Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
i
Ref. No. [UMCES]CBL 08-020
DEVELOPMENT AND IMPLEMENTATION OF THE
CHESAPEAKE BAY FISHERY-INDEPENDENT MULTISPECIES SURVEY (CHESFIMS)
Final Report
Submitted to: Mr. Derek Orner
Chair, Chesapeake Bay Fishery Steering Committee NOAA Chesapeake Bay Office 410 Severn Avenue Annapolis, MD 21401
Submitted by Thomas J. Miller, ,J. A. Nye, and D. L. Loewensteiner,
Chesapeake Biological Laboratory, University of Maryland Center for Environmental Science, P. O. Box 38, Solomons, MD 20688;
Sponsor Project Number NA07FU0534
UMCES Project Number 07-5-25412
Submitted on: February 26, 2008
Technical Report Series No TS-545-08- of the University of Maryland Center for Environmental Science
ii
EXECUTIVE SUMMARY Here we report on the results of a large fishery independent survey of fishes in the
Chesapeake Bay supported through a research grant to the University of Maryland Center for
Environmental Science Chesapeake Biological Laboratory from the National Oceanic and
Atmospheric Administration’s Chesapeake Bay Office. Bay wide surveys were conducted three
times a year during April, July and September from 2001-2006. Sampling during the surveys
employed an 18m2 midwater trawl that was deployed in a stepwise fashion at approximately 50
sites throughout the mainstem of the Chesapeake Bay. Additional surveys were conducted in the
Patuxent River in an effort to expand the scope of inference. Data collected by the project were
entered into a data management system and have been made available to the research community
and the general public through a web-based interface at http://hjort.cbl.umces.edu/chesfims.html.
We completed 17 surveys between April 2001 – October 2006. During this period,
baywide average temperatures ranged from 13.41-25.95 oC, baywide salinities varied from10.96-
20.36 and baywide oxygen concentrations from 4.25-9.81 mg.L-1. These levels of variability are
similar, although less extreme than those reported by earlier baywide surveys conducted during
the period 1995-2000. Over the course of the 17 surveys, we occupied 693 stations and collected
520,786 fish weighing 2,205 kg. Only eight deployments (1.15%) resulted in a null catch. The
average 20 minute tow resulted in a catch of 811±1135 (mean ± SD) fish. We collected 110
different species of fish. Analysis of these data indicated clear seasonal and spatial patterns of
variability with temperature, salinity and dissolved oxygen being implicated as important
environmental drivers.
iii
Studies of the ecology of individual species based on specimens conducted during the
baywide surveys were also completed. These studies provide data on species-specific patterns of
abundance, distribution, growth and diet that will be important inputs to efforts to implement
ecosystem-based management within the Chesapeake Bay.
We also conducted extensive statistical analysis of alternative survey designs to estimate
variances for design- and model-based estimation approaches. This information will be useful to
the development of a complemented baywide fishery survey.
This project was a collaboration of scientists from the University of Maryland Center for
Environmental Science, the Department of Animal and Avian Sciences at the University of
Maryland College Park, the Maryland Department of Natural Resources, and Versar Corp. This
report is divided into six individual chapters that each present a different aspect of the project. In
each case, authorship on the work is identified on the chapter title page.
iv
CHAPTER 1 BACKGROUND, MOTIVATION, AND REPORT STRUCTURE……………...1 1.1. BACKGROUND ............................................................................................................ 2 1.2 REPORT STRUCTURE...................................................................................................... 7
1.2.1. Chapter 2 – Data management and Dictionary ....................................................... 7 1.2.2. Chapter 3. Patterns in the distribution and composition of the fish assemblage in the Chesapeake Bay........................................................................................................ 7 1.2.3. Chapter 4. Fishery-independent multispecies assessment of the fish assemblage in the Patuxent River, MD .................................................................................................. 8 1.2.4 Chapter 5. Distribution and diet of Atlantic croaker Micropogonias undulatus in Chesapeake Bay.............................................................................................................. 8 1.2.5 Chapter 6. Design efficiencies of transect and stratified random trawl surveys 9
1.3 PRODUCTS...................................................................................................................... 10 1.3.1 Data Sources ......................................................................................................... 10 1.3.2 Student Dissertations and Theses.......................................................................... 10 1.3.3 Presentations ......................................................................................................... 11
1.4. LITERATURE CITED ..................................................................................................... 13 CHAPTER 2 CHESFIMS DATA MANAGEMENT AND DICTIONARY…………………...16
2.1. ABSTRACT...................................................................................................................... 17 2.2 INTRODUCTION ............................................................................................................. 18 2.3. METHODS ....................................................................................................................... 19
2.3.1. Survey Design............................................................................................................ 19 2.3.2. CTD Profiling ............................................................................................................ 20 2.3.3. Midwater Trawl Deployment..................................................................................... 21 2.3.4. Laboratory Procedures ............................................................................................... 22
2.3.4.1. Ageing................................................................................................................. 23 2.3.4.2. Stomach Contents ............................................................................................... 23 2.3.4.3. Tissue Energy Density .................................................................................. 24 2.3.4.4. RNA: DNA-based condition......................................................................... 24
2.4. DATABASE DESIGN AND MANAGEMENT.............................................................. 25 2.4.1. Database Design......................................................................................................... 25 2.4.2. Data Reporting........................................................................................................... 27 2.4.3. Development of Web Interface.................................................................................. 28
3.4. LITERATURE CITED ..................................................................................................... 30 APPENDIX 2.1......................................................................................................................... 52
CHAPTER 3 PATTERNS IN THE DISTRIBUTION AND COMPOSITION OF THE FISH ASSEMBLAGE IN THE CHESAPEAKE BAY………………………………………………..54
3.1 INTRODUCTION ............................................................................................................. 55 3.2 METHODS ........................................................................................................................ 58
3.2.1. Field ............................................................................................................................ 58 3.2.2. Statistical analysis....................................................................................................... 59
3.3. RESULTS ......................................................................................................................... 62 3.4 DISCUSSION.................................................................................................................... 65 3.5 LITERATURE CITED ...................................................................................................... 68
v
CHAPTER 4. FISHERY-INDEPENDENT MULTISPECIES ASSESSMENT OF THE FISH ASSEMBLAGE IN THE PATUXENT RIVER, MD…………………………………………...86
4.1 INTRODUCTION ............................................................................................................. 87 4.2 METHODS ................................................................................................................... 91
4.2.1. Field Methods ............................................................................................................ 91 4.2.2. Laboratory Methods................................................................................................... 93 4.2.3. Statistical Analysis..................................................................................................... 93
4.2.3.1 Abundance and distribution .................................................................................. 93 4.2.3.2. Size distributions and growth........................................................................ 95
4.3 RESULTS ........................................................................................................................... 96 4.3.1 Abundance and distribution ........................................................................................ 96 4.3.2 Size distributions and growth.................................................................................... 100 4.3.3. Multivariate Analyses .............................................................................................. 102
4.4 DISCUSSION.................................................................................................................. 104 4. 5. LITERATURE CITED ............................................................................................... 109
CHAPTER 5. DISTRIBUTION AND DIET OF ATLANTIC CROAKER MICROPOGONIAS UNDULATUS IN CHESAPEAKE BAY……………………………………………………….133
5.1 INTRODUCTION ........................................................................................................... 134 5. 2. METHODS .................................................................................................................... 137
5.2.1. Data collection ......................................................................................................... 137 5.2.2. Spatial distribution ................................................................................................... 140 5.2.3. Diet analysis............................................................................................................. 141 5.2.4. Effect of environmental variables and diet on croaker presence and abundance .... 142
5.3. RESULTS ....................................................................................................................... 145 5.5 DISCUSSION.................................................................................................................. 151 5.5. LITERATURE CITED ................................................................................................... 183
CHAPTER 6 DESIGN EFFICIENCIES OF TRANSECT AND STRATIFIED RANDOM TRAWL SURVEYS……………………………………………………………………………207
6.1 ABSTRACT..................................................................................................................... 208 6.1 INTRODUCTION ............................................................................................................ 209 6.2. MATERIAL AND METHODS...................................................................................... 210
6.2.1. Study area................................................................................................................. 210 6.2.2. Survey methods........................................................................................................ 211 6.2.3. Estimating mean catch per unit effort ...................................................................... 213 6.2.4. Analytical evaluation of design efficiency .............................................................. 215 6.2.5. Development of a composite estimator..................................................................... 216
6.3 RESULTS ........................................................................................................................ 217 6.4 DISCUSSION AND CONCLUSIONS ...................................................................... 220 6.5. LITERATURE CITED ................................................................................................... 224 Appendix 6.1........................................................................................................................... 240
1
CHAPTER 1
BACKGROUND, MOTIVATION, AND REPORT STRUCTURE
2
1.1. BACKGROUND
Fisheries in Chesapeake Bay contribute significantly to U.S. catches at the national and
regional levels. Recent National Marine Fisheries Service (NMFS) statistics indicate that
between 250,000 - 350,000 metric tonnes (t) of fish and shellfish are harvested annually from
Chesapeake Bay waters, with a dockside value of more than $100 Million (Miller 2006a).
Maintaining the health of these fisheries is an important but difficult task given the considerable
interannual variability in catches of component species (Fig. 1.1). Scientists, managers and the
general public recognize that species targeted by fisheries are components of complex
ecosystems, whereby the removal of individual targeted species may have ramifications for the
entire system (Pikitch et al. 2004). Moreover, changes in ecosystem function or health, may
impact the sustainability and profitability of multiple fisheries simultaneously. Yet, the often
marked differences in the biology, ecology and pattern of exploitation of different species lead to
the development of fisheries management techniques that are based on individual species (Smith
1994). The majority of species subject to fisheries within the Chesapeake Bay are managed on a
single species basis, although increasingly there are efforts to develop multispecies or
ecosystem-based approaches (Miller et al. 1996, Latour et al. 2003, CFEPTAP 2004, Chesapeake
Fishery Ecosystem Plan Technical Advisory Panel 2006).
The potential for biological interactions and technical interactions within traditional
single species management has motivated an interest in multispecies and ecosystem-based
approaches in the Chesapeake region (Miller et al. 1996). Many of the principal approaches to
traditional single species management focus on quantifying maximum sustainable yields so that
fisheries can harvest the surplus production. However, when viewed from an ecosystem point of
3
view, were the surplus production allowed to remain, it would impact other components of the
food web. For example, selective removal of a piscivore may allow increased production of its
principal prey. In contrast, selective removal of a forage species may cause decreased
production of its principal predators. The diversity of fisheries on many fishing grounds means
that both predators and prey are often being targeted simultaneously. Thus, the pattern of
exploitation in one fishery likely impacts the potential harvest in other fisheries as a result of
biological interactions among the individual species. The existence of technical interactions has
been another common concern motivating the development of multispecies fisheries. As a
result of often similar life histories and habitat requirements, many fisheries catch a variety of
species. Often only a few species are the target of the fishery, the other species represent by-
catch. However, in some fisheries, all species are potentially of value and are harvested. In such
fisheries, management on a species by species basis is clearly unsatisfactory.
In 1996, Miller et al. conducted a literature review and synthesis of information to
determine whether the forces that have motivated the development of multispecies approaches in
other regions were present in the Chesapeake Bay. They concluded that both biological and
technical interactions were present in Chesapeake Bay fisheries. Several species harvested from
Chesapeake Bay are either potential competitors (e.g., striped bass Morone saxatilis, bluefish
Pomotomus saltatrix, and the weakfish Cyonscion regalis) or predators and prey (e.g., striped
bass and menhaden Brevoortia tyrannus and Atlantic croaker Micropogonias undulatus), and
thus their fisheries may experience biological interactions. Miller et al. (1996) provide a
hypothetical example in which the linkage between menhaden and croaker is mediated through
predation pressure exerted by striped bass, bluefish and weakfish which varies with hydrography
4
(Fig. 1. 2). Several fisheries, notably the pound net, are inherently multispecies by nature, and
thus technical interactions also clearly exist in Chesapeake Bay (Miller et al. 1996).
Motivated by the Miller et al (1996)white paper, the NOAA Chesapeake Bay Office
funded an international workshop to assess the desirability of implementing multispecies or
ecosystem-based approaches in Chesapeake Bay (Houde et al. 1998). The workshop report
indicates that adoption of multispecies fisheries management would bring the regulation of
fisheries in line with the ecosystem-level focus of the Chesapeake Bay Program. Many
participants felt that explicit recognition of species interactions, by-catch concerns, competing
users, and habitat issues is needed to move fisheries toward a multispecies management ideal
that will be desirable in the future. Several important conclusions were reached during the
workshop. The workshop participants recommended that coordinated, baywide surveys be
performed to estimate key species abundances and that biological data on both economically and
ecologically important species that are currently lacking be obtained (Houde et al. 1998). These
surveys should form a vehicle within which to estimate the temporal and spatial dynamics of key
predator-prey relationships and trophic interactions (Houde et al. 1998). However, at that time
no such surveys existed from which the status and trends of the fish community could be
assessed.
Following the Houde et al report, the NOAA Chesapeake Bay program committed to
supporting research to study the design of Baywide fishery independent surveys. The intention
of the funding was to study alternative survey gear, designs and protocols that would lead
eventually to a complemented Baywide survey supported by the States. One element of this
funding initiative was a detailed review of ongoing fishery-independent surveys (Bonzek et al.
2007). In addition, two baywide trawl surveys were funded as a result of this initiative.
5
Scientists with the Virginia Institute of Marine Sciences developed and implemented a trawl
program focusing on adult sized fishes. The program called CHESMAPP is still underway.
Details of the program can be found at http://www.fisheries.vims.edu/multispecies/. NOAA
funded a second trawl program, the results of which we report on here.
With support from the NOAA Chesapeake Bay Office, we conducted a six year fishery-
independent survey that focused on forage and juvenile fish in the pelagic realm called
CHESFIMS – the Chesapeake Bay Fishery Independent Multispecies Survey. The program
build on from a National Science Foundation funded Land Margin Ecosystem study of trophic
interaction in the Chesapeake that had surveyed the fish community between 1995-2000. Data
from this NSF program revealed the importance of environmental forcing in structuring the
forage and juvenile fish assemblage in the Chesapeake (Jung and Houde 2003). By building
from this platform we hoped to leverage more information on patterns and scales of variability in
the Chesapeake Bay fish assemblage. The CHESFIMS program had four core objectives
Objective 1. Conduct a baywide survey of the bentho-pelagic fish community, focusing on young
(juveniles, and yearling) fishes in the mainstem of Chesapeake Bay.
Objective 2. Conduct pilot surveys of the benth0-pelagic fish community in key tributaries and in
the mainstem to generate sampling statistics that will of use in subsequent design improvements.
Objective 3. Determine trophic interactions among key components of the pelagic fish
community, and examine the implication of the relationships uncovered in empirical studies
using bioenergetic modeling.
6
Objective 4. Conduct statistical analyzes of existing and new data to optimize the complemented
bentho-pelagic survey with respect to consistency and accuracy of key parameters.
7
1.2 REPORT STRUCTURE
The report is divided into five chapters, each one focusing on elements of each core
objective. It is important to note that the information presented in this report is by no means
comprehensive. A considerable amount of analysis and interpretation remains to be done on the
data accumulated. Rather each chapter provides an example of the kind of information and
interpretation that can be gleaned from the CHESFIMS data that increases our understanding of
the fish assemblage in Chesapeake Bay and may guide the design of future fishery-independent
surveys. All chapters are numbered sequentially. Their content is summarized below
1.2.1. Chapter 2 – Data management and Dictionary
One of the central challenges to undertaking a six-year baywide fishery independent
survey is the management and dissemination of the data collected. Each survey generated a huge
amount of primary data, including physical oceanographic conditions and fish catches, and
secondary data, such as gut content information. Ensuring that data are managed to maintain
integrity while at the same time ensuring open access is a central challenge of all large data
programs. We invested considerable effort in developing a data management system that
achieves this goal. Accordingly, this chapter defines the generally sampling and subsequent
sample processing procedures and presents the data management structure that was used in the
CHESFIMS program
1.2.2. Chapter 3. Patterns in the distribution and composition of the fish assemblage in the Chesapeake Bay
8
This chapter presents and summarizes information on the fish assemblage that we
sampled over the six years of the program. Principal findings of the chapter relate to the role of
spatial and temporal variability in structuring the fish community. While the analyses are
preliminary and further work needs to be completed, the results do indicate the opposing
structuring forces of the central up-Bay – down-Bay axis against the seasonal variation in
temperature and salinity in structuring the Chesapeake Bay fish assemblage.
This chapter is authored by Thomas Miller and David Loewensteiner and it is anticipated
that it will be submitted for peer-review in Estuaries and Coasts
1.2.3. Chapter 4. Fishery-independent multispecies assessment of the fish assemblage in the Patuxent River, MD
This chapter presents the results of the attempt to expand the baywide survey into a
principal tributary of the Chesapeake Bay. Using support from the NOAA Chesapeake Bay
Office and the State of Maryland, we conducted an intensive seasonal survey of the Patuxent
River. This chapter presents the results of these samples.
The data presented in this chapter are at a preliminary level of analysis, but it is expected
that they will be submitted for publication at some point in the future.
1.2.4 Chapter 5. Distribution and diet of Atlantic croaker Micropogonias undulatus in
Chesapeake Bay
This chapter presents the results of an analysis of the trophic relationships within the
Chesapeake Bay fish assemblage. Here we focus on Atlantic croaker, although we note that
9
similar analysis have been completed for hogchoker (Curti 2005) and weakfish (Miller and
Loewensteiner in prep). The analyses presented here are simply representative of the class of
analyses that could be conducted. We have selected croaker for inclusion because this study
demonstrates the importance of continuing these “basic” science studies because they do provide
new insight and evidence of change in processes we thought we understood well. This was the
case here where our work on the diets of croaker, a species that has been described as
omnivorous (Murdy et al. 1997), indicates that fully 50% of the diet from an energetic point of
view of croaker in the mid Bay was derived from bay anchovy.
This chapter represents a chapter in the PhD dissertation of Janet A. Nye. It is anticipated
that it will be submitted for publication to Transaction of the American Fisheries Society shortly.
1.2.5 Chapter 6. Design efficiencies of transect and stratified random trawl surveys
This chapter presents the statistical exploration of the survey design used in CHESFIMS
and develops appropriate statistical estimators for complemented survey designs. The
manuscript was submitted to Fishery Bulletin for consideration for publication but was rejected.
It is in revision currently for resubmission to an alternative journal
10
1.3 PRODUCTS
In addition to the chapters described above the CHESFIMS programs has resulted in
several notable products which are summarized here
1.3.1 Data Sources Perhaps most importantly from the communities point of view all CHESFIMS data are
publicly available and accessible through a custom-designed web interface – accessible at
http://hjort.cbl.umces.edu/cfdata.html. This site provides maps of distributions of key species
throughout the program, and more importantly access to the raw data through a simple query
tool. We are not tracking usage but know from email queries that usage is widespread among
agencies, academics and the public.
1.3.2 Student Dissertations and Theses Four students have completed degree based largely or in part on funding through CHESFIMS.
They are
Nye, Janet. A.. Ph.D. Bioenergetic and ecological consequences of diet variability in Atlantic
croaker (Micropogonias undulatus) in Chesapeake Bay. April 2008
Michael G. Frisk. Ph.D. Biology, life history and conservation of elasmobranchs with an
emphasis on western Atlantic skates . September 2004
Kiersten. L. Curti. MS. Biology of hogchoker in Chesapeake Bay. April 2005
Olaf P. Jensen. MS. Spatial ecology of blue crab (Callinectes sapidus) in Chesapeake Bay.
September 2004.
11
1.3.3 Presentations Ostrowski, A. and T. J. Miller. The effects of food availability on the RNA:DNA ratio of
juvenile Atlantic menhaden (Brevoortia tyrannus) in the Patuxent River system.
American Society of Limnology and Oceanography Meeting. Honolulu, HI, February
2006.
Nye, J. A. and T. J. Miller. Evaluation and application of a bioenergetic model of Atlantic
croaker, Micropogonias undulatus, in Chesapeake Bay. American Fisheries Society
Meeting. Anchorage, AK, Sept. 2005.
Nye, J. A., D. L. Loewensteiner and T. J. Miller. Biotic and abiotic factors influencing the
distribution and diet of Atlantic croaker in Chesapeake Bay. American Fisheries Society
Meeting. Anchorage, AK, Sept. 2005.
Peer, A., and T. J. Miller. Incorporating climate-induced correlated recruitments in coupled
age-structured models: a case study in the Chesapeake Bay. American Fisheries Society
Meeting. Anchorage, AK, Sept. 2005.
Bauer, L. and T. J. Miller. Winter mortality of the blue crab (Callinectes sapidus) in the
Chesapeake Bay. American Fisheries Society Meeting. Madison, WI. August 2004.
Curti, K. L. and T. J. Miller. Patterns in the distribution, size and age structure of the hogchoker,
Trinectes maculatus, in the Chesapeake Bay. American Fisheries Society Meeting.
Madison, WI. August 2004.
Nye, J. A. and T. J. Miller. Interspecific interactions between Atlantic croaker (Micropogonias
undulatus) and weakfish (Cynoscion regalis) in Chesapeake Bay. American Fisheries
Society Meeting. Madison, WI. August 2004.
12
Curti, K. L., and T. J. Miller. Patterns in the abundance, size structure and distribution of the
hogchoker, Trinectes maculatus, in the Chesapeake Bay, USA. Annual Meeting of the
Tidewater Section of the American Fisheries Society Meeting. Salisbury, MD. Jan.
2004.
Curti, K. L., T. J. Miller and E. D. Houde. Patterns in the abundance, size structure and
distribution of the hogchoker, Trinectes maculatus, in the Chesapeake Bay, USA.
American Fisheries Society Meeting, Quebec Cité, Quebec, Canada. August 2003
Volstad, J. H., M. C. Christman and T. J. Miller. Comparison And combination of
spatially-overlapping transect and stratified random surveys of finfish abundance in the
Chesapeake Bay. American Fisheries Society Meeting, Quebec Cité, Quebec, Canada.
August 2003.
Jensen, O. P., G. Moglen, M. Christman and T. J. Miller. Application of Geostatistics to for
Estimating Stock Size in Estuaries: A Non-Euclidean Approach to Variogram
Calculation and Kriging. 2nd International Conference on GIS and Fisheries. Brighton,
UK. September 2002.
Christman, M. C., K. Donaldson and T. J. Miller. Spatial distribution of biodiversity of juvenile
finfish. American Fisheries Society, Baltimore, MD August 2002. ,
13
1.4. LITERATURE CITED LITERATURE CITED
Bonzek, C. F., and coauthors. 2007. Baywide and coordinated Chesapeak Bay fish stock
monitoring. Chesapeake Research Consortium, Edgewater, MD. CFEPTAP. 2004. Fisheries ecosystem planning for Chesapeake Bay. Chesapeake Fisheries
Ecosystem Plan Technical Advisory Panel, NOAA Chesapeake Bay Office, Annapolis. Chesapeake Fishery Ecosystem Plan Technical Advisory Panel. 2006. Fishery Ecosystem
Planning For Chesapeake Bay, volume 3. American Fisheries Society, Bethesda, MD. Curti, K. L. 2005. Patterns in the distribution, diet and trophic demand of the hogchoker,
Trinectes maculatus in the Chesapeake Bay, USA. MS. University of Maryland, College Park.
Houde, E. D., M. J. Fogarty, and T. J. Miller. 1998. Prospects for Multispecies Management in Chesapeake Bay: A workshop. Scientific and Technical Committee of the Chesapeake Bay Program, 98-002, Edgwater, MD.
Jung, S., and E. D. Houde. 2003. Spatial and temporal variabilities of pelagic fish community structure and distribution in Chesapeake Bay, USA. Estuarine Coastal and Shelf Science 58(2):335-351.
Latour, R. J., M. J. Brush, and C. F. Bonzek. 2003. Toward ecosystem-based fisheries management: Strategies for multispecies modeling and associated data requirements. Fisheries 28(9):10-22.
Miller, T. J. 2006. Total Removals. Pages 189-212 in C. B. F. E. A. Panel, editor. Fisheries Ecosystem Planning for the Chesapeake Bay, volume 3. American Fisheries Society, Bethesda, MD.
Miller, T. J., E. D. Houde, and E. J. Watkins. 1996. Perspectives on Chesapeake Bay Fisheries: Prospects for multispecies management and sustainability. Scientific & Technical Advisory Committee, Chesapeake Bay Program, Annapolis, MD.
Murdy, E. O., R. S. Birdsong, and J. A. Musick. 1997. Fishes of Chesapeake Bay. Smithsonian Institution Press, Washington.
Pikitch, E. K., and coauthors. 2004. Ecosystem-based fishery management. Science 305(5682):346-347.
Smith, T. D. 1994. Scaling Fisheries, the science of measuring the effects of fishing, 1855- 1955. Cambridge Univ. Press, Cambridge, UK.
14
E) Striped bass
0
1
2
3
4
1860 1880 1900 1920 1940 1960 1980 2000
Year
A) Blue crab
0
20
40
60
1880 1900 1920 1940 1960 1980 2000
B) Menahden
0
100
200
300
400
C) Oyster
0
20
40
60
D) American shad
0
2
4
6
8
Land
ings
(ton
nes
x 10
3 )
Figure 1.1. Time series of Chesapeake Bay catches of selected important species. Data from
Miller et al. 1996
15
0123456
012
012
Bay Anchovy
ZooplanktonMenhaden
Phyotplankton
Spot &Croaker
StripedBass WeakfishBluefish
0123456
012
012
Bay Anchovy
ZooplanktonMenhaden
Phyotplankton
Spot &Croaker
StripedBass WeakfishBluefish
A
B
Figure 1.2. Conceptual example of complex interactions in the Chesapeake Bay piscivore guild
( striped bass, bluefish and weakfish) in A) normal years when sufficient energy flows through
the bay anchovy and menhaden pathways and B) years when menhaden abundance is reduced
leading to increasing competition for spot and croaker by the top piscivores. Numbers in boxes
represent the age classes of the predators. The importance of the interaction with respect to
energy flow is indicated by the weight of the arrow
16
CHAPTER 2
CHESFIMS DATA MANAGEMENT AND DICTIONARY
Thomas J. Miller, David A. Loewensteiner and Michael Kent
Chesapeake Biological Laboratory
University of Maryland Center for Environmental Science
P. O. Box 38
Solomons, MD 20688
17
2.1. ABSTRACT
A central challenge of any large field program is management of the flow of metadata, data and
derived data into the project. Importantly, these aspects of the project must be fully addressed if
the data collected during the research are to be of lasting benefit to the research community and
beyond. Researchers involved in the Chesapeake Bay Fishery Independent Monitoring Survey
(CHESFIMS) were committed to these goals. Here we describe the protocols and procedures
used to select stations visited during the survey, the operation of equipment while the station was
occupied, the preservation techniques of all collected samples, and laboratory techniques used to
process the samples subsequently. A data management and reporting system was developed for
CHESFIMS that provides for easy quality assurance and quality checking. The data
management system also affords easy, reliable and open access to the data to project staff and to
the general public through a web based interface. The structure of the database and analytical
tools used to query the database are described.
18
2.2 INTRODUCTION
Ecosystem-based fisheries management places a high demand on data. For
example Murawski (Murawski 2000) notes that expansion from traditional single species
approaches to ecosystem-based approaches requires not an abandonment of the traditional data
collection systems, but their expansion to include data on a wider array of species, information
on trophic relationships, and habitat relationships. Thus progress toward ecosystem-based
approaches requires analysis of a wider range of data than are typically available (Choi et al.
2005). Such analyses rely on access to data from a wide range of sources.
In 1996, Miller et al. (1996) conducted a literature review and synthesis of information to
determine whether the forces that have motivated the development of multispecies and
ecosystem-based approaches in other regions were present in the Chesapeake Bay. They
concluded that both biological and technical interactions were present in Chesapeake Bay
fisheries. Several species harvested from Chesapeake Bay are either potential competitors (e.g.,
striped bass Morone saxatilis, bluefish Pomotomus saltatrix, and the weakfish Cyonscion
regalis) or predators and prey (e.g., striped bass and menhaden Brevoortia tyrannus and Atlantic
croaker Micropogonias undulatus), and thus their fisheries may experience biological
interactions. Miller et al. (1996) provide a hypothetical example in which the linkage between
menhaden and croaker is mediated through predation pressure exerted by striped bass, bluefish
and weakfish which varies with hydrography. Several fisheries, notably the pound net, are
inherently multispecies by nature, and thus technical interactions also clearly exist in Chesapeake
Bay (Miller et al. 1996).
Houde et al. (1998) reported the recommendations of workshop to explore the utility and
advisability of adopting multispecies approaches in Chesapeake Bay. The workshop report
indicated that adoption of multispecies fisheries management would bring the regulation of
fisheries in line with the ecosystem-level focus of the Chesapeake Bay Program. Many
participants felt that explicit recognition of species interactions, bycatch concerns, competing
users, and habitat issues was needed to move fisheries toward a multispecies management ideal
that will be desirable in the future. Several important conclusions were reached during the
19
workshop. The workshop participants recommended that coordinated, baywide surveys be
performed to estimate key species abundances and that biological data on both economically and
ecologically important species that were currently lacking at time (Houde et al. 1998).
With funding from the NOAA Chesapeake Bay Office, we initiated the Chesapeake Bay
Fishery Independent Multispecies Survey to specifically address the deficiency identified in the
Houde et al workshop. CHESFIMS was designed to extend and complement a baywide
investigation of biological production potential and its temporal and spatial variability that had
been conducted 1995-2000. The TIES (Trophic Interactions in Estuarine Systems) research has
been conducted in a six-year program that includes baywide surveys three times each year
(April, July and October). A broad suite of samples were collected on each cruise including
measures of water quality, primary production and community metabolism, zooplankton
abundances and biomass, gelatinous zooplankton, ichthyoplankton, and juvenile/adult fish.
2.3. METHODS
2.3.1. Survey Design
The CHESFIMS survey is a complemented design involved both fixed and random
stations. The fixed stations were located along fixed transects that had been defined during the
early TIES program (Jung and Houde 2003). The TIES program occupied more stations and
transects than could be occupied during CHESFMS, and accordingly we sampled a subset of
these original stations, selected to ensure a broad spatial coverage throughout the Chesapeake
Bay. In addition to these fixed stations, we also included a similarly number of stratified random
stations selected in proportion to strata area. Individual strata have distinctive characteristics,
and their boundaries broadly corresponding to ecologically relevant salinity regimes and depths
above 5 m. The upper Bay (38 45’ to 39 25’N) is generally shallow, with substantial areas with
depths less than 5 m, and has well mixed waters with high nutrient concentrations. The bottom
topography in the mid Bay includes a narrow channel in the middle of the Bay (37 55’N to 38
45’N) with a stratified water column and broad flanking shoals. This region has relatively clear
waters and experiences seasonally high nutrient concentrations and periods of hypoxia. The
20
lower Bay (37 05’ to 37 55’N) has the clearest waters, greatest depths and lowest nutrient
concentrations (Kemp et al., 1999). The strata volumes are 26,608 km3 (Lower), 16,840 km3
(Mid) and 8,664 km3 (Upper).
CHESFIMS was initiated in 2001, employing the TIES trawling procedures, transect
design and stratification. Sampling was conducted during spring, summer and autumn from the
University of Maryland Center for Estuarine and Environmental Science’s R/V ‘Aquarius. All
data from each cruise were identified to cruise as CFYYSS, where CF is for CHESFIMS, YY is
for year, and MM is a numeric for season. On most cruises we occupied 31 stations allocated
according to the original TIES transect design, and 20 stations allocated according to a stratified
random scheme. The transect stations were fixed for the duration of the CHESFIMS program,
whereas the stratified random stations were selected anew for each cruise. For the transect
design, the im stations within the ith transect were selected by restricted random sampling. Each
transect was divided into im segments of equal size, with one station allocated randomly within
each segment. For the transect surveys, the clustering of stations and variable transect lengths
resulted in heterogeneous selection probabilities for stations within strata. Starting in 2002, the
transect sampling was augmented with an independent, stratified random trawl survey in an
effort to optimize the monitoring design. We were restricted to maintaining the transects used for
monitoring and so chose stratified random sampling as an appropriate additional sampling design
for optimizing the geographic distribution of sampling locations and as a counterpoint to the
transect design. In the stratified random surveys, conducted simultaneously with the transect
surveys, 20 stations covering the entire bay were allocated to the each stratum proportional to
their volumes. Latitude and longitude of stations within strata were randomly generated. Weather
precluded complete sampling of all 20 stations during some surveys. To the extent possible all
activity at each station was standardized. All information obtained during a station was
identified as CFYYMMTTSSS where TT is the transect number (1-30 with RN being used for
Random) and SSS is station number (1-999). An example of the allocation of stations in the
2002 stratified random survey is shown in Figure 2.1.
2.3.2. CTD Profiling
21
Immediately prior to each trawl deployment, a Sea-Bird SBE 25 conductivity,
temperature, depth profiler was used to profile the water column. The dissolved oxygen probe
was recalibrated before each cruise, and all instruments were calibrated by Seabird annually. To
profile the water column the profiler was lowered to the surface and held for 2-5 mins while the
instrument equilibrated. The instrument was then lowered slowly and steadily to the bottom.
Once the bottom was achieved, the instrument was raised approximately 1 m off the bottom and
held for 2-5 minutes at depth to re-equilibrate. Following this second period of re-equilibration,
the profiler was slowly raised to the surface.
Each profile resulted in two data files: a communication file and a hexadecimal raw data
file. The communication file contained all the station specific details for the deployment. The
HEX file contains raw instrument data, often in mV. Following a cruise, the “Seasave” software
(Sea-Bird Electronics Inc, Bellevue, WA) was used to convert the data to provide a downcast
profile in ASCII format for subsequent analysis. Filename standards generated from the CTD
profiling are provided in table 2.1.
2.3.3. Midwater Trawl Deployment
Survey deployments followed the TIES trawling procedures (Jung and Houde, 2003).
We continued to use an 18-m2 midwater trawl (MWT) with 3-mm cod end mesh as the primary
survey gear. MWTs are commonly used to survey pelagic fish including anchovy (Komatsu et
al. 2002), walleye pollock (Wilson 2000), and Pacific whiting (Sakuma and Ralston 1997)and in
commercial fisheries for many of these same species. The MWT employed in the Bay samples
fish 30-256 mm total length of most species effectively, but appears to be less effective for
Atlantic menhaden (Brevoortia tyrannus) of all sizes(Jung and Houde 2003).
We used a standardized 20-minute oblique, stepped tow in all deployments. The trawl
was towed for two minutes in each of ten depth zones distributed throughout the water column
from the surface to the bottom, with minimum trawlable depth being 5 m. The section of the tow
conducted in the deepest zone sampled epibenthic fishes close to or on the bottom. The
22
remaining portion of the tow sampled pelagic and neustonic fishes. All tows were conducted
between 19:00 and 7:00 Eastern Standard Time to minimize gear avoidance and to take
advantage of the reduced patchiness of multiple target species at night. A temperature-depth
mini-logger was attached to the head rope of the trawl for all deployments to record the tow
profile. Data from this minilog was inspected immediately after each haul to determine whether
the haul had largely met survey criteria. From the autumn 2003 survey onwards an additional
minilog was attached to the foot rope of the trawl. By combining data from the top and bottom
minilogs, the net opening could be estimated for the duration of the haul using a custom MatLab
script (Appendix 2.1). Filename standards generated from the minilog data are provided in table
2.1.
Catches were identified, enumerated, measured and weighed onboard. For species for
which there were less than 100 individuals caught in the haul, all individuals were measured to
the nearest mm for total length (TL), and a total weight of the species caught was measured on a
precision spring balance (nearest g). For some species, individuals were of sufficient size that
individual weights could be determined, but this was not common. For abundant species (e.g.,
bay anchovy, white perch etc), a subsampling procedure was used to estimate the catch. In most
cases a random subsample of 100 individuals was measured for TL. This subsample was
weighed, and the total catch of the species was also weighed. The total number in the catch was
then estimated by proportion.
Following complete enumeration of the catch, selected subsamples of target species were
preserved. The subsamples of the majority of species were preserved in ethanol. However, for
some targeted species, and for larger individuals, subsamples were immediately frozen at -5oC.
Information on each haul was initially recorded on log sheets in the field. Immediately
following the cruise, the information was transcribed to an Excel spreadsheet for subsequent
export to the project database.
2.3.4. Laboratory Procedures
23
In the laboratory, individual specimens were removed from ethanol or thawed prior to
further processing. Individual length (TL nearest mm) and weight (W nearest 0.1g) and maturity
status were recorded for these processed fish. Subsequently, the carcasses were processed for
some of the following characteristics: otolith-based ageing, stomach contents, tissue energy
density, RNA:DNA ratio. To relate information gathered on individual fish each fish was given
a unique identification of the form CFYYMMTTSSS-SpCodeNNN, where SpCode is a unique
two letter code for each species, and NNN is a numeric identifier.
