Laboratory Stream Research Articles

Longitudinal effects of herbivory on lotic periphyton assemblages

SUMMARY 1. The longitudinal effects of herbivory on stream periphyton assemblages were examined in laboratory stream channels, each of which consisted of an upstream chamber, which either contained snail grazers or not, and downstream chambers, none of which contained grazers. Periphyton assemblages of two ages (0–21 days old and 21–42 days old) were sampled in both upstream and downstream chambers to detect proximate (i.e. localized) and longitudinal (i.e. downstream) effects of herbivory. 2. Both proximate and longitudinal effects were detected, although they differed in their impact on the periphyton assemblage. Periphyton biomass and cell accumulation were lower in grazed than in ungrazed upstream chambers throughout the experimental period. Accumulation rates on initially bare tiles were substantially higher downstream of grazed than of ungrazed chambers, but grazing had no effect on cell densities in established (21–42 day old) assemblages downstream. 3. Longitudinal effects of herbivory were not due to quantitative differences in the flux of propagules or nutrients from grazed and ungrazed chambers. Although not tested in this study, it is hypothesized that differences in the physiological condition of exported propagules may have contributed to differences in downstream colonization rates in grazed and ungrazed streams. 4. The magnitude of longitudinal impacts of herbivory and the importance of different causal mechanisms are predicted to vary depending on the standing crop and productive capacity of the periphyton assemblages as well as the consumptive demand of the herbivore guild.

Foraging and Evading Predators: The Effect of Predator Species on a Behavioural Trade-Off by a Lotic Mayfly

We performed laboratory experiments to test the hypothesis that predator species influenced a forager's trade-off between predation risk and food reward. Larvae of the stream mayfly, Baetis tricaudatus Dodds, were provided with simultaneous access to feeding patches in laboratory streams that differed in predation risk (safe, risky) and food reward (low, high). Predation risk was imposed using either: (1) a fish predator, the longnose dace, Rhinichthys cataractae (Valenciennes), (2) a stonefly predator, Claassenia sabulosa (Banks), and (3) both predators is risky patches at the same time. We hypothesized that predator species would influence the behavioural trade-off between predation risk and food reward because mortality risk on Baetis from Rhinichthys exceeds that from Claassenia

Direct and Indirect Effects of Fish Predation on the Replacement of a Native Crayfish by an Invading Congener

In Ohio streams, the crayfish Orconectes rusticus is replacing O. sanborni, and herein we test how predators influence this replacement. In a field survey, crayfish were scarce when fish were abundant, suggesting that predators can adversely affect these prey. In laboratory experiments, we examined underlying mechanisms for this inverse relationship; specifically, we tested how crayfish species, adult aggression, and habitat heterogeneity influenced the predator–prey interaction. In a laboratory stream, smallmouth bass (Micropterus dolomieu) ate similar numbers of equal-sized O. rusticus and O. sanborni, but when sizes mimicked those in the field (i.e., O. rusticus 4 mm > O. sanborni), fewer O. rusticus were eaten. Fish also reduced juvenile activity and behaviors whereas adult aggression increased the frequency of these risky responses. More affected by adult crayfish, O. sanborni should suffer disproportional predation where adults and juveniles interact. Thus, fish predators should increase replacement rates and adult aggression should further accelerate this process. Manifested through crayfish size, both indirect and direct predator effects contribute to the replacement of O. sanborni by O. rusticus.

Food Limitation of Two Lotic Grazers: Quantity, Quality, and Size-Specificity

A food augmentation experiment demonstrated that growth and lipid accumulation by grazing snails (Elimia clavaeformis) and caddisflies (Neophylax etnieri) from a snail-dominated stream in eastern Tennessee were strongly food limited. Elimia (two size classes) and third-instar Neophylax from White Oak Creek were given one of four diets in laboratory streams: (1) low-biomass periphyton (from White Oak Creek), (2) high-biomass periphyton (from an unshaded, nutrient-enriched stream that lacked effective grazers), (3) low-biomass periphyton and leaves, and (4) high-biomass periphyton and leaves. Small Elimia grew 2× faster and large Elimia grew 5× faster on high-biomass periphyton than on their normal diet of low-biomass periphyton. Neophylax grew 7× faster on the high-biomass periphyton. The high-biomass periphyton also produced significantly higher ratios of lipid to ash-free dry mass for both grazers. Leaf additions did not increase growth or lipid accumulation, even for Elimia, a feeding generalist. Measur...

