PALMER, MARGARET A. Incorporating lotic meiofauna into our
Transcription
PALMER, MARGARET A. Incorporating lotic meiofauna into our
Oceanogr., 37(2), 1992,329-34 1 0 1992, by the American Society of Limnology and Oceanography, Inc. Limnol. Incorporating lotic meiofauna faunal transport processes into our understanding of Margaret A. Palmer Department of Zoology, University of Maryland, College Park 20742-44 15 Abstract This work represents the first detailed study using field data and flume experiments to evaluate the magnitude and pattern of meiobenthic drift in a stream, the evidence that both active and passive processes operate during entry and transport, and how meiofaunal transport in streams compares with what is currently known about the transport of lotic macrofauna and of the meiofauna and larval macrofauna of marine systems. Abundances of drifting meiofauna in Goose Creek, Virginia, were extremely high (up to 250,000 per 100 m3) and represented a significant amount of biomass when compared to literature biomass estimates for macrobenthic drift in other warmwater streams. Drift of the dominant taxa in this study (rotifers, early instar chironomids, oligochaetes, copepods) was flow-dependent. Flume experiments examining live and dead entry processes provided evidence that a critical threshold velocity existed (-9-l 2 cm s-l), above which the number of animals entering the water increased dramatically. This threshold was less than the critical threshold velocity of the streambed substrate (- 14-17 cm s-l), suggesting that whenever the streambed is scoured, potentially large numbers of meiofauna may enter the water. Despite the evidence that animals may enter the water passively due to entrainment by flow, active control over drift entry also operated since all four of the dominant taxa exhibited significant increases in drift at night. These diel drift patterns were reflected in reduced streambed abundances at night for chironomids and oligochaetes but not for copepods and rotifers. Additionally, in flume experiments which examined the return of live vs. dead drifting fauna to the bottom, copepods were able to influence their return to the bottom once in transport, a finding consistent with field observations. The downstream transport or drift of benthic stream invertebrates has been studied extensively (reviewed by Brittain and Eikeland 1988). Most of the studies have focused on macroinvertebrates, despite the fact that total stream drift may be underesti- mated dramatically by ignoring the smaller invertebrates (Sandlund 1982). In marine environments, meiofauna are known to be regularly transported in the water and their drift has been shown to have important ecological consequences including controlling their availability as prey to macrofauna and fish and influencing benthic recruitment Acknowledgments (Walters and Bell 1986; Palmer and Gust This research was supported by grants from the Na198 5). Because meiofauna are extremely tional Science Foundation (RI1 86-20422, BSR 90abundant in streams (e.g. Pennak and Ward 06002) and the Maryland Agricultural Experiment Sta1986), occur in the drift, and may be food tion. Thanks go to Peter Arensburger, Alexa Bely, Kathsources for macroinvertebrates and fish leen Berg, Dan Cristini, Chris Hakenkamp, Lies1 King, (O’Doherty 1988; Vadas 1990), their drift and Linda Mitchell for field and laboratory assistance. dynamics should influence their availability The University of Maryland stream group made suggestions that improved the manuscript as did Dave as prey and their recruitment, just as is true for marine systems. Allan, Art Benke, Susan Bell, Cheryl Ann Butman, LeRoy Poff, David Strayer, and an anonymous reAlthough stream meiofauna have reviewer. Special thanks go to Dave Allan who played a ceived little study, considerable progress has pivotal role in helping with the transition from the been made in the study of the transport of study of marine systems to inland streams and with whom I had many hours of discussions. I thank Paul marine meiofauna. These studies have foTurner, Janet Reid, and Kathy Coates who offered their cused on the same sorts of questions which expertise in the identification of rotifers, copepods, and have been asked concerning the drift of oligochaetes. Fred J. Easton of the Virginia State Water stream macrofauna, e.g. why do the animals Control Board made discharge data easily available. I enter the water column and what are the thank the sisters of the Notre Dame Academy and Mr. and Mrs. Ridgley White for freely permitting access to consequences of this entry? (e.g. Allan et al. the study site. 1986; Walters and Bell 1986). Because 329 330 Palmer freshwater and marine meiofauna inhabiting sandy bottoms are similar in size and may live in similar flows, it is tempting to assume that stream meiofauna exhibit transport patterns similar to their marine counterparts. However, oceanic and stream habitats are quite different physically and evolutionarily, so selective pressures that influence drift patterns may also be very different. Indeed, stream meiofauna may exhibit drift patterns (e.g. high nocturnal drift) more similar to their macrobenthic counterparts. Support for this comes from Schram et al. (1990) who showed that “zooplankton” in Arkansas streams drift in high numbers, particularly at night. Although this study did not involve benthic sampling, the specieslists they provided suggestthat many of the drifting animals they collected were probably members of the meiobenthos. I designed a study to examine meiobenthic drift in lotic systems to answer the following. specific questions with respect to the magnitude of drift. Is the drift of stream meiofauna significant