From census to survey – changing monitoring methodology without compromising time series integrity

Abstract Time series arising from long-term monitoring constitute a key source for ecological knowledge, and their value continues to increase as new data are being added. Yet, in many cases data collection methodologies applied are rooted in historical decisions. To adapt to new questions, opportunities or challenges, methodologies may have to be changed. Such changes, however, can not jeopardize the integrity of long-term time series. Here we present a statistical protocol that can be used to optimize and validate the switch in methodology for estimating abundances in a designated area from a full census approach to a sample survey with permanent transects. To evaluate the generality of our method, we first implemented our methodology on a suite of simulated abundance data, and second on a real-life case study on high Arctic collared lemmings (Dicrostonyx groenlandicus). For the simulated data series, our approach generally allowed for the transition from census-based to survey-based monitoring without compromising time series integrity, even with a relatively low number of transects. Accuracy, however, depends on overall abundance of the focal species. For the real-life data, we demonstrate that the new transect-based approach indeed was able to capture the dynamics and cyclicity of the population abundance present in the census data collected over almost 30 years. The integrity of the time series is thus maintained, whilst 94% less effort is being invested in the collection of data. Using the provided R code, the approach is easily implemented in other monitoring programs that require a change in methodology.