Lotic Ecosystem Response to a Chlorine Disturbance.

The influence of repeated chlorine additions on the resistance and resilience of laboratory stream ecosystems was investigated. Prior to the disturbance, four treatments were imposed on eight laboratory streams in a 2 x 2 factorial design, resulting in two replicate streams per treatment. Treatments included two circulation regimes (once-through or 98% recirculated) and two levels of grazer density (1000 individuals/m2 or 0 individuals/m2 , using the snail Elimia clavaeformis). Resistance of periphyton biomass to an initial set of chlorine additions was greater in once-through than recirculated streams. Periphyton resistance following a second set of chlorine additions was greater in streams without snails compared to those with snails. This appeared to be a biomass-mediated response; the greater biomass levels in no-snail streams provided a larger buffer to the oxidizing action of chlorine. Resilience of most structural parameters and area-specific carbon-fixation rates was not influenced to a significant degree either by flow regime or snail density. Resilience of chlorophyll-specific carbon fixation and exoproteolytic activity (an index of microbial activity) was greater in recirculated streams compared to once-through streams, but only when snails were present. Over the entire experimental period (15 wk), most structural parameters and area-specific carbon fixation rates were significantly greater in (1) streams without snails compared to those with them and (2) streams with once-through circulation compared to those that were recirculated, but only for those streams with snails. Our chlorine additions appeared to have relatively little influence on snails, and snail growth over the experiment was significantly greater in once-through compared to recirculated streams. These data are consistent with other findings that total residual chlorine concentrations of <1.0 mg/L can significantly affect periphyton structure and function. However, nutrient cycling rates appeared to be little affected by the chlorine perturbation, which allowed recirculated systems to recover faster than predicted. The findings from this study suggest the periphyton mat physiognomy and integrity strongly influence system resistance (due to buffering capacity) and resilience (because of nutrient cycling).

Predation by rock bass on other stream fishes: experimental effects of depth and cover

Effects of water depth and cover availability on predation rates by adult rock bass, Ambloplites rupestris, on juvenile central stoneroller, Campostoma anomalum, pumpkinseed, Lepomis gibbosa, and fantail darter, Etheostoma flabellare, were measured in a laboratory stream. A predation rate experiment and a prey activity experiment were conducted. In the first experiment, each prey type shared experimental chambers with predatory rock bass under all combinations of two depths (shallow and deep) and two cover levels (absent and present). Predation rates after 72 h were greater in deep (35–38 cm) than in shallow (7–10 cm) water for fantail darter and two sizes of central stoneroller, but not for pumpkinseed. Presence of cover (opaque plastic tubes) on the stream bottom had no direct effect on predation rates, which were higher for pumpkinseed and small stoneroller (35–64 mm total length) than for fantail darter and large stoneroller (70–89 mm total length). In the second experiment, diel patterns of small stoneroller activity were monitored under all combinations of two water depths and three levels of predation risk. Small stonerollers were more active during the day and in the absence of rock bass, but were not affected directly by water depth. My results suggest that effects of habitat features (e.g., depth, cover) on predator-prey interactions vary according to the natural history and behavior of particular prey and predators. Future research should integrate habitat-specific responses of prey to predation risk into models that predict the distribution of prey among available habitats.

Interactions between Native Brook Trout and Hatchery Brown Trout: Effects on Habitat Use, Feeding, and Growth

Habitat use, feeding behavior, and growth of native brook trout Salvelinus fontinalis and hatchery brown trout Salmo trutta were compared in an experiment in which the two species were held alone and together in laboratory stream channels at 14°C. Microhabitat location and vertical distribution of brook trout within the stream channels shifted in the presence of brown trout, and the frequency with which brook trout initiated aggressive interactions declined. Prey capture rates were higher for brown trout than for brook trout in single- and mixed-species trials, and higher for both species when they occurred alone than when they were together. Brook trout and brown trout in single-species trials maintained weight, but did not grow. In the presence of brown trout, however, brook trout lost weight. Instantaneous growth rates of brown trout in mixed species trials were positive. In the presence of brown trout, 33% of the brook trout contracted the fungus Saprolegnia sp. and died; brown trout were never infected nor were brook trout in single-species trials. If these changes in behaviors and growth rates extend to natural settings in which brook trout and brown trout co-occur, they may contribute to an explanation for the observed decline of brook trout abundance in eastern North America.

