Science Article Sheds Light on Causality in Complex Ecosystems
An international team of researchers, including Prof. Chih-hao Hsieh of the Institute of Oceanography and Institute of Ecology and Evolutionary Biology, published the results of its cuttingedge research aimed at identifying causal networks in the October 26 issue of the prestigious journal Science. The team’s article, “Detecting causality in complex ecosystems,” introduces a new method for analyzing the causal relations between biological and non-biological elements within ecological systems.
A causal relationship does not necessarily exist between two things just because they appear to be correlated. This can be illustrated through a simple example from life: During the summer, people become easily impatient and irritable, while sales of ice cream go up as well, so there is a strong statistical relationship between impatience and irritability and ice cream sales volumes. Nonetheless, there is in fact no causal connection between the two; they are both simply influenced by the weather.
When we conduct research on ecological systems, for instance, and we observe a significant correlation between the population dynamics of two species in nature, it is possible there are inter-species interactions between the two species, yet it is also possible that the two species are being affected by a similar environment at the same time.
This issue came to light only through the work of Clive Granger and Robert Engle. They abandoned traditional correlation in favor of prediction as the basis in their effort to analyze causal relationships, and they verified this approach through their analysis of financial and economic data. The pair ultimately shared the 2003 Nobel Prize in Economics for making this breakthrough.
Prof. Hsieh and his collaborators developed convergent cross mapping to detect causality in complex systems. Simply speaking, this method is designed based on the principle that anything will leave a trace where it has passed. The researchers used time series data sets to test whether A ever left a trace in B’s history; if so, then A is the cause that caused B.
In the Science article, the team applied this approach in analyzing historical data on two vital fisheries—anchovies and sardines—in the ecological system of the ocean current near California to shed light on the causal factors behind the variations in the populations of these fisheries. Over the last century, the populations of anchovies and sardines experienced anti-synchronistic fluctuations on repeated occasions. There was an obvious negative correlation, with one population expanding when the other shrank.
It has long been debated whether these variations were due to inter-species competition or environmental factors (e.g., changes in ocean temperature). The team’s method showed that the anti-synchronistic fluctuations in these populations were caused by the different reactions of the two species to changes in ocean temperature and not inter-species competition.