Breaking with Tradition: Redefining Measures for Diet Description with a Case Study of the Aleutian Skate Bathyraja Aleutica (Gilbert 1896)
Total Page:16
File Type:pdf, Size:1020Kb
Environ Biol Fish (2012) 95:3–20 DOI 10.1007/s10641-011-9959-z Breaking with tradition: redefining measures for diet description with a case study of the Aleutian skate Bathyraja aleutica (Gilbert 1896) Simon C. Brown & Joseph J. Bizzarro & Gregor M. Cailliet & David A. Ebert Received: 1 September 2010 /Accepted: 27 October 2011 /Published online: 16 November 2011 # Springer Science+Business Media B.V. 2011 Abstract Characterization of fish diets from stomach Aleutian skate Bathyraja aleutica from specimens content analysis commonly involves the calculation of collected from three ecoregions of the northern Gulf multiple relative measures of prey quantity (%N,%W,% of Alaska (GOA) continental shelf during June- FO), and their combination in the standardized Index of September 2005–2007. Aleutian skate were found to Relative Importance (%IRI). Examining the underlying primarily consume the commonly abundant benthic structure of dietary data matrices reveals interdependen- crustaceans, northern pink shrimp Pandalus eous and cies among diet measures, and obviates the advantageous Tanner crab Chionoecetes bairdi, and secondarily use of underused prey-specific measures to diet charac- consume various teleost fishes. Multivariate variance terization. With these interdependencies clearly realized partitioning by Redundancy Analysis revealed spa- as formal mathematical expressions, we proceed to tially driven differences in the diet to be as influential isolate algebraically, the inherent bias in %IRI, and as skate size, sex, and depth of capture. Euphausiids provide a correction for it by substituting traditional and other mid-water prey in the diet were strongly measures with prey-specific measures. The resultant new associated with the Shelikof Strait region during 2007 index, the Prey-Specific Index of Relative Importance (% that may be explained by atypical marine climate PSIRI), is introduced and recommended to replace %IRI conditions during that year. for its demonstrated more balanced treatment of the relative measures of prey quantity, and less erroneous Keywords Skate . Bathyraja aleutica . Diet . Gulf behavior across taxonomic levels of identified prey. As a of Alaska . Food habits . Index of relative importance case study, %PSIRI was used to examine the diet of the Introduction S. C. Brown (*) : J. J. Bizzarro : G. M. Cailliet : D. A. Ebert Pacific Shark Research Center, Diet information is crucial to our understanding of Moss Landing Marine Laboratories, species ecology, trophic interrelationships, food webs, 8272 Moss Landing Road, and ultimately, the flow of energy through ecosys- Moss Landing, CA 95039, USA e-mail: [email protected] tems. Stomach content analysis remains a universal technique for sampling the diets of fishes and these Present Address: studies contribute large amounts of species-specific J. J. Bizzarro diet data for potential use in trophic ecosystem School of Aquatic and Fishery Sciences, University of Washington, modeling that provide ecosystem-based fishery man- PO Box 355020, Seattle, WA 98195-5020, USA agement advice (Ainsworth et al. 2010). Although 4 Environ Biol Fish (2012) 95:3–20 there has been an arguably successful call (294 large skate, reaching a maximum size of 150 cm total citations in Web of Science) for consistency in length (TL). Diet data reported from the western reporting diet compositions of elasmobranchs with North Pacific and Aleutian Islands indicate that percent number (%N), percent weight (%W), frequency Aleutian skates consume primarily decapod crusta- of occurrence (%FO), and the standardized Index of ceans, with fishes and cephalopods also represented Relative Importance or %IRI (Cortès 1997), there (Orlov 1998; Yang 2007); however, scant dietary remain critical unresolved weaknesses in this widely information is available for the Aleutian skates in the used and accepted methodology (Cortès 1998; northern GOA. The findings of this study represent Hansson 1998). There are, in fact, not only weak- the first detailed trophic information on the Aleutian nesses, but also serious unrecognized mathematical skate population from the northern GOA shelf flaws in the presentation of diet data both graphically ecosystem. and by indices that have gone largely unnoticed in the To improve current data reporting methods for published literature. These methodological problems stomach contents analysis of elasmobranchs, our are a direct consequence of diet researchers’ incorrect objectives for this study are to: 1) define the structure mathematical understanding of diet measures. A of diet data; 2) elucidate mathematical relationships deeper understanding of the structure of dietary data