1. The annual input, contribution to the diet of salmonids, and quantitative input of terrestrial invertebrates to four reaches with contrasting forest (n=2) and grassland riparian vegetation (n=2) were compared in a Japanese headwater stream.
2. The annual input of terrestrial invertebrates falling into the forest reaches (mean±1 SE=8.7×103±0.3×103 mg m−2 year−1) was 1.7 times greater than that in the grassland reaches (5.1×103±0.8×103 mg m−2 year−1), with clear seasonality in the daily input of invertebrates in both vegetation types. The daily input, however, differed between the vegetation types only in summer, when it rose to a maximum in both vegetation types.
3. Fish biomass also differed among the seasons in both vegetation types, being less in the grassland reaches. The contribution of terrestrial invertebrates to the salmonid diet in the forest and grassland reaches was 11 and 7% in spring, 68 and 77% in summer, 48 and 33% in autumn, and 1 and 1% in winter, respectively. The prey consumption rate of fish, which was similar between the vegetation types, increased with stream temperature and was highest in summer. Terrestrial invertebrates supported 49% (mean±1 SE=5.3×103±0.4×103 mg m−2 year−1) of the annual, total prey consumption (10.9×103±1.7×103 mg m−2 year−1) by salmonids in the forest and 53% (2.0×103±0.3×103 mg m−2 year−1) (3.8×103±0.6×103 mg m−2 year−1) in the grassland reaches.
4. Salmonids were estimated to consume 51 and 35% of the annual total (falling plus drift) input of terrestrial invertebrates in the forest and grassland reaches, respectively. The input of terrestrial invertebrates by drift, however, was almost equal to the output in both vegetation types, suggesting that the reach-based, in-stream retention of terrestrial invertebrates almost balanced these falling in.
5. Difference in the riparian vegetation, which caused spatial heterogeneity in the input of terrestrial invertebrates, could play an important role in determining the local distribution of salmonids.