlanticcroaker were plotted on separate panelsbecause each species pair exhibited the sametemporal trend over the time periodoriginally considered. We have updated theRothschild et al. plots to include recent data(Fig. 22). Some trends illustrated byRothschild et al. (1981) have beenmaintained but others have diverged whendata for the period 1980-1992 are includedin the plots. Similar approaches were usedby Tyler (1971) in his analysis of fishcommunities along the eastern seaboard ofNorth America, and by Regier and Hartman(1973) in their analysis of successionalpatterns of species in the Laurentian GreatLakes. These simple techniques still have arole in modern multispecies research. Theyare an essential first step before moredetailed analysis (Shelton 1992).

Statistical analyses approaches alsohave been used. Gabriel and Murawski(1985), Mahon (1985) and MacPherson andGorda (1992) all used cluster analysis to

identify groups of species that exhibit thesame trends in abundance. In these analyses,species that are within the same statisticalcluster exhibit similar dynamics. Speciesfrom adjacent clusters share moresimilarities in their dynamics than do speciesfrom distant clusters. Kerr and Ryder (1989)note that ordination of species within clustergroups may primarily illustrate responses tophysical forcing factors rather than tobiological controlling agents.

Principal components analysis(PCA) has been used to determine patternsof similarity in abundances among species. The traditional approach is to producescatter plots of species loadings against pairsof principal component axes (Koslow 1984;Pepin 1990; Fig. 18, this report). Theseplots are then inspected for groupings;species within the same group exhibitsimilar dynamics. Koslow (1984) used PCAto analyze 14 recruitment time series innorthwest Atlantic fish stocks and concludedthat large-scale physical forcing, rather thanlocal biological interactions, predominatelyregulate recruitment to these fisheries. Pepin(1990) also used PCA to conclude thatrecruitment variability in North Sea fisheswas independent of changes in theabundance of plankton over the long term. In a different approach, Rothschild (1991)plotted the PCA eigenvectors themselves toshow how abundance of George’s Bankstocks had varied over time. He found thatthe temporal pattern of abundance wasdominated by the patterns of variation inthree species: spiny dogfish (Squalusacanthias), winter skate (Raja ocellata) andhaddock (Melanogrammus aeglefinus).

Time series approaches have beenused to analyze patterns in multispeciesinteractions. One of the first applications


53

0

50

100

150

200

250

300

1880 1900 1920 1940 1960 1980 20000

0.5

1

1.5

2

2.5

3

3.5

0

10

20

30

40

50

60

1880 1900 1920 1940 1960 1980 20000

1

2

3

4

5

0

2

4

6

8

10

12

1880 1900 1920 1940 1960 1980 20000

5

10

15

20

25

30

0

5

10

15

20

25

30

35

40

45

1880 1900 1920 1940 1960 1980 20000

1

2

3

4

5

A B

C D

Land

ings

(t x

103)

Figure 22. Landings time series through 1992 for selected species (updated from Rothschild et al.1981). Ordinates on the left side of each panel are associated with the species plotted as a bold line,ordinates on the right side of each panel are associated with the species plotted as a dotted line. Species plotted are A) menhaden (bold) and striped bass (dotted), B) weakfish (bold) and croaker(dotted), C) oyster (bold) and spot (dotted) and D) hard shell blue crabs (bold) and soft clams


54

was Saila and Erzini’s (1987) Markov modelof multispecies systems in the Gulf ofThailand and Gulf of Maine. Saila andErzini (1987) concluded that, under heavyexploitation, a system configurationdominated by squid and other non-valuablefinfish would continue to dominate, and thatthe system is most likely to return to a squid-dominated state after any disturbance.

More sophisticated time-seriesapproaches have been applied recently to themultispecies management issue. Criddle(1991) proposed a state-space, time-seriesapproach to model the dynamics of theNorth Pacific groundfish complex of walleyepollock (Theragra chalcogramma), Pacifichalibut (Hippoglossus stenolepis), Pacificcod (Gadus macrocephalus), yellowfin sole(Limanda aspera) and red king crab(Paralithodes camtschatica). This approachyielded accurate predictions of the short-term dynamics of this system. However, theapplication of multivariate time series is nota well established procedure in ecological orfisheries science, and its uses have beeninfrequent to date (Chatfield 1989). Theprincipal drawback of the approach is that itis very data-intensive. Consequently, longand detailed time series are required todevelop a model, and still retain portions ofthe data for model evaluation.

All of the approaches described aboveare essentially empirical. They have limitedforecasting capabilities, given the potentialfor large environmental, biological oranthropogenic perturbations that can occurannually or on longer time scales. However,these descriptive approaches have yieldedimportant insights into the regulation andstability of multispecies fisheries (Table 6).

2. Dynamical multivariate.The dynamical multivariate approach

includes techniques that extend knowledge

of single-species dynamics to predict themultispecies case (Kerr and Ryder 1989 -Table 6). This approach assumes that thedynamics of multispecies systems arepredictable from a detailed knowledge of thedynamics of system components. Thus,multispecies assemblages are assumed not topossess emergent properties. Within thisbroad area are included both closed-formanalytical approaches (May et al. 1979;Shepherd 1988) and simulation models(Andersen and Ursin 1977).

Multispecies systems that can beanalyzed analytically are, of necessity,relatively simple. Typically such systemsinclude few species with well definedtrophic relationships. May et al. (1979) usedsimple Lotka-Volterra predator-prey modelsto analyze the influence of harvest levels onthe dynamics of the interaction betweenbaleen whales and krill in the SouthernOcean ecosystem. They concluded that themaximum sustainable krill yield can betaken at intermediate krill harvest levels, andhigh levels of whale harvest. Competitionmodels from basic ecological theory alsohave been applied to multispecies fisheries. Silvert and Crawford (1988) developed amodel of two competing species, thatexplained species replacements in the south-east Atlantic when the individual specieswere exploited by separate fleets (i.e. notechnical interaction). In a laterdevelopment, Shelton (1992) developed amodel for two competing species and asingle fleet exploiting both species. BothSilvert and Crawford (1988) and Shelton(1992) conclude that man’s exploitation inmultispecies systems is equivalent to indirectcompetition.

In research more directly related tofisheries management models, Shepherd(1988) examined the dynamics of a systemof single-species, yield-per-recruit models in


55

a mixed species fishery with predator-preyinteractions. He found that, with respect toparameterizing the models, that a lack ofknowledge of the dynamics of themathematical system is a smaller constraintthan the lack of comparable biologicalknowledge. Moreover, increases in thevariety of fleets exploiting the multispeciessystem has more impact on the overallmodel dynamics than does increasing thenumber of interacting species. Interestingly,Shepherd (1988) also concluded that themultispecies dynamics modeled were moresensitive to the parameters of underlyingstock-recruitment relationships than to theeffects of predation.

Attempts have been made to extendRicker’s spawner-recruit approach andBeverton and Holt’s yield-per-recruit modelsto multispecies situations. Ricker (1958)considered the problem of simultaneousexploitation of three stocks in a graphicalanalysis. Based upon spawner-recruitrelationships, he showed that a mixed-stockmaximum sustainable yield (MSY) was onlypossible at levels that would drive one of thestocks to extinction. However, Ricker(1958) assumed that the three stocks musthave the same replacement size. Paulik etal. (1967) extended Ricker’s analysis toconsider the case when the three speciesdiffered in their stock-recruitment functionsand found that the risk of extinction ofindividual species increased when thespecies differed substantially in abundanceor productivity. Murawski (1984) developeda multispecies yield-per-recruit model forthe groundfish complex of George’s Bank. He considered the effect of increasing meshsize on the multispecies yield per annualrecruitment as a function of fishing effort. He concluded that larger mesh sizesproduced larger total yields at highersustained fishing efforts. However, these

models suffer from the same shortcoming asthose noted by Shepherd (1988) - the lack ofknowledge or inclusion of biologicalinteractions leads to uncertainty in accuracyand precision (Table 6).

