J. N. Am. Benthol. Soc., 2005, 24(4):858–871 ᭧ 2005 by The North American Benthological Society

Chimney construction by riparius larvae in response to hypoxia: microbial implications for freshwater sediments

PETER STIEF1 Max Planck Institute for Marine Microbiology, Celsiusstrasse 1, 28359 Bremen, Germany

LARISA NAZAROVA2 Ecological Faculty, Kazan State University, Kremlyevskaya Str. 18, 420008 Kazan, Russia

DIRK DE BEER3 Max Planck Institute for Marine Microbiology, Celsiusstrasse 1, 28359 Bremen, Germany

Abstract. Many shallow aquatic ecosystems with high nutrient loads experience periods of O2 depletion that evoke behavioral responses by macrobenthic organisms. The sediment-dwelling larva Chironomus riparius reduces its deposit-feeding activity and allocates more time to burrow ven- tilation during periods of hypoxia. We investigated another striking behavioral adaptation of this species, i.e., the elongation of U-shaped sediment burrows to chimneys that tower above the sediment surface. Chironomus riparius larvae gradually abandoned burrow construction and took up chimney construction when exposed to hypoxic conditions in laboratory microcosms. Microsensors were used to show that the chimneys were oxic sediment compartments that were periodically irrigated by the

larvae with oxygenated surface water. O2 uptake rates per unit interface area were significantly higher for chimneys than for the flat sediment surface. This observation was consistent with the dense

colonization of the chimneys by bacteria. Chimneys may facilitate the larval acquisition of both O2 for respiration and microbial biomass for food. Given the mass abundance of C. riparius in many

polluted and O2-deficient habitats, the chimneys also may contribute significantly to the patchiness of the benthic microbial community in terms of structure and function. In particular, the presence of chimneys might favor aerobic bacterial populations and their metabolism. Key words: , macrofauna, freshwater sediment, oxygen depletion, biogenic structure, sediment bacteria, diffusive oxygen uptake, –microbe interaction, sediment microcosm, micro- sensor.

Periods of O2 depletion are common in many (e.g., in streams, ponds, and lakes, Hellmann aquatic ecosystems, including coastal upwelling 1994, Hamburger et al. 2000). regions (Levin 2003), fjords (Rosenberg et al. Benthic macro- and microorganisms must 2001), streams receiving sewage water (Hell- cope with the consequences of hypoxic events, mann 1994), and lake hypolimnia during tem- including shortage of O2 and accumulation of perature stratification (Hamburger et al. 2000). toxic substances (e.g., sulfide). Macrofauna may Common features of ecosystems susceptible to have physiological adaptations to these condi- hypoxia or anoxia are high nutrient concentra- tions (e.g., hemoglobin, Choi et al. 2000; anaero- tions, extensive organic sedimentation, and poor biosis, Hamburger et al. 2000), or they may re- exchange of bottom water (Gray et al. 2002, Lev- spond to hypoxia with changes in behavior. in 2003). High nutrient concentrations can be of Ventilation of sediment burrows is intensified in natural origin (e.g., coastal upwelling regions, shallow waters (Leuchs 1986, Heinis and Crom- Bru¨ chert et al. 2003), or they can result from mentuijn 1992), bioturbation activity is reduced anthropogenic eutrophication and pollution (Levin 2003), and sediment burrows are elon- gated above the sediment surface (Kon and Hi- 1 Present address: Department of Microbiology, In- daka 1983, Jørgensen and Revsbech 1985). How- stitute of Biological Sciences, Aarhus University, Ny Munkegade, Building 540, 8000 Aarhus C, Denmark. ever, stress avoidance and behavioral compen- E-mail: [email protected] sation are of little importance to sediment bac- 2 E-mail addresses: [email protected] teria (except for mobile taxa). Instead, external 3 [email protected] conditions regulate the metabolism of bacteria

