Turra Alexander, Corte Guilherme N, Amaral Antonia Cecília Z, Yokoyama Leonardo Q, Denadai Márcia R
Departamento de Oceanografia Biológica / Instituto Oceanográfico, Universidade de São Paulo, São Paulo, SP, Brasil.
Departamento de Biologia Animal / Instituto de Biologia, Universidade Estadual de Campinas, Campinas, SP, Brazil.
PeerJ. 2018 Jun 27;6:e5070. doi: 10.7717/peerj.5070. eCollection 2018.
Evaluation of relative (allometric) growth provides useful information to understand the development of organisms, as well as to aid in the management of fishery-exploited species. Usually, relative growth analyses use classical models such as the linear equation or the power function (allometric equation). However, these methods do not consider discontinuities in growth and may mask important biological information. As an alternative to overcome poor results and misleading interpretations, recent studies have suggested the use of more complex models, such as non-linear regressions, in conjunction with a model selection approach. Here, we tested differences in the performance of diverse models (simple linear regression, power function, and polynomial models) to assess the relative growth of the trigonal clam , an important fishing resource along the South American coast. Regressions were employed to relate parameters of the shell (length (), width (), height () and weight ()) among each other and with soft parts of the organism (dry weight (DW) and ash-free dry weight ()). Then, model selection was performed using the information theory and multi-model inference approach. The power function was more suitable to describe the relationships involving shell parameters and soft parts weight parameters (i.e., vs. , , and and vs. ). However, it failed in unveiling changes in the morphometric relationships between shell parameters (i.e., vs. and ; vs. ) over time, which were better described by polynomial functions. Linear models, in turn, were not selected for any relationship. Overall, our results show that more complex models (in this study polynomial functions) can unveil changes in growth related to modifications in environmental features or physiology. Therefore, we suggest that classical and more complex models should be combined in future studies of allometric growth of molluscs.
相对（异速）生长的评估为理解生物体的发育以及协助管理渔业开发物种提供了有用信息。通常，相对生长分析使用经典模型，如线性方程或幂函数（异速生长方程）。然而，这些方法没有考虑生长中的不连续性，可能会掩盖重要的生物学信息。作为克服结果不佳和误导性解释的替代方法，最近的研究建议使用更复杂的模型，如非线性回归，并结合模型选择方法。在这里，我们测试了不同模型（简单线性回归、幂函数和多项式模型）在评估三角蛤相对生长方面的性能差异，三角蛤是南美海岸一种重要的渔业资源。采用回归分析来关联贝壳的参数（长度（）、宽度（）、高度（）和重量（））之间以及与生物体的软组织（干重（DW）和无灰干重（））之间的关系。然后，使用信息论和多模型推断方法进行模型选择。幂函数更适合描述涉及贝壳参数和软组织重量参数的关系（即与、和以及与）。然而，它未能揭示贝壳参数之间形态计量关系（即与和；与）随时间的变化，而多项式函数能更好地描述这些变化。反过来，线性模型没有被选用于任何关系。总体而言，我们的结果表明，更复杂的模型（在本研究中为多项式函数）可以揭示与环境特征或生理变化相关的生长变化。因此，我们建议在未来软体动物异速生长的研究中应将经典模型和更复杂的模型结合起来。