2.3.4.1. Ageing
Sagitall otoliths were dissected from thawed carcasses and cleaned in distilled water. The
otolith was weighed, and its maximum dimension recorded. The otolith was affixed to a triple
thickness of glass using thermoplastic cement and sections of the otolith were cut. A double
blade technique was used to make thin sections through the core of the otolith. This approach
requires only one cut for each section and ensures production of uniformly flat sections. Briefly,
two 4” diameter diamond wafering blades were mounted on a low speed wheel saw (South Bay
Technologies, San Clemente, CA). The two blades were separated by an 0.5 mm stainless
spacer. A range of thicknesses of spacers was available, and changed according to the species
being sectioned.
Otolith sections were examined under a dissecting microscope at low magnification and
age was determined. We employed a double blind technique that we have used successfully in
other studies (Frisk and Miller 2006). Briefly, each otolith was read blind by at least two
different readers, and the ages compared. If age estimates differed, a third read was taken. If the
age of this third reader did not agree with one of the previous estimates, the otolith was discarded
as unreadable.
2.3.4.2. Stomach Contents
Whole stomachs were dissected from the carcass. For this application the
stomach was defined as that section between the termination of the esophagus and the pyloric
caecae. The remainder of the alimentary canal was inspected to ensure no undigested food was
24
present. Prior to removal of the stomach contents, full stomachs were blotted dry and weighed to
obtain a full stomach weight (nearest 0.01g). Stomach contents were then removed and the
stomach was subsequently re-weighed to obtain an empty stomach weight (nearest 0.01g). The
difference between these two weights represents an estimate of the total weight of prey in the
stomach. Stomachs were scored for the presence/ absence of food. A feeding incidence of 1.0
indicated the presence of food in the stomach.
Stomach contents were sorted and identified to the lowest practical taxonomic level under
a dissecting microscope. Individual items comprising each prey group were blotted dry and
weighed to obtain an estimate of the total weight of that prey type in the stomach. For larger
prey, length and weight of each individual prey item was also recorded.
2.3.4.3. Tissue Energy Density
For targeted species, tissue energy density measurements were taken. Individual
carcasses, less stomachs and otoliths were ground in a commercial meat grinder. For Atlantic
croaker, liver tissues were processed separately. Homogenized tissue samples were then freeze-
dried at -70C and X Atm for 24 hrs in a 2.5 L capacity freeze drier (Labonco Corp, Kansas City,
MO). Dried samples were ground in a commercial coffee grinder and formed into 0.5 – 1.0 g
pellets. The pellets were burned in an oxygen rich-environment in a bomb calorimeter (Model
6200, Parr Instruments, Moline, IL) to determine the energy content of the tissue.
2.3.4.4. RNA: DNA-based condition
RNA:DNA has been shown to be a reliable indicator of short term growth in juvenile fish (Peck
et al. 2003). We used a fluorometric technique to quantify RNA:DNA levels in samples of axial
muscle (Caldarone et al. 2006). Breifly, each tissue sample was extracted in N-lauroylsarcosine
(final concentration 1%) in Tris-EDTA (TE) buffer (pH 7.5). After diluting and centrifuging, a
portion of the supernatant was combined in a microplate with the fluorophore EB and the total
nucleic-acid fluorescence was recorded with a microplate fluorometer. After adding RNase to
each well, the plate was read a second time and the resulting fluorescence was attributed to
25
DNA. RNA concentrations were calculated from the difference in fluorescence between the first
and second readings (Caldarone et al. 2001).
2.4. DATABASE DESIGN AND MANAGEMENT The CHESFIMS program developed a substantial volume of data. Initially, the program
used the data management tools that had been developed for the TIES program. However, it was
soon apparent that the existing data management scheme was insufficient with respect to data
security and integrity, data quality assurance and did not allow for simultaneous multiuser access
to the most recent data.
Following discussions with CBL IT professionals we contracted with an independent
database consultant to design and implement the CHESFIMS database. Subsequently,
discussions among the consultant, CHESFIMS personnel and the CBL IT group it was decided
to implement the database as a MySQL database on a SUN Solaris Server at CBL. The
advantage of this choice is that routine database management would be handled by the CBL IT
group, who would be responsible for archiving, software updates, and any hardware difficulties.
The CBL IT group also handles user profiles and ensures security, thereby preventing any
unauthorized access. MySQL is an appropriate choice because it can serve as an ODBC server
and the majority of software applications can serve as ODBC clients.
2.4.1. Database Design
A database design that strongly reflected the sampling program was developed. The final
database design is shown in Fig. 2.2. The database design focuses on station events as the
central design element. A unique event number is assigned each time the research vessel stopped
to undertake sampling. Physical data collected during each event are stored in one of three
tables: CTD, MINILOG_TOP and MINILOG_BOTTOM. Each datalogger records its data by
unique scan numbers, thus the combination of event number and scan number provides unique
keys for these tables. A CRUISE table was also established to provide convenient hierarchical
subsetting of the data. Cruise ID served as a unique key for the CRUISE table. Biological data
collected at each event were recorded in separate tables. For each event we recorded the catch
26
summary by species in the CATCHSUMMARY table. For this table event number and species
IDs served as a unique key. To ensure the widest possible use of the data we used ITIS codes as
species ID codes. ITIS, the Integrated Taxonomic Information System, is the result of a
partnership of federal agencies, included NOAA, formed to satisfy their mutual needs for
scientifically credible taxonomic information. The goal of the partnership is to create an easily
accessible database with reliable information on species names and their hierarchical
classification. The database is reviewed periodically to ensure high quality with valid
classifications, revisions, and additions of newly described species. The ITIS includes
documented taxonomic information of flora and fauna from both aquatic and terrestrial habitats.
The taxonomic information on the catch and species in the diet of the catch are recorded in the
SPECIES table. The subset of the catch that was measured onboard the vessel is recorded in the
SIZE table. A dummy specimenID variable, combined with event number and species ID
provides the key for this table. Information from the secondary processing of the sample is
recorded in PROCESSED_FISH and DIET tables. Full details of each table, its data fields,
formats and default values are provided in Tables 2.2-2.12.
Following extensive prototyping which involved running frequent dummy queries, the
consultant delivered the final database design in May 2006. The initial prototyping was all
conducted in MS Access. Subsequently, CHESFIMS staff input all CHESFIMS data into the
database in MS Access for the ease that its visual interface afforded. This exercise provided
ample opportunity for quality assurance / control. Subsequently, data were exported to populate
the final MySQL database structure.
We note that this database has now become the default data management tool for all field
collections undertake in the Miller Lab at CBL. Thus, the database continues to expand, now
holding data from a wide variety of programs including a parallel midwater trawl survey in the
Patuxent River in 2004 that was funded by Maryland Department of Natural Resources, an
icthyoplankton survey of the Patuxent River conducted in 2006 and 2007, seine survey
collections for striped bass and menhaden conducted in Maryland waters of Chesapeake Bay in
2006 and 2007.
27
2.4.2. Data Reporting
Protocols for data access were checked for all principal software clients that were likely
to be used by CHESFIMS staff. We verified data access in MS Access, MS Excel, SAS, R, and
ArcView. Currently, these software clients can only be used internally within CBL due to
security firewalls in place at CBL. However, as described below, we have also implemented a
web interface that provides access to all but the processed fish and diet data to the general public
via the internet.
We have developed SQL queries for all local clients able to support data queries from the
master CHESFIMS database. For example the following code implements the database
connection in SAS and extracts all spring cruises
libname survey ODBC datasrc="miller_survey" user=xxx password=xxx; libname cf 'c:\files\cf'; proc sql; Create table cf.testcatch as Select CruiseID FROM survey.cruise WHERE month>4; run; Similarly, it is simple to make the connection in the statistical package R Library(ODBC) cfdata<-odbcconnect(“Miller_Survey”, uid=“xxx”, pwd=“xxx”) Cruiselist<-sqlQuery(cfdata, “select CruiseID from Cruise where month >4” Of course more complex queries are possible. The next example of the implementation in R extracts data to generate length frequencies for a single species from the database. ## produces a length frequency plot for a single species for all CF cruises ## edit the Size.Species= to change the species - may have to change x and y label limits ## anchovy = 161839, wp=167678, croaker=169283, sb=167680, hc=172982 library(RODBC) ## connect to the database lfdata<-odbcConnect("Miller_Survey", uid="xxx", pwd="xxx") ## use SQL query to retrieve required data
28
spdata<-sqlQuery(lfdata,"SELECT Cruise.CruiseID, Cruise.Program, Cruise.Month, Station.EventNum, Station.Gear, Size.SpeciesID, Size.Length, Size.Weight FROM (Cruise INNER JOIN Station ON Cruise.CruiseID = Station.CruiseID) INNER JOIN Size ON Station.EventNum = Size.EventNum GROUP BY Cruise.CruiseID, Cruise.Program, Cruise.Month, Station.EventNum, Station.Gear, Size.SpeciesID, Size.Length, Size.Weight HAVING (((Cruise.Program)='CF') AND ((Station.Gear)='MW') AND ((Size.SpeciesID)=167678)) ORDER BY Size.SpeciesID") close(lfdata)
2.4.3. Development of Web Interface
Following the successful development and implementation of the SQL database and
development of ODBC clients for a variety of commonly used software, our next goal was to
develop a web interface for the database. It was quickly determined following discussions with
CBL IT staff that off-the-shelf solution would likely not be sufficiently flexible for use.
Accordingly, we contracted with an independent web-developer to design and code a specific
implementation that we could tailor to our own use.
The contractor was provided with a limited number of minimum requirements. The web
interface had to be secure and not provide a portal to the actual data itself. Further, a specific
design criterion was that the portal should be sufficiently simple that any potential stakeholder
could access the data. Following discussions it was decided that these goals could be met if we
restricted the tool to the following data types
Total Catch (number) Total Catch (weight) Length data Average physical parameters (temperature, salinity, dissolved oxygen) Average surface parameters (temperature, salinity, dissolved oxygen) Average bottom parameters (temperature, salinity, dissolved oxygen)
The web site was coded using a combination of php, java script and SQL language. The
web site is completely open: there is no login or restriction on any user. The web site is
accessible from any web browser and has been tested in the principal commercially available
29
browsers included Internet Explorer v 6, Firefox v 2., Netscape v 7, and Apple Safari. The site is
registered with Google and other search engines. A GOOGLE search for CHESFIMS takes the
user directly to the main project page, or secondarily to the data tool itself. The main page of the
web interface is shown in Fig 2.3. The main page provides instructions about how to use the
remainder of the data selection tool. The data tool does not produce results directly to a web,
rather the result of any query is an Excel file that provides the raw data. The data tool also
provides access to a complete list of all stations visited during the survey program. This can be
used to expand estimates of survey CPUEs by accounting for those stations at which there was
no catch of the species selected. By navigating through the tabs from the main page, the user can
subset the CHESFIMS data by Season, Area, Date or Species. We are particularly pleased that
the Area tab allows the user to select regions of the Chesapeake Bay from which to query data
graphically by clicking and dragging to define the area of interest (Fig. 2.4).
The results of a query using the data selection tool are shown in Figure 2.5. Prior to
saving the data, the user can inspect the data prior to saving. Once the user has ensured the data
selected meet the criteria, the user can then save the data The data are saved to a compressed
folder that can be saved locally on the users own hard disk. When uncompressed, the data are
saved in a series of comma separated files that can be opened in excel or imported into a
software package of the users choice. The compressed files include a master file – called
index.csv which provides information on each cruise, and then separate csv files for each cruise.
An example of the data included in an individual cruise file is shown in Figure 2.6.
30
3.4. LITERATURE CITED
3. Literature Cited
Baird, D., R. E. Ulanowicz, and W. R. Boynton. 1995. Seasonal Nitrogen Dynamics in Chesapeake Bay - a Network Approach. Estuarine Coastal and Shelf Science 41:137-162.
Bonzek, C. F., E. D. Houde, S. Giordano, R. J. Latour, T. J. Miller, and K. G. Sellner. 2007. Baywide and coordinated Chesapeak Bay fish stock monitoring. Chesapeake Research Consortium, Edgewater, MD.
Caldarone, E. M., C. Clemmeson, T. J. Miller, A. Folkvord, J. Berdalet, G. J. Holt, and I. Suthers. 2006. Intercalibration of four spectrofluormetric methods for measuring RNA:DNA ratios in larval and juvenile fish. Limnology and Oceanography Methods 4:153-163.
CFEPTAP. 2004. Fisheries ecosystem planning for Chesapeake Bay. Chesapeake Fisheries Ecosystem Plan Technical Advisory Panel, NOAA Chesapeake Bay Office, Annapolis.
Chesapeake Bay Fisheries Ecosystem Advisory Panel. 2006. Fisheries Ecosystem Planning for the Chesapeake Bay. American Fisheries Society, Bethesda, MD.
Chesapeake Fishery Ecosystem Plan Technical Advisory Panel. 2006. Fishery Ecosystem Planning For Chesapeake Bay. American Fisheries Society, Bethesda, MD.
Choi, J. S., K. T. Frank, B. D. Petrie, and W. C. Leggett. 2005. Integrated assessment of a large marine ecosystem: A case study of the devolution of the Eastern Scotian Shelf, Canada. Oceanography and Marine Biology - an Annual Review, Vol. 43 43:47-+.
Clarke, K. R., and R. N. Gorley. 2001a. PRIMER. in. PRIMER-E Ltd, Plymouth, UK.
Clarke, K. R., and R. N. Gorley. 2001b. PRIMER v5: User Manual/Tutorial. Pages 91 in. PRIMER-E Ltd, Plymouth, UK.
Clarke, K. R., and R. M. Warwick. 2001. Change in Marine Communities: An Approach to Statistical Analysis and Interpretation, 2nd edition. PRIMER-E Ltd, Plymouth, UK.
Curti, K. L. 2005. Patterns in the distribution, diet and trophic demand of the hogchoker, Trinectes maculatus in the Chesapeake Bay, USA. MS. University of Maryland, College Park.
31
Frisk, M. G., and T. J. Miller. 2006. Age, growth, and latitudinal patterns of two Rajidae species in the northwestern Atlantic: little skate (Leucoraja erinacea) and winter skate (Leucoraja ocellata). Canadian Journal of Fisheries and Aquatic Sciences 63:1078-1091.
Hare, J. A., and K. W. Able. 2007. Mechanistic links between climate and fisheries along the east coast of the United States: explaining population outbursts of Atlantic croaker (Micropogonias undulatus). Fisheries Oceanography 16:31-45.
Houde, E. D., M. J. Fogarty, and T. J. Miller. 1998. Prospects for Multispecies Management in Chesapeake Bay: A workshop. 98-002, Scientific and Technical Committee of the Chesapeake Bay Program, Edgwater, MD.
Jensen, O. P., M. C. Christman, and T. J. Miller. 2006. Landscape-based geostatistics: a case study of the distribution of blue cab in Chesapeake Bay. Environmetrics 16:1-17.
Jensen, O. P., and T. J. Miller. 2005. Geostatistical analysis of blue crab (Callinectes sapidus) abundance and winter distribution patterns in Chesapeake Bay. Transactions of the American Fisheries Society 134:1582-1598.
Jensen, O. P., R. Seppelt, and T. J. Miller. 2005. Winter distribution of blue crab (Callinectes sapidus) in Chesapeake Bay: application of a two-stage generalized additive model (GAM). Marine Ecology - Progress Series 299:239-255.
Jung, S., and E. D. Houde. 2003. Spatial and temporal variabilities of pelagic fish community structure and distribution in Chesapeake Bay, USA. Estuarine Coastal and Shelf Science 58:335-351.
Jung, S., and E. D. Houde. 2004. Recruitment and spawning-stock biomass distribution of bay anchovy (Anchoa mitchilli) in Chesapeake Bay. Fishery Bulletin 102:63-77.
Kaiser, M. J., and S. Jennings. 2002. Ecosystem effects of fishing. Pages 342-366 in P. J. B. Hart and J. B. Reynolds, editors. Handbook of Fish Biology and Fisheries. Blackwell Science, Oxford.
Kemp, W. M., W. R. Boynton, J. E. Adolf, D. F. Boesch, W. C. Boicourt, G. Brush, J. C. Cornwell, T. R. Fisher, P. M. Glibert, J. D. Hagy, L. W. Harding, E. D. Houde, D. G. Kimmel, W. D. Miller, R. I. E. Newell, M. R. Roman, E. M. Smith, and J. C. Stevenson. 2005. Eutrophication of Chesapeake Bay: historical trends and ecological interactions. Marine Ecology-Progress Series 303:1-29.
Komatsu, T., T. Sugimoto, K. Ishida, K. Itaya, P. Mishra, and T. Miura. 2002. Importance of the Shatsky Rise Area in the Kuroshio Extension as an offshore nursery ground for Japanese anchovy (Engraulis japonicus) and sardine (Sardinops melanostictus). Fisheries Oceanography 11:354-360.
32
Latour, R. J., M. J. Brush, and C. F. Bonzek. 2003. Toward ecosystem-based fisheries management: Strategies for multispecies modeling and associated data requirements. Fisheries 28:10-22.
McErlean, A. J., S. G. O'Connor, J. A. Mihursky, and C. I. Gibson. 1973. Abundance, diversity and seasonal patterns of estuarine fish populations. Estuarine, Coastal and Shelf Science 1:19-36.
Miller, T. J. 2006a. Total Removals. Pages 189-212 in C. B. F. E. A. Panel, editor. Fisheries Ecosystem Planning for the Chesapeake Bay. American Fisheries Society, Bethesda, MD.
Miller, T. J. 2006b. Total Removals. Pages xx-xxx in NOAA, editor. Fisheries Ecosystem Planning for the Chesapeake Bay. American Fisheries Society, Bethesda, MD.
Miller, T. J., E. D. Houde, and E. J. Watkins. 1996. Perspectives on Chesapeake Bay Fisheries: Prospects for multispecies management and sustainability. Scientific & Technical Advisory Committee, Chesapeake Bay Program, Annapolis, MD.
Montane, M. M., and H. M. Austin. 2005. Effects of hurricanes on Atlantic croaker (Micropogonias undulatus) recruitment to Chesapeake Bay. Pages 185-192 in K. G. Sellner, editor. Hurricane Isabel: Perspectives. Chesapeake Research Consortiumm, Edgewater. MD.
Murawski, S. A. 2000. Definitions of overfishing from an ecosystem perspective. ICES Journal fo Marine Science 57:649-658.
Murdy, E. O., R. S. Birdsong, and J. A. Musick. 1997. Fishes of Chesapeake Bay. Smithsonian Institution Press, Washington.
Nixon, S. W. 1988. Physical energy inputs and the comparative ecology of lake and marine ecosystems. Limnology and Oceanography 33:1005-1025.
Oksanen, J., R. Kindt, P. Legendre, B. O'Hara, and M. H. H. Stevens. 2007. Vegan: Community Ecology Package. in. R Foundation for Statistical Computing.
Pauly, D., V. Christensen, S. Guenette, T. J. Pitcher, U. R. Sumaila, C. J. Walters, R. Watson, and D. Zeller. 2002. Towards sustainability in world fisheries. Nature 418:689-695.
Peck, M. A., L. J. Buckley, E. M. Caldarone, and D. A. Bengtson. 2003. Effects of food consumption and temperature on growth rate and biochemical-based indicators of growth in early juvenile Atlantic cod Gadus morhua and haddock Melanogrammus aeglefinus. Marine Ecology-Progress Series 251:233-243.
Pikitch, E. K., C. Santora, E. A. Babcock, A. Bakun, R. Bonfil, D. O. Conover, P. Dayton, P. Doukakis, D. Fluharty, B. Heneman, E. D. Houde, J. Link, P. A. Livingston, M. Mangel,
33
M. K. McAllister, J. Pope, and K. J. Sainsbury. 2004. Ecosystem-based fishery management. Science 305:346-347.
R Core Development Team. 2007. R: A Language and Environment for Statistical Computing. in. R Foundation for Statistical Computing, Vienna, Austria.
Rothschild, B. J., J. S. Ault, P. Goulletquer, and M. Heral. 1994. Decline of the Chesapeake Bay oyster population: a century of habitat destruction and overfishing. Marine Ecology Progress Series 111:29-39.
Sakuma, K. M., and S. Ralston. 1997. Vertical and horizontal distribution of juvenile Pacific whiting (Merluccius productus) in relation to hydrography off California. California Cooperative Oceanic Fisheries Investigations Reports 38:137-146.
Secor, D. H., and H. M. Austin. 2006. Externalities. Pages 277-324 in C. B. F. E. A. Panel, editor. Fishery Ecosystem Planning for the Chesapeake Bay. American Fisheries Society, Bethesda, MD.
Secor, D. H., and E. D. Houde. 1995. Temperature effects on the timing of striped bass egg production, larval viability, and recruitment potential in the patuxent river (chesapeake bay). Estuaries 18:527-544.
Smith, T. D. 1994. Scaling Fisheries, the science of measuring the effects of fishing, 1855- 1955. Cambridge Univ. Press, Cambridge, UK.
Ter Braak. 1986. Canonical correspondance analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67:1167-1179.
Wilson, M. T. 2000. Effects of year and region on the abundance and size of age-0 walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska, 1985-1988. Fishery Bulletin 98:823-834.
Wood, S. N., and N. H. Augustin. 2002. GAMs with integrated model selection using penalized regression splines and applications to environmental modelling. Ecological Modelling 157:157-177.
Worm, B., E. B. Barbier, N. Beaumont, J. E. Duffy, C. Folke, B. S. Halpern, J. B. C. Jackson, H. K. Lotze, F. Micheli, S. R. Palumbi, E. Sala, K. A. Selkoe, J. J. Stachowicz, and R. Watson. 2006. Impacts of biodiversity loss on ocean ecosystem services. Science 314:787-790.
34
Table 2.1. Filename standards and content for electronic information developed on each cruise
Filename Extension Instrument Format Information CTD CFYYMMTTSS .con CTD ASCII Communication
information for CTD
CFYYMMTTSS .hex CTD Hexadecimal Raw data from CTD profile
CFYYMMTTSS .cnv CTD ASCII, CSV delimited
Converted Hex data. File contains converted data for the downcast together with details on each instrument
MiniLog BinXXXX .XXX Minilog Hexadecimal Raw data for
each minilog deployment. The numbers following the BIN identify the minilog used, and the extension is simply a numeric counter that identifies the number of hauls since the memory was last wiped.
CFYYMMTTSS_btm .txt Minilog ASCII, csv delimited
Converted minilog data for a minilog attached to the foot rope. Data columns are identified in the file header
CFYYMMTTSS_top .txt Minilog ASCII, csv Converted
35
delimited minilog data for a minilog attached to the header rope. Data columns are identified in the file header
36
Table 2.2. Details of CRUISE Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface
Field Name Data type Explanatione CruiseID Text, 6 Cruise Number (CF0203) Program Text, 2 CF Vessel Text, 50 Aqaurius Year Integer, 4 2002 Season Text, 20 Summer Month Integer, 2 7 Comment Text, 50
37
Table 2.3. Details of STATION Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface
Field Name Data Type Explanation EventNum Number StationID Number Original station number CruiseID Text Cruise Number (CF0302) Btime Date/Time Arrival time on station AirTemp Number Air temperature (C – from
Bridge log) Barometer Number Air pressure (From Bridge
log) WindDirection Text Wind direction – compass rose
– From Bridge Log WindSpeed Text Wind Speed (From Bridge
Log) Tide Text State of tide (From Bridge
Log) Weather Text Weather conditions (From
Bridge Log) CloudCover Text (From Bridge Log) SeaState Text (From Bridge Log) Transect Number Transect for station, cruise
sampling design Region Text Region of Bay (Upper, Mid,
Lower) Order Number Sequential number of station
during cruise Date Date/Time Date of arrival on station Depth Number Station depth recorded from
aft mounted depth sounder (m)CTDLat Number Latitude of CTD cast CTDLong Number Longitude of CTD cast Gear Text Gear used (MW – midwater) TLATBegin Number Latitude at beginning of tow TLongBegin Number Longitude at beginning of tow TLatEnd Number Latitude at end of tow TLongEnd Number Longitude at beginning of tow Ttime Number Duration of tow CTDFile Text Name of CTD conv file MLTopFile Text Name of mimilog top file MLBottomFile Test Name of minilog bottom file
38
Table 2.4. Details of CTD Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface
Field Name Data type Explanation EventNum Number EventNumber from StationID Sca Integer Sequential number indicating
line of data Time Number Time of scan Pressure Number Pressure in water column Depth Number Estimated depth from pressure Salinity Number Salinity in water column Temp Number Temperature in water column Oxygen Number Oxygen concentration in water
column Oxysat Number Percent saturation of oxygen
in water column Transmis Number Transmissivity in water
column Flourom Number Flourometer value in water
column Flag Number Error flag recorded by CTD
39
Table 2.5. Details of MiniLogTop Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface
Field Name Data type Explanation EventNum Number EventNumber from StationID Time Text Time at which data was
scanned Temp Number Temperature in water column Depthj Number Depth recorded
40
Table 2.6. Details of MiniLogBottom Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface
Field Name Data type Explanation EventNum Number EventNumber from StationID Time Text Time at which data was
scanned Temp Number Temperature in water column Depth Number Depth recorded
41
Table 2.7. Details of CatchSummary Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface Field Name Data Type Explanation EventNum Number EventNumber from StationID StationID Number Original station number SpeciesID Number Species number from ITIS
database Split Text Sample fraction worked up SplitBasis Text Sample fraction based on
volume of weight SubWt Number Weight of subsample SubVol Number Volume of subsample TotVol Number Total volume of sample SubCt Number Count of subsample TotWt Number Total weight of sample TotCt Number Total count of sample Comment Text
42
Table 2.8. Details of Size Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface Field Name Data Type Explanation SizeSpecimenID Number Unqiue ID for each dataline EventNum Number EventNumber from StationID StationID Number Original station number SpeciesID Number Species number from ITIS
database Length Number Total length of fish Weight Niumber Weight of fish Sext Text Sex of fish Comment Text
43
Table 2.9. Details of PROCESSEDFISH Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface Field Name Data Type Explanation EventNum Number EventNumber from StationID SpeciesID Number Species number from ITIS
database FishID Number Unique sequential number for
each species, eventnum combination
ProcessedDate Date/Time Date when fish was processed Length Number Total length of fish Weight Niumber Weight of fish Sex Text Sex of fish FullStomachWt Number Weight of stomach with
contents EmptyStomachWt Number Weight of stomach with
contents removed Age Number Estimated age Comments Text
44
Table 2.10. Details of DIET Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface Field Name Data Type Explanation DietSpecimenID Number Unqiue ID for each dataline FishID Number Unique ID from Processed
fish table PreyID Number Species number from ITIS
database r PreyCt Number Count of prey type in stomach Length Number Total length of fish Weight Niumber Weight of fish Presence Number 0,1 index indicating presence % Number Percent contribution Comment Text
45
Table 2.12. Details of DIET Table Structure for CHESFIMS SQL Database. The key for each table is shown in boldface Field Name Data Type Explanation SpeciesID Number Species number from ITIS
database r e Genus Text Genus Species Text Species TaxName Text Full latin binomial Family Text Family name CommonName Text Full common name Guild Text Spawning guild LifeHistory Text Life history mode Trophic Text Trophic mode SampleAbbreviation Text Abbreviation used in sample
processing SurveySpecies Text Abbreviation used in survey
data sheet PreySpecies Text Abbreviation used in diet data
sheet Comment
46
Figure 2.1 Example of the distribution of samples on an individual cruise
47
Fig 2.2. Relationships among tables in the CHESFIMS database. Keys for each table are shown in boldface
48
Figure 2.3. Main page of the web tool for accessing the CHESFIMS data
49
Figure 2.4. The area selection tab on the CHESFIMS data web tool
50
Figure 2.5. Preliminary results of the data selection tool for a query selecting white perch during the summer in the upper Bay.
51
Figure 2.6. An example of the data included in a single cruise file produced by a search for data on white perch catches in the upper bay in the summer.
52
APPENDIX 2.1.
Matlab scripts to estimate average net openings
Aveop.m % a script file that reads a list of stations for a cruise % and then executes a function - mlcombst for each station % to calculate the average net opening for that station stalist=textread('H:\chesfims\cf0303\minilog\0303sta.txt','%s'); stas=char(stalist); cmax=length(stas); for i=1:cmax sta=char(stalist(i)) aveop(i,2)=mlcombst(cruise,sta); end
mlcombst.m
function meanopen=mlcombst(cruise,sta) %mlcomb takes two data matrices read from minilog files on the top and %bottom ropes of a trawl, aligns time stamps and creates a single output %file %Inputs are: % top = the ml data matrix from the top rope - data obtained % from a call to ml2mat(top rope file name) % bottom = the ml data matrix from the bottom rope - data obtained % from a call to ml2mat(top rope file name) % top=ml2mat(strcat('H:\chesfims\cf',cruise,'\minilog\CF',cruise,sta,'_top.txt')); bottom=ml2mat(strcat('H:\chesfims\cf',cruise,'\minilog\CF',cruise,sta,'_btm.txt')); % compare the initial time stamps if top(1,1)<bottom(1,1) is=find(top(:,1)<bottom(1,1)); indx=max(is); top=top(indx:end,:); elseif top(1,1)>bottom(1,1) is=find(bottom(:,1)<top(1,1)); indx=max(is); bottom=bottom(indx:end,:); end
53
% % now merge the two data sets % if length(top)>length(bottom) maxrow=length(bottom); elseif length(bottom)>length(top) maxrow=length(top); end top=top(1:maxrow,:); bottom=bottom(1:maxrow,:); mldat=horzcat(top, bottom); mldat(:,7)=mldat(:,6)-mldat(:,3); means=mean(mldat,1); meanopen=means(1,7); return
54
Chapter 3
PATTERNS IN THE DISTRIBUTION AND COMPOSITION OF THE FISH ASSEMBLAGE IN THE
CHESAPEAKE BAY
Thomas J. Miller and David A. Loewensteiner
Chesapeake Biological Laboratory
University of Maryland Center for Environmental Science
P. O. Box 38
Solomons, MD 20688
55
3.1 INTRODUCTION
Agencies, at all levels, are seeking to adopt ecosystem approaches to management (EAM)
of fisheries in efforts to ensure long term sustainability of the exploited marine resources and the
ecosystems from which these resources are taken (Pauly et al. 2002). The need to adopt EAM is
motivated by the recognition that fish populations are components of complex marine
ecosystems. Removals by fisheries from these ecosystems have consequences to ecosystem
services generally, and in turn ecosystem processes affect the productivity and sustainability of
individual fish stocks (Pikitch et al. 2004, Worm et al. 2006). However, EAM require analysis
of a wider range of data than are typically available (Choi et al. 2005). Such analyses rely on
access to data from a wide range of sources.
The Chesapeake Bay, located on the mid-Atlantic seaboard, is the largest estuary in
North America. It is shallow, temperate, coastal plain estuary with high productivity (Nixon
1988). Since European settlement, the bay has experienced heavy exploitation, extensive
physical habitat alteration, invasion of exotics and anthropogenic nutrient enrichment (Kemp et
al. 2005, Secor and Austin 2006). These forces have had significant impacts on its fishery
resources (Miller 2006a). For example, oyster (Crassostrea virginica) fisheries flourished at the
end of the 19th Century, with catches in excess of 600,000 t. As a result of both overfishing and
disease (Rothschild et al. 1994), current catches are less than 1% of this maximum level. Several
other species (e.g., striped bass) have experienced similarly substantial changes in abundance.
Together these changes in the Chespaeake Bay ecosystem motivated the regional management
jurisdictions to commit to an EAM for the Chesapeake Bay (Chesapeake Bay Fisheries
Ecosystem Advisory Panel 2006). A central requirement for the success implementation of
EAM is the availability of fishery-independent survey data for both economically and
ecologically important species from which biological reference points can be developed.
In 1996, researchers at the University of Maryland Center for Environmental Science
instituted a baywide survey of the fish community in Chesapeake Bay as part of a larger project
to document and quantify trophic interactions in the estuary (TIES). Surveys of the fish
community focused on the small pelagic and benthopelagic components of the fish fauna
because these sizes and species represent an important energetic pathway for the ecosystem as a
56
whole (Baird et al. 1995). Additionally, this component of the fish community had been largely
neglected by existing surveys. Jung and Houde (2003) published a description of the baywide
spatial and temporal pattern in community structure of the pelagic and benthopelagic fishes
throughout the Bay based on surveys conducted between 1995-2000. Jung and Houde
documented an assemblage comprising 36 pelagic and 21 benthic species. They characterized
the assemblage as a low diversity, but high dominance system, with catches being dominated by
bay anchovy (Anchoa mitchelli) by number and by bay anchovy, Atlantic croaker
(Micropogonias undulatus) white perch (Morone Americana) by weight. Jung and Houde (op.
cit.) used a multivariate correspondence analysis to quantify changes in the fish assemblage over
time. This analysis clearly demonstrated an important seasonal and spatial progression in survey
catches. Interannual differences in the abundances, when corrected for season appeared to
respond most to variation in the freshwater input to the Bay. Most surprisingly, Jung and
Houde’s analysis also indicated an inverse relationship between fish biomass and mean, depth-
integrated oxygen concentration. This relationship would appear to indicate that assemblage
biomass was highest under the lowest oxygen concentrations. Jung and Houde hypothesized
that the oxygen levels in this analysis were serving as a surrogate for benthic production: when
benthic production is high, it acts to lower oxygen levels. Thus they suggest that the high
benthic production supports high biomass in the fish assemblage. The results of Jung and
Houde’s analyses represent hypotheses relating to the bentho-pelagic fish community in
Chesapeake Bay. Specifically they hypothesized that
H1: Interannual variation in assemblage structure is caused by differences in the flow of
freshwater into the Chesapeake Bay.
H2: Spatial variation in the assemblage responds to the salinity distribution such that clear
oligohaline (0-5 salinity), mesohaline (5-18 salinity) and polyhaline (>18 salinity) assemblages
can be identified.
H3: There is a strong seasonal succession evident in the fish assemblage that reflects the role of
the Bay as a nursery area for anadromous species in the spring /summer period and then the
recruitment of bay anchovy in the autumn.
57
With funding from the NOAA Chesapeake Bay Office, we initiated the Chesapeake Bay
Fishery Independent Multispecies (CHESFIMS) Survey to complement the TIES survey. The
objectives of the survey were to extend the TIES survey for an additional five years (2001-2006)
using the same sampling protocols and to determine the trophic interactions among key
components of the pelagic fish community. Here we use the CHEFIMS data to to test the
specific hypotheses suggested by Jung and Houde’s analysis of the 1995-2000 data.
58
3.2 METHODS
3.2.1. Field
The CHESFIMS survey is a complemented design involved both fixed and random
stations. The fixed stations were located along fixed transects that had been defined during the
early TIES program (Jung and Houde 2003). The TIES program occupied more stations and
transects than could be occupied during CHESFMS, and accordingly we sampled a subset of
these original stations, selected to ensure a broad spatial coverage throughout the Chesapeake
Bay. In addition to these fixed stations, we also included a similarly number of stratified random
stations selected in proportion to strata area. Individual strata have distinctive characteristics,
and their boundaries broadly corresponding to ecologically relevant salinity regimes and depths
above 5 m. The upper Bay (38 45’ to 39 25’N) is generally shallow, with substantial areas with
depths less than 5 m, and has well mixed waters with high nutrient concentrations. The bottom
topography in the mid Bay includes a narrow channel in the middle of the Bay (37 55’N to 38
45’N) with a stratified water column and broad flanking shoals. This region has relatively clear
waters and experiences seasonally high nutrient concentrations and periods of hypoxia. The
lower Bay (37 05’ to 37 55’N) has the clearest waters, greatest depths and lowest nutrient
concentrations (Kemp et al., 2005). The strata volumes are 26,608 km3 (Lower), 16,840 km3
(Mid) and 8,664 km3 (Upper).
CHESFIMS was initiated in 2001, employing the TIES trawling procedures, transect
design and stratification. Sampling was conducted during spring, summer and autumn from the
University of Maryland Center for Estuarine and Environmental Science’s R/V ‘Aquarius. On
most cruises we occupied 31 stations allocated according to the original TIES transect design,
and 20 stations allocated according to a stratified random scheme. The transect stations were
fixed for the duration of the CHESFIMS program, whereas the stratified random stations were
selected anew for each cruise. In the stratified random surveys, conducted simultaneously with
the transect surveys, 20 stations covering the entire bay were allocated to the each stratum
proportional to their volumes. Latitude and longitude of stations within strata were randomly
generated. Weather precluded complete sampling of all stations during some surveys.
