NOAA COASTAL OCEAN PROGRAM
Decision Analysis Series No. 16

Topic 2 Report for the Integrated Assessment on Hypoxia in the Gulf of Mexico

Robert J. Diaz, Virginia Institute of Marine Science
Andrew Solow, Woods Hole Oceanographic Institution
May 1999

U.S. DEPARTMENT OF COMMERCE
National Oceanic and Atmospheric Administration
National Ocean Service
Coastal Ocean Program

We would like to acknowledge the assistance of many managers and scientists who have contributed in-sight and/or data on the ecological and economic effects of hypoxia, particularly the Task 2 Ecological Effects and Economic Effects Working Group members.

Ecological Effects Working Group Members
Denise Breitburg, Philadelphia Academy of Science
Churchill Grimes, National Marine Fisheries Service, Southeast Fisheries Science Center
Jim Hanifen, Louisiana Fish and Wildlife Department
Don Harper, Texas A & M University
Nancy Marcus, Florida State University
Nancy Rabalais, Louisiana Universities Marine Consortium
Gilbert Rowe, Texas A & M University
Gene Turner, Louisiana State University
Roger Zimmerman, National Marine Fisheries Service

Economic Effects Working Group Members
Otto Doering, Agricultural Economics Department, Purdue University
Joseph Glauber, Economic Analysis Section, U.S. Department of Agriculture
Walter Keithley, Coastal Fisheries Institute, Louisiana State University
Jim Nance, National Marine Fisheries Service, Galveston
Kenneth Roberts, Louisiana Cooperative Extension Service, Louisiana State University
Roger Zimmerman, National Marine Fisheries Service

Other Contributors
Fred Bryan, U.S. Geological Survey
Ed Chesney, Louisiana Universities Marine Consortium
Randal Cutter, Virginia Institute of Marine Science
John Downing, Iowa State University
Drew Fillingame, Gulf of Mexico Program
Rutger Rosenberg, Gothenberg University, Sweden
Mark Schexnayder, Louisiana Fish and Wildlife Department

Nutrient overenrichment from anthropogenic sources is one of the major stresses on coastal ecosystems. Generally, excess nutrients increase algal production and the availability of organic carbon within an ecosystem--a process known as eutrophication. Scientific investigations in the northern Gulf of Mexico have documented a large area of the Louisiana continental shelf with seasonally depleted oxygen levels (< 2 mg/l). Most aquatic species cannot survive at such low oxygen levels. The oxygen depletion, referred to as hypoxia, forms in the middle of the most important commercial and recreational fisheries in the conterminous United States and could threaten the economy of this region of the Gulf.

As part of a process of considering options for responding to hypoxia, the U.S. Environmental Protection Agency (EPA) formed the Mississippi River/Gulf of Mexico Watershed Nutrient Task Force during the fall of 1997, and asked the White House Office of Science and Technology Policy to conduct a scientific assessment of the causes and consequences of Gulf hypoxia through its Committee on Environment and Natural Resources (CENR). A Hypoxia Working Group was assembled from federal agency representatives, and the group developed a plan to conduct the scientific assessment.

The National Oceanic and Atmospheric Administration (NOAA) has led the CENR assessment, although oversight is spread among several federal agencies. The objectives are to provide scientific information that can be used to evaluate management strategies, and to identify gaps in our understanding of this complex problem. While the assessment focuses on hypoxia in the Gulf of Mexico, it also addresses the effects of changes in nutrient concentrations and loads and nutrient ratios on water quality conditions within the Mississippi-Atchafalaya River system.

As a foundation for the assessment, six interrelated reports were developed by six teams with experts from within and outside of government. Each of the reports underwent extensive peer review by independent experts. To facilitate this comprehensive review, an editorial board was selected based on nominations from the task force and other organizations. Board members were Dr. Donald Boesch, University of Maryland; Dr. Jerry Hatfield, U.S. Department of Agriculture; Dr. George Hallberg, Cadmus Group; Dr. Fred Bryan, Louisiana State University; Dr. Sandra Batie, Michigan State University; and Dr. Rodney Foil, Mississippi State University. The six reports are entitled:

Topic 1: Characterization of Hypoxia. Describes the seasonal, interannual, and long-term variations of hypoxia in the northern Gulf of Mexico and its relationship to nutrient loadings. Lead: Nancy N. Rabalais, Louisiana Universities Marine Consortium.

Topic 2: Ecological and Economic Consequences of Hypoxia. Evaluates the ecological and economic consequences of nutrient loading, including impacts on the regional economy. Co-leads: Robert J. Diaz, Virginia Institute of Marine Science, and Andrew Solow, Woods Hole Oceanographic Institution, Center for Marine Policy.

Topic 3: Flux and Sources of Nutrients in the Mississippi-Atchafalaya River Basin. Identifies the sources of nutrients within the Mississippi-Atchafalaya system and Gulf of Mexico. Lead: Donald A. Goolsby, U.S. Geological Survey.

Topic 4: Effects of Reducing Nutrient Loads to Surface Waters Within the Mississippi River Basin and Gulf of Mexico. Estimates the effects of nutrient-source reductions on water quality. Co-leads: Patrick L. Brezonik, University of Minnesota, and Victor J. Bierman, Jr., Limno-Tech, Inc.

Topic 5: Reducing Nutrient Loads, Especially Nitrate–Nitrogen, to Surface Water, Ground Water, and the Gulf of Mexico. Identifies and evaluates methods for reducing nutrient loads. Lead: William J. Mitsch, Ohio State University.

Topic 6: Evaluation of the Economic Costs and Benefits of Methods for Reducing Nutrient Loads to the Gulf of Mexico. Evaluates the social and economic costs and benefits of the methods identified in Topic 5 for reducing nutrient loads. Lead: Otto C. Doering, Purdue University.

These six individual reports provide a foundation for the final integrated assessment, which the task force will use to evaluate alternative solutions and management strategies called for in Public Law 105-383.

As a contribution to the Decision Analysis Series, this report provides a critical synthesis of the best available scientific information regarding the ecological and economic consequences of hypoxia in the Gulf of Mexico. As with all of its products, the Coastal Ocean Program is very interested in ascertaining the utility of the Decision Analysis Series, particularly with regard to its application to the management decision process. Therefore, we encourage you to write, fax, call, or e-mail us with your comments. Our address and telephone and fax numbers are on the inside front cover of this report.

David Johnson, Director
Coastal Ocean Program

Donald Scavia, Chief Scientist
National Ocean Service

In this report we have attempted to evaluate the ecological and economic consequences of hypoxia in the northern Gulf of Mexico. Although our initial approach was to rely on published accounts, we quickly realized that the body of published literature dealing with hypoxia was limited, and we would have to conduct our own exploratory analysis of existing Gulf data, or rely on published accounts from other systems to infer possible or potential effects of hypoxia.

