Multivariate geometric designs for mixture experiments and response surface methodology (RSM) were tested as a means of optimizing plant mixtures to support generalist predatory arthropods. The mixture design included 14 treatment groups, each comprised of six planters and having a proportion of 0.00, 0.17, 0.33, 0.66, or 1.00 of each plant species. The response variable was the frequency of predators trapped on sticky card traps placed in each group and replaced 2 times per week. The following plant species were used: Spring 2017: Euphorbia milii, E. heterophylla, and Phaseolus lunatus; Summer 2017: E. milii, Fagopyrum esculentum, and Chamaecrista fasciculata; and, Summer 2018: E. milii, F. esculentum, and Portulaca umbraticola. Predator occurrence was influenced by: 1) Linear mixture effects, which indicated that predator occurrence was driven by the amount of a single plant species in the mixture; or, 2) Nonlinear blending effects, which indicated that the plant mixture itself had emergent properties that contributed to predator occurrence. Predator abundance was highest in the Spring 2017 experiment and both linear mixture effects and nonlinear blending effects were observed. Predator occurrence decreased in subsequent experiments, which were conducted in the warmer summer months. In both Summer experiments, only linear mixture effects were observed, indicating that predator occurrence was driven by the amount of a single plant species in the test mixtures: Euphorbia milii in 2017 and Portulaca umbraticola in 2018. The results showed that not only did the species composition of a plant mixture drive predator occurrence but that proportionality of species contributed to the outcome as well. This suggests that, when formulating a plant mixture to aid in conservation biological control consideration should be given to the proportion of each plant species included in the mixture. RSM can be an important tool for achieving the goal of optimizing mixtures of plants for conservation biological control.