Towards an ecological understanding of biological nitrogen fixation

N limitation to primary production and other ecosystem processes is widespread. To understand the causes and distribution of N limitation, we must understand the controls of biological N fixation. The physiology of this process is reasonably well characterized, but our understanding of ecological controls is sparse, except in a few cultivated ecosystems. We review information on the ecological controls of N fixation in free-living cyanobacteria, vascular plant symbioses, and heterotrophic bacteria, with a view toward developing improved conceptual and simulation models of ecological controls of biological N fixation. A model (Howarth et al. 1999) of cyanobacterial fixation in lakes (where N fixation generally increases substantially when N:P ratios are low) versus estuaries (where planktonic N fixation is rare regardless of N:P ratios) concludes that an interaction of trace-element limitation and zooplankton grazing could constrain cyanobacteria in estuaries and so sustain N limitation. Similarly, a model of symbiotic N fixation on land (Vitousek & Field 1999) suggests that shade intolerance, P limitation, and grazing on N-rich plant tissues could suppress symbiotic N fixers in late-successional forest ecosystems. This congruence of results raises the question - why do late-successional tropical forests often contain many potentially N-fixing canopy legumes, while N fixers are absent from most late-successional temperate and boreal forests? We suggest that relatively high N availability in lowland tropical forests permits legumes to maintain an N-demanding lifestyle (McKey 1994) without always being required to pay the costs of fixing N. Overall, both the few simulation models and the more-numerous conceptual models of ecological controls of biological N fixation suggest that there are substantial common features across N-fixing organisms and ecosystems. Despite the many groups of organisms capable of fixing N, and the very different ecosystems in which the process is important, we suggest that these common controls provide a foundation for the development of regional and global models that incorporate ecological controls of biological N fixation.