and, if so, what percentage of the meiobenthos occurs in the drift? .Are drift rates comparable to those found for stream macrobenthos or marine meioflwna? I also asked specific questions regarding the evidencefor active and passive components to drift. Do drift rates increase with flow (suggestive of passive drift)? Is drift a nonlinear function of benthic density (evidence animals actively influence drift)? Do the meiofauna show nocturnal peaks in drift as do their macrobenthic counterparts (evidence animals actively influence drift)? Do live meiofauna enter the drift at higher threshold current velocities and exhibit shorter drift distances than dead fauna (evidence: animals actively influence drift)? Pasc;ive drift as used herein means animals enter and exit the water as if they are inert particles; active (influence over) drift means that through their behaviors (e.g. swimrning into the water, clinging to substrates on the bottom, etc.), animals reduce or enhance drift entry or exit. 38”57’N). Goose Creek is tributary to the Potomac River and the creek is situated in the Marshall Formation geologic region; the underlying bedrock is biotite granitic gneiss. The drainage area is -750 km2; the stream has year-round flow and a width of lo-20 tn. Water depth at the study site varies from - 15 to 150 cm during the year. The stream channel has not been altered by humans; however, portions of the catchment area are subject to human activity for recreation and farming. The riparian zone is dominated by deciduous trees common to the area especially oaks, sycamore, and box elder. L,ong stretches (- 100-300 m) of sandy bottom grade into occasional, short riffles (- 10 m) with cobble and pebbles. Sampling was con‘ducted in the sandy channels where the median grain size is 1 mm, the hyporheic zone extends from the sand surface to a depth of 40-60 cm, and the meiofauna are extremely abundant (Palmer 1990a,b). The site is just downstream from a USGS gauging station, thus long-term records of daily discharge and water quality were available. Mean annual discharge is - 8.5 m3 s-’ and peak discharge during floods is - 100 m3 s-l. Each tirne we sampled, vertical velocity profiles were measured with a Marsh-McBirney model 20 1 electromagnetic meter. Field sampling- On each sampling date, streambed meiofauna ‘weresampled with 35 replicate cores (1.8-cm diam) taken in 1Ocm intervals between the sand surface and bedrock. Separate cores were collected at each depth (O-10 cm, 10-20 cm, etc.) to ensure that data from various depth levels were independent. Replicate samples of meiofauna in the drift were collected at midday with drift nets (44+m mesh) positioned 5 cm above the sediment-water interface for 15-30 min. Nets were anchored above and not on the bottom to minimize flow disturbances which might artificially enhance erosion. The exact volume filtered was determined midway through each sampling by measuring mean velocity through the known cross-sectional area of the net. Nets were 2 Materials and methods m long with a 0.1 m2 opening and a widened Study site-The study site was Goose area 0.5 m below the mouth to minimize Creek, a low-gradient, fourth-order stream turbulence at the net entrance. Preliminary sampling showed that drift nets made of in Eoudoun County, Virginia (77’45’W, Lotic meiofauna drift mesh finer than 44 pm became quickly clogged and did not result in significantly greater meiofauna catches (20-pm nets vs. 44-ym nets: abundance of meiofauna not significantly different, P > 0.05). Samples were collected monthly from January 1988 to January 1989 and more frequently when discharge changed markedly; however, I was unable to collect drift samples on three dates (September, October, November) when flows were too low. Diel patterns were investigated with replicate streambed cores and drift samples collected at midday and 2 h after sunset on four dates (22 April, 27 April, 13 May, 23 August). After collection, samples were immediately placed in 1% MgC12 for - 3 min to promote meiofaunal relaxation, then rinsed and those fauna retained on a 44-pm sieve preserved in 10% Rose Bengal-Formalin. Fauna were identified to major taxonomic category; five taxa made up 90% of the community by abundance: rotifers, oligochaetes, copepods, early instar chironomids, and nematodes. For three of the sampling dates, rotifers, oligochaetes, and copepods were identified to species. Flume experiments-To investigate the importance of active vs. passive drift, I measured drift distances and critical threshold drift velocities of live vs. dead animals in a Plexiglas flume (5.5 m long, 3 5 cm wide, 50 cm deep) with freshly collected water at field temperature. Measurements of drift distance served as a tool for assessing whether the animals act like passive particles in transport. Specifically, if live animals released into the water returned to the bottom just as quickly as dead animals then both would drift the same distances at the same flow. A low flow setting (2.5 cm s-l) was used in the flume for these measurements because it allowed animals that were released into the water to settle to the channel bed before they were transported out of the channel itself. The channel was ~6 m long and the gravitational sinking rates of the animals were slow (-0.02-0.39 cm s-l; Palmer 1990a), so using higher velocities would have required an unrealistically long flume to allow settlement (e.g. at flows of 30 cm s-l, rotifers 331 released 3 cm from the bed would drift -45 m). It was not my goal to make drift distance measurements for extrapolation to the field although the measurements made here probably are reasonable for the long periods of low flow that generally occur in Goose