Realism, generality, or testability: The ecological modeler's dilemma

Going big by going small: Trade-offs in microbiome explanations of cancer

Microbial factors have been implicated in cancer risk, disease progression, treatment and prevention. The key word, however, is "implicated." Our aim in this paper is to map out some of the tensions between competing methods, goals, and standards of evidence in cancer research with respect to the causal role of microbial factors. We discuss an array of pragmatic and epistemic trade-offs in this research area: prioritizing coarse-grained versus fine-grained explanations of the roles of microbiota in cancer; explaining general versus specific cancer targets; studying model organisms versus human patients; and understanding and explaining cancer versus developing diagnostic tools and treatments. In light of these trade-offs and the distinctive complexity and heterogeneity on both sides of the microbiome-cancer relationship, we suggest that it would be more productive and intellectually honest to frame much of this work, at least currently, in terms of generating causal hypotheses to investigate further. Claims of established causal connections between the microbiome and cancer are in many cases overstated. We also discuss the value of "black boxing" microbial causal variables in this research context and draw some general cautionary lessons for ongoing discussions of microbiomes and cancer.

Modeling behavioral adaptations

Optimization models have often been useful in attempting to understand the adaptive significance of behavioral traits. Originally such models were applied to isolated aspects of behavior, such as foraging, mating, or parental behavior. In reality, organisms live in complex, ever-changing environments, and are simultaneously concerned with many behavioral choices and their consequences. This target article describes a dynamic modeling technique that can be used to analyze behavior in a unified way. The technique has been widely used in behavioral studies of insects, fish, birds, mammals, and other organisms. The models use biologically meaningful parameters and variables, and lead to testable predictions. Limitations arise because nature's complexity always exceeds our modeling capacity.