For the economic analysis, we developed a conceptual model of how hypoxia-related impacts could affect fisheries. Our model included both supply and demand components. The supply model had two components: (1) a physical production function for fish or shrimp, and (2) the cost of fishing. If hypoxia causes the cost of a unit of fishing effort to change, then this will result in a shift in supply. The demand model considered how hypoxia might affect the quality of landed fish or shrimp. In particular, the market value per pound is lower for small shrimp than for large shrimp.

Given the limitations of the ecological assessment, the shallow continental shelf area affected by hypoxia does show signs of hypoxia-related stress. While current ecological conditions are a response to a variety of stressors, the effects of hypoxia are most obvious in the benthos that experience mortality, elimination of larger long-lived species, and a shifting of productivity to non-hypoxic periods (energy pulsing). What is not known is whether hypoxia leads to higher productivity during productive periods, or simply to a reduction of productivity during oxygen-stressed periods.

The economic assessment based on fisheries data, however, failed to detect effects attributable to hypoxia. Overall, fisheries landings statistics for at least the last few decades have been relatively constant. The failure to identify clear hypoxic effects in the fisheries statistics does not necessarily mean that they are absent. There are several possibilities: (1) hypoxic effects are small relative to the overall variability in the data sets evaluated; (2) the data and the power of the analyses are not adequate; and (3) currently there are no hypoxic effects on fisheries.

Lack of identified hypoxic effects in available fisheries data does not imply that effects would not occur should conditions worsen. Experience with other hypoxic zones around the globe shows that both ecological and fisheries effects become progressively more severe as hypoxia increases. Several large systems around the globe have suffered serious ecological and economic consequences from seasonal summer-time hypoxia; most notable are the Kattegat and Black Sea. The consequences range from localized loss of catch and recruitment failure to complete system-wide loss of fishery species. If experiences in other systems are applicable to the Gulf of Mexico, then in the face of worsening hypoxic conditions, at some point fisheries and other species will decline, perhaps precipitously.

Catch per unit effort for brown shrimp, while variable, has trended down since the late 1970s. While fish and shrimp may avoid hypoxia, the effects of avoidance on metabolism and predator avoidance are unknown, including the time and space scales over which avoidance is possible. In fact, a major difference between the northern Gulf of Mexico hypoxic zone and other such zones around the world is that for the last few decades, the Gulf ecosystem has managed to maintain energy flow to productive fisheries (crabs and shrimps) that depend on the bottom.

Any effect of hypoxia in the northern Gulf of Mexico is intertwined with other environmental stressors. To understand specifically how hypoxia affects populations in the Gulf, we first need to determine the contribution of all natural and anthropogenic sources of mortality and growth to population dynamics. We also need to determine what functional aspects of the ecosystem are specifically affected by hypoxia. A comprehensive research plan is needed as a focus for efforts directed at assessing both the ecological and the economic effects of hypoxia in the Gulf. This plan must include elements for both directing new research and synthesizing existing data. Our efforts to identify hypoxia-related problems were severely hampered by lack of published data. A total ecosystem approach to the problem will be the only successful approach. Consideration of how hypoxia and other stressors interact with all aspects of the Gulf ecosystem is essential for planning restoration efforts.

An important goal of the CENR hypoxia science assessment was to document the state of knowledge of the extent, characteristics, causes, and effects (both ecological and economic) of hypoxia in the northern Gulf of Mexico. This report specifically deals with the ecological and economic effects of annual hypoxia, which is apparently procreated by excess nutrient loading delivered to the Gulf via the Mississippi River Basin. It also examines the impacts of hypoxia on Gulf of Mexico fisheries and the regional and national economies, and articulates both the ecological and the economic consequences and, to the extent appropriate, their interaction.

Other reports in the assessment series compiled existing information on nutrient sources, identified alternatives for reducing nutrient inputs, and examined the costs and benefits associated with reducing the nutrient loads. The ecological and economic assessment was built from the work of many Gulf of Mexico researchers.

The direct connection between land and sea is best exemplified by the relationship between estuarine and coastal fisheries production and land-derived nutrients. The most productive fisheries zones around the world are always associated with significant inputs of either land- (runoff) or deep oceanic- (upwelling) derived nutrients. The basic nutrients carried by land runoff and oceanic upwelling are essential elements that fuel primary production passed through marine food webs to species of economic importance. This basic scenario has been played out for eons around the world, including the northern Gulf of Mexico (Rabalais et al. 1996). The importance of the linkage between land and sea is clearly seen in the Gulf of Mexico, which in 1996 accounted for 16% of the total commercial landings in the United States, with over half of this total harvested from waters surrounding the Mississippi River Delta (Holliday and O'Bannon 1997).

Problems begin when the nutrients entering the system exceed the capacity of the food chain to assimilate them. At first, increased nutrients lead to increased fisheries production. But as production of organic matter production increases, changes occur in the food web that leads to different endpoints. These changes are very predictable and have followed the same path in many marine ecosystems (Figure 1.1). The relationship between nutrient loads delivered to the northern Gulf of Mexico and basic ecological responses (e.g., increased primary productivity in the water column, increased flux of organic matter to the bottom, bottom-water hypoxia, altered energy flow, and stressed fisheries) are typical of other systems’ responses around the world (see

reviews by Brongersma-Sanders 1957; Caddy 1993; Diaz and Rosenberg 1995). Basically, the northern Gulf of Mexico hypoxic zone, popularly know as the "Dead Zone," is a secondary manifestation of the larger problem of excess nutrients, which leads to increased production of organic matter and, in the case of the Gulf of Mexico, hypoxia. (For details, see the Topic 1 report, Characterization of Hypoxia.) The increased loading of organic matter is referred to as eutrophication (Nixon 1995).

Many marine species in the northern Gulf of Mexico are targets of commercial and recreational fishing and, therefore, have direct economic value. These species deserve special attention in an assessment of the ecological and economic consequences of hypoxia. Moreover, the information on which such an assessment could be based is generally more detailed for commercially exploited species.

Despite considerable concern and speculation, to date there appears to have been no rigorous assessment of ecological or economic effects. Accordingly, the decision was made to focus on identifying economic effects in existing fisheries data for selected commercial stocks that could reasonably be attributed to hypoxia. For identifying ecological effects, the focus was on fisheries-independent data and other environmental studies conducted primarily for oil and gas development. Although this focus is certainly narrow, it was necessitated by a combination of the requirement for some level of scientific rigor and limitations on project resources.

