R. Nortvedt et S. Tuene, MULTIVARIATE EVALUATION OF FEED FOR ATLANTIC HALIBUT, Chemometrics and intelligent laboratory systems, 29(2), 1995, pp. 271-282
The present investigation tested three physical properties of fish fee
d: dry matter content, floating vs. sinking feed, pellet size and thei
r cross-terms upon growth and feed utilization of Atlantic halibut (Hi
ppoglossus hippoglossus) in the ongrowing phase. A factorial design (F
D, 2(3)) with four center points was applied. Each of three size subgr
oups within each rearing tank were modeled across 12 tanks. Nine singl
e growth and feed utilization responses and three combined responses (
super Y's) were modeled with the aid of PLS (partial least squares reg
ression). The fish were individually tagged and reared for three month
s at a mean temperature of 8.7 degrees C. The models of growth and fee
d utilization were improved by sorting the individual fish, according
to blindness, and by splitting the modeling in size and time groups. T
he application of weighted multiresponse models gave more relevant inf
ormation than single response models. High dry matter content (dry fee
d) caused the most unambiguous and positive effect on growth and feed
utilization. The floating property of the feed was negatively correlat
ed to growth rates, especially for small fish. The pellet size of 8-9
mm was too big for the small fish (< 55 g at start of the experiment).
Relatively small pellets should only be used if they are sinking. Lar
ge pellets resulted in the highest conversion of feed into flesh, espe
cially by the big fish. Pellet size (var. No. 3) and the combined effe
ct from pellet size and high dry matter content (var. Nos. 1 x 3) were
the most significant positive parameters upon GFCE (gross feed conver
sion efficiency) and protein utilization. High dry matter content (var
. No. 1) and the combined effect from pellet size and high dry matter
content (var. Nos. 1 X 3) were most important to fat utilization. The
sinking/dry/large pellets showed the highest super Y response. Future
feeds should consequently maximize these parameters within biological
reasonable limits to achieve the best utilization of the feed.