59
To the extent possible all activity at each station was standardized. Immediately prior to
each trawl deployment, a Sea-Bird SBE 25 conductivity, temperature, depth profiler was used to
profile the water column. We used an 18-m2 midwater trawl (MWT) with 3-mm codend mesh as
the primary survey gear (Jung and Houde, 2003). The MWT employed in the Bay samples fish
30-256 mm total length of most species effectively, but appears to be less effective for Atlantic
menhaden (Brevoortia tyrannus) of all sizes(Jung and Houde 2003). We used a standardized 20-
minute oblique, stepped tow in all deployments. The trawl was towed for two minutes in each of
ten depth zones distributed throughout the water column from the surface to the bottom, with
minimum trawlable depth being 5 m. The section of the tow conducted in the deepest zone
sampled epibenthic fishes close to or on the bottom. The remaining portion of the tow sampled
pelagic and neustonic fishes. All tows were conducted between 19:00 and 7:00 Eastern Standard
Time to minimize gear avoidance and to take advantage of the reduced patchiness of multiple
target species at night. A temperature-depth minilogger was attached to the head rope of the
trawl for all deployments to record the tow profile. Data from this minilog was inspected
immediately after each haul to determine whether the haul had largely met survey criteria. An
additional minilog was attached to the foot rope of the trawl. By combining data from the top
and bottom minilogs, the net opening could be estimated for the duration of the haul. Catches
were identified, enumerated, measured and weighed onboard. For species for which there were
less than 100 individuals caught in the haul, all individuals were measured to the nearest mm for
total length (TL), and a total weight of the species caught was measured on a precision spring
balance (nearest g). For some species, individuals were of sufficient size that individual weights
could be determined, but this was not common. For abundant species (e.g., bay anchovy, white
perch etc), a subsampling procedure was used to estimate the catch. In most cases a random
subsample of 100 individuals was measured for TL. This subsample was weighed, and the total
catch of the species was also weighed. The total number in the catch was then estimated by
proportion.
3.2.2. Statistical analysis
Catch data were used to generate catch per unit effort estimates for each species in the
survey. These estimates were expanded by including zeros for hauls where that species was not
60
collected. We calculated the species richness, S, as the number of unique species caught per tow.
Estimates of S could be aggregated to the region, cruise, season and program level. Because the
sensitivity of species richness to the number of individuals caught, we also calculated the
Shanon’s diversity index, H, for all tows. H is given by
∑−=i
ii ppH 2log
where pi is the proportion of species i in the tow. It is not possible to aggregate H at levels
higher than the individual tow because H is sensitive to differences in sampling effort (Clarke
and Warwick 2001).
Following Jung and Houde (2003), we also calculated k-dominance curves. These are a
graphical approach to comparing diversity among different sites. In this approach the ranked
abundance, expressed as the proportion of total abundance that is represented by species i, is
plotted against the species rank. K-dominance curves were calculated using the software Primer
(Clarke and Gorley 2001a).
We also used canonical correspondence analysis (Ter Braak 1986) to explore
relationships between environmental determinants and species distributions. Prior to conducting
canonical correspondence analysis, we examined the correlation structure of the environmental
data collected during the cruises. For most parameters, there were strong correlations among
surface, bottom and average conditions at each station. For spring and autumn cruises, the
correlation between surface and bottom values for each parameter were the strongest in the entire
correlation matrix (r>0.42 for all pair wise surface vs. bottom comparisons). The relationship
was less strong during summer months. Relationships between surface and bottom temperature,
oxygen and transmissivity were not strong. However, despite this pattern in summer values,
initial, exploratory canonical correspondence analyses were conducted with region, season, year,
average water column values for salinity, temperature (oC), oxygen (mg.L-1), transmissivity and
fluorometry values. Catch data for all species were included in the analysis. The constraint that
data for all parameters had to be available for a station to be included, restricted the number of
stations available for analysis. Missing environmental data meant that only 464 stations could be
included. An exploratory analysis tested for the significance of each environmental parameter
61
using a randomization process with 500 permutations. Based on these analyses, non-significant
terms were dropped from subsequent canonical correspondence analyses. Canonical
correspondence analysis was conducted using the Vegan package (v. 1.8-6) in R.
To address hypotheses regarding the relationship between fish abundance and dissolved
oxygen, we regressed Bay wide average catch per unit effort, estimated in terms of weight
(g.min-1) or number (fish.min-1) against depth-weighted dissolved oxygen. Regressions were
conducted for both summer and fall separately.
All statistical analyses, except for K-dominance curves estimation, were conducted in R
(R Core Development Team 2007) using the base package for standard analysis and graphics.
The Vegan package was used for the canonical correlation analysis (Oksanen et al. 2007).
Finally the K-dominance curves were implemented in Primer v.6 (Clarke and Gorley 2001b)
62
3.3. RESULTS
We completed 17 surveys between April 2001 – October 2006 (Table 3.1), During this
period, baywide average temperatures ranged from 13.41-25.95 oC, baywide salinities varied
from10.96-20.36 and baywide oxygen concentrations from 4.25-9.81 mg.L-1. These levels of
variability are similar, although less extreme than those reported by Jung and Houde (Jung and
Houde 2003) for the period 1995-2000. There was a tendency for those cruises that were on
average warmer, to also be on average saltier (Fig. 3.1), but no other strong correlations were
apparent in the data.
Over the course of the 17 surveys, we occupied 693 stations (Fig. 3.2), and collected
520,786 fish weighing 2,205 kg. Only eight deployments (1.15%) resulted in a null catch. We
collected 110 different species of fish (Table 3.2). The average 20 minute tow resulted in a catch
of 811±1135 (mean ± SD) fish. When analyzed by logarithmic catch intervals, the frequency
distribution of catches appeared multimodal, with modes at approximately 200 and at 2000
fish.tow-1 (Fig. 3.3). We calculated species-specific average biomass CPUEs at the annual,
seasonal and regional levels (Figs 3.4-3.6). At the annual level, catches exhibited considerable
variability (Fig 3.4).. In four of six years Atlantic croaker exhibited the highest CPUES. In the
two years in which croaker was not dominant, white perch was the dominant species. In
combination, croaker, white perch and bay anchovy comprised at least 60% of the total CPUE in
all years. Spot represented substantial CPUE levels in 2005-2006, but otherwise all other species
individually accounted for less than 10% of the total CPUE , and often less than 5%. At the
seasonal level, average biomass-based CPUEs were again dominated by croaker, white perch and
bay anchovy (Fig 3.5). Average CPUEs for croaker and white perch declined over the course of
the year, whereas average bay anchovy CPUE increased. At the regional level, CPUE clearly
indicate the known salinity tolerances of individual species (Fig. 3.6). White perch dominated
catches in the upper Bay, bay anchovy those in the mid Bay and croaker those in the lower Bay.
The average species richness on each cruise was 36.06±7.06 species. Species richness
increased over the course of the year, with an average of 32.6±4.72, 35.83±5.04 and 39.16±9.62
63
species collected in spring, summer and autumn cruises respectively. There was also an increase
in the species richness over the course of the course of the sampling program. However,
inspection suggests that this increase reflects two aspects of the data: the seasonal effect and the
sensitivity of species richness, S, to sample size. Indeed, there is a weakly significant positive
relationship between the number of stations occupied on a cruise and the species richness
(S=0.287+23.39*Stations, R2=0.17, n=17, p <0.1). Otherwise, there was no obvious change in
design, gear or vessel that might have contributed to the increase. At the level of the individual
tow and ignoring the null hauls, species richness per haul ranged from 1-21 with a mean of
6.17±3.57. The distribution of species richness per haul was strongly negatively skewed (Fig.
3.7). When averaged by cruise, the seasonal effect on species richness is clear (Fig. 3.7). The
distribution of Shanon’s diversity index, H, for each tow was strongly negatively skewed also
(Fig. 3.7). The average value of H was 0.524 ± 0.455 (range 0 – 2.12).
K-dominance curves indicated that the Chesapeake Bay fish assemblage is dominated by
a relatively few species when assessed either by abundance (Fig 3.8A) or biomass (3.8B). In
either case, the top five species never account for less than 60% of the cumulative distribution,
and often more than 90%. This is the particularly the case for abundance in fall cruises (Fig
3.8A) for which a single species (bay anchovy) often accounted for in excess of 90% of the
cumulative catch by itself. Analysis of similarities indicated that the seasonal differences were
significant for biomass (Global R=0.205, p=0.05), and marginally insignificant for abundance
(Global R=0.119, p=0.08). A similar pattern was evident when the data were aggregated to
examine K-dominance curves at the regional level (Fig. 3.9). As with individual cruise data, at
the regional level, the assemblage was less evenly distributed when measured in terms of
abundance (Fig. 3.9 A) than when measured in terms of biomass (Fig. 3.9 B). The top five
species combined to represent more than 90% of the total abundance, and more than 60% of the
total biomass.
Canonical correspondence analysis was conducted using region, season, year, water
temperature, salinity, oxygen concentration, and transmissivity. The analysis yielded an
ordination dominated by the first five unconstrained axes, with the first axes dominating all
64
others (Table 3.3) The CCA triplot indicated clear patterns in the weightings of the
environmental variables (Fig. 3.10). The first axis which explained 11.8% of the variation in the
data, was highly positively correlated with region and salinity, and highly negatively correlated
with salinity. The second axis was most strongly related to temperature (+ve) and oxygen (-ve).
The effects of region and season were almost orthogonal to each other, as were temperature and
salinity. On the plot the effects of temperature and season were in a similar direction, and
opposing that of oxygen, indicating that seasonal changes are primarily associated with changes
increases in temperature and decreases in dissolved oxygen levels. Similarly, the effects of year
and salinity were in the same direction and opposed the effects of region. This combination
indicates that salinity is the principal factor associated with interannual differences in community
structure, and is the primary factor affecting community difference in the three regions.
We also quantified the relationship between fish catches and oxygen concentrations. The
relationship between summertime (July) catch per unit effort and summer time depth weighted
oxygen concentration was negative (Fig 3.11A). However, the coefficient of determination was
not significant (CPUE=497.36 - 64.08* Oxygen, n=5, r2=-0.15, p=0.539), suggesting a weak
relationship. A model of October CPUEs vs oxygen concentrations was however significant, but
the slope was reversed (Fig. 3.11 B).
65
3.4 DISCUSSION
We have documented variability in the fish assemblage of the Chesapeake Bay at annual,
seasonal and regional scales. The variation we document is similar in magnitude to that
previously demonstrated for the Chesapeake Bay by Jung and Houde (2003), but there are
important differences in the pattern between the two studies. For example, in the six years
analyzed by Jung and Houde bay anchovy was the dominant species in terms of biomass in four
of the six years. In none, of the years we studied, was bay anchovy dominant in terms of
biomass. In contrast, croaker dominated only in one year of the period examined by Jung and
Houde, whereas it dominated in four of the six years we report on hear. Croaker is known to
exhibit substantial interannual variation in abundances (Montane and Austin 2005, Hare and
Able 2007). Thus it is possible that in combination the 12 year data set reflect periods of
contrasting croaker abundances. The possibility of this is reinforced by the observation that
croaker were in high abundance in the later years of the TIES program, and in the earlier years of
the CHESFIMS program. However, we also note that Jung and Houde took a slightly different
approach to analyzing the TIES data than selected here. For example, they chose to expand their
catches by a catchability coefficient derived from a comparison of bay anchovy abundance in
TIES survey catches and that estimated from an anchovy egg production model (Jung and Houde
2004). Thus, at the moment, we are unable to determine whether the reported differences do
indeed reflect underlying ecological shifts or differences in methodology. At present, we are
converting the TIES data to the same data management system used by CHESFIMS and thus we
should be able to distinguish between these two alternatives in the near future.
Our analyses reinforce the overall spatial patterns demonstrated by Jung and Houde (Jung
and Houde 2003). There are clear spatial differences in the fate of production in Chesapeake
Bay. In the upper portion of the Bay, most of the primary and secondary production supports
substantial populations of white perch. In both the analyses reported here and those reported by
Jung and Houde, white perch represented approximately 70% of the total fish biomass in the
survey. As one moves southward the dominance of white perch declines and shifts to first bay
anchovy and then to croaker, which comprises approximately 40% of the biomass in the lower
Bay region. This shift in dominance reflects changes in known salinity tolerances of these
66
species, as demonstrated by the canonical correlation analysis. However, this pattern also
suggests the likelihood that the ultimate fate of secondary production in the Chesapeake Bay
ecosystem is likely dependent on the salinity distribution in the Bay. Thus the balance of
freshwater input and marine inflow determine whether the production is internalized within the
system (high run-off years leading to high white perch abundance) or potentially exported out of
the system (low run-off years leading to high croaker abundance).
Jung and Houde (Jung and Houde 2003) clearly demonstrated the importance of the
environmental oxygen levels in controlling the distribution and abundance of fish. These
findings were reinforced by the analysis of the CHESFIMS data. Environmental oxygen
exhibited its influence orthogonally to salinity, and in opposition to season. In addition, Jung
and Houde found that summertime baywide fish biomass was strongly negatively linearly related
to summertime environmental oxygen concentrations. This unexpected finding suggested that
fish biomass could be expected to be lower under conditions of more favorable oxygen
concentrations. They hypothesized that the pattern they reported indicated higher levels of
community respiration under more favorable oxygen environments, thereby leading to an
increased volume of hypoxic water during the strongly stratified summer months. We repeated
this analysis using the CHESFIMS database. A negative relationship was found between
baywide fish biomass and depth-averaged oxygen concentration during summer, but this
relationship was not as strong or as compelling as that reported by Jung and Houde. Inspection
of the results suggest the weaker pattern results from unexpectedly low biomasses for the
measured oxygen concentration in 2002 and 2003. We extended Jung and Houde’s analysis by
considering the correlation between fish biomass and oxygen in September. Surprisingly, this
relationship exhibited an opposite sign to that found in summer. The striking change in the sign
of the correlation points to a need to understand the role of the spatial and temporal distribution
of oxygen in the Chesapeake Bay on the production and distribution of fish. We suggest several
alternative hypotheses. First, the reported patterns may simply reflect differential vulnerability
to the midwater trawl caused by differences in the distribution of oxygen. The mid water trawl
has lower efficiency in the bottom waters. Thus if oxygen concentrations served to alter only the
vertical distribution of fish and not their overall abundance, then period of low DO would result
in a higher fraction of the fish assemblage being vulnerable to the midwater trawl. This
67
hypothesis could be tested by examining regional relationships between sub pycnocline oxygen
and catch. If the hypothesis is correct, there should be a negative relationship between these two
variables. Alternatively, environmental oxygen levels might, as Jung and Houde hypothesized
lead to higher production at higher DO levels. This would increase the rain of organic materials
to sub-pycnocline levels, thereby leading to an overall decline in DO. If this hypothesis is
correct, water column stratification should be a significant covariate in the relationship between
fish biomass and DO. At stations were stratification is weak, we should find the expected
positive relationship between fish biomass and Do, and the slope of this relationship should
become increasing more negative as the stratification index increases.
The analyses reported herein have been conducted at a coarse spatial scale. We have
designated only regional level spatial variation. However, it is likely that there is considerable
finer scale spatial pattern. Analysis at a finer spatial scale has the potential to develop predictive
models of fish abundance from environmental covariate using General Additive Models (Curti
2005, Jensen et al. 2005). More significantly, geostatistical approaches to interpolating the
distribution of abundance offer the possibility of more reliable estimates of abundance than are
possible from traditional design-based approaches (Jensen and Miller 2005, Jensen et al. 2006).
It is envisaged that both approaches will be applied to the CHESFIMS data for selected species.
Not only do the model-based geostatistical approaches offer the potential of developing
abundance indices, but they also offer the potential of yielding insight into the likely spatial
overlap of predators and prey. This is an important requirement for estimates of trophic demand
and predator-prey interactions that are the foundation for ecosystem-based approaches to
fisheries management (CFEPTAP 2004).
In summary, we have documented patterns of spatial and temporal variability in the fish
assemblage of the Chesapeake Bay. These patterns suggest important differences in the fate of
secondary production regionally, and potentially inter-annually. Moreover, we have also
documented the relationship between environmental oxygen and fish biomass. However, these
analyses are merely exploratory. More detailed analyses of the combined TIES and CHESFIMS
data would be beneficial to distinguish real pattern from methodological differences. Such
analyses are forthcoming.
68
3.5 LITERATURE CITED
LITERATURE CITED
Baird D, Ulanowicz RE, Boynton WR (1995) Seasonal Nitrogen Dynamics in Chesapeake Bay - a Network Approach. Estuar Coast Shelf S 41:137-162
CFEPTAP (2004) Fisheries ecosystem planning for Chesapeake Bay, Chesapeake Fisheries Ecosystem Plan Technical Advisory Panel, NOAA Chesapeake Bay Office, Annapolis
Chesapeake Bay Fisheries Ecosystem Advisory Panel (2006) Fisheries Ecosystem Planning for the Chesapeake Bay, Vol 3. American Fisheries Society, Bethesda, MD
Choi JS, Frank KT, Petrie BD, Leggett WC (2005) Integrated assessment of a large marine ecosystem: A case study of the devolution of the Eastern Scotian Shelf, Canada. Oceanogr Mar Biol 43:47-+
Clarke KR, Gorley RN (2001a) PRIMER. PRIMER-E Ltd, Plymouth, UK Clarke KR, Gorley RN (2001b) PRIMER v5: User Manual/Tutorial. PRIMER-E Ltd, Plymouth,
UK, p 91 Clarke KR, Warwick RM (2001) Change in Marine Communities: An Approach to Statistical
Analysis and Interpretation, Vol. PRIMER-E Ltd, Plymouth, UK Curti KL (2005) Patterns in the distribution, diet and trophic demand of the hogchoker, Trinectes
maculatus in the Chesapeake Bay, USA. MS, University of Maryland Hare JA, Able KW (2007) Mechanistic links between climate and fisheries along the east coast
of the United States: explaining population outbursts of Atlantic croaker (Micropogonias undulatus). Fisheries Oceanography 16:31-45
Jensen OP, Christman MC, Miller TJ (2006) Landscape-based geostatistics: a case study of the distribution of blue cab in Chesapeake Bay. Environmetrics 16:1-17
Jensen OP, Miller TJ (2005) Geostatistical analysis of blue crab (Callinectes sapidus) abundance and winter distribution patterns in Chesapeake Bay. Transactions of the American Fisheries Society 134:1582-1598
Jensen OP, Seppelt R, Miller TJ (2005) Winter distribution of blue crab (Callinectes sapidus) in Chesapeake Bay: application of a two-stage generalized additive model (GAM). Marine Ecology - Progress Series 299:239-255
Jung S, Houde ED (2003) Spatial and temporal variabilities of pelagic fish community structure and distribution in Chesapeake Bay, USA. Estuar Coast Shelf S 58:335-351
Jung S, Houde ED (2004) Recruitment and spawning-stock biomass distribution of bay anchovy (Anchoa mitchilli) in Chesapeake Bay. Fishery Bulletin 102:63-77
Kemp WM, Boynton WR, Adolf JE, Boesch DF, Boicourt WC, Brush G, Cornwell JC, Fisher TR, Glibert PM, Hagy JD, Harding LW, Houde ED, Kimmel DG, Miller WD, Newell RIE, Roman MR, Smith EM, Stevenson JC (2005) Eutrophication of Chesapeake Bay: historical trends and ecological interactions. Marine Ecology-Progress Series 303:1-29
Miller TJ (2006) Total Removals. In: Panel CBFEA (ed) Fisheries Ecosystem Planning for the Chesapeake Bay, Vol 3. American Fisheries Society, Bethesda, MD, p 189-212
Montane MM, Austin HM (2005) Effects of hurricanes on Atlantic croaker (Micropogonias undulatus) recruitment to Chesapeake Bay. In: Sellner KG (ed) Hurricane Isabel: Perspectives. Chesapeake Research Consortiumm, Edgewater. MD, p 185-192
69
Nixon SW (1988) Physical energy inputs and the comparative ecology of lake and marine ecosystems. Limnology and Oceanography 33:1005-1025
Oksanen J, Kindt R, Legendre P, O'Hara B, Stevens MHH (2007) Vegan: Community Ecology Package. R Foundation for Statistical Computing
Pauly D, Christensen V, Guenette S, Pitcher TJ, Sumaila UR, Walters CJ, Watson R, Zeller D (2002) Towards sustainability in world fisheries. Nature 418:689-695
Pikitch EK, Santora C, Babcock EA, Bakun A, Bonfil R, Conover DO, Dayton P, Doukakis P, Fluharty D, Heneman B, Houde ED, Link J, Livingston PA, Mangel M, McAllister MK, Pope J, Sainsbury KJ (2004) Ecosystem-based fishery management. Science 305:346-347
R Core Development Team (2007) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria
Rothschild BJ, Ault JS, Goulletquer P, Heral M (1994) Decline of the Chesapeake Bay oyster population: a century of habitat destruction and overfishing. Marine Ecology Progress Series 111:29-39
Secor DH, Austin HM (2006) Externalities. In: Panel CBFEA (ed) Fishery Ecosystem Planning for the Chesapeake Bay. American Fisheries Society, Bethesda, MD, p 277-324
Ter Braak (1986) Canonical correspondance analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67:1167-1179
Worm B, Barbier EB, Beaumont N, Duffy JE, Folke C, Halpern BS, Jackson JBC, Lotze HK, Micheli F, Palumbi SR, Sala E, Selkoe KA, Stachowicz JJ, Watson R (2006) Impacts of biodiversity loss on ocean ecosystem services. Science 314:787-790
70
Table 3.1. Summary of cruises included in the analysis
Average Physical Conditions Fish Caught Cruise
Year Season Start End N0. Stations Temp
(oC) Salinity Oxygen (mg.L-1)
Oxygen Saturation (%)
N Kg
CF0101 2001 Spring 5/1/01 5/4/01 30 14.96 15.60 7.62 83.20 1474 65947CF0103 2001 Fall 9/26/01 9/28/01 15 22.40 14.55 7.96 NA 73619 76316CF0201 2002 Spring 4/30/02 5/7/02 46 15.76 18.08 7.41 83.58 5986 113022CF0202 2002 Summer 7/8/02 7/14/02 49 25.78 17.04 5.19 70.20 31513 70053CF0203 2002 Fall 9/9/02 9/17/02 51 24.29 20.37 6.62 88.92 73439 189432CF0301 2003 Spring 4/28/03 5/4/03 49 13.65 14.16 7.90 83.32 4636 113421CF0302 2003 Summer 7/14/03 51 24.58 14.84 4.93 62.80 6981 60844CF0303 2003 Fall 9/9/03 9/16/03 29 23.71 14.59 6.76 86.81 32532 38241CF0401 2004 Spring 4/19/04 4/25/04 48 13.41 12.17 9.82 101.69 18193 146318CF0402 2004 Summer 7/6/04 7/13/04 51 25.95 15.11 6.45 86.92 15465 102596CF0403 2004 Fall 9/13/04 9/22/04 46 24.45 16.06 7.47 97.97 64422 153517CF0501 2005 Spring 4/28/05 5/4/05 47 13.62 10.97 7.96 82.37 14601 152487CF0502 2005 Summer 7/5/05 7/14/05 36 25.41 15.28 4.25 56.90 13710 213742CF0503 2005 Fall 9/20/05 9/25/05 48 25.46 16.09 7.56 100.68 73046 208243CF0602 2006 Summer 7/5/06 7/11/06 47 24.86 16.61 5.57 74.10 17039 171179CF0603 2006 Fall 10/2/06 10/11/06 50 21.06 17.20 7.71 95.79 74130 330420
71
Table 3.2. Collected numbers, and average catches per unit effort for the duration of the CHESFIMS program
Species Number collected
Total weight
collected
Average CPUE
(#.min-1)
Relative CPUEn
(%)
Average CPUE
(g.min-1)
Relative CPUEw
(%)
Anchoa mitchilli (Bay anchovy) 543912 428113.89 36.5124 89.2207 28.73284 17.6320 Anchoa hepsetus (Striped anchovy) 13544 38159.242 0.9612 2.3488 2.728175 1.6742 Morone americana (White perch) 7566 551131 0.5046 1.2331 36.79013 22.5764 Leiostomus xanthurus (Spot) 6839 260523.6 0.4554 1.1127 17.3496 10.6466 Cynoscion regalis (Weakfish) 6109 164610.3 0.4068 0.9940 10.97347 6.7339 Micropogonias undulatus (Croaker) 5577 491336.5 0.3713 0.9073 32.71208 20.0739 Alosa aestivalis (Blueback herring) 4375 6580.55 0.2912 0.7116 0.438053 0.2688 Brevoortia tyrannus (Atlantic menhaden) 4017 87834.65 0.2675 0.6536 5.864841 3.5990 Alosa sp. (Alosid) 3351 8513.1188 0.2231 0.5451 0.567192 0.3481 Clupeidae (Clupeid) 2554 2167 0.1702 0.4159 0.144285 * Opisthonema oglinum (Atlantic thread herring) 2359 43737 0.1576 0.3850 2.942466 1.8057 Peprilus triacanthus (Butterfish) 807 14437.25 0.0537 0.1312 0.961268 0.5899 Urophycis regia (Spotted hake) 769 8713.65 0.0511 0.1249 0.58007 0.3560 Morone saxatilis (Striped bass) 768 75424.5 0.0512 0.1250 5.048709 3.0982 Peprilus alepidotus (Harvestfish) 731 15638.4 0.0486 0.1188 1.041627 0.6392 Callinectes sapidus (Blue crab) 700 2 0.0466 0.1139 6.66E-05 * Loligo sp. (Loligo squids) 678 1374.7 0.0451 0.1102 0.091508 * Prionotus sp. (Searobin) 624 3018.9 0.0415 0.1014 0.201234 0.1235 Trinectes maculatus (Hogchoker) 577 15711.15 0.0386 * 1.050084 0.6444 Alosa pseudoharengus (Alewife) 410 11071 0.0272 * 0.737017 0.4523 Anchoa (Anchovy) 363 104 0.0241 * 0.006858 * Stenotomus chrysops (Scup) 337 5631.5 0.0226 * 0.377278 0.2315 Menticirrhus saxatilis (Northern kingfish) 304 5450.75 0.0202 * 0.364778 0.2238 Bairdiella chrysoura (Silver perch) 280 11620.5 0.0186 * 0.773602 0.4747 Menidia menidia (Atlantic silverside) 259 390.45 0.0172 * 0.025932 * Pomatomus saltatrix (Bluefish) 182 5398.2 0.0121 * 0.359353 0.2205 Symphurus plagiusa (Blackcheek tonguefish) 145 3264.1 0.0096 * 0.21725 0.1333 Anguilla rostrata (American eel) 132 5202 0.0088 * 0.351283 0.2156 Menticirrhus americanus (southern kingfish) 118 2038.5 0.0078 * 0.135652 * Paralichthys dentatus (Summer flounder) 100 22329.75 0.0066 * 1.486601 0.9123 Scophthalmus aquosus (Windowpane flounder) 80 1415.25 0.0053 * 0.094946 * Centropristis striata (Black sea bass) 56 1672.25 0.0037 * 0.111268 * Etropus microstomus (Smallmouth flounder) 56 304.55 0.0037 * 0.020336 * Ictalurus punctatus (Channel catfish) 48 6561.5 0.0031 * 0.439062 0.2694 NA (Sciaenid) 45 13.94898 0.0030 * 0.000862 * Sphoeroides maculatus (Northern puffer) 43 1214.2 0.0028 * 0.080779 * Ophidion marginatum (Striped cusk-eel) 31 287.5 0.0020 * 0.019131 * Syngnathus fuscus (Northern pipefish) 31 28.7 0.0020 * 0.001848 * Gobiosoma bosc (Naked goby) 28 36.5 0.0018 * 0.002364 * Chaetodipterus faber (Spadefish) 27 2029.5 0.0017 * 0.135053 * Raja eglanteria (Clearnose skate) 27 29136 0.0017 * 1.947106 1.1948 Hippocampus erectus (Lined seahorse) 25 19.01 0.0016 * 0.001199 * Prionotus scitulus (Leopard searobin) 20 299.5 0.0013 * 0.019874 * Etheostoma olmstedi (Tessellated darter) 20 41 0.0013 * 0.002663 * Etrumeus teres (Round herring) 19 56 0.0013 * 0.003844 *
72
Ictalurus nebulosus (Brown bullhead) 18 1240 0.0011 * 0.082732 * Alosa sapidissima (American shad) 15 741 0.0009 * 0.049268 * Atherinidae (Silverside) 14 8.5 0.0009 * 0.000499 * Larimus fasciatus (Banded drum) 14 41.5 0.0009 * 0.002696 * Etropus crossotus (Fringed flounder) 13 41.5 0.0008 * 0.002696 * Dorosoma cepedianum (Gizzard shad) 12 9392.2 0.0007 * 0.625246 0.3837 Scomberomorus maculatus (Spanish mackerel) 12 205.5 0.0007 * 0.013615 * Ameiurus catus (White catfish) 10 829.5 0.0006 * 0.05516 * Orthopristis chrysoptera (Pigfish) 10 486 0.0006 * 0.03229 * Gobiidae (Goby) 10 6 0.0006 * 0.000333 * Citharichthys spilopterus (Bay whiff) 10 259 0.0006 * 0.017177 * Synodus foetens (Inshore lizardfish) 9 775 0.0005 * 0.051531 * Prionotus evolans (striped searobin) 9 207 0.0005 * 0.013715 * Alosa mediocris (Hickory shad) 8 689 0.0005 * 0.045806 * Rhinoptera bonasus (Cownose ray) 7 66601 0.0004 * 4.434088 2.7210 Opsanus tau (Oyster toadfish) 7 392 0.0004 * 0.026032 * Clupea (Atlantic herring) 6 21 0.0003 * 0.001332 * Astroscopus guttatus (Northern stargazer) 6 67 0.0003 * 0.004394 * Bothidae (Lefteye flounder) 6 198 0.0003 * 0.013116 * Dasyatis sabina (Atlantic stingray) 4 10201 0.0002 * 0.679095 0.4167 Selene vomer (lookdown) 4 40 0.0002 * 0.002597 * Lagodon rhomboides (Pinfish) 4 86 0.0002 * 0.005659 * Archosargus probatocephalus (Sheepshead) 4 61 0.0002 * 0.003995 * Pogonias cromis (Black drum) 4 5661 0.0002 * 0.377584 0.2317 Gymnura altavela (Spiny butterfly ray) 3 2876 0.0001 * 0.191411 0.1175 Notropus hudsonius (Spottail shiner) 3 17 0.0001 * 0.001065 * Hyporhamphus unifasciatus (Halfbeak) 3 23.5 0.0001 * 0.001498 * Strongylura marina (Atlantic needlefish) 3 266 0.0001 * 0.017643 * Membras martinica (Rough silverside) 3 10.5 0.0001 * 0.000632 * Prionotus carolinus (northern searobin) 3 115 0.0001 * 0.00759 * Perca flavescens (Yellow perch) 3 27 0.0001 * 0.001731 * Selene setapinnis (Atlantic moonfish) 3 21 0.0001 * 0.001332 * Squalidae (Dogfish) 2 1451 0.0001 * 0.096538 * Squalus acanthias (Spiny dogfish) 2 421 0.0001 * 0.027963 * Dasyatis americana (Southern stingray) 2 1 0.0001 * 0 * Notemigonus crysoleucas (Golden shiner) 2 2 0.0001 * 6.66E-05 * Ictaluridae (Freshwater catfish) 2 189 0.0001 * 0.012517 * Ictalurus furcatus (Blue catfish) 2 179 0.0001 * 0.011851 * Ameiurus natalis (Yellow bullhead) 2 56 0.0001 * 0.003662 * Lepomis gibbosus (Pumpkinseed sunfish) 2 2 0.0001 * 6.66E-05 * Caranx crysos (Blue runner) 2 52 0.0001 * 0.003395 * Trachinotus carolinus (florida pompano) 2 64 0.0001 * 0.004194 * Cynoscion nebulosus (Spotted seatrout) 2 94 0.0001 * 0.006192 * NA (Blennies) 2 1 0.0001 * 0 * Trichurus lepturus (Atlantic cutlassfish) 2 94 0.0001 * 0.006192 * Monocanthus hispidus (Planehead filefish) 2 2 0.0001 * 6.66E-05 * Myliobatis freminvillii (bullnose ray) 2 951 0.0001 * 0.063249 * Chilomycterus schoepfii (Striped burrfish) 2 381 0.0001 * 0.0253 * Sphyraena borealis (Northern sennet) 2 2 0.0001 * 7.01E-05 *
73
Table 3.3. Results of canonical correspondence analysis of station level average environmental conditions, and the composition of the catch
Source Mean Square Contingency
Proportion
Total 5.497 1 Constrained 0.541 0.098 Unconstrained 4.956 0.902 Unconstrained Eigenvalues Axis CCA1 CCA2 CCA3 CCA4 CCA5 Eigenvalue 0.631 0.505 0.4425 0.378 0.342 Percent variation explained
11.4% 9.18 8.05 6.88 6.21
Tests of Factor Significance Source DF Chi Sq F Pr(>F) Region 1 0.138 2.598 <0.002 Season 1 0.163 3.058 <0.002 Year 1 0.0314 0.589 0.008 Salinity 1 0.0931 1.747 <0.002 Temperature 1 0.0635 1.191 <0.002 Oxygen conc 1 0.0278 0.522 0.028 Transmissivity 1 0.0244 0.457 0.020 Residual 93 4.956
74
Figure 3.1. Scatterplot of baywide average physical conditions for the 17 CHESFIMS cruises. Lines are loess fits to the data, and are shown for illustration only.
0 5 10 15 20 25
05
1015
2025
AvgOfTemp
05
1015
2025
AvgOfSalinity
0 5 10 15 20 25
05
1015
2025
0 5 10 15 20 25 0 5 10 15 20 25
05
1015
2025
AvgOfOxygen
Cruise Averages
75
Figure 3.2 The distribution of stations visited during CHESFIMS from Spring 2001 – Autumn 2006.
76
0
10
20
30
40
50
60
70
80
90
100
0 2 4 10 30 50 70 90 200
400
600
800
1000
3000
5000
7000
9000
Numerical catch
Num
ber o
f sta
tions
Fig. 3.3. The distribution of fish per tow in CHESFIMS survey tows from Spring 2001 – Autumn 2006
77
Fig 3.4. Annual species composition in fish biomass (CPUE – g.min-1) based on MWT collections during CHESFIMS cruises from Spring 2001- Autumn 2006
2001
Weakfish
Striped bass
Spot
Bay anchovy
White perch
Croaker
HogchokerStriped anchovy
AlosidOther
2002
Weakfish
Striped bass
Spot
Striped anchovy
Bay anchovy
White perch
Croaker
Atlantic thread herring
Atlantic menhaden
Other
2003
Weakfish
Striped bass
Spot
Croaker
Bay anchovy
White perch
Atlantic menhaden
Blueback herring
Clearnose skate
Other 2004
Weakfish
Striped bass
Spot Croaker
White perch
Bay anchovy
Atlantic menhaden
Summer flounder
Atlantic thread herring
Other
2005
Weakfish
Spot
Cownose ray
Atlantic menhaden
Atlantic stingray
Clearnose skate
Bay anchovy
Croaker
White perch
Other 2006
Weakfish
Striped bass
Spot
Atlantic menhaden
Cownose raySilver perch
Bay anchovy
White perch
Croaker
Other
78
Fig 3.5. Seasonal species composition in fish biomass (CPUE – g.min-1) based on MWT collections during CHESFIMS cruises from Spring 2001- Autumn 2006
Spring
Atlantic menhaden
Weakfish
Clearnose skate
OtherSpotted hake
Striped bass
Bay anchovy
Alewife
Croaker
White perch
Summer
Atlantic menhaden
Weakfish
Spot
Other
White perch
Croaker
Cownose ray
Bay anchovyStriped bass
Atlantic thread herring
Fall
Atlantic menhaden
Weakfish
Spot
OtherStriped anchovy
Striped bass
Bay anchovy
Atlantic thread herring
Croaker
White perch
79
Fig 3.6. Regional species composition in fish biomass (CPUE – g.min-1) based on MWT collections during CHESFIMS cruises from Spring 2001- Autumn 2006
Upper bay
Striped bass
Alewife
Gizzard shad
OtherBay anchovyCroaker
Cownose ray
Spot
White perch
Atlantic menhaden
Mid bay
Striped anchovy
Cownose ray
Silver perch
OtherAtlantic menhaden
Spot
Weakfish
Striped bass
Croaker
Bay anchovy
Lower bay
Atlantic thread herring
Striped anchovy
Clearnose skate
OtherAtlantic menhaden
Spot
Weakfish
Summer flounder
Croaker
Bay anchovy
80
Fig. 3.7. The distribution of species richness based on MWT collections during CHESFIMS cruises from Spring 2001- Autumn 2006
Number of species
Num
ber
of s
tatio
ns
0 5 10 15 20 25
020
4060
8010
0
81
Fig. 3.8 The distribution of species richness per cruise based on MWT collections during CHESFIMS cruises from Spring 2001- Autumn 2006
82
1 10 100
20
40
60
80
100
Cum
ulat
ive
Dom
inan
ce%
CF0101CF0102CF0103CF0201CF0202CF0203CF0301CF0302CF0303CF0401CF0402CF0403CF0501CF0502CF0503CF0602CF0603
A) Abundance ranks
1 10 100Species rank
20
40
60
80
100
Cum
ulat
ive
Dom
inan
ce%
B) Biomass ranks
Fig. 3.9 K-dominance curves for A) abundance and B) biomass based on MWT collections during CHESFIMS cruises from Spring 2001- Autumn 2006. Note data are color-coded by season: springtime cruises are shown in blue, summer cruises are shown in red, and autumn cruises in orange. All cruises conducted in the same year have the same symbol.
