The importance of terrestrial-aquatic linkages was evaluated by a large-scale, 3-year exclusion of terrestrial leaf litter inputs to a forest stream. Exclusion of leaf litter had a strong bottom-up effect that was propagated through detritivores to predators. Most invertebrate taxa in the predominant habitat declined in either abundance, biomass, or both, compared with taxa in a nearby reference stream. However, fauna in moss habitats changed little, indicating that different food webs exist in habitats of different geomorphology. Thus, the ecosystem-level consequences of excluding detrital inputs to an ecosystem were demonstrated. Inputs of riparian detritus are essential for conservation or restoration of diverse stream food webs. Detritus, or dead organic matter (I), is the major carbon pathway in most ecosystems: 70 to 90% of all primary production eventually enters the detrital food web (2). The addition of detritus to food webs complicates classical plant-herbivore-predator relationships (3). Indeed, in most headwater streams draining forests in eastern North America, inputs of detritus from the surrounding forest exceed within-stream primary production (4). One of the basic tenets of stream ecology for more than two decades has been the importance of terrestrial-aquatic linkages (5). Although details about linkages between detritivores and detritus processing in streams are well known (2, 6, 7), there is little direct evidence supporting the importance of terrestrial detrital inputs and ecosystem productivity (8) and it is limited to short-term studies in artificial channels (9). It has also been suggested that several generations of consumers are required to detect responses to detrital manipulations (10). We studied the role of detritus in ecosystem productivity by excluding inputs of terrestrial litter to a 180m-long headwater stream, using an overhead canopy and a lateral fence for 3 years (11). We evaluated the impact of the basal resource (terrestrial litter inputs) in this forest stream on abundance, biomass, and production of animals. In addition to examining numerical abundances of populations, we calculated secondary production as the flow (or flux) of mass • area" 1 • time" 1 , which incorpo

We define disturbance in stream ecosystems to be: any relatively discrete event in time that is characterized by a frequency, intensity, and severity outside a predictable range, and that disrupts ecosystem, community, or population structure and changes resources or the physical environment. Of the three major hypotheses relating disturbance to lotic community structure, the dynamic equilibrium hypothesis appears to be generally applicable, although specific studies support the intermediate disturbance hypothesis and the equilibrium model. Differences in disturbance frequency between lentic and lotic systems may explain why biotic interactions are more apparent in lakes than in streams. Responses to both natural and anthropogenic disturbances vary regionally, as illustrated by examples from the mid-continent, Pacific northwest, and southeastern United States. Based on a generalized framework of climatic-biogeochemical characteristics, two features are considered to be most significant in choosing streams for comparative studies of disturbance: hydrologic regimes and comparable geomorphology. A method is described for quantifying predictability of the hydrologic regime based on long-term records of monthly maximum and minimum stream flows. Different channel forms (boulder and cobble, alluvial gravelbed, alluvial sandbed) have different responses to hydrologic disturbance from spates. A number of structural and functional components for comparing disturbance effects within regions and across biomes are presented. Experimental approaches to studying disturbance involve spatial-scale considerations, logistic difficulties, and ethical questions. General questions related to disturbance that could be addressed by stream ecologists are proposed.

Ahstnict. Estimation of invertebrate biomass is a critical step in addressing many ecological questions in aquatic environments. Length-dry mass regressions are the most widely used approach for estimating benthic invertebrate biomass because they are faster and more precise than other methods. A compilation and analysis of length-mass regressions using the power model, M (mass) = n L (length)h, are presented from 30 y of data collected by the authors, primarily from the southeastern USA, along with published regressions from the rest of North America. A total of 442 new and published regressions are presented, mostly for genus or species, based on total body length or other linear measurements. The regressions include 64 families of aquatic insects and 12 families of other invertebrate groups (mostly molluscs and crustaceans). Regressions were obtained for 134 insect genera (155 species) and 153 total invertebrate genera (184 species). Regressions are provided for both body length and head width for some taxa. In some cases, regressions are provided from multiple localities for single taxa. When using body length in the equations, there were no significant differences in the mean value of the exponent b among 8 insect orders or Amphipoda. The mean value of b for insects was 2.79, ranging from only 2.69 to 2.91 among orders. The mean value of b for Decapoda (3.63), however, was significantly higher than all insects orders and amphipods. Mean values of n were not significantly different among the 8 insect orders and Amphipoda, reflecting considerable variability within orders. Reasons for potential differences in b among taxa are explained with hypothetical examples showing how b responds to changes in linear dimensions and specific gravity. When using head width as the linear dimension in the power model, the mean value of b was higher (3.11) than for body length and more variable among orders (2.8-3.3). Values of b for Ephemeroptera (3.3) were significantly higher than those for Odonata, Megaloptera, and Diptera. For those equations in which ash-free dry mass was used, % ash varied considerably among functional feeding groups (3.3-12.4%). Percent ash varied from 4.Oo/o to 8.5% among major insect orders, but was 18.9% for snails (without shells). Family-level regressions also are presented so that they can be used when generic equations are unavailable or when organisms are only identified to the family level. It is our intention that these regressions be used by others in estimating mass from linear dimensions, but potential errors must be recognized.

The diversity of life in headwater streams (intermittent, first and second order) contributes to the biodiversity of a river system and its riparian network. Small streams differ widely in physical, chemical, and biotic attributes, thus providing habitats for a range of unique species. Headwater species include permanent residents as well as migrants that travel to headwaters at particular seasons or life stages. Movement by migrants links headwaters with downstream and terrestrial ecosystems, as do exports such as emerging and drifting insects. We review the diversity of taxa dependent on headwaters. Exemplifying this diversity are three unmapped headwaters that support over 290 taxa. Even intermittent streams may support rich and distinctive biological communities, in part because of the predictability of dry periods. The influence of headwaters on downstream systems emerges from their attributes that meet unique habitat requirements of residents and migrants by: offering a refuge from temperature and flow extremes, competitors, predators, and introduced species; serving as a source of colonists; providing spawning sites and rearing areas; being a rich source of food; and creating migration corridors throughout the landscape. Degradation and loss of headwaters and their connectivity to ecosystems downstream threaten the biological integrity of entire river networks.