Interactions between fungi, bacteria and beech leaves in a stream microcosm.

Immigration and colonization of isolates of naturally occurring stream bacteria and hyphomycetes on beech leaves were studied in a laboratory stream microcosm. Fungal spores were more successful immigrants, especially on new leaves, than bacteria, which were more repelled than attracted by the substrate. Fewer bacteria immigrated to older leaves than to new, and bacteria multiplied faster in water than on leaves. Fungi and bacteria showed synergistic relationships so that each group grew significantly faster in presence of the other group. If one considers, differences in immigration, colonization and synergism patterns, fungal mycelia doubled about 10 times faster than bacterial cells which might explain the dominance of fungi usually found on leaves in early decay. The individual fungal species could be assigned to one of three colonization groups; one of fugitive species, preceding a second group of species that grew from rarity to dominance, and a third group of very slow colonizers. The leachate was fractionated in different molecular size classes by gel chromatography, and the fraction around 2500D in the new leaf leachate was associated with a high concentration of polyphenols. High-pressure liquid chromatography (HPLC) and gas chromatography-mass spectrometry (GC-MS) revealed the presence of 16 phenolic acids in the new leaf leachate at concentrations ranging from < 1 to 640 μg l-1. All fungi except the species with the slowest growth rate grew faster on leachate with the fraction around 2500D removed, and the density of bacteria was significantly reduced when pure stream water was supplemented with compounds from the same fraction.

The water velocity preferences of Austrosimulium furiosum and Simulium ornatipes (Diptera: Simuliidae) larvae, and the implications for micro-habitat partitioning

The aquatic larvae of two simuliid species, Austrosimulium furiosum (Skuse) and Simulium ornatipes Skuse, which often occur together in Victorian streams, were shown in laboratory experiments to have preferences for different water velocities: larvae of A. furiosum preferred water velocities of 0.2–0.3 m s−1, and S. ornatipes preferred water velocities of 0.9–1.3 m s −1 . Final instar larvae of both species selected ‘slow’ water speeds of less than 0.25 m s −1 prior to pupation. Flow patterns around a cylinder in a laboratory stream were mapped, and the distribution of A. furiosum larvae within the wake, paired vortices and horseshoe vortex was recorded. Larvae selected areas with suitable water velocities and aligned with the flow, providing flow visualization of micro-current speed and direction. The potential for micro-habitat partitioning is discussed in relation to benthic sampling strategies.

Effects of predation risk on foraging behavior of juvenile rainbow trout (Oncorhynchus mykiss)

The effects of a simulated avian predator attack on the food intake and foraging behavior of wild juvenile rainbow trout, Oncorhynchus mykiss, were examined in laboratory streams. Exposure to a simulated predator attack reduced the amount of food eaten for several minutes after an attack. The percent reduction in foraging rate was not related to the base-line foraging rate of the fry, a measure of feeding motivation, indicating that fry were not balancing risk and hunger. Fry at risk responded differently from safe fry to individually presented food items. Following exposure to a simulated attack, fry initially ate fewer items than did nonattacked fry, but over the course of the 10-min experiment, fry at risk ate as much as safe fry, which suggests a degree of compensation for lost foraging opportunities. I hypothesized that fry at risk from overhead avian predators should avoid risk by temporarily suspending foraging, since the risk is acute, short-term suspension of foraging has low energetic costs, and compensation is often possible. Fry at risk from piscivorous fish should be sensitive to predator behavior and should balance risk and energy intake, since risk is more likely to be sustained, and prolonged suspension of foraging has a high energetic cost.

Behavioural interaction and habitat use in juvenile chinook, Oncorhynchus tshawytscha, and coho, O. kissutch, salmon