Network analysis is another form ofdynamical multivariate approach thatquantifies the relationship among elementsin complex food webs. Baird and Ulanowicz(1989) have applied network analysis to theChesapeake Bay. In this approach themultispecies system of interest is representedas a series of components linked by flows ofenergy. Components may be individualspecies, or trophic aggregations (Ulanowicz1996). In general, network analysis mapsthe flow of energy through an ecosystem, byassigning values to 1) the energy inputs, 2)all rates of flow between elements in thesystem, 3) the rates of dissipation of energy,usually through respiration, and 4) anyexports of energy from the system(Ulanowicz and Kay 1991; Christensen andPauly 1992). Application of such modelsyields estimates of productivity andefficiency for individual components andentire systems. One emerging modelingframework for this approach is ECOPATH II(Christensen and Pauly 1992), which hasbeen widely applied in tropical marine andfreshwater systems to estimate total fisheriesyields (Christensen and Pauly 1993).

Another dynamical multispeciesapproach is based upon simulation modelingof trophic interactions. This approach isexemplified by Andersen and Ursin’s (1977)model, which modeled the principalcompetitive and predatory interactions for allof the major species within the North Seaecosystem. However, the complexity ofsuch efforts insures that a full understandingof a system’s dynamics is unlikely. Thus,while such simulation models do permitexploration of the integrated effects of


56

(3)

fishing on the ecosystem, the large numberof unsubstantiated assumptions required area cause for concern. Simulation models donot oten lend themselves to confidentexploration of management strategies(Sissenwine and Daan 1991).

Independently, Pope (1979) andHelgason and Gislason (1979) proposed amodeling approach that incorporated thecomplexity of the fish community, yet wassufficiently simple to give managementagencies a degree of confidence in itspredictions. The approach, multispeciesvirtual population analysis (MSVPA), is anextension of the VPA technique commonlyapplied to single-species systems. Althoughthis new approach was simpler than the fullecosystem model proposed by Andersen andUrsin (1977), it still required voluminousdietary data for the complete range of agesand sizes of principal fish species. Thesedata were provided by a “model-driven”program to collect stomach data from fishesover the widest possible range of species andsites in the North Sea (Daan 1987). Together, the development of the modelingframework and the implementation of thestomach-analysis program provided theimpetus for application of MSVPA to theNorth Sea (Sissenwine and Daan 1991, Pope1991).

Specific applications of MSVPA havebeen reviewed thoroughly by Sparre (1991)and Magnusson (1995). For diverse fishcommunities, MSVPA is very data- andcomputer-intensive, and is often dependentupon initializing assumptions of parameterestimates. Fortunately, experience hasshown that results are largely insensitive tothe initial parameter estimates of relativeabundances and size distributions (Magnusand Magnusson 1987; Rice et al. 1991). Incontrast, the results are very sensitive toassumptions regarding the partitioning of the

total consumption among the suite ofpredators (Helgason and Gislason 1979;Gislason and Helgason 1985; Pope 1979;Sparre 1991), and on the assumption thatpredator ingestion rates can be accuratelypredicted from stomach analysis surveys(Hilden 1988). Furthermore, the temporaland spatial scales over which preyabundance and predator diets are estimatedalso has been shown to be important (Anon.1992). To date, MSVPA models have beendeveloped for several prominent fisheries,including the North Sea (Daan 1987), BalticSea (Sparholt 1991), George’s Bank(Overholtz et al. 1991), and the Benguelaupwelling system (Blinov 1991).

3. Multivariate system.Techniques included in this approach

rely upon aggregating information onindividual species to produce data for fisheryyield, trophic efficiency and productivity(Table 6). Kerr and Ryder (1989) note thatthis approach is exemplified by the FAO’smultispecies, surplus production model(FAO 1978). The multispecies and single-species models do not differ in theirmathematical form. Both are of the form:

Typically the middle three terms of the right-hand-side of the equation are combined torepresent the “surplus production.” Themanagement objective is to adjust catch tomaximize surplus production. In single-species models, each term of the equation isapplied to each of the species. In themultispecies form, the biomasses of allspecies are summed to create an aggregate


57

value for the ecosystem. Similarly,aggregate values for surplus production andcatch also are produced. The analysisproceeds, as in the single species case, byplotting surplus production against stockbiomass and estimating the multispeciesbiomass at which surplus production ismaximized. To be successful at theecosystem level, productivity must be aconservative property, or perhaps theaveraging processes effectively damp outvariability. Brander (1988) argued that theimproved fit of the aggregate model mayresult from the fact that the approach isbased upon long-term steady state conditionsand thus implicitly includes interactionsamong species.

Application of aggregate surplusproduction models to the Gulf of Thailandfishery has been very successful (FAO1978). Brander (1977, 1989) has appliedaggregated, multispecies surplus productionmodels to determine acceptable quotas forthe Irish Sea and Bristol Channel demersalfisheries. Although his model fit the dataextremely well, it was never used in amanagement context. The demersal fisheryoff the northwest coast of Australia has beenmanaged using an aggregate surplusproduction model (Sainsbury 1988, 1991). The model was used to explore the limitingrole of habitat availability. The aggregatedmodel was one of four different multispeciesmodels developed to determine managementstrategies. The other models all requiredmore precise formulations of the nature ofinter-specific interactions. When the modelswere compared over a seven-year period, allmodels produced similar qualitative results. But, the aggregated model indicated thathigher sustainable yields were possible thanthe non-aggregated models.

Hightower (1990) compared the yieldpredictions of single-species models and a

multispecies model based upon aggregatebiomass to develop harvesting policies forthe multispecies rockfish (Sebastes spp)fishery of the northwest Pacific coast ofNorth America. He found that the predictedvariance in yield was lower for multispeciesmodels than for single-species models,suggesting that the multispecies model waseffective in accommodating interannualvariations in relative abundances of the fivetargeted species. A final example of theapplication of an aggregated multispeciesmodel is Ralston and Polovina’s (1982)exploration of the deep-sea longline fisheryoff Hawaii. Cluster analysis of catch-per-unit-effort produced 3 groups of speciescharacterized by different depths of capture. Ralston and Polovina compared thepredictions of models at varying levels ofaggregation. Surplus production modelswere fit to individual species of the threegroups suggested by the cluster analysis. The fit of data to the aggregated speciesmodel was much better for both spatiallyexplicit and spatially aggregated data in eachcase.

4. Integral Systems.A final category of multispecies

models considers the properties of theecosystem to be emergent, i.e. they are notdetermined by the interactions at lowerorganizational levels (Dickie et al. 1987 -Table 6). This assumption may be justifiedas the total yield from a mixed speciesfishery is often relatively conservative,despite the high variability of yields ofindividual species (Regier and Hartman1973; Ryder et al 1974; Sutcliffe et al. 1977;Holden 1978; Pauly and Christensen 1995). Furthermore, Humphreys (1979) detailedanalysis of production and respiration ratessuggests that the ratio of these two isconstant within taxonomic groups. If such


58

predictable ecosystem-level, emergentproperties are common, they could be thebasis of multispecies management strategies. To our knowledge, no direct managementapplication of this approach has beenattempted (Table 6).

Several general, organizing principleshave been suggested as being potentiallyimportant for management, such as theallometry of mortality with respect to size(McGurk 1986), and niche breadth (Pearre1986). Here, we review the potentialapplication of one particular property, thedistribution of organism sizes in aquaticsystems, and its relationship to multispeciesmanagement. Indeed, Dickie et al. (1987)suggest that body-size distributions may be“one of the most sensitive criteria forbiological management effectiveness.” Many biological processes show clearallometric patterns (Peters 1983). However,it was not until Sheldon et al.’s (1972, 1973)analyses that clear allometric patterns weredemonstrated at the ecosystem level. Sheldon et al.’s analysis demonstrated that,in an ecosystem, organism concentrationsdeclined exponentially and predictably asorganism size increased. Not only doabundance patterns demonstrate regularitiesat the ecosystem level, but so too doesbiomass (Sprules et al. 1983, 1991;Schwinghammer 1981, 1985;Schwinghammer et al. 1986). These initialanalyses and ideas were related to potentialfisheries production by Dickie andcolleagues (Dickie et al. 1987; Boudreau andDickie 1989, 1992). However, the size-spectra had considerable variability, makingit uncertain if the spectra were conservative.

