Temporal variability: An intersite comparison of several key facets

Mark Harmon (AND) and Tim Kratz (NTL)

Attendance: 25

Although the LTER network was established to examine long-term temporal trends, there has been surprisingly little comparison of key facets of temporal variability across sites. Aspects of temporal variability such as lags, modulation, and temporal cohesion (i.e., synchrony) are essential to understand to properly scale results both in time and well as space. However, to date more attention has been placed on understanding how to scale in space (probably driven by LTER regionalization efforts) than in time. Past intersite comparison may have been limited by lack of similar records at sites. The pioneering work of John Magnuson and others indicates that relative temporal variation allows a very diverse set of variables to be compared.

This workshop started with a review results from Andrews and North Temperate Lakes on some of these temporal phenomena. The examples from Andrews, presented by Mark Harmon, focused on modulation versus amplification of signals, particularly net primary production and heterotrophic respiration. The work from North Temperate Lakes was presented by Tim Kratz and examined spatial coherence or site-to-site synchronicity. Many aspects of ecosystem and landscape structure of lakes have been revealed by these analyses including differential responses of abiotic, chemical, and biotic variables; and degree of landscape connectivity.

During the discussion that followed it was agreed that examinations of temporal behavior, particularly those of site-to-site synchronicity could provide a structure for the predictions of the synthesis themes. This idea would be essential to include in the research planning for these themes because it will require a specific type of design. For example, it would require replication within the sites and would benefit from the co-location of measurements (i.e., in the exact same place and not in the same general area). Despite these constraints, the method would allow a wide assortment of variables to be compared. The ones listed by the participants included NPP (above- and belowground), decomposition (including soil respiration), populations of native and invasive species, and hydrology-related processes. In addition, it would be important to examine processes that drive the system, such as climate and disturbance (biotic as well as abiotic). Several methodological concerns were expressed by participants that have not generally been considered in past analyses. More appropriate statistical methods would be useful as smoothing of data can strongly influence results.

Two strategies, one for post-facto and one for future analyses were discussed. For post-facto analysis, a next step would be to inventory or assess the amount of time series data available in the LTER network. A great deal has been collected since the Trout Lake workshop that made the first attempt on this subject. In addition to knowing the number of data points available for analysis at sites, it would be useful to know the degree that measures are co-located at the sub-LTER site level. For future analysis it would be important to sell the benefits of the design (e.g., it should help identify general versus specific responses) and to examine analysis models that could be used in these comparisons. It would also be important to start introducing these concepts of analysis into the synthesis topic discussions.

Several volunteered to pursue the next steps of analyzing temporal behaviors at the LTER network level. These included John Yarie, John Bradford, Tom Philippi, Joel Trexler, and Mike Anatolin in addition to Mark Harmon and Tim Kratz.