Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 Data Requirements for Population Assessment Three kinds of data are required in developing population assessment models: (1) total catch (or total removals if including bycatch); (2) demographic information on the size, age, and taxonomic composition of the fish removed; and (3) indices of relative abundance. Total removals by size and age are used to measure the level of mortality incurred by different components of the population. Abundance indices serve to denote relative change in the fish population over time. These indices can be based on data collected directly from the fishery (i.e., fishery- dependent indices, such as catch rate indices from fishery logbooks) or data collected independent of the fishery (i.e., fishery-independent indices, such as research surveys). Information on these three kinds of data ideally should be obtained from all fisheries and gear types involved in removals from the population. Commercial fisheries have been the main source of fishery- dependent data used in developing quantitative population assessments; however, more and more often, data from recreational fisheries are relied on to complement data collected from other sources or as the sole source of information for some assessments. This increased demand on recre- ational fisheries data necessitates a discussion of the survey methods used in recreational fisheries and whether these methods provide data adequate for assessment purposes. Certainly the survey design and data collection recommendations outlined in earlier chapters are likely to improve the information used for population assessments. In particular, the establishment of mandatory logbooks to monitor catch for all vessels in the for-hire sector would provide more in-depth data--the kind of data that would be ideal for use in population assessments. Logbooks could provide fishing location, time of day, and weather conditions, all of which could be helpful in inter- 83
84 REVIEW OF RECREATIONAL FISHERIES SURVEY METHODS preting catch rate estimates. Onboard observers also could be used for one-time studies into catch rate, such as investigations into the influence of the kind of bait used or the depth being fished, or they could assist in the collection of specific oceanographic and meteorological data during the fishing trip. In addition to identifying reliable data sources, data quality must be assessed and accounted for appropriately. Modern statistical population assessment models are capable of dealing with data characterized by different variance structures, or even unknown variance. Not surpris- ingly, what goes into the model influences what comes out, and the accuracy of population estimates is influenced by the accuracy of the data used. Statistical fitting procedures used in these models often assume variance structure for data inputs that are not likely to be met by most recreational fisheries sampling programs. Assessment models can be modified to accommodate such data characteristics, but these characteristics first must be identified and quantified at the source level of the surveys. Inconsistencies in how dockside samples are collected can be particularly aggravating when conducting population assessments. For example, the lack of a common knowledge base among anglers, data collectors, and data users with regard to taxonomic identification will bias mortality estimates for all species concerned. Population assessment scientists must have confidence that species designations are accurate and applied consistently in the sampling process. Therefore, biological data obtained from intercept surveys must be consistent with categories used in assessments. Two additional issues complicate the usefulness of recreational fisheries data for population assessment. One concern is the challenge faced by population scientists in interpreting catch and effort data recorded from recreational fisheries surveys in ways that are analogous to commercial and scientific indices to measure changes in relative abundance. Obtaining a measure of catch per unit effort (CPUE) that is a true measure of relative abundance is challenging since the measures for these different data sources are compiled with different purposes in mind. For commercial fisheries, catch and effort are obtained simul- taneously from individuals in association with an area fished and species targeted; thus, CPUE can be seen as a direct measure of relative abundance for a given area and species, as long as fishing efficiency and catchability do not change. In recreational fisheries surveys, such as the Marine Recreational Fisheries Statistics Survey (MRFSS), CPUE is obtained from individuals and is expanded by an estimate of effort across
DATA REQUIREMENTS FOR POPULATION ASSESSMENT 85 all individuals to develop an estimate of total catch. For these surveys, recreational CPUE typically is not associated with specific areas or even with specific target species; thus its applicability as a relative abundance measure is clouded by its design as a means to obtain total catch in the survey. To better address these issues, a closer look needs to be taken at effort and CPUE calculations as they are carried out in a recreational fisheries context. The other complicating issue is how catch and release influences the accuracy of total removals reported and the subsequent underestimation of fishing mortality. Underreported removals occur when fish are released but subsequently die from capture and handling. If catch-and- release survivorship rates are not known, the proportion of releases that die is not known. This is further complicated by the fact that the number of releases (by species) probably is not estimated accurately either. The numbers of released fish are obtained from the intercept