The properties invoked by theseecosystem-level models are now beingapplied to specific problems in fisheriesecology (Table 6). For example, Pope et al.(1994) have used size-spectra theory to

explain regularities in spawning time ofmarine fishes. Heath (1995) adopted theseconcepts and applied them to a particularfjord system, Loch Linnhe in Scotland. Byaccounting for net import and export to theAtlantic, Heath could show that themeasured size spectra demonstratedrepeatable patterns in agreement with theSheldon et al. (1972, 1973) originalsuggestions. Investigations of trophicinteractions at the ecosystem level, andapplication of size-spectra analysis andmodeling, presently are the basis of NSF-sponsored research on Chesapeake Bay thatseeks to quantify secondary production andits relationship to both nutrient inputs andthe coupling of biological processes tophysical features (Boynton et al. 1994).

C. Practical Considerations ForMultispecies Approaches.

The intuitive biological appeal ofmultispecies approaches is undeniablyattractive (Gulland et al. 1991). Current,single-species approaches largely fail toincorporate or consider interactions betweenspecies that are important determinants ofstock dynamics. Why have fisherymanagement agencies not moved towardadoption of what Gulland et al. (1991)termed “the Holy Grail of perfect fisherymanagement.” There are several reasons.

Dealing with uncertainty has becomeincreasingly important in fisherymanagement (Hilborn and Walters 1992). The uncertainty inherent in biologicalsystems limits our ability to exploit speciesto their estimated maximal sustainable levels(Larkin 1963). Management already ishampered by a lack of reliable informationand the adoption of multispecies models canonly make the situation more difficult (Table6). As with any mathematical system the


59

more complex a model becomes, the morelikely that input errors propagate within themodel to produce erroneous results. Thus, ingeneral the most parsimonious model ispreferred. As we have seen, while they areintuitively pleasing, all multispeciesapproaches that account for biological andtechnical interactions are more complex thantheir single-species equivalents (Table 6). Consequently, process error is a concern inmultispecies models. Single-species modelsare simply likely to be more robust. Withinmultispecies approaches, those aimed ataccounting for technical interactions aremost similar to single-species models andwe suspect that, at least in the foreseeablefuture, technical interaction models will beadopted most widely.

A second issue that limits theapplicability of multispecies models is theirutility (Brugge and Holden 1991). In thisaspect also, biologically- and technically-motivated models differ substantially. Technical interaction models deal withproblems that managers face directly (e.g.bycatch problem). Thus, the terms of themodel are directly translatable into currentmanagement practices. In sharp contrast,biologically motivated models often forcethe manager to consider new issues. Moreover, these models often predictoutcomes that are beyond the scope of management control. For example, in manyorganisms, abundances are regulated duringearly life stages; yet, managers have nocontrol over survival at this stage.

Another factor which limits the utilityof multispecies models, and also

differentiates them from the single-speciesapproaches, is the time scale at which theyoperate. Single-species models operatebasically at the annual level. Quotas can beand are adjusted annually or even within aseason. Multispecies models incorporateentire life cycles and are inherently multi-year models. It is uncertain if multispeciesmodels would be flexible or robust enoughto account for mid-season corrections ineffort allocation or interannual differences infleet activity. Brugge and Holden (1991)suggested that multispecies models will beused best as exploratory tools which permitmanagers to explore the possible outcomesof alternative strategies prior to applicationof traditional, single-species management.

Finally, multispecies models willonly be adopted if they permit us to addresseffectively fundamental questions. In apessimistic note, Gulland et al. (1991)suggested that multispecies models will beadopted only if they produce managementpredictions in which there are no clearlosers. Certainly, technical interactionmodels may allow us to limit the negativeimpacts of bycatch by regulated effort on thetarget species. However, the advantages ofbiologically-motivated multispecies modelsare less certain. For example, they may wellsuggest that increased harvest of preyspecies is attainable, in the face of reductionin abundance of predators by other fisheries. Many doubt that such fine-scale and directedmanagement is possible in multispeciesfisheries (Gulland et al. 1991) which aresubject to environmental uncertainties andpoorly understood biological interactions.

V. THE POTENTIAL FOR MULTISPECIES APPROACHES IN CHESAPEAKE BAY

The commercial fisheries ofChesapeake Bay, like most regional fisheries

throughout the world, target a relatively fewspecies which, because of their abundance,


60

traditional food value, high economic value,or recreational importance, dominate theoverall catch. Yet, many species are landed,if only in minor quantities. And, somespecies that are not fished, yet are abundantforage species such as bay anchovy, areimportant contributors to production ofeconomically valuable piscivores as well askey elements in the Bay’s food web. Single-species fishery management plans (FMPs) inChesapeake Bay do not consider thebiological or technical interactions thatpotentially can control productivity ofindividual species. In this regard,Chesapeake FMPs are no different thanthose elsewhere in the world. Yet, thepossibility of increasing productivity,optimizing economic return, and insuringsustainability is leading fishery scientistsand managers to consider broad ecosystemapproaches to fisheries management. Ecosystem management is a goal of theChesapeake Bay Program and, as such,requires consideration of multispeciesmanagement of its fisheries.

Menhaden dominate the landingsfrom Chesapeake Bay and all but 2-3% ofmenhaden landings are from purse-seinecatches. Purse seines, which are highlyselective for menhaden (Austin et al. 1994),are restricted in their use to Virginia watersof Chesapeake Bay; consequently, thegeographical area of the Bay’s major (byweight) fishery is limited by regulation. Other gears take more diverse catches. Forexample, NMFS commercial statistics for1992 indicated that 61 species were landedin Bay fisheries. Poundnets accounted for45 species and gillnets for 53 species. Recreational fishermen also account for adiverse catch, although they too target a fewdesirable species. It is clear that the Baysupports a multispecies fishery. However, itis uncertain if multispecies approaches will

improve overall resource management in theshort term because of inadequate knowledgeof biological interactions at the ecosystemlevel. Research aimed at understanding suchinteractions is much desired as the BayProgram moves toward full implementationof an ecosystem approach in living resourcesmanagement.

Do the forces that are leadingtowards multispecies approaches elsewhereexist in the Chesapeake Bay? Several linesof evidence suggest that they do (Table 5). Evidence comes from both technical andbiological interactions. Although the purse-seine menhaden fishery is quite free ofbycatch, it accounts for >65% of the annual,coast wide landings of menhaden(Ahrenholtz 1991; Ahrenholtz et al. 1987). There is concern, although no evidence atthis time, that the fishery competes withpiscivore predators (e.g. bluefish, weakfish,striped bass) for the menhaden resource. Hartman and Brandt (1995a, b) have clearlyshown that menhaden constitute a majorfraction of the diet of piscivores inChesapeake Bay and that the three majorspecies potentially compete for prey. Wehave indicated how these interactionspotentially can have complex and sometimescounter-intuitive effects on overallabundances (Fig. 21). In another example,concern has been expressed recently over thepredatory impact of the recovering stripedbass population on the stock of blue crabs. Available scientific evidence for thisinteraction is weak (Booth and Gary 1993;Goshorn and Casey 1993), but possiblysignificant. Historically, it is certain thatblue crab and striped bass coexisted in highabundances in Chesapeake Bay. However,blue crab catches, fishing effort, and fishingmortality rates have increased markedly overthe past 60 years. It is conceivable thatmortality of prerecruit blue crabs from


61

predation by striped bass and other predatorsnow constitutes a significant control onabundance of fishable crabs. In thisexample, man and striped bass compete for alimited supply of blue crabs, in which the“bycatch” of blue crabs in the striped bassdiet generates the interaction.