The economic assessment was originally envisioned as consisting of two steps: identifying effects in the fisheries data that could be attributed to hypoxia, and attaching an economic value to those effects. There appears to be only one quantitative assessment of the effects of hypoxia on a commercial fishery in the Gulf of Mexico (Zimmerman et al. 1996). Briefly, the main findings of that study were that (1) there is a weak relationship between shrimp catch and a measure of the frequency of hypoxic conditions, and (2) catch per unit effort is unrelated to the frequency of hypoxia.

In the absence of an existing body of analysis, the study described here involved what should be viewed as an exploratory or preliminary analysis of existing data to identify possible hypoxic effects. As described in more detail below, this analysis uncovered only weak evidence of such effects. In interpreting these results, it is important to emphasize three points:

• First, the failure to identify hypoxic effects does not necessarily mean that they are absent. Rather, it only suggests that their magnitude is small in relation to the overall variability in the data. In technical terms, given the data and sample size, the statistical power of this analysis may be low. There are many potential sources of variability in the fisheries data other than hypoxia. These include measurement errors and other problems with the data; variation due to weather and the compounding effects of coastal habitat modifications; and economic and other factors affecting fishing behavior, including variation through time and between localities in fisheries regulations.
• Second, in connection with the first point, the ability to account empirically for different sources of variability in the fisheries data is limited by information about these sources. Notably, consistent quantitative data on the severity of hypoxic conditions were limited to estimates of the area experiencing oxygen concentrations lower than 2 mg/l (AREA) over the period 1985–95, with little to no data for 1989 (see the Topic 1 report, Characterization of Hypoxia). In light of the high variability of fisheries data, this time series is simply too short to establish a credible relationship between the severity of hypoxia and variables relating to fisheries. In an attempt to overcome this problem, the study group used a proxy for hypoxia to extend the analysis over a longer period. However, this proxy, described below, is certainly rough.
• Third, even if the failure to identify clear hypoxic effects in the historical data is taken as evidence of their absence, this does not imply that effects would not occur should conditions worsen. As noted elsewhere in this report, experience with other hypoxic areas shows that ecological effects and fisheries effects become progressively greater as hypoxia worsens (Caddy 1993).

The analysis focused primarily on the two main shrimp species--brown shrimp (Penaeus aztecus) and white shrimp (Penaeus setiferus)--for three reasons:

• These species are commercially important (Table 1.1).
• As benthic (i.e., bottom-dwelling) species, the potential for experiencing effects due to hypoxia is relatively high.
• The data for these fisheries are relatively good, particularly in terms of geographical resolution. Additional analysis was performed for two other species--Gulf menhaden (Brevoortia patronus) and red snapper (Lutjanus campehanus)--to address public concerns that have been raised about the effects of hypoxia on them.

Source: National Marine Fisheries Service.

Evaluation of ecological effects from the northern Gulf of Mexico hypoxic zone was based primarily on data collected by various investigators and federal programs and secondarily on meetings convened to assess hypoxia in the Gulf (Table 2.1). Most of the data appeared as agency reports or gray literature of limited circulation. In almost all cases these studies were commissioned for reasons other than evaluating the effects of hypoxia, usually for compiling fishery statistics or for information related to oil and gas activities. This particular aspect of the available data made it difficult to single out either ecological or economic effects associated with excess nutrient loading and hypoxia in the Gulf of Mexico.

Excess nutrient loading leading to eutrophication and subsequent hypoxia and anoxia are well studied phenomena in other places around the world (see Rosenberg 1985; Colombo et al. 1992; Nixon 1995; Howarth et al. 1996; Vitousek et al. 1997) and provide information pertinent to the northern Gulf of Mexico. In particular, much of the information dealing with estuarine and marine ecosystem-level effects of these phenomena on fisheries and benthos was recently summarized by Caddy (1993) and Diaz and Rosenberg (1995).

The SEAMAP (Southeast Area Monitoring and Assessment Program) Environmental and Biological Atlas of the Gulf of Mexico database of the Gulf States Marine Fisheries Commission contains large amounts of data on trawlable species that are relevant to the current ecological and economic assessment. SEAMAP is a state, federal, and university program for collection, management, and dissemination of fisheries-independent data in the United States. The Gulf program data collection started in 1982, with many stations located in the hypoxic zone. Details on the sampling strategy used in collecting these data are contained in a series of summary atlases published each year (see Donaldson et al. 1997 as an example). The original data are available at the program's FTP site (ftp://conch.ssc.nmfs.gov/pub/seamap). Our ecological assessment was based on data from 1987 and 1988 (a time when hypoxia was less extensive) and from 1993 and 1994 (a time when hypoxia was more extensive).

To focus the analysis on the central issue--namely, the economic impact on fisheries of hypoxia--it was necessary to adopt a conceptual model of how these impacts might arise. Without such a model, the appropriate data analysis would be unclear. The conceptual model adopted here is as follows.

Potential economic effects of hypoxia arise through shifts in either the supply or the demand for fish. The supply model has two components. The first component is the physical production function for fish. A simple model of the production function is:

CATCH = q EFFORT STOCK (1)

where q is the catchability coefficient (e.g., Clark 1985). Under this model, catch per unit effort (CPUE) is proportional to stock size:

CPUE = CATCH/EFFORT = q STOCK (2)

As described below, the analysis of shrimp data was performed for three areas for which estimates of STOCK are not available, but for which estimates of CATCH and EFFORT--and, therefore, CPUE--are. Under the assumption that q does not depend on the level of hypoxia, an effect of hypoxia on stock can be detected by establishing an empirical relationship between CPUE and a hypoxia measure.

It is important to emphasize that this approach to detecting a hypoxic effect on STOCK does not require that EFFORT be independent of hypoxia. This point is sometimes missed. For example, Zimmermann et al. (1996) found no empirical relationship between CPUE and a measure of the frequency of hypoxia in highly spatially resolved data. As shrimp are known to avoid hypoxic waters, they suggested that this reflected a combination of reduced STOCK and reduced EFFORT in areas of high hypoxia. However, under the model (1), this argument cannot be correct. Specifically, if a reduction in EFFORT leads to a proportional reduction in CATCH, so that CPUE remains constant, then under this model STOCK cannot have changed. The approach outlined above does assume that catchability q is independent of hypoxia. If a decline in STOCK due to hypoxia is, in fact, offset by an increase in q, then CPUE may remain unchanged, thereby obscuring the STOCK effect. There is anecdotal evidence that shrimp leaving the hypoxic zone congregate in high densities at the border of this zone. By increasing catchability, this may obscure an effect of hypoxia on STOCK. This possibility deserves further attention.

