When possible, ¢sh were sampled during the rice
crop. As such we obtained SGRs for di¡erent developments
of the rice crop: vegetative stage (0^40 DAS),
reproductive stage (41^70 DAS), ripening stage
(71^100 DAS), fallow stage (before rice seeding and
after rice harvest) and ratoon stage (after wet season
rice harvest). However, not enough ¢sh could be
sampled very often, as large number of ¢sh forage in
the rice ¢eld where they cannot be caught. Nevertheless,
we were able to increase the data set to 202
entries for the C. carpio multiple regression. An
overview of the dependent and all independent variables
for the SGR and survival rate multiple regression
can be found inTables 3 and 4 respectively.
The variable‘extra feed’stood for a dailyapplication
(at a rate of 5% of the total ¢sh biomass) of a mixture
of ¢ne rice bran (40%), rough rice bran (30%) and
freshwater spinach (Ipomoea aquatica Forsk.) (30%) to
the trench. The variable ‘pig manure’ stood for the
number of pigs that provided manure per plot. The
variable ‘inorganic fertilizer’ stood for an application
of inorganic fertilizer to the trench at a rate of
7 kgNha1 and 10 kg P2O5 ha1, once every 2
weeks.The variables rice vegetative phase, rice reproductive
phase, rice-ripening phase, trench and ratoon
indicate to which period of the rice^¢sh cycle
each data entry belongs. Some entries span more
than one period. For the SGR multiple regressions,
we used the C. carpioweight as recorded at the beginning
of each period (Costa-Pierce, van Dam& Kapeleta1993).
For the survival-rate multiple regression, we
used the average ¢sh weight at stocking.The wild ¢sh