The fisheries in the Bay are managedby complex sets of largely independentregulations in individual FMPs. No formalprocedures exist in which the effects ofchanging regulations for one species areconsidered with respect to potential effectson other species. As we pointed out, thisinsufficiency was a motivating factor that isleading towards imposition of multispeciesapproaches in the New England and otherfisheries. In this consideration, preliminarysteps that include workshops to define thespecific issues, research programs to addressthe issues, and development of multispeciesmodels to understand implications arerequired before multispecies managementmight be adopted in the Bay.

What species or species groups are the"strong interactors" and thus likely to becentral to a multispecies approach? While itis unlikely that a complete, predictiveunderstanding of the food-web dynamicswill be achieved for the Chesapeake Bay, itis already apparent that some species aremore likely to interact than others andtherefore may be critical in a multispeciesplan. For example, the suite of toppiscivores clearly is linked by their commonreliance on menhaden and bay anchovy prey. Moreover, blue crab, because of its relianceand probable impact on benthic prey, as wellas its possible control through predation onit by predatory fishes, is a candidate forinclusion in multispecies planning. Thecomplex of alosid species (shads, riverherrings), and other planktivore fishes(menhaden, bay anchovy, gizzard shad) also

are potential strong interactors. Oysters andmenhaden form a species pair that dependsupon filtration of phytoplankton to meetnutritional demands. To what extent dooysters and menhaden substitute for orcomplement each other in the Bay’s trophicstructure and dynamics? Finally, specieswith no apparent direct link to harvestablefisheries, e.g. jellyfishes which are majorconsumers of zooplankton andichthyoplankton, may indirectly limitfisheries productivity and thus are “stronginteractors,” deserving to be included inplanning for management of multispeciesfisheries in Chesapeake Bay.

What multispecies approach mightbe most appropriate, and what advantageswould be gained through its adoption in theChesapeake Bay? As discussed earlier, theintensely seasonal nature of the Bay’s foodweb complicates making a choice. Largeportions of the life cycle of many species inthe Bay’s fisheries are being driven byprocesses external to the Bay. For example,spawning and early life dynamics ofmenhaden, spot, and croaker occur offshoreand away from the influence of the Bay’sestuarine environment. Moreover, age-specific and season-specific migratorybehaviors of many species that resideseasonally in the Bay occur on a coast widescale (e.g. menhaden, bluefish, weakfish,striped bass). This suggests that techniquessuch as the MSVPA will be difficult toapply. Application of such techniques willrequire information from a broad region ofthe mid-Atlantic Bight in addition toinformation from the Bay. We suggest that,initially, descriptive multivariate techniquesand multivariate system approaches may bemost successful (Kerr and Ryder 1989). Supporting this suggestion, much of the dataon commercial species that is required toimplement a multispecies surplus production


62

model already exists. However, thereremain serious deficiencies in the data basethat must be addressed before anymultispecies management approach could beenvisaged.

We identify the following areas asprincipal impediments to implementation ofmultispecies approaches in the ChesapeakeBay, and as areas where research is neededto guide the Bay Program toward decisionsregarding the advisability of multispeciesmanagement.

1 Systematic information on catch andeffort for exploited stocks. Rothschild et al. (1981) previouslyidentified this concern. For many ofthe Bay’s commercial stocks, effortdata for both commercial andrecreational components of the catchare unavailable or of unknownquality. A major assumption in therecent CBSAC stock-assessment ofblue crab was the level of therecreational harvest. Untilcommercial and recreational landingsand associated effort data arecollected consistently, the ability toinfer trends or changes in abundancewill be compromised. Moreover,improvements in this area are requiredeven if multispecies approaches arenot adopted. Such data arefundamental to effective, long-termsingle-species management.

2. Fishery-independent estimates ofabundance for the principal species inthe Chesapeake Bay. A majorproblem in identifying multispeciespatterns in the landings data is thelack of reliable time series ofabundances for many species. Insome cases, landings may be a poorindictor of abundance. It is critical to

obtain Baywide abundanceinformation on key species such asthe heavily exploited menhaden andthe unexploited bay anchovy, whichalmost certainly dominate thenumbers and biomasses of fish in thebay. Some of the shortcomings inabundance estimates are currentlybeing addressed through increaseduse of hydroacoustic assessments inBaywide research programs. It isunlikely that any multispeciesapproach can be fully implementeduntil the abundances of major targetand non-target species are known.

3. Basic life history information. Thereremains a lack of basic knowledgeon the life histories of manyharvested organisms in ChesapeakeBay, which impedes movingconfidently towards multispeciesmanagement. For example, it isdifficult to know if restricting crabpots to water depths less than 40'will conserve female blue crabs, ashas been suggested by some, becausetoo little is known about thebehavior, pathways or cues thatcontrol migrations of blue crabs. Wegenerally know little about causes ofvariability in recruitment andabundances of many species in theBay community. Better informationis required on stage-specificmortality rates and on bioenergeticsrelationships for key fisheriesspecies and for organisms thatsupport productivity of the harvestedspecies. It is especially important togain a better comprehension ofpredator-prey relationships andcompetitive interactions that canexercise control over speciesabundances, productivity, and


63

potential yields.4. Effects of habitat alteration. Loss,

degradation, or alteration of habitatcan lead to changes in communitystructure and productivity. There isan insufficient knowledge of effects ofhabitat changes in Chesapeake Baythrough habitat destruction (e.gdeclines in SAV and oyster reefs) orhabitat deterioration brought about bydeliberate alteration, the presence oftoxicants, or disease organisms. Inthe case of anadromous fishes, theproductivity of this multispeciesgroup may have been permanentlydamaged by dams and blockages onspawning tributaries, although a majoreffort by the Bay Program isunderway to alleviate theseconditions. Understanding the spatialextent of suitable habitats of keyharvested species and of stronglyinteracting species is critical for long-term multispecies management.

5. Lack of detailed understanding of

multispecies models. Much oftraditional single-speciesmanagementrelies on an extensive body ofexperience in applying traditionalmodels to a wide variety of fisheries. Despite the long tradition, a highdegree of uncertainty remains. In thecase of multispecies models,experiences are few and knowledgeoftheir behavior and probability ofsuccess are limited. Consequently,their utility is more uncertain in theeyes of managers than are morefamiliar single-species models. Theuncertainties should not deter effortsto improve and understand themultispecies models. In the longterm, fisheries managementregionallyand globally seems certain to movetoward multispecies approaches thatare compatible with ecosystemmanagement goals.

VI. ACKNOWLEDGMENTS.

The authors thank colleagues at the Maryland Department of Natural Resources, VirginiaInstitute of Marine Science and the National Marine Fisheries Service for access to sources ofdata and reports. We also thank Anne Lange, NOAA/NMFS Chesapeake Bay office forreviewing the synthesis proposal and an early draft of this review. This is contribution number2778 from the Center for Environmental and Estuarine Studies, University of Maryland System.


64

VII. LITERATURE CITED.

Alverson, D. L., M. H. Freeberg, S. A.Murawski, J. G. Pope. 1994. A globalassessment of fisheries bycatch and discards. FAO Paper No. 339, Rome Italy. 233 p.

Alverson , D. L. and Hughes. 1995. Bycatch: from emotion to effective naturalresource management. Natural ResourcesConsultants Inc. Seattle WA. 3 3 p.

Andersen, K. P. and E. Ursin. 1977. Amultispecies extension of the Beverton andHolt theory of fishing; with accounts ofphosphorus, circulation and primaryproduction. Meddr. Damn. Fisk. -og.Havunders. N. S. 7:319-435.

Anderson, T. W. 1984. An Introduction toMultivariate Statistical Analysis. WileyInterscience. Chichester, UK. 675 p.

Anon. 1992. Report of the MultispeciesAssessment Working Group. Copenhagen,16-25 June 1992. ICES C. M.1992/Assess:16 152 p.

Apollonio, S. 1994. The use of ecosystemcharacteristics in fisheries management. Reviews in Fish. Sci. 2:157-180.