Creek for -5 months of the year (see Fig. I and Palmer et al. 1992). The procedure used to measure drift distance was as follows. The flume channel was filled with a layer of azoic sand and a drift net positioned at the end of the channel. Water depth in the flume was kept at 6 cm, flow was set at 2.5 cm s-* , and animals were released into the flow at 3 cm above the bottom. For release, animals were extracted from freshly collected stream sand by swirling it in streamwater and decanting the water and suspended animals into a dish. The contents of the dish were poured into a sample splitter and divided into four aliquots (each with - 1,000 animals in 500 ml of water). Before use, each aliquot was examined to be certain that animals were active. Aliquots were poured into a Plexiglas chamber (11 x 30 cm) which was lowered into the flowing water of the flume channel, and the two ends of the chamber normal to the flow were immediately raised from the chamber base so that the animals entered the flow. Only the chamber base (0.96 cm thick and beveled to 1 mm thickness at its edges to minimize flow disturbance) intercepted the flow (designed by Woods Hole Oceanographic flume group; C. Butman pers. comm.). One aliquot of animals was released at a distance of 0.5 m upstream of the catchment net and allowed to drift-settle for 5 min. The drift net was then retrieved and all captured animals (i.e. not settled in the flume) were preserved in Formalin with Rose Bengal stain. This procedure was repeated at additional release points upstream: the second, third, and fourth aliquots of animals were released at 1, 2, and 3 m upstream of the drift net. Before each release, the sand downstream of the last release point was removed and replaced with new azoic sand. The entire procedure (aliquot release at four distances) was repeated three times with live animals and three times with animals that had been killed with ethanol. Nematode drift 332 Palmer from the box cores. Box cores (20-cm diam X 10 cm deep) consisted of sand and associated fauna collected from the field ‘by inserting the core into the streambed and then sliding a bottom plate below the core. All cores were collected 24 h before their J, use in the flume and held in a living stream. Previous experiments have shown that the vertical distribution of IGoose Creek meiofauna in box cores collected and allowed to equilibrate before use in the flume was not significantly different frlom the vertical distribution in the field (P.almer et al. 1992). Box cores fit into the flume channel 4.5 > 20000 CHIRONOMIDS m downstream of the entrance, and the surI1 bz 15000 face of the substrate in the core was flush with the bottom of the: flume. Azoic sand 8 10000 was placed upstream (evenly all along the tbottom of the channel) of and surrounding 5000 & the core to ensure no changes in bottom D 0 I-~,,,,,, roughness as flow approached the core. The maximum velocity used in these experiCOPEPODS ments (20 cm s-l) was chosen to exceed critical erosion velocity for the substrate. The drift vs. flow measurements were made for three replicate box cores with live animals and three “dead” box cores that had been injected with ethanol 2 h before use. This 7 120 technique did not alter the vertical distrim bution of animals in the box cores (live vs. 20 DISCHARGE L II I I dead, percentage of animals in top 5 mm not different, P > 0.05). Data analysis -Abundances of fauna in the drift are expressed as drift densities with those units most commonly reported-No. of fauna per 100 m3. For comparison of drift Fig. :I. Temporal patterns of drift for 1988-1989. Mean values for the four dominant taxa with standard between dates and with physical variables, error bars; data points absent if flow was too low for drift was also expressed as drift rate (drift drift sampling. Mean daily discharge from USGS gaug- density x discharge >: time), which is the ing station near sample site. total number of animals drifting in a cross section of the stream per unit time. Drift rate is less dependent on the dilution and distances were not measured because there concentration effects that changing flow rewere not enough live animals. gimes have on drift d’ensity. Throughout, parametric statistical tests If drift entry is passive, a critical entrainment velocity (hereafter, threshold) should were used after the data were examined to exist analogous to critical erosion velocities determine that appropriate assumptions for se,diments. To investigate, I exposed box were valid. When necessary, abundances cores of fresh stream sand to increasing flows were log& + 1) or arcsin-transformed be(O-20 cm s-l in 2 cm s-l intervals) for 2 min fore tests. ANOVA and linear regressions at ealch flow. Drift nets positioned at the were performed with general linear model end of the flume and changed before each techniques (SAS 19(35) and comparisons made with the Tukey-Kramer method when flow jincrease caught all animals originating I 333 Lotic meiofauna drift Table 1. Percentage of bottom meiofauna in drift. Percentages are calculated for each taxon by dividing numerical abundances found in the drift (no. vol.