Knowledge of the diet composition of a fish species at a specific locality is useful in assessing its ecological function and impact, for the construction of ecosystem models (see Box 21), and to help define the nutritional requirements of potential aquaculture species. In FishBase, data in the DIET table are also used to estimate the trophic level of species (see Box 23).
On the other hand, most demersal fish have developed various specialized methods of food gathering. Those ‘browsing on the substrate’, ‘sucking food-containing materials’ or ‘grazing on aquatic plants’ often live near the bottom and have developed specialized morphologies adapted to this (see e.g., de Groot 1984 for flatfishes). More specialized feeding techniques are used by those fish that depend on other organisms to feed, i.e., the parasites, commensals, cleaners and scavengers. Fish with ‘variable’ feeding types also exist (see for example Tiews et al. 1972 on the feeding habits of leiognathids). The choice ‘other’ is provided for fish with specialized food gathering habits not in the choice list; in such cases, the specific feeding type is indicated in the Remarks field.
There is a huge number of references in the literature which provide information on the frequency of occurrence of food items in fish stomachs, which some readers may view as providing useful data on diet compositions. However, except perhaps in fish larvae, whose food items are all uniformly small, frequency of occurrence is not a good indicator of how much a food item contributes to the diet of a given population. For example, a small copepod that occurs in 50% of the examined stomachs may contribute much less to the diet than large polychaetes that are found in only 40% of the stomachs. The many indices applied to frequency of occurrence data do not remedy this basic flaw and rather confuse the topic. Editors and referees should reject submitted manuscripts dealing with stomach contents that do not present diet data in terms of weight, volume or energy.
We have limited our entries to those quantitative reports which do not suffer from the flaw described above. Records entered in this table deal only with studies on the stomach content of fish as they occur in the wild, and not under experimental conditions. Thus, most of the information entered in the DIET table was obtained from relatively few (>460) references, notably Stevens (1966), Randall (1967), Hobson (1974), Armstrong (1982), Sano et al. (1984), Randall (1985), Gonzalez and Soto (1988), Laroche (1982), Sierra et al. (1994) and Valtysson (1995).
Diet composition data have been compiled for more than 1,400 species. We would like to have diet data for as many finfish species as possible, and would appreciate reprints for species that we have missed so far.
The taxonomic classifications of the food items of the more than 3,000 records for over 1,400 species with diet compositions were checked against the Taxonomic Code of Hardy (1993), the Taxonomic Authority List of the Aquatic Sciences and Fisheries Information System (de Luca 1988) and Barnes (1980). Inconsistencies may arise in the functional classification of some animal food items. We have tried to reduce as much as possible the inconsistencies resulting from this by inferring the functional group of a food item from the habitat and behavior of the species that consumed it, but have probably failed to resolve all of them.
The DIET table consists of the following fields:
The Stage of fish sampled (choice) field has 4 options, i.e., larvae; recruits or juveniles (recruits/juv.); juveniles and adults (juv./adults) which is the default option for cases where the life stage is not specified); adults.
The Mean length field refers to the average length of fish in the sample in centimeters and is coupled with a length type field. The Number of fish sampled refers to the individual fish specimens; the percentage with empty stomachs is stated whenever available.
The Locality field refers to the specific site where the study was undertaken, further identified by the Country field before it.
The Months covered by the study, which appear as highlighted fields define the period of the year when the samples were obtained. Such information can help interpret the presence or abundance of specific food items in the habitat.
The Remarks field is used for ancillary information required if the option ‘other’ in choice fields within the table has been clicked or, for information which may further explain and/or describe particular food items.
Fig. 37. Diet composition, in % volume or weight of Oreochromis niloticus niloticus in Lake Awasa, Ethiopia. Further
breakdown of the given categories is available in FishBase.
To accommodate the range of information found in the literature, food items are classified in three categories, from very general groupings in Food I to taxonomic groups in Food III (see FOOD ITEMS table and Box 24 for details on the hierarchy). Finally, the species name of the food item and/or other information can be seen in a text field by clicking on More button (if highlighted). For recomputed volumes, the More button also shows the original % in which a food item appears in the diet.
The Prey stage field refers to the life stage of a prey, i.e., eggs; larvae or pupae; recruits/juv.; juv./adults; adults; or to a specific part of a plant food, i.e., roots; stem; leaves/blades; fruits/seeds. The ‘n.a./others’ option is provided for cases in which the life stage is not stated by the source (and cannot be deduced), or when several stages are consumed.
The % diet (numeric) field refers to the percent weight or volume contributed by a food item to the stomach content of a fish; the percentages of the various items must add up to 100%, which is ascertained by a calculated field. Unidentified items in the diet are excluded (see Other items in Ref. field) and the contributions to the diet of all identified items are readjusted to bring the total back to 100%. The percentages can also be viewed in form of a pie chart (see Fig. 37).
The troph estimated from a diet composition (and from the trophs of the food items; see the FOOD ITEMS table) is displayed in a computed field, along with its standard error.
We used the over 3,000 records in the DIET table to estimate the typical contribution of various food items to the diet of fish, if they were the main food, i.e., had rank 1 in a diet study based on percent contribution to total stomach content in weight or volume. The results are shown in Table 2 and Fig. 38. As can be seen from the pie chart the most common main foods of fish are animals, mainly other fish, and benthic and pelagic crustaceans. Table 2 shows the typical (here: median) contribution of functional food groups I and II to the diet when they are the main food. We plan to turn this into a software that will add percentages to food items if the ranking of their contribution (but not the actual percentage) is indicated in the literature, as is often the case. That software will also consider the typical contributions of food items if they occupy, e.g., rank 2 or 3 of a certain number of reported food items. At the beginning of the DIET chapter, we explained why diet studies expressed in frequency of occurrence are rather useless. However, if we can show that the ranking of the frequency of food items in stomachs is the same as the ranking of their contribution in percent volume or weight, then this software can make the huge ‘frequency of occurrence’ literature useful for food web studies. See also Box 24 which explains how we use a Monte Carlo routine to estimate trophic level from food items for which no ranking is known.
Maria Lourdes D. Palomares and Rainer Froese
Table 2. Median contribution to stomach contents (% volume or % weight), if the respective item was the dominant food, i.e., first item in the diet composition record; 2,420 records were obtained from the DIETS table and only functional groups with more than 10 records as dominant food were used in this analysis.