83
A) Abundance ranks
B) Biomass ranks
1 10 100Species rank
80
85
90
95
100C
umul
ativ
eD
omin
ance
%LowerMidUpper
1 10 100Species rank
20
40
60
80
100
Cum
ulat
ive
Dom
inan
ce%
LowerMidUpper
Fig. 3.10 K-dominance curves by region for A) abundance and B) biomass based on MWT collections during CHESFIMS cruises from Spring 2001- Autumn 2006. Note data are color-coded by region: upper bay data are shown in turquoise, mid bay in blue, and lower bay in green.
84
Fig. 3.11. Results of a canonical correlation analysis of CHESFIMS MWT and CTD data. Symbols represent individual MWT hauls
85
Figure 3.12. Relationship between baywide average catch per unit effort (g.min-1) and depth weighted dissolved oxygen for A) July and B) September. The relationship for July was CPUE = 497.36 – 64.08* Oxygen, n=5, r^2=-0.15, p=0.539. The relationship for September was CPUE = -719.46 +125.31 * Oxygen, n=6, r2=0.49, p=0.07.
86
Chapter 4
FISHERY-INDEPENDENT MULTISPECIES ASSESSMENT OF THE FISH ASSEMBLAGE
IN THE PATUXENT RIVER, MD
Thomas J. Miller and David A. Loewensteiner
Chesapeake Biological Laboratory
University of Maryland Center for Environmental Science
Solomons, MD 20688
87
4.1 INTRODUCTION
Fisheries in the Chesapeake Bay contribute significantly to U.S. catches at the
national and regional levels. Recent National Marine Fisheries Service (NMFS) statistics
indicate that between 250,000 - 350,000 metric tonnes (t) of fish and shellfish are
harvested annually from Chesapeake Bay waters, with a dockside value of more than
$100 Million (Miller 2006b). Maintaining the health of these fisheries is an important
but difficult task given the considerable interannual variability in catches of component
species. Scientists, managers and the general public recognize that species targeted by
fisheries are components of complex ecosystems, whereby the removal of individual
targeted species may have ramifications for the entire system (Kaiser and Jennings 2002).
These changes in ecosystem function or health may impact the sustainability and
profitability of multiple fisheries simultaneously. Yet, fisheries management techniques
were developed for the single species case (Smith 1994). This is true of exploited species
in the Chesapeake Bay for which all exploited species are managed on a single species
basis under the aegis of The 1987 Chesapeake Bay Agreement.
In 1996, Miller et al. conducted a literature review and synthesis of information to
determine whether the forces that have motivated the development of multispecies
approaches in other regions were present in the Chesapeake Bay. They concluded that
both biological and technical interactions were present in Chesapeake Bay fisheries.
Several species harvested from Chesapeake Bay are either potential competitors (e.g.,
striped bass Morone saxatilis, bluefish Pomotomus saltatrix, and the weakfish Cyonscion
regalis) or predators and prey (e.g., striped bass and menhaden Brevoortia tyrannus and
88
Atlantic croaker Micropogonias undulatus), and thus their fisheries may experience
biological interactions. Additionally, several fisheries, notably the pound net, are
inherently multispecies by nature, and thus technical interactions also clearly exist in
Chesapeake Bay (Miller et al. 1996).
Houde et al. (1998) reported the recommendations of workshop to explore the
utility and advisability of adopting multispecies approaches in Chesapeake Bay. The
workshop report indicated that adoption of multispecies fisheries management would
bring the regulation of fisheries in line with the ecosystem-level focus of the Chesapeake
Bay Program. Many participants felt that explicit recognition of species interactions,
bycatch concerns, competing users, and habitat issues is needed to move fisheries toward
a multispecies management ideal that will be desirable in the future. Several important
conclusions were reached during the workshop. The workshop participants
recommended that coordinated, baywide surveys be performed to estimate key species
abundances and that biological data on both economically and ecologically important
species that are currently lacking be obtained (Houde et al. 1998). These surveys should
form a vehicle within which to estimate the temporal and spatial dynamics of key
predator-prey relationships and trophic interactions (Houde et al. 1998).
Several fishery-independent surveys for the assessments of important fish and
shellfish stocks in the Chesapeake Bay are currently ongoing. The Virginia Trawl survey
has been conducted and regularly expanded since the 1950's, so that it now encompasses
all of the main tributaries and the mainstem of the Chesapeake Bay in Virginia. The data
89
are widely available (www.fisheries.vims.edu/vimstrawldata/), and have been utilized in
the assessment of several key species. Similarly, striped bass seine surveys have been
conducted since 1954. These surveys conducted in both Maryland and Virginia focus on
sampling recruiting juvenile striped bass in tributaries to the Chesapeake Bay. However,
other species, common in the littoral zone, are also sampled in this survey. In response to
gaps in survey coverage, the NOAA Chesapeake Bay Office funded two, Baywide
surveys. The Chesapeake Bay Fishery Independent Multispecies Survey (CHESFIMS)
built on from an early NSF-funded study of the fish production in the Chesapeake Bay.
CHESFIMS. This survey samples all species of fish in the bentho-pelagic realm, but
focuses on anchovy and young-of-year of commercially important species. Likewise, the
Chesapeake Bay Multispecies Monitoring and Assessment Program (CHESMAP -
http://www.fisheries.vims.edu/chesmmap/) samples the fish community on a baywide
basis, but focuses on commercially-important sizes of fish. Together these two programs
provide a synoptic view of the entire fish community in the mainstem of the Chesapeake
Bay.
The tributaries of the Chesapeake Bay are important features for the ecology of
many fishes that utilize the Bay. Anadromous fish utilize the tributaries for spawner and
they serve as important nursery areas for many such species (Secor and Houde 1995).
Other species use the tributaries are feeding areas later in the year (e.g. menhaden and
Bay anchovy – Thomas Miller, pers. Comm). Tributaries in the Virginia portion of the
Chesapeake Bay are extensively sampled as a part of the VIMS trawl program. This
sampling platform has lead to considerable insight into the dynamics of fish in
90
Chesapeake Bay tributaries. However, no systematic effort has been mounted to assess
the importance of tributaries in the Maryland portion of Chesapeake Bay. Maryland
tributaries are sampled as a part of the juvenile striped bass survey, but this focuses on
shallow littoral habitat. Earlier work associated with the permitting of power generation
plants on several tributaries do provide data on the fish community and patterns of habitat
usage in the 1960s. For example, McErlean et al. (1973) report the results of an
extensive four year sampling program involving seines, beam trawls and otter trawls in
the Patuxent River. They report clear cycles in seasonal diversity in the River during the
period of study (Summer 1963 – Spring 1967), with peaks in diversity in the Spring
associated with the immigration of anadromous speces, and lower diversity in the Winter.
Even though these studies were extensive, they focused on patterns in fish abundance and
did not provide an ecosystem context for their work.
Here we report on a one-year study of the fish community in the Patuxent River,
MD. The objectives of the study were to conduct frequent sampling of the fish
community, using gear that parallel those being used in the mainstem of the Chesapeake
Bay by the CHESFIMS survey, to quantify patterns of movement, production and trophic
dependencies in the river, and between the river and the mainstem. Thus the study was
termed PAXFIMS. The objectives of the program were (i) to conduct a temporally-
intensive survey pelagic and epi-benthic fish community, focusing on young (juveniles,
and yearling) fishes in the Patuxent River and (ii) to determine trophic interactions
among key components of the pelagic fish community, and examine the implication of
the relationships uncovered in empirical studies using bioenergetic modeling.
91
4.2 METHODS 4.2.1. Field Methods
Survey work was conducted year round from January 2004 – February 2005.
Cruise scheduling was based on the thermal regime in the river (Fig. 4.1). The average
temperature cycle in the mesohaline section of the river varies from a low of 2.6 C in the
fifth week of the year (February) to 27.66 C in the 31st week of the year (August). When
the expected daily temperature change in the river is plotted, two broad peak rates of
temperature change are apparent (Fig. 4.2). Using the 2 oC.day-1 isoline as a criterion,
the weeks of peak change are between weeks 9 - 22 in the spring, and between weeks 39
- 49 in the autumn / winter. Accordingly the 14 planned surveys were scheduled to focus
most heavily on these times when temperature changes are most rapid. This scheme
resulted in more frequent surveys in the periods of rapid change during spring and
autumn when the system is expected to change most rapidly. All surveys were conducted
from the University of Maryland Center for Environmental Sciences’ RV Aquarius. An
engine failure on the fourth cruise forced us to abandon the cruise. Delays produced by
the necessary repairs precluded any rescheduling of ship time.
All surveys were conducted from the University of Maryland Center for
Environmental Sciences’ RV Aquarius. Each survey was conducted as a completely
randomized design. Twenty stations were randomly selected based on river kilometer,
and distributed from Solomons to Abingdon Cove. On each survey, we sampled both the
pelagic fish community using a midwater trawl, and the benthic community using an otter
trawl. We sampled up to 20 benthic stations, and a subset of up to 10 of pelagic stations.
92
The benthic component of the survey was conducted during daylight hours (~0600 –
1700) of the first day, and the midwater trawl was deployed during nighttime (~1900 –
0500) on the evening of the second day. The midwater trawl had a mouth opening of 5.7
x 4 m, an 3.8 – 7.6 mm mesh body , and an 3 mm mesh cod-end. The net was
instrumented with depth-temperature loggers on the head rope and the foot rope during
all deployments. These data were used subsequently to calculate the deployment profile
and average net opening for all deployments. The midwater trawl was fished in a single,
stepped oblique tow for 10 minutes. Where station depth permitted the net was held at
each of 10 depth intervals for 1 minute, to ensure the entire water column was sampled.
At shallow stations the number of depth intervals was adjusted according to conditions.
The benthic survey employed a 9m semi-ballon otter trawl with 38 mm mesh in the body
and 7mm mesh in the cod end. The trawl was equipped with a tickler chain. This net
was fished for 5 minutes on the bottom at 1kn. Following all deployments the water
column was profiled with a SEABIRD SBE 25 CTD which recorded temperature,
salinity, depth, fluorescence profiles for the station.
All fish and crabs collected in the the midwater and otter trawls were identified,
enumerated, and the total weight for each species recorded. For species with high levels
of abundance, random sub-samples of between 30-100 individuals were measured for
length and preserved on ice for subsequent dietary analysis. For less abundant species,
the entire catch was measured for length and preserved. All fish were handled according
to procedures approved by the University of Maryland Center for Environmental
Sciences’ Animal Care and Use Committee.
93
4.2.2. Laboratory Methods
In the laboratory, individual fish were thawed and their individual length and
fresh weight determined. The stomach (between the oesaphagus and the pyloric
sphincter) was removed and weighed. A qualitative assessment of the stomach fullness
was made. The stomach was then opened, and its contents transferred to a clean glass
Petri dish. The empty stomach was reweighed. The stomach contents were identified the
lowest taxonomic level practical. For large prey items, the weight and size of individual
prey were determined. For smaller prey items, the bulk weight and where possible to the
total number of prey were determined.
For selected species the age of fish in the sub-sample was also determined.
Ageing was based on otolith section readings. For fish assumed to be age 1+ the otolith
was mounted in Spurr resin, sectioned through the core using a South Bay Technologies
Wafering Saw to yield a thin ( ~ 1mm) section. The number of annulae in the section
was determined under a dissecting microscope at 40X magnification. Each otolith was
read twice, blind to improve aging accuracy. For YOY fish, individual otoliths were
mounted in Spurr resin, sectioned through the core, mounted and polished using a graded
series of polishing media and read at 400 – 1000X under a compound microscope.
4.2.3. Statistical Analysis
4.2.3.1 Abundance and distribution Catches per station for individual species were converted to catch per unit effort
(CPUE) by dividing either the catch in number or weight by the duration of the tow.
94
Mean (±SD) CPUEs were calculated for each survey as the simple average of station
CPUE. A general linear model was fit to the CPUEs for each species to assess whether
CPUEs were significantly different among cruises. Station-specific CPUEs were
considered independent observations.
Two approaches were used to quantify the distribution of each species. Simple
bubble plots were generated for each species where the area of the bubble is directly
proportional to CPUE at the station. The plots were inspected visually to describe
changes in the spatial distribution of each species. Additionally, two-stage general linear
models were used to quantify the impacts of temperature, salinity and dissolved oxygen
on the distribution of each species (Jensen et al. 2005). This modeling approach identifies
the effect of the environmental variables on predicting presence (1st stage) and
subsequently on predicting abundance given presence (2nd stage). The model uses
penalized regression splines to describe the response of either presence or abundance
given presence (Wood and Augustin 2002). Independent fits are estimated for each stage
of the model. Thus, each environmental variable can affect presence and abundance
differently. The first stage of the model can be described as
( ) ( ) ( ) ε+++= DOssalinitysetemperaturspresenceP 321)(
Where the s’s are spline fits for each environmental variable. Similarly the
second stage of the model is given by
( ) ( ) ε+++= DOssalinitysetemperatursAbundance 654 )(
95
The individual s terms reflect the degree of non-linearity in the responses.
Values of s=1 indicate a linear fit, values greater than 1 indicate increasing non-
linearities.
4.2.3.2. Size distributions and growth
Estimates of individual total lengths, measured on the vessel, were used to
construct size distributions. Distributions were developed for individual cruises, without
regard to the location of fish within the river on that cruise. Distributions were developed
for midwater and bottom trawls separately where possible.
Length frequency distributions were analyzed using model-based, hierarchical
clustering as an approach to modal analysis. The analysis identified the mixture of
Gaussian distributions that best described the observed length frequency. An iterative
expectation-maximization method was used to provides the maximum-likelihood
estimate of a suite of parameterized Gaussian mixture models. The algorithm used a
Bayesian Information Criterion to determine the number of distributions in the mixture.
The algorithm was implemented using the EMClust package within R.
Based on the fitted mixture model, the modes from individual cruises were related
longitudinally to provide estimates of growth. Mean sizes of appropriate modes were
plotted against the mid-date of the cruise to develop approximate growth curves.
96
4.3 RESULTS
We completed 11 cruises over a 12 month period. One of the cruises had
to be abandoned because of an engine failure on the research vessel (Table 4.1). Given
the way in which the sampling was tied to the thermal regime in the river, it was not
possible to reschedule this cruise.
4.3.1 Abundance and distribution
Bay anchovy was the most abundant species collected in trawls during the
PAXFIMS cruises. The average catch per unit effort in the midwater trawl was 26.2
fish.min-1 (19.9 g.min-1). The equivalent catches in the bottom trawl were substantially
lower (3.45 fish.min-1; 5.66 g.min-1) reflecting differences in selectivity between the two
gear. Catches in both gear varied over the course of the year. Figure 4.4 shows the
seasonal pattern of CPUEs in the midwater trawl. Patterns were similar, although lower
in the bottom trawl. CPUEs were significantly different in the midwater trawl
(F9,109=13.28, p<0.001) and bottom trawl (F10,158=3.46, p<0.004). The increase in
abundance evident from late summer to late autumn represent the recruitment of anchovy
spawned that summer to the gear.
The seasonal distribution of bay anchovy indicated a broad distribution
from April – December (Fig. 4.5). In the initial winter surveys, bay anchovy was
confined to the stations near the confluence of the Patuxent River and the Chesapeake
Bay (Fig. 4.5). Recruitment of the new year class was evident in the distributions, but
recruitment did not appear to be spatially concentrated as it was in white perch. Bay
97
anchovy remained abundant in midwater trawl catches throughout the remained of the
year.
Application of two-stage GAMs to the distribution of bay anchovy indicated
significant effects of environmental variables on the distribution of bay anchovy in the
Patuxent River. The first stage GAM which seeks to predict presence of bay anchovy at
individual locations. Results indicated significant contributions of bottom temperature,
salinity and dissolved oxygen to forecasting presence (Table 4.2). Overall the model
37.3% of the deviation in the data. Spline fits for individual environmental parameters
were variable (Fig 4.6). The fit for temperature indicated bay anchovy abundance was a
linear increasing function of temperature, but a linear decreasing function of salinity.
There was an indication of a broad peak in abundance with respect to dissolved oxygen
of approximately 7-8 mg/L. (Fig 4.6).
The second stage of the GAM for bay anchovy seeks to predict abundance given
presence in the sample. All three environmental parameters were again assessed for their
contribution to the predictive ability of the mode. Results indicated a significant effect of
temperature and dissolved oxygen, but not of salinity. Overall the model 37.9% of the
deviation in the data. Spline fits for individual environmental parameters were variable
(Fig 4.7). The fit for temperature indicated bay anchovy abundance was a linear
increasing function of temperature, but a linear decreasing function of salinity. There
was an indication of a broad peak in abundance with respect to dissolved oxygen of
approximately 7-8 mg/L. (Fig 4.7).
98
The GAM analysis of the bottom trawl data was used to predict catches in the
midwater trawl to examine the forecasting ability of the GAM. The forecasting ability of
the GAM should be judged by the r2 value rather than a direct comparison of observed
and predicted values. The Pearson correlation between observed and predicted was
r=0.43, indicated that the GAM could explain just over half of the variability in the
observed catches of white perch in the midwater trawl. The relationship between the two
variables is shown in Figure 4.8.
White perch was the next most abundant species collected in trawls during the
PAXFIMS cruises. White perch CPUEs were higher in the bottom trawl than in the
miwater trawl. The average catch per unit effort in the bottom trawl was 9.68 fish.min-1
(608.6 g.min-1). The equivalent catches in themidwater trawl were substantially lower
(3.55 fish.min-1; 261.1 g.min-1) reflecting differences in selectivity between the two
gear. Catches in both gear varied over the course of the year. Figure 4.9 shows the
seasonal pattern of CPUEs in the midwater trawl. There was no clear seasonal pattern in
total CPUE either expressed in number or weight. Although, because the midwater trawl
selects for smaller sized individuals, this gear was able to detect the pulse of biomass of
white perch recruits (Fig. 4.10). CPUEs were not significantly different through the
season either in the midwater trawl (F9,109=0.82, p<0.6) and bottom trawl (F10,77=0.96,
p<0.4).
The distribution of white perch changed over the course of the surveys (Fig 4.10).
Initially, the perch were distributed downriver toward its confluence with the Chesapeake
99
Bay. By March, adults had started to move upstream. The center of mass of the perch
distribution remained in the upper portion of the sampling domain until reproduction in
early summer. Young-of-year white perch gradually became dispersed throughout the
Patuxent, until the species was almost uniformly distributed throughout the river but early
Autumn.
Application of two-stage GAMs to the distribution of white perch indicated
significant effects of environmental variabiles on the distribution of white perch in the
Patuxent River. The first stage, which seeks to predict presence of white perch in bottom
trawl catches, indicated significant contributions of bottom temperature and salinity to
forecasting presence, but not of bottom oxygen. Overall the model explained 31.8% of
the deviation in the data. Spline fits for individual environmental parameters were
variable (Fig 4.11). The fit for temperature indicated a peak of predicted presence at
about 10 C. Spline fits forecasting white perch presence for salinity was linear, with
higher probabilities of presence at lower salinities (Fig 4.11).
The second stage of the GAM for white perch seeks to predict abundance given
presence in the sample. All three environmental parameters were again assessed for their
contribution to the predictive ability of the mode. Results indicated a significant effect of
both temperature and salinity, but no significant effect of dissolved oxygen. Overall the
model did not explain a substantial fraction of the deviation in the data. Spline fits for
individual parameters were broadly similar to those observed in stage I, with a non-linear
spline for temperature, and broadly linear effects for salinity and oxygen (Figure 4.12)
100
The GAM analysis of the bottom trawl data was used to predict catches in the
midwater trawl to examine the forecasting ability of the GAM. The forecasting ability of
the GAM should be judged by the r2 value rather than a direct comparison of observed
and predicted values. The Pearson correlation between observed and predicted was
r=0.51, indicated that the GAM could explain just over half of the variability in the
observed catches of white perch in the midwater trawl. The relationship between the two
variables is shown in Figure 4.13.
4.3.2 Size distributions and growth
The E-M clustering algorithm provided an adequate fit to the observed length
frequency distributions in the midwater trawl (Fig. 4.14). The algorithm identified a
single mode in the length frequency distributions for PF0402, PF0403 and PF0406. A
second mode was identified in PF0405, but it was not felt to be reliable as it accounted
for only 3% of the observed catch (Table 4.2). Thus, overall the model represented the
bay anchovy population as being comprised by a single age class from March – June.
This is in line with what is known about the reproductive biology of bay anchovy. The
average of the single mode increased from 44.6 mm TL on March 4, 2004 to 66 mm TL
on June 16, 2004 (Table 4.2). These data suggest a growth rate for this cohort during this
period of 0.21 mm.day-1.
Subsequently, the algorithm identified two clear modes in the length
frequency distributions for cruises PF0407 - PF0411 (Fig. 4.14). The smaller size mode
101
reflects the recruitment of young of year bay anchovy, likely spawned in the mainstem of
the bay in June and July, to the survey gear. This mode was first observed in August
when it comprised 86% of the observed length frequency distribution. The proportion of
the population represented by this mode remained approximately constant throughout the
remainder of the survey. The size of this mode increased from 39.6 mm TL in August to
49.4 mm TL in November, yielding an estimated growth rate of 0.11mm.day-1.
Length frequency fits for bay anchovy collected in the bottom trawls were not as
compelling as those observed for the midwater trawl. For bottom trawl length
frequencies, insufficient numbers of fish were caught in PF0402, PF0403, and PF0412.
Even in other cruises, abundances were low compared to midwater trawl catches.
Accordingly, no further analysis of these data will be presented.
Length frequency fits to white perch data from the bottom trawl and midwater
trawl were similar. The bottom trawl length frequencies will be the focus of the
discussion that follows, and midwater trawl data will only be referenced if it contrasts
with or provides additional support for the bottom trawl patterns.. The E-M clustering
algorithm provided an adequate fit to the observed length frequency distributions in the
bottom trawl (Fig. 4.15). The algorithm identified a minimum of two distributions in all
of the length frequencies, except that for PF0410 (Table 4.3). This was similar in the
midwater trawl length frequency modeling, except that it was PF0409 that was found to
comprise a single mode. A third, smaller mode was initially identified in PF0405 in the
midwater trawl and in PF0406 in the bottom trawl length frequencies (early June). This
mode was detected intermittently until the end of the surveys in December. It represents
the production and recruitment of young of perch to the river. The size modes that
102
dominate length frequencies for the bulk of the sampling reflect separate age classes.
The smaller of the two modes (mean size =88mm TL in March) are the survivors from
the 2003 year class. The larger and broader size mode (mean size = 177mm TL in
March) likely represents a mixture of age classes from earlier years.
Individual size modes were assigned to putative year class to permit a
rudimentary growth analysis (Fig. 4.16). Simple linear regressions fitted to size at date of
capture data were used to estimate daily growth rates. These analyses indicated that the
Age-0 fish (2004 year class) were likely growing at 0.23 mm.day-1 from late summer to
early winter. The age-1 fish (2003 year class) appeared to be growing approximately
linearly throughout the year. Linear regression of length on date of capture provides an
estimated growth of 0.19 mm.day-1. The mixed age-2+ mode shows no steady increase
in size during the survey, inidicating that these fish have attained approximate maximum
size.
4.3.3. Multivariate Analyses
A non-metric multidimensional scaling analysis was able to detect
differences in the fish community caught in the midwater trawl. The two-dimensional
ordination had a low stress value (Fig. 4.17), indicating that there were clear groupings,
but the fact that the stress value was not equal to zero indicates that there residual
variation, not explained by the ordination still remains. Inspection of Fig. 4.17 reveals a
clear seasonality in the ordination of stations, suggesting a clear seasonality in the
midwater trawl catches. In general “winter” cruises, i.e., those conducted in March, April
and December are clustered in the lower left of Fig. 4.17, whereas, “summer” stations are
103
in the top right. A cluster analysis of these data however, did not indicate distinct
groupings of stations by cruise, suggesting the existence of a substantial amount of non
seasonal pattern in the data set. Some of this variability can be explained by patterns in
the catches of individual species. For example, Fig 4.18 shows the CPUE of bay anchovy
overlain on the nMDS ordination. From this plot it is clear that stations with a high
anchovy catch are found in the upper right corner of the ordination. A similar pattern
results for other species with higher salinity affinities, e.g., croaker and weakfish.
Conformation plots using white perch and hogchoker (Fig. 4.19) as representative of
freshwater and mesohaline end members further clarify that salinity is another dominant
response in the nMDS ordination. A broadly similar pattern was evident in an nMDS
analysis of the catches in the bottom trawl portion of the survey, and are thus not
presented here.
To further explore the pattern in the distribution of catches a PCA was used. The
first three axes of the ordination explained 88% of the variability in the data (Table 4.5 ).
The loadings indicate that temperature was strongly positively correlated with the first
PCA axis, whereas salinity exhibited no correlation with this axis. Conversely, salinity
was highly correlated with the second principal component axis, whereas temperature
exhibited only a weak correlation with this axis. The relationships between the data and
their loadings is exhibited in Figure 4.20
104
4.4 DISCUSSION
We conducted a year-long survey of the fish community in the Patuxent
River as part of an initiative to support the increased sampling needs for ecosystem-based
approaches to fisheries. The results of the work indicate the potential for a single
multispecies survey to replace on going single species approaches. However, given that
we have only a single year of data, further parallel sampling would be required to
determine whether the temporal pattern in existing single species surveys is replicated in
the multispecies approaches.
As expected, there was a strong seasonal pattern in survey catches. The
seasonality was evident in time series of catches of individual species (such as bay
anchovy and white perch) and in the multivariate analysis where water temperature
proved to be the dominate factor in explaining the ordination of survey catches. In
most cases the seasonal response evident in the data resulted from recruitment of juvenile
fish to either the midwater or bottom trawls, as evidenced by both the bay anchovy and
white perch data. The second most important factor explaining the distribution of catches
was salinity, or river mile. Again, this should not be viewed as a surprising factor given
the strong salinity gradients that characterize estuarine habitats.
Thirty-three different species of fish were collected during the survey overall.
Species ranged from freshwater residents such as pumpinkseed sunfish (Lepomis
gibbosus to marine end members such as the harvest fish (Peprillus alepidotus).
However, although the diversity was relatively high,the distribution of abundance was
not even. In midwater trawl catches, the top ten species accounted for 99% of total
105
average catch per effort. Within this group, bay anchovy accounted for 81% of the
average CPUE in the midwater trawl. Only white perch, which accounted for 10% of the
remaining average CPUE was anywhere near as abundant. The eight other relatively
abundant species were hogchoker, coraker, blue crab, spot, weakfish, striped bass and
brown bullhead. The pattern was similar in the bottom trawl component of the survey,
although less marked. In bottom trawl catches, the top ten species still accounted for
98% of the total average CPUE. However, the most dominant species in bottom trawl
catches, white perch, only accounted for 58.1% of the total average CPUE. Bay anchovy
accounted for an additional 21.5% of the total average CPUE – together combining for
79% of the total average CPUE. Hogchoker, Atlantic croaker, blue crab, spot, weakfish,
striped bass and brown bullhead completed the top ten most abundant species in the
bottom trawl survey. .
We employed a complemented survey involving both a midwater trawl and a
bottom trawl. The intent of using both gear was to fully capture the entire fish
community in the River. In previous surveys where only a single gear had been used,
concern had been expressed over gear avoidance. Concerns similar to these are what
motivated McEarlean et al to use a bottom trawl, beam trawl and seine in their survey of
the Patuxent River (McErlean et al. 1973). However, as can be seen by the summary
provided above, the relative indices of community composition were not sensitive to the
gear used. Thus, if one seeks only to develop a community metric to serve as an
indicator of overall status a single gear will likely suffice.
106
We explored the application of general additive models to predict the
distribution and abundance of individual fish species (Jensen et al. 2005). In general
these models had relatively low predictive power. It is likely that the relatively linear
nature of the Patuxent River system accounts for the lack of success of these applications,
which have been successful elsewhere. GAMs were dominated by relatively linear
salinity and temperature terms, reflective of the one dimensional nature of the
environment. However, the results of the model also indicated the broad similarity of the
samples in the midwater and bottom trawls.
We were more successful in using multi-modal size decomposition algorithms to
provide estimates of the growth of the principal species collected in the survey. The
approach used did not rely on directly ageing individual fish, but rather makes the
assumption that the modal assignments remain constant through time, so that one may
use the time different between samples, and modal means as surrogates for size at age.
Using this assumption, our analysis showed the presence of two distinct cohorts of
croaker recruiting to the Patuxent River in 2004. It is not clear whether this is a
consistent feature of recruitment of croaker to the river or simple a feature of the pattern
for 2004. However, it is known that the recruitment of croaker to the Chesapeake Bay
system generally is tied to meterology (Montane and Austin 2005), suggesting that the
presence of multiple recruiting cohorts may be a common feature of the system. The
growth rates of the two cohorts appeared to differ substantially. The first cohort grew at
approximately 1 mm.day, whereas the later cohort appeared to have recruited to the river
and ceased growing perhaps due to declining water temperatures. The modal size
107
analysis approach was also capable of estimating growth rates of different year classes in
some species as well as in different cohorts. For example, in white perch we were able to
identify three different age classes, age-0, age-1 and age 2+ as well as estimate growth
rates for each year class. The youngest age class was growing most rapidly (g=0.231
/day), whereas the oldest age class had a very low growth rate (g=0.06 /d). These
estimates are in line with what is known of the growth of this species.
The survey was designed to sample more often during times of rapid temperature
change in the river, as it was suspected that this was when the community would be
changing most rapidly. The sampling scheme did not appear to provide any advantage in
terms of increased resolution in community structure. In part, this may have resulted
from the vessel engine failure in the proposed PF04 survey in mid May. This survey may
have provided a clearer pattern. However, it is more likely that the dynamics of the
system are driven by the production arising during summer and early autumn than by
changes in temperature per se. This would suggest that subsequent surveys concentrate
more effort in the July – September period, reducing effort in the early spring months.
During these early surveys catches and diversity were low and the effort did not provide
much resolution of community structure. Investment of some of the effort expended in
early cruises to increase the number of stations on individual cruises, or by adding
additional cruises during the Autumn months would have been preferred.
In summary, we completed a complemented survey of the Patuxent River
to collect data in support of ecosystem-based fisheries management. The survey was
108
capable of providing estimates of relative abundance of numerous fish species during a
single calendar year. Estimates of absolute abundance could not be developed from the
data. Additional survey work would be required to determine whether a single
multispecies survey could replace the multitude of single species surveys currently
employed by regional agencies in support of fisheries management. Of additional
benefit, samples collected during the survey have been retained and are currently
undergoing age and diet analyses. It is anticipated that these data will be available within
the next 6-12 months, and they will provide additional useful information for
parameterizing food-web based ecosystem models.
109
4. 5. LITERATURE CITED
Literature Cited
Houde ED, Fogarty MJ, Miller TJ (1998) Prospects for Multispecies Management in Chesapeake Bay: A workshop. Report No. 98-002, Scientific and Technical Committee of the Chesapeake Bay Program, Edgwater, MD
Jensen OP, Seppelt R, Miller TJ (2005) Winter distribution of blue crab (Callinectes sapidus) in Chesapeake Bay: application of a two-stage generalized additive model (GAM). Marine Ecology - Progress Series 299:239-255
Kaiser MJ, Jennings S (2002) Ecosystem effects of fishing. In: Hart PJB, Reynolds JB (eds) Handbook of Fish Biology and Fisheries, Vol 2. Blackwell Science, Oxford, p 342-366
McErlean AJ, O'Connor SG, Mihursky JA, Gibson CI (1973) Abundance, diversity and seasonal patterns of estuarine fish populations. Estuarine, Coastal and Shelf Science 1:19-36
Miller TJ (2006) Total Removals. In: NOAA (ed) Fisheries Ecosystem Planning for the Chesapeake Bay, Vol 3. American Fisheries Society, Bethesda, MD, p xx-xxx
Montane MM, Austin HM (2005) Effects of hurricanes on Atlantic croaker (Micropogonias undulatus) recruitment to Chesapeake Bay. In: Sellner KG (ed) Hurricane Isabel: Perspectives. Chesapeake Research Consortiumm, Edgewater. MD, p 185-192
Secor DH, Houde ED (1995) Temperature effects on the timing of striped bass egg production, larval viability, and recruitment potential in the patuxent river (chesapeake bay). Estuaries 18:527-544
Smith TD (1994) Scaling Fisheries, the science of measuring the effects of fishing, 1855- 1955, Vol. Cambridge Univ. Press, Cambridge, UK
Wood SN, Augustin NH (2002) GAMs with integrated model selection using penalized regression splines and applications to environmental modelling. Ecological Modelling 157:157-177
110
Table 4.1. Schedule of cruises conducted as a part of the PAXFIMS program
Dates Cruise Beginning Ending Number of Stations
PF0401 1/22/2004 1/22/2004 9 PF0402 3/3/2004 3/5/2004 27 PF0403 4/6/2004 4/8/2004 28 PF0404 Abandoned due to engine failure on vessel PF0405 6/2/2004 6/4/2004 28 PF0406 6/15/2004 6/17/2004 28 PF0407 8/3/2004 8/5/2004 28 PF0408 9/8 /2004 9/10/2004 28 PF0409 10/5/2004 10/7/2004 28 PF0410 10/20/2004 10/22/2004 28 PF0411 11/2/2004 11/4/2003 28 PF0412 11/30/2004 12/2/2004 28
111
Table 4.2. Results of analysis of length frequencies of bay anchovy in the midwater trawl
for each cruise. Modes are fitted by maximum likelihood methods
Ist mode 2nd Mode CruiseID
Mean date Prop. Mean Varianc
e Prop. Mean Variance
Weighted average length (mm)
PF0402 3/4/04 1.000 44.667 11.222 44.667 PF0403 4/7/04 1.000 61.560 129.286 61.560 PF0404 PF0405 6/2/04 0.968 65.555 44.417 0.032 87.599 44.417 66.267 PF0406 6/16/04 1.000 66.723 43.647 66.723 PF0407 8/3/04 0.137 75.485 81.603 0.863 39.580 81.603 44.483 PF0408 9/8/04 0.194 56.790 59.972 0.806 38.225 59.972 41.818 PF0409 10/6/04 0.140 64.774 94.112 0.860 43.584 94.112 46.546 PF0410 10/21/0
4 0.204 67.290 93.084 0.796 44.985 93.084 49.542
PF0411 11/3/04 0.206 68.355 78.264 0.794 49.367 78.264 53.284 PF0412 12/2/04 1.000 45.879 97.054 45.879
112
Table 4.3. Results of analysis of length frequencies of white perch in the bottom trawl for each cruise. Modes are fitted by maximum likelihood methods
Ist mode 2nd Mode 3rd mode CruiseID Mean data Proportion Mean Variance Proportion Mean Variance Proportion Mean Variance
Weighted average length (mm)
PF0402 3/4/2004 0.7805 88.10663 130.7426 0.2195 177.1809 551.0106 107.6585 PF0403 4/7/2004 0.502373 93.09016 101.3245 0.497627 182.138 1343.974 137.4027 PF0404 PF0405 6/2/2004 0.568785 116.9223 98.00178 0.431215 161.4028 730.7882 136.1029 PF0406 6/16/2004 0.021898 44.16667 5.555554 0.458536 121.5551 144.1783 0.519566 169.0627 473.2473 144.5438 PF0407 8/3/2004 0.328298 130.791 98.48014 0.671702 164.1738 753.2124 153.2143 PF0408 9/8/2004 0.103867 63.84065 353.0089 0.848135 133.4566 353.0089 0.047997 204.8223 353.0089 129.6512 PF0409 10/6/2004 0.458505 133.2231 69.42351 0.541495 155.8772 633.1339 145.4902 PF0410 ######## 1 139.0027 365.0915 139.0027 PF0411 11/3/2004 0.88914 143.0676 186.2329 0.11086 194.7341 186.2329 148.7953 PF0412 12/2/2004 0.086744 83.23931 194.3714 0.824917 137.4564 194.3714 0.088339 195.6449 194.3714 137.8937
113
Table 4.4. Results of PCA analysis of environmental data from midwater trawl
Importance PCA 1 PCA 2 PCA 3 PCA 4 PCA 5 Standard deviation 1.508 1.097 0.961 0.683 0.377
Proportion of Variance 0.454 0.239 0.185 0.0933 0.0284
Cumulative Proportion 0.454 0.693 0.878 0.971 1.000
Loadings:
Salinity 0.739 0.520 0.407 0.101 Temp 0.597 0.228 0.145 -0.290 -0.697 Oxygen -0.616 -0.133 0.316 -0.707 Transmis -0.394 0.536 -0.314 -0.675 Fluorom -0.317 -0.334 0.770 -0.441
114
0
5
10
15
20
25
30
35
0 13 26 39 52
Week
Tem
pera
ture
(o C
)TemperatureSalinity
Figure 4.1. Average seasonal temperature and salinity for the Patuxent River (1985 – 2001) based on CBP monuitoring data
115
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0 13 26 39 52
Week
Tem
pera
ture
gra
dien
t (o C.
day
-1)
1 5 3 5
Figure 4.2. Rate of change of temperature in the Patuxent River (data from Figure 1.) Sampling intensity shown in bottom panel – see text for details
116
0
20
40
60
80
100
120
140
1/1/2004 3/2/2004 5/2/2004 7/2/2004 9/1/2004 11/1/2004 1/1/2005
Date
Mea
n CP
UE
(#.m
in-1)
0
10
20
30
40
50
60
70
80
90
100
1/1/2004 3/2/2004 5/2/2004 7/2/2004 9/1/2004 11/1/2004 1/1/2005
Date
Mea
n CPU
E (g
.min
-1)
B)
Figure 4.4. Average catch per unit effort for bay anchovy collected on PAXFIMS cruises expressed as A) numbers, or B) weight. Error bars represent ± 1 SD
117
Jan Mar Apr June
Aug Sept Oct Nov Dec
Figure 4.5. Distribution of bay anchovy in midwater trawl samples in the Patuxent River in 2004
118
Fig. 4.6. Spline fits for bay anchovy presence in the Patuxent River for bottom temperature (top), salinity (middle) and dissolved oxygen (bottom)
119
Fig. 4.7. Spline fits for bay anchovy abundance given presence in the Patuxent River for bottom temperature (top), salinity (middle) and dissolved oxygen (bottom)
120
Fig. 4.8. Relationship between observed bay anchovy catch per unit effort and catch per unit effort predicted by the GAM
121
Fig. 4.9. Average catch per unit for white perch collected in PAXFIMS cruises for A) numbers in the bottom trawl, B) weight in the bottom trawl, C) numbers in the midwater trawl and d) weight in the midwater trawl. Error bars indicated ± 1 SD
0
20
40
60
80
100
120
1/1/2004 3/2/2004 5/2/2004 7/2/2004 9/1/2004 11/1/2004 1/1/2005
Date
Mea
n CP
UE (#
.min
-1)
0
1000
2000
3000
4000
5000
6000
1/1/2004 3/2/2004 5/2/2004 7/2/2004 9/1/2004 11/1/2004 1/1/2005
Date
Mea
n CP
UE (g
.min
-1)
0
5
10
15
20
25
1/1/2004 3/2/2004 5/2/2004 7/2/2004 9/1/2004 11/1/2004 1/1/2005
Date
Mea
n CP
UE (#
.min
-1)
C)
0
200
400
600
800
1000
1200
1400
1600
1800
1/1/2004 3/2/2004 5/2/2004 7/2/2004 9/1/2004 11/1/2004 1/1/2005
Date
Mea
n C
PUE
(g.m
in-1
)
D)
122
Jan Mar Apr June
Aug Sept Oct Nov Dec
Figure 4.10. Distribution of white perch in benthic trawl samples in the Patuxent River in 2004
123
0 5 10 15 20 25
-50
510
AvgOfTemp
s(Av
gOfT
emp,
3.5)
2 4 6 8 10 12 14
-50
510
AvgOfSalinity
s(Av
gOfS
alin
ity,1
)
2 4 6 8 10 12 14
-50
510
AvgOfOxygen
s(Av
gOfO
xyge
n,1)
Fig, 4.11. Spline fits for white perch presence in the Patuxent River for bottom temperature (top), salinity (middle) and dissolved oxygen (bottom)
124
0 5 10 15 20 25
-6-2
02
4
AvgOfTemp
s(Av
gOfT
emp,
4.33
)
2 4 6 8 10 12
-6-2
02
4
AvgOfSalinity
s(Av
gOfS
alin
ity,1
.55)
4 6 8 10 12 14
-6-2
02
4
AvgOfOxygen
s(Av
gOfO
xyge
n,1.