-‘) by numerical abundances found in an equal volume of sediment from the top 10 cm of the hyporheic zone. Estimates are mean values (N = 34) for entire year; SE in parentheses. Rotifers Chironomids Oligochaetes Copepods Nematodes . Drifting benthos w Max drifting w 0.019 (0.015) 0.004 (0.003) 0.00 1 (0.00 1) 0.054 (0.03) 0.005 (0.008) 0.05 0.02 0.02 0.14 0.03 significant main effects were present from an ANOVA. Results The magnitude of drift -Large numbers of meiofauna were found in all drift samples collected at Goose Creek. Over the entire year mean abundance of total meiofauna drifting was N 89,000 per 100 m3, but in April drift densities exceeded 250,000 per 100 m3. The number of drifting meiofauna varied significantly by date (ANOVA, P < 0.001) with highest drift abundances generally in late spring (Fig. 1). The percentage of benthos drifting was estimated from field data on drift and benthic densities, which ranged from 0.00 1 to 0.14% depending on taxon and date (Table 1). The predominant taxonomic group in the water was rotifers, which constituted > 86% of the drift by numerical abundance (Table 2). Evidence for active and passive components to drift-The temporal pattern in drift over the year (Fig. 1) typically showed greatest drift densities in April, May, and June. Statistically, drift rates and drift densities increased with current velocity for all groups (P < 0.01; e.g. Fig. 2, drift rate for rotifers). Drift was not density-dependent, i.e. there was not a significant increase in percent drifting when benthic density increased (P > 0.05) for any of the taxa. When samples were collected to determine whether there was diel variation in drift, flow was not significantly different between night and day on any of the sampling dates (P > 0.05). However, discharge magnitude was significantlv~ lower - -~ during - - ----” the _--_ Table 2. Percentage representation of common meiobenthic taxa in drift and bottom samples. Percentages are calculated from numerical abundances for each taxon divided by the total faunal abundance (all taxa) found in the sediment (bottom) or drift. N = 34. SE in parentheses. Rotifers Chironomids Oligochaetes Copepods Nematodes Bottom w Drift v-4 47.7 (5.2) 12.9 (2.3) 19.1 (2.8) 4.2 (0.9) 3.1 (0.5) 86.2 (1.5) 6.9 (1.1) 2.9 (0.4) 1.5 (0.2) 2.4 (0.4) August sampling (1.4 m3 s-l) than during the April and May samplings (X = 8.1 m3 s-l). Diel variation in drift was pronounced for all taxonomic groups except nematodes. Significantly more chironomids, oligochaetes, and copepods were drifting at night than during the day for all four dates (Fig. 3) on which day-night samples were collected (ANOVA, P < 0.01 for each taxon). Abundances of chironomids and oligochaetes in the streambed (Fig. 3) also exhibited a diel pattern, with lower numbers in the hyporheic zone at night (ANOVA, P < 0.00 1). Diel patterns in abundance of copepods in the streambed were quite variable between dates and overall there was no statistically significant diel effect (P > 0.05). Rotifers exhibited higher night than day drift rates, but this diel pattern was highly dependent on date (ANOVA, date x time-ofday interaction significant at P < 0.001) with no diel variation in August (Fig. 3). Q 6000 a t 4000. B L 2000 . . . . :a z 0L L%.,..L.., .. ...._.....,.__..... T 0 20 VELOCITY 40 60 (‘cm 60 s- ’ 1 Fig. 2. Drift rate of rotifers (total drift past a point in the stream over 24 h) vs. mean current velocity for same time period. Drift samples were not collected when- flows were too low (September, October, No. vember). 334 Palmer DRIFT DAY 0 NICXT m Chronomlds STREAMBED DAY 0 NIGHT a Fig. 3. Diel patterns of meiofaunal abundance in the drift and the streambed on 22 and 27 April, 13 May, and 23 .4ugust 1988. Values are means per unit volume sampled based on benthos samples taken from the entire hyporheic zone (O-50 cm) and drift samples taken with nets positioned 5 cm above the bottom. Standard error bars shown. Diel patterns were not detectable in streambed abundances of rotifers (P > 0.05; Fig. 3). For the rotifers and copepods, there was not co:mplete concordance between species found in the drift and those found in the benthic samples, indicating that some drifters originated from benthic habitats not sampled such as debris dams or streambanks. Over 80 species of rotifers have been identified from Goose Creek benthic and drift samples to date (Turner and Palmer in prep). Most of the species found in the drift were also found in benthic samples (N 70%); however, several of the dominant drift species were among those species never found in the sandy benthos [e.g. Lecane (Monostyla) cornuta, Macrochaetus subquadratus]. For the copepods, species diversity was higher in the drift samples than in benthic samples. One species, Diacyclops albus, was the overwhelming dominant in the benthic samples yet this species was no more common in the drift than Eucyclops agilis and Macrocyclops albidus. Among the oligochaetes, most of the species found in the sediment (e.-g. Chaetogaster diaphanus, Chaetogaster diastrophus, Slavina appendiculata, Nais variabilis, .4ulodrilus pluriseta) were also found in the drift. In the drift distance experiments, the number of animals caught in drift nets gen- erally decreased exponentially as the distance from the point of release to the drift net increased (Fig. 4). In order to determine the mean drift distance for each group, I regressed log catch (number of animals caught in the drift) on :releasedistance. The inverse of the slope of that regression line gives mean drift distance (Elliott 197 1). All regressions were signilircant with r2 values of 0.87-0.99. Because three live and three dead replicate release e:xperiments were run for each group, there were six regression lines per taxonomic group. The overall mean drift distance (n = three replicates) for dead animals was similar to live for all taxa except copepods (Fig. 5) which drifted significantly farther before returning to the bottom than did live copepods (P < 0.05). Differences in mean drift distance between taxa were apparent (Fig. 5). Oligochaetes drifted the shortest distances; relative to oligochaetes, chironomids drifted CU1.6 x farther, copepods -2.5~ farther, and rotifers -5.5~ farther. Similar differences in