604 
Fishery Bulletin 96(3), 1 998 
The goal of the study was to investigate the effects 
of dissolved oxygen and temperature on growth, sur- 
vival, and respiration of juvenile (young-of-the-year) 
Atlantic sturgeon. High temperatures are known to 
amplify negative effects of hypoxia on growth and 
survival of estuarine fishes (Coutant, 1985). Habi- 
tats in Chesapeake Bay that satisfy both tempera- 
ture and dissolved oxygen (DO) preferences of At- 
lantic sturgeon may be limiting during the summer 
as have been demonstrated for striped bass (Coutant 
and Benson, 1990; Brandt and Kirsch, 1993). In this 
study we define hypoxia as oxygen concentrations 
<4.0 mg/L. Hypoxia has been defined previously as 
<2 mg/L for Chesapeake Bay (Phil et al., 1991; 
Harding et al., 1992), an ambient oxygen level that 
is detrimental to benthic infaunal production. This 
definition, however, may be too stringent for fishes 
because oxygen concentrations at this level are of- 
ten lethal (Brett, 1970; Jordan et al., 1992). 
A nested-multifactorial experiment was designed 
to investigate the effects of temperature and dis- 
solved oxygen on growth, respiration, and survival 
of young-of-the-year Atlantic sturgeon. During the 
course of our experiments we observed that fish in 
hypoxic conditions frequently surfaced, often break- 
ing the surface of the water with their snout. There- 
fore we included surface access as a third factor in 
our investigation — whether this behavior might ben- 
efit growth and survival under conditions of oxygen 
stress. 
grams wet weight (>10 cm total length) and large 
enough to be handled with little or no stress for scute- 
clips (see below). 
Nested growth and survival experiments 
A nested multivariate experiment (Fig. 1) evaluated 
survival and growth rates in relation to access to 
surficial water (sealed or unsealed tank), tempera- 
ture (~20°C and ~26°C), dissolved oxygen (~3 mg/L 
or ~7 mg/L), and tank replication. The nested design 
directed the analysis of variance to occur in hierar- 
chical order at four levels. The model of the nested 
design for growth rate was 
yijklm - L 1 + T i + Pj(i) + Yk(ij) + 0\(,ijk) + £ (ijkl)m, 
where y ijklm = 
V , 
T i 
Pj(i) ~ 
Y k(ij) - 
® l(ijk) ~ 
£ (ijkl)m ~ 
the growth rate response by indi- 
vidual juveniles; 
overall mean; 
the effect of the ith surface access 
category; 
the effect of the yth temperature; 
the effect of the Mh oxygen level; 
the effect of the Zth tank (replicate); 
and 
the random error component. 
The model of the nested design for survival rate was 
s ijki = f J + l 'i + fij(i) + Ykdj) + £ iujk). 
Methods 
Experimental material 
Juvenile Atlantic sturgeon were obtained from the 
U.S. Fish and Wildlife Service, Northeast Fishery 
Center, Lamar, Pennsylvania (Hendrix, 1995). Dur- 
ing June 1995, Center personnel collected a large 
female (2.4-m total length) and three male Atlantic 
sturgeon from the Hudson River near Hyde Park 
(River km 135). Fish were transported to the Center 
for artificial spawning and for larval rearing. Lar- 
vae and early juveniles were reared at the Center at 
17°C and 1 ppt salinity. A failure in the water heat- 
ing system at the Center caused juveniles, age 45 to 
60 days after hatching, to experience low water tem- 
peratures, ca. 10°C. At 60 days after hatching, 500 
juveniles (0. 3-2.0 g wet weight) were transported in 
an oxygenated container to Chesapeake Biological 
Laboratory and acclimated to 19°C and 2 ppt salin- 
ity over a 10-d period. Juveniles were reared in six- 
teen 40-liter tanks and fed Biokyowa©fry feed (700- 
2000 pm diameter) ad libitum until they were ca. 5 
where S ijk[ = the arcsine-transformed survival rate 
for each replicate. 
Statistical significance for factors was accepted at 
a = 0.05 (type-I sum of squares for type-I error). To 
remove possible effects due to differences in fish size 
among experiments and experimental levels, initial 
fish weight was included in ANOVAs as a covariate. 
Calculations of variances and significance tests were 
performed by using PC-SAS, PROC NESTED (SAS, 
1982). Survival data were arcsine transformed to 
meet assumptions of normally distributed error. 
Four 10-day experiments were conducted under the 
four combinations of surface access, temperature, and 
dissolved oxygen, each replicated twice. In the high- 
temperature hypoxia treatments, high mortality was 
observed. Because we wished to have greater confi- 
dence in associating low survival with high-tempera- 
ture hypoxia, we repeated the treatments with surface 
access and high temperature (at both low and high DO 
levels) for a total of four replicates (Fig. 1). 
Experimental tank dimensions were 78-cm diam- 
eter, 46-cm height, and 220-liter volume. External 