37)
Figure 4.12. Spline fits for white perch abundance given presence in the Patuxent River for bottom temperature (top), salinity (middle) and dissolved oxygen (bottom)
125
Figure 4.13. Relationship between observed white perch catch per unit effort and catch per unit effort predicted by the GAM
126
Fig. 4.14. Modal analysis of bay anchovy length frequencies from the midwater trawl samples.
127
Fig. 4.15. Modal analysis of white perch length frequencies from the bottom trawl samples.
128
y = 0.2312x - 8776.7R2 = 0.9999
y = 0.1946x - 7311.9R2 = 0.8972
y = 0.061x - 2151.3R2 = 0.112
0
50
100
150
200
250
1/14/2004 3/4/2004 4/23/2004 6/12/2004 8/1/2004 9/20/2004 11/9/2004 12/29/2004
Date
Tota
l len
gth
(mm
)Age-0Age-1Age 2+Linear (Age 2+)
Fig 4.16. Mean size (mm TL) of white perch cohorts identified through modal analysis as a
function of date of capture. Linear regressions are fitted to each age class to provide an estimate of growth
129
Fig. 4.17. Results of a nMDS analysis of CPUEs for all species at every station in the survey. Stations are identified by cruise.
Transform: Fourth rootResemblance: S17 Bray Curtis similarity
CruisePF02PF11PF05PF06PF07PF08PF09PF10PF03PF12
2D Stress: 0.19
130
Fig. 4.18. Results of nMDS analysis of CPUEs with the station specific catches of bay anchovy identified. The size of the bubbles is proportional to the anchovy CPUE
Transform: Fourth rootResemblance: S17 Bray Curtis similarity
Bay anchovy
20
80
140
200
2D Stress: 0.19
131
Fig. 4.19. Conformation plots resulting the nMDS analysis showing the relationship with white perch CPUE (upper panel) and hogchoker CPUE (lower panel)
Transform: Fourth rootResemblance: S17 Bray Curtis similarity
Hogchoker
0.6
2.4
4.2
6
2D Stress: 0.19
Transform: Fourth rootResemblance: S17 Bray Curtis similarity
White perch
6
24
42
60
2D Stress: 0.19
Transform: Fourth rootResemblance: S17 Bray Curtis similarity
Hogchoker
0.6
2.4
4.2
6
2D Stress: 0.19
Transform: Fourth rootResemblance: S17 Bray Curtis similarity
White perch
6
24
42
60
2D Stress: 0.19
132
Fig .4.20. Results of PCA analysis of environmental data collected during midwater trawling
133
CHAPTER 5:
DISTRIBUTION AND DIET OF ATLANTIC CROAKER MICROPOGONIAS UNDULATUS IN
CHESAPEAKE BAY
Janet A. Nye and Thomas J. Miller
Chesapeake Biological Laboratory
University of Maryland Center for Environmental Science
Solomons, MD 20688
134
5.1 INTRODUCTION
The relative effect of biotic and abiotic factors in determining the distribution and diets of
organisms is a fundamental question of ecology. The distribution and abundance of an organism
is ultimately determined by its ecological niche. Hutchinson (1957) was the first to describe and
stress the importance of the multifaceted niche as the ecological space in which an organisms
lives building on the works of Grinell (1917) and Elton (1927). While Grinell (1917) was the
first to use the term "niche" to describe the geographic location of an organism in its
environment, Elton (1927) emphasized food availability and predators in determining the
ecological niche of a species. Hutchinson in a sense combined the ideas of these and other works
and conceived the ecological niche as defined by many biotic and abiotic variables. As such he
defined a niche as a multifaceted "hypervolume" or a multidimensional space occupied by an
organism.
Estuaries are good places to study to understand the complexities of niche theory. These
highly dynamic physio-chemical environments are influenced by energetic tidal flows and wind-
induced turbulence with strong seasonal effects and variability in freshwater input (Kennish
1986, Mann and Lazier 1996). Because of their characteristic circulation patterns, there are
strong gradients that provide the full spectrum of physical properties that might define an
organisms' niche. For example, the full range of salinities are found in estuaries as freshwater
rivers and tributaries flowing out of the estuary meet and mix with marine waters flowing into
the estuary. Thus, physiological tolerances in defining the niche can be determined. However,
estuaries introduce challenges in understanding an organism's niche because these systems are
135
not closed systems and have strong annual and seasonal changes in temperature, salinity, and
even dissolved oxygen.
In estuarine environments three abiotic factors: temperature, salinity and dissolved
oxygen, are likely the dominant regulators of fish distributions (Jung 2002, Lankford and Targett
1994, Rueda 2001) and their prey (Bottom and Jones 1990, Seitz and Schaffner 1995). These
studies exemplify the rich body of research on abiotic factors that affect species distribution.
Although temperature and salinity may influence population abundance and distribution based
on the physiology of each species, substrate and habitat structure are also important for fish
feeding and may influence distribution (Gibson and Robb 1992, Methratta 2006, Stoner et al.
2001). Such studies are important because they are informative at the scale on which a fishery
operates and can be used in management decisions such as delineating essential fish habitat and
marine reserves (Methratta and Link 2006). However, few studies exist that attempt to
quantitatively delineate which biotic and abiotic factors influence species abundance and
distribution.
Atlantic croaker Micropogonias undulatus, hereafter croaker, is a common, abundant
bottom-associated fish species that is distributed in marine and estuarine systems from the Gulf
of Mexico to Delaware Bay (ASMFC 1987). Numerous diet studies have been conducted on
croaker. Adult croaker have been described as opportunistic bottom-feeders that occasionally eat
small fishes (Murdy et al. 1997, Hildebrand and Schroeder 1928). Young of year (YOY) croaker
rely heavily on polychaetes in their diets, but also consume other benthic food such as detritus,
nematodes, insect larvae and amphipods (Homer and Boynton 1978, Nemerson 2002, Overstreet
136
and Heard 1978, Sheridan 1979). Croaker appear to change feeding habits as they get larger,
relying more heavily on large organisms such as mysids and fish (Nemerson 2002, Overstreet
and Heard 1978, Sheridan 1979). Hildebrand and Schroeder (1928) noted that of 392 fish whose
stomach contents were examined only three contained fish. Studies also indicate strong
ontogenetic patterns in diets. These data studies suggest less reliance on benthic prey than is
typically expected of this demersal Sciaenid (Chao and Musick 1977). Despite many diet studies
the trophic ecology of croaker, the ontogenetic, seasonal and spatial patterns in diet remain
poorly understood. More significantly, the consequences of this variability have been
completely ignored particularly with regard to the spatial distribution and abundance of croaker
in the Chesapeake Bay estuary.
The objectives of this study were first, to describe the distribution and diet of croaker in
the Chesapeake Bay and secondly, to understand how distribution and diet are related. In
quantifying these patterns, I seek specifically to determine the role of abiotic and biotic factors in
determining both aspects of croaker ecology. Quantification of the patterns and trends in diet is
challenging from both a sampling and statistical view points (Cortes 1997, Tirasin and Jorgensen
1999). No single approach or technique fully captures the spatial and temporal diversity in
dietary patterns. Accordingly, I used multivariate analyses to quantify seasonal, regional and
inter-annual patterns in diet. Subsequently, I used a two-stage Generalized Additive Model
(GAM) to determine biotic and abiotic factors that influence spatial distribution. The first stage
of the model predicts the probability of occurrence based on environmental variables using
presence/absence data as the response variable. The second stage of the GAM predicts the
abundance of croaker using the same explanatory variables that were used in the first stage, but
137
only using stations where croaker were present. I have used GAMs to relate distribution and diet
because they allow for linear and nonlinear relationships between explanatory and response
variables. GAMs have been widely used to quantify distributions of estuarine organisms (Jensen
et al. 2005, Jowett and Davey 2007, Stoner et al. 2001). However, few have attempted to
connect diet and distribution using GAMs to elucidate the relative importance of environmental
factors and the prey field to understand how each influences distribution. Using GAMs I
hypothesize that 1) croaker presence/absence is determined by physiological tolerances to abiotic
factors and, 2) that croaker abundance is influenced by availability of suitable prey.
Accordingly, abiotic factors should be the most important factors describing croaker occurrence
in the 1st stage of the GAM and biotic factors the most important in predicting croaker
abundance in the 2nd stage of the GAM.
5. 2. METHODS
5.2.1. Data collection
Croaker and environmental data were collected from 1995-2005 as part of two fishery-
independent sampling programs in the Chesapeake Bay. The Trophic Interactions in Estuarine
Systems (TIES) program surveyed the fish community in Chesapeake Bay from 1995- 2000
(Jung and Houde 2003). Subsequently, the Chesapeake Fishery-Independent Multispecies trawl
survey (CHESFIMS) extended the TIES sampling protocols for the fish community from 2001-
2005. In both programs, research cruises occurred over 5-7 day periods three times annually, the
only difference being that cruises occurred in May, July, and October from 1995 to 2000 and in
May, July, and September from 2001 to 2005. During both programs additional cruises
supplemented the three annual cruises opportunistically. The survey design changed very little
138
during the eleven year time series. Trawl stations in the TIES program were located along 15
fixed transects spaced approximately 18.5 km (10 nm) apart from the head of the Bay to the Bay
mouth to ensure bay wide coverage (Jung and Houde 2003). Within each season, 11 of the 15
transects were occupied. Transects were identified as falling within one of three strata: upper,
middle, and lower Bay (Figure 5.1).
The individual strata have distinctive characteristics, and their boundaries broadly
correspond to ecologically relevant salinity regimes. The upper Bay is generally shallow, with
substantial areas less than 5 m in depth, and well mixed waters with high nutrient concentrations.
The bottom topography in the mid Bay includes a narrow channel in the middle of the Bay with a
stratified water column and broad flanking shoals. This region has relatively clear waters and
experiences seasonally high nutrient concentrations and periods of hypoxia. The lower Bay has
the clearest waters, greatest depths and lowest nutrient concentrations (Kemp et al., 1999). The
strata volumes are 26,608 km3 (Lower), 16,840 km3 (Mid) and 8,664 km3 (Upper). The
CHESFIMS sampling design consisted of both fixed and random stations with stations
proportionally allocated to strata according to strata volume. Some fixed stations from the TIES
program were kept in CHESFIMS sampling for continuity. However, an additional number of
stations were randomly allocated based on strata volumes were added to each survey, resulting in
a complementary survey design.
Survey deployments throughout the 11-year time series followed the TIES trawling
procedures (Jung and Houde 2003) with standardized 20-minute oblique, stepped tows
conducted at each station using midwater trawls of the same design. Croaker was one of the
139
most frequently caught species caught in this time series. A midwater trawl with an 18-m2
mouth-opening with 6-mm cod end was deployed to collect primarily pelagic and benthopelagic
fishes. Oblique tows of the net were fished from top to bottom, and were 20 minutes in duration.
The trawl was towed for two minutes in each of ten depth zones evenly distributed throughout
the water column from the surface to the bottom, with minimum trawlable depth being 5 m. The
section of the tow conducted in the deepest zone sampled epibenthic fishes close to or on the
bottom. The remaining portion of the tow sampled pelagic and neustonic fishes. A minilog was
attached to the float line of the net and measured depth, temperature, and time during each tow.
The depth profile from the minilog was inspected after each tow to ensure that the trawl was
deployed in the manner described above and that the net fished the bottom portion of the water
column, important in the case of the demersal croaker. All tows were conducted between 18:00
and 7:00 Eastern Standard Time to minimize gear avoidance and to take advantage of the
reduced patchiness of multiple target species at night. At each station, a CTD was deployed to
measure dissolved oxygen, salinity, and temperature in the water column.
Catches at every station were identified, enumerated, measured and weighed onboard.
For each species, all fish or for large catches a subsample of 50-100 fish were measured (total
length in mm). Total weight of the catch of each species was measured. Croaker from the 2002-
2005 cruises were collected from each tow when present and were frozen for subsequent
processing in the laboratory. At each station, a CTD was deployed to measure dissolved oxygen,
salinity, and temperature throughout the water column. Data from the CHESFIMS collections
were used to map spatial distributions and describe diets. Data from the combined TIES and
140
CHESFIMS collections were used to develop two-stage GAM models to predict croaker
distributions.
5.2.2. Spatial distribution
To visualize the spatial distribution of croaker, spatial maps of croaker were developed. I
modeled adult croaker, defined as croaker greater than 100mm because of sporadic catch and
abundance estimation of YOY croaker. There were many stations where no croaker were
caught, causing the data to be zero-inflated. Thus, to adequately model the spatial distribution of
croaker I used indicator kriging to map the probability of croaker occurrence in the mainstem of
the bay. Indicator kriging changes abundance data into presence/absence data and does not
require the data to meet the assumptions of normality or stationarity (Chica-Olmo and Luque-
Espinar 2002). The abundance variables are transformed to categorical presence/absence
variables before the kriging process by picking a threshold level, in this case an abundance equal
to one fish. Points above this threshold are given a value of one and points below are given a
value of zero. Thus, indicator kriging is robust to outliers (Journel 1983). This analysis provides
maps of probability of occurrence, rather than spatial abundance estimates of Atlantic croaker.
Maps were developed for each of the three annual CHESFIMS cruises from 2002-2005
using the indicator kriging option in ArcMap using a spherical semivariogram in all cases. The
semivariogram was adjusted by changing the number of nearest neighbors and geometry of the
search sectors in ArcMap. By changing these parameters, the model with the lowest Root Mean
Square (RMS) and lowest average standard error was chosen to represent croaker distribution.
In most cases, the search geometry had four sectors with a 45o offset.
141
5.2.3. Diet analysis
Frozen croaker collected during the CHESFIMS cruises (2002-2005) were thawed and
individual fish were weighed (wet weight, g), measured for total length (TL, nearest mm), and
their otoliths and stomachs removed. To quantify diets, the preserved stomach was blotted dry
and weighed with contents intact. The stomach contents were removed and the remaining
stomach tissue reweighed. The dissected stomach contents were examined and quantified under
a dissecting microscope at 10-40x magnification. Prey items were identified to the lowest taxon
feasible. Each prey type was weighed and the number of individuals determined. Diet was
quantified using percent composition by weight (%W). Mean proportional contribution of a prey
type by weight was calculated for each experimental unit or station with a two-stage clustering
scheme (Buckel et al. 1999, Cochran 1977). For each group, i, the total weight wik, of prey item
k was divided by the total weight of all identifiable prey items at the station, wi. Thus, the mean
proportional contribution of a prey type (Wk) was calculated as:
Wk = ΣMi(wik/wi)/ ΣMi
where Mi is the number of fish >100mm caught at the station. This method was used to calculate
%W for two clustering schemes, 1) where group (i) were equal to the year and strata and 2)
where the group (i) was simply the cruise (or year and season).
I used simple graphic analyses and summary statistics to describe croaker diet
composition by age, season, region and year. To quantify patterns in croaker diets more fully, I
applied Canonical Correspondence Analysis (CCA) to analyze patterns in %W (ter Braak 1986).
CCA is an ordination technique, but unlike ordination approaches such as principal components
analysis, CCA does not seek to explain all the variation in the data, rather it seeks to explain only
142
that variation directly associated with specified factors. For my analyses I examined
contributions of year, season, and strata of the bay. Analyses were conducted using the Vegan
package (Version 1.8.8) in R (Oksanen et al. 2007).
To understand trends in croaker diet composition by size, two-stage clustering was not
used and data was pooled from 2002-2005. Instead total weight of each prey item was divided
by the total weight of all prey items to arrive upon %W for each individual fish. Subsequently,
%W for each individual was averaged by 10mm length class and displayed graphically. To
determine if the incidence of anchovy in croaker stomachs exhibited diel trends the average total
weight (not %W) of anchovy in stomach was plotted against the time of capture for each season.
The average weight of anchovy in stomachs was also compared between males and females
using a non-parametric Mann-Whitney U test. Percent occurrence (%O) is also reported for each
prey category for individual fish pooled from 2002-2005 and is calculated by dividing the
number of stomach in which a prey item occurred by the total number of stomachs.
5.2.4. Effect of environmental variables and diet on croaker presence and abundance
To understand the biotic and abiotic factors that influence spatial distribution of adult
croaker as illustrated in maps produced by indicator kriging, I developed two-stage Generalized
Additive Models (GAMs). I chose four environmental parameters and two biotic parameters to
include in Generalized Additive Models (GAM). The parameters selected were chosen to reflect
parameters believed to influence the distribution of croaker. Salinity, temperature, and dissolved
oxygen were averaged over the entire water column for each CTD cast at each station.
Maximum depth was determined as the maximum depth from the CTD cast. Average grain size
143
was estimated using data from the Chesapeake Bay Program data collected from 1975-1981.
Grain size was reported on log2 (phi) scale where a value of 1 is the grain size for gravel and a
grain size of 8 and above corresponds to clay. Most of the area of the Chesapeake Bay floor
consists of sand (Phi ~0 to 4). The locations of stations at which sediment analyses were
conducted differed from TIES and CHESFIMS stations. Therefore, a map of interpolated phi
values for the entire Chesapeake Bay mainstem was created. Subsequently, I overlaid the TIES
and CHESFIMS station locations on the interpolated grain size map and the appropriate
interpolated values of phi were obtained using Hawth Tools
(http://www.spatialecology.com/htools/) in ArcGIS 8.1 software.
Maximum depth and grain size are physical properties that may represent a habitat
quality that croaker prefer. However, I have interpreted these variables as proxies for benthic
food resources available to croaker. Anchovy biomass was also used as a biotic variable because
it is a frequent food item in adult croaker stomachs. Anchovy biomass was log transformed so
that the data would be normally distributed and values would be within an order of magnitude of
the other variables in the model. The Pearson correlation coefficients among these variables
were quantified to understand the relationships among biotic and abiotic parameters used in the
model.
I first conducted a two stage GAM for data pooled over all years (1995-2005) and
seasons to explore broad trends in distribution. The predictions from the two stage GAM using
pooled data allowed evaluation of the method to predict croaker abundance. However, the
144
purpose of the two-stage GAM was to determine factors that influence distribution other than
seasonal migrations as timing of seasonal migrations can be easily discerned from distribution
maps. Therefore, I conducted three separate two stage GAMS for spring, summer, and fall to
understand factors that influence croaker distribution on a smaller temporal scale.
To evaluate how important each factor was in predicting croaker presence and
secondarily abundance I took 100 random samples of 79% of the data (n=1000), fit the GAM,
and then tallied the number of times a parameter was significant. Those factors that were
consistently significant in the GAMs were considered more important factors in determined
croaker distribution. All statistical analysis was done in R (Version 2.4.1) using the mgcv
package (Wood 2007). It should be noted that in the mgcv library the degree of smoothing is
part of model fitting so rather than set the degrees of freedom a priori, the best model is chosen
in part by changing the degrees of freedom. Model fits with more degrees of freedom indicate
more "curviness" and the overall model fit is penalized by high degrees of freedom.
145
5.3. RESULTS
The incidence of croaker occurrence exhibits seasonal and annual variation (Figure 2).
However, Atlantic croaker were consistently located in the lower to middle part of the
Chesapeake Bay. As indicated by the overall low probabilities of occurrence in spring cruises,
there are relatively few croaker in the Bay in the spring as adult croaker are just beginning to
migrate into the Bay. In the summer months, there are higher incidences of occurrence with
large aggregations of croaker in the low to mid section of the bay. However, in some years -
notably 2002 and 2003, there is another aggregation of croaker in the Upper Bay.
Eleven categories of prey were recognized in croaker diets collected between 2002 and
2005 (Table 5. 1). Overall, polychaetes were the dominant component of croaker by weight
(61.5%) and by occurrence (83.6%). Anchovy (8.9%) and mysids (8.2%) followed polychaetes
in importance by weight. However, in combination mysids, amphipods, and other benthic
organisms were more common in croaker stomachs than anchovy. Detritus and miscellaneous
pelagic prey were the least common food items and in many years were not recorded in stomachs
at all. The diet of Atlantic croaker varied annually and seasonally (Figure 5.3). Croaker
consumed more anchovies, fish, and mysids in the summer and fall of several years. In the
summer, at least 20% of the diet of croaker consistently consisted of anchovies and fish. In
particular, in the summer of 2002, about 50% of the diet of croaker by weight consisted of
anchovy in the middle strata of the bay.
146
The CCA of croaker diet explained approximately 4.1% of the data, but reinforced annual
and seasonal trends (Figure 5.4). Polychaetes and other organisms which were consistently
present in croaker stomachs were located centrally in the ordination. Anchovy, fish, and detritus
occurrence in diet was attributable to most of the explained variation on an interannual basis, as
reflected by the strong coherence of these three prey categories and the year variable in the
ordination. The presence of crabs in croaker diet was more strongly associated with season than
with region, but the coherence was not strong. In general, it appears that bivalves were more
frequently eaten in the upper part of the Bay and shrimp in the lower part of the Bay (Figure 5.3).
Correlations of environmental variables (temperature, salinity, dissolved oxygen, and
grain size) with prey categories were tested, but all correlation coefficients were very low and
only one comparison was significant at the P=0.001 level (Bonferoni adjustment,
P=0.05/44=0.001). Proportion of amphipods in diets was positively correlated with salinity
(r=0.246, P=0.001), indicating that amphipods are consumed in waters of higher salinity, perhaps
in the lower Bay. Grain size and other benthic prey category were positively correlated (r=0.17,
P=0.0248). Dissolved oxygen and %W of anchovy was weakly negatively correlated (r=-0.15,
P=0.0468).
There was an ontogenetic change in croaker diets with small croaker eating small
crustaceans, particularly amphipods (Figure 5. 5). As croaker got larger their diet seemed to
become more diverse, but this may in part be a result of a greater number of individual stomachs
examined in moderate size classes. Size classes were pooled for fish <100m and >390 because
147
of small sample size. Larger croaker tended to have higher proportion of anchovies and fish in
their diet. Polychaetes were the staple diet item in all size classes.
The weight of anchovy in the stomachs of croaker was highest following sunset in spring
and summer (Figure 5.6). In the spring, the weight of anchovy in croaker stomachs was also
high near sunrise. However, this trend was not seen in other seasons. In contrast, there did not
appear to be any diel trend in polychaetes consumption.
Croaker occupied waters of the Bay exhibiting a wide range of temperatures, salinities,
and dissolved oxygen (Figure 5.7). The log of croaker abundance was weakly, but significantly
positively correlated with salinity and negatively correlated with grain size (Table 5.2). Croaker
biomass was not significantly correlated with any other factors examined and appeared to be
present and abundant at a wide range of values for all physical parameters examined (Figure
5.3). There were several correlations between variables used in the GAMs (Table 5.2). Salinity
was negatively correlated with dissolved oxygen and grain size, but positively correlated with
depth, croaker biomass, and anchovy biomass. The significant negative correlation with grain
size can be explained by the estuarine gradients in both salinity and grain size from the
freshwater input at the head of the estuary to the mouth of the bay. Grain size decreases from
large to small grain sizes in general from the head to the mouth of the bay (Figure 5.8). Other
correlations with salinity were relatively low. The correlation between dissolved oxygen and
temperature was relatively high which can be explained by the decrease in oxygen solubility as
temperature increases. Interestingly, salinity was correlated with both anchovy and croaker
148
biomass, reflecting the high abundance of croaker in the lower to middle parts of the Bay (Figure
5.2).
Bootstrapping each stage of the GAM with data pooled over all seasons indicated that of
all the included main effects, croaker presence was most influenced by temperature and salinity
when year and the interaction of temperature and salinity were not included in the model (Table
5.3). In 100 iterations, temperature was significant at the p=0.01 level 100% of the time and
salinity 93% of the time. However, when year and the interaction of temperature and salinity
were included, the main effects of both temperature and salinity were significant only 16 and
12% of the time in predicting croaker presence respectively. This suggests that the main effects
of temperature and salinity are reflective of the seasonal migrations of croaker. Interestingly,
anchovy biomass was a predictor of croaker presence in every run with or without year effects
included in the model.
In the second stage of the model where croaker abundance was modeled, temperature and
salinity were again important factors in the model when the effect of year or the interaction of
temperature and salinity was not included (Table 5.3). In contrast to the first stage bootstrapping
results, when year and the interaction of temperature and salinity were included in the model, the
main effects of temperature and salinity remained the most important factors in predicting
abundance. While anchovy biomass and grain size were frequently incorporated in the 1st stage
GAM, these factors were rarely significant in predicting croaker abundance in the 2nd stage of the
149
GAM. Dissolved oxygen was never a significant factor for either the 1st or 2nd stage GAM.
Depth was occasionally a significant factor in the 1st stage, but never in the 2nd stage.
After this bootstrapping exercise on 100 subsets of the data, a two stage GAM was run
with all data (n=1258) to evaluate the predictive ability of the model. In the first stage,
significant factors in predicting croaker presence were temperature, depth, grain size, anchovy
biomass, year and the interaction of temperature and salinity (Table 5.4, Figure 5.9). The
relationships of croaker occurrence with temperature, depth and year were curvilinear (Figure
5.10). The relationship appears dome shaped with depth and anchovy weight. In the second
stage, temperature, salinity, grain size, year, and the temperature and salinity interaction were
incorporated to predict croaker abundance (Figure 5.10). The relationship of croaker abundance
predicted by the second stage of the GAM was curvilinear with temperature, dome shaped with
grain size and year, and linear with salinity. Deviance explained in the second stage of the
model was 43.2%, much higher than the deviance explained in the first stage of the model,
18.7%.
Predicted croaker abundance was calculated in two ways: 1) by the 2nd stage GAM itself
using only stations where croaker were present and 2) by the product of the presence and
abundance predicted by the 1st and 2nd stage models respectively. The explanatory variables
from the original data were used in both cases and observed croaker biomass was compared to
these predictions. The second stage of the two stage GAM alone predicted croaker abundance
much better than the full two stage GAM (Figure 5.11). However, neither captured the range of
150
values of croaker biomass and the GAM seemed to dampen much of the variability in abundance
that was observed.
To eliminate the effects of seasonal migrations, two stage GAMs were run for the spring,
summer, and fall. Year and the interaction between temperature and salinity and Year were
important factors in almost all of the seasonal models even though the data was separated by
season (Table 5.4). The relationship of both croaker occurrence and abundance with year was
highly curvilinear especially in the spring and fall (Figures 5.12-17). In general, croaker
occurrence and abundance increased linearly or approached linearity with salinity. In the second
stage of the seasonal GAMs, croaker abundance increases linearly with dissolved oxygen and
depth in the Spring (Figure 5.13). Most other relationships of explanatory variables with croaker
occurrence and abundance were curvilinear reflecting the patchiness in croaker distribution. The
deviance explained and R2 values were higher for the seasonal models than for the pooled model
(Table 5.4). The seasonal two stage GAMs also predicted observed croaker abundance better
than the pooled model, but again, the modeling approach dampened the range of croaker
abundance estimates (Figure 5.18). The maximum observed croaker biomass was much higher
than the maximum predicted value in both the pooled and seasonal models.
151
5.5 DISCUSSION
Croaker feeds on a wide variety of organisms, but in contrast to previous studies croaker
were found to eat a substantial amount of anchovy during the summer months in Chesapeake
Bay. Fish have been reported as small components of the diet of adult croaker in previous studies
(Darnell 1961, Nemerson 2002, Overstreet and Heard 1978, Sheridan 1979). The work herein
suggests that about 20% of the diet of croaker by weight consists of anchovy. While croaker still
consistently feed on benthic portions of the food web, these results suggest that a substantial
portion of their bioenergetic needs (as indicated by %W) are met by anchovy in the summer
months and that croaker predation could influence both the benthic and pelagic portions of the
foodweb.
Anchovy could be more common in croaker diets because of the effects of eutrophication
on the ecosystem. Estuarine ecosystems worldwide are increasingly subject to anthropogenic
stresses that have lead to eutrophication, which induces widespread alterations in the ecosystem
(e.g. Kemp et al. 2005). De Levia Moreno et al. (2000) proposed that one of the effects of
eutrophication was to increase the ratio of biomasses of pelagic to benthic associated fishes,
indicative of general system wide change from benthic to pelagic production. Indeed in the
Chesapeake Bay, the ratio of pelagic to benthic fishery removals increased from 1.90 to 2.66
between the 1960’s and the 1990s. Eutrophication and the change from a more pelagic to
benthic ecosystem may cause alteration of diet patterns. Powers et al. (2005) found that the diet
of Atlantic croaker shifted from clams to less nutritious food sources such as detritus and plant
tissue after summer hypoxic events in the Neuse River estuary (NC, USA). Studies on other
152
benthic feeders in Chesapeake Bay illustrated that the ability of a benthic predator to prey upon
clams was reduced during periods of even sporadic low dissolved oxygen events (Seitz 2003).
The earliest of croaker diet studies, that of Hildebrand and Schroeder (1928) reported less
than 1% of the stomachs that were examined had fish in them. In contrast, this study and other
studies since the 1970s report fish as a relatively small, but common part of croaker diet (Chao
and Musick 1977, Nemerson 2002, Overstreet and Heard 1978, Sheridan 1979). The increase in
fish reported in the diet of croaker might be a result of different sampling, but it might also
reflect increasingly poor water quality in coastal areas where croaker live. There was no
statistically significant correlation between dissolved oxygen and the amount of anchovy in
croaker diet. However, croaker eat more anchovies in the summer when hypoxia is more
common. In the summer of 2003 the middle and upper regions of the Bay experienced very low
oxygen conditions, which is coincident with a high proportion of anchovy and fish in the diets of
croaker in the same regions. However, the highest incidence of anchovy feeding was in 2002,
when hypoxia was not as severe as 2003. Factors that influence diet were difficult to detect in
this and other diet studies. Therefore, it is possible that a general shift from a benthic to a
pelagic Chesapeake Bay ecosystem may explain the higher incidence of anchovy in present day
croaker diets.
It is more likely that the incidence of anchovy and fish in the diet of croaker were higher
in this study because croaker are crepuscular predators on anchovy. This crepuscular feeding
was captured by sampling with a midwater trawl at night while other studies of croaker diet have
used bottom trawls during the day. The only other diet study where samples were collected at
153
night probably did not capture this because it was conducted in shallow waters and there was a
notable decrease in croaker catches at night presumably because croaker moved to deeper water
at night (Homer and Boynton 1978). In this study, there was a higher weight of anchovy in
croaker stomachs following sunset indicating crepuscular feeding behavior. The adjustment in
sight and behaviors of many fish during the twilight period after sunset and before sunrise is
thought to provide an opportunistic feeding time for some predators in aquatic environments.
Indeed diel variations in diet have been detected in other studies (Clark et al. 2003, Johnson and
Dropkin 1993). Taylor et al. (2007) also found that swimming speeds of bay anchovy were
lower and less variable at night than during the day, which may enable a demersal fish such as
croaker to feed upon prey that is much more mobile than its traditional benthic prey. While
some consumption of anchovy could be from net-feeding in the midwater trawl, this is unlikely.
The relative degree of digestion was recorded in 2004 and 2005 and all stages of digestion were
present, indicating that the consumption of anchovy is not simply a result of net feeding.
Distribution of croaker varied seasonally and annually and is reflected in the maps of
probability of occurrence and in the two stage GAMs. Croaker occurrence and abundance
fluctuated annually so that the effect of year was included in all but two of all the first and
second stage GAMs produced. Temperature and salinity and/or their interaction was also a
consistent contributor to predict croaker distribution. We hypothesized that presence of croaker
would be predicted by physical properties of the water column because the presence of croaker
should be bounded by its tolerance to water chemistry. However, croaker was tolerant of a wide
range of salinity, temperature, and dissolved oxygen. Furthermore, the prey field seemed to be
154
important in determining croaker occurrence. Anchovy was a consistent predictor of croaker
occurrence in these models.
We secondarily hypothesized that croaker would be more abundant where prey resources
were high. However, the second stage of the GAMs indicated that both abiotic and biotic factors
were important in predicting abundance. In fact, anchovy biomass was not included in any of the
second stage models and grain size was included only in the second stage GAM pooled over
seasons. These results do not mean that prey field is not important in determining croaker
distribution. Grain size was used as a proxy for benthic food resources, but it would have been
better to use actual abundance estimates of benthic organisms upon which croaker frequently
feed. Estimates of benthic biomass are available but do not overlap temporally with our
sampling scheme. Furthermore, the estimates of grain size were obtained from the 1980s and
there may have been changes in sediment characteristics since that time. However, the overall
trends in grain size are probably similar. While anchovy was a consistent predictor of croaker
presence it is possible that anchovy abundance is influenced by the same factors as croaker and
these factors may or may not have been incorporated into the model.