The second component of the supply model is the cost of fishing. If hypoxia causes the cost of a unit of fishing effort to change, then this will result in a shift in supply. In particular, concern has been raised about the possibility that the hypoxia off the Louisiana coast has caused an offshore shift in the distribution of shrimp (and possibly other commercial species). Such a distributional shift could affect the cost of fishing by changing the distance that fishing vessels must travel. Detecting such a distributional shift requires relatively detailed information about the location of fishing activity. Some information along these lines is available for the shrimp fishery. As described below, an attempt was made to detect a distributional shift in that fishery that could be attributed to hypoxia. Incidentally, an increase in catchability q due to hypoxia could reduce the cost of fishing.

Finally, hypoxia could cause the demand for fish to shift if it affects the quality of the landed fish. In particular, the market value per pound is lower for small shrimp than for large shrimp. As a result, if the size of landed shrimp declines as a result of hypoxia (e.g., through an effect on the growth of shrimp or on the location of shrimp harvest), then this would result in a loss in value, even if the total weight of landings remains unchanged.

The analysis in this part of the study involved identifying empirical relationships between fisheries variables and a measure of hypoxia. One problem with basing the identification of such relationships on the time series of AREA (Figure 2.1) is that this time series is short. Moreover, the correlation between the time series of AREA with time itself is nearly 0.8, so that any other time series exhibiting a secular change over this relatively short period will be correlated with AREA. To overcome this problem, a proxy for hypoxia covering the period 1960-96 was constructed.

Briefly, hypoxia results from the decay of phytoplankton and other kinds of organic material in the bottom waters by oxygen-using bacteria. The amount of phytoplankton in the Gulf depends in part on the amount of nutrients supplied to the Gulf by the Mississippi and Atchafalaya Rivers. The amount of nutrients, in turn, depends on the product of the nutrient concentration in these rivers and their discharge into the Gulf (see the Topic 3 report, Flux and Sources of Nutrients in the Mississippi-Atchafalaya River Basin). The hypoxia index (INDEX) for the period 1960-96 is given by:

INDEX(t) = d(t) exp(0.02 (t-1960)) (3)

where INDEX(t) is the hypoxia index in year t, d(t) is the combined Mississippi-Atchafalaya spring discharge, and the exponential term assumes a 2% growth rate of nutrient concentration over this period (Turner and Rabalais 1991).

The time series of INDEX (Figure 2.1) is in rough qualitative agreement with the time series of AREA (Figure 2.2) over the period 1985-95, with a correlation of 0.60. The correlation between INDEX and time is around 0.7, due mostly to the assumed exponential growth in nutrient concentration. As with AREA, there is a danger of detecting a spurious relationship with a variable that also exhibits a secular change over this period. However, in comparison to AREA, the danger is reduced somewhat by the length of the INDEX time series.

In addition to the problem of secular variation, in interpreting analyses based on this index, it is important to stress that INDEX--and true hypoxic conditions--are expected to be related to climatic conditions, such as temperature and precipitation, and to primary production in the Gulf. These factors can, by themselves, affect coastal fisheries. For example, Griffin et al. (1976) developed a yield model for brown shrimp in which Mississippi River discharge served as a proxy for environmental conditions in estuarine nursery grounds. As a result, the establishment of an empirical relationship between a fisheries variable and INDEX--or even between a fisheries variable and true hypoxic conditions, were they known--does not necessarily imply the existence of an effect due to hypoxia. Instead, the effect may be due to a climatic or other covariates of INDEX. As described below, an attempt was made to avoid this problem in the analysis of the shrimp data by using fisheries data from different parts of the northern Gulf.

A series of meetings and workshops dealing specifically with hypoxia-related issues provided insight and data regarding recent changes to the global picture and the northern Gulf of Mexico. The participants in these meetings were very helpful, and many provided relevant insight into the ecological and economic effects of hypoxia (Table 2.1).

As part of the CENR assessment, we formed two hypoxic effects working teams that specifically provided insight and data on the Gulf of Mexico ecological and economic dynamics. We met and communicated with our teams through the process of developing this assessment. All members are listed in the Acknowledgments section in the beginning of this report.

Excess nutrient loading leads to eutrophication of coastal seas, a widespread problem around the globe in general (Nixon 1980, 1995; Howarth et al. 1996). The primary factor driving this marine coastal eutrophication is an imbalance in the nitrogen cycle that can be directly linked to increased population whether through urbanization in coastal river drainages or expanded agricultural activities. In many areas hypoxia follows from eutrophication, which results from the underlying nutrient problem. An examination of the distribution of hypoxic zones around the world showed that they were closely associated with developed watersheds or coastal population centers that deliver large quantities of nutrients to coastal seas, with nitrogen being the most important nutrient. (See the Topic 1 report, Characterization of Hypoxia, and the Topic 3 report, Flux and Sources of Nutrients in the Mississippi-Atchafalaya River Basin.) Although agriculture and, to a much lesser degree, industry are regarded as the key causes, population growth and rising living standards are in fact driving the need for industry and agriculture to produce.

The scenario linking nutrient additions to the formation of hypoxia and impacts to fisheries via eutrophication can be summarized as follows. Excess nutrients lead to increased primary production, which is new organic matter added to the ecosystem. Because shallow estuarine and coastal systems tend to be tightly coupled (benthic-pelagic coupling), much of this organic matter reaches the bottom. This increased primary productivity may also lead to increased fisheries production (Caddy 1993). At some point, however, the ecosystem's ability to process organic matter in a balanced manner is exceeded, and if physical dynamics permit stratification, water quality is degraded and hypoxic conditions develop. Initially, the increased fisheries production may offset any detrimental effects of hypoxia. But as eutrophication increases and hypoxia expands in duration and area, the fisheries production base is affected and declines. This graded reaction to the combined problems of excess nutrients and hypoxia has been documented for many systems around the globe.

The linkage of the Mississippi River and northern Gulf of Mexico continental shelf has led to a highly productive system that yields significant landings of fish and shellfish to the region. Annual landings have exceeded 1 billion pounds since 1969 (U.S. Department of Commerce Fishery Statistics; see Holliday and O'Bannon 1997 as an example). In the Caddy (1993) model that relates fishery yield to nutrients supplied, the northern Gulf of Mexico is currently somewhere in the eutrophic category (Figure 1.1). To a point, nutrient enrichment may increase fishery yields; but beyond a certain level, it is negative in effect (Caddy 1993).

The dissolved oxygen conditions of many major coastal ecosystems around the world have been adversely affected through the process of eutrophication. Most of these coastal systems recorded a steady (monotonic) decline in dissolved oxygen through time, in most cases starting from initial oxygen measurements, usually in the 1950s (Rosenberg 1990). For systems that have historical data from the turn of the century, the declines in oxygen levels started in the 1950s and 1960s. However, for the Baltic Sea, declining dissolved oxygen levels were noted as early as the 1930s (Fonselius 1969).

