Get reliable rangeland science

Complex systems exhibit inconsistent variable relationships that give rise to threshold behavior. Thresholds have been evaluated extensively in relation to state and transition boundaries, but are not commonly evaluated within states. Linear or curvilinear graphical trends inflect sharply at thresholds. Thresholds are targets for management to attain or avoid since a small variation of a management variable can give large response of the dependent variable at thresholds. Extrapolation of empirical results across a threshold can drive controversy. Threshold behavior may help explain the rotational grazing controversy. The relationship between grazing recovery period and plant diversity (over years of grazing) may display threshold behavior in some settings. Initial seral state and scale may also be threshold variables in grazing trials. Managers are forced to make tens to hundreds of decisions (trials) per month in adaptive grazing systems with multiple paddocks. The large number of uncontrolled trials over time can lead management to recognize threshold relationships in complex biological systems. In contrast, the small number of trials at high expense available in controlled scientific studies provides limited ability to find thresholds. Science knowledge is especially limited when trials are set up to answer a yes/no question such as, "is one grazing protocol statistically different than another". Induction driven management knowledge from a few individuals has very limited value to range science. However, uncontrolled management knowledge may have significant value if tested over time by hundreds or thousands of managers. Over time production managers sample management knowledge and utilize methods that give results. In a competitive environment they cannot (and will not) commit time and capital resources to ineffective management tools. Effective communication between management and science requires that science value time-tested management knowledge. Management knowledge can provide hypothesis for controlled experiments that evaluate a wide range of variable values