How many species of prokaryotes are there?

T microorganisms classified in the two prokaryotic domains of the tree of life, Bacteria and Archaea, possess immense metabolic diversity, and their activities are critical in processes ranging from sewage treatment to regulating the composition of the atmosphere. Especially in light of the rate of modern climate change, it is essential to understand how microbial communities affect ecosystem functioning and how human activities, such as agriculture, waste management, and climate modification, affect microbial communities. Thus discovering and understanding the diversity of microbial communities (the number of species and their relative abundances) is a high priority in ecology. In this issue of PNAS, Curtis et al. (1) address what is therefore one of the most fundamental questions in microbial ecology—how many species of prokaryotes are there in nature? Only in the last 10–15 years has it even been possible to pose the question and hope realistically for an answer in the case of prokaryotes. Less than 30 years ago, the answer to the even more fundamental question ‘‘How many individuals are there?’’ was revised in such a way as to change the entire focus of environmental microbiology. So to appreciate the significance of the question and answer provided by Curtis et al. (1), it is useful to review that recent history briefly. Before the mid-1970s, microbial ecologists assessed the population size of bacteria in soils, sediments, and natural waters by culturing the microbes and counting the number of colonies that grew on nutrient agar plates. For seawater, the cultivable prokaryotic population size was a few hundred cells per milliliter (2, 3). That was an almost inconsequential number relative to the thousands or tens of thousands of planktonic algal cells that could be seen (literally, with a microscope) in the same milliliter of water. The primary importance then ascribed to environmental bacteria was their potential as pathogens, and research focused on survival of pathogens in natural environments. But when researchers saw millions of cells per milliliter, or per gram of soil or sediment, with electron microscopy (4) and epifluorescence microscopy [using DNA-binding fluorochromes such as acridine orange and later 4 ,6-diamidino-2phenylindole (DAPI)] (5, 6), they realized that bacteria must have much broader and potentially more important roles in natural systems. Those additional cells could not all be pathogens, but microbial ecologists were stymied in their efforts to identify them by the very fact that they did not grow on plates. So the answer to the question ‘‘How many are there?’’ led to intense interest in ‘‘Who are they?’’ and ‘‘What are they doing?’’ The answer to ‘‘What are they doing?’’ was provided by sensitive stable isotope and radiotracer methods: prokaryotes play essential roles in the primary production and consumption of organic matter and the cycling of nutrient elements in the modern environment and were perhaps even more important in the evolution of the atmosphere and hydrosphere before the appearance of eukaryotes. The answer to ‘‘Who are they?’’ became infinitely more approachable with the advent of molecular biological methods in environmental microbiology. Woese (7) first recognized the utility of the ribosomal RNA molecule, with its universal distribution, and its high conservation coupled with moderate variability, for constructing global phylogenies of all living things. He also introduced the now widely accepted threedomain tree of life (8), in which prokaryotes constitute two branches (Bacteria and Archaea) and Eukarya the third. On the basis of a few hard-won sequences, the first ‘‘universal’’ primers for use in the polymerase chain reaction (PCR) were designed and used to pluck the 16S ribosomal RNA genes of uncultivated organisms right out of the soil and water of their natural environments. Once those genes were sequenced, their evolutionary relationships to cultivated and other uncultivated organisms could be determined. ‘‘Who are they?’’ became synonymous with ‘‘Where do they fall in the 16S rRNA-based tree of life?’’ Diversity surveys based on cultivation had identified 10 or 12 major divisions within the Bacteria and two or three in the Archaea (9). There are now more than 40 major bacterial divisions recognized (10) and 12 or more in the Archaea (11). The tree is still growing and not just at the twig level—just last month (May 2002), a new phylum of Archaea was discovered at a hydrothermal vent, its rRNA sequence so different from known groups that it could not be detected with the ‘‘universal’’ probes (12). Perhaps the most astounding finding, and one that will keep microbial ecologists busy for many years, is that most of the sequences retrieved from the environment without cultivation are not represented by any cultivated organisms. Entirely new microbial worlds have always been out there, beneath our feet, in our water and air. Many remarkable and fascinating new organisms have been discovered (on the basis of their 16S rRNA sequence) in extreme environments such as hydrothermal vents and hot springs, desert sand, and Antarctic oceans, but the smallest drop of temperate seawater or a grain of agricultural soil will also yield myriad 16S rRNA sequences that are new to science. As these data on the immense diversity of life began to accumulate, the question that Curtis et al. (1) attempt to answer became unavoidable. ‘‘How many different kinds can there possibly be?’’ If with every milliliter of water or gram of sediment we discover new prokaryotic sequences, is there any limit to the diversity in nature? If we cannot enumerate all of the different sequences in one sample, how can we compare that sample to another to ask whether the community compositions of the two samples are different? The calculation of community diversity by any conventional diversity index requires two fundamental pieces of information: the number of species and the number of individuals in each species. These data are often presented in a species abundance curve in which the number of species is plotted vs. the number of individuals per species. Although microbial ecologists have made much progress toward identifying large numbers of species, the quantitative information on how many of each is present is still an almost impossible goal. Curtis et al. (1) show that by assuming a log-normal distribution (i.e., most species have an intermediate number of individuals and few species have very small or very large populations), the area under the curve can be used to estimate the number of species from the total number of individuals. It is then possible to relate diversity of prokaryotic communities to the ratio of two potentially