survey, and the accuracy of this information may be dependent upon the memorableness of the release event. Most anglers would remember releasing a marlin but may be uncertain as to how many of a more common species, such as striped bass or mackerel, were released. EFFORT AND CATCH PER UNIT EFFORT CALCULATIONS Stock assessment scientists often use the reported CPUE from commercial logbooks as a fishery-dependent index of abundance. The basic assumption is that catch (C) is a function of fishing effort (E), and catchability of the fish to the fishing gear used (q) is constant over time, such that: C = f(E,q,N) where N is the population size. Effort is usually a function of time spent actively fishing (trawling) or the time a specific amount of passive gear (e.g., traps, pots, longlines) was in the water. The simplest form for this function is a proportional relationship: C = E × q× N and as a result:
86 REVIEW OF RECREATIONAL FISHERIES SURVEY METHODS CPUE = C = q × N E Assuming a constant catchability, CPUE should track changes in population size over time. In the MRFSS, the catch per trip from the onsite interview survey is often referred to as a catch rate or CPUE, but this estimate is rarely the one used in stock assessments when defining a recreational CPUE index of abundance. The definition of fishing effort, and hence, fishery catch rate as an index of abundance needs to take into account what species the effort was directed for and not just the total catch over a set amount of time. Holiman (1996) defines three types of effort that can be calculated from the MRFSS data: target effort, catch effort, and directed effort. Population assessment scientists must be aware of the presence of these three different types of effort in the database and understand how they relate to the problem of estimating relative abundance. Those in charge of data collection and monitoring also must be aware of these effort types in order to document them properly but also to insure that the right type of information can be made available to those in need of it. Target effort is based on the anglers' identification of their primary or secondary target species to the intercept interviewer, whether or not they were successful in catching any of that species. Interviews occur after the fishing trip is completed; therefore, accepting the angler's designation of target species after the fact may result in biased estimates, as some people may report only what they caught as being what they targeted. This is often referred to as "prestige" bias since it is a result of anglers not wanting to admit that they were unsuccessful in catching what they were targeting. In cases where there are multiple anglers (e.g., head boats) and catch cannot be separated by angler, total catch is attributed to one angler (termed "leader" in the MRFSS) who represents the other anglers on the trip (designated as "followers"). Generally, it is assumed that all followers fish when the leader fishes for the target species (Holiman, 1996). If followers do not fish when the leader does, the amount of target effort will be overestimated. However, if the leader does not report a target species but one of the followers does, it is not assumed that all followers also targeted that species. This assumption may result in an underestimate of target effort. The most recent Atlantic bluefish stock assessment used target effort to define catch rate indices (Lee, 2003) and
DATA REQUIREMENTS FOR POPULATION ASSESSMENT 87 therefore might suffer from the difficulties mentioned above. In addition, for schooling fish species (such as bluefish), there is an increased probability that if a follower reports targeting that species, the other anglers did as well. Catch effort is the effort associated with the successful catch of a species, whether or not it was targeted. In addition to the issues raised above for target effort, assumptions have to be made when calculating the total effort for groups of anglers. Again, all catch, either kept for interviewer inspection or not available (i.e., filleted, released dead or alive, given away), is attributed to the leader. However, the number of angler trips associated with catching these fish is not recorded. That is, assuming that bag limits exceed one fish per angler, one person may have caught more than one of these fish. For example, Holiman's code assumes that if the number of fish is less than the number of anglers, then the number of trips equals the number of fish because the focus here is the successful catch of a specific species; otherwise, the number of trips equals the number of anglers. Effort calculations for the red grouper assessment simply use all of the anglers when dealing with multiple angler intercepts (Southeast Fisheries Science Center, 2002). Directed effort is the effort associated with all catch of a particular species whether targeted (including unsuccessful catch) or not. The difference between target effort and catch effort is referred to as effort associated with incidental catch. All of the above deal with effort estimates obtained from anglers interviewed during intercept surveys. However, there is no information available in the MRFSS on target or other kinds of effort for anglers who have private access. At present, it must be assumed that this portion of the angler effort is represented adequately by the sampled portion from intercepts. Most stock assessments try to use some form of target effort, and the main issue is how to calculate the "target but no catch" portion of the effort in a way that does not rely on the anglers' identification of target species. Ralston and Dick (2003) use location data from the California commercial passenger fishing vessel (CPFV) data to restrict black rockfish data to only those locations where black rockfish had been caught in at least five separate locations. The latest assessment for red snapper (Gulf of Mexico Fishery Management Council, 2005a) uses only trips where at least one snapper was caught or where a catch of species typically associated with snapper was caught in the past. No-catch reef trips for hogfish are defined as reef trips using hook and line or spear in counties where hogfish were not caught in the current trip but had been
