Twenty-first century Ecology requires statistical literacy. Observational studies routinely gather multivariate data at many spatiotemporal scales and experimental studies routinely include multiple blocked and nested factors. Our journals are replete with likelihood and state-space models, Bayesian and frequentist inference, and complex multivariate analyses, and publish papers on statistical theory and methods. We test hypotheses, model data, and forecast future environmental conditions. And many statistical methods cannot be automated in software packages. Developing statistical literacy among ecologists requires overcoming challenges in recognition and understanding. First, we must recognize that fundamental ecological theories are best phrased in terms of stochastic differential-equation models, but our textbooks have not yet caught up with these models. Second, we must understand statistical modeling well enough to construct, or collaborate with statisticians who construct, nonstandard statistical models and apply various types of inference – estimation, hypothesis testing, model selection, and prediction – to our models and scientific questions. How can ecologists successfully meet these challenges when teaching and learning statistics?
Results/Conclusions
Ecologists must first appreciate that statistics is a mathemically-based research discipline and that statistical tools evolve; it is neither a static entity nor an off-the-shelf toolkit. Some ecologists will keep up with the statistical literature and keep their students and colleagues abreast of changes in the field. Others will establish fruitful collaborations with these statistically-aware ecologists or with professional statisticians. We expect that the collaborative approach will be more common. Thus, we suggest that literate ecologists at a minimum should master core statistical concepts, including probability and likelihood, principles of data visualization and reduction, fundamentals of sampling and experimental design, the difference between design-based and model-based inference, model formulation and construction, and basic programming. Because mathematics is the language of statistics, familiarity with essential mathematical tools – matrix algebra and especially calculus – is a must and will facilitate collaborations between ecologists and statisticians. Our experience suggests that statistical concepts are best illustrated in computational laboratories using a diversity of real datasets whose analysis requires different models and approaches. Students must learn statistical concepts before their study designs have been finalized and the data have been collected. Pilot studies are ideal case-studies to use to illustrate statistical concepts, and their analysis can lead to refinements in full-scale design and subsequent analysis. Our experience suggests that students benefit most from statistics courses taught jointly by teams of statisticians and ecologists.