From a historical perspective it is clear that many of the systems that are currently hypoxic were not when they were first studied. The best examples of systems with long-term data come from Europe, where it can be seen that benthic hypoxia was not reported prior to the 1950s in the Baltic Sea (Fonselius 1969), 1960s in the northern Adriatic (Justic´ 1987), 1970s in the Kategatt (Baden et al. 1990), and 1980s on the Northwest continental shelf of the Black Sea (Mee 1992). Except in areas of natural upwelling, coastal hypoxia is not a natural condition.

By the 1970s ecosystems around the world were becoming saturated with organic matter, and many of them manifested hypoxia for the first time. Once it had occurred, hypoxia quickly became an annual event and a prominent feature that controlled ecosystem energy flow (Diaz and Rosenberg 1995). From the 1980s to the present, the distribution of hypoxia around the world has not appreciably changed. Only in systems that have experienced intensive regulation of nutrient inputs have oxygen conditions improved. There are many examples of small-scale hypoxia reversals associated with improvements in treatment of sewage and pulp mill effluents (Rosenberg 1972, 1976).

In the United States, the improved water quality in Lake Erie is the best example and evidence that large ecosystems do respond positively to nutrient regulation, but that the time interval for achieving noticeable improvements many be long (Boyce et al. 1987; Charlton et al. 1993). The extent of hypoxia in Lake Erie was similar between 1970 and 1990, despite the reduced nutrient loads. The delayed improvement in oxygen conditions may be consistent with mechanisms and processes that contribute to the ecosystem's resilience (Charlton et al. 1993). Improvements in oxygen may not be noticed for decades and could be complicated by climatic changes (DiToro and Blumberg 1990). The Lake Erie example points to the need to have knowledge of a system's response to the complex problems associated with eutrophication before conclusions can be drawn as to the effectiveness of management actions.

Because of their geomorphology and circulation patterns, some marine systems have a greater tendency for hypoxic conditions to develop. The basic features of a system that make it prone to hypoxia are low physical energy (tidal, currents, or wind) and large freshwater input. These features combine to form stratified or stable water masses near the bottom that become hypoxic when they are isolated from reoxygenation with surface waters. The first investigations of bottom water quality in Chesapeake Bay in the 1930s reported hypoxia in deep channel areas of the mainstem (Newcombe and Horne 1938). In Mobile Bay there are accounts of "Jubilees"--likely hypoxia-related then as they are now (Schroeder and Wiseman 1988)--from the 1860s (J. Pennock, personal communication). Better mixed or flushed systems do not have a tendency toward hypoxia. The Baltic Sea and the Kattegat showed no historical tendency toward hypoxia (Pearson et al. 1985; Elmgren 1989). It was not until the 1950s and 1970s, respectively, that oxygen was found to be a problem, even though there were oxygen readings in both systems that go back to the turn of the century (Rosenberg unpublished data). Similarly, the northern Adriatic, with oxygen data from the 1910s, did not exhibit hypoxia until the 1960s (Justic´ 1987).

A historical picture of oxygen conditions for the northern Gulf of Mexico, derived from reading the geochronology of sediment cores, indicates that hypoxia was most likely not a prominent feature of the shallow continental shelf prior to the 1950s (Rabalais et al. 1996 and Sen Gupta et al. 1996 in the Topic 1 report, Characterization of Hypoxia). A longer (2,000-year) geochronology done in Chesapeake Bay pointed to early European settlement of the Bay's watershed as a key feature that led to changes in most paleoenvironmental indicators and set the stage for current oxygen problems as much as 300 years ago (Cooper and Brush 1991).

Annual summertime hypoxia was the most common form of low dissolved oxygen event recorded around the globe (30 of 47 known anthropogenic hypoxic zones, Diaz and Rosenberg 1995). So in this respect, the northern Gulf of Mexico is not unique. Interestingly, the degree of obvious ecological and economic effects related to the hypoxia varies from system to system. The most serious ecological and economic effects of the combined problems of eutrophication and hypoxia are seen in the Black Sea and Baltic Sea, where demersal trawl fisheries have been either eliminated or severely stressed (Mee 1992; Elmgren 1984). A comparison of effects from three similar coastal hypoxic zones with the Gulf indicates that, at least for now, significant declines in fishery production attributable to hypoxia have not been documented for the Gulf of Mexico (Table 3.1).

In the Kattegat, the sea between Denmark and Sweden, indications of troubled waters were seen by the mid-1970s (increased frequency of algal blooms), with seasonal summertime hypoxia observed since the early 1980s. Initially, hypoxia caused mass mortality of commercial and noncommercial species. Now large-scale migrations and/or mortality among demersal fish and the Norway lobster (Nephrops) continue, resulting in a changed species composition and reduced growth and biomass. Hypoxia in this area is believed to be partly responsible for the overall decline in the stock size, recruitment, and landings of commercial fish over the last two decades (Baden et al. 1990, Pihl unpublished data). However, hypoxia is not the only stress factor. Other factors implicated in declining stocks or populations are eutrophication (Caddy 1993), bycatch (Andrew and Pepperell 1992; Chesney et al. in press), trawl disturbance (Currie and Parry 1996), fishing pressure (Turkstra et al. 1991), habitat loss (Chesney et al. in press), and harmful algal blooms (Karup et al. 1993). Given the complexity and potential synergism of stressors, the effects of hypoxia are more clearly expressed in other coastal systems around the world than the northern Gulf of Mexico. Like the Gulf of Mexico, many systems around the world have gradually become eutrophic and hypoxic, but other systems have reached a point were fisheries are clearly negatively affected.

If hypoxia in the Gulf gradually increased in size and duration from its inception--most likely in the 1950s (Sen Gupta et al. 1996)--then the ecosystem's response may also have been gradual and to date not catastrophic. In this scenario the northern Gulf of Mexico ecosystem, while certainly affected by hypoxia and other stressors, has for the last few decades maintained fishery production at a level that we have not been able to distinguish from that experienced previously (Chesney et al. in press). Any scenario that had hypoxia appearing suddenly would have precipitated an ecosystem response similar to the 1976 hypoxic event off the coast of New York/New Jersey, which caused mass mortality of many commercial and non-commercial species (Azarovitz et al. 1979; Boesch and Rabalais 1991). While mass mortality events have been reported in the northern Gulf of Mexico (McEachron et al. 1994), current ecological conditions and lack of any recorded hypoxia-related mass mortality of fishery species, other than "Jubilees," tend to support the scenario of hypoxia developing gradually through time.

