Doctoral Dissertations

Date of Award

5-1997

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Ecology and Evolutionary Biology

Major Professor

James A. Drake

Abstract

An important issue facing ecologists is whether or not individual species matter in maintaining the function of ecosystems. This issue is not a simple one to address experimentally however because it involves either removing a species from an ecosystem or assembling an ecosystem from scratch without a species in it to see how the system would develop. Both approaches are plagued with difficulties and complications that prevent a direct answer to the question from ever being found. Because a direct approach is not possible, an indirect examination of the issue is taken using two different approaches to the problem.

The first approach to the problem was based on assembling ecosystems in different ways using periphyton algae in artificial streams and examining the development of the system in terms of species composition, biomass accumulation, and water chemistry. The approach is based on the premise that systems assembled in different ways but with the same pool of species will obtain different species compositions. If the systems also develop different ecosystem processes and patterns, then it will be possible to argue that species playing different roles in different ecosystems caused the difference in ecosystem function to occur. The second approach examined the role of landscape processes and the migration of individuals between local patches, or metapopulation dynamics, as a mechanism maintaining species richness and influencing species composition. The rationale behind this approach was to find a mechanism that influences species composition and particularly species richness that is independent of ecosystem function. If such a mechanism exists, then there is no reason to believe that there should be a relationship between species richness and ecosystem function.

The assembly approach involved introducing sixteen species of periphyton algae into artificial streams in three different sequences. This approach did indeed result in the development of different species composition and different patterns of biomass accumulation and water chemistry among the three treatments. After about 50 days of development the bluegreen alga Phormidium retzii began to become the numerical dominant in all treatments while a green alga Scenedesmus quadricaudata was also increasing in importance. By day 98 the species composition of the assemblages in the three treatments were nearly identical as expressed in the popular community similarity index SIMI (Standler, 1970). Despite the apparent convergence of species composition, the biomass and water chemistry variables failed to converge. In addition, a closer inspection of the data revealed that some species were likely exhibiting metapopulation dynamics and existing on a small proportion of the available substrate sites in the streams. This list of species included many of the species that had dominated each of the introduction sequences at the beginning of the experiment. The influence of dispersal and migration was observed directly by following the recolonization of substrates that were inserted into the streams to replace those that were removed for sampling. The role of dispersal was particularly strong for Phormidium and Scenedesmus and was probably responsible for their rise to dominance in all the streams.

The experiment was changed on day 99 by altering rates of nutrient loading to the streams. This was done to observe how an external constraint would influence the systems. Increasing the loading rates of nitrogen and phosphorus by an order of magnitude in some streams and decreasing them by an order of magnitude in others produced an amazing lack of response in terms of species composition though biomass levels were affected.

These results indicate that assembly processes do influence the development of species composition and probably the pattern of metapopulation dynamics as well despite the fact that dispersal capability of certain species became the overriding factor controlling species composition later on. The patterns of species composition determined early on during assembly nevertheless established the characteristics of the developing ecosystem and continued to do so even after convergence of the assemblages had occurred. The lack of response to altered nutrient inputs further suggests that species do not simply fit into a template regulated by ecosystem processes but rather that species and assembly processes define the development of ecosystem function.

The second approach to addressing the role of species in ecosystem function involving an examination of metapopulation dynamics in stream periphyton turned out to be a continuation of the first approach. The basic design of this approach was to develop specific predictions of what metapopulation dynamics should look like in periphyton assemblages using a standard model of metapopulation dynamics (Hanski, 1982) and test the predictions experimentally. To make sure the role of immigration was carefully considered in the examination, the periphyton assemblages were subjected to scour disturbances of three different recurrence frequencies to make the role of immigration different in each treatment. The model was parameterized using data from experiments that had been done in the same streams that were to be used for the experimental part of the study.

The model was used to predict that systems experiencing a high frequency of disturbance would be dominated by high immigration rates so that all of the species would be found on all of the available patches. In contrast, systems experiencing low disturbance frequency were expected to have low immigration rates so that all species would be found on only a few patches. Systems with an intermediate disturbance frequency were expected to be intermediate so that some of the species would occur on most of the patches while the rest occurred on only a few patches. In other words, the frequency distribution of patch occupancy for an assemblage of species was expected to be bimodal at intermediate disturbance frequency.

An experimental test of these predictions failed to support the metapopulation model for periphyton assemblages. The frequency distribution of patch occupancy was the same for all disturbance frequencies. The frequency distribution was bimodal but highly skewed with many more species occurring on only a few habitat patches than were found on a high proportion of patches. This result suggests that; 1) the parameters in the model are wrong, 2) the model is wrong, or 3) periphyton do not exhibit metapopulation dynamics. The next step was to examine the parameters in the model as well as the suitability of the model itself. If the problem was neither of these, then the third possibility above must be correct.

The only parameters in the metapopulation model are the probability of immigration and extinction for each species. Immigration in stream periphyton was carefully examined experimentally for differences between species and changes that occur over time. An evaluation of extinction revealed immediately that extinction is problematic to quantify especially for microscopic organisms. This indicated a reformulation of the model was also in order. The careful examination of immigration revealed that there are at least 5 types of immigration patterns in stream periphyton algae. This showed that the immigration parameter had far more variation than had originally been appreciated. The removal of the extinction term from the model was accomplished by recognizing that a rescue effect can be achieved by individuals migrating between patches. Extinction is thus related to the rate of immigration and can be replaced in the model. The new term was further enhanced by recognizing that propagules attempting to immigrate come from two sources, within the immediate landscape area and from outside the immediate landscape area. Incorporation of this scale factor into the model converted it from a closed two scale model (local processes and landscape patterns) to an open multiscale model.

Despite these improvements, the model predictions still failed to adequately explain the experimental frequency distributions for patch occupancy although the model fit was much improved. The final improvement to the model was an attempt to include the effects of ecosystem level factors such as resource levels which regulate productivity and the intensity of competitive interactions among species. Inclusion of a term to account for ecosystem level factors in the model produced frequency distributions for patch occupancy that were not statistically different from the experimental results for any of the three disturbance frequencies.

These results suggest that stream periphyton exhibit metapopulation dynamics in a manner that can be predicted within reasonable accuracy by the patterns of immigration onto local habitat patches for each of the species. If the model developed here turns out to be correct, characteristics of the ecosystem that regulate resources and productivity of the assemblage are also important in determining the patterns of metapopulation dynamics. Thus it is not possible to argue that metapopulation dynamics provide a mechanism for regulating species composition that is independent of ecosystem processes and function. However, it was also determined that merely changing the order in which species were introduced into a system results in different patterns of metapopulation dynamics and even in different patterns of development of ecosystem processes and function. Not only that but the effect of different species introduction sequences on metapopulation dynamics and ecosystem function persisted even after effect on species composition of the assembly process was eliminated by high dispersal rates.

From these observations it is possible to conclude that species are important to the development of ecosystem function. Even after dispersal drove Phormidium and Scenedesmus to numerical dominance, biomass levels and water chemistry remained different. Thus species do not have to be dominant to influence ecosystem properties. Because of dispersal and the role of assembly processes in determining the development of ecosystem properties, a species does not even need to be important in the ecosystem where is presently occurs. A species importance may well lie in its ability to disperse and influence assembly processes being initiated somewhere else. At this point in our state of knowledge, it seems premature to believe that any species can be considered unimportant.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS