This work is devoted to studying a class of biological control problems in a stochastic environment. Specifically, it focuses on stochastic Lotka-Voltera systems. Our effort is on treating average cost per unit time controlled diffusions. It is natural to use a vanishing discount argument. However, in contrast to the existing literature, neither the 'near-monotone' nor the 'stable' condition is satisfied in the current set up. In reference to one of our recent works, we divide the domain into two parts. In one sub-domain, the 'near-monotone' condition is satisfied, whereas in the other sub-domain, the 'stable' condition is satisfied. We then carefully work out the analysis in the two domains so as to obtain the desired optimal control.