On the exponent in the Von Bertalanffy growth model

• Published in: PeerJ (San Francisco, CA) Vol. 6; p. e4205
• Main Authors: Renner-Martin, Katharina, Brunner, Norbert, Kühleitner, Manfred, Nowak, Werner Georg, Scheicher, Klaus
• Format: Journal Article
• Language: English
• Published: United States PeerJ. Ltd 04-01-2018; PeerJ, Inc; PeerJ Inc
• Subjects: Age; Agricultural Science; Akaike’s information criteria (AIC); Analysis; Animal populations; Animal sciences; Applied research; Aquaculture, Fisheries and Fish Science; Biology; Computational Biology; Datasets; Differential equations; Fish; Fisheries management; Mathematical Biology; Metabolic scaling exponent; Metabolism; Multi-model inference; Scaling; Von Bertalanffy growth function (VBGF); Weak universality; Zoology; Metabolic scaling exponent; Multi-model inference; Weak universality; Akaike’s information criteria (AIC); Von Bertalanffy growth function (VBGF)
• ISSN: 2167-8359; 2167-8359
• DOI: 10.7717/peerj.4205

Von Bertalanffy proposed the differential equation
'(
) = 
 × 
(
)
 - 
 × 
(
) for the description of the mass growth of animals as a function
(
) of time
. He suggested that the solution using the metabolic scaling exponent
= 2/3 (Von Bertalanffy growth function VBGF) would be universal for vertebrates. Several authors questioned universality, as for certain species other models would provide a better fit. This paper reconsiders this question. Based on 60 data sets from literature (37 about fish and 23 about non-fish species) it optimizes the model parameters, in particular the exponent 0 ≤ 
 < 1, so that the model curve achieves the best fit to the data. The main observation of the paper is the large variability in the exponent, which can vary over a very large range without affecting the fit to the data significantly, when the other parameters are also optimized. The paper explains this by differences in the data quality: variability is low for data from highly controlled experiments and high for natural data. Other deficiencies were biologically meaningless optimal parameter values or optimal parameter values attained on the boundary of the parameter region (indicating the possible need for a different model). Only 11 of the 60 data sets were free of such deficiencies and for them no universal exponent could be discerned.