858 2005] CHIRONOMID RESPONSE TO HYPOXIA 859 from full activity to dormancy. On the micro- uptake vary between inhabited and uninhabited scale, sediments can be considered model habi- sediments and the chimneys? Laboratory micro- tats for studying the bacterial response to hyp- cosm experiments were conducted using ho- oxia and anoxia because sharp O2 gradients de- mogenized freshwater sediments that were ex- termine the metabolic activity of aerobic and an- posed to different O2 concentrations (factor 1) aerobic bacteria within the sediment (Wieringa and to which C. riparius larvae (factor 2) were et al. 2000, Altmann et al. 2003). added or not. The number and size of burrows Here we present an example of the behavioral and chimneys were quantified, total bacterial response of an abundant macrofaunal organism abundance was determined, and O2 concentra- to hypoxic conditions and the implications of tions were measured on a microscale in all treat- this behavior for bacterial colonization patterns ment combinations. in freshwater sediments. Chironomus riparius,a nonbiting midge with aquatic larval stages, is a Methods typical inhabitant of polluted and temporarily Origin of sediment and O2-deficient shallow waters (Pinder 1986, Bair- lein 1989, Penttinen and Holopainen 1995). The Sediment was sampled in a freshwater ditch larvae regularly occur in masses, construct U- ϳ20 km west of Ko¨ln (North-Rhine Westphalia, shaped burrows within soft sediments, and feed Germany). The ditch runs through an intensely on the detritus particles and the attached bac- farmed area. The water column is characterized teria that are deposited onto the sediment sur- by high nutrient concentrations and transient O2 face (Pinder 1986). Chironomus riparius larvae tol- depletions. Large populations of various Chiron- omus species inhabit the silty and organic-rich erate O2 deficiency because their hemolymph sediment of the ditch. A flat shovel (10 ϫ 15 cm) contains hemoglobin with a high affinity for O2 (Choi et al. 2000). The larvae also can switch to was pushed 10 cm into the sediment and the anaerobiosis, but they are less efficient than lar- load of sediment was lifted carefully out of its vae of other species (e.g., Chironomus anthracinus, place. Five loads were placed in a 10-L plastic Penttinen and Holopainen 1995, Hamburger et bucket and covered with 2 L of ditch water. The al. 2000). Chironomus riparius larvae increase sediment was sieved through a 1-mm-mesh Њ their undulation activity during periods of low screen, homogenized, and stored at 4 Cinthe dark until use. Egg masses of C. riparius were O2 concentrations in the water column (Leuchs 1986). Occasionally, the larvae elongate their provided by the Institute for Inland Water Man- burrows to reach above the sediment surface, a agement and Waste Water Treatment (RIZA) in behavior described as chimney-projecting behavior Lelystad, The Netherlands. Larvae were bred in ϫ ϫ (Kon and Hidaka 1983, von Danwitz 1983). Bio- laboratory tanks (40 50 15 cm) filled with genic structures that pierce through the diffu- aerated, aged tap water and a 5-mm thick layer of sediment. A food suspension of ground sive boundary layer are thought to improve O2 availability for the inhabiting animal (Jørgensen leaves of stinging nettle (Urtica sp.) and fish food (Trouvit, Tetramin௡) was added 3 times a and Revsbech 1983). Projecting biogenic struc- Њ tures also may generate small-scale habitat het- week. The culture was kept at 20 C under a erogeneity for the benthic microbial community light:dark cycle of 16:8 h. Chironomus riparius lar- by promoting advective nutrient transport and vae grew to a body length of 5 to 7 mm within rd trapping organic particles (Eckman and Nowell 10 to 15 d (3 -instar larvae) and were then 1984, Eckman 1985, Eckman and Thistle 1991, ready for use in the hypoxia experiments. A group of larvae was tested for its sensitivity to Soltwedel and Vopel 2001). a standard toxicant (K Cr O , Grootelaar et al. Our study addressed the following questions: 2 2 7 1996) before the actual experiment. The sensitiv- 1) Does a threshold O concentration induce the 2 ity of larvae taken from our culture lay within chimney-projecting behavior at the expense of the frame of acceptability for use in laboratory sediment burrow construction or is the response exposure experiments. to increasing hypoxia graded? 2) Does the erect shape of chimneys make them a residual oxic sediment compartment under hypoxic condi- Laboratory microcosms tions? 3) Are chimneys preferential sites for bac- Sediment was added to the center compart- ϫ ϫ terial colonization? 4) How does sedimentary O2 ment (10 5 5 cm) of transparent aquaria 860 P. STIEF ET AL. [Volume 24

FIG. 1. Microcosms used for hypoxia experiments (side view). Numbered parts are acrylic aquarium (1), O2 macroelectrode (2), oxymeter (3), O2 control unit (4), magnetic valve (5), gas dispenser (6), stir bar (7), and magnetic stirrer (8).

(20 ϫ 5 ϫ 10 cm; Fig. 1). The remaining volume trol treatments) were done in random order. of the aquarium was filled with aged tap water. Each run began with a 2-wk preincubation pe- Two stir bars, rotating on either side of the cen- riod during which 4 aquaria were filled with tral sediment compartment, homogenized the sediment and allowed to equilibrate at 100% air Њ water. The O2 content of the water body was saturation and 15 C in the dark. After this pre- rd regulated using an O2 control unit (Aquastar V incubation period, fifty 3 -instar larvae (1 lar- 2.0, iks ComputerSysteme GmbH, Karlsbad, va/cm2) were added to each aquarium in ani-