drift distance among taxa were obtained from calculations based on gravitational sinking rates made in still water (Palmer 1990a): relative to oligochaetes (with the shortest drift distance), chironomids would drift 1.4 X farther, copepods 2.7 x farther, and rotifers 7 X farther. The experiments to determine the critical 335 Lotic meiofauna drift Oligochaetes 1 2 1 3 Drift Distance 2- 3 (m) Fig. 4. Relationship between total catch of meiofauna and horizontal distance between drift net and the release point in a laboratory flume. Mean water velocity was 2.5 cm s-l and release height was 3 cm above the bottom. Three replicates for live animals (dashed lines) and three for dead (solid lines). drift thresholds of the animals as a function of velocity yielded variable results between taxa and, for some taxa, between live and dead animals (Fig. 6). If we assume passive entry into the water via erosion, few or no animals should be present in the water at low flows and, then, above a “threshold” velocity, the numbers drifting should increase dramatically. The threshold points were visually approximated from the graphs (arrowheads on Fig. 6) as is commonly done to estimate critical erosion velocity of sediments. All of the animals had entry thresholds between m9 and 12 cm s-l, which was less than the critical erosion threshold for the sediment (range, 14-17 cm s-l; mean, 16.4 cm s-l). There was little difference between the curves for live vs. dead for the oligochaetes and chironomids, suggesting they did not exert active control over entry in these experiments. For the copepods, the threshold flow at which the number of animals entering the water increased noticeably was slightly higher for live animals than for dead animals, which also may have been true for rotifers; however, there was large variability between the three threshold curves for live rotifers (Fig. 6). Discussion The magnitude of drift-Drift densities of total meiofauna in Goose Creek were ex- tremely high: 250,000 per 100 m3 during spring spates and an average of -89,000 per 100 m3 for the remainder of the year. These densities vastly exceed drift densities of macrobenthic invertebrates in streams, which usually fall between 100 and 1,000 per 100 m3 (Allan 1987); however, the Goose Creek values are comparable to the highest meiobenthic drift found in marine systems (Hagerman and Rieger 198 1; Palmer and Gust 1985). Additionally, Schram et al. (1990) reported extremely high drift densities for “zooplankton” in a third-order stream in Arkansas. Although they did not sample the benthos, their species list indi- Oligochaetes Chmnomids Rotifers Taxa Fig. 5. Mean drift distance for live and dead meiofauna calculated from slopes of regression lines in Fig. 6; seetextforjiirther details. Standard error bars shown. 336 Palmer Oligochaetes 24 d Live 6 -2 :I,,p: 24 / / /’ /’ Dead A’ /’ /’ /’ ,/--’ ,/’ ,y”-r----* --- ___C-- 6 /’ LLII-2 12 -5% c--- 16 20 24 20 24 Copepods 0 24 c ----I 12 Velocity km s-’ 16 ) Fig. 6. Number of meiofauna drifting (per 2 min) from flume box-core sediments as a function of current velocity. Three replicate box cores with live animals and three with dead animals. For each line, drift threshold was approximated visually (H) as the velocity at which drift began to increase most dramatically. Arrowheads indicate mean threshold velocity for the three lines. cates that many of the rotifers collected in their drift nets (80qm mesh) were probably meiobenthic in origin, not planktonic. Clearly, the present study and that of Schram et al. indicate that meiofaunal drift in streams can be substantial and that use of large-mesh drift nets has caused most stream investigators to underestimate drift (seealso Benke et al. 1986). The magnitude of meiofaunal drift in Goose Creek was estimated in terms of biomass using literature values for similarly sized freshwater meiofauna. I used the yearly average drift densities for each taxon (Fig. 1) and the range in individual biomass reported in the literature (Table 3) and found that meiofaunal biomass in the drift in the creek averaged 8-9 1 mg dry mass per 100 m3. When I used peak drift densities (generally found in spring) for these calculations, meiofaunal biomass in the drift was 24-308 mg per 100 m3. Ranges are given to emphasize that these are merely estimates because species-specific: biomass and lengthweight data are not yet available for Goose Creek. The estimates indicate that, in terms of biomass, meiofaunal drift is not insignificant as has often been assumed. The bio- Lotic meiofauna drift 337 Table 3. Literature estimates of mean individual dry mass @g ind.-‘) for benthic taxa of meiofaunal size that are important in Goose Creek. Values shown cover entire range of values reported. Biomass range (pg ind.-‘) Rotifers Chironomids Oligochaetes Copepods Nematodes 0.07-0.44 0.50-4.00 0.03-l 3.7 0.20-6.96 0.04490 References Oden 1979; Quiglcy and Nalepa 1983; Strayer 1985 Benke et al. 1984* Giere 1975; Strayer 1985 Goodman 1980; Quigley and Nalepa 1983; O’Doherty 1988; Strayer 1985 Quigley and Nalepa 1983; Strayer 1985 * Data from two smallest size classes used. mass of macrofauna drifting in warm-water streams such as Goose Creek has been reported to be 20-50 mg per 100 m3 for the Satilla River (Benke et al. 1986) and 240246 for the Ogeechee River (Benke et al. 199 1). A comparison of the range in biomass estimates for meiofauna in Goose Creek with the range in biomass reported by Benke and coworkers indicates that meiofaunal drift may be 1O-l 00% of macrofaunal drift! The percentage of bottom meiofauna drifting at any point in time was within the range generally reported for most stream macrobenthos (usually ~0. 