TrophLab (version of June 2000)
Introduction
Trophic levels (here
abbreviated to ‘troph’, to avoid overlap with ‘TL’, used for total length),
express where fish and other organisms tend to operate in their respective food
webs. To estimate the trophs of fish, we must consider both their diet
composition, and the trophs of their food item(s). The troph of a given group
of fish (individuals, population, species) is then estimated from
Troph = 1 + mean troph of the food items …1)
where the mean is weighted
by the contribution of the different food items.
This can be expressed more formally by
…2)
where Dcij is the fraction of prey (i) in the diet of consumer (i), troph j is the trophic level of j, and G is the number of groups in the diet of i.
Following a convention
established in the 1960s by the International Biological Program, we attribute
primary producers and detritus (including associated bacteria) a definitional
troph of 1 (Matthews 1993).
Thus for example, an anchovy
whose diet would consist of 50% phytoplankton (troph = 1) and 50% herbivorous
zooplankton (troph = 2) would have a troph of 2.5. The last value is an
estimated, fractional troph, differing conceptually and numerically from the
integer values that are often assumed for higher trophs, and which we think are
too imprecise and inaccurate to be useful in any kind of analyses.
An omnivore is a “species which
feeds on more than one trophic level” (Pimm 1982). Thus, an omnivory index
(O.I.) can be derived from the variance of the trophs of a consumer’s food
groups i.e., from
...
3)
The O.I. takes values of
zero when all feeding occurs at the same troph, and increases with the variety
of food items’ trophs. Its square root is a standard error, i.e., (Christensen and
Pauly 1992).
Routines for estimation of
trophs and O.I. values are incorporated in the Ecopath software, which has been
applied to a large number of ecosystems (see Pauly and Christensen 1995; Pauly
et al. 1998). Troph estimates from Ecopath i.e., from diet compositions and
food web studies have been found to correlate closely with troph estimates
based on stable isotope ratios of Nitrogen (Kline and Pauly 1998; Pauly et al.
2000). This has led to numerous troph estimates for a wide range of taxa
becoming available, notably for the invertebrates, fish, marine mammals and
other groups covered by FAO statistics, and now included in FishBase. However,
a tool was lacking so far which would allow estimation of trophs outside of
Ecopath applications.
To remedy this situation, TrophLab
was developed as a stand-alone application for estimating the troph of any
consumer, given quantitative or qualitative information on the composition of
its food.
TrophLab has two basic routines:
1) Estimation of troph (and their s.e.) from quantitative diet composition data; and
2) Estimation of approximate trophs (and their s.e.) from list of items known to occur in the diet of an animal, i.e., from qualitative information.
Quantitative
data
The routine in (1) estimates trophs (and s.e.) from equation (1) given the percentage (in weight, volume or energy) of each item in the diet of the organisms in question, and the trophic level of their preys. In case such trophic levels are missing, default trophs (and s.e.) can be taken from the table below. [Note that this table was optimized for fish diets, and may not well cover other, e.g., terrestrial predators.]
The routine in (2) rests on the idea that it is possible to obtain a rough estimate of a species troph and its s.e. based on individual prey items (rather than a complete diet composition) granted that enough food groups or items are known for that species, and that one is willing to accept certain assumptions on the relative importance of these food items in the overall diet of the species.
Examination of diet compositions entered into FishBase until mid 1999 (n = >1,800) showed that, typically, the relative contribution of different food groups or items to the overall diet composition follows a pattern described by the equation
log10P = 2 – 1.9log10R – 0.16log10G …4)
where P is the contribution of an item to the total diet in percent; R is the rank of the food item (in terms of its relative contribution to the total diet); and G is the number of food groups or items (Note that the equation is defined only for 1<G<10).
In the following, a description of the resampling routine is provided which is used in TrophLab to estimate trophs and their s.e. from individual food items. This routine involves three cases:
Case 1: all food items are plants or detritus
Then: troph = 2.0 and s.e. = 0;
Case 2: there is only one food item, and it is neither a plant nor detritus.
Then: troph = 1 + troph of food item & s.e. = s.e. of food item (see Food Items Table for trophic levels and s.e. of food items; use Food III if possible, else Food II, else Food I).
