Blood samples were taken from one parr in each tank (n = 2) on day −1, from five parr on days 9 and 20 (n = 10) and from six parr in each tank (n = 12) on day 37, by caudal venous puncture using 1 mL sodium heparinised syringes. Plasma chloride was immediately analysed by an i-STAT 1 analyser (Abbot Norge AS, N-1361 Billingstad, Norway), using EC8+ cassettes. Individual fish was taken from each tank consecutively to minimise disturbance. Second gill arch on the left side and mid-kidney samples were taken from one parr in each tank (n = 2) on day −1 and from six parr in each tank on day 37 (n = 12). Kidneys were examined for nephrocalcinosis. Gills for Al analysis were sampled on day 37. The gill filament on the third gill arch from the left side of each fish was dissected and stored frozen before being freeze-dried. After microwave-assisted digestion, yttrium was added as internal standard. Gill Al was analysed using ICP-OES in axial viewing mode. The limit of detection for the Al measurements was 0.9 μg L−1 or about 0.6 μg Al g−1 dry weight gill. Body weight and fork length were measured on days −7, 23 and at the end of the experiment (day 38). The condition factors (K) and specific growth rate (SGR) were calculated as described in Hosfeld et al. (2008).
The statistical analysis of the blood data and growth parameters (weight, length and condition factor) included linear and nonlinear regression analysis. Statgraphics Centurion 16 (Statpoint technologies Inc., Virginia) was applied in these analyses. Leverage was used to investigate if any of the data points had too much influence in determining