PILOT STUDY INFORMATION

Project Title: Evaluation of two methods to discriminate Pacific herring
(Clupea pallasi) stocks along the northern Gulf of Alaska

Principal Investigator: Ted Otis, Alaska Dept. of Fish and Game, Homer, AK
ted_otis@fishgame.state.ak.us            http://www.adfg.state.ak.us/

Co-Principal Investigator: Ron Heintz, National Marine Fisheries Service-Auke Bay Lab, Juneau, AK
Ron.Heintz@noaa.gov                         http://www.afsc.noaa.gov/abl/

Funding Agency: Exxon Valdez Oil Spill Trustee Council
http://www.evostc.state.ak.us/

Abstract:

Understanding the stock structure of northern Gulf of Alaska (NGA) herring is relevant to how these exploited populations should be assessed and managed.  We evaluated the capabilities of otolith microchemistry and heart tissue fatty acid profile to identify the stock of origin for herring sampled from four focal spawning aggregations in the NGA, and two outside the NGA (Sitka and Togiak).  Otolith microchemistries were measured (ppt) using an electron microprobe equipped with four wavelength dispersive spectrometers. Fatty acid profiles were determined by performing trans-esterification and fatty acid chromatography on purified lipids from whole hearts.  A MANOVA revealed significant regional differences among the mean fatty acid profiles of heart lipids from Sitka, NGA and Togiak herring (P < 0.0001) and among the four NGA stocks (P < 0.0001).  Cross-validation of discriminant functions demonstrated that fatty acid profiles could be used to correctly identify unknown samples more than 90% of the time.  Otolith microchemistries were also significantly different between Sitka and Togiak (P<0.0001) and among some NGA stocks (P<0.0001), however, otolith microchemistries could not effectively discriminate between NGA and other regions, or between two of the four NGA stocks.  Heart tissue fatty acid profiles can be used to discriminate among NGA Pacific herring, but further work is necessary to verify the temporal stability of the biomarkers our study identified before they can be used to determine stock contributions from mixed-stock herring fisheries in this region.

Executive Summary:

Pacific herring (Clupea pallasi) are an important component of the marine ecosystem providing a trophic pathway for energy flowing from secondary producers to apex predators, including humans. Two related aspects of herring life history that remain poorly described are the degree to which herring return to natal areas to spawn and the scale at which population structure exists within large geographic areas.  These key characteristics are directly relevant to how exploited herring stocks should be assessed and managed. 

In this study, we evaluated two tools for discriminating herring stocks over relatively fine (< 100 km) spatial scales.  Tools that can discriminate stocks at this scale are necessary to understand the underlying structure of herring populations in the Northern Gulf of Alaska (NGA).  For instance, commercial fisheries that target herring that spawn in Kamishak Bay (Cook Inlet) and the northwestern side of Kodiak Island (Shelikof Strait) are managed separately, however, there is evidence suggesting these stocks mix in northern Shelikof Strait as rearing juveniles and over-wintering adults (C. Burkey, Alaska Department of Fish and Game, unpublished scale pattern analysis data).  Despite a management strategy that treats these stocks as independent entities, the degree to which they are separated is unknown.  Similar issues surround herring management in Prince William Sound (PWS) where temporally and spatially separated spawning aggregations are managed as a single stock.  Previous efforts to identify population structure among PWS herring using microsatellite DNA variation found little genetic divergence among the four spawning aggregations sampled, with the possible exception of Port Chalmers (O’Connell et al. 1998). 

Many diverse techniques have been proposed as stock identification tools (Pawson and Jennings 1996; Begg et al. 1999a).  Two of the more promising methods recently developed include elemental analysis (EA) of otoliths and fatty acid analysis (FAA) of select body tissues.  Because otoliths are acellular, their microchemistry is stable (Campana and Neilson 1985) and can help identify subtle differences in aquatic environments fish are exposed to during their life (Gunn et al. 1992; Radtke and Shafer 1992; Secor 1992).  This record of environmental exposure relies on the assumption that otolith microchemistry is, in part, determined by the chemistry of the water occupied by the fish (Radtke and Shafer 1992; Campana and Gagne 1995).  Recent reports have also suggested that the fatty acid composition of phospholipids in some body tissues (e.g., heart tissue, brain, eggs) have a stable genetic basis, making these tissues appropriate as stock identifiers (Joensen and Grahl-Nielsen 2000; Joensen et al. 2000).  

