Intrinsic and Extrinsic Variability in Community Dynamics II

Members of the Intrinsic and Extrinsic Variability in Community Dynamics Working Group met for their second meeting 5 to 17 January 1997 at NCEAS in Santa Barbara. Following on our first meeting in fall 1996, this session had two general objectives:

Much of our time was spent on developing several questions that had originally been proposed in our first meeting (see summary of first meeting on NCEAS Homepage). Several of our main activities in this area are summarized below.

1. Examining the relationship between zooplankton community diversity and responses to ecological factors in the ecosystems in our dataset. In particular, we sought to determine whether there were any relationships between species diversity and community responses to stress and resilience particularly in several manipulated lakes. Our initial results did not suggest any straightforward relationships between zooplankton diversity and ecosystem function in our study systems. We will also use the comparative data listed in 5 below to test for any relationship between diversity and patterns of intra-annual variability.

One valuable spin-off of this work was an investigation of the rate of change of zooplankton communities as measured by similarities of the communities in two fairly close time periods (typically two weeks) which we have termed "sequential similarity". Some interesting patterns were apparent particularly when examining similarities based upon simple functional groups.

2. Developing a categorization scheme for classifying zooplankton into about ten functional groups which were common to most of the lakes in our dataset. Tim Wootton leads the development of this classification scheme and will use it to test interactions strengths in zooplankton communities across the different lakes in our dataset. The categorization scheme was developed by the investigators familiar with each lake set in a "Delphi" session in which a common scheme was developed that did not depend upon the particular taxa in any lake.

3. Testing the applications of multivariate, autoregressive models in predicting a vector of abundances of taxa at time t from their abundances at time t-1. The basic form of the autoregressive models is Xt = A + BXt-1 + CUt + Et, where Xt is a vector of species abundances, A is a vector of constants, B is the species interaction matrix, Ut, is the matrix containing environmental drivers, and C is a matrix of constants describing the effect of the environmental drivers on the species in the system. This work was spearheaded by B. Dennis, K. Cottingham, and T. Ives who developed Matlab program to explore these AR(1) models.

4. Assessing the extensive dataset available from Artemia population in Mono Lake using both linear and nonlinear models of population dynamics. The Mono Lake zooplankton community, consisting of only one species, is much more simple that those from other lakes in our dataset. It therefore provides a good testing ground on which to compare linear and nonlinear time-series approaches.

5. Investigating whether there are general characteristics common to all of the reference lakes in our dataset. This will initially focus on some simple statistical properties (e.g., mean, variance, and autocorrelation) of a wide range of common lake variates particularly for zooplankton (e.g., biomass, density, species richness, size indices, compare taxonomic and functional groupings). These features will be compared at several scales including year to year, lake to lake, and region to region.

6. Following on (5) the same characteristics will be evaluated for manipulated lakes. We will test if there are certain responses that are common to all manipulated lakes. We will also examine how responses to manipulations can be related to a) lake morphometry (size, depth, flushing), b) species richness (see (1) above), c) chemistry, d) dominant predators, e) dominant prey, f) habitat or refuges.

7. Determining whether zooplankton abundances in the spring predict abundances later in the year, and whether abundances in the fall influence abundances in the following year.


Some short-term targets (nearing completion within 6 months) for our working group include

a) Quantify indirect effects of a Bythotrephes invasion in Harp lake using an AR(1) approach (Yan, Rusak, Ives, Wooton).

b) Mono Lake models (Hastings, Melack, Ives, et al.).

c) Examine the application of AR(1) models on the Cascade Lakes (Dennis, Cottingham, Carpenter, Ives).

d) Develop a paper on the use of sequential similarity indecies for zooplankton communities for a paper at ESA in August 97 (Micheli, Frost, Fischer, Klug, et al.).

e) Develop a paper on some general features of our multi-reference-lake dataset for a paper at ESA in August 97 (Cottingham, Frost, Carpenter, Ives, Yan, Patterson, et al.).

Longer term targets include

a) Extending the comparisons across all lakes, including Mono.

b) Extending the development of models for Mono Lake.

c) Expanding the comparison of the lake data into considerations of functional groups and exploring the applications of AR(1) and Alternate State assessments for these data.

d) Continue the evaluations of Spring vs. Fall and Fall vs. the following Spring comparisons.

Summary Compiled by Tom Frost with assistance from Kathy Cottingham, Tony Ives, and Norm Yan.