88 REVIEW OF RECREATIONAL FISHERIES SURVEY METHODS caught at least once in the period 19822001 (Ault et al., 2003). As another approach, Stephens and MacCall (2004) describe a logistic regression approach based on multispecies presenceabsence information that predicts the probability that the target species would be present based on other species caught, even if no catch of the target species was recorded. Methods that depend upon species-complex indices to determine targeted non-catch trips may be confounded by differential targeting of the fishery from year-to-year (or season-to-season) on more desirable species or by changes in the complex arising from different dynamics of the component species. Also, regulation changes for associ- ated species may complicate interpretation of species-complex catch information. While all of these methods are genuine attempts to measure target effort, confidence in their use in stock assessments will require more research with respect to multi-species associations and the impact of species-specific catch regulations. In recent west coast Stock Assessment Review Panel (STAR Panel) reports (Pacific Fishery Management Council, 2006), the use of recreational catch rate data from CPFV logs or Recreational Fisheries Information Network (RecFIN) data for rockfish and similar species has come under close scrutiny. The usual assumption of proportionality between catch rate and abundance used for commercial indices has not been tested for recreational fisheries. These reports note that recreational fishing, especially when conducted by the for-hire sector, focuses on giving the angler a successful fishing experience with respect to the desirability of and the challenge of landing the species being sought. This behavior may lead to the targeting of fish in high density areas, resulting in catch rate indices exhibiting a slower decline than what the actual population is experiencing (i.e., hyperstability). There are many other factors also at play in determining what makes a successful fishing experience (Holland and Ditton, 1992) that may further complicate the link between recreational catch rate and population size. Technological improvements (e.g., Global Positioning System [GPS]) are not usually taken into account when using recreational catch rates as indices of abundance. In addition, changes in fisheries regulations for the targeted or associated species may change the relationship between catch rate and population size. As fisheries become more restrictive with respect to bag and size limits, the increasing number of releases may result in CPUE being prone to recapture bias (e.g., lingcod) (King and Haggarty, 2004). For many fisheries where there, currently, is a small or no com- mercial component (e.g., rockfish on the U.S. west coast), recreational catch rates are usually the only abundance indices available for the recent
DATA REQUIREMENTS FOR POPULATION ASSESSMENT 89 years. Despite the problems that have been identified with using recre- ational effort data, assessment scientists need to have access to the best effort data possible either to determine whether these kinds of data can be used to monitor abundance or to make the necessary modifications so the data are useful. CATCH AND RELEASE For stock assessment purposes, the total number of fish removed from the population by the fishery is of more interest than just the number of fish landed. Total removals are calculated as the sum of those fish caught and landed or known to be dead upon capture and those fish that were released (or discarded) but did not survive. There are two types of catch records in the MRFSS database: type A and B. Type A records account for fish that were caught, landed whole, and available for identification by the intercept interviewers. These fish are available for weight and length measurements, although these measurements may not always be taken. For type B records, the fish were caught but were either not kept or were unavailable for identification. These records are further identified as either type B1 or B2. The former type refers to fish that were filleted, released dead, given away, or disposed of in some way other than for types A or B2. Those fish that were caught and released alive are coded as B2. Total landings from the recreational fishery are calculated as A+B1 for stock assessments where there are recreational components. For example, in 2003, recreational landings (A+B1) in the striped bass fishery were estimated at 2.4 million fish or 11,486 metric tons (25.3 million pounds) from the MRFSS. These landings constituted 74 percent of the total landings of striped bass by the recreational and commercial fishery (Atlantic States Marine Fisheries Commission, 2004). Some fish released alive, as recorded in the B2 records, are expected to die after being released. This subsequent mortality is often referred to as a hooking or release mortality and can arise for a number of reasons, including swim bladders expanding too quickly as a result of fish being brought up from significant depths. There is also the possibility that fish exhausted from fighting the angler are more susceptible to predation. In 2003, for striped bass, the B2 catch was estimated at 14.6 million fish. Assuming an 8 percent hooking mortality rate, catch and release resulted in an estimated removal of 1.2 million additional fish.