10%) (Williams 1980). The percentages presented here (Table 1) can be considered conservative for two reasons. First, the calculations were made by using average benthic abundances per volume of sediment in the top 10 cm of the streambed. Because abundances there were slightly higher than in deeper sediment layers (Palmer 1990b), this would yield a smaller percentage than if calculations were made using average benthic abundances per volume of sediment based on the entire (O50 cm) streambed. Second, the day-night sampling of the drift and benthos indicated that the drift nets may have seriously underestimated total drift. The number of animals lost from the streambed at night was much higher than the nighttime increase in drift (Fig. 3). Drift nets apparently did not capture all animals exiting the sandy bottom, probably because they were anchored 5 cm above the bottom and thus did not capture the large numbers of meiofauna that may be transported near the bed. Although percent drifting is a useful indication of the approximate magnitude of drift, comparisons between bottom fauna and the drift (in this and other papers) should be interpreted with caution becausenets may sample the drift inefficiently and because the origin of the animals caught in drift nets is usually unknown, In my study, only the dominant benthic habitat-the main sandy channel-was sampled for benthos, yet individuals captured in drift nets were likely to have been advected into nets from areas other than where benthic samples were collected. Indeed, the species identifications of drift vs. bottom samples indicated that at least rotifers and copepods were advected in from other areas. Even though the dominant benthic habitat is the sandy channel, debris dams are present along much of the streambank and they may be important habitats for meiofauna, perhaps contributing substantially to the drift (see Benke et al. 1986 for macrofauna). Evidencefor active andpassive control over drift-The present study provides a starting point for examining the factors controlling drift of lotic meiofauna (Table 4). There is evidence that both active and passive mechanisms operate. The positive relationship between drift and current velocity in Goose Creek suggests that animals may enter the water column by passive entrainment as flow increases. Passive entrainment has been demonstrated for marine meiofauna (Hagerman and Rieger 198 1; Palmer and Gust 1985) and there is also evidence that the drift of lotic macrofauna increases as flow increases (Brittain and Eikeland 1988). Passive entry into the water for Goose Creek meiofauna during the daytime was also supported by the flume experiments in which critical entry thresholds (9-l 2 cm SS’)could be identified. Interestingly, these thresholds were less than the entrainment threshold for the sandy bottom, probably reflecting differences in lift and drag experienced by an- 338 Palmer Table 4. Summary of evidence that stream meiofauna can actively influence transport based on results of the field and flume studies. This active influence may involve various behaviors which affect the transport process at several stages (e.g. entry into the water, exit, time spent in transport). Support found for an active component-(+); no evidence for an active component-(-); no data available-NA.. Diel patterns* Dl-if-t Sediment Sinking rate? (live > dead) Drift distance* (dead > live) Drift threshold4 (live Z dead) Densitydependent driflll * Fig. 3. t Palmer 1990~2. :z* 1. 1)Text. . imals and sediments due to their different specific gravities, sizes, and shapes. Regardless of the underlying cause, this difference in thresholds is an important result because it implies that whenever the streambed is eroded (e.g. during spates), meiofauna may also be transported. Despite evidence that passive mechanisms do operate, drift in streams and in marine environments clearly is not purely passive (Walters and Bell 1986; Allan et al. 1986). My findings that some meiofauna exhibited diel patterns in drift, differences in live vs. dead sinking rates, shorter drift distances for live than dead, and drift thresholds which differed for live vs. dead show that some lotic meiofauna can actively influence drift entry or exit (Table 4), just as appears true for some stream macrofauna and some marine fauna. The strong diel drift pattern observed for Goose Creek meiofauna was less surprising for chironomids than for copepods and rotifers because diel patterns #arewell known for stream macrobenthos, and chironomids are merely juvenile macrofauna. This diel pattern suggeststhat selective pressures favoring nighttime over daytime entry for macrofauna (e.g. predator avoidance) may operate for stream meiofauna as well. Schram et al. (1990) also found diel platterns in stream drift for rotifers and copepods. They made the important point that drift patterns were quite variable between species of rotifers and, thus, there is a need to focus more effort at the species level in such studies. Perhaps the differences between dates in the diel drift patterns of rotifers found here (Fig. 3) could be explained if species-level data were available. The meiofaunal behaviors responsible for high nighttime drift are unknown. Animals may actively leave the bottom, swimming up into the water at night or may increase their susceptibility to passive transport by moving closer to the sediment-water interface at night. Indeed, the fact that migrations within the sediment may influence transport is supported by several studies showing that some meiofauna move down into the sediments as flow increases (see Palmer et al. 1992). Such migrations may also be responsible for the differences in live vs. dead drift thresholds observed for