In a study of the joint toxicity of copper and zinc, researchers randomly allocated 25 beakers containing minnow larvae to receive one of 25 treatment combinations. The treatment levels were all combinations of 5 levels of zinc and 5 levels of copper added to a beaker. Following a four-day exposure, a sample of the minnow larvae were homogenized and analyzed for protein. These data are available on the course webpage under the filename protein. txt. The data appear in three columns as copper level (ppm), zinc level (ppm), and protein (µg/larva), respectively. Treating protein as the response and the copper and zinc levels as the explanatory variables, fit a full second-order (including interactions) linear model, and perform backward elimination based on partial significance (p-value) with α = 0.05. Summarize which variables were removed at each step. [Since there are higher-order terms in this model, I would avoid running any R-automated backward elimination procedures you might find.]