Ahrenholtz, D. W. Population biology andlife history of the North Americanmenhadens, Brevoortia spp. Mar. Fish. Rev.53:3-19.

Ahrenholtz, D. W., W. R. Nelson and S. P.Epperly. 1987. Population and fisherycharacteristics of Atlantic menhaden,Brevoortia tyrannus. Fish. Bull. 85:569-600.

Austin, H., J. Kirkley and J. Lucy. 1994.

By-catch and the fishery for Atlanticmenhaden (Brevoortia tyrannus) in the Mid-Atlantic Bight. Virginia Sea Grant MarineResource Advisory no 53. Gloucester Point,VA. 39 p.

Baird, D. and R. E. Ulanowicz. 1989. Theseasonal dynamics of the Chesapeake Bayecosystem. Ecological Monographs59(4):329-364.

Baird, D., R. E. Ulanowicz and W. R.Boynton. 1995. Seasonal Nitrogendynamics in Chesapeake Bay: A networkanalysis. Estuar. Coast. Shelf Sci. 41:137-162.

Bartleson, R. D. and W. M. Kemp. 1990. Preliminary ecosystem simulations ofestuarine planktonic-benthic interactions: theplanktonic submodel. pp 243-252 In J. A.Mihursky and A. Chaney (Eds.) NewPerspectives in the Chesapeake System: AResearch and Management Partnership. Chesapeake Research Consortium.Solomons, MD. 780 p.

Baumgartner, T. R., A. Soutar and V.Ferreira-Bartrina. 1992. Reconstruction ofthe history of Pacific sardine and northernanchovy populations over the last twomillennia from sediments of the SantaBarbara Basin, California. CalCOFIReports. 33: 24-40.

Bax, N. J. 1991. A comparison of the fishbiomass flow to fish, fisheries and mammalsin six marine ecosystems. ICES Mar. Sci.Symp. 193:217-224.

Beddington, J. R. 1986. Shifts in resourcepopulations in large marine ecosystems. pp


65

9-18 In K. Sherman and L. M. Alexander(eds.), Variability and Management of LargeMarine Ecosystems. Am Assoc. Adv. Sci.Sel. Symp. 99.

Beverton. R. J. H. and S. J. Holt. 1957. Onthe dynamics of exploited fish populations. Min. Ag. Fish. Food. (UK) Fisheries Invest.19

Blinov, V. V. 1991. A robust approach tothe analysis of feeding data and an attemptto link it to the Volterra predator-preyanalysis. ICES Mar. Sci. Symp. 193:80-85.

Boicourt, W. C. 1992. Influences ofcirculation processes on dissolved oxygen inthe Chesapeake Bay. pp 7- 60 In: Smith, D.E., M. Leffler and G. Mackeiernan (eds.),Oxygen Dynamics in the Chesapeake Bay. Maryland Sea Grant, College Park, MD. 234 p.

Booth, K. J. and M. L. Gary. 1993. Stripedbass feeding behavior and the potentialeffect on the blue crab population in theChesapeake Bay. Fisheries TechnicalMemorandum Series No. 2. TidewaterAdmin., Maryland Dept. of Nat. Res.,Annapolis. MD. Boudreau, P. R. and L. M. Dickie. 1989. Biological models of fishereis productionbased on physiological and ecologicalscalings of body size. Can J. Fish. Aquat.Sci. 46:614-623.

Boudreau, P. R. and L. M. Dickie. 1992. Biomass spectra of aquatic ecosystems inrelation to fishery yield. Can. J. Fish.Aquat. Sci. 49:1528-1538,

Boynton, W. R., W. C. Boicourt, E. D.Houde, W. M. Kemp and M. R. Roman.

1994. Trophic transfers in large estuarineecosystems: Pulsed inputs, physical featuresand biological patchiness. Proposal toNational Science Foundation, 31 Jan. 1994,[UMCEES]Ref. 94-005.

Boynton, W. R., J. H. Garber, R. Summersand W. M. Kemp. 1995. Inputs,transformations, and transports of Nitrogenand Phosphorus in Chesapeake Bay andselected tributaries. Estuaries 18: 285-314.

Brander, K. 1977. The management of IrishSea fisheries. - a review. Min. Ag. Food &Fish. (UK) Fish Bull. 36. 40 p.

Brander, K. 1988. Multispecies fisheries ofthe Irish Sea. pp 303-328, In Gulland, J. A.(ed.). Fish Population Dynamics (2ndEdition). John Wiley and Sons, London.

Brander, K. 1989. Norway lobsters in theIrish Sea: Modeling one component of amultispecies resource. pp 183-204. In J. F.Caddy (Eds). Marine InvertebrateFisheries: Their Assessment andManagement.

Brandt, S. B. and J. Kirsch. 1993. Spatially-explicit models of striped bassgrowth in the mid-Chesapeake Bay. Trans.Am. Fish. Soc. 122:845-869.

Brandt, S. B., D. M. Mason and E. V.Patrick. 1992. Spatially explicit models offish growth. Fisheries (Bethesda) 17:23-35.

Brandt, S. B., W. R. Boynton, W. M. Kemp,R. Wetzel, R. Bartleson, J. D. Hagy, K. J.Hartman, J. Luo, C. J. Madden, M. Myers,T. Rippetoe, D. D. Fugate and R. Batiuk. 1995. Chesapeake Bay ecosystem modelingprogram 1993-1994 CBP/TRS 133/95. Chesapeake Bay Program, Annapolis


66

Maryland. 150 p.

Breitburg, D. L. 1990. Nearshore hypoxiain the Chesapeake Bay: Patterns andrelationships among physical factors. Estuar. Coast. Shelf Sci. 30(6):593-609.

Breitburg, D. L. 1992. Episodic hypoxia inChesapeake Bay: Interacting effects ofrecruitment, behavior and physicaldisturbance. Ecol. Monogr. 62:525-546.

Breitburg, D. L., N. Steinberg, S. DuBeau,C. Cooksey and E. D. Houde. 1994. Effectsof low dissolved oxygen on predation onestuarine fish larvae. Mar. Ecol. Prog. Ser.104:235-246.

Brugge, W. J. and M. J. Holden. 1991. Multispecies management: a manager’spoint of view. ICES Mar. Sci. Symp. 193:353-358.

Carpenter, S. R., J. F. Kitchell and J. R.Hodgson. 1985. Cascading trophicinteractions and lake productivity. BioScience 35:634-639.

Chatfield, C. 1989. The Analysis of TimeSeries. 4th Ed. Chapman and Hall, London. 241 p.

Chesapeake Bay Program. 1995. Introduction to an Ecosystem. 29 p.

Chesapeake Bay Program. 1996. Chesapeake Bay Blue Crab FisheryManagement Plan (FMP). 61 p.

Chowning, L. S. 1990. Harvesting theChesapeake: Tools and Tradtions. Tidewater Publishers, Centreville, MD. 284p.

Christensen, V. and D. Pauly. 1992. ECOPATH II. A software for balancingsteady-state ecosystem models andcalculating network characteristics. Ecol.Model. 61:169-185.

Christensen, V. and D. Pauly. 1993. Trophic models in aquatic ecosystems. ICLARM, Manila, Phillipines

Coutant, C. C. and D. L. Benson. 1990. Summer habitat suitability for striped bass inChesapeake Bay: Reflections on apopulation decline. Trans. Am. Fish. Soc.119:757-778.

Cowan, J. H., Jr. and E. D. Houde. 1992. Size-dependent predation on marine fishlarvae by ctenophores, scyphomedusae, andplanktivorous fish. Fish. Oceanogr. 1:113-126.

Cowan, J. H., Jr. and E. D. Houde. 1993. Relative predation potentials ofscyphomedusae, ctenophores andplanktivorous fish on ichthyoplankton inChesapeake Bay. Mar. Ecol. Prog. Ser.95:55-65.

Crawford, R. J. M. 1991. Factorsinfluencing population trends of someabundant vertebrates in sardine-rich coastalecosystems. S. Afr. J. mar. Sci. 10:365-381.

