About

What is Ecological Niche Modeling?

I have long been a researcher of ecological niche modeling. To comprehend the ecological conditions in which a species can survive is the foundation for ensuring the survival of ecosystems where millions exist.

What we look for in ecological niche models are the critical factors in a species’ environment. This often requires evaluating a set of conditions that occur naturally with a species. We observe, assess and collect data that allows us to capture occurrence points and how they impact lifecycles. This information can include habitat, temperatures and seasonal precipitation. Occurrences will be used to determine those environmental conditions where a species persists. And while our estimates cannot be considered absolute, they give us adequate localized information that can be effectively applied to other areas.

Once the data has been collated, it can be used to build mathematical models of an environment’s tolerances on a species. While there are many approaches to this, the end result will be a set of points in environmental space based on data culled from points in geographic space.

It is a complicated process, but an important one. I — and others like me — are designing algorithms to model various species and use these results to make geographic determinations of environmental effects. This gives us a greater idea of habitat suitability across geographic space.

Of course, the science behind ecological niche modeling cannot be fully introduced in the space allowed here. There are concepts and methodological issues that require extensive study. I myself have a Ph.D. in Ecosystem Dynamics. I have worked with the WHO, Parks and Wildlife, Land and Natural Resources services, as well as the San Diego Supercomputer Center at the University of California San Diego. Through grants from the DOT, USGS and NSF, I was involved in the development of data and computational ecological niche modeling infrastructures.

My contributions to the academic field have been recognized by the US Immigration Service as an Outstanding Researcher. I have developed a number of software systems including The GARP modelling system, a genetic algorithm for rule=set prediction, and WhyWhere, an algorithm for predicting and explaining the distribution of species. I am the writer of many peer reviewed articles, and the author of Niche Modeling: Predictions from Statistical Distributions. You can find out more about my research here.

While understanding ecological niche modeling may not be for everyone, I do believe it is important that we all know how vital it is to the sustainability of our ecosystems and species.

What is Research Bias?

Another theme of my research has been the recognition of research bias. Avoiding research bias in modelling and data analysis is essential if we are to have confidence in the results of any area of science.

What I look for are poor control of the critical factors that determine the reliability of studies. This often requires comparing model results with real and simulated data. When we observe results that deviate significantly from normal expected patterns, probably bias has occurred. While bias does not necessarily disprove a study, it does bring its reliability into question.

There are many examples of bias in research on the rate and consequences of global warming. For example, when a researcher builds a mathematical model of an environment’s tolerances on a species, and uses this model to predict what will happen when the climate changes, some species will gain and some species will lose under normal conditions. When a study reports only the consequences for species that lose – such as extinctions – then this is an example of researcher bias.

Another example is the prediction of change in frequency of droughts in Australia in the future. If the climate models used to predict the trend in droughts in the future cannot adequately predict the trend in drought in the past, then the models are invalid and should be rejected as unfit for use. To do otherwise is an example of researcher bias.

Another example of researcher bias is the exaggeration of the rate of global warming. If a researcher makes an explicit prediction about the rate of warming in the future, then a retrospective examination will identify bias when that prediction disagrees with observations. For example, extrapolations of the high rate of warming in the early 1990’s in Recent Climate Observations compared to Predictions were shown to be exaggerated in Recent climate observations: Disagreement with projections due to statistical errors.

Another example is a tendency to correct errors in weather station data – a process known as homogenization – in the preferential direction of increased warming, leading to datasets that exaggerate warming in Australia this century.

Revealing the errors in scientific studies is a thankless and difficult task, but it is an important one. Without it, we would not be aware of the limitations and would not develop improved algorithms that give us more confidence in our results. And clearly when it comes to issues that have important implications such as climate change, the discovery of researcher bias is even more vital.