Young anadromous chinook salmon and coho salmon reside sympatrically in many streams and rivers tributary to the North Pacific Ocean. This study tested the hypothesis that behavioural domination by coho salmon, which reside in freshwater for at least 1 year before seaward migration, promotes variability in habitat use and in the duration of freshwater residence by chinook salmon, which migrate seaward in their first year of life (‘ocean-type’) or after a year or more in freshwater (‘stream-type’ chinook). In laboratory stream channels coho behaviourally dominated chinook; they spent more time attacking chinook than vice versa and numerically dominated upstream sections of the channels where food was introduced. Principal components analysis was used to study habitat use by coho and chinook in four natural streams. In two streams where they were sympatric, coho and chinook used different habitats: coho used slow water, deep ‘pool’ areas while chinook used faster water, shallow ‘riffle’ areas. In two streams where chinook were allopatric, however, they made greater use of pool habitats than when in sympatry. Coho and chinook introduced into a controlled-flow stream channel also preferred pool and riffle habitats respectively. The differences in habitat use in the stream channel were probably the result of species-specific differences in habitat preference and not behavioural interaction because chinook made greater use of riffle habitats whether they were introduced alone or with coho. Co-introduction with coho had no significant effect on the degree of emigration by chinook from the stream channel compared to chinook introduced alone at the same density. While coho may socially dominate chinook in simple poollike habitats, differences in habitat preference between the species that have developed during sympatric evolution probably minimize the extent to which coho influence the duration of freshwater residence by chinook.

Species Interactions in Cave Stream Communities: Experimental Results and Microdistribution Effects

-The interspecific interactions of three pairs of cave isopods and amphipods, previously thought to be competitors, were examined by analyzing washouts and losses from laboratory streams and from manipulation experiments in streams of Organ Cave, West Virginia. None of the pairs were in fact competitors. The isopod Caecidotea holsingeri was an amensalist of the amphipod Stygobromus spinatus in the laboratory, but the two did not interact in the field. The amphipod Gammarus minus was a predator on C. holsingeri both in laboratory streams and in the field, and C. holsingeri was a competitor of G. minus, at least in laboratory streams. Stygobromus emarginatus was an amensalist of G. minus in laboratory streams and in the field; but in the field G. minus was a commensalist of S. emarginatus. This combination of commensalism and amensalism produces the dynamics of a predator-prey system. Interactions in both laboratory streams and cave streams were asymmetric and often had characteristics of several kinds of interspecific interactions. Negative interactions dominated in laboratory streams while positive interactions were more common in the field experiments. Negative interactions had an effect on microdistribution only at the level of individual stones in a riffle. Positive interactions may contribute to the positive association of some species at larger scales.

Interactive Effects of Nutrient Reduction and Herbivory on Biomass, Taxonomic Structure, and P Uptake in Lotic Periphyton Communities

Four treatments were imposed on eight laboratory streams in a factorial design to examine the roles of nutrient reduction and herbivory on periphyton communities. Treatments included two flow regimes (once-through flow or 90% recirculated water) and two levels of grazer density (1000 or 0∙m−2, using the snail Elimia clavaeformis). Periphyton biomass was significantly greater in streams without snails than in those with them, but water supply had no overall significant effect on biomass, even though inorganic P and N concentrations were significantly lower in recirculated than in once-through streams. Areal-specific P uptake rates (measured with 33P) were significantly greater on two dates in no-snail streams compared with snail streams, presumably because of the greater biomass levels in the former systems. Differences in biomass-specific P uptake rates were not significantly affected by either grazer density or water supply. Relative abundances of most algal species were unaffected by the water supply treatment, although percent biovolume of two Epithemia species was greater in no-snail, recirculated than in no-snail, once-through streams. Grazing activity dramatically reduced the percent biovolume of species with upright growth forms, resulting in dominance by species with prostrate growth forms.

Effects of Hydraulic Refuge and Irradiance on Grazer-Periphyton Interactions in Laboratory Streams

A 54-d experiment was conducted in laboratory streams to examine relationships between temporal and spatial patterns of algal development and grazing behavior of the snail Juga silicula in a patchy physical environment. Each stream contained nine channel-spanning blocks (22.5 × 22.5 × 4 cm) spaced 22.5 cm apart and was illuminated with a photon flux density of 375, 90 or 20 μ E m -2 s -2 . We monitored periphyton succession and the number of snails in fast current regimes on tops of blocks (exposed substrata) and in slower current regimes on surfaces recessed between blocks (sheltered substrata). At the end of the experiment, algal biomass in streams illuminated with high irradiance was approximately 80× and 8× greater than in streams with low and intermediate irradiance, respectively. Growth rates of Juga in streams with high and intermediate irradiances were similar and approximately 15× greater than rates at low irradiance. At all irradiances, snail densities were higher and algal bi...

