Per Aspera ad Astra: Through Complex Population Modeling to Predictive Theory

TY - JOUR

T1 - Per Aspera ad Astra

T2 - Through Complex Population Modeling to Predictive Theory

AU - Topping, Christopher J.

AU - Alroe, Hugo Fjelsted

AU - Farrell, Katharine N.

AU - Grimm, Volker

PY - 2015/11

Y1 - 2015/11

N2 - Population models in ecology are often not good at predictions, even if they are complex and seem to be realistic enough. The reason for this might be that Occam's razor, which is key for minimal models exploring ideas and concepts, has been too uncritically adopted for more realistic models of systems. This can tic models too closely to certain situations, thereby preventing them from predicting the response to new conditions. We therefore advocate a new kind of parsimony to improve the application of Occam's razor. This new parsimony balances two contrasting strategies for avoiding errors in modeling: avoiding inclusion of nonessential factors (false inclusions) and avoiding exclusion of sometimes-important factors (false exclusions). It involves a synthesis of traditional modeling and analysis, used to describe the essentials of mechanistic relationships, with elements that arc included in a model because they have been reported to be or can arguably be assumed to be important under certain conditions. The resulting models should be able to reflect how the internal organization of populations change and thereby generate representations of the novel behavior necessary for complex predictions, including regime shifts.

AB - Population models in ecology are often not good at predictions, even if they are complex and seem to be realistic enough. The reason for this might be that Occam's razor, which is key for minimal models exploring ideas and concepts, has been too uncritically adopted for more realistic models of systems. This can tic models too closely to certain situations, thereby preventing them from predicting the response to new conditions. We therefore advocate a new kind of parsimony to improve the application of Occam's razor. This new parsimony balances two contrasting strategies for avoiding errors in modeling: avoiding inclusion of nonessential factors (false inclusions) and avoiding exclusion of sometimes-important factors (false exclusions). It involves a synthesis of traditional modeling and analysis, used to describe the essentials of mechanistic relationships, with elements that arc included in a model because they have been reported to be or can arguably be assumed to be important under certain conditions. The resulting models should be able to reflect how the internal organization of populations change and thereby generate representations of the novel behavior necessary for complex predictions, including regime shifts.

KW - complexity

KW - error avoidance

KW - agent-based models

KW - model development

KW - modest approach

KW - INDIVIDUAL-BASED ECOLOGY

KW - NARRATIVE EXPLANATION

KW - DYNAMICS

KW - MANAGEMENT

KW - LESSONS

KW - SCIENCE

KW - PATTERN

KW - ENERGY

U2 - 10.1086/683181

DO - 10.1086/683181

M3 - Journal article

C2 - 26655779

SN - 0003-0147

VL - 186

SP - 669

EP - 674

JO - American Naturalist

JF - American Naturalist

IS - 5

ER -