Long-term environmental monitoring: some perspectives from lakes
As part of a workshop sponsored by the Sustainable Biosphere Initiative on Change and Trend Detection, we have reviewed some general considerations pertinent to establishing long-term monitoring programs. A general goal of most long-term monitoring is to assign causality and establish mechanisms for patterns in observed data. However, mechanistic inference can be difficult in complex, often uncontrolled, natural systems. We present examples from three lakes representing (1) highly manipulated, intensively monitored systems, (2) management situations where many simultaneous manipulations may be occurring, and (3) at-large surveillance programs with no deliberate manipulations. We show that at one end of this continuum, in controlled, intensively monitored lakes, delayed and counterintuitive responses can make mechanistic inference difficult. At the other end of the continuum, where no management or experimental manipulations are underway, it is still possible to ascertain mechanism in the presence of apparently conflicting data, given sufficient information about the system. Management situations are often difficult to decipher; many simultaneous manipulations often confound mechanistic interpretations, although patterns in the data are readily apparent.