Premium
Nonlinear relationships between vital rates and state variables in demographic models
Author(s) -
Dahlgren Johan P.,
García María B.,
Ehrlén Johan
Publication year - 2011
Publication title -
ecology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.144
H-Index - 294
eISSN - 1939-9170
pISSN - 0012-9658
DOI - 10.1890/10-1184.1
Subject(s) - nonlinear system , mathematics , vital rates , population , state variable , population model , statistics , linear model , regression analysis , econometrics , regression , spline (mechanical) , polynomial , linear regression , population growth , demography , mathematical analysis , physics , structural engineering , quantum mechanics , sociology , engineering , thermodynamics
To accurately estimate population dynamics and viability, structured population models account for among‐individual differences in demographic parameters that are related to individual state. In the widely used matrix models, such differences are incorporated in terms of discrete state categories, whereas integral projection models (IPMs) use continuous state variables to avoid artificial classes. In IPMs, and sometimes also in matrix models, parameterization is based on regressions that do not always model nonlinear relationships between demographic parameters and state variables. We stress the importance of testing for nonlinearity and propose using restricted cubic splines in order to allow for a wide variety of relationships in regressions and demographic models. For the plant Borderea pyrenaica , we found that vital rate relationships with size and age were nonlinear and that the parameterization method had large effects on predicted population growth rates, λ (linear IPM, 0.95; nonlinear IPMs, 1.00; matrix model, 0.96). Our results suggest that restricted cubic spline models are more reliable than linear or polynomial models. Because even weak nonlinearity in relationships between vital rates and state variables can have large effects on model predictions, we suggest that restricted cubic regression splines should be considered for parameterizing models of population dynamics whenever linearity cannot be assumed.