Nutrient-use efficiency (NUE) within forests of the mangroves Rhizophora stylosa and Avicennia marina was estimated in and Western Australia using litter fall rates and rates of leaf CO2 exchange. Litter fall rates ranged from 9.8 to 34.4 t DW ha(-1) y(-1) but equated to only 13-41% (mean = 30%) of net canopy primary production. Foliar N:P ratios were in most instances >= 16, suggesting P limitation. NUE for N based on litter fall rates were significantly less (NUEL = 167-322 g DW g(-1) N) than those based on photosynthesis measurements (NUEP = 234-448 g DW g(-1) N), suggesting that NUE estimates for nitrogen based on litter fall data are underestimates. NUEp estimates for N were significantly greater for R. stylosa than for A. marina. NUE for P were not significantly different, with NUEL ranging from 2905 to 5053 g DW g(-1) P and NUEP ranging from 1632 to 4992 g DW g(-1) P. Both sets of NUE are at the higher end of the range of estimates calculated for most other forests and equivalent to those for wet tropical mangroves. These arid-zone trees live in low-nutrient habitats, but it appears that selection on components of NUE (i.e. traits that reduce nutrient loss) rather than on NUE itself equates to a lack of clear patterns in NUE between different environments, emphasizing the flexible nature of nutrient allocation in woody plants. NUE in R. stylosa correlated inversely with mature leaf N and P content, implying that NUE in this species is maximized by the synthesis of low-nutrient leaves, i.e. a nutrient retention strategy, whereas such does not appear to be the case for A. marina. This strategy translates into a direct advantage in terms of net primary productivity for R. stylosa. This idea is supported by evidence of longer nutrient residence times for R. stylosa than for A. marina. (c) 2004 Elsevier B.V. All rights reserved.