From an ecological standpoint, hypoxia in the northern Gulf of Mexico has become one of the many factors that at present control or influence the population dynamics of the regions of many pelagic and benthic species (Chesney et al. in press). Hypoxia exerts its control at two levels:

• First, and most obvious, is the loss of bottom and near-bottom habitat through the seasonal depletion of oxygen levels. From all indications, few if any, mobile organisms stay on bottoms that are hypoxic (Pavela et al. 1983; Leming and Stuntz 1984). This forced movement could increase population losses due to predation of species dependent on the seabed for their survival. But when hypoxia dissipates, mobile organisms return (SEAMAP database).
• Second, and not so obvious, is the alteration of energy flows. During hypoxia, significant amounts of the system's energy are shunted to microbial decomposition (Rosenberg and Diaz 1993; Grant et al. 1995), the primary biological process that creates and maintains the hypoxia. In the northern Gulf of Mexico the major sources of this energy seem to be organic matter produced in the surface waters and benthic biomass (mostly small worms, snails, and clams) (Harper et al. 1991; Rabalais et al. 1993, 1995). Energy flows between trophic levels (mediated largely by predator/prey interactions) are also altered. In Chesapeake Bay, demersal fish and blue crab abruptly stopped feeding when dissolved oxygen declined below approximately 2.1 mg/l (Nestlerode and Diaz 1998). But they quickly returned to affected areas when hypoxia dissipated, and resumed feeding activity before stressed infauna were able to return to normal living positions in the sediment (Pihl et al. 1992). Breitburg et al. (1997) also found hypoxia altered predator-prey interactions to affect the survival of larval fish.

Considering all the potential sources of stress to the Gulf ecosystem, the most likely contemporary ecological impacts of hypoxia involve altered energy flows and community structure. Annual hypoxia affects energy flows in three primary ways, which unfortunately are not exclusive responses to hypoxia: (1) pelagic species not dependent on direct contact with the bottom take on a more prominent role; (2) energy is diverted from invertebrates to microbes (Pearson and Rosenberg 1992); and (3) the pulsing of energy flow through the ecosystem favors opportunistic species with shorter life cycles, which can take advantage of the shortened time bottom habitats are available. A reduction in the overall biodiversity of the ecosystem is also associated with energy-pulsed systems, since longer-lived species tend to be eliminated (Odum 1981). The first point, enhancement of pelagic species, is a basic ecosystem response to eutrophication and not hypoxia; the third point is a response to the hypoxia; and the second point is a response to both.

Zooplankton are an important trophic link between primary production and fisheries species, particularly for larvae, and also contribute to the flux of organic matter to the bottom through the sinking of fecal pellets. Unfortunately, the effects of hypoxia on zooplankton are not very well understood. In addition to mortality, which reduces their population numbers, the documented responses of zooplankton to hypoxia are avoidance for adults, interference with vertical migration, and changes in the composition of smaller species that carry their eggs (Powers 1997; Zaitsev 1992; Qureshi 1995; Stalder and Marcus 1997). Copepods, the dominant zooplankton in the northern Gulf of Mexico, were lower in abundance or absence when dissolved oxygen was <1 mg/l (Qureshi 1995). A similar reaction to hypoxia has been documented in Chesapeake Bay (Roman et al. 1993). While scant data indicate that there may be some avoidance of hypoxia and interference with vertical migration (Powers 1997; Qureshi 1995), the overall abundance and community composition of zooplankton populations appear to be similar to other coastal areas.

The response of northern Gulf of Mexico benthic communities to hypoxia--primarily noncommercial species, which form the base of the food web for shrimp and most demersal fish--has been similar to that of communities from other systems around the globe. In all cases of summertime hypoxia, varying degrees of benthic community mortality were reported, with recolonization of the bottom after hypoxia dissipated. (For Gulf studies, see Gaston 1985; Murrell and Fleeger 1989; Harper et al. 1991; Rabalais et al. 1991, 1993, 1995; Gaston et al. 1998; Vittor & Assoc. 1998. For global studies, see Diaz and Rosenberg 1995.)

There are no historical benthic data sets that can be used for a comparison of pre- and post-hypoxic conditions. The earliest data come from a series of surveys from 1951 to 1957 of the Louisiana-Texas continental shelf by Parker (1960), which mostly described the large fauna (molluscs, echinoderms, crustaceans) associated with shelf regions. Most of the species that Parker (1956, 1960) listed as common were collected by various recent studies (summarized in appendix C of Gaston et al. 1998), but because of gear differences and area sampled, direct quantitative comparison was not possible. The small opportunistic bivalve Mulinia lateralis stands out as having been recorded as a dominant component of the fauna in shallow muddy areas in the 1950s and in the 1980s-90s.

Recent data on benthos in the northern Gulf of Mexico come mostly from areas on the eastern side of the hypoxic zone, but showed general signs of hypoxic stress that are most likely found throughout the hypoxic zone. Mortality of benthos was the prominent effect, with the degree of mortality directly related to the severity of hypoxia (Boesch and Rabalais 1991; Rabalais et al. 1993, 1995). Giammona and Darnell (1990) studied the effects of brine discharge on benthos and found that the community around the brine diffuser experienced less mortality due to the mixing effect of the diffuser, which did not allow the dissolved oxygen to decline as much relative to areas away from the diffuser. Areas that experience severe hypoxia, with near-anoxic conditions, experienced greater mortality and had fewer species and lower bio-mass than areas with intermittent or less severe hypoxia (Table 3.2). Even though mortality from hypoxia was widespread, no study of Gulf benthos has recorded azoic bottom. Recolonization of benthos, while variable from year to year, appeared to occur annually.

Source: Rabalais et al. 1995.

In 1980, the Louisiana Department of Wildlife and Fisheries started a long-term benthic monitoring program to investigate the possible effects of oil activities associated with the Louisiana Offshore Oil Port (LOOP). Two study areas located in the Gulf were subjected to hypoxia and provided insight into how benthos respond in the long term. While both study areas were 27 km apart, the shallower nearshore site experienced more frequent and more severe hypoxia than the deeper offshore site on the offshore edge of the hypoxia. Recruitment at the nearshore stations was pronounced and occurred mainly in the spring. Bivalves (mostly Mulinia lateralis) dominated the spring pre-hypoxia recruitment, with polychaetes (mostly Paraprionospio pinnata) dominating the summer and fall post-hypoxia recruitment. At the offshore stations recruitment was not pronounced and tended to occur in the summer (Table 3.3). While hypoxia was more severe at the nearshore area, it also had higher densities of organisms relative to the offshore area (Table 3.3). However, much of the difference between the nearshore and offshore areas could also be related to timing of recruitment or differences in predation pressure, depth, and/or sediment type. Considering this mix of physical and biological processes that are known to affect community structure, the lack of obvious long-term hypoxic effects on benthic community structure is not surprising. The LOOP study also did not find any benthic effects attributable to oil development activities (Vittor & Assoc. 1998). Most of the dominant species reported are opportunists, which are tolerant of a wide range of environmental stressors, including hypoxia (Diaz and Rosenberg 1995), and adapted to rapidly recolonize disturbed areas (Boesch and Rosenberg 1981).