While pooling data across all seasons provided a large number of data points to fit the
GAMs, seasonal GAMs predicted the abundance of croaker much better. The two stage GAM
did predict general trends in croaker abundance, but was unable to capture the wide range of
estimates of croaker biomass. In particular, GAMs were unable to capture the number of stations
with zero values. GAMs have been used to predict the spatial distribution in much of the marine
ecology literature (Hedger et al. 2004, Jensen et al. 2005, Stoner et al. 2001). While it is possible
155
to create spatially explicit maps of croaker abundance based on abiotic and biotic factors, in this
application GAMs were used to identify factors that influence croaker presence and secondarily
abundance. GAMs allowed the incorporation of many possible explanatory variables, different
distributions of data (Poisson in the first stage and Gaussian in the second stage), and the ability
to fit curvilinear relationships to predict distribution, which is more biologically realistic.
Here, we have documented clear trends and levels of variability in the distribution and
diet of Atlantic croaker in Chesapeake Bay. While the patterns were clear, the consequences of
these patterns to the fitness of individual fish remain uncertain. For example, does the variability
in croaker diet observed at the regional and interannual levels have any fitness consequence for
the individual croaker? Specifically, does a higher proportion of anchovy in the diet confer a
growth advantage, or does it reflect changes in diet driven by exclusion of croaker from preferred
habitats, and therefore the presence of anchovy in croaker diets actually confers a fitness cost.
To explore these and other potential hypotheses, it would be necessary to assess the condition of
croaker with different observed diets. The challenge of such analyses will be matching the
temporal resolution of indices of condition with that of the diet. Dietary information derived
from analysis of stomach contents represents a "snapshot" of consumption, but do not necessarily
represent what a fish is consistently eating and more importantly assimilating. Similarly, indices
of condition also have characteristic response and latency times (Ferron and Leggett 1994,
Suthers et al. 1992). Thus, addressing the consequences of the patterns in distribution and diet
observed here will require additional studies that seek to match observations on diet and
condition at appropriate spatial and temporal scales.
Table 5.1: Description of prey categories used to describe croaker diet.
Prey category Description %W %O
Polychaetes Many unidentified species, but include trumpet worms Pectinaria gouldi, clam worms, Neries spp. and terebellid worms Terebellidae 61.5% 83.6%
Anchovy Mostly bay anchovy, Anchoa mitchilli, but may include striped anchovy Anchoa hepsetus 8.9% 13.2%
Mysids Mostly Neomysis americanus, but may include Mysidopsis bigelowi 8.2% 36.5%
Amphipods Many species including Gammarus spp, Leptocheirus plumulosus, Corophium lacustre, Monoculodes edwardsi 5.2% 21.0%
Other benthic Hydroids, molluscs, gastropods, barnacles, cumaceans, isopods,
Cyathura spp., skeleton shrimp, other crustaceans, sea squirts, and ribbon worms
4.3% 20.2%
Bivalves Unidentified bivalves, clams and seedling mussels 3.5% 12.7%
Fish Unidentified fish and fish remains, and YOY weakfish Cynoscion regalis 3.4% 12.3%
Crabs Unidentified crab remains and white fingered mud crab Rhithropanopeus harrisii 1.8% 4.5%
Shrimp Unidentified shrimp remains, Caridean shrimps, Pugeo spp., sand shrimp Crangon septemspinosa, and mantis shrimp Squilla empusa 1.6% 6.5%
Detritus and macroalgae Unidentified algae, inorganic matter, and plant matter 1.3% 11.4%
Other pelagic Squids, sea nettles, insects 0.3% 1.1%
Table 5.2: Pearson correlations between explanatory variables used in the 1st stage of the
GAM, all seasons and years combined. Pairwise comparisons were considered significant at
the P=0.002 level to maintain an experiment-wise error rate of P=0.05 (Bonferroni adjustment
P =0.05/21=0.002).
Salinity TemperatureDissolved Oxygen Depth Grainsize
Log Anchovy Biomass
Log Croaker Biomass
Salinity 1
Temperature 0.099
0.0151 1
Dissolved Oxygen
-0.18344 <0.0001
-0.679<0.0001 1
Depth 0.129
0.0016 0.01934
0.637-0.0820.0442 1
Grainsize -0.611 <0.0001
-0.069620.089
0.1120.0062
-0.0820.044 1
Log Anchovy Biomass
0.319 <0.0001
0.010380.8
-0.173<0.0001
0.00750.855
-0.105 0.011 1
Log Croaker Biomass
0.19928 <0.0001
0.0750.0673
-0.0360.377
0.0140.733
-0.134 0.001
0.08250.044 1
158
Table 5.3: The percentage of simulations (number out of 100 iterations obtained from
bootstrapping) where each explanatory variable was significant in 1st and 2nd stage GAMS of
data pooled over all seasons 1995-2005.
1st stage 2nd stage Explanatory
Variable No year effects Year effect included No year effects Year effect
included Temperature 100 16 100 92
Salinity 93 12 100 99
Dissolved Oxygen 0 0 0 0
Depth 3 9 0 0
Grain size 69 75 4 5 Log Anchovy
Biomass 100 100 13 4
Year 83 56 Temperature*Salin
ity 100 78
159
Table 5.4: Significant variables used in final two stage models for all seasons combined and then for each separate season.
1st stage 2nd stage
Explanatory Variable All
seasons Spring Summer Fall All seasons Spring Summer Fall
Temperature X X X X
Salinity X X X X
Dissolved Oxygen X
Depth X X X
Grain size X X X X
Anchovy Biomass X X X
Year X X X X X X
Temp*Salinity X X X X X X
Deviance Explained 18.7% 36.60% 20.4% 24.3% 43.2% 65.4% 34.60% 46.6%
Adjusted R2 0.195 0.351 0.188 0.262 0.383 0.602 0.272 0.417
N 1258 396 415 447 446 120 122 204
160
Figure 5.1: An example of the TIES and CHESFIMS survey design using the stations from the
spring of 2001. Fixed stations are indicated with stars. Random stations are indicated with
circles. The three strata of Chesapeake Bay (Upper, Middle, and Lower Bay) are separated
with horizontal lines and labeled accordingly.
161
Figure 5.2: Maps of the probability of occurrence of Atlantic croaker in Chesapeake Bay as estimated by indicator kriging for a) 2001, b)
2002, c) 2003 and d) 2004. Numbers in parentheses indicate the number of stations sampled in that particular cruise.
162
Error! Objects cannot be created from editing field codes.Error! Objects cannot be created from editing field codes. Error!
Objects cannot be created from editing field codes.Error! Objects cannot be created from editing field codes.
Figure 5.3: Diet of Atlantic croaker (proportion by weight) by year, season, and strata of the Bay a) 2002 Spring, b) 2002 Summer, c)
2002 Fall, d) 2003 Spring, e) 2003 Summer, f) 2003 Fall, g) 2004 Spring, h) 2004 Summer, i) 2004 Fall, j) 2005 Spring, k) 2005
Summer, and l) 2005 Fall. Panels where a figure is missing indicates that no croaker were collected in that sampling period.
2a)
2c)
163
2002
Lower Bay
0
0.2
0.4
0.6
0.8A
ncho
vy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
Middle Bay
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
Upper Bay
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
a)
b)
c)
164
2003
Lower Bay
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
Middle Bay
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
Upper Bay
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
d)
f)
e)
165
2004
Lower Bay
0
0.2
0.4
0.6
0.8A
ncho
vy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
Middle Bay
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
h)
g)
i)
166
2005
Lower Bay
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
Middle Bay
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
0
0.2
0.4
0.6
0.8
Anc
hovy
Fish
Pel
agic
Mys
id
Shr
imp
Am
phip
ods
Cra
b
Ben
thic
Biv
alve
Pol
ycha
ete
j)
k)
l)
167
Figure 5.4: Biplot from Canonical Correspondence Analysis of the factors influencing diet
composition of Atlantic croaker. Arrows represent factors while labels in blue are centroids of scores
for the prey species.
168
Figure 5.5: Diet composition by weight for each 10mm size class of Atlantic croaker examined 2002-
2005.
0
0.2
0.4
0.6
0.8
1
<100 11
012
013
014
015
016
017
018
019
020
021
022
023
024
025
026
027
028
029
030
031
032
033
034
035
036
037
038
0>3
90
Per
cent
com
posi
tion
by w
eigh
t (%
W)
Anchovy Fish Mysid Shrimp Other pelagic Other benthic
Amphipods Bivalves Crab Polychaetes Detritus
169
Figure 5.6: Total weight (g) of polychaetes and anchovies in croaker stomachs by one hour
time periods. X-axis labels represent the beginning of each time interval (i.e. 19:00 indicates
the time period from 19:00-20:00). Arrows indicate sunset.
0
1
2
3
4
5
19:00 20:00 21:00 22:00 23:00 0:00 1:00 2:00 3:00 4:00 5:00
PolychaetesAnchovies
0
1
2
3
4
5
19:00 20:00 21:00 22:00 23:00 0:00 1:00 2:00 3:00 4:00 5:00
Tota
l wei
ght o
f pre
y ite
m (g
)
0
1
2
3
4
5
19:00 20:00 21:00 22:00 23:00 0:00 1:00 2:00 3:00 4:00 5:00
170
Figure 5.7: Relationship of the log of croaker biomass with explanatory variables used in the two stage
GAMs.
0
4
8
12
0 10 20 30 40Salinity
Log
of c
roak
er b
iom
ass
0
4
8
12
10 15 20 25 30Temperature (oC)
0
4
8
12
0 4 8 12 16Dissolved Oxygen (mg/l)
Log
of c
roak
er b
iom
ass
0
4
8
12
0 2 4 6 8 10Depth (m)
0
4
8
12
0 4 8 12 16Grainsize (phi)
Log
of c
roak
er b
iom
ass
0
4
8
12
0 2 4 6 8 10Log of Anchovy abundance
171
Figure 5.8: General trends in a) salinity in the summer months and b) grain size (phi)
172
Figure 5.9: Smooth functions from 1st stage of GAM pooled over year and season
173
Figure 5.10: Spline functions for significant terms in the second stage of the GAM pooled over
year and season.
174
Figure 5.11: Prediction of croaker biomass by multiplying the two stages of the GAM (black
circles) and by the second stage of the GAM alone (Grey). The dashed line is the 1:1 line for
reference and regression lines are shown for both predictions.
0
2
4
6
8
10
0 2 4 6 8 10Observed Log Croaker Biomass
Pre
dict
ed L
og C
roak
er B
iom
ass
175
Figure 5.12: Spline smoothed plots of Atlantic croaker presence generated by the first stage
spring GAM. Y-axes represent the effect of the explanatory variable on croaker occurrence.
Tick marks (or rugs) on the x-axis indicate sampling intensity. Points are residuals for each
observation and dashed lines are twice the standard error.
Depth Grain size
Log Anchovy Biomass
Temperature
Year
Salin
ity
176
Figure 5.13: Spline smoothed plots of Atlantic croaker presence generated by the second stage
spring GAM. Y-axes represent the effect of the explanatory variable on croaker abundance.
Tick marks (or rugs) on the x-axis indicate sampling intensity. Points are residuals for each
observation and dashed lines are twice the standard error.
Temperature
Year
Dissolved Oxygen Depth
Salinity
177
Figure 5.14: Spline smoothed plots of Atlantic croaker presence generated by the first stage
GAM in the summer. Y-axes represent the effect of the explanatory variable on croaker
occurrence. Tick marks (or rugs) on the x-axis indicate sampling intensity. Points are residuals
for each observation and dashed lines are twice the standard error.
Temperature Salinity
Grain size Year
Yea
r
Salinity
178
Figure 5.15: Spline smoothed plots of Atlantic croaker presence generated by the second stage
GAM in the summer. Y-axes represent the effect of the explanatory variable on croaker
abundance. Tick marks (or rugs) on the x-axis indicate sampling intensity. Points are residuals
for each observation and dashed lines are twice the standard error.
Salinity
Depth
Temperature
Salin
ity
179
Figure 5.16: Spline smoothed plots of Atlantic croaker presence generated by the first stage fall
GAM. Y-axes represent the effect of the explanatory variable on croaker occurrence. Tick
marks (or rugs) on the x-axis indicate sampling intensity. Points are residuals for each
observation and dashed lines are twice the standard error.
180
.
Figure 5.17: Spline smoothed plots of the distribution of Atlantic croaker biomass generated by
the second stage fall GAM. Y-axes represent the effect of the explanatory variable on croaker
Year
Log Anchovy biomass
Temperature
Salin
ity
181
abundance. Tick marks (or rugs) on the x-axis indicate sampling intensity. Points are residuals
for each observation and dashed lines are twice the standard error.
Temperature
182
Figure 5.18: Comparison of observed biomass with biomass predicted by the two stage GAMs for a) spring, b) summer, and c) fall. Regressions and equations are provided to quantify model fit.
y = 0.399x + 1.1641R2 = 0.4402
0
2
4
6
8
10
0 2 4 6 8 10
y = 0.204x + 1.4667R2 = 0.2299
0
2
4
6
8
10
0 2 4 6 8 10
Pre
dict
ed L
og C
roak
er B
iom
ass
y = 0.2276x + 1.6185R2 = 0.2428
0
2
4
6
8
10
0 2 4 6 8 10Observed Log Croaker Biomass
183
5.5. LITERATURE CITED
Able, K.W., and Fahay, M.P. 1998. The First Year in the Life of Estuarine Fishes in the Middle Atlantic Bight. Rutgers University Press, New Brunswick, NJ.
Able, K.W., Hagan, S.M., Kovitvongsa, K., Brown, S.A., and Lamonaca, J.C. 2007. Piscivory by the mummichog (Fundulus heteroclitus): Evidence from the laboratory and salt marshes. Journal of Experimental Marine Biology and Ecology 345(1): 26-37.
Alheit, J., and Niquen, M. 2004. Regime shifts in the Humboldt Current ecosystem. Progress In Oceanography. Regime shifts in the ocean: Reconciling observations and theory 60(2-4): 201-222.
Allen, J.I., and Clarke, K.R. 2007. Effects of demersal trawling on ecosystem functioning in the North Sea: a modelling study. Marine Ecology-Progress Series 336: 63-75.
Anderson, R.O., and Neumann, R.M. 1996. Length, weight, and associated indices. In Fisheries Techniques. Edited by B.R. Murphy and D.W. Willis. American Fisheries Society, Bethesda, MD. pp. 447-482.
ASMFC. 2005. Atlantic Croaker stock assessment and peer review reports, Washington, DC.
Bailey, K.M. 2000. Shifting control of recruitment of walleye pollock Theragrea chalcogramma after a major climatic and ecosystem change. Marine Ecology Progress Series 198: 215-224.
Baird, D., and Ulanowicz, R.E. 1989. The seasonal dynamics of the Chesapeake Bay Ecosystem. Ecological Monographs 59(4): 329-364.
Barbieri, L.R., Chittenden, M.E., Jr., and Jones, C.M. 1994. Age, growth, and mortality of Atlantic croaker, Micropogonias undulatus, in the Chesapeake Bay region, with a discussion of apparent geographic changes in population dynamics. Fishery Bulletin 92: 1-12.
Bartell, S.M., Breck, J.E., Gardner, R.H., and Brenkert, A.L. 1986. Individual parameter perturbation and error analysis of fish bioenergetics models. Canadian Journal of Fisheries and Aquatic Sciences 43: 160-168.
Basson, M., and Fogarty, M.J. 1997. Harvesting in discrete-time predatory-prey systems. Mathematical Bioscience 141(1): 41-74.
184
Beddington, J.R., and May, R.M. 1982. The Harvesting of Interacting Species in a Natural Ecosystem. Scientific American 247(5): 62-89.
Berlow, E.L. 1999. Strong effects of weak interaction in ecological communities. Nature 398: 330-334.
Bevelheimer, M.S., Stein, R.A., and Caline, R.F. 1985. Assessing the significance of physiological differences amony 3 esocids with a bioenergetics model. Canadian Journal of Fisheries and Aquatic Sciences 42: 57-69.
Bishop, M.J., and Wear, S.L. 2005. Ecological consequences of ontogenetic shifts in predator diet: Seasonal constraint of a behaviorally mediated indirect interaction. Journal of Experimental Marine Biology and Ecology 326(2): 199-206.
Blaber, S.J.M., Cyrus, D.P., Albaret, J.-J., Ching, C.V., Day, J.W., Elliott, M., Fonseca, M.S., Hoss, D.E., Orensanz, J., Potter, I.C., and Silvert, W. 2000. Effects of fishing on the structure and functioning of estuarine and nearshore ecosystems. ICES Journal of Marine Science 57: 590-602.
Blackwell, B.G., Brown, M.L., and Willis, D.W. 2000. Relative Weight (Wr) Status and Current Use in Fisheries Assessment and Management. Reviews in Fisheries Science 8(1): 1-44.
Boggs, C.H. 1991. Bioenergetics and growth on Northern Anchovy Engraulis mordax. Fishery Bulletin 89: 555-566.
Boisclair, D., and Leggett, W.C. 1989. The importance of activity in bioenergetics models applied to actively foraging fishes. Canadian Journal of Fisheries and Aquatic Sciences 46: 1859-1867.
Boisclair, D., and Sirois, P. 1993. Testing assumptions of fish bioenergetics models by direct estimation of growth, consumption, and activity rates. Transactions of the American Fisheries Society 122: 784-796.
Bolger, T., and Connolly, P.L. 1989. The selection of suitable indices for the measurement and analysis of fish condition doi:10.1111/j.1095-8649.1989.tb03300.x. Journal of Fish Biology 34(2): 171-182.
185
Bonzek, C.F., Latour, R.J., and Gartland, J. 2007. Annual progress report: Data collection and analysis in support of single and multispecies stock assessments in Chesapeake Bay: The Chesapeake Bay Multispecies Monitoring and Assessment Program, School of Marine Science, College of William and Mary, Virginia Institute of Marine Science.
Bottom, D.L., and Jones, K.K. 1990. Species Composition, Distribution, and Invertebrate Prey of Fish Assemblages in the Columbia River Estuary. Progress in Oceanography 25(1/4): 243-270.
Boynton, W.R., Garber, J.H., Summers, R., and Kemp, W.M. 1995. Inputs, Transformations, and Transport of Nitrogen and Phosphorus in Chesapeake Bay and Selected Tributaries. Estuaries 18(1B): 285-314.
Brabrand, A. 2004. Food of roach (Rutilus rutilus) and ide (Leusiscus idus): signficance of diet shift for interspecific competition in omnivorous fishes. Oecologia 66(4): 461-467.
Brown, M.L., and Murphy, B.R. 1991. Relationship of relative weight (Wr) to proximate body composition of juvenile striped bass and hybrid striped bass. Transactions of the American Fisheries Society 120: 509-518.
Buckel, J.A., Conover, D.O., Steinberg, N.D., and McKown, K.A. 1999. Impact of age-0 bluefish (Pomatomus saltatrix) predation on age-0 fishes in the Hudson Rover estuary: evidence for density-dependent loss of juvenile striped bass (Morone saxatilis). Canadian Journal of Fisheries and Aquatic Sciences 56: 275-287.
Buckley, L., Calderone, E., and Ong, T.L. 1999. RNA-DNA ratio and other nucleic acid-based indicators for growth and condition of marine fishes. Hydrobiologia 401: 265-277.
Buckley, L.J. 1984. Rna-DNA Ratio - an Index of Larval Fish Growth in the Sea. Marine Biology 80(3): 291-298.
Caddy, J.F. 2000. Marine catchment basin effects versus impacts of fisheries on semi-enclosed seas. ICES Journal of Marine Science 57: 625-640.
Caldarone, E.M. 2006. Estimating growth in haddock larvae Melanogrammus aeglefinus from RNA:DNA ratios and water temperature. Marine Ecology Progress Series 293: 241-252.
Campbell, S.J., and Pardede, S.T. 2006. Reef fish structure and cascading effects in response to artisanal fishing pressure. Fisheries Research 79(1-2): 75-83.
186
Carpenter, S.R., Kitchell, J.F., and Hodgson, J.R. 1985. Cascading trophic interactions and lake productivity. Biological Science 35: 634-639.
Chao, L.N., and Musick, J.A. 1977. In the York River Estuary, Virginia. Fishery Bulletin 75(4): 657-702.
Chesapeake Bay Fisheries Ecosystem Advisory Panel. 2006. Fisheries ecosystem planning for Chesapeake Bay. American Fisheries Society, Trends in Fisheries Science and Management 3, Bethesda, MD.
Chipps, S.R., Einfalt, L.M., and Wahl, D.H. 2000. Growth and food consumption by Tiger Muskellunge: Effects of Temperature and Ration level on bioenergetic model predictions. Transactions of the American Fisheries Society 129: 186-193.
Christensen, V., and Walters, C.J. 2004. Ecopath with Ecosim: methods, capabilities and limitations. Ecological Modelling 172(2-4): 109-139.
Clark, K.L., Ruiz, G.M., and Hines, A.H. 2003. Diel variation in predator abundance, predation risk and prey distribution in shallow-water estuarine habitats. Journal of Experimental Marine Biology and Ecology 287: 37-55.
Clemmesen, C. 1994. The Effect of Food Availability, Age or Size on the Rna/DNA Ratio of Individually Measured Herring Larvae - Laboratory Calibration. Marine Biology 118(3): 377-382.
Cochran, W.G. 1977. Sampling techniques. Wiley, New York.
Collie, J.S., and Gislason, H. 2001. Biological reference points for fish stocks in a multispecies context. Canadian Journal of Fisheries and Aquatic Sciences 58: 2167-2176.
Collie, J.S., Richardson, K., and Steele, J.H. 2004. Regime shifts: Can ecological theory illuminate the mechanisms? Progress In Oceanography. Regime shifts in the ocean: Reconciling observations and theory 60(2-4): 281-302.
Cortes, E. 1997. A critical review of methods of studying fish feeding based on analysis of stomach contents: application to elasmobranch fishes. Canadian Journal of Fisheries and Aquatic Sciences 54: 726-738.
187
Cox, M.K., and Hartman, K.J. 2005. Nonlethal estimation of proximate composition in fish. Canadian Journal of Fisheries and Aquatic Sciences 62(2): 269-275.
Cummins, K.W., and Wuycheck, J.C. 1971. Calorific equivalents for investigations in ecological energetics. Mitteilungen der Internationalen Vereinigung fur Theoretische und Angewandte Limnologie (Communications of the International Association of Theoretical and Applied Limnology).
Cury, P., and Shannon, L. 2004. Regime shifts in upwelling ecosystems: observed changes and possible mechanisms in the northern and southern Benguela. Progress In Oceanography Regime shifts in the ocean. Reconciling observations and theory 60(2-4): 223-243.
Cutts, C.J., Betcalfe, N.B., and Caylor, A.C. 1998. Aggression and growth depression in juvenile Atlantic salmon: the consequences of individual varaion in standard metabolic rates. Journal of Fish Biology 52(5): 1026-1037.
Dame, R.F., and Allen, D.M. 1996. Between estuaries and the sea. Journal of Experimental Marine Biology and Ecology 200(1-2): 169-185.
Darnell, R. 1961. Trophic spectrum of an estuarine community, based on studies of Lake Pontchartrain, Louisiana. Ecology 42(3): 553-568.
de Juan, S., Cartes, J.E., and Demestre, M. 2007a. Effects of commercial trawling activities in the diet of the flat fish Citharus linguatula (Osteichthyes : Pleuronectiformes) and the starfish Astropecten irregularis (Echinodermata : Asteroidea). Journal of Experimental Marine Biology and Ecology 349(1): 152-169.
de Juan, S., Thrush, S.F., and Demestre, M. 2007b. Functional changes as indicators of trawling disturbance on a benthic community located in a fishing ground (NW Mediterranean Sea). Marine Ecology-Progress Series 334: 117-129.
de Leiva Moreno, J.I., Agostini, V.N., Caddy, J.F., and Carocci, F. 2000. Is the pelagic-demersal ratio from fishery landings a udeful proxy for nutrient availability? A preliminary data exploration for the semi-enclosed seas around Europe. ICES Journal of Marine Science 57: 1091-1102.
Deegan, L.A., Bowen, J.L., Drake, D., Fleeger, J.W., Friedrichs, C.T., Galvan, K.A., Hobble, J.E., Hopkinson, C., Johnson, D.S., Johnson, J.M., Lemay, L.E., Miller, E., Peterson, B.J.,
188
Picard, C., Sheldon, S., Sutherland, M., Vallino, J., and Warren, R.S. 2007. Susceptibility of salt marshes to nutrient enrichment and predator removal. Ecological Applications 17(5): S42-S63.
Diaz, R.J., and Rosenberg, R. 1995. Marine benthic hypoxia: A review of its ecological effects and the behavioural responses of benthic macrofauna. In Oceanography and Marine Biology - an Annual Review, Vol 33. pp. 245-303.
Duncan, M., Craig, S.R., Lunger, A.N., Kuhn, D.D., Salze, G., and McLean, E. 2007. Bioimpedance assessment of body composition in cobia Rachycentron canadum (L. 1766). Aquaculture 271(1-4): 432-438.
Durbin, E.G., and Durbin, A.G. 1983. Energy and nitrogen budgets for the Atlantic menhaden, Brevoortia tyrannus (Pisces: Clupeidae), a filter feeding planktivore. Fishery Bulletin 8(177-199).
Eby, L.A., Crowder, L.B., McClellan, C.M., Peterson, C.H., and Powers, M.J. 2005. Habitat degradation from intermittent hypoxia: impacts on demersal fishes. Marine Ecology Progress Series 291: 249-261.
Elton, C.S. 1927. Animal Ecology. University of Chicago Press, Chicago.
Essington, T.E., Beaudreau, A.H., and Wiedenmann, J. 2006. Fishing through marine food webs 10.1073/pnas.0510964103. Proceedings of the National Academy of Sciences 103(9): 3171-3175.
Ferron, A., and Leggett, W.C. 1994. An appraisal of condition measures for marine fish larvae. Advances in Marine Biology 30: 217-303.
Finn, R.N., Ronnestad, I., Meeren, T.v.d., and Fyhn, H.J. 2002. Fuel and metabolic scaling during the early life stages of Atlantic cod Gadus morhua. Marine Ecology Progress Series 243: 217-234.
Fox, M.G. 1991. Food consumption and bioenergetics of young-of-the-year walleye (Stizostedion vitreum vitreum): Model predictions and population density effects. Canadian Journal of Fisheries and Aquatic Sciences 48(3): 434-441.
Fraley, C., and Raftery, A.E. 2007. MCLUST Version 3 for R: Normal Mixture Modeling and Model-Based Clustering. Available from http://www.stat.washington.edu/mclust/.
189
Fraser, A.J. 1986. Triacylglycerol content as a condition indec for fish, bivalve, and crustacean larvae. Canadian Journal of Fisheries and Aquatic Sciences 46(11): 1868-1873.
Fu, C., and Quinn, T.J., II. 2000. Estimability of natural mortality and other population parameters in a length-based model: Pandalus borealis in Kachemak Bay, Alaska. Canadian Journal of Fisheries and Aquatic Sciences 57: 2420-2432.
Fulton, T.W. 1904. The rate of growth of fishes, Fisheries Board of Scotland.
Garrison, L.P., and Link, J.S. 2000. Dietary guild structure of the fish community in the Northeast United States continental shelf ecosystem. Marine Ecology Progress Series 202: 231-240.
Gendron, L., Fradette, P., and Godbout, G. 2001. The importance of rock crab (Cancer irroratus) for growth, condition and ovary development of adult American lobster (Homarus americanus). Journal of Experimental Marine Biology and Ecology 262(2): 221-241.
Gibson, R.N., and Robb, L. 1992. The relationship between body size, sediment grain size and the burying ability of juvenile plaice, Pleuronectes platess L. Journal of Fish Biology 40(5): 771-778.
Gilliers, C., Amara, R., Bergeron, J.P., and Le Pape, O. 2004. Comparison of growth and condition indices of juvenile flatfish in different coastal nursery grounds. Environmental Biology of Fishes 71(2): 189-198.
Gislason, H. 1999. Single and multispecies reference points for Baltic fish stocks. ICES Journal of Marine Science 56: 571-583.
Govoni, J.J., Hons, D.E., and Chester, A.J. 1983. Comparative feeding of three species of larval fishes in the Northern Gulf of Mexico: Brevoortia patronus, Leiostomus xanthurus, and Micropogonias undulatus. Marine Ecology Progress Series 13: 189-199.
Grimes, C.B. 2001. Fishery production and the Mississippi River discharge. Fisheries 26(8): 17-26.
Grinell, J. 1917. The niche-relationships of the California Thrasher. Auk 34: 427-433.
190
Hagy, J.D. 2002. Eutrophication, hypoxia, and transfer efficiency in Chesapeake Bay. PhD Dissertation, University of Maryland Center for Environmental Science, College Park, MD.
Hansen, M.J., Boisclair, D., Brandt, S.B., Hewett, S.W., Kitchell, J.F., Lucas, M.C., and Ney, J.J. 1993. Applications of Bioenergetics models to fish ecology and management: Where do we go from here? Transactions of the American Fisheries Society 122(5): 1019-1030.
Hanson, P.C., Johnson, T.B., Schindler, D.E., and Kitchell, J.F. 1997. Fish Bioenergetics, 3.0. University of Wisconsin Sea Grant Institute, Madison, WI.
Hare, J.A., and Able, K.W. 2007. Mechanistic links between climate and fisheries along the east coast of the United States: explaining population outbursts of Atlantic croaker (Micropogonias undulatus). Fisheries Oceanography 16(1): 31-45.
Hare, J.A., Thorrold, S.R., Walsh, H., Reiss, C.S., Valle-Levinson, A., and Jones, C.M. 2005. Biophysical mechanisms of larval fish ingress into Chesapeake Bay. Marine Ecology Progress Series 303: 295-310.
Hartman, K.J. 1993. Striped Bass, Bluefish and Weakfish in the Chesapeake Bay: Energetics, Trophic Linkages, and Bioenergetics Model Applications. Dissertation, Marine Estuarine and Environmental Sciences Program, University of Maryland, College Park.
Hartman, K.J. 2003. Population-level consumption by Atlantic coastal striped bass and the influence of population recovery upon prey communities. Canadian Journal of Fisheries and Aquatic Sciences 10(5): 281-288.
Hartman, K.J., and Brandt, S.B. 1993. Systematic Sources of Bias in a Bioenergetics Model: Examples for the Age-0 Striped Bass. Transactions of the American Fisheries Society 122: 912-926.
Hartman, K.J., and Brandt, S.B. 1995a. Comparative energetics and the developmentof bioenergetics models for sympatric estuarine predators. Canadian Journal of Fisheries and Aquatic Sciences 52(8): 1647-1666.
Hartman, K.J., and Brandt, S.B. 1995b. Predatory demand and impact of striped bass, bluefish, and weakfish in the Chesapeake Bay: applications of bioenergetics models. Canadian Journal of Fisheries and Aquatic Sciences 52: 1667-1687.
191
Hartman, K.J., and Brandt, S.B. 1995c. Trophic Resource Partitioning, Diet, and Growth of Sympatric Estuarine Predators. Transactions of the American Fisheries Society 124: 520-537.
Hartman, K.J., and Margraf, F.J. 1993. Evidence of predatory control of yellow perch recruitment in a large system. Journal of Fish Biology 42: 109-119.
Harvey, C.J., Hanson, P.C., Essington, T.E., Brown, P.B., and Kitchell, J.F. 2002. Using bioenergetics models to predict stable isotope ratios in fishes. Canadian Journal of Fisheries and Aquatic Sciences 59: 115-124.
Hedger, R., McKenzie, E., Heath, M., Wright, P., Scott, B., Gallego, A., and Andrews, J. 2004. Analysis of the spatial distributions of mature cod (Gadus morhua) and haddock (Melanogrammus aeglefinus) abundance in the North Sea (1980-1999) using generalised additive models. Fisheries Research 70(1): 17-25.
Heyer, C.J., Miller, T.J., Binkowski, F.P., Caldarone, E.M., and Rice, J.A. 2001. Maternal effects as a recruitment mechanism in Lake Michigan yellow perch (Perca flavescens). Canadian Journal of Fisheries and Aquatic Sciences 58(7): 1477-1487.
Hightower, J.E. 1990. Multispecies harvesting policies for Washington-Oregon-California Rockfish Trawl Fisheries. Fishery Bulletin 88: 645-656.
Hoffman, J.C., Bonzek, C.F., and Latour, R.J. 2006. Gear efficiency estimate of ChesMMAP's bottom trawl for Chesapeake Bay multispecies management. 20th Annual Meeting of the Tidewater Chapter of the American Fisheries Society, Atlantic Beach, NC.
Holland, A.F., Shaughnessy, A.T., and Hiegel, M.H. 1987. Long-Term variation in mesohaline Chesapeake Bay macrobenthos: spatial and temporal patterns. Estuaries 10(3): 227-245.
Hollowed, A.B., Bax, N., Beamish, R., Collie, J.S., Fogarty, M., Livingston, P., Pope, J., and Rice, J.C. 2000. Are multispecies models an improvement on single-species models for measuring fishing impacts on marine ecosystems. ICES Journal of Marine Science 57: 707-719.
Homer, M., and Boynton, W.R. 1978. Stomach Analysis of Fish Collected in the Calvert Cliffs Region, Chesapeake Bay - 1977 78-154-CBL, University of Maryland Center for Environmental Studies Chesapeake Biological Laboratory, College Park.
192
Houde, E.D., and Madon, S.P. 1995. Fish in MEERC. Top-down controls-Bioenergetic models and experimental protocols, University of Maryland Center for Environmental Science, Cambridge, MD.
Houde, E.D., and Schekter, R.C. 1982. Oxygen uptake and comparative energetics among eggs and larvae of three subtropical marine fishes. Marine Biology 72(3): 283-293.
Hutchinson, G.E. 1957. Population Studies - Animal Ecology and Demography (Reprinted from Cold Spring Harbor Symposia on Quantitative Biology, Vol 22, Pg 415-427, 1957). Bulletin of Mathematical Biology 53(1-2): 193-213.
Iverson, R.L. 1990. Control of Marine fish production. Limnology and Oceanography 35(7): 1593-1604.
Jackson, J.B.C., Kirby, M.X., Berger, W.H., Bjorndal, K.A., Botsford, L.W., Bourque, B.J., Bradbury, R.H., Cooke, R., Erlandson, J., Estes, J.A., Hughes, T.P., Kidwell, S., Lange, C.B., Lenihan, H.S., Pandolfi, J.M., Peterson, C.H., Steneck, R.S., Tegner, M.J., and Warner, R.R. 2001. Historical overfishing and the recent collapse of coastal ecosystems. Science 293(5530): 629-638.
Jackson, S., Duffy, D.C., and Jenkins, J.F.G. 1987. Gastric digestion in marine vertebrate predators: in vitro standards. Functional Ecology 1: 287-291.
Jennings, S., Kaiser, M.J., and Reynolds, J.D. 2001. Marine Fisheries Ecology. Blackwell Science Ltd., Oxford, UK.
Jensen, O.P., Seppelt, R., Miller, T.J., and Bauer, L.B. 2005. Winter distribution of blue crab Callinectes sapidus in Chesapeake Bay: application and cross-validation of the two-stage generalized additive model. Marine Ecology Progress Series 299: 239-255.
Jobling, M. 1983. Effect of feeding frequency on food intake and growth of Arctic charr, Salvelinus alpinus L. Journal of Fish Biology 23(2): 177-185.
Johnson, J.H., and Dropkin, D.S. 1993. Diel Variation in Diet Composition of a Riverine Fish Community. Hydrobiologia 271(3): 149-158.
193
Jones, P.W., Wilson, J.S., Morgan, R.A., Lunsford, H.R., and Lawson, J. 1977. Potomac River fisheries study: striped bass spawning stock assessment. Interpretive report 1974-1976., University of Maryland CEES Reference Number 75-56-CBL., Solomons, MD.
Joseph, E.B. 1972. The status of Sciaenid Stocks of the Middle Atlantic Coast. Chesapeake Science 13(2): 87-100.
Jowett, I.G., and Davey, A.J.H. 2007. A comparison of composite Habitat Suitability Indices and Generalized Additive Models of invertebrate abundance and fish presence-habitat availability. Transactions of the American Fisheries Society 136: 428-444.
Juanes, F., Buckel, J.A., and Scharf, F.S. 2001. Predatory behaviour and selectivity of a primary piscivore: comparison of fish and non-fish prey. Marine Ecology-Progress Series 217: 157-165.
Jung, S. 2002. Fish Community Structure and the Spatial and Temporal Variability in Recruitment and biomass production in Chesapeake Bay. Dissertation, Marine-Estuarine-Environmental Sciences, University of Maryland, College Park.
Jung, S., and Houde, E.D. 2004. Production of bay anchovy Anchoa mitchilli in Chesapeake Bay: application of size-based theory. Marine Ecology Progress Series 281: 217-232.
Kahn, D.M., Uphoff, J.H., Crecco, V., Vaughan, D.S., Murphy, B., Brust, J.C., O'Reilly, R.L., and Paramore, L.M. 2006. Weakfish stock assessment report for peer review (Part 1), Atlantic States Marine Fisheries Commission, Washington, DC.
