Premium
An analysis of the analysis of herbivore population dynamics
Author(s) -
Månsson Lena,
Lundberg Per
Publication year - 2006
Publication title -
oikos
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.672
H-Index - 179
eISSN - 1600-0706
pISSN - 0030-1299
DOI - 10.1111/j.2006.0030-1299.14362.x
Subject(s) - herbivore , statistical inference , covariate , population , ecology , inference , statistics , statistical analysis , statistical model , mathematics , econometrics , computer science , biology , artificial intelligence , demography , sociology
Time series analysis of herbivore data with weather included as covariate is commonly used as a mean to shed light on the state and ecology of the studied population. Conclusions about the herbivore population are drawn from statistical parameter values and presence/absence in the most parsimonious model. However, this procedure is only reliable if the statistical parameters have general interpretations regardless of system characteristics. Here we investigated the extent to which this is true by deriving six different vegetation–herbivore‐systems and analyzing their respective statistical parameters. The analysis was done in both continuous and discrete time. It turned out that both density parameters (a 1 and a 2 ) and rainfall coefficients change with biological interactions and amount of average rainfall, and they do so in different ways in different systems. This means that there is no valid general interpretation of them and, most important, the probability of detecting density dependence and effects of rainfall vary between systems. Hence, you can not make inference about the biological processes from statistical analysis without knowing the system that you study and what model best describes the interactions within it.