Source: Vittor & Associates 1998.

Overall, the range of abundance reported for hypoxia-stressed bottoms is well within that reported throughout Louisiana's coastal waters (Table 3.4). Comparison of benthic communities between the area affected by hypoxia with those unaffected in nearby Mississippi Sound indicated that a functional shift has occurred, away from a better-developed, late successional stage, "equilibrium"-type community toward an early successional stage disturbance, adapted community (Pearson and Rosenberg 1978; Rhoads and Germano 1986). Few large, long-lived, deep-burrowing species were present in the areas affected by hypoxia, where there are more small, short-lived species (Rabalais et al. 1993, 1995; Gaston et al. 1998; Vittor & Assoc. 1998). This influences the ability of the benthos to process sediments (D'Andrea et al. 1996) and store energy (Diaz and Schaffner 1990). However, there is evidence that shrimp feed heavily on these short-lived species (Harper et al. 1991) and may even benefit from the shift to a smaller-bodied, shallower-dwelling infauna (McTigue and Zimmerman 1998). This basic difference between the LA and MS shallow continental shelf is at least consistent with stress from hypoxia.

The reduction in numbers of long-lived, deep-burrowing species has a direct negative effect on the linkage between benthic and pelagic communities within the system. Long-lived species tend to be larger than short-lived species and perform two important ecosystem functions. First, they can store large amounts of energy as biomass, buffering the ecosystem against the pulsing of energy (Odum 1981). Second, they burrow deep into the sediments and aid in the cycling of nutrients and other sediment-bound substances (Rhoads 1974). This "aeration" of sediment in turn helps to prevent the buildup of organic matter (Aller and Aller 1998). Without long-lived species, it is more difficult for the ecosystem to regulate and store the energy that comes to the bottom. Energy is then recycled over shorter time intervals because it is passed on to smaller short-lived species that undergo boom-and-bust cycles (very quickly building large populations that crash to near zero when energy is exhausted or other factors come into play, such as predation or hypoxia). Short-lived species are not good biomass revisions for energy and are of minimal help in oxygenating sediments.

As far as benthic communities are concerned, hypoxia takes the bottom out of production for a period of Time--possibly as long as half a year for the northern Gulf of Mexico, depending on the duration of hypoxia. (See the Topic 1 report, Characterization of Hypoxia.) Mortality of affected communities is obvious, but the question that remains is: Is the overall secondary productivity of the system affected?

The quality and quantity of secondary production are important, since benthic invertebrates are the major prey base for brown shrimp and other bottom-feeding fish and crabs. Unfortunately, there are no measurements from other systems that are directly comparable. In general, however, bottoms that are exposed to extensive periods of anoxia have low annual production (Rainer and Fitzhardinge 1981). The amount of productivity in these severely stressed habitats is a function of how quickly benthos can recruit and grow during periods of normal oxygen.

In a near-anoxic basin--Port Hacking, Australia--productivity was almost 16 times lower than at a nearby station that only experienced hypoxia (Rainer and Fitzhardinge 1981). In Chesapeake Bay, areas that experience anoxia have about half the annual productivity of areas that are only hypoxic. While some areas affected by seasonal anoxia have lower productivity, the trend is inconsistent across habitat types. This results from a combination of duration of exposure to anoxia and rapid recovery of benthos. Annual secondary production of Chesapeake Bay habitats known to experience only seasonal hypoxia was of the same magnitude as habitats that always had normal oxygen levels, although the qualitative nature of the production differed (Diaz and Schaffner 1990; Neubauer 1993).

Hypoxia clearly disrupts the movement and migration patterns of many species. Trawl data from the fishery-independent SEAMAP database (1987, 1988, 1993, 1994) showed a very consistent pattern that whenever dissolved oxygen declined and approached 3 and 1 mg/l, catch of shrimp and fish, respectively, rapidly declined to zero (Figures 3.1 and 3.2).

In laboratory experiments, both white and brown shrimp were able to detect and attempted to avoid hypoxic water (<2 mg/l) (Renaud 1986). Because shrimp must move from inshore wetland nurseries to off-shore feeding and spawning grounds, hypoxia poses a major problem to their migration. Gazey et al. (1982) found that juvenile brown shrimp leaving marsh nurseries moved farther offshore when hypoxia was not present--an indication that if hypoxia were not present shrimp could exploit more bottom area. Analyses of the distribution of shrimp catch and the localization of catch on the shelf in relation to hypoxia also suggest that hypoxia interferes with shrimp migration (Zimmerman et al. 1996), and where hypoxia was extensive, catch of shrimp was consistently low (Figure 3.3).

The blocking effect of hypoxia on shrimp migration reveals that hypoxia impedes offshore movement, so nearshore abundance of shrimp may be higher. Shrimp are actually capable of moving far distances relative to their small size. Sheridan et al. (1987) reported brown shrimp were capable of moving 1.3-1.6 km/day. The maximum distance traveled was 620 km in 320 days. Hypoxic conditions can easily cause brown shrimp to redistribute hundreds of kilometers to more favorable conditions elsewhere, such as the Texas or Mississippi shelves. A similar blocking effect for bluefish was recorded for the 1976 New York/New Jersey hypoxic event. This event clearly demonstrated that coastal hypoxia cannot only displace demersal fish but can also block the migration of pelagic fish. Northward-migrating bluefish that encountered the hypoxic zone did not pass through or go around it, but stayed to the south, waiting for it to dissipate and then continued their migration north (Azarovitz et al. 1979). If it is important for fish and shrimp to reach critical nursery or feeding areas at certain times in their lives, then hypoxia may have an

indirect effect on populations by delaying arrival to spawning or feeding grounds. In such cases, the cost of delayed migration in terms of population mortality and production is unknown.

The brown shrimp catch in the Gulf (mostly caught in Louisiana and Texas) declined from a record high in 1990 to below the historical average during 1992-97. This decline coincides with years of greatly increased hypoxia. By comparison, white shrimp catch within the same area and time period did not exhibit a similar decline. A greater negative effect on adult and large subadult brown shrimp would be expected from enlargement of the hypoxic zone because they require offshore shelf habitat, whereas white shrimp do not (they require nearshore habitat). Thus, the differences in life-cycle requirements are more likely to expose brown shrimp than white shrimp to the offshore effects of hypoxia.

Also, there continues to be the pattern of low shrimp catch and shrimping effort in offshore Louisiana. Geographic comparison of distribution of shrimping effort in Texas and Louisiana shows lower shrimping effort in Louisiana to be coincident with shelf hypoxia. Moreover, although Louisiana's management strategy favors inshore fisheries, it does not preclude offshore shrimping and does not adequately explain the low catch patterns on the Louisiana offshore shelf, where subadult and adult brown shrimp are expected to occur.