Germany). The O2 concentration of the water mal treatment runs. No larvae were added to was measured next to the sediment compart- aquaria in control treatment runs. Chironomus ri- ment with a calibrated macrosensor. A magnetic parius typically dug into the sediments in Ͻ30 valve opened automatically and N2 gas was min. The O2 content of the aquaria was adjusted blown into the water via a gas dispenser when after 2 d of acclimation. One aquarium was kept

O2 concentrations were above the selected value. at 100% air saturation, and one each was kept at The valve closed as soon as the selected value 45, 30, or 15% air saturation. Larvae were fed rd was reached. The O2 concentration (expressed as every 3 day with 5 mL of food suspension. % air saturation) of the water could be held con- Aquarium water was replaced the day after stant with an accuracy of 5%. food addition to prevent accumulation of detri- Ϫ mental NO2 concentrations. Daily checks of the Ϫ Experimental schedule overlying water showed that NO2 concentra- tions remained well below the tolerance level for Experiments were carried out in 6 consecutive the larvae (20 ␮mol/L; Neumann et al. 2001) at runs. Three runs with chironomid larvae (ani- all times. All manipulations carried out in ani- mal treatments) and 3 runs without larvae (con- mal treatments also were carried out in control 2005] CHIRONOMID RESPONSE TO HYPOXIA 861

FIG. 2. Representative positions of O2 microsensor during measurements for O2 concentration profiles near burrows and chimneys constructed by Chironomus riparius larvae in hypoxia experiments. Letters refer to probe positions when measuring vertical profiles across the sediment–water interface (A), profiles within the diffusive boundary layer surrounding a chimney (B), and time-series profiles at the opening of a chimney (C). Repre- sentation of the chimney (C) is a simplification: chimney is actually an extension of the U-shaped burrow. Microsensor drawings are not to scale.

treatments. Larval performance, O2 uptake rates ments and were found in the water on either of sediments and chimneys, and total bacterial side of the sediment compartment. These larvae abundance in sediments and chimneys were were regarded as having escaped from adverse studied during the 3-wk experimental incuba- conditions in the sediment, and their number tion period after the addition of larvae. was recorded during daily inspections. Dead larvae and pupae and the exuviae left behind Larval performance from emerged animals were collected and counted. The cumulative % of escaped, dead, The 2 types of larval tubes (burrows and and emerged animals was calculated over the chimneys) were quantified in terms of total duration of the experiments. The periodic un- number and average size (length, diameter, and dulation behavior of the larvae inside the bur- thickness of walls) on day 14 of the experimen- rows was verified qualitatively with observa- tal incubations. A cylindrical shape was as- tions through the transparent aquarium wall. sumed for the chimneys, and the dimensions Larval undulation inside chimneys was docu- were used to calculate the relative gain in sed- mented in the form of fluctuating O2 concentra- iment–water interface area and oxic sediment tions near the chimney opening (Fig. 2C). Mi- volume. A few C. riparius larvae left the sedi- crosensors (see below) were used to make sev- 862 P. STIEF ET AL. [Volume 24 eral 240-min time-series recordings (temporal NaCl) and stored in 3 volumes of a 1:1 mixture ϭ resolution 3s)ofO2 concentrations near chim- of phosphate-buffered saline and 96% ethanol at ney openings. Ϫ20ЊC. Samples were diluted 100ϫ with phos- phate-buffered saline and sonicated with an O uptake rates UW70 probe (Bandelin Electronic GmbH, Ber- 2 lin, Germany) at a setting of 3 ϫ 60 s, 20% pulse, Microsensors were used to quantify the dif- and 109 ␮m amplitude. A volume of 75 ␮Lof fusive O2 uptake by sediments and chimneys the sonicated dilutions was immobilized on during weeks 2 and 3 of the experimental in- black polycarbonate membrane filters (pore size ϭ ␮ cubations. O2 microsensors were constructed, 0.2 m; Osmonics, Minnetonka, Minnesota). calibrated, and used for concentration profiling Cells were stained with 4Ј,6-diamino-2-pheny- as described previously (Revsbech 1989, Stief lindole (DAPI) at a concentration of 0.5 ␮g/mL and de Beer 2002). The sensors had tip diame- for 5 to 10 min. Bacterial cells were counted in ters of 10 to 20 ␮m and allowed measurements 20 to 30 randomly chosen microscopic fields on with a spatial resolution of 50 to 200 ␮m. Sensor each filter, giving a total of 700 to 1500 cells tips were positioned relative to the sediment or counted/filter. Cell counts were related to the chimney surface (Fig. 2) with a dissection mi- wet volumes of the sediment slices and the croscope and a motorized micromanipulator chimneys. In treatments with larvae, the chim- that drove the sensors stepwise into the sedi- ney dimensions were used to calculate the rel- ments or towards the chimney surface. Sensor ative increase in oxic sediment volume (see signals were logged on a computer using the above) and the relative increase in the number program InSight (R. Thar, University of Copen- of bacteria in the oxic sediment volume. hagen, Denmark). Diffusive O2 uptake by the sediments and chimneys was calculated from Statistical analysis linear concentration gradients within the boundary layers covering their surfaces using One-way analysis of variance (ANOVA) was Fick’s 1st Law of Diffusion. The diffusion coef- used to identify significant effects of % air sat- Њ ϫ Ϫ5 ficient of O2 at 15 C was taken as 1.83 10 uration on the number of burrows and chim- 2 cm /s (Broecker and Peng 1974). O2 micropro- neys built by the larvae. Each experimental run files were repeated at as many different posi- delivered a single number of burrows and chim- tions as possible to average out lateral hetero- neys in each % air saturation treatment and was Ϫ2 geneities of the samples. H2S and total S con- treated as a replicate. Homogeneity of variances centrations and pH were measured using mi- was checked with Levene’s test. Scheffe´and crosensors (Revsbech et al. 1983, Jeroschewski et Games–Howell post hoc tests were run for pair- al. 1996) in 100% and 15% air saturation treat- wise comparisons when variances were homog- ments. enous or nonhomogenous, respectively. Two- way ANOVA was used to account for the block Bacterial abundance design of the experiment whenever multiple val- ues were obtained in a single treatment (i.e.,

