Microbial ecology is often regarded as the epitome of ecological reductionism. Although many microbial ecologists adopt reductionist approaches to achieve causal explanation, I feel this is due to the small size of the organisms and systems they work with rather than to any necessary preoccupation with small scale entities or approaches. Further, I believe that the relative proportion of holists and reductionists in microbial ecology does not differ from other disciplines. Microbial ecologists, however, must brave a nebulous episteme when laboratory studies of processes are to be related to field measures of the results of these processes. Two very different aspects of my research, one clearly reductionist (in the sense of small vs large) and the other more holistic, eventually brought me to the same point in understanding how tundra ecosystems function. Here I will discuss how observations and measurements of climate variables and respiration in organic residues, on the one hand, and data on microbial taxa, physiology and biomass on the other, gave me the same insight as to how and to what extent microbes contribute to or relate to system functions such as primary productivity, organic matter turnover and nutrient cycling. These ecosystem perspectives emerged fortuitously from studying some of the parts of the system. My basic questions were: 1) How are the processes which result in CO2 emission from soils influenced by climate? 2) To what degree and how efficiently do microbes use energy and mineral resources in the field? To answer the first question we measured the seasonal climate variables of temperate and moisture and the respiration response of various litters, organic soils and litterbag weight losses (Flanagan and Veum 1974). Using climatic data we attempted to predict microbial activities in decaying tundra substrates and relate them to the turnover of organic matter and plant productivity. This could be considered a more holistic approach because it concentrated on microbial activities (respiration) and ignored microbial taxa. As a result of this work, three simulation models were constructed relating microbial activities quantitatively and unambiguously to the environmental phenomena that govern them. One model (GRESP, Flanagan and Bunnell 1976, Bunnell et al. 1977a) relates the respiratory response of different kinds of litter to changing climatic conditions. The second (DECOMP, Flanagan and Bunnell 1976, Bunnell et al. 1977b) expresses the respiration rate as a function of the quality of the litter. The third (ABISKO II, Bunnell and Scoullar 1975) integrates the effects of changing meteorological condition and litter quality within the ecosystem. These models document relationships between weight loss and microbial activities, as they are influenced by abiotic variables and litter quality and the relationship of the processes of microbial populations to primary production and turnover of organic matter. Although the development of these models are based on research in tundra (Flanagan and Bunnell 1976), their predictive abilities have also been tested for conditions found in taiga forest and moors (Bunnell et al. 1977a, Bunnell and Scoullar 1981). Thus the models represent the inference of causation, involving both smaller and larger scales of organization than that of the measurements and observations that went into the formation of the models. The second question (energy and mineral processing by microbes) was approached in my laboratory from what is commonly perceived as a reductionist approach. I isolated the major species components of the microflora and subjected them to experiments on their growth and respiration, on field substrate (cellulose, hemicellulose, sugars, etc.) and evaluation of their yield and maintenance demands on different substrates at varying temperatures (consistent with field conditions) (Flanagan and Scarborough 1974, Flanagan 1978. Having established yield coefficients and maintenance demands in the laboratory (Flanagan 1978) these numbers were used, together with estimates of the total fungal biomass of the study sites, to compute the amount of organic matter necessary to maintain an average standing crop of microbes. The equation used was that of Marr et al. (1963). The remainder of the annual input from plants or the measure of decay in g m-2 yr-' was used as a numerator to be divided by an average microbial standing crop (denominator) to calculate potential annual microbial turnover (productivity). Using the measured yield coefficient (Y) with the calculated productivity gave an estimate of average CO2 emission from decaying organic matter. The potential number of microbial generations was estimated using the equation of Gray and Williams (1971). This equation includes a term allowing for recycling of dead microbial tissues. The microbial biomass (per square meter) in the tundra system of Alaska was 18.1 g and the yield efficiency (g microbial tissue per g of substrate) was 0.35. Based on an average (litter)