Kaiser, M.J., Clarke, K.R., Hinz, H., Austen, M.C.V., Somerfield, P.J., and Karakassis, I. 2006. Global analysis of response and recovery of benthic biota to fishing. Marine Ecology-Progress Series 311: 1-14.
Karlson, K., Rosenberg, R., and Bonsdorff, E. 2002. Temporal and spatial large-scale effects of eutrophication and oxygen deficiency on benthic fauna in Scandinavian and Baltic waters - A review. In Oceanography and Marine Biology, Vol 40. pp. 427-489.
Kemp, W.M., Boynton, W.R., Adolf, J.E., Boesch, D.F., Boicourt, W.C., Brush, G., Cornwell, J.C., Fisher, T.R., Glibert, P.M., Hagy, J.D., Harding, L.W., Houde, E.D., Kimmel, D.G., Miller, W.D., Newell, R.I.E., Roman, M.R., Smith, E.M., and Stevenson, J.C. 2005. Eutrophication of Chesapeake Bay: historical trends and ecological interactions. Marine Ecology-Progress Series 303: 1-29.
194
Kennish, M.J. 1986. Ecology of Estuaries: Physical and Chemical Aspects. CRC Press, Inc., Boca Raton, FL.
Kerr, S.R. 1982. Estimating the energy budgets of actively predatory fishes. Canadian Journal of Fisheries and Aquatic Sciences 39: 371-379.
Kitchell, J.F., Stewart, D.J., and Weininger, D. 1977. Applications of Bioenergetics Model to Yellow Perch (Perca flavescens) and Walleye (Stizostedion vitreum vitreum). Journal of Fisheries Research Board of Canada 34: 1922-1935.
Knowlton, N. 2004. Multiple "stable" states and the conservation of marine ecosystems. Progress In Oceanography Regime shifts in the ocean. Reconciling observations and theory 60(2-4): 387-396.
Labar, G.W. 1993. Use of bioenergetic models to predict the effect of increased lake trout predatin on rainbow smelt following sea lamprey control. Transactions of the American Fisheries Society 122: 942-950.
Lamare, M.D., and Wing, S.R. 2001. Calorific content of New Zealand marine macrophytes. New Zealand Journal of Marine and Freshwater Research 35: 335-341.
Lambert, Y., and Dutil, J.D. 1997. Condition and energy reserves of Atlantic cod (Gadus morhua) during the collapse of the northern Gulf of St. Lawrence stock. Canadian Journal of Fisheries and Aquatic Sciences 54(10): 2388-2400.
Lankford, J., Thomas E., and Targett, T.E. 2001a. Physiological Performance of Young-of-the-year Atlantic Croakers from Different Atlantic Coast Estuaries: Implications for Stock Structure. Transactions of the American Fisheries Society 130: 367-375.
Lankford, T.E., and Targett, T.E. 1994. Suitability of estuarine nursery zones for juvenile weakfish (Cynoscion regalis): effects of temperature and salinity on feeding, growth, and survival. Marine Biology 119: 611-620.
Lankford, T.E., and Targett, T.E. 2001b. Low-temperature tolerance of Age-0 Atlantic Croakers: Recruitment implications for U.S. Mid-Atlantic Estuaries. Transactions of the American Fisheries Society 130: 236-249.
195
Latour, R.J., Brush, M.J., and Bonzek, C.F. 2003. Towards Ecosystem-Based Fisheries Management: Strategies for Multispecies Modelling and Associated Data Requirements. Fisheries 28(9): 10-22.
Laurence, G.C. 1977. A bioenergetic model for the analysis of feeding and survival potential of winter flounder, Pseudopleuronectes americanus, larvae during the period from hatching to metamorphosis. U.S. National Marine Fisheries Bulletin 75: 529-546.
Libralato, S., Christensen, V., and Pauly, D. 2006. A method for identifying keystone species in food web models. Ecological Modelling 195(3-4): 153-171.
Limburg, K.E. 1996. Modelling the ecological constraints of growth and movement of juvenile American Shad (Alosa sapidissima) in the Hudson River Estuary. Estuaries 19(4): 794-813.
Link, J.S. 2002. Ecological considerations in fisheries management: When does it matter? Fisheries 27(4): 10-17.
Link, J.S., and Garrison, L.P. 2002. Changes in piscivory associated with fishing induced changes to the finfish community on Georges Bank. Fisheries Research 55: 71-86.
Lowerre-Barbieri, S.K., Chittenden, M.E., Jr., and Barbieri, L.R. 1995. Age and growth of weakfish, Cynoscion regalis, in the Chesapeake Bay region with a discussion of historical changes in maximum size. Fishery Bulletin 93: 643-656.
Lowerre-Barbieri, S.K., Chittenden, M.E., Jr., and Barbieri, L.R. 1996. The multiple spawning pattern of weakfish in the Chesapeake Bay and Middle Atlantic Bight. Journal of Fish Biology 48: 1139-1163.
Lukaski, H.C. 1987. Methods for the assessment of human body composition: traditional and new. American Journal of Clinical Nutrition 46: 537-556.
Luo, J., and Brandt, S.B. 1993. Bay anchovy Anchoa mitchilli production and consumption in mid-chesapeake Bay based on a bioenergetics model and acoustic measures of fish abundance. Marine Ecology Progress Series 98: 223-236.
Luo, J., Hartman, K.J., Brandt, S.B., Cerco, C.F., and Rippetoe, T.H. 2001. A spatially-explicit approach for estimating carrying capacity: An application for the Atlantic menhaden (Brevoortia tyrannus) in Chesapeake Bay. Estuaries 24(4): 545-556.
196
Malloy, K., and Targett, T. 1994. The use of RNA:DNA ratios to predict growth limitation of juvenile summer flounder (Paralichthys dentatus) from Delaware and North Carolina estuaries. Marine Biology 118: 367-375.
Malloy, K.D., Yamashita, Y., Yamada, H., and Target, T.E. 1996. Spatial and temporatl patterns of juvenile stone flounder Kareius bicoloratus growth rates during and after settlement. Marine Ecology Progress Series 131: 49-59.
Mangel, M., and Levin, P.S. 2005. Regime, pahse and paradigm shifts: making community ecology the basic science for fisheries. Philosophical Transactions of the Royal Society 360: 95-105.
Mann, K.H., and Lazier, J.R. 1996. Dynamics of Marine Ecosystems: Biological-Physical Interactions in the Oceans. Blackwell Science, Inc., Cambridge, Massachusetts.
Mansueti, R.J. 1961. Age, growth, and movements of the striped bass, Roccus saxatilis, taken in size selective fishing gear in Maryland. Chesapeake Science 2: 9-36.
Marshall, C.T., Kjesbu, O.S., Yaragina, N.A., Solemdal, P., and Ulltang, O. 1998. Is spawner biomass a sensitive measure of the reproductive and recruitment potential of Northeast Arctic cod? Canadian Journal of Fisheries and Aquatic Sciences 55: 1766-1783.
Marshall, C.T., Needle, C.L., Thorsen, A., Kjesbu, O.S., and Yaragina, N.A. 2006. Systematic bias in estimates of reproductive potential of an Atlantic cod (Gadus morhua) stock: implication for stock-recruit theory and management. Canadian Journal of Fisheries and Aquatic Sciences 63: 980-994.
Marshall, C.T., Yaragina, N.A., Adlandsvik, B., and Dolgov, A.V. 2000. Reconstructing the stock-recruitment relationship for Northeast Arctic cod using a bioenergetic index of reproductive potential. Canadian Journal of Fisheries and Aquatic Sciences 57: 2433-2442.
Marshall, C.T., Yaragina, N.A., Lambert, Y., and Kjesbu, O.S. 1999. Total lipid energy as a proxy for total egg production by fish stocks. Nature 402: 288-290.
Mason, D.M., Johnson, T.B., and Kitchell, J.F. 1998. Consequences of prey fish community dynamics on lake trout (Salvelinus namaycush) foraging efficiency in Lake Superior. Canadian Journal of Fisheries and Aquatic Sciences 55(5): 1273-1284.
197
Massman, W.H., and Pcheco, A.L. 1961. Movements of striped bass in Virginia waters of Chesapeake Bay. Chesapeake Science 2: 37-44.
May, R.M., Beddington, H.J.R., Clark, C.W., Holt, S.J., and Laws, R.M. 1979. Management of Multispecies Fisheries. Science 205: 267-277.
McCann, K.S. 2000. The diversity-stability debate. Nature 405: 228-233.
McCann, K.S., Hastings, A., and Huxel, G.R. 1998. Weak trophic interactions and the balance of nature. Nature 395: 794-798.
Megrey, B.A., Rose, K.A., Klumb, R.A., Hay, D.E., Werner, F.E., Eslinger, D.L., and Smith, S.L. 2007a. A bioenergetics-based population dynamics model of Pacific herring (Clupea harengus pallasi) coupled to a lower trophic level nutrient-phytoplankton-zooplankton model: Description, calibration, and sensitivity analysis. Ecological Modelling 202: 144-164.
Megrey, B.A., Rose, K.A., Klumb, R.A., Hay, D.E., Werner, F.E., Eslinger, D.L., and Smith, S.L. 2007b. A bioenergetlics-based population dynamics model of Pacific herring (Clupea harengus pallasi) coupled to a lower trophic level nutrient-phytoplankton-zooplankton model: Description, calibration, and sensitivity analysis. Ecological Modelling 202(1-2): 144-164.
Melville, A.J., and Connolly, R.M. 2003. Spatial analysis of stable isotope data to determine primary sources of nutrition for fish. Oecologia 136(4): 499-507.
Melzner, F., Forsythe, J.W., Lee, P.G., Wood, J.B., Piatkowski, U., and Clemmesen, C. 2005. Estimating recent growth in the cuttlefish Sepia officinalis: are nucleic acid-based indicators for growth and condition the method of choice? Journal of Experimental Marine Biology and Ecology 317(1): 37-51.
Methratta, E.T. 2006. Association between surficial sediments and groundfish distributions in the Gulf of Maine-George Bank Region. North American Journal of Fisheries Management 26: 473-489.
Miller, J.M., Crowder, L.B., and Moser, M.L. 1985. Migration and utilization of estuarine nurseries by juvenile fishes: An evolutionary perspective. In Contributions in Marine Science. Migration: Mechanisms and adaptive significance. Edited by M. Rankin, D. Checkley, J. Cullen, C. Kitting and P. Thomas, Port Aransas, TX. pp. 338-352.
198
Miller, T.J., Houde, E.D., and Watkins, E.J. 1996. Chesapeake Bay fisheries: Prospects for multspecies management and sustainability. Chesapeake Research Consortium (CRC), Edgewater, MD.
Mittelbach, G.G., Osenberg, C.W., and Wainwright, P.C. 1992. Variation in resource abundance affects diet and feeding morphology in the pumpkinseed sunfish (Lepomis gibbosus). Oecologia 90: 8-13.
Mittelbach, G.G., and Persson, L. 1998a. The ontogeny of piscivory and its ecological consequences. Canadian Journal of Fisheries and Aquatic Sciences 55(6): 1454-1465.
Mittelbach, G.G., and Persson, L. 1998b. The ontogeny of piscivory and its ecological significance. Canadian Journal of Fisheries and Aquatic Sciences 55: 1454-1465.
Moellmann, C., Kornilovs, G., Fetter, M., Koester, F., and Hinrichsen, H. 2003. The marine copepod, Pseudocalanus elongatus, as a mediator between climate variability and fisheries in the Central Baltic Sea. Fisheries Oceanography 12(4-5): 360-368.
Montane, M.M., and Austin, H.M. 2005. Effects of hurricanes on Atlantic croaker (Micropogonias undulatus) recruitment to Chesapeake Bay. In Hurricane Isabel in Perspective. Edited by K.G. Sellner. Chesapeake Research Consortium, Edgewater, MD.
Mukherjee, S., and Jana, B.B. 2007. Water quality affects SDH activity, protein content and RNA : DNA ratios in fish (Catla catla, Labeo rohita and Oreochromis mossambicus) raised in ponds of a sewage-fed fish farm. Aquaculture 262(1): 105-119.
Nemerson, D.M. 2002. Trophic dynamics and habitat ecology of the dominant fish of Delaware Bay (United States of America) marsh creeks, Rutgers, the State University of New Jersey, New Brunswick, NJ.
Ney, J.J. 1990. Trophic economics in fisheries: assessment of demand-supply relationships between predators and prey. Reviews of Aquatic Sciences 2: 55-81.
Niklitschek, E.J. 2001. Bioenergetics Modeling and Assessment of Suitable Habitat for Juvenile Atlantic and Shortnose Sturgeons in the Chesapeake Bay. Dissertation, Marine Estuarine and Environmental Science, University of Maryland, College Park.
199
Nislow, K.H., Folt, C.L., and Parrish, D.L. 2000. Spatially explicit bioenergetic analysis of habitat quality for Age-0 Atlantic salmon. Transactions of the American Fisheries Society 129: 1067-1081.
Nixon, S.W. 1988. Physical energy inputs and the comparative ecology of lake and marine ecosystems. Limnology and Oceanography 33(4, part 2): 1005-1025.
Nixon, S.W., and Buckley, B.A. 2002. "Strikingly Rich Zone"-Nutrient enrichment and secondary production in coastal marine ecosystems. Estuaries 25(4b): 782-796.
Nixon, S.W., and Jones, C.M. 1997. Age and growth of larval and juvenile Atlantic croaker, Micropogonias undulatus, from the Middle Atlantic Bight and estuarine waters of Virginia. Fishery Bulletin 95: 773-784.
Norcross, B.L. 1991. Estuarine Recruitment Mechanisms of Larval Atlantic Croakers. Transactions of the American Fisheries Society 120(6): 673-683.
North, E.W., and Houde, E.D. 2006. Retention mechanisms of white perch (Morone americana) and striped bass (M. saxatilis) early-life stages in an estuarine turbitiy maximum: an integrative mapping and Eulerian approach. Fisheries Oceanography 15(6): 429-450.
Oksanen, J., Kindt, R., Legendre, P., O'Hara, B., and Steven, H.H. 2007. The vegan Package. Available from http://cran.r-project.org/src/contrib/Descriptions/vegan.html.
Osse, J.W.M., Boogaart, J.G.M.c.d., Snik, G.M.J.v., and Sluys, L.v.d. 1997. Priorities during early growth of fish larvae. Aquaculture 155: 249-258.
Overstreet, R.M., and Heard, R.W. 1978. Food of the Atlantic Croaker, Micropogonias undulatus, from Mississippi Sound and the Gulf of Mexico. Gulf Research Reports 6(2): 145-152.
Page, H.M., Dugan, J.E., Schroeder, D.M., Nishimoto, M.M., Love, M.S., and Hoesterey, J.C. 2007. Trophic links and condition of a temperate reef fish: comparisons among offshore oil platform and natural reef habitats. Marine Ecology-Progress Series 344: 245-256.
Paine, R.T. 1966. Food Web Complexity and Species Diversity. American Naturalist 100(910): 65-&.
200
Paine, R.T. 1995. A Conversation on Refining the Concept of Keystone Species. Conservation Biology 9(4): 962-964.
Pandolfi, J.M., Bradbury, R.H., Sala, E., Hughes, T.P., Bjorndal, K.A., Cooke, R.G., McArdle, D., McClenachan, L., Newman, M.J.H., Paredes, G., Warner, R.R., and Jackson, J.B.C. 2003. Global Trajectories of the Long-Term Decline of Coral Reef Ecosystems 10.1126/science.1085706. Science 301(5635): 955-958.
Parsons, T.R. 1992. The removal of marine predators by fisheries and the impact of trophic structure. Marine Pollution Bulletin 25(1-4): 51-53.
Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., and Torres, F., Jr. 1998. Fishing Down Marine Food Webs 10.1126/science.279.5352.860. Science 279(5352): 860-863.
Peck, M.A., Buckley, L.J., Caldarone, E., M., and Bengtson, D.A. 2003. Effects of food consumption and temperature on growth rate and biochemical indicators of growth in early juvenile cod Gadus morhua and haddock Melangogrammus aeglefinus. Marine Ecology Progress Series 251: 233-243.
Peck, M.A., Clemmesen, C., and Herrmann, J.P. 2005. Ontogenetic changes in the allometric scaling of the mass and length relationship in Sprattus sprattus. Journal of Fish Biology 66(3): 882-887.
Persson, A., and Hansson, L.-A. 1998. Diet shift in fish following competitive release. Canadian Journal of Fisheries and Aquatic Sciences 56: 70-78.
Pihl, L. 1994. Changes in the diet of demersal fish due to eutrophication-induced hypoxia in the Kattegat, Sweden. Canadian Journal of Fisheries and Aquatic Sciences 51: 321-336.
Pihl, L., Baden, S.P., and Diaz, R.J. 1991. Effects of hypoxia on distribution of demersal fish and crustaceans. Marine Biology 108: 349-360.
Pihl, L., Baden, S.P., Diaz, R.J., and Schaffner, L.C. 1992. Hypoxia-Induced Structural-Changes in the Diet of Bottom-Feeding Fish and Crustacea. Marine Biology 112(3): 349-361.
201
Power, M.E., Tilman, D., Estes, J.A., Menge, B.A., Bond, W.J., Mills, L.S., Daily, G., Castilla, J.C., Lubchenco, J., and Paine, R.T. 1996. Challenges in the quest for keystones. Bioscience 46(8): 609-620.
Powers, S.P., Peterson, C.H., Christian, R.R., Sullivan, E., Powers, M.J., Bishop, M.J., and Buzzelli, C.P. 2005. Effects of eutrophication on bottom habitat and prey resources of demersal fishes. Marine Ecology Progress Series 302: 233-243.
Quinn, T., and Deriso, R.B. 1999. Quantitative fish dynamics. Oxford University Press, New York.
Rand, P.S., and Stewart, D.J. 1998a. Dynamics of salmonine diets and foraging in Lake Ontario, 1983 1993: a test of a bioenergetic model prediction. Canadian Journal of Fisheries and Aquatic Sciences 55: 307-317.
Rand, P.S., and Stewart, D.J. 1998b. Prey fish exploitation, salmonine production, and pelagic food web efficiency in Lake Ontario. Canadian Journal of Fisheries and Aquatic Sciences 55: 318-327.
Reid, P.C., Borges, M.d.F., and Svendsen, E. 2001. A regime shift in the North Sea circa 1988 linked to changes in the North Sea horse mackerel fishery. Fisheries Research 50(1-2): 163-171.
Rice, J.A., and Cochran, P.A. 1984. Independent evaluation of a bioenergetics model for largemouth bass. Ecology 65(3): 732-739.
Ricker, W.E. 1975. Computation and interpretation of biological statistics of fish populations. Bulletin of the Fisheries Research Board of Canada 191: 1-382.
Roff, D.A. 1983. An allocation model of growth and reproduction in fish. Canadian Journal of Fisheries and Aquatic Sciences 40: 1395-1404.
Roff, D.A. 1984. The evolution of life history parameters. Canadian Journal of Fisheries and Aquatic Sciences 41: 989-1000.
Rooker, J.R., Holt, G.J., and Holt, S.A. 1997. Condition of larval and juvenile red drum (Sciaenops ocellatus) from estuarine nursery habitats. Marine Biology 127(3): 387-394.
202
Rothschild, B.J., and Shannon, L.J. 2004. Regime shifts and fishery management. Progress In Oceanography. Regime shifts in the ocean. Reconciling observations and theory 60(2-4): 397-402.
Rowan, D.J., and Rasmussen, J.B. 1996. Measuring the bioenergetic cost of fish activity in situ using a globally dispersed radiotracer (137CS). Canadian Journal of Fisheries and Aquatic Sciences 53: 734-745.
Rueda, M. 2001. Spatial distribution of fish species in a tropical estuarine lagoon: a geostatistical appraisal. Marine Ecology Progress Series 222: 217-226.
Sargent, J.R., McEvoy, L.A., and Bell, J.G. 1997. Requirements, presentation and sources of polyunsaturated fatty acids in marine fish larval feeds. Aquaculture 155: 117-127.
Secor, D.H., and Piccoli, P.M. 2007. Oceanic migration rates of Upper Chesapeake Bay striped bass (Morone saxatilis), determined by otolith microchemical analysis. Fishery Bulletin 105: 62-73.
Seitz, R.D., Jr, L.S.M., Hines, A.H., and Clark, K.L. 2003. Effects of hypoxia on predator-prey dynamics of the blue crab Callinectes sapidus and the Baltic clam Macoma balthica in Chesapeake Bay. Marine Ecology Progress Series 257: 179-188.
Seitz, R.D., and Schaffner, L.C. 1995. Population Ecology and Secondary Production of the Polychaete Loimia-Medusa (Terebellidae). Marine Biology 121(4): 701-711.
Setzler, E.M., Boynton, W.R., Wood, K.V., Zion, H.H., Lubbers, L., Mountford, N.L., Frere, P., Tucker, L., and Mihursky, J.A. 1980. Synopsis of biological data on striped bass, Morone saxatilis (Walbaum). NOAA/NMFS Technical Report 433, 69 pp.
Sheridan, P.F. 1979. Trophic resource utilization by three species of Sciaenid fishes in a northwest Florida Estuary. Northeast Gulf Science 3(1): 1-15.
Shoji, J., and Tanaka, M. 2001. Strong piscivory of Japanese Spanish mackerel larvae from their first feeding. Journal of Fish Biology 59(6): 1682-1685.
Shuter, B.J., and Post, J.R. 1990. Climate, Population Viability, and the Zoogeography of Temperate Fishes. Transactions of the American Fisheries Society 119(2): 314-336.
203
Simpson, A.W., and Watling, L. 2006. An investigation of the cumulative impacts of shrimp trawling on mud-bottom fishing grounds in the Gulf of Maine: effects on habitat and macrofaunal community structure. Ices Journal of Marine Science 63(9): 1616-1630.
Smith, T.R., and Buckley, L.J. 2003a. RNA-DNA Ratio in Scales from Juvenile Cod Provides a nonlethal measure of feeding condition. Transactions of the American Fisheries Society 132: 9-17.
Sparre, P., and Venema, S.C. 1998. Introduction to tropical fish assessment FAO Fisheries Technical Paper, 306/1, FAO.
St. John, M.A., and Lund, T. 1996. Lipid biomarkers: linking the utilization of frontal plankton biomass to enhanced condition of juevnile North Sea cod. Marine Ecology Progress Series 131: 75-85.
Steele, J.H. 2004. Regime shifts in the ocean: reconciling observations and theory. Progress In Oceanography. Regime shifts in the ocean. Reconciling observations and theory 60(2-4): 135-141.
Stephens, D.W., and Krebs, J.R. 1986. Foraging Theory. Princeton University Press, Princeton, NJ.
Stierhoff, K.L., Targett, T.E., and Miller, K. 2006. Ecophysiological responses of juvenile summer and winter flounder to hypoxia: experimental and modeling analyses of effects on estuarine nursery quality. Marine Ecology Progress Series 325: 255-266.
Stoner, A., Manderson, J., and Pessutti, J. 2001. Spatially explicit analysis of esturaine habitat for juvenile winter flounder: combining generalized additive models and geopgraphic information systems. Marine Ecology Progress Series 213: 253-271.
Striped Bass Technical Committee for the Atlantic Striped Bass Management Board. 2005. 2005 Stock Assessment Report for Atlantic Striped Bass: Catch at Age Based VPA and Tag Release/Recovery Bases Survival Estimation, Washington, D.C.
Suthers, I.M. 1998. Bigger? Fatter? Or is faster growth better? Considerations on condition in larval and juvenile coral-reef fish. Australian Journal of Ecology 23(3): 265-273.
204
Suthers, I.M., Fraser, A., and Frank, K.T. 1992. Comparison of Lipid, Otolith and Morphometric Condition Indexes of Pelagic Juvenile Cod Gadus-Morhua from the Canadian Atlantic. Marine Ecology-Progress Series 84(1): 31-40.
Sutton, S.G., Bult, T.P., and Haedrich, R.L. 2000. Relationships among fat weight, body weight, water weight, and condition factors in wild Atlantic salmon parr. Transactions of the American Fisheries Society 129: 527-538.
Taylor, J.C., Rand, P.S., and Jenkins, J. 2007. Swimming behavior of juvenile anchovies (Anchoa spp.) in an episodically hypoxic estuary: implications for individual energetics and trophic dynamics. Marine Biology 152(4): 939-957.
ter Braak, C.J.F. 1986. Canonical Correspondence Analysis: A New Eigenvector Technique for Multivariate Direct Gradient Analysis. Ecology: 1167-1179.
Thorrold, S.R., Latkoczy, C., Swart, P.K., and Jones, C.M. 2001. Natal homing in a marine fish metapopulation. Science 291(5502): 297-299.
Tillin, H.M., Hiddink, J.G., Jennings, S., and Kaiser, M.J. 2006. Chronic bottom trawling alters the functional composition of benthic invertebrate communities on a sea-basin scale. Marine Ecology-Progress Series 318: 31-45.
Tirasin, E.M., and Jorgensen, T. 1999. An evaluation of the precision of diet description. Marine Ecology Progress Series 182: 243-252.
Trudel, M., and Bosclair, D. 1994. Seasonal consumption by dace (Phoxinus eos x P. neogaeus): A comparison between field and bioenergetic model estimates. Canadian Journal of Fisheries and Aquatic Sciences 51: 2558-2567.
Tyler, R.M., and Targett, T.E. 2007. Juvenile weakfish Cynoscion regalis distribution in relation to diel-cycling dissolved oxygen in an estuarine tributary. Marine Ecology Progress Series 333: 257-269.
Ueberschar, B., and Clemmesen, C. 1992. A comparison of the nutritional condition of herring larvae as determined by two biochemical methods - tryptic enzyme activity and RNA/DNA ratio measurements 10.1093/icesjms/49.2.245. ICES J. Mar. Sci. 49(2): 245-249.
205
Uphoff, J.H. 2006. Weakfish stock assessment report for peer review (Part 2). Atlantic States Marine Fisheries Commission, Washington, D.C.
Uphoff Jr., J.H. 2003. Predator-prey analysis of striped bass and Atlantic menhaden in upper Chesapeake Bay. Fisheries Management and Ecology 10: 313-322.
Virnstein, R.W. 1977. The importance of predation by crabs and fishes on benthic infauna in Chesapeake Bay. Ecology 58: 1199-1217.
Wagner, M., Durbin, E., and Buckley, L. 1998. RNA : DNA ratios as indicators of nutritional condition in the copepod Calanus finmarchicus. Marine Ecology-Progress Series 162: 173-181.
Walters, C., Christensen, V., and Pauly, D. 1997. Structuring dynamic models of exploited ecosystems from trophic mass-balance assessments. Reviews in Fish Biology and Fisheries 7(2): 139-172.
West, J.B., Brown, G.J., Cerling, T.E., and Ehleringer, J.R. 2006. Stable isotopes as one of nature's ecological recorders. Trends in Ecology and Evolution 21(7): 408-414.
Whipple, S.J., Link, J.S., Garrison, L.P., and Fogarty, M.J. 2000. Models of predation and fishing mortality in aquatic ecosystems. Fish and Fisheries 1: 22-40.
Whitledge, G.W., Hayward, R.S., and Zweifel, R.D. 2003. Development and Laboratory Evaluation of a bioenergetics model for subadult and adult smallmouth bass. Transactions of the American Fisheries Society 132: 316-325.
Winberg, G.G. 1956. Rate of metabolism and food requirements of fishes, Belorussian University.
Wood, S.J. 2007. The mgcv Package. Available from http://cran.r-project.org/src/contrib/Descriptions/mgcv.html.
Wootton, R.J. 1998. Ecology of Teleost Fishes. Kluwer Academic Publishers, Norwell, MA.
Yaragina, N.A., and Marshall, C.T. 2000. Trophic influences on interannual and seasonal variation in the liver condition index of Northeast Arctic cod. ICES Journal of Marine Science 57: 42-55.
206
Yodzis, P. 1994. Predator-Prey Theory and management of multispecies fisheries. Ecological Applications 4(1): 51-58.
207
CHAPTER 6
DESIGN EFFICIENCIES OF TRANSECT AND
STRATIFIED RANDOM TRAWL SURVEYS
Jon H. Vølstad1, Mary Christman2, and Thomas J. Miller3
1Versar, Inc., 9200 Rumsey Rd., Columbia, MD 21045, USA, Tel: 410-964-9200,
Fax: 410-964-5156, E-mail address: [email protected]
2Department of Statistics – Institute for Food and Agricultural Sciences, University of Florida,
Gainesville, FL 32611, USA
3 Chesapeake Biological Laboratory, University of Maryland Center for Environmental Science,
Solomons, MD 20688, USA.
208
6.1 ABSTRACT
Spatially overlapping transect and stratified random trawl surveys were conducted
simultaneously in the Chesapeake Bay during spring, summer, and fall during 2002 and 2003 to
quantify the relative abundance, diversity, distribution and trophic interactions of economically
and ecologically important fish species. We evaluated the design efficiency of each survey with
respect to the precision of relative abundance estimates for a given sampling effort. We used
Kish’s design effects and effective sample sizes to assess the efficiency and cost-effectiveness of
each survey with respect to estimates of relative abundance for selected species and overall. The
empirical design effects suggested that the stratified random survey, with proportional allocation,
was more effective than the transect sampling in estimating mean CPUE across species, and for
most target species. However, for weakfish, trawling along transects appeared to be more
effective than the stratified random survey. We developed a composite estimator to combine the
estimates of mean CPUE from the two independent surveys within season for each year, using
weights that are based on their respective effective sample sizes. The weights are optimized with
respect to achieving minimum variance. Our method for combining estimates across surveys is
robust to heterogeneous variances such as might be found when the means and variances are
correlated, which is often the case for CPUE from marine trawl surveys. A similar approach,
with possible modifications for gear differences, could also be used to combine multiple surveys
to develop a composite index for use in stock assessments.
Keywords: Marine trawl surveys; Survey design; Design effect; Transect surveys; Effective
sample size; Cluster-sampling; Intra-cluster correlation; Composite estimator.
209
6.1 INTRODUCTION
Fisheries-independent trawl surveys are widely used to provide estimates of relative abundance
and other population parameters crucial for effective fisheries management (Gunderson, 1993).
The average number of fish caught per area or volume swept has long been used as an index of
relative abundance (Grosslein, 1969; Pennington, 1985, 1986; Smith, 2002). Standardized trawl
surveys, coupled with appropriate allocation of stations, can provide reliable estimates of
changes in abundance, if the catch efficiency of the survey gear remains approximately constant
by depth and over time. However, given the expense of trawl programs, it is important to
optimize the survey design so that the precision of key parameters is maximized for a fixed total
survey cost.
The distribution of marine resources is generally highly patchy (cf. Seber, 1986), and
often results in strong short-range spatial autocorrelation in measures of relative abundance
(Polacheck and Vølstad, 1993; Kingsley et al., 2002). Accordingly, a considerable body of
research has sought to determine survey designs appropriate for temporal and spatial variability
with possible clustering. A classic example includes the application of stratification and sample
allocation schemes that seek to increase inter-stratum variability while minimizing intra-stratum
variability, thereby increasing the precision of parameter estimates for single species (Cochran,
1977; Gavaris and Smith, 1987; Harbitz et al., 1998). Designs for surveys that target multiple
species are more complicated as distributions of the individual species are likely to vary at
differing spatial and temporal scales. The recent interest in multispecies fisheries management
will require development of multispecies, fishery-independent surveys. However, guidance on
optimization of multispecies surveys is lacking.
The presence of biological and technical interactions among Chesapeake Bay fisheries
(Miller et al. 1996) has motivated management agencies to establish a goal of implementing
multispecies management in the Chesapeake Bay by 2007 (C2K agreement). To support the
development of multispecies management in the Chesapeake Bay an international workshop
recommended the development of coordinated, baywide surveys to estimate key species
abundances and to provide biological data on both economically and ecologically important
species that are currently lacking (Houde et al. 1998).
210
We conducted seasonal midwater trawl surveys in the Chesapeake Bay to evaluate the
efficiency of different survey designs for a range of finfish species as a component of the
Chesapeake Bay Fishery Independent Multispecies Survey (CHESFIMS). Information for the
initial survey design came from an earlier research program to quantify ‘Trophic Interactions in
Estuarine Systems’ (TIES). The TIES program included baywide midwater trawl surveys,
conducted from 1995 - 2000 during spring, summer and fall (Jung and Houde, 2003). The TIES
data demonstrated that baywide seasonal surveys are required to characterize the communities in
the Bay because of the spatial and temporal variability in fish abundance and community
structure (Jung and Houde, 2003).
Our goals for the design of CHESFIMS were to: (1) maintain the time series from the
TIES program while improving survey efficiency (2) complement existing fishery-independent
surveys and (3) expand survey coverage to ecologically important forage species to aid
development of multispecies fisheries management in the Chesapeake Bay. We sought to
develop a cost-effective design for the monitoring of abundance, diversity, distribution and
trophic interactions of economically and ecologically important fish species in the Chesapeake
Bay over broad spatial and temporal scales given limited resources. Accordingly, we conducted
spatially and temporally overlapping transects and stratified random trawl surveys in the
Chesapeake Bay during spring, summer, and fall in 2002 and 2003. Here we evaluate the
efficiency of both survey designs for estimating relative abundance using design effects and
effective sample sizes (Lehtonen and Pakinen, 1994; Kish, 2003). We also demonstrate how two
survey indices of abundance can be combined to yield an overall index with minimum variance.
6.2. MATERIAL AND METHODS
6.2.1. Study area
The Chesapeake Bay is the largest estuary in the United States, with an area of
approximately 6,500 km2, a length of over 300 km, a mean depth of 8.4 meters, and a volume of
52,112 km3. Fisheries in Chesapeake Bay contribute significantly to U.S. catches at the national
and regional levels. Fisheries statistics from the National Marine Fisheries Service (NMFS)
shows that around 220,000 metric tons (t) of fish and shellfish were harvested in 2002 from
211
Chesapeake Bay waters, with a dockside value of more than $172 million. All exploited species
in the Chesapeake Bay are currently managed and regulated on a single species basis.
Bay anchovy (Anchoa mitchilli) is the most abundant and ubiquitous fish in the Bay
(Jung and Houde, 2004). Although not of commercial importance, bay anchovy has an important
role in the ecosystem because it is a major prey of piscivores, including several economically
important fishes (Baird and Ulanowicz, 1989; Luo and Brandt, 1993; Hartman and Brandt,
1995). Several anadromous species were also common in survey catches including white perch
(Morone americana), and members of the Sciaenidae family such as Atlantic croaker
(Micropogonias undulatus), and weakfish (Cynoscion regalis).
6.2.2. Survey methods
Standardized trawl surveys have been conducted during spring, summer, and fall in the
mainstem of the Chesapeake Bay since 1995 using an 18-m2 midwater trawl (MWT) with 3-mm
codend mesh. MWTs are commonly used to survey pelagic fish including anchovy (Anchoa
japonica -- Komatsu et al. 2002), walleye pollock (Theragra chalcogramma --Wilson 2000), and
Pacific whiting (Merluccius productus – Sakuma and Ralston 1997) and in commercial fisheries
for many of these same species. The MWT employed in the Bay samples fish 30-256 mm total
length of most species effectively, but appears to be less effective for Atlantic menhaden
(Brevoortia tyrannus) of all sizes (Jung and Houde, 2003). Trawl stations in the TIES program
were located along 15 fixed transects spaced approximately 18.5 km (10 nm) apart from the head
of the Bay to the Bay mouth to ensure bay wide coverage (Jung and Houde, 2003). Within each
season, 11 of the 15 transects were occupied. Transects were identified as falling within one of
three upper, middle, and lower Bay strata (Figure 6.1). Individual strata have distinctive
characteristics, and their boundaries broadly corresponding to ecologically relevant salinity
regimes and depths above 5 m. The upper Bay is generally shallow, with substantial areas with
depths less than 5 m, and has well mixed waters with high nutrient concentrations. The bottom
topography in the mid Bay includes a narrow channel in the middle of the Bay with a stratified
water column and broad flanking shoals. This region has relatively clear waters and experiences
seasonally high nutrient concentrations and periods of hypoxia. The lower Bay has the clearest
212
waters, greatest depths and lowest nutrient concentrations (Kemp et al., 1999). The strata
volumes are 26,608 km3 (Lower), 16,840 km3 (Mid) and 8,664 km3 (Upper).
CHESFIMS was initiated in 2001, employing the TIES trawling procedures, transect
design and stratification, with 2-4 trawl stations sampled within 11 transects during spring,
summer, and fall. Sampling was conducted from the University of Maryland Center for Estuarine
and Environmental Science’s R/V ‘Aquarius. The im stations within the ith transect were
selected by restricted random sampling. Each transect was divided into im segments of equal
size, with one station allocated randomly within each segment. For the transect surveys, the
clustering of stations and variable transect lengths resulted in heterogeneous selection
probabilities for stations within strata. Starting in 2002, the transect sampling was augmented
with an independent, stratified random trawl survey in an effort to optimize the monitoring
design. We were restricted to maintaining the transects used for monitoring and so chose
stratified random sampling as an appropriate additional sampling design for optimizing the
geographic distribution of sampling locations and as a counterpoint to the transect design. In the
stratified random surveys, conducted simultaneously with the transect surveys, 20 stations
covering the entire bay were allocated to the each stratum proportional to their volumes. Twenty
new locations were chosen during each survey. Latitude and longitude of stations within strata
were randomly generated. Weather precluded complete sampling of all 20 stations during some
surveys. An example of the allocation of stations in the 2002 stratified random survey is shown
in Figure 6.1.
Survey deployments followed the TIES trawling procedures (Jung and Houde, 2003),
with standardized 20-minute oblique, stepped tows conducted at each station using the same
midwater trawl. The trawl was towed for two minutes in each of ten depth zones distributed
throughout the water column from the surface to the bottom, with minimum trawlable depth
being 5 m. The section of the tow conducted in the deepest zone sampled epibenthic fishes close
to or on the bottom. The remaining portion of the tow sampled pelagic and neustonic fishes. All
tows were conducted between 19:00 and 7:00 Eastern Standard Time to minimize gear
avoidance and to take advantage of the reduced patchiness of multiple target species at night.
Catches were identified, enumerated, measured and weighed onboard.
213
6.2.3. Estimating mean catch per unit effort
To estimate the mean catch per unit effort (CPUE) and the associated variance, we
treated the transect survey as a stratified, two-stage design, with primary sampling units of
unequal size (e.g., Cochran, 1977; Wolter, 1985) to account for the variable transect lengths. For
estimating purposes, we assumed that the hn primary sampling units (transects) were selected
randomly from all possible transects within each stratum h ( 1, 2,3h = ). Let hiy denote the mean
CPUE for the him stations within transect i in stratum h , and let hil denote the transect length.
We applied a combined (across strata) ratio estimator for a two-stage survey (Cochran, 1977) to
estimate the overall mean CPUE 3
13
1
h h hh
T
h hh
w l yy
w l
=
=
=∑
∑ [6.1]
where, for stratum h ,
1
1
h
h
n
hi hii
h n
hii
l yy
l
=
=
=∑
∑
is the weighted mean CPUE across transects,
1
1 hn
h hiih
l ln =
= ∑
is the mean transect length within a stratum, and
3
1
hh
hh
VwV
=
=
∑ [6.2]
is the stratum weight, with hV being the stratum size (volume). The transect-wise variation of
mean CPUE as well as the variation in estimated mean transect length by stratum contributes the
variance of Eq. 6.1. An approximate estimator of the variance of Eq. 6.1 is (see Sukhatme and
Sukhatme, 1970, p. 307)
214
22 232
21 1
1( )hn
hb hi hiT h
h ih h h hi
s l sVar yn n l m
λ= =
⎧ ⎫⎛ ⎞⎪ ⎪= +⎨ ⎬⎜ ⎟⎝ ⎠⎪ ⎪⎩ ⎭
∑ ∑ , [6.3]
where
3
1
h hh
h hh
w l
w lλ
=
=
∑
is the proportion of second-stage units in stratum h ,
2 2
1
1 ( )1
hn
hb hi hih
s y yn =
= −− ∑
is the between transect variance in CPUE for stratum h , and
2 2
1
1 ( )1
him
hi hij hijhi
s y ym =
= −− ∑
is the within transect variance in CPUE for transect i in stratum h . We used SUDAAN (RTI
2001), a specialized software for the analysis of complex surveys and cluster-correlated data
(Brogan, 1998; Carlson, 1998), to estimate the variance of Ty (Eq. 6.3).
In the stratified random surveys (STR), stations were allocated proportional to the
volume of each stratum. We applied the standard estimators for the stratified mean and its
variance (Cochran, 1977) 3
,1
str h h srsh
y w y=
=∑ [6.4]
and 3
2,
1( ) ( )str h h srs
hVar y w Var y
=
=∑ [6.5]
where the weight for stratum h is based on its fraction of the total volume as in equation 6.2,
,h srsy is the ordinary mean CPUE for simple random sampling within stratum h, and ,( )h srsVar y
is the corresponding variance of the stratum mean CPUE. The spring survey in 2002, and the fall
survey in 2003 had incomplete sampling coverage for logistical reasons. For these surveys we
collapsed the strata and treated all stations as a simple random sample (SRS), and then used the
ordinary estimators of the mean and variance.
215
6.2.4. Analytical evaluation of design efficiency
The efficiency of each survey design was evaluated by comparing the respective design-
based variance of the estimated mean CPUE ( y ) with the expected variance obtained under
simple random sampling. Kish (1965; 1995; 2003) defined the “design effect” as the ratio of the
two variances,
( ) / ( )c c srs srsdeff Var y Var y= [6.6]
where ( )c cVar y is the expected variance based on the actual (complex) survey design, and
( )srs srsVar y is the expected variance under simple random sampling (SRS) for a sample of equal
size. The design-based variance, ( )c cVar y , reflects the effects of stratification and, for the transect
survey, clustering of stations. Kish (1995) and Potthoff et al. (1992) provide a general discussion
on the calculation of design effects and effective sample sizes. We estimated ( )c cVar y for mean
CPUE for the stratified random survey from equation 6.5, while ( )srs srsVar y was estimated by
treating the stations as a simple random sample from the total survey area. This estimate of the
variance that would have been obtained under a simple random sample of the same size is
justified because stations in general were allocated proportionally to the strata sizes.
The “effective sample size” for estimation of the mean CPUE ( y ) using data from the complex
survey design C is defined as * /Cn n deff= . [6.7]
The effective sample size *Cn is the number of stations selected by simple random sampling that
would be required to achieve the same precision obtained with n stations under the actual
complex sampling design. If, for example, the design effect equals two for the estimated mean
CPUE for a transect survey with 30 stations, then a simple random sample of 15 stations (the
effective sample size) would have achieved the same precision.
To further evaluate the efficiency of the survey designs we also compared the lengths of
the cruise track for a given sample size. We quantified the inter-station distance in our
alternative survey designs based on actual and simulated sample selections, using ArcMap (v8.
216
ESRI Corp, Redlands, CA). To assess the impact of alternative designs, we increased the number
of stations in the stratified random survey to levels comparable to the combined survey.
6.2.5. Development of a composite estimator
We used a composite estimator to take a weighted average of the mean CPUE for the
independent stratified random and transect surveys (e.g., Korn and Graubard, 1999; Rao, 2003).
The estimator for the combined mean is given by:
(1 )comb STR Ty y yφ φ= + − [6.8]
with the weight (0 1)φ φ≤ ≤ being chosen to minimize the variance of comby
2 2( ) ( ) (1 ) ( )comb STR TVar y Var y Var yφ φ= + − [6.9]
where ( )STRVar Y and ( )TVar Y are the design-based expected variances of the mean CPUE
estimators for the stratified random and transect surveys, respectively. The optimum weight
( optφ ), expressed as a function of the effective sample sizes for each survey ( * *,STR Tn n ), is
*
* *
11
STRopt
STR T
nn n R
φ = =+ +
,
where * */T STRR n n= (See appendix 6.1 for derivation). Thus, the optimal weight depends only on
the ratio of the effective sample sizes R . We used the sample data to estimate the variances and
the optimal weight optφ in equation 6.9.
217
6.3 RESULTS
Our comparisons of the transect and stratified random survey designs focused on mean
catch per unit efforts for all species combined, and for bay anchovy, white perch, Atlantic
croaker, and weakfish individually (Tables 6.1-6). The relative standard error (RSE), defined as
the ratio of the standard error (SE) to the survey estimate ( y ), was used as a measure of
precision (Jessen, 1978). The comparison of design effects shows that stratified random sampling
with proportional allocation generally is more effective than transect sampling for estimating the
mean CPUE across all species in the mainstem of Chesapeake Bay. The transect survey resulted
in less precise estimates of mean CPUE across species compared to the stratified random survey
in each season, even though the former survey completed 55% to 81% more trawling stations.
The two independent surveys produced comparable estimates of mean catch per unit
effort (CPUE) of all species combined (Figure 6.2) but differed for the individual species that
were considered (compare Tables 6.1-2, and Tables 6.4-5). The combined estimates shows that
bay anchovy was most abundant and widely distributed during all seasons in both years, but with
significantly lower (p<0.5) summer abundance in 2003 as compared to 2002 (Tables 6.3 and
6.6). White perch dominated catches in the upper Bay, but was rare or absent in the mid and
lower bay strata. In the mid and lower bay regions, weakfish and croaker were common. Overall,
diversity of catches was highest at the northern-most and southern-most stations.
The respective design effects suggest that the stratified random survey consistently is
more efficient than the transect survey for estimating mean CPUE in every season for all species
combined (Figure 6.3), and for bay anchovy and white perch specifically (Tables 6.1-2, and 6.4-
5). The relative standard errors of these estimates from the stratified random survey were on par,
or lower than those from the transect survey, although the latter survey occupied from 55 % to
81% more stations. This supports the conclusion that transect sampling generally is less efficient
than stratified random sampling for the system studied here. The effective sample sizes for
estimating overall mean CPUE for the transect surveys were lower than the number of stations
during all seasons, and similar to the number of transects during summer and fall. In contrast, the
design effects suggested that transect sampling was at least as effective as stratified random
sampling for estimating the abundance of weakfish during all three seasons (deff ≤ 1) (Tables
218
6.1-2 and 6.4-5; Figure 6.4). Here the effective sample sizes were larger than the actual number
of stations for some seasons, especially for spring and fall. These design effects may be biased
downwards because of the high frequency of zero catches within some strata. However, an
inspection of the distribution of weakfish and the allocation of stations (Figure 6.1) confirms that
transects effectively captured the spatial variability in abundance. For bay anchovy, in contrast,
stations within transects had similar CPUE and thus did not capture both sources of the overall
variance (Eq. 6.3). White perch were patchily distributed, but some locations appear to have
persistent high density over time. Trawl stations along transects had significantly higher mean
CPUE than the stratified random stations in both years (Tables 6.1-2, and 6.4-5). This suggests
that white perch are highly patchy in distribution and that, while the deff was very poor, the fixed
transects probably covered more of the patchy habitats that are favored by white perch, while the
random stations missed these high-density areas by chance.
The inter-sample distances varied with the number of stations in the survey. The
stratified random survey had the smallest sample size, involving approximately 20 stations in
each survey. The average inter-station distance was 11.49 ±2.77 nm (mean ±SD). The fixed
transect survey involved 28 stations, and had a corresponding longer inter-station distance of
7.86 ± 6.72 nm. When the two surveys were combined, the average inter-station distance was
5.65±0.411 nm. A stratified random survey with approximately 50 stations could be expected to
have an average inter-station distance of 5.48 ± 0.94 nm, slightly shorter than the inter-station
distance for the combined survey design. However, this difference was not significant. When all
survey designs were compared, the average inter-station distance decreased as the number of
stations in the survey increased (Fig. 6.5).
The composite estimator produced precise estimates of mean CPUE for all species
combined, with relative standard errors (RSE) between 13% and 22% (Tables 6.3 and 6.6). The
relative standard errors of the composite estimates of mean CPUE were 18% to 27% lower than
the most precise individual component estimate. The combined estimates for white perch shows
that the use of effective sample sizes to determine weights can yield more accurate results than
the use of traditional weights based on the variances of each survey estimate. In the 2002 fall
surveys, the stratified random trawl stations caught zero white perch (Table 6.1), compared to a
219
mean of 7.9 fish per tow for the transects. Weighting based on the variances would have assigned
all the weight to the stratified random survey, resulting in a poor combined CPUE estimate of
zero. The combined estimate (2.3 fish per tow) based on effective sample sizes is more reliable
than the former.
220
6.4 DISCUSSION AND CONCLUSIONS
Our approach to the use of the design effect is superior to using the relative standard error
(RSE) for evaluating survey efficiency, as it is independent of the sample size n. Hence, it can be
used to determine the sample size required to achieve an adequate level of precision in key
estimates resulting from a given design. The efficiency of sampling along transects depends
inversely on the homogeneity in catch per tow for clusters of stations within transects. Sampling
at stations along transects is an efficient design if stations within each transect are as variable as
those in the general population of trawling stations. In the transect surveys analyzed here,
homogeneity is a common problem and so selecting an additional station from the same transect
generally adds less new information than would a completely independent selection. With the
exception of weakfish and croaker, results from the fixed transect surveys show that the design
effect tends to be higher than unity. Hence, a simple random survey would generally be expected
to produce more precise estimates for a similar survey effort. There was no evidence in our
simulations of any systematic effect of the survey design on the expected inter-sample distance.
Thus, in Chesapeake Bay, a stratified random survey with proportional allocation would be
expected to achieve at least the same number of stations as the combined transect and stratified
random survey for a fixed cost. A possible advantage of stratified random sampling over
systematic sampling in open waters is that the former tends to result in a shorter sailing distance
to occupy all stations (Harbitz and Pennington, 2004), and thus reduces survey cost for a fixed
number of stations.
We treated transects and stations within transects as a two-stage cluster sampling design.
In fact, the actual design was slightly more structured in the sense that transects were chosen so
that spacing would be somewhat even. Such a design does not involve replication, and thus there
is no unbiased design-based estimator of the variance of the mean CPUE. One reason for this
allocation of stations in the TIES program was to ensure maximal spatial coverage over the bay.
This is the same reasoning often used for taking systematic samples.
Effective estimation of means using complex survey designs requires estimators that fully
account for the sampling design. Data collected from clustered samples often result in a
221
reduction in the effective sample size for estimating a statistic such as mean CPUE because of
the tendency of measurements collected within clusters to be more similar than measurements
taken between clusters (e.g., Pennington and Vølstad 1994; Williams 2000). For complex survey
designs, the estimation of the variance of the mean CPUE under the assumption of independence
between all observations will generally underestimate the true variance in the presence of
positive intra-cluster correlation. For complex surveys, we concur with Brogan (1998) and
recommend that specialized sample survey software such as SUDAAN be used for design-based
estimation of population parameters, descriptive analyses and analytical analyses. Alternatively,
a model-based approach such as geostatistical models could be used to determine the sampling
error in the form of the global estimation variance for a wide range of sampling designs (Petitgas
1999; Rivoirard et al. 2000; Christman et al., in preparation).
The effective sample size for the transect survey when estimating CPUE for all species
combined is closer to the number of transects than to the overall number of stations. This again
suggests that CPUE from stations within the same transect tend to be similar, while any two
observations from different transects are different.
The designated strata and proportional allocation of random stations did not substantially
reduce variability in CPUE between stations in the 2002 surveys, but generally resulted in higher
precision than for simple random sampling in the 2003 surveys. Thus, the relative efficiency of
the stratified random and transect sampling vary across years. However, the stratified random
design appears to be robust to variations in the spatial distribution of different species and hence
is recommended over simple random sampling.
For stratified random sampling the theoretical optimum allocation (Neyman allocation) of
a given number of hauls is to sample each stratum in proportion to its standard deviation
multiplied by the stratum size (Cochran, 1977). However, in practice it may be better to allocate
stations proportional to stratum size since the standard deviation only can be approximated from
previous surveys, and the spatial distribution of fish exhibits temporal variability. Stratification
with proportional sampling nearly always leads to gain in precision (Cochran, 1977), with the
largest gains being achieved when the strata means exhibit large variation.
222
This study suggests that a composite estimator with weights based on the effective
sample sizes of individual survey estimates can yield more accurate results than weights based
on variances. For trawl surveys, patchy distributions can result in zero catch even though some
habitats have high densities. If an estimate from a simple random survey with zero catch is
combined with another survey estimate with mean and variance greater than zero, then the
commonly used weighting based on their respective variances will produce a poor combined
estimate of zero. In this case, weights based on the variances would clearly yield an
inappropriate estimate of the mean if both surveys employ simple random surveys with equal
sample sizes. Our method of assigning weights based on their respective effective sample sizes,
in contrast, will yield a more equitable combined estimate because the number of stations will
determine the weight of the simple random survey. In the simplest case of combining two
independent surveys with simple random sampling, our weighs would appropriately yield the
same estimate as the simple mean for the pooled samples. The composite estimator is unbiased
as long as the variances are independent of the means. For trawl surveys, catch per tow tend to
have a skewed distribution because of patchiness of fish populations, and the variance thus is
likely related to the mean (Seber, 1986; Pennington and Vølstad, 1991). This is a serious
problem when combining two surveys if the weight used in equation (1.8) is based on the
variances obtained from the two surveys. Since we are using the effective sample sizes which
depend on both the variance and the sample size for each survey, the differences are greatly
attenuated. For example, had we used the variance weighting for the fall 2003 white perch data,
the weight would have been 0.000079 and hence the transect results would have dominated the
combined estimator. Instead, the use of effective sample size provided a weight of 0.46 and so
both TSTR yy and contribute to comby almost equally.
Our analysis suggest that the variances of the mean CPUE for the transect survey were
relatively high because the stations were clustered, and not because the means were higher than
for the stratified random (except for white perch). For a series of combined surveys, we
recommend that the weight be fixed to eliminate bias caused by the dependence between the
variance and the mean. The maximum reduction of 50 % in the variance of the combined mean
CPUE is achieved when the two surveys have equal effective sample sizes. However, the
precision of the combined estimate is robust to deviation from the optimal ratio ( R ) of effective
223
sample sizes; a change to R from unity to 6 does not significantly increase the variance of the
composite estimator (Rao, 2003, p. 58). In practice, hence, the use of a fixed weight across years
should not appreciably increase the variance of combined estimates.
Acknowledgements
This study (Contribution No XXX of the University of Maryland Center for Environmental
Science) was funded by the National Marine Fisheries Service, through a grant from the
Chesapeake Bay Stock Assessment Committee (Grant NA07FU0534). This research would not
have been possible without the help of the students and technicians at CBL who participated in
the research cruises. In particular, we thank Kiersten Curti, Chris Heyer, and Dave
Loewensteiner who served as chief scientists on CHESFIMS cruises. We appreciate comments
from John Hoenig (VIMS) that helped improve this manuscript.
224
6.5. LITERATURE CITED
Baird, D., and R.E. Ulanwitz. 1989. The seasonal dynamics of the Chesapeake Bay ecosystem.
Ecol. Monogr. 59:329-364.
Brogan, D. 1998. Software for sample survey data, misuse of standard packages. In:
Encyclopedia of Biostatistics, Volume 5 (P. Armitage and T. Colton, Eds.). John Wiley
& Sons, New York, NY pp. 4167-4174.
Carlson, B.D.1998. Software for Statistical Analysis of Sample Survey Data. In: Encyclopedia
of Biostatistics, Volume 5 (P. Armitage and T. Colton, Eds.). John Wiley & Sons, New
York, NY: pp. 4160-4167.
Cochran, W.G. 1977. Sampling Techniques 3rd ed. John Wiley & Sons, New York, NY.
Gavaris, S., and S.J. Smith.1987. Effect of allocation and stratification strategies on precision of
survey abundance estimates for Atlantic cod (Gadus morhua) on the eastern Scotian
Shelf. J. Northw. Atl. Fish. Sci. 7:137-144.
Grosslein, M.D.1969. Groundfish survey program of BFC Woods Hole. Comm. Fish. Rev.
31:22-30.
Gundersen, D.R.1993. Surveys of Fisheries Resources. John Wiley & Sons, New York, NY.
Harbitz, A., M. Aschan, and K. Sunnanå. 1998. Optimal effort allocation in stratified, large area
trawl surveys, with application to shrimp surveys in the Barents Sea. Fish. Res. 37:107-
113.
Harbitz, A., and M. Pennington. 2004. Comparison of shortest sailing distance through random
and regular sampling points. ICES Journal of Marine Science 61:140-147.
225
Hartman, K.J. and S.B. Brandt. 1995. Predatory demand and impact of striped bass, bluefish,
and weakfish in the Chesapeake Bay: applications of bioenergetics models. Can. J. Fish.
Aquat. Sci. 52: 1667-1687.
Houde, E.D., M.J. Fogarty, and T.J. Miller. 1998. Prospects for multispecies management in
Chesapeake Bay: A workshop. STAC Publication 98-002. Chesapeake Research
Consortium, Edgewater, MD.
Jessen, RJ. 1978. Statistical Survey Techniques. John Wiley & Sons, New York, NY.
Jung, S., and E.D. Houde. 2003. Spatial and temporal variabilities of pelagic fish community
structure and distribution in Chesapeake Bay, USA. Fish. Bull. 102:63-77.
Jung, S., and E.D. Houde. 2004. Recruitment and spawning-stock distribution of bay anchovy in
Chesapeake Bay. Fish. Bull. 102:63-77.
Kalton, G. 1979. Ultimate cluster sampling. J. Royal. Stat. Soc. A 142, 210-222.
Kemp, W.M., J. Faganelli, S. Ouscaric, E.M. Smith, and W.R.Boyton. 1999. Pelagic-benthic
coupling of nutrient cycling. In: Malone, T.C., Malej, A., Harding jr., L.W., Smodlaka,
N., Turnber, R.E. (eds.). 1999. Coastal and estuarine Studies: Ecosystems and the land-
sea Margin Drainage Basin to Coastal Sea. American Geophysical Union, Washington
D.C. pp. 295-340
Kish, L. 1965. Survey Sampling. John Wiley & Sons, New York, NY.
1995. Methods for design effects. Journal of Official Statistics 11:55-77.
2003. Selected Papers. Edited by G. Kalton, and S. Heeringa. John Wileys & Sons.
Wileys Series in Survey Methodology.
226
Kingsley, M.C.S., D.M. Carlsson, P. Kanneworff, and M. Pennington. Spatial structure of the
resources of Pandalus borealis and some implications for trawl survey. Fish. Res.
58:171-183.
Komatsu T, T. Sugimoto, K. Ishida, K. Itaya, P. Mishra, and T. Miura 2002. Importance of the
Shatsky Rise Area in the Kuroshio Extension as an offshore nursery ground for Japanese
anchovy (Engraulis japonicus) and sardine (Sardinops melanostictus). Fish. Oceanogr.
11: 354-360 2002
Korn, E.L., and B.I. Graubard. 1999. Analysis of Health Surveys. John Wiley & Sons, New
York, NY.
Lehtonen, R., and E.J. Pahkinen Practical Methods for Design and Analysis of Complex
Surveys. John Wiley & Sons, Chichester.
Luo, J. and S.B. Brandt. 1993. Bay anchovy Anchoa mitchilli production and consumption in
mid-Chesapeake bay based on a bioenergetics model and acoustic measurements of fish
abundance. Mar. Ecol. Prog. Ser. 98:223-236.
Miller, T. J., E.D. Houde, and E.A. Watkins. 1996. Chesapeake Bay Fisheries: Prospects for
multispecies management and sustainability. CRC Publication 154B. Chesapeake
Research Consortium, Edgewater MD.
Pennington, M. 1985. Estimating the relative abundance of fish from a series of trawl surveys.
Biometrics 41:197-202.
1986. Some statistical techniques for estimating abundance indices from trawl surveys.
Fish. Bull. 84(3):519-525.
Pennington, M., and J.H. Vølstad. 1991. Optimum size of sampling unit for estimating the
density of marine populations. Biometrics 47:717-723.
227
1994. The effect of intra-haul correlation and variable density on estimates of population
characteristics from trawl surveys. Biometrics 50:725-732.
Petitgas, P. 1999. A review of linear geostatistics for fisheries survey design and stock
assessment. Quantitative geology and geostatistics 10: 1-12.
Polacheck, T., and J.H. Vølstad. 1993. Analysis of spatial variability of George Bank haddock
(Melanogrammus aeglefinus) from trawl survey data using a linear regression model with
spatial interaction. ICES Journal of Marine Science 50:1-8.
Potthoff, R.F., M.A. Woodbury, and K.G. Manton, K.G. 1992. “Equivalent sample size” and
“equivalent degrees of freedom” refinements for inference using survey weights under
superpopulation models.” Journal of the American Statistical Association 87:383-396.
Rao, J.N.K. 2003. Small Area Estimation. John Wiley & Sons, New York, NY.
Research Triangle Institute (RTI).2001. SUDAAN User's Manual, Release 8.0. Research
Triangle Park, NC: Research Triangle Institute.
Rivoirard, J., E.J. Simmonds, K.F. Foote, P.G. Fernandes, and N. Bez. 2000. Geostatistics for
estimating fish abundance. Blackwell Science Ltd., Oxford, 206 pp.
Sakuma K.M. and S. Ralston. Vertical and horizontal distribution of juvenile Pacific whiting
(Merluccius productus) in relation to hydrography off California. CaCOFI Reports 38:
137-146
Seber, G.A.F. 1986. A review of estimating animal abundance. Biometrics 42:267-292.
Skinner, C.J. 1986. Design effects of two-stage sampling. J.R. Statist. Soc. B 48:89-99.
Skinner, C.J., D. Holt, and T.M.F. Smith. 1989. Analysis of Complex Surveys. John Wiley &
Sons, New York, NY.
228
Snedecor, G.W., and W.G. Cochran. 1980. Statistical Methods. 7th ed. The Iowa State
University Press. Ames, Iowa, U.S.A.
Särndal, C.E., B. Swensson, and J. Wretman. 1992. Model Assisted Survey Sampling.
Springer-Verlag, New York, NY.
Smith,T. The Woods Hole bottom-trawl resource survey: development of fisheries-independent
multispecies monitoring. ICES Marine Science Symposium 215:474-482.
Sukhatme, P.V. and B.V. Sukhatme. 1970. Sampling Theory of Surveys With Applications. 2nd
edition. Iowa State University Press. Ames, Iowa.
Taylor, L.R., 1961. Aggregation, variance and the mean. Nature 189:732-735.
Williams, R.L. 2000. A note on robust variance estimation for cluster-correlated data.
Biometrics 56:645-646.
Wilson MT 2000. Effects of year and region on the abundance and size of age-0 walleye
pollock, Theragra chalcogramma, in the western Gulf of Alaska, 1985-1988 Fish. Bull.
98: 823-834
Wolter, K.M. 1985. Introduction to Variance Estimation. Springer-Verlag. New York, NY.
229
Table 6.1
Mean catch-per-unit-effort (CPUE) and measures of precision and design effects for the
2002 stratified random surveys.
Season Species deff N effn STRy SE RSE
Spring Bay Anchovy 1.0 16 16 84.2 19.4 0.23
Summer Bay Anchovy 1.0 19 19 570.8 143.6 0.25
Fall Bay Anchovy 1.0 20 20 1541.8 445.6 0.29
Spring Croaker 1.0 16 16 4.8 2.0 0.42
Summer Croaker 1.0 19 19 1.6 0.7 0.44
Fall Croaker 1.0 20 20 3.2 1.4 0.44
Spring White Perch 1.0 16 16 0.6 0.5 0.83
Summer White Perch 1.0 19 19 0.7 0.6 0.86
Fall White Perch 1.0 20 20 0 0.0
Spring Weakfish 1.0 16 16 1.6 1.0 0.63
Summer Weakfish 1.0 19 19 2.6 1.2 0.46
Fall Weakfish 1.1 20 18 8.3 2.7 0.33
Spring All Species 1.0 16 16 103.4 18.8 0.18
Summer All Species 1.1 19 17 733.0 161.9 0.22
Fall All Species 0.9 20 22 1,643.6 442.0 0.27
230
Table 6.2
Mean catch-per-unit-effort and measures of precision and design effects for the 2002
transect surveys. The design effects (deff) are obtained from SUDAAN.
Season Species deff N effnTy SE RSE
Spring Bay Anchovy 1.3 29 22 98.0 27.1 0.28
Summer Bay Anchovy 2.2 31 14 482.6 157.4 0.33
Fall Bay Anchovy 3.0 31 10 1452 422.8 0.29
Spring Croaker 1.5 29 19 3.3 1.1 0.33
Summer Croaker 3.9 31 8 2.6 1.6 0.62
Fall Croaker 0.7 31 44 7.8 2.2 0.28
Spring White Perch 3.5 29 8 14.1 9.6 0.68
Summer White Perch 4.2 31 7 4.8 4.1 0.85
Fall White Perch 3.7 31 8 7.9 7.0 0.89
Spring Weakfish 0.6 29 48 1.6 0.7 0.44
Summer Weakfish 1.0 31 31 1.1 0.5 0.45
Fall Weakfish 0.4 31 78 13.4 1.3 0.10
Spring All Species 1.3 29 22 145.2 25.5 0.18
Summer All Species 2.2 31 14 565.0 177.6 0.31
Fall All Species 3.1 31 10 1,557.8 419.6 0.27
231
Table 6.3
Mean catch-per-unit-effort and measures of precision and design effects for the 2002
combined stratified random and transect surveys. The estimates are obtained by using the
optimal weights ( optφ ) in eqs. 1.8 and 1.9. The ratio r is the RSE for the composite
estimate divided by the smaller RSE for the individual survey estimates.
Season Species deff n effn Cy SE RSE optφ r
Spring Bay Anchovy 1.2 45 38 92.2 17.7 0.19 0.42 0.83
Summer Bay Anchovy 1.5 50 33 533.4 106.2 0.20 0.58 0.79
Fall Bay Anchovy 1.7 51 30 1,511.9 335.0 0.22 0.67 0.77
Spring Croaker 1.3 45 35 4.0 1.1 0.27 0.46 0.83
Summer Croaker 1.9 50 27 1.9 0.7 0.36 0.70 0.82
Fall Croaker 0.8 51 64 6.4 1.6 0.25 0.31 0.88
Spring White Perch 1.9 45 24 5.1 3.2 0.63 0.67 0.93
Summer White Perch 1.9 50 26 1.8 1.2 0.66 0.73 0.77
Fall White Perch 1.8 51 28 2.3 2.0 0.89 0.71
Spring Weakfish 0.7 45 64 1.6 0.6 0.36 0.25 0.83
Summer Weakfish 1.0 50 50 1.7 0.6 0.33 0.38 0.73
Fall Weakfish 0.5 51 96 12.4 1.2 0.09 0.19 0.94
Spring All Species 1.2 45 38 127.6 16.7 0.13 0.42 0.73
Summer All Species 1.6 50 31 657.1 119.8 0.18 0.55 0.82
Fall All Species 1.6 51 32 1,616.8 332.0 0.21 0.69 0.76
232
Table 6.4.
Mean catch-per-unit-effort (CPUE) and measures of precision and design effects for the
2003 stratified random surveys.
Season Species deff N effn STRy SE RSE
Spring Bay Anchovy 0.7 20 29 91.8 17.6 0.19
Summer Bay Anchovy 0.6 20 32 570.8 143.6 0.25
Fall Bay Anchovy 1.0 9 9 1,113.6 456.5 0.41
Spring Croaker 0.8 20 27 1.6 0.5 0.31
Summer Croaker 0.9 20 22 4.2 2.0 0.47
Fall Croaker 1.0 9 9 0.1 0.1 1.00
Spring White Perch 0.7 20 27 0.7 0.4 0.60
Summer White Perch 1.1 20 18 1.0 1.0 1.00
Fall White Perch 1.0 9 9 0.2 0.2 1.00
Spring Weakfish 0.9 20 22 0.2 0.1 0.55
Summer Weakfish 1.0 20 20 0.2 0.1 0.53
Fall Weakfish 1.0 9 9 1.3 1.2 0.91
Spring All Species 0.6 20 31 59.3 11.2 0.19
Summer All Species 0.2 20 91 123.0 13.0 0.11
Fall All Species 1.0 9 9 1,122.9 455.4 0.41
233
Table 6.5.
Mean catch-per-unit-effort and measures of precision and design effects for the
2003 transect surveys. The design effects (deff) are obtained from SUDAAN.
Season Species deff N effnTy SE RSE
Spring Bay Anchovy 1.1 31 30 73.7 21.3 0.29
Summer Bay Anchovy 1.9 31 16 64.7 20.3 0.31
Fall Bay Anchovy 2.1 20 9 1,228.9 378.5 0.31
Spring Croaker 1.0 31 32 3.0 1.5 0.51
Summer Croaker 1.0 31 33 1.7 1.0 0.57
Fall Croaker 1.7 20 12 0.2 0.2 0.94
Spring White Perch 2.7 31 11 17.3 13.4 0.78
Summer White Perch 6.3 31 5 51.7 6.3 0.12
Fall White Perch 1.9 20 11 12.3 10.1 0.82
Spring Weakfish 0.6 31 53 0.3 0.2 0.61
Summer Weakfish 1.0 31 32 0.8 0.5 0.60
Fall Weakfish 0.8 20 24 1.2 0.6 0.52
Spring All Species 1.1 31 29 103.2 22.2 0.22
Summer All Species 4.8 31 6 223.2 108.9 0.49
Fall All Species 2.4 20 8 1,412.7 529.8 0.38
234
Table 6.6.
Mean catch-per-unit-effort and measures of precision and design effects for the 2003
combined stratified random and transect surveys. The estimates are obtained by using the
optimal weights ( optφ ) in eqs. 1.8 and 1.9. The ratio r is the RSE for the composite
estimate divided by the smaller RSE for the individual survey estimates.
Season Species deff n effn Cy SE RSE optφ r
Spring Bay Anchovy 0.9 51 59 82.7 13.8 0.17 0.50 0.9
Summer Bay Anchovy 1.1 51 48 402.6 96.1 0.24 0.67 0.9
Fall Bay Anchovy 1.6 29 18 1172.8 295.2 0.25 0.49 0.8
Spring Croaker 0.9 51 59 2.4 0.9 0.37 0.45 1.2
Summer Croaker 0.9 51 55 2.7 1.0 0.36 0.41 0.8
Fall Croaker 1.4 29 21 0.1 0.1 0.72 0.43 0.8
Spring White Perch 1.3 51 39 5.6 4.0 0.71 0.71 1.2
Summer White Perch 2.2 51 23 11.9 1.6 0.13 0.78 1.1
Fall White Perch 1.5 29 20 6.8 5.5 0.81 0.46 1.0
Spring Weakfish 0.7 51 75 0.3 0.1 0.49 0.29 0.9
Summer Weakfish 1.0 51 51 0.5 0.3 0.54 0.38 1.0
Fall Weakfish 0.9 29 33 1.2 0.6 0.46 0.27 0.9
Spring All Species 0.9 51 60 80.3 12.1 0.15 0.52 0.8
Summer All Species 0.5 51 97 129.6 14.1 0.11 0.93 1.0
Fall All Species 1.7 29 17 1,263.1 347.8 0.28 0.52 0.7
235
Figure 6.1. Stratification of the Chesapeake Bay mainstem (Lower, Mid, and Upper Bay), and an example of station allocation for the
transect survey (╬) and stratified random survey (o) and the distribution of relative abundance of (a) Bay anchovy and (b) weakfish
during fall 2002.
236
Figure 6.2. Seasonal design-based estimates of mean catch per unit effort (cpue) across all
species, and the combined estimates across surveys based on the composite estimator. Error bars
represent ± SE.
2002 surveys
0
500
1000
1500
2000
2500
Spring Summer Fall
Mea
n C
PUE
STR Transect Combined
2003 surveys
0
500
1000
1500
2000
2500
Spring Summer Fall
Mea
n C
PUE
STR Transect Combined
237
Figure 6.3. Design effects for estimating seasonal mean catch per unit effort (CPUE) across all species by survey. The combined estimate is based on the composite estimator
2002 surveys
0.01.02.03.04.05.0
Spring Summer Fall
DEF
F
STR Transect Combined
2003 surveys
0.01.02.03.04.05.0
Spring Summer Fall
DEF
F
STR Transect Combined
238
Figure 6.4. Design effects for estimating seasonal mean catch per unit effort (CPUE) of weakfish
by survey. The combined estimate is based on the composite estimator.
2002 surveys
0.0
0.5
1.0
1.5
Spring Summer Fall
DEF
F
STR Transect Combined
2003 surveys
0.0
0.5
1.0
1.5
Spring Summer Fall
DEF
F
STR Transect Combined
239
All datay = -5.50Ln(x) + 26.63
R2 = 0.98
0
2
4
6
8
10
12
14
16
18
0 10 20 30 40 50 60
Number of stations
Inte
rsam
ple
dist
ance
(NM
)
TransectRandomCombinedAll random
Figure 6.5. Estimated inter-station distances for the fixed transect and stratified random surveys designs, and for combined surveys for actual and simulated sample selections.
240
Appendix 6.1 The optimum weight, obtained by minimizing equation Error! Reference source not found.
with respect to φ is a function of the expected variances of the mean CPUE under each design,
( )( ) ( )
Topt
STR T
Var YVar Y Var Y
φ =+
when the two surveys are independent (Rao, 2003). By definition, the expected variance of the
estimated mean CPUE from each survey can be expressed by dividing the population variance of
CPUE under simple random sampling, 2SRSσ , with the effective sample size,
2
*( ) SRST
T
Var Ynσ
= (A.1)
and
2
*( ) SRSSTR
STR
Var Ynσ
= (A.2)
where *Tn and *
STRn are the effective sample sizes for the transect and stratified random surveys,
respectively. Replacing ( )TVar Y and ( )STRVar Y in optφ with their equivalents from (A.1) and
(A.2) yield the desired result.