Fish food habits data summary system

Background

Currently, our research group collects stomachs for later laboratory analysis while conducting two major fishery independent trawl programs. As described on other pages (ChesMMAP Info, NEAMAP info, NEAMAP Page at ASMFC ) ChesMMAP and NEAMAP conduct operations in Chesapeake Bay, and in the nearshore Atlantic Ocean between Cape Cod and Cape Hatteras, respectively. 

The data base you are about to access is based on analyses of approximately 110,000 fish stomachs (as of 2018), about 30% ChesMMAP and 70% NEAMAP.  New specimens are added to the group’s data base nearly every day.

Results you will select are not based on accessing this ‘raw’ data base of individual specimen examinations. Instead, you will select data that result from specific analyses and summarizations of the raw data.  Data are offered in this way because of the complex sampling, subsampling, and analysis methods employed on these surveys.  Specifically, we sample stations according to a stratified random sampling regime, we subsample specimens for stomach analysis using methods to assure adequate sampling across all sizes, and we analyze the results based on an assumption of ‘cluster sampling’ as explained below and as described by Buckel (1999). All of these factors must be accounted for when analyzing these raw data.  So, to provide users with the quickest response times, the data base you will access represents pre-summarized data. Full descriptions of collection and data analysis methodologies are explained in our project reports (ChesMMAP Reports and NEAMAP Reports).

Cluster Sampling

It is important to understand the cluster sampling analysis methods mentioned above. This concept assumes that fish are not randomly distributed throughout their habitat but concentrate themselves in time and space according to size groups, habitat types, prey concentrations, and other factors. It further assumes that the specimens you sample at one location are more likely to be similar to each other than they would be to specimens sampled at another location or at a different time. Therefore, traditional summarization of data based on an assumption of a random distribution would be inappropriate.  Let’s look at a hypothetical, though potentially very real, example. Assume we sampled striped bass in two locations.  At the first we captured five fish and at the second we caught 100. At both locations our subsampling protocols state that we will examine the stomachs of five specimens. If we were to simply add up the weights (or numbers) of prey in the stomachs of each group of five we would not be giving proper consideration to the fact that at one site we caught 20 times the number of fish that we did at the other. The same concept holds for different sizes of fish captured within and among sampling stations.  On a population level, we believe that we must account for such differences in order to properly describe the food habits of the fishes we are sampling. A real example can help illustrate the differences in analytical methods and why they are important:

Our overall ChesMMAP (i.e. all ages, all locations, all years) diet summaries for striped bass show that about 9% of the diet by weight consists of Atlantic menhaden. This differs substantially from other food habits studies of striped bass in Chesapeake Bay (Griffin et. al 2003, Hartman and 1995, Walter et. al 2003). If we were to treat our data without regard to the cluster sampling assumptions (i.e. simply add up the total weight of menhaden in striped bass stomachs and divide by the total weight of all prey items), which is the traditional way that diet studies have analyzed their data, menhaden would contribute about 40% by weight to the diet of striped bass (other factors such as sampling methods, times, and locations likely account for this specific difference as well).

As previously stated, under these cluster sampling assumptions, fish captured together are assumed to be more similar to one another than they would be to fish captured elsewhere or at a different time. In effect, the cluster becomes the sampling unit rather than the individual specimen. For this reason, when reporting on sample size, it is more appropriate to report the number of clusters sampled than the number of individual fish stomachs examined. Both figures are provided on the data reports you will obtain.

Predator-Centered Data Retrieval

The data base offers you the opportunity to choose to retrieve fish food habits analyses from either a predator or prey point of view.  If you choose to view data from the predator’s perspective you can select data by Survey, Predator Species, Year, Age, and State, summarized by either prey weight or prey numbers (we do not currently offer online summaries by frequency of occurrence though such data are available by request). Once you enter the site you will see the following screen:

Within the “Predator-Centered Data Retrieval” simply make your choices within each drop-down box, hit the submit button, and you will be presented with a food habits report that looks like this:

In the report, data for prey types are presented first by major taxonomic group (e.g. fishes, crustaceans, molluscs, worms, etc.) in decreasing group order by percent consumed, and then within each major group by specific prey items, also in decreasing order.  In some cases, prey types at the prey item level are grouped, even if more detailed data may be available in our raw data base. For example, data for all amphipod species are combined in these reports even if species-specific data may be available. More detailed summaries may be obtained upon request. For each major group, only specific prey items which represent 1.0% or more of the diet are presented individually. All prey types that are less than 1.0% are grouped into an ‘xxxxxxxx – other’ category where ‘xxxxxxxx’ represents the taxonomic group.

A few conditions in the predator-based data retrieval system require explanation:

• The first choice you will make when selecting your data will be to pick from which survey you wish to see data (ChesMMAP or NEAMAP). After that, you will select a predator species, then, you can make choices for Year, Age, and State if such data are available for the species of interest. You must also choose whether you want data summarized by prey weight or prey number. Choices in the drop-down boxes are linked such that values presented in each field are filtered based on your choices in previous fields (this fixes a bug that existed in earlier versions of the data base). However, you must make your choices for Year, Age, and State in that order so that the filtering works as expected.
• Similarly, when making selections for Year, Age, and State, at least one of these choices must be ‘All’ or no data will be returned. While in many cases data may exist in our raw data base to allow such specific analyses to be accomplished, in general the sample sizes would be too small to offer reliable estimates of diet.
• In the drop-down list for predator Ages you may see a figure of -1. This is a special circumstance which only exists in the data base for Atlantic croaker. It results from differences in spawning patterns in the northern (Chesapeake Bay) portion of the species’ range which make it possible for us to capture specimens in the fall of the year preceding their official birth date of 1 January. In our data bases these specimens are assigned an age of -1 so that the proper year class can be calculated.

Prey-Centered Data Retrieval

The choices available to view data from the prey perspective are much simpler. The only menu choice available is to select the prey species for which you want to retrieve data. The data report is also simplified, listing all predator species in which the prey appears and the percent of the overall diet, both by number and weight (for All Years, All Ages, All States), separately for each Survey. Sample size measurements (from the predator side) are also listed. Percentages listed in this report DO NOT add up in any meaningful way. Researchers are encouraged to use the information on this report to explore more deeply in the predator-based query system.

Updates

As explained above, data are being added to our raw food habits data base on a daily basis. We will update this online version of our data periodically (2-4 times per year) and without notice. Therefore, results that you download today may differ somewhat from results using the same menu choices that you obtain in the future. You should make note of the date listed at the bottom of the data report.

Future Improvements

Currently, the application only allows users to view or print out the data report. You may be able to select the report text and copy and paste into an application such as MS Word but formatting changes are likely to make the resulting document difficult to view. If there is sufficient demand we may make results available in the future in spreadsheet or other more copy-able form.

Citation

We ask that all uses of this data base be given proper credit and citation. For scientific publications please cite as follows:

VIMS Multispecies Research Group. Year, Month, Day. Fish food habits online data base.  Retrieved from https://www.vims.edu/fisheries/fishfood

Disclaimer

When you have finished reading this page, to proceed to the data base, please click the link below. This will take you to an email page where we ask that you send us just a bit of information about yourself and your intended use of these data. This will allow us to keep track of the uses and users of our data.

Data included in this data base have undergone routine and extensive QA/QC and care has been taken to limit the number of errors present. However, it is likely that some number of identification and/or coding errors do exist. Users should use care in interpreting these data. VIMS and the Multispecies Research Group take no responsibility for the users’ use or misuse of these data.