selected publications

A Bayesian model takes the physiology to the landscape

The power of a model lies in its ability to predict. The paradox is that, the best model needs the most complete information to forecast, but when we already complete information we would not need the model to make predictions anymore. So what should we do when there is incomplete information? We developed a Bayesian modeling framework that takes physiological knowledge to inform an existing model. Because the physiology represents the mechanisms rather than the correlations, so such model will have a better transferability. [online soon]

The crisis of reproducibility

Can you believe that? A recent Nature survey of 1,576 researchers from various disciplines found that more than 70% of researchers were unable to reproduce research by others, and 50% were not even able to reproduce their own results. We believe that to address this issue, we need a better way to document the Methods, thus we proposed a "checklist" (essentially a metadata standard) to help address the reproducibility crisis [read more]

degree of collinearity vs. collinearity shift

Degree of collinearity (i.e. the linear dependence among predictor variables) is known to affect modeling calibration. But the change of the collinearity structure (i.e. collinearity shift) also negatively affect the model prediction. Both aspects are important factors to consider in modeling practices. [read more]

The headache of model transfer!

Statistical models are commonly used in ecology and biodiversity research. The beauty of model lies in its power to forecast, but the underlying risk is high uncertainty, and such uncertainty is usually not seriously treated in literature. We looked into this issue. Based on simulations, we quantified the "behavior" of 11 algorithms under model transfer scenarios. [read more]

Physiological thinking of the fundamental niche

"Ecological Niche" is a concept proposed by predecessors in ecology (e.g. Hutchinson). There has been many definition of this term, but key aspect is indefinite existence of a species. We proposed a conceptual framework to refine the boundaries of ecological niche, differentiating the species’ existence status (survival, growth, and reproduction) based on physiological limits of major life history characteristics. [read more]

The range expansion of nine-banded armadillo

If you have been southern US, you may have been nine banded-armadillo (Dasypus novemcinctus) running along the highways. Actually, armadillos are not native to US. They were first introduced to Texas in 1849, then and after that they rapidly expanded their distribution northward & eastward. They are observed in Florida, Kansas, Missouri, and also recently in Smokies in Tennessee. The obvious interesting question to me is, where will this species be given the current climate? So I built a model to forecast its potential distribution. [read more]

A nine-banded armadillo, foraging along the shore line of Lake McMurtry, Stillwater, OK [2013]