some of the Goose Creek fauna. The behavioral basis for the shorter drift distances of live vs. dead copepods may include differences in the posture assumed by live and dead animals while being transported, result from downward swimming of drifting animals, or be due to active re-entry into the water after initial contact with the bottom. Little is known about postures assumed during settlement; however, there is evidence from marine studies that the supply of meiofaunal-sized organisms to the bottom is a flow-domlinated process (i.e. not controlled by swimming) because faunal swim speeds are substantially lower than near-bottom flow speeds (Butman 1987; Palmer 1990a) and th[e animals settle where they are hydrodynamically retained (Butman 1987; Eckman 1990). Even if settlement is passive, Mullineaux and Butman (199 1) have shown that active control over final settlement site can be altered by behaviors that come into play immediately postsettlement. The compelling evidence that stream Lotic meiofauna drift meiofauna may actively enhance nighttime drift contrasts with what has been found for marine meiofauna in environments with swift flows similar to those experienced by lotic meiofauna. Active nighttime entry by meiofauna has never been found in openchannel, high-flow marine environments where, instead, animals burrow down or cling to substrates to avoid transport (reviewed by Palmer 1988). There has been speculation that behaviors which decrease water-column transport may actually be advantageous in such systems. Suspension into the water may lead to transport to unsuitable habitats (Palmer and Gust 1985), expose the animals to pelagic predators (D’Amours 1988), and interfere with feeding (D. Thistle pers. comm.). Active nocturnal entry by meiofauna has been found in seagrass beds (Walters and Bell 1986) where flow is greatly reduced and predation risk may be decreased due to high structural complexity of the habitat. Significance of mei0faunal transport in streams-I have shown that extremely large numbers of meiofauna are transported in the water column in Goose Creek but that these drifters represent a small proportion of the total meiofauna in the sediment. Despite the potential problems with such proportion calculations (see above), this percentage is comparable to what has been reported for lotic macrofauna and for marine meiofauna (Williams 1980; Palmer 1988). The small proportion (<O.lO%) should not be taken to mean that drift is unimportant. Because these are quantifications of the percentage of bottom fauna in the water at an instant in time, others (e.g. Wilzbach and Cummins 1989) have pointed out that the cumulative effect is nonetheless great: 24 h of drift over a unit area can reach ten or even a hundred times benthic density. The transport of lotic meiofauna should have great significance for both meiofauna population dynamics and streamwide community structure. First, drift allows meiofauna to colonize new and denuded habitats, which is of particular importance in streams that are often subjected to frequent and intense flooding (Townsend and Hildrew 1976). Given the prevalence of asexual reproduction by many freshwater meiofau- 339 na (e.g. rotifers), only a few individuals may be required for invasion of a habitat. Second, water-column transport is the primary dispersal mechanism for meiofauna (Palmer 1988) and, as such, has important genetic consequences. Others have shown that there is generally an inverse relationship between the duration of the pelagic phase for dispersing marine larvae and the degree of genetic differentiation between intraspecific populations (seeBurton and Feldman 1982). For sexually reproducing meiofauna (e.g. copepods), water-column transport from one stream habitat to another surely enhances gene flow, even if only a small fraction of the population is moving. For example, genetic homogeneity across a number of demes is theoretically ensured even at extremely low dispersal rates (e.g. one migrant per deme per generation; Smith 1989). Such dispersal rates are certainly achieved by drifting meiofauna. Third, from a community perspective, meiofauna that have entered the water are easily available as food for filter-feeding macroinvertebrates and drift-feeding fish (e.g. D’Amours 1988). Marine studies have shown that even when only a small fraction (%) of a meiofaunal population is eaten, the fauna may still be extremely important in overall community trophodynamics: because of the vast numbers of meiofauna, they can represent a major portion of the diets of predators (e.g. meiobenthic copepods made up > 80% of the diet of juvenile chum salmon in a sixth-order stream in Canada; Sibert et al. 1977). Although we do not yet know which lotic macrofauna or fish depend primarily on meiofauna for food, we do know that stream meiofauna serve as prey for these larger animals (O’Doherty 1988; Vadas 1990). This study represents the first examination of both the significance of meiofaunal drift and the extent to which entry into the water column in lotic environments is influenced by active or passive processes. Clearly both processes influence drift, often in complex ways (Table 4). No simple dichotomy between active and passive entry or exit exists. Evidence for passive entry came from a general increase in drift with flow and from the presence of critical flow thresholds for drift. Evidence for active en- 340 Palmer try included nighttime increases in drift and, special reference to meiobenthic species. Mar. Biol. 31: 139-156. for copepods, the ability to influence return to the bottom once in the drift. The general GOODMAN, K. S. 1980. The estimation of individual dry weight and standing crop of harpacticoid coabsence of such active entry in fast-flowing pepods. Hydrobiologia 72: 253-259. marine systems suggests that the relative HAGERMAN,G. M., AND R. M. RIEGER. 