Criddle, K. R. 1991. Modelling dynamicsfish populations. NOAA Tech Memo.NOAA-TM-NMFS-F/NWC-197. 82 p.

Daan, N. 1973. A quantitative analysis ofthe food uptake of North Sea cod Gadusmorhua. Neth. J. Sea Res. 6:479-517.

Daan, N. 1975. Consumption andproduction in North Sea cod, Gadus morhua;


67

an assessment of the ecological status of thestock. Neth. J. Sea. Res. 89:24-55.

Daan, N. 1980. A review of replacement ofdepleted stocks by other species and themechanisms underlying such replacements. Rapp. P-V Reun. Cons. Int. Explor. Mer.177:405-421.

Daan, N. 1987. Multispecies versus singlespecies assessment of North Sea fish stocks. Can. J. Fish. Aquat. Sci. 44 (Supple 2):360-370.

Dickie, L. M., S. R. Kerr and P. R.Boudreau. 1987. Size-dependent processesunderlying regularities in ecosystemstrucutre. Ecol. Monogr. 57:233-250.

Fairbanks, W. L. and W. S. Hamil. 1932.The Fisheries of Maryland. MarylandDevelopment Bureau, Baltimore, MD. 198p.

FAO. 1978. Some scientific problems ofmultispecies fisheries. FAO Fish. Tech.Pap. 181:42 p.

Feigenbaum, D. and M. Kelly. 1984. Changes in the lower Chesapeake Bay foodchain in presence of the sea nettle Chrysaoraquinquecirrha (Scyphomedusa). Mar. Ecol.Prog. Ser. 19:39-47.

Funderburk, S. L., S. J. Jordan, J. A.Mihursky and D. Riley (Eds.) 1991. Habitat Requirements for Chesapeake BayLiving Resources. Chesapeake BayProgram, Annnapolis, MD.

Gabreil, W. L. and S. A. Murawski. 1985. The use of cluster analysis in identificationand description of multispecies systems. In:Mahon, R.(ed.), Towards the inclusion of

fishery interactions in management advice. Can. Tech. Rep. Fish. Aquat. Sci. 1347

Gislason, H. And T. Helgason. 1985. Species interactions in assessment of fishstocks with special attention to the NorthSea. Dana 5:1-44.

Goodyear, C. P. 1995. Red snapper in U.S.waters of the Gulf of Mexico. MIA-95/96-05. National Marine Fisheries Service,Southeast Fisheries Center, Miami.

Goshorn, D. and J. Casey. 1993. Anexamination of the relationship betweenstriped bass and blue crabs. FisheriesTechnical Memorandum Series No. 3.Tidewater Admin., Maryland Dept. of Nat.Res., Annapolis. MD.

Govoni, J. J. and J. E. Olney. 1990. Potential predation on fish eggs by the lobatectenophore Mnemiopsis leidyi within andoutside the Chesapeake Bay plume. Fish.Bull., U. S. 89:181-186.

Gulland, J. A. 1965. Estimation of mortalityrates. ICES C.M. 1965 3. 9 p.

Gulland, J. A. 1988. Fish PopulationDynamics. (2nd Edition). WileyInterscience, Chichester. 422 p.

Gulland, J. A., N. Daan and M. P.Sissenwine. 1991. Under what conditionswill multispecies models lead to betterfisheries management. ICES Mar. Sci.Symp. 193:348-352.

Gulland, J. A. And S. Garcia. 1986. Observed patterns in multispecies fisheries. pp. 155-190, In: May, R. M. (Ed). Exploitation of Marine Communities. Springer Verlag, Berlin.


68

Hairston, N. G., F. E. Smith and L. B.Slobodkin. 1960. Community structure,population control, and competition. Am.Nat. 94:421-25.

Harding, L. W., Jr. 1994. Long-term trendsin the distribution of phytoplankton inChesapeake Bay: roles of light, nutrients andstreamflow. Mar. Ecol. Prog. Ser. 104:267-291.

Hartman, K. J. and S. B. Brandt. 1995a. Predatory demand and impact of stripedbass, bluefish and weakfish in theChesapeake Bay: application of abioenergetics model. Can. J. Fish. Aquat.Sci. 52:1667-1687.

Hartman, K. J. and S. B. Brandt. 1995b. Trophic resource partitioning, diets, andgrowth of sympatric estuarine predators. Trans. Am. Fish. Soc. 124:520-537. Heath, M. R. 1995. Size spectrum dynamcisand the planktonic ecosystem of LochLinnhe. ICES J. Mar.Sci. 52:627-642.

Helgason, T. and H. Gislason. 1979. VPA-analysis with species interactions due topredation ICES C.M. 1979: Gadoid FishComm. (3) 9 p.

Hightower, J. E. 1990. Multispeciesharvesting policies for Washington, Oregonand California rockfish trawl fisheries. Fish.Bull. 88:645-656.

Hilborn, R. and C. J. Walters. 1992. Quantitative Fisheries Stock Assessment:Choice, Dynamics and Uncertainty. Chapman Hall, New York , NY. 570p.

Hilden, M. 1988. Significance of thefunctional response to changes in prey

abundance in multispecies virtual populationanalysis. Can. J. Fish. Aquat. Sci. 45:89-106.

Holden, M. J. 1978. Long-term changes inlandings of fish from the North Sea. Rapp.P-v. Reún. du Cons. Internat. Explor Mer. 172:11-26.

Horton, T. and W. M. Eichbaum. 1991. Turning the Tide: Saving the ChesapeakeBay. Island Press. Washington, D.C. 328 p.

Humphreys, W. F. 1979. Production andrespiration in animal populations. J. Anim.Ecol. 48:427-453.

Hutchings, J. A. and R. A. Myers. 1994. What can be learned from the collapse of arenewable resource: Atlantic cod Gadusmorhua off Newfoundland and Labrador. Can. J. Fish. Aquat. Sci. 51: 2126-2146.

Kawasaki, T. 1992. Mechanisms governingfluctuations in pelagic fish populations. S.Afr. J. mar. Sci. 12:873-879.

Kawasaki, T. And M. Omori. 1995. Possible mechanisms underlying fluctuationsin the Far Eastern sardine populationinfereed from time series of two biologicaltraits. Fish. Oceanogr. 4:238-242.

Kemp, W. M. and R. D. Bartleson. 1990. Preliminary ecosystem simulations ofestuarine planktonic-benthic interactions: thebenthic submodel. pp 253-264, In:Mihursky, J. A. and A. Chaney (eds.), NewPerspectives in the Chesapeake System: AResearch and Management Partnership. Chesapeake Research Consoortium.Solomons, MD. 780 p.

Kemp, W. M. and W. R. Boynton. 1992.


69

Benthic -pelagic interactions: Nutrient andoxygen dynamics. pp.149-221, In: Smith, D.E., M. Leffler and G. Mackiernan (eds.),Oxygen Dynamics in the Chesapeake Bay: ASynthesis of Recent Research. Maryland SeaGrant, College Park, MD. 234 p.

Kennedy, V. S. 1991. Eastern Oyster. pp3.1-3.20, In: Funderburk, S. L. et al. (eds.),Habitat Requirements of Chesapeake BayLiving Resources. 2nd Edition. ChesapeakeResearch Consortium. Solomons, MD.

Kerr, S. R. and R. A. Ryder. 1989. Currentapproaches to multispecies analyses ofmarine fisheries. Can. J. Fish. Aquat. Sci.46:528-534.

Koslow, J. A. 1984. Recruitment patternsin Northwest Atlantic fish stocks. Can J.Fish Aquat. Sci. 41:1722-1729.

Larkin, P. A. 1963. Interspecificcompetition and expoitation. J. Fish Res.Bd. Can. 20:647-678.

Laurec, A., A. Biseau and A. Charuau. 1991. Modeling technical interactions. ICES Mar. Sci. Symp. 193:225-236.

Lippson, A. J. and R. L. Lippson. 1984. Life in the Chesapeake Bay. The JohnsHopkins University Press, Balitmore, MD.220p.

