The cellular composition of tumors is highly heterogeneous, involving not only divergent lineages of transformed cancer cells, but also host cells of the stroma and immune system. The complicity of protumoral host cells is essential for conferring malignancy and promoting progression in tumor in a wide range of solid tissues, including breast, pancreas, and brain. This understanding has led to the concept of ecological treatment: that is, molecular therapies aimed not directly at the destruction of cancer cells, but at disrupting interactions between tumor cells and host cells or the microenvironment, in effect creating a microenvironment unfavorable for tumor progression. In order to design effective eco- logical interventions and predict the course of progression of complex tumors, a quantitative understanding of the interactions between cellular subpopulations in tumors is essential. The theoretical branch of ecology has long established a history of using mathematical modeling to describe and predict the behavior of heterogeneous populations consisting of myriad interacting individuals each susceptible to noise in their responses to local stimuli, and complex systems comprised of different subpopulations engaged in asymmetrical interactions. We adapt some of these models--specifically, an agent-based self-propelled particle model and a population dynamics differential equation model--to the problems of stromal cell-dependent cancer cell migration and growth. From the former study, I find that paracrine signaling between tumor cells and in- creases the stability and efficiency of a preexisting tumor cell collective migration phenotype, rendering the net comigratory behavior more robust against microenvironmental fluctuations. From the latter study, I find that gliomas de- pendent upon protumoral tumor-associated macrophages for growth undergo multi-phasic growth dynamics. I also conclude that macrophage-targeted treat- ment of such tumors in a linear stage of progression leads to tumor reduction dependent on the size and composition of the tumor at the time of treatment initiation, and that tumors exhibiting weak response to such a treatment may harbor hidden vulnerability to combinatorial therapy. In addition, I infer from these theoretical studies possible methods of intervening in protumoral ecological interactions.