90 REVIEW OF RECREATIONAL FISHERIES SURVEY METHODS The issue of hooking or release mortalities has been the subject of a number of studies (e.g., Lucy and Studholme, 2002; Policansky, 2002), but these mortalities have not been estimated for all recreational species. Specific studies have often limited their attention to particular combinations of factors, such as hook type, depth range, and temper- atures, that can be manipulated in an experimental setting. Not all factor combinations that could be significant may have been studied, but more to the point, there may not be enough detail on the released fish from the intercept surveys to determine which, if any, of these factors are operating at any one time. Although some hooking mortality studies appear to be reliable for the restricted conditions they apply to, not all are reliable, and for many species, the hooking mortality associated with particular gear types under a variety of conditions simply is not well known. In addition to providing information for stock assessment scientists, hooking mortality studies also have the potential to improve management by advising anglers on how to handle and release hooked fish to increase their chances of survival. The lack of accurate information for estimating release (or discard) mortality has been identified as problematic for the red snapper assessment in the Gulf of Mexico. For red snapper, recreational data were obtained from three sources: · The MRFSS (19811998) with some exceptions: (1) no wave 1 data in 1981, (2) no Texas boat mode in 19821984, (3) no Texas data after 1986, and (4) no head boat sampling after 1985 · The National Marine Fisheries Service's Beaufort Laboratory head boat survey for all states after 1985 · The Texas Parks and Wildlife Department's coastal sport fishing survey Recreational discards data are collected by the MRFSS in the Gulf of Mexico but are not available for Texas landings or for landings from head boats. Mortality rates used for discarded live red snapper differ according to depth and area and therefore depend upon accurate location information of where the discards occur across the whole range of the fishery (see Appendix C). Many stock assessments convert numbers caught to weight caught. Weight conversions are based on the length and weight information obtained from the type A catch; the size compositions of the type B1 and B2 catch are assumed to be similar to the type A catch--a potentially
DATA REQUIREMENTS FOR POPULATION ASSESSMENT 91 strongly biased approach given that one of the main reasons for releasing fish is that they are below the size limit. At present, there are some limited programs to capture length, weight, or age data from the recreational discards (e.g., striped bass lengths are available from vol- unteer angler logbooks and American Littoral Society data), and, starting in 2003, California Recreational Fisheries Survey (CRFS) samplers have measured length and weight of discarded fish from CPFVs and from onshore anglers. (See Appendix B for more detail on CRFS.) Another problem is the comparability of discard data between different recreational surveys that may be combined into a stock assessment. As an example, recreational catch data for lingcod on the Pacific coast come from a variety of sources, but not all sources provide the same level of detail with respect to the condition of the fish caught or released. For California, the RecFIN database (including the MRFSS) was used for 19801989 and 19932003. Beginning in 2004, CRFS has been used in place of the MRFSS in California. Oregon recreational catch data are provided by the Oregon Department of Fish and Wildlife, and Washington catch data are obtained from the Washington Department of Fish and Wildlife (WDFW) Ocean Sampling Program (see Appendix B). Discard information on numbers and disposition (released alive or dead) is available from CRFS. On the other hand, only the number released is available from the Oregon Recreational Boat Survey data (see Appendix B). The WDFW has collected discard information from the recreational fishery since 2002 but does not collect information on the portion of the catch discarded live or dead. In Washington, 57 percent of the lingcod catch is estimated to be discarded, but it is unknown how many of the live releases survive. Various adjustments are made to the catch and projections in the assessment to account for discard mortality. Yet, recent stock assessments for lingcod identified the need for better coastwide enumeration of at-sea discards and mortality of released recreational fish to account for total removals from the population more accurately. CONCLUSIONS AND RECOMMENDATIONS Documentation of the source of the effort available in the MRFSS, as to whether it falls into the category of target effort, catch effort, or directed effort, would go a long way in helping population scientists use this data in an appropriate manner. Population scientists should work in collaboration with those involved with data
92 REVIEW OF RECREATIONAL FISHERIES SURVEY METHODS collection and monitoring to design data collection protocols and sum- mary statistics that are appropriate for use in population assessment. Information on target species and area fished also would make catch and effort data collected from recreational fisheries more amenable for use in the development of assessment indicators. The establishment of mandatory logbooks to monitor catch for all vessels in the for-hire sector also would be appropriate for the collection of target effort data. Basic data recorded by the vessel captain on the number of anglers, actual hours spent fishing, and target species would get around the complications of the leaderfollower designations currently being used in the MRFSS. The logbooks also could record position, time of day, and weather conditions, all of which could be helpful in interpreting catch rate estimates. These logbooks would not be considered the sole source of information, and similar to the commercial fishery, onboard observers should be used on a sample of the vessels to validate the information, especially in the case of numbers, species, condition, and size composition of the released fish. Recall bias of released fish has been identified as an issue in recreational fisheries (Pollock et al., 1994) and shown to be significant for salmon fisheries in the Strait of Georgia (Diewert et al., 2005). These observers also could be used for one-time studies into catch rate, such as investigations into the influence of the kind of bait used, depth being fished, and discard mortality, or perhaps they could assist in the collection of specific oceanographic and meteoro- logical data during the fishing trip. Information on targeted effort, such as discussed for the for-hire sector, could be obtained for private access anglers as part of a panel survey. Panel surveys could be used to collect a wide range of detailed data from the previously unsampled private access mode. Participants could be contacted by telephone or as part of an internet survey. It may be possible to design these panel surveys in a way that detailed infor- mation on catch rate and targeted species can be related back to the larger telephone survey of private sector anglers providing fisherywide or regional estimates of catch rate for stock assessments. These surveys also can be used to collect information on the sizes of kept or released fish. This may require significant training to ensure accurate species data, but since data will be collected from each participant over a long period time, this investment in training may be worthwhile.