SPATIAL VARIABILITY IN THE POTENTIAL FOR SYMBIOTIC N-2 FIXATION BY WOODY-PLANTS IN A SUBTROPICAL SAVANNA ECOSYSTEM

1. Root infection by symbiotic N-2-fixing Frankia and Rhizobium strain s was quantified in relation to light and soil properties for seedling s of 12 woody species from a subtropical savanna in southern Texas, US A. 2. None of four rhamnaceous species nodulated, despite the fact tha t bioassays with a known actinorhizal species yielded 13 nodules per s eedling. Celtis pallida (Ulmaceae), Acacia greggii and Acacia berlandi eri (Leguminosae) also failed to nodulate even though field population s of these species were characterized by high (2.7-4.2%) foliar nitrog en concentration. 3. Infective rhizobia occurred in all soils studied regardless of soil depth, distance from a host plant or type of plant cover. Plant growth in N-free media and acetylene reduction activity s uggested that all nodules were capable of N-2-fixation. 4. The extent of nodulation varied by species. However, nodulated seedlings were tal ler, produced more biomass and allocated less biomass to root systems than their non-nodulated counterparts. 5. Numbers of nodules on seedli ngs of Prosopis glandulosa, the dominant woody species in this subtrop ical savanna and throughout the south-western USA, were reduced by low light (15% full sunlight) regardless of soil N level; at medium and f ull sunlight nodule biomass expressed as a fraction of whole plant bio mass decreased with increasing soil N. Nodulation of held-grown P. gla ndulosa appears to be ephemeral, apparently varying with changes in so il moisture. 6. Nodulation and N-2 fixation among woody legumes in sub tropical savannas can occur across a broad range of soil conditions an d depths with significant impacts on local and regional N-cycles. 7. F ield levels of foliar N in species that failed to nodulate in the labo ratory were comparable to or greater than those in species capable of nodulation, suggesting that leaf N is not a reliable indicator of N-2 fixation.