Survival of fecal coliforms and fecal streptococci in storm drain sediment

This study defined major factors influencing the survival of the bacterial indicators, fecal coliforms (FC) and fecal streptococci (FS), in separate storm drain sediments during dry weather periods. FC and FS abilities to survive were determined by field sampling and experimental treatments of laboratory stream models. The effects of interspecific competition, antagonism, and predation on FC and FS survivals were determined using recirculating models that contained either untreated drain water and sediment, cycloheximide-treated drain sediment, or autoclaved drain water and sediment seeded with drain isolates of FC and FS. Storm drain sediment FC and FS counts were stable at 10 5/100 ml during dry weather (4–6 days), with little or no bacterial supplementation from the water (10 1–10 2/100 ml). Untreated (control) model FC and FS populations stabilized at 10 3/100 ml for 7–9 days. In cycloheximide-treated models, FS and FC survived at 10 4 and 10 5/100 ml, respectively. In autoclaved-seeded models, FS and FC initial counts of 10 4/100 ml increased to and stabilized at 10 7–10 9/100 ml. Separate storm drain sediments serve as reservoirs of high concentrations of FC and FS during warm, dry weather periods. FC exhibited some ability to multiply in drain sediment. Native microfloral competition/antagonism (including bacterial predation) and protozoan predation are major biotic factors influencing FC and FS survivals. During warm, dry weather periods, abiotic factors were generally stable and had minimal affects on FC and FS survivals.

Resilience of Lotic Ecosystems to a Light‐Elimination Disturbance

Resilience of eight laboratory stream ecosystems was evaluated following a 92—d light elimination disturbance. Prior to the disturbance, four treatments (snails/once—through flow, snails/recirculated flow, no snails/once—through flow, no snails/recirculated flow) were imposed on the streams, resulting in systems with different biomass levels, nutrient concentrations, and recycling indices. Based on results from models of ecosystem response to disturbance, we hypothesized a priori that once—through streams would recover more quickly than recirculated streams within each grazing regime and that grazed streams would recover more quickly than ungrazed streams within each flow regime. Our results indicated that once—through streams did have a higher resilience than recirculated streams when snails were absent, but not when snails were present. Indeed, most parameters recovered faster in streams without snails than those with them, irrespective of flow regime, in contrast to our prediction. Despite the faster initial recovery rates in once—through than recirculated streams without snails, final biomass levels were similar between these streams. Measurements of phosphorus recycling indices suggested that higher rates of nutrient recycling near the end of the experiment in recirculated streams compensated for the lower inputs of new nutrients in the incoming water, allowing biomass to reach levels similar to those in once—through streams. Results from this empirical work are consistent with the theory that resilience is positively associated with the rate of nutrient input to a nutrient—limited system, but are inconsistent with the idea that resilience is positively associated with the addition of a herbivore trophic level to the system. Grazing intensity (i.e., density), as well as behavioral (e.g., vagility) and life history characteristics may need to be considered in future models.

Habitat Use by Steelhead and Coho Salmon and Their Responses to Predators and Cover in Laboratory Streams

Subyearling coho salmon Oncorhynchus kisutch and steelhead O. mykiss were placed together and separately in laboratory stream channels to assess their habitat use, interactions, and response to cover and predators. We studied the fish in six replicated riffle-pool-riffle units in the laboratory channels during 12 trials of 7 d each. In all test situations, we allowed the fish to emigrate from the test units. More fish remained in the units of the laboratory streams when both species were together than when either species was alone. The presence of larger fish (potential predators) caused both species to change habitats, regardless of cover available, and many left the test units or were eaten. Coho salmon used mainly the pools, whether stocked alone or together with steelhead; steelhead used both the riffles and pools when alone but shifted slightly toward increased use of the riffles when both species were present.

Role of Nutrient Cycling and Herbivory in Regulating Periphyton Communities in Laboratory Streams