Loesch, J. G. and S. M. Atran. 1994. History of Alosa fisheries management:Virginia, a case study. pp 1-6, In: Cooper, J.E., R. T. Eades, R. J. Klauda and J. G.Loesch (eds.). 1994. Anadromous AlosaSymposium. American Fisheries Society,Bethesda, MD. 169 p.

Luo, J. and S. B. Brandt. 1993. Bay

anchovy Anchoa mitchilli production andconsumption in mid-Chesapeake Bay basedon a bioenergetics and acoustic measures offish abundance. Mar. Ecol. Prog. Ser.98:223-236.

MacCall, A. D. 1990. Dynamic Geographyof Marine Fish Populations. University ofWashington Press, Seattle, WA. 153 p.

MacPherson, E. and A. Gorda. 1992. Trends in demersal fish community offNamibia from 1983-1990. S. Afr. J. Mar.Sci. 12:635-650.

Magnus, R. and K. G. Magnusson. 1987. Existence and uniqueness of solution to theMSVPA equations. IMA J. Math. Appl.Med. Biol. 1987:247-263.

Magnusson, K. G. 1995. An overvie of themultispecies VPA- theory and applications. Rev. in Fish Biol. Fisheries 5:195-212.

Mahon, R. 1985. Groundfish assemblageson the Scotian Shelf. pp. 153-162, In:Mahon, R. (ed.), Towards the inclusion offishery interactions in management advice. Can. Tech. Rep. Fish. Aquat. Sci. 1347.

Malone, T. C., W. M. Kemp, H. W.Ducklow, W. R. Boynton, J. H. Tuttle and R.B. Jonas. 1986. Lateral variation in theproduction and fate of phytoplankton in apartially stratified estuary. Mar. Ecol. Prog.Ser. 32:149-160.

May, R. M. 1972. Will a large complexsystem be stable. Nature 238: 413-414.

May, R. M., J. R. Beddington, C. W. Clark,S. J. Holt and R. M. Laws. 1979. Management of multispecies fisheries. Science 205:267-277


70

Murawski, S. A. 1984. Mixed-speciesyield-per-recruitment analyses accountingfor technological interactions. Can. J. Fish.Aquat. Sci. 41(6):897-916.

Murawski, S. A., A. M. Lange and J. S.Idoine. 1991. An analysis of technologicalinteractions among Gulf of Maine mixedspecies fisheries. ICES Mar. Sci. Symp.193:237-252.

McGurk, M. D. 1986. Natural mortality ofmarine pelagic fish eggs and larvae: role ofspatial patchiness. Mar. Ecol. Prog. Ser.34:227-242.

Newell. R. I. E. 1988. Ecological changesin the Chesapeake Bay: are they the resultsof over-harvesting the American oysterCrassostrea virginica? pp. 536-546, In:Understanding the Estuary: Advances inChesapeake Bay Research. CRCPublication 129, Gloucester Point, VA.

Nixon, S. W. 1988. Physical energy inputsand the comparative ecology of lake andmarine ecosystems. Limnol. Oceanogr.33:1005-1025.

Oveholtz, W. J., S. A. Murawski and K. L.Foster. 1991. Impact of predatory fish,marine mammals and seabirds on the pelagicfish ecosystem of the northeastern USA. ICES. Mar. Sci. Symp. 193:198-208.

Paine, R. T. 1989. Food webs: Road mapsof interactions or grist for theoreticaldevelopment. Ecol. 69:1648-1654.

Paulik, G. J. , A. S. Hourston and P. A.Larkin. 1967. Exploitation of multiplestocks by a common fishery. J. Fish Res.Bd. Can. 24:2527-2537.

Pauly, D. 1988. Anecdotes and the shiftingbaseline syndrome of fisheries. TREE10:430.

Pauly, D. 1996. One hundred milliontonnes of fish, and fisheries research. Fish.Res. 25:25-38.

Pauly, D. and V. Christensen. 1995. Primary production required to sustainglobal fisheries. Nature 374:255-257.

Pearre, S., Jr. 1986. Ratio-based nichebreadths of fish, the Sheldon spectrum andthe size-efficiency hypothesis. Mar. Ecol.Prog. Ser. 27:299-314.

Pepin, P. 1990. Biological correlates ofrecruitment variability in North Sea fishstocks. J. Cons Int. Explor. Mer. 47:89-98.

Peters, R. H. 1983. The EcologicalImplications of Body Size. Cambridge Univ.Pres, Cambridge. 329 p.

Pikitch, E. K. 1988. Objective forbiologically and technically interrelatedfisheries. pp. 107-136, In: Wooster, W. S.(ed.), Fisheries Science and Management:Objectives and Limitations . Lecture Noteson Coastal and Estuarine Studies 28:Springer-Verlag, Berlin.

Pimm, S L. and R. L. Kitching. 1989. Foodweb patterns: trivial flaws or the basis of anactive research program? Ecol. 69:1669-1672.

Polis, G. A., C. A. Myers and R. D. Holt. 1989. The ecology and evolution ofintraguild predation: potential competitorsthat eat each other. Ann. Rev. Ecol. & Syst.20:297-330.


71

Pope, J. G. 1972. An investigation of theaccuracy of virtual population analysis. IntComm. Northwest Atl. Fish. Res. Bull. 9:65-74.

Pope. J. G. 1979. A modified cohortanalysis in which constant mortality isreplaced by estimates of predation levels.ICES C.M. 1979/H:16. 8 p.

Pope, J. G. 1991. The ICES multispeciesassessment working group: evolution,insights, and future problems. ICES Mar.Sci. Symp. 193:22-33.

Pope, J. G. and C. T. Macer. 1991. Multispecies virtual populations analysis ofNorth Sea predators for the years 1963-1978. ICES Mar. Sci. Symp. 193:46-49.

Pope, J. G., J. G. Shepherd and J. Webb. 1994. Successful surf-riding on size spectra:the secrets of survival in the sea. Proc. Phil.Roy. Soc. Lond. B. 343:41-49.

Purcell, J. E. and J. H. Cowan, Jr. 1995. Predation by the scyphomedusan Chrysoaraquinquecirrha on Mnemiopsis leidyictenophores. Mar. Ecol. Prog. Ser. 129:63-70.

Purcell, J. E., D. A. Nemazie, S. E. Dorsey,E. D. Houde and J. C. Gamble. 1994. Predation mortality of bay anchovy Anchoamitchilli eggs and larvae due toscyphomedusae and ctenophores inChesapeake Bay. Mar. Ecol. Prog. Ser.114:47-58.

Ralston, S. and J. J. Polovina. 1982. Amultispecies analysis of the commercialdeep-sea handline fishery in Hawaii. Fish.Bull. 80:435-448.

Regier, H. A. and W. L. Hartman. 1973. Lake Erie’s fish community: 150 years ofcultural stresses. Sci. 180:1248-1255.

Rice, J. C., N. Daan, J. G. Pope and H. G. Gislason. 1991. The stability of estimatesof suitabilities in MSVPA over four years ofdata from predator stomachs. ICES Mar.Sci. Symp. 193:34-45.

Richkus, W. A., H. M. Austin and S. J.Nelson. 1992. Fisheries assessment andmanagement synthesis: lessons forChesapeake Bay. pp 75-114, In:Perspectives on the Chesapeake Bay:Advances in Estuarine Studies. ChesapeakeResearch Consortium, Solomons, MD.

Ricker, W. E. 1954. Stock and recruitment. J. Fish. Res. Bd. Can. 11:559-623.

Ricker, W. E. 1958. Maximum sustainedyields from fluctuating environments andmixed stocks. J. Fish. Res. Bd. Can. 15-991-1006.

Riffenburgh. R. H. 1969. A stochasticmodel of interpopulation dynamics in marineecology. J. Fish Res. Bd. Can. 26:2843-2880.

Rothschild, B. J. 1991. Multispeciesinteraction on Georges Bank. ICES Mar.Sci. Symp 193:86-92.