1981. Discost-benefit ratio of transport may be higher persal of benthic meiofauna by wave and current action in Bogue Sound, North Carolina, U.S.A. in the more open marine systems than in Mar. Ecol. Publ. Staz. Napoli 2: 245-270. streams. The importance of meiofaunal drift L. S., AND C. A. BUTMAN. 199 1. Initial in terms of abundance and biomass was great MULLINEAUX, contact, exploration, and attachment of barnacle in this study, as it generally is in marine cyprids settling in flow. Mar. Biol. 110: 93-104. environments and for lotic macrofauna. ODEN, B. J. 1979. The freshwater littoral meiofauna in a South Carolina reservoir receiving thermal There are both ecological and evolutionary effluents. Freshwater Biol. 9: 29 I-304. consequences associated with this waterE. C. 1988. The ecology of meiofauna column transport. These consequences have @DOHERTY, in an Appalachian headwater stream. Ph.D. thesis, important implications not only for underUniv. Georgia. 113 p. standing meiofauna population dynamics PALMER,M. A. 1988. Dispersal of marine meiofauna: A review and conceptual model explaining passive but for the study of trophic transfers from transport and active emergence with implications the benthic to the pelagic realm. for recruitment. Mar. Elcol. Prog. Ser. 48: 8 l-9 1. 1990a. Understanding the movement dynamics of a stream-dwelling meiofauna commuALMN, J. D. 1987. Macroinvertebrate drift in a Rocky nity using marine analogs. Stygologia 5: 67-74. Mountain stream. Hydrobiologia 144: 26 l-268. --. 1990b. Temporal and spatial dynamics of ---, A.S. FLECCKERANDN.C.MCCLINTOCK. 1986. meiofauna within the hyporheic zone of Goose Diel epibenthic activity of mayfly nymphs, and its Creek, Virginia. J. N. Am. Benthol. Sot. 9: 17-25. nonconcordance with behavioral drift. Limnol. --, A.E. BELY,AND K. E. BERG. 1992. Response Oceanogr. 32: 1057-1065. of invertebrates to lotic disturbance: A test of the BENKE,A.C.,R.J. HUNTER,ANDF. K. PARRISH. 1986. hyporheic refuge hypothesis. Oecologia 89: 182194. Invertebrate drift dynamics in a subtropical black--AND G. GUST. 1985. Dispersal of meiofauna water river. J. N. Am. Benthol. Sot. 5: 173-190. in ‘a turbulent tidal creek. J. Mar. Res. 43: 179---, K.A. PARSONS,AND~.M. DHAR. 1991. Pop210. ulation and community patterns of invertebrate drift in an unregulated coastal plain river. Can. J. PENNAK,R. W., AND J. V. WARD. 1986. Interstitial faunal communities of the hyporheic and adjacent Fish. Aquat. Sci. 48: 81 l-823. groundwater biotopes of a Colorado mountain ----? T.C.VAN ARSDALL,JR., lack D.M. GILLESPIE. stream. Arch. Hydrobiol. Suppl. 74(3), p. 3561984. Invertebrate productivity in a subtropical 396. blackwater river: The importance of habitat and QUIGLEY, T. F., AND M. A. NALEPA. 1983. Abunlife history. Ecol. Monogr. 54: 25-63. dance and biomass of the meiobenthos in nearBRITTAIN,J. E., AND T. J. EIKEL~~ND.1988. Invertcshore Lake Michigan with comparisons to the brate drift-a review. Hydrobiologia 166: 77-9 3. macrobenthos. J. Great Lakes Res. 9: 53oL547. BURTON,R. S., AND M, W. FELDMAN. 1982. PopuSANDLUND,0. T. 1982. The drift of zooplankton and lation genetics of coastal and estuarine invertemicrozoobenthos in the River Strandaclva, westbrates: Does larval behavior influence population ern Norway. Hydrobiologia 94: 33-48. structure?, p. 537-55 1. In V. S. Kennedy [ea.], SAS INSTITUTE,INC. 1985. SAS user’s guide: StatisEstuarine comparisons. Academic. tics. Version 5 ed. BUTMAN, C. A. 1987. Larval settlement of soft-sedSCHRAM, M. D., A. V. BB~OWN, AND D. C. JACKSON. iment invertebrates: The spatial scales of pattern 1990. Diel and seasaNnadrift of zooplankton in explained by active habitat selection and the a headwater stream. Am. Midl. Nat. 123: 135emerging role of hydrodynamical processes. 143. Oceanogr. Mar. Biol. Annu. Rev. 25: 113-l 65. &BERT, J., T. J. BROWN, M. C. HEALEY, AND B. A. D'AM~uRs, D. 1988. Vertical distribution and abunKASK. 1977. Detritus-based food webs: Exploidance of natant harpacticoid copepods on a vegtation by juvenile chum salmon (Oncorhynchus etated tidal flat. Neth. J. Sea Res. 22: 16l-l 70. keta). Science 196: 649-650. ECKMAN,J. E. 1990. A model of passive settlement SMITH, J. M. 1989. Evolutionary genetics. Oxford. by planktonic larvae onto bottoms of differing STRAYER,D. 1985. The benthic micrometazoans of roughness. Limnol. Oceanogr. 35: 887-90 1. Mirror Lake, New Hampshire. Arch. Hydrobiol. ELLIOTT,J. M. 197 1. The distances travelled by driftSuppl. 72(3), p. 287-126. ing invertebrates in a Lake District stream. OecoTOWNSEND,C. R., AND A. B. HJLDREW. 1976. Field logia 6: 350-379. experiments of the drifting, colonisation and continuous redistribution of stream benthos. J. Anim. GIERE.,0. 1975. Population structure, food relations, and ecological role of marine oligochaetes with Ecol. 45: 759-773. References ---. Lotic meiofauna drift VADAS,R. L., JR. 1990. The importance of omnivory and predator regulation of prey in freshwater fish assemblages of North America. Environ. Biol. Fishes 27: 285-302. WALTERS,K., AND S. S. BELL. 1986. Diel patterns of active vertical migrations in seagrass meiofauna. Mar. Ecol. Prog. Ser. 34: 95-103. WILLIAMS,D. D. 1980. Invertebrate drift lost to the sea during low flow conditions in a small coastal stream in western Canada. Hydrobiologia 75: 25 l254. 341 WILZBACH,M. A., AND K. W. CUMMINS. 1989. An assessment of the short term depletion of stream macroinvertebrate benthos by drift. Hydrobiologia 185: 29-39. Submitted: 27 December 1990 Accepted: I1 September 1991 . Revised: 3 October 1991