between diet measures; 3) demonstrate how appropri- and the resulting mathematical relationships between ate operationalization of these relationships can diet measures not only resolves current methodolog- correct current flaws in compound diet indices and ical weaknesses, but also leads to increased extraction graphical displays and; 4) illustrate the application of of information about food habits in graphical displays this approach in reporting the diet composition on and diet composition tables. dietary data from the Aleutian skate. Additionally, we Proper metrics are necessary to characterize diets of provide an in-depth statistical analysis of ontogenetic, abundant, data-poor species, to avoid faulty or incom- regional, and interannual variation in the diet of the plete conclusions about their trophic roles. Skates as Aleutian skate from the northern GOA ecosystem abundant mesopredators (Ebert and Bizzarro 2007) during 2005–2007. likely play important trophic roles in demersal fish communities and may be able to overtake the resource The structure of diet data niche left open by depleted teleost stocks (Stevens et al. 2000; Link and Sosebee 2008). For large marine Stomach content analysis typically incorporates meas- ecosystems with large industrial fisheries, like the urements of numerical abundance (i.e. aggregate counts northern Gulf of Alaska (GOA) shelf, attaining of individual prey items in each designated prey trophic information on skates is beneficial for consid- category), gravimetric (or volumetric) abundance (i.e. eration in ecosystem based modeling and manage- aggregate weights or volumes of prey items in each ment. In the GOA, skates contribute substantially to designated prey category), and the frequency of occur- the bycatch (retained and discarded) of directed rence of prey categories among all stomach samples fisheries such as those for Pacific Halibut Hippo- (Hyslop 1980). The resulting diet data matrix, by glossus stenolepis and other commercially valuable numerical abundance or biomass, is composed of groundfish species (Ormseth and Matta 2009), but columns of prey categories (i) by rows of individual their ecology is poorly understood. stomach samples (j) standardized to proportion by Skates of the genus Bathyraja are widely distrib- total individual stomach content (i.e. by row): uted in the North Pacific. In the GOA the Aleutian skate Bathyraja aleutica (Gilbert 1896) is one of the 2 prey category ðiÞ 3 most abundant skate species (Ormseth and Matta 10 0 0 6 : : : 7 2009). The Aleutian skate ranges from the northern ð Þ 6 0025 0 5025 7 stomach sample j 4 : : 5 Sea of Japan into the Sea of Okhotsk and Bering Sea 000505 00:75 0 0:25 to the eastern GOA, but has been reported as far south as Cape Mendocino, California, U.S.A. (Hoff 2002), typically occupying shelf and slope depths of 100– This stomach sample by prey item diet matrix 800 m (Mecklenburg et al. 2002). It is a relatively is well-suited for multivariate statistical analysis Environ Biol Fish (2012) 95:3–20 5 (de Crespin de Billy et al. 2000), but also conve- of %N or %W cannot exceed its corresponding %FO niently serves in calculation of diet measures. The value for any prey category. The determined value of % average percent number (%N) and average percent FO represents an upper limit to %N and %W values weight (%W) for each prey item are column averages because discrete absences are averaged into all of this matrix and are additive, meaning that they sum measures, this creates a mathematical dependence to 100% for all different prey items of a prey between diet measures whose strength increases with category. For example, in the dietary data matrix the increasing frequency of zero values in a diet data provided above, the average percent value for each matrix. Amundsen et al. (1996), fully realizing this prey category (i.e. each column) is 25%, and the graphical limitation of diet measures in constructing aggregate values for all prey categories sum to 100%. feeding strategy diagrams (Costello 1990), proposed a The Percent Frequency of Occurrence (%FO) can new measure termed prey-specific abundance. Prey- be calculated as the column averages from the same specific abundance is defined as the percent numerical diet matrix, but translated to discrete conditions of abundance of a prey item averaged over the stomach presence/absence (i.e. binary data). From the example samples in which it occurs (i.e. excluding zero diet matrix above, the %FO for each prey category values). Like %FO, the value of the prey-specific (from left to right) is: 25%, 50%, 50%, and 75% and abundance for a prey item may take any value >0% to sums to an indeterminate value which, is 200% in this 100% independent of the values for all other prey example. Because different prey categories occur items. For example, from the diet data matrix