TY - JOUR
T1 - Simultaneous identification and correction of systematic error in bioenergetics models
T2 - Demonstration with a white crappie (Pomoxis annularis) model
AU - Bajer, Przemyslaw G.
AU - Hayward, Robert S.
AU - Whitledge, Gregory W.
AU - Zweifel, Richard D.
PY - 2004/11
Y1 - 2004/11
N2 - Recent evidence indicates that important systematic error exists in many fish bioenergetics models (BEMs). An approach for identifying and correcting this error is demonstrated with a white crappie (Pomoxis annularis) BEM. Model-predicted trajectories of growth and cumulative consumption for 39 individual white crappie obtained from six 60-day laboratory experiments diverged from observed values by up to 42.5% and 227%, respectively, indicating systematic error in the BEM. To evaluate correlates of the systematic error, model prediction errors were regressed against three major input/output variables of BEMs that were covered by the laboratory experiments: fish body weight (80-341 g), temperature (23-30 °C), and consumption level (0.5%-6.2% daily). Consumption level explained >80% of the prediction error for growth and consumption. Two multiple regression equations containing body weight, temperature, and consumption variables were developed to estimate growth prediction error (R2 = 0.96) and consumption prediction error (R 2 = 0.86), and incorporated into the white crappie BEM to correct its predictions. Cross-validation indicated that growth and consumption prediction error was reduced 2- to 4-fold by correction. Given recent evidence of widespread systematic error and increasing application rates of BEMs, the efficient error-identification and -correction approach described appears broadly applicable and timely.
AB - Recent evidence indicates that important systematic error exists in many fish bioenergetics models (BEMs). An approach for identifying and correcting this error is demonstrated with a white crappie (Pomoxis annularis) BEM. Model-predicted trajectories of growth and cumulative consumption for 39 individual white crappie obtained from six 60-day laboratory experiments diverged from observed values by up to 42.5% and 227%, respectively, indicating systematic error in the BEM. To evaluate correlates of the systematic error, model prediction errors were regressed against three major input/output variables of BEMs that were covered by the laboratory experiments: fish body weight (80-341 g), temperature (23-30 °C), and consumption level (0.5%-6.2% daily). Consumption level explained >80% of the prediction error for growth and consumption. Two multiple regression equations containing body weight, temperature, and consumption variables were developed to estimate growth prediction error (R2 = 0.96) and consumption prediction error (R 2 = 0.86), and incorporated into the white crappie BEM to correct its predictions. Cross-validation indicated that growth and consumption prediction error was reduced 2- to 4-fold by correction. Given recent evidence of widespread systematic error and increasing application rates of BEMs, the efficient error-identification and -correction approach described appears broadly applicable and timely.
UR - http://www.scopus.com/inward/record.url?scp=17144368468&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=17144368468&partnerID=8YFLogxK
U2 - 10.1139/F04-160
DO - 10.1139/F04-160
M3 - Article
AN - SCOPUS:17144368468
SN - 0706-652X
VL - 61
SP - 2168
EP - 2182
JO - Canadian Journal of Fisheries and Aquatic Sciences
JF - Canadian Journal of Fisheries and Aquatic Sciences
IS - 11
ER -