Rothschild, B. J., P. W. Jones and J. S.Wilson. 1981. Trends in Chesapeake BayFisheries. Proc. 46th N. Am. Wildl. Conf.284-298.

Rothschild, B. J., J. S. Ault, P. Goulletquerand M. Heral. 1994. Decline of theChesapeake Bay oyster population: a centuryof habitat destruction and overfishing. Mar.


72

Ecol. Prog. Ser. 111:29-39.

Ryder, R. A., S. R. Kerr, W. W. Taylor andP. A. Larkin. 1974. The morphoedaphicindex, a fish yield estimator - review andevaluation. J. Fish. Res. Bd. Can. 38:1856-1866.

Saila, S. B. and K. Erzini. 1987. Empiricalapproaches to multispecies stock assessment. Trans. Am. Fish. Soc. 116:601-611.

Sainsbury, K. J. 1988. The ecological basisof multispecies fisheries, and management ofa demersal fishery in tropical Australia. pp.349-381, In: Gulland, J. A. (ed.), FishPopulation Dynamcis (2nd Ed). John Wileyand Sons, Chichester. 422 p.

Sainsbury, K. J. 1991. Application of anexperimental approach to management of atropical multispecies fishery with highlyuncertain dynamics. ICES Mar. Sci. Symp.193:301-326.

Schaeffer, M. B. 1954. Some aspects of thedynamics of populations important to themanagement of commercially importantmarine fisheries. Inter-Am. Trop. TunaCom. Bull. 1:27-56.

Schwinghammer, P. 1981. Characteristicsize distributions of integral benthiccommunities. Can J. Fish Aquat. Sci.38:1255-1263.

Schwinghammer. P. 1985. Observations ofsize-structure and pelagic coupling of someshelf and abyssal benthic communities. pp.347-359, In: Gibbs, P. E. (ed.), Proc. 19thEuropean Mar. Biol. Sympo. CambridgeUniv. Press, Cambridge.

Schwinghammer, P., B. Hargrave and C. M.

Hawkins. 1986. Partitioning of productionand respiration among size groups oforganisms in an intertidal benthiccommunity. Mar. Ecol. Prog. Ser. 31: 131-142.

Secor, D. S. 1995. Chesapeake BayAtlantic Sturgeon: Current Status andFuture Recovery. University of Maryland,Chesapeake Biological LaboratoryPublication, Ref. No. [UMCEES]CBL 95-040. 10 p.

Secor, D. S. and E. D. Houde. 1995. Temperature effect on the timing of stripedbass egg production, larval viability andrecruitment potential in the Patuxent River(Chesapeake Bay). Estuaries 18:527-544.

Setzler-Hamilton, E. M. and L. Hall, Jr. 1991. Striped bass, Morone saxatilis. Chapter 13, In: Funderburk, S. L., S. J.Jordan, J. A. Mihursky and D. Riley (eds.), Habitat Requirements for Chesapeake BayLiving Resources. Chesapeake ResearchConsortium, Solomons, MD.

Sheldon, R. W., A. Prakash and W. H.Sutcliffe. 1972. The size distributon ofparticles in the oceans. Limnol Oceanogr.17:327-340.

Sheldon, R. W., W. H. Sutcliffe and A.Prakash. 1973. The production of particlesin surface waters of the ocean with particularreference to the Sargasso Sea. LimnolOceanogr. 18:719-733.

Shelton, P. A. 1992. Detecting andincorportating multispecies effects intofisheries management in the North-west andSouth-east Atlantic. S. Afr. J. Mar. Sci.12:723-737.


73

Shepherd, J. G. 1988. An exploratorymethod for the assessment of multispeciesfisheries. J. Cons. Int Explr. Mer. 44:189-199.

Shiah, F.-K. and H. W. Ducklow. 1994. Temperature and substrate regulation ofbacterial abundance, production and specificgrowth rate in Chesapeake Bay, USA. Mar.Ecol. Prog. Ser. 103:297-308.

Silvert, W. and R. J. M. Crawford. 1989. The periodic replacement of one fish stockby another. p. 161-180 In T. Wyatt and M.G. Larraneta (Eds.) Proc. Int. Symp. onLong Term Changes in Marine FishPopulations. Insitituto de InvestigacionesMarinas de Vigo.

Sissenwine, M. P. 1984. Why do fishpopulations vary? pp. 59-94, In: May, R. M.(ed). Exploitation of Marine Communities. Springer Verlag, Berlin.

Sissenwine, M. P. 1986. Perturbation of apredator controlled continental shelfecosystem., p. 55-86, In: Sherman, K. and L.M. Alexander (eds.), Variability and theManagement of Large Marine Ecosystems. Am Assoc. Adv. Sci. Sel. Symp. 99.

Sissenwine, M. P. and N. Daan. 1991. Anoverview of multispecies models relevant tomanagement of living resources. ICES Mar.Sci. Symp. 193:6-11.

Smith, D. E., M. Leffler and G. Mackiernan(Eds.). 1992. Oxygen Dynamics in theChesapeake Bay. Maryland Sea Grant,College Park, MD 234 p. Soutar, A. and J. D. Isaacs. 1974. Abundance of pelagic fish during the 19thand 20th centuries as recorded in anaerobic

sediments off the Californias. Fish Bull.72:257-274.

Sparholt, H. 1991. Multispecies assessmentof the Baltic fish stocks. ICES Mar. Sci.Symp. 193:64-79.

Sparholt, H. 1994. Fish species interactionsin the Baltic Sea. Dana 10:131-162.

Sparre, P. 1991. Introduction tomultispecies virtual population analysis. ICES Mar Sci. Symp. 193:12-21.

Sprules, W. G., S. B. Brandt, D. J. Stewart,M. Munawar, E. H. Jin and J. Love. 1991. Biomass size spectrum of the Lake Michiganpelagic food web. Can. J. Fish. Aquat. Sci.48:105-115.

Sprules, W. G., J. M. Casselman and B. J.Shuter. 1993. Size distribution of pelagicparticles in lakes. Can. J. Fish. Aquat. Sci.40:1761-1769.

Sutcliffe, W. H., Jr., K. Drinkwater and B. S.Muir. 1977. Correlations of fish catch andenvironmental factors in the Gulf of Maine. J. Fish Res. Bd. Can. 34:19-30.

Tyler, A. V. 1971. Periodic and residentcomponents in communities of Atlanticfishes. J. Fish Res. Bd. Can. 29:935-946

Ulanowicz, R. E. 1996. Ecosystem trophicfoundations: Lindeman Exonerata. pp.549-560, In: Pattern, B. C. and S. E.Jorgensen (eds.), Complex Ecology: thepart-whole relation in ecosystems. PrenticeHall, Englewood Cliffs, NJ.

Ulanowicz, R. E. and J. J. Kay. 1991. Apackage for the analysis of ecosystem flownetworks. Environmental Software 6:131-


74

142.

Walters, C. J. and A. Maguire. 1996. Lessons for stock assessment from thenorthern cod collapse. Rev. Fish. Biol.Fisheries 6:125-137.

Wang, S. B. and E. D. Houde. 1994. Energy storage and dynamics in bayanchovy Anchoa mitchilli. Mar. Biol.121:219-227.

Watanabe, Y. H. Zenitani and R. Kimura. 1996. Causes of population decline in thejapanese sardine (Sardinops melanostictus):

Overfishing or early mortality? pp. 83-94,In: Watanabe, Y., Y. Yamashite and Y.Oozeki (Eds.), Survival Strategies in EarlyLife Stages of Marine Resources. Proc.International Workshop, Yokohama, Japan. 367 p.

Wharton, J. 1957. The Bounty of theChesapeake, Fishing in Colonial Virginia. Univ. Press Virginia. Charlottesville, VA. 79 pp.

White, C. P. 1989. Chesapeake Bay: AField Guide. Tidewater Publishers,Centreville, MD.

chesapeake bay fisheries: prospects for multispecies ... · multispecies fisheries approaches in...

Documents