Case 3: There are several food items, and at least one is not a plant or detritus.
Then: run Subroutine A.
1) Count the food items, and call their number G;
2) Select at random one of these food items, and give it the rank 1 (R = 1);
3) Given G, and R, solve equation (1) for P;
4) Select at random one of the remaining food items, give it a rank of 2 (R = 2) and again solve equation (1) for P;
5) Repeat (2) – (4) until all items have been selected (R = 3, 4 . . . . G);
6) From the P values, and the trophs specific to each items, estimate a mean troph from:
…
2)
7) Compute s.e. of
troph from Sachs (1984)
…
3)
7) Save troph and s.e.; repeat (2) – (8), using different random numbers to select first, second, etc. item; stop after 100 loops.
8) Take grand mean of computed trophs and of their standard, errors output these and stop.
The key point of this subroutine is that the grand mean s.e. that is estimated considers all possible permutations of the food items in term of the relative abundance they could have had in a ‘real’ diet composition. Note that the standard errors and corresponding troph estimates obtained from this routine are tentative, and should be replaced by estimates from quantitative diet compositions whenever possible.
Often, data exist that are intermediate in nature between the quantitative data required for the routine in (1) and the qualitative data analyzed in (2). Example is information that a given food items is “very abundant” or “dominant” in the diet, while others “occur occasionally” or are “infrequent”.
We recommend for such case to either attempt to turn such information into numbers (see Table 2) and then to use the quantitative routine in (1) or to enter the abundant or dominant food item several times into the file required for the qualitative routine in (2), such that it will be given more weight in the analysis. This process can be rendered less subjective by splitting the relatively more abundant groups in the diet (as indicated by qualitative statements) into its component taxa e.g., by splitting abundant ‘fishes’ into, say, 2 species of common prey fishes that are demersal, and one that is a small pelagic, or conversely, given the known habits of the predator.
Table 1. Default troph values
used by TrophLab for various prey (arranged by 3 levels of aggregation). (Based
on data in FishBase; Froese and Pauly 1999). Note that TrophLab allows this
values to be overwritten when better estimates are available.
Food I |
Troph |
s.e. |
Food II |
Troph |
s.e. |
Food III |
Troph |
s.e. |
detritus |
1.00 |
0.00 |
detritus |
1.00 |
0.00 |
debris; carcasses |
1.00 |
0.00 |
plants |
1.00 |
0.00 |
phytoplankton |
1.00 |
0.00 |
blue-green algae; diatoms;
etc. |
1.00 |
0.00 |
|
|
|
other
plants |
1.00 |
0.00 |
benthic algae/weeds;
periphyton; terrestrial plants |
1.00 |
0.00 |
zoobenthos |
2.50 |
0.50 |
sponges/tunicates |
2.00 |
0.00 |
sponges; ascidians |
2.00 |
0.00 |
|
|
|
cnidarians |
2.50 |
0.52 |
hard corals; n.a./other
polyps |
2.34 |
0.61 |
|
|
|
worms |
2.06 |
0.26 |
polychaetes; n.a./other
annelids; non-annelids |
2.06 |
0.26 |
|
|
|
mollusks |
2.80 |
0.46 |
chitons bivalves gastropods octopi n.a./other mollusks |
2.38 2.10 2.37 3.50 2.60 |
0.51 0.3 0.58 0.51 0.5 |
|
|
|
benthic crustaceans |
2.50 |
0.50 |
ostracods benthic copepods isopods amphipods stomatopods |