In this study we examined the role of nutrient cycling and herbivory in regulating stream periphyton communities. Population, community, and ecosystem—level properties were studied in laboratory stream channels that had nutrient inputs reduced compared to channels where ambient nutrient levels were maintained. We reduced nutrient inputs in four of eight channels by recirculating 90% of the flow, whereas the other four channels received once—through flow of spring water. We examined the interaction between herbivory and nutrients by varying the number of snails (Elimia clavaeformis) among streams with different nutrient input (circulation) regimes. Reduction in nutrient input viar recirculation resulted in lower concentrations of nutrients in the water but did not result in significant differences in biomass, carbon fixation, or algal taxonomic composition. However, herbivory had large effects on these characteristics by reducing biomass and areal rates of carbon fixation and simplifying periphyton taxonomy and physiognomy. Lower rates of nutrient input significantly affected characteristics associated with nutrient cycling. Streams with reduced nutrient inputs had lower periphyton nutrient contents, higher ratios of total : net uptake of P from water, and higher rates of phosphatase activity than streams with ambient nutrient inputs. However, the effects of reduced nutrient input on cycling characteristics were reduced or eliminated by intense herbivory. Our results indicate that nutrient cycling can increase in response to reductions in nutrient input once sufficient biomass or detritus accumulates to facilitate cycling, thereby allowing biomass accrual and carbon fixation rates to be maintained at levels achieved at higher nutrient inputs. In streams with low nutrient input, cycling can meet a large fraction of the nutrient demand of the periphyton. Because our experimental design involved a reduction from ambient nutrient levels rather than the usual enrichment techniques (e.g., additions to water, nutrient—diffusing substrates) in aquatic ecosystems, we have been able to observe how biological structure and processes are influenced by external nutrient inputs at the low natural levels of most undisturbed streams.

Influence of Nutrients and Grazing on the Response of Stream Periphyton Communities to a Scour Disturbance

We examined the effect of nutrient input and grazing intensity on resistance and resilience of periphyton communities in laboratory streams to a scour disturbance. Previous theoretical studies using stream models indicated that resilience of nutrient-limited streams, as defined by the return time to the initial state, is strongly related to the residence time of nutrients in the system and to the presence of herbivores. To experimentally test this prediction we established two levels of nutrient input by altering the flow regime (once-through flow and 98% recirculated flow) and two levels of grazing intensity by the presence or absence of the snail Elimia clavaeformis, resulting in a 2 × 2 factorial design with two replicate streams per treatment. After the streams reached an approximately stable level of periphyton biomass, a scour disturbance was imposed on all streams. Measures of the resistance and resilience of the periphyton to this disturbance were made. The resistance of most structural characteristics of the periphyton was significantly higher in streams with snails than in streams without snails. Nutrient input had little effect on resistance of all periphyton characteristics. Resilience was measured by two parameters: rate of increase of a particular characteristic over time following the disturbance (r), and time to return to the pre-disturbance state (T_s). The experimental results on resilience were complex. In general, high grazing intensity resulted in low values of r for most periphyton characteristics, but also shorter T_s for some characteristics. The shorter T_s values were attributed to lower pre-disturbance biomass and metabolism in the streams with snails compensating for lower r for these characteristics following the disturbance. Contrary to model predictions, low nutrient input generally did not result in low resilience for most characteristics in the streams without snails, probably because increased nutrient recycling compensated for low inputs. However, for the streams with snails, resilience of most periphyton characteristics was lower in streams with low nutrient input compared with streams with higher nutrient input. This contrast in resilience suggests a positive feedback between biomass accrual and nutrient recycling that helps maintain high productivity despite low nutrient inputs during recovery of stream periphyton communities from disturbance.

Reduction of predation risk under the cover of darkness: Avoidance responses of mayfly larvae to a benthic fish.

Mayfly larvae of Paraleptophlebia heteronea (McDunnough) had two antipredator responses to a nocturnal fish predator (Rhinichthys cataractae (Valenciennes)): flight into the drift and retreat into interstitial crevices. Drift rates of Paraleptophlebia abruptly increased by 30 fold when fish were actively foraging in the laboratory streams but, even before fish were removed, drift began returning to control levels because larvae settled to the substrate and moved to areas of low risk beneath stones. This drifting response was used as an immediate escape behavior which likely decreases risk of capture from predators which forage actively at night. Surprisingly, drift most often occurred before contact between predator and prey, and we suggest that in darkness this mayfly may use hydrodynamic pressure waves for predator detection, rather than chemical cues, since fish forage in an upstream direction. Although drifting may represent a cost to mayfly larvae in terms of relocation to a new foraging area with unknown food resources, the immediate mortality risk probably out-weighs the importance of staying within a profitable food patch because larvae can survive starvation for at least 2 d. In addition to drifting, mayflies retreated from upper, exposed substrate surfaces to concealed interstitial crevices immediately after a predator encounter, or subsequent to resettlement on the substrate after predator-induced drift. A latency period was associated with this response and mayflies remained in these concealed locations for at least 3 h after dace foraging ceased. Because this mayfly feeds at night and food levels are significantly lower in field refugia under stones, relative to exposed stone surfaces, predator avoidance activity may limit foraging time and, ultimately, reduce the food intake of this stream mayfly.