Total bacterial abundance on sediments and depth of burrows, height of chimneys, O2 up- chimneys was measured at the end of each ex- take rates, and bacterial numbers). In these cas- periment. Up to three 2.5-cm-diameter sediment es, the experimental run (block) was used as a cores were taken from each aquarium. Sediment 2nd fixed factor for the analysis. Significant in- cores were positioned to avoid inclusion of teractions between block and % air saturation chimneys in the sample. A 2-mm-thick sediment were not detected. The dependence of cumula- slice (i.e., the feeding layer of C. riparius larvae, tive percentages of escaping, emerging, and dy- Stief and de Beer 2002) was removed from the ing larvae on % air saturation was analyzed us- top of each core and fixed in 4% paraformal- ing a ␹2 test. The dependence of the % gain in dehyde for 1 h. Chimneys Ͼ3 mm in length bacterial cell abundance in the oxic sediment were removed as whole units using tweezers volume of animal treatments on % air saturation and fixed. Fixed samples were washed 3ϫ in also was analyzed using a ␹2 test. Multiple pair- ␣ ϭ phosphate-buffered saline (3 mmol/L NaH2PO4, wise comparisons were made, so ( 0.05) was ␣ ϭ 7 mmol/L Na2HPO4 [pH 7.2], 130 mmol/L Bonferroni-corrected ( /6 0.0083). 2005] CHIRONOMID RESPONSE TO HYPOXIA 863

Results df ϭ 3, p ϭ 0.207). At Ն30% air saturation, only a small number of larvae (Ͻ10%) escaped from Larval performance the sediment compartment (Table 1). At 15% air saturation, the cumulative proportion of es- Burrows and chimneys were observed inside caped larvae was 23.6%. Emergence of adult and on top of the sediments, respectively, with- was independent of % air saturation (␹2 in 24 h after introducing the larvae into the ϭ 0.317, df ϭ 3, p ϭ 0.957) and was Ͻ13.1% in aquaria. Surface sculpturing of the sediment by all % air saturation treatments. Mortality was the larvae was fully developed after 2 wk. Some independent of % air saturation (␹2 ϭ 5.827, df of the burrows were filled with sediment and ϭ 3, p ϭ 0.120). Mortality was ϳ25% at 45% and some of the chimneys had disappeared by the 100% air saturation, ϳ38% at 30% air saturation, 3rd week of the experiments. Therefore, the 2 and ϳ18% at 15% air saturation. types of tubes were routinely quantified on day Spurts of larval undulation in burrows oc- 14 of all experimental runs (Fig. 3). The number curred every 3 to 4 min and lasted for ϳ20 s of U-shaped burrows was not affected by % air each (PS, unpublished data). Spurts of undula- ϭ ϭ saturation (ANOVA, F3,8 0.710, p 0.573), tion in chimneys, inferred from O2 concentration whereas the number of chimneys increased sig- measurements made at the opening of the chim- nificantly as % air saturation decreased (ANO- neys, were less frequent (every 7–8 min) but ϭ ϭ VA , F3,8 6.306, p 0.017; Fig. 3A). The number lasted longer (3–6 min). A current of oxygenat- of chimneys was significantly greater at 15% ed water was drawn into the chimney during than at 100% air saturation (Scheffe´, p ϭ 0.020). each ventilation spurt, whereas anoxic condi- The depth of burrows and height of chimneys tions prevailed at the chimney opening during varied significantly with % air saturation (AN- resting periods (Fig. 4). ϭ Ͻ OVA