Problem: Kay would like to run a study on the diversity of leaf-litter insects at different locations throughout Papua New Guinea. At each location she sets up 5 quadrats and collects all the insects in each quadrat. She later identifies every species and averages the number of species per quadrat for each of her locations. In total, she samples 66 locations in the rainforests of New Guinea. She hypothesises that the following covariates will have an effect on insect diversity: leaf litter weight (continuous variable of leaf litter dry weight/square cm), average daily temperature (degrees C), average annual rainfall (mm) and human impact. Human impact is recorded as minimal (no nearby settlements), medium (small nearby settlements with populations <1000), large (nearby towns of >1000 people) and extra large (nearby cities). She produces a linear regression model using the mean number of insect species as the response variable and the above covariates, which finds a significant impact from leaf litter weight, temperature and rainfall, but not human impact. The R-squared is quite good at 0.81. All residual assumptions are met and there is no significant collinearity between covariates. What is the major issue with this study?