2.50 2.00 2.29 2.29 3.09 |
0.61 0.00 0.53 0.53 0.53 |
|
|
|
|
|
|
shrimps/prawns lobsters crabs n.a./other benthic
crustaceans |
2.60 3.20 2.50 2.50 |
0.59 0.41 0.60 0.50 |
|
|
|
insects |
2.10 |
0.40 |
insects |
2.20 |
0.40 |
|
|
|
echinoderms |
2.40 |
0.35 |
sea stars/brittle stars sea urchins sea cucumbers |
3.10 2.00 2.00 |
0.60 0.00 0.00 |
|
|
|
|
|
|
n.a./other echinoderms |
2.40 |
0.35 |
|
|
|
other
benthic invertebrates |
2.50 |
0.43 |
n.a./other benthic
invertebrates |
2.50 |
0.37 |
zooplankt. |
2.10 |
0.28 |
jelly
fish/hydroids |
3 |
0.28 |
jellyfish/hydroids |
3.00 |
0.28 |
|
|
|
planktonic crustaceans |
2.1 |
0.3 |
planktonic copepods cladocerans mysids euphausiids |
2.00 2.00 2.20 2.20 |
0.00 0.00 0.40 0.40 |
|
|
|
|
|
|
n.a./other planktonic
crustaceans |
2.10 |
0.30 |
|
|
|
other planktonic |
2.2 |
0.17 |
n.a./other
planktonic invertebrates |
2.40 |
0.45 |
|
|
|
invertebrates |
2.2 |
0.4 |
|
|
|
|
|
|
fish
(early stages) |
3.5 |
0.8 |
fish eggs/larvae |
3.50 |
0.80 |
nekton |
3.50 |
0.60 |
cephalopods |
3.5 |
0.37 |
squids/cuttlefish |
3.50 |
0.37 |
|
|
|
finfish |
3.5 |
0.8 |
bony fish |
3.50 |
0.80 |
|
|
|
|
|
|
n.a./other finfish |
3.50 |
0.80 |
others |
2.40 |
0.50 |
herps |
2.6 |
0.52 |
salamanders/newts toads/frogs turtles n.a./other reptiles |
2.60 2.58 2.10 3.00 |
0.68 0.68 0.30 0.30 |
|
|
|
birds |
3.6 |
0.62 |
sea birds shore birds n.a./other birds |
3.77 3.36 3.56 |
0.34 0.81 0.58 |
|
|
|
mammals |
4.1 |
0.03 |
dolphins pinnipeds n.a./other mammals |
4.18 3.97 4.00 |
0.02 0.04 0.50 |
|
|
|
others |
2.4 |
0.44 |
n.a./others |
1.50 |
0.50 |
Table 2. Suggested percentage corresponding to words used for describing relative contribution to diet composition.
Word(s) |
Approx. % |
Rationale |
“Almost completely” |
>95 |
As in parametric statistics |
“Dominant” |
80 |
Based on the 80/20 rule |
“Mostly” |
50-60 |
--- |
“Fairly Common” |
50 |
As for bird sightings |
“Frequent” |
20-50 |
--- |
“Occasional” |
5-20 |
--- |
“Rare” |
<5 |
As in parametric statistics |
References
Christensen, V. and D. Pauly. 1992. ECOPATH II – a
software for balancing steady-state ecosystem models and calculating network
characteristics. ICLARM, Manila.
Kline, T. and D. Pauly. 1998. Cross-validation of
trophic level estimates from a mass-balance model of Prince William Sound using
15N/14N data. In
Fishery stock assessment models. Alaska Sea Grant College Program. AK-SG-98-01.
Mathews, C.P. 1993. Productivity and energy flows at
all trophic levels in the River Thames, England: Mark 2, p. 161-171. In V. Christensen and D. Pauly (eds.)
Trophic models of aquatic ecosystems. ICLARM Conf. Proc. 26. 390 p.
Pauly, D and V. Christensen. 1995. Primary production
required to sustain global fisheries. Nature 374:255-257.
Pauly, D., V. Christensen, J. Dalsgaard, R. Froese
and F. Torres, Jr. 1998. Fishing down the food webs. Science 279:860-863.
Pauly, D., V. Christensen, R. Froese and M.L.
Palomares. 2000. Fishing down aquatic food webs. American Scientist
88(Jan.-Feb.):46-51.
Pimm, S.
1982. Food webs. Chapman and Hall, London and New York. 219 p.
Sachs, L. 1984. Angewandte Statistik. [Applied statistics, 6th Edition]. Springer-Verlag, Berlin. 552 p.
Daniel Pauly,
Rainer Froese, Pascualita Sa-a, Maria Lourdes Palomares, Villy Christensen and
Josephine Rius