In this study, we evaluated the capabilities of otolith microchemistry and heart tissue fatty acid composition to identify the stock of origin for herring sampled from four focal spawning aggregations within the NGA, and two outside the NGA (Sitka and Togiak).  Specifically, our objectives were to 1) determine whether EA or FAA will allow discrimination among Alaska’s three major herring stocks (Sitka, NGA, and Togiak); and if so, 2) determine whether EA or FAA can detect finer scale structuring of putative herring stocks inside PWS and elsewhere in the NGA.

In this pilot study, we maximized the resolution of these techniques by limiting our analysis to pre-spawning female herring aged 4 to 7.  Given our expectation that Kamishak and Kodiak herring resided within their respective natal spawning areas for only a short period, and probably occupied the same over-wintering areas, we focused our EA on the microchemistry of the otolith accreted prior to the first annulus.  Otoliths were examined between the focus and the first annulus for the presence of nine elements- Na, Mg, P, S, Cl, K, Ca, Fe, and Sr. Lipids were extracted from the hearts of the same herring and examined for the presence of 39 fatty acids.

Differences in the chemical composition of tissues from different groups were examined by multivariate analysis of variance (MANOVA), univariate ANOVA, and discriminant function analysis (DFA). In addition, we cross-validated the DFA results using the leave-one-out method (Huberty 1994) to estimate the reliability of each stock identification method.

The elemental compositions of herring otoliths from Sitka, Togiak and NGA were significantly different (MANOVA; P < 0.0001).  DFA indicated that elemental analysis could be used to reliably separate herring from Togiak and Sitka, however, there appeared to be little discriminating power between Sitka and NGA or Togiak and NGA.  On a fine spatial scale, EA provided detectable differences among NGA stocks (P < 0.0001), but DFA indicated the existence of two groups (Kodiak/PWS-Montague and Kamishak/PWS-NE) that did not conform to our a priori stock designations.  More detailed analysis indicated the group designations determined by EA derived from the Na and Cl content of the otoliths (ANOVA; P < 0.004).  . 
           

Our results suggest that fatty acid analysis of heart lipids may be a reliable method for discriminating herring stocks at fine spatial scales (i.e., < 100 km).  Analysis by MANOVA revealed significant differences between the mean fatty acid compositions of heart lipids from Sitka, NGA and Togiak herring (P < 0.0001).  DFA revealed that the fatty acid data could be used to correctly identify unknown samples from these stocks more than 94% of the time.  Heart lipids from the four NGA stocks- Kodiak, Kamishak, PWS-Montague and PWS-NE, were also easily discriminated by their fatty acid compositions.  MANOVA results indicated significant differences among the average fatty acid compositions of the four stocks (P < 0.0001).  Using DFA, erroneous stock identifications occurred less than 10% of the time and stock specific error rates ranged between 8% and 12%. Unlike the elemental data, our a priori stock identities appeared to best describe the fatty acid data structure.

Although these results are quite promising, follow up work is necessary in order to verify the utility of fatty acid analysis for discriminating herring stocks.  This follow up work should include all members of each population (e.g., all cohorts, sexes, and maturity levels) and aim to satisfy Galvin et al.’s (1995) third criterion for successful application of a mixed stock analysis (MSA) technique by establishing the temporal stability of the stock specific chemical signatures we identified in our pilot study.  The potential value of a tool for discriminating herring stocks over fine spatial scales cannot be overestimated.  In the near term, this approach can be used to resolve a number of pressing management questions regarding stock structure in Prince William Sound, Kodiak/Kamishak, Togiak/Dutch Harbor, and in southeastern Alaska.  Ultimately, the ability to identify the stock of origin for herring collected away from their natal spawning areas would provide a basis for understanding larval dispersal patterns, home ranges of individual populations, locations of stock specific over-wintering areas, and perhaps the degree to which Pacific herring home back to their natal spawning areas.

If you would like to read the full final report for this pilot study: Full Report (IN PDF FORMAT)