Unfortunately, there have been no studies of how hypoxia affects fish populations in the northern Gulf of Mexico. However, data collected by the SEAMAP program will be instrumental in determining any hypoxic effects. Preliminary analysis of this data set indicates that there may be a herding effect, with higher abundance of fish on the edge of the hypoxic zone (Crowder and Craig, personal communication).

Some effects of hypoxia that have been seen in the Kattegat, Baltic Sea, and Chesapeake Bay are difficult to detect without directed studies, but could occur in the northern Gulf of Mexico. Pelagic eggs of economically and ecologically important species can sink into hypoxic bottom waters and die. For example, oxygen concentrations from 4 to 5 mg/l reduced the survival of cod eggs (Nissling and Vallin 1996). Eggs sinking into hypoxic water is thought to be an important factor that leads to the decline of cod in the Baltic Sea (Nissling et al. 1994) and may be an important source of mortality for Chesapeake Bay anchovy (Kiester et al. unpublished). Habitat restriction can also increase the predation mortality of fish (Breitburg 1992; Breitburg et al. 1997) and can force fish into areas with poor growth potential (Coutant 1985). And hypoxia may increase the incidence of disease and parasitism in exposed fish populations. Mellergaard and Nielsen (1995) found the relative risk of contracting lymphocystis, a viral disease that affects many marine fish worldwide, significantly increased in the Kattegat from 1984 to 1986, compared to 1987 to 1991, years of severe hypoxia.

Also of concern to fisheries populations is the increased incidence of harmful algal blooms in coastal waters resulting from eutrophication (Goldberg 1995). Harmful algal blooms have long been linked to fish kills (Brongersma-Sanders 1957), but recent developments along the Atlantic East Coast indicate how serious they can be (Burkholder et al. 1992). There is also a strong relationship between bloom-related mortality and hypoxia. (See Burkholder et al. 1995 as a recent example. For more details on phytoplankton community changes, see the Topic 1 report, Characterization of Hypoxia.)

As noted, the analysis focused mainly on brown shrimp (Penaeus aztecus) and white shrimp (Penaues setiferus). The biology and fisheries characteristics of these species are described in detail in the Fishery Management Plan for the Shrimp Fishery of the Gulf of Mexico, United States Waters (Gulf of Mexico Fishery Management Council 1981) and in subsequent amendments. The data used in this analysis were extracted from the extensive database of monthly statistics maintained by the National Marine Fisheries Service (NMFS) Laboratory in Galveston, Texas (Poffenberger 1991).

Statistics are maintained for 21 statistical areas (Figure 3.4). The offshore waters of each area are divided into depth zones of 5-fathom increments out to 40 fathoms. In the analysis described here, the statistical areas were grouped into three zones: eastern Louisiana (ELA), covering zones 11 and 12; Louisiana (LA), covering zones 13 through 17; and Texas (TX) covering zones 18 through 21. As the hypoxic zone in the northern Gulf lies within the zone designated as LA, the other two zones serve as a kind of control. It is important to have such a control to distinguish between effects due to hypoxia--which would be confined to LA--and effects due to climatic or other covariates--which could also appear in ELA, TX, or both. The decision to group the statistical areas in this way was based on an extensive analysis of the data for each of the 21 statistical areas. In particular, there is considerable variability between statistical areas, and combining them in this way into zones reduces this considerably, while retaining the value of comparisons between zones.

The analysis involved the calculation of correlations between INDEX and selected fisheries variables for each of the three zones--ELA, LA, and TX. Correlations were based on annual values, although, as with INDEX, the annual values of the fisheries variables were calculated using data for selected months (e.g., brown shrimp CPUE (cost per unit effort) during July-August). As the effect of hypoxia in a particular year may be felt on shrimp in the following year, for certain fisheries variables, both contemporaneous and lagged correlations were calculated, with the largest in magnitude reported below. In the latter case, INDEX lagged the fisheries variable by one year. For completeness, correlations between AREA and the fisheries variables are also presented. However, the time series of AREA was felt to be too short to serve as an adequate basis for empirical analysis, and the analytical effort was directed at the correlations with INDEX.

It is well known that the presence of secular trend can produce spurious correlations or obscure genuine correlations between time series. To avoid this problem, estimated trends were removed from both the INDEX and the fisheries variables by subtracting seven-year moving averages. As noted, this analysis should be viewed as exploratory. The basic idea is that an effect could be attributed to hypoxia if it involved a correlation of reasonably large magnitude that occurred in LA, but not in ELA or TX. To supplement the ordinary correlations, the partial correlation between INDEX and each fisheries variable for LA, with the "effects" of the same fisheries variable in ELA and TX removed, was also calculated. This amounts to treating the time series for ELA and TX as proxies for the effects of climatic and other factors that do not involve hypoxia. This involves the strong assumption that the effects of these factors in LA are linearly related to their effects in ELA and TX.

Although no formal assessment of statistical significance was attempted, as a rough guide, under the unrealistic assumption that each of the individual time series is serially independent, a correlation or partial correlation would have to be around 0.35 in magnitude to achieve significance at the 0.05 level. This critical value would have to be adjusted to account for serial dependence. The assessment of statistical significance would also be complicated by the consideration of so large a number of (not necessarily independent) correlations and by the decision to select (e.g., via the consideration of multiple lags) the largest magnitude correlations.

In Figure 3.5, annual time series of combined July and August catch per unit effort (CPUE) for brown shrimp are shown for the three zones for the period 1960-96. The July-August period covers the most intensive fishing period for this species. Figure 3.6 shows the same time series for white shrimp during the August-September period, which covers the most intensive fishing for this species. In Table 3.5, lag 1 correlations between AREA and log CPUE and between detrended INDEX and detrended log CPUE are reported for each species within each of the three zones. Also reported in Table 3.5 are the partial correlations between detrended INDEX and detrended log CPUE in LA, with the effects of detrended log CPUE in ELA and TX removed. The logarithmic transformation of CPUE reflects the natural assumption of a multiplicative, rather than additive, effect.

The partial correlations (Table 3.5) are the best evidence of a hypoxic effect uncovered in this analysis. While both are rather modest in magnitude, each has the expected sign. However, in this regard, it is important to point out that these estimated partial correlations are not independent of each other and, therefore, cannot be taken as independent pieces of evidence of a hypoxic effect. To illustrate the potential for overinterpretation, the partial correlation between detrended INDEX and detrended log CPUE for brown shrimp in ELA with the effects of LA and TX removed is -0.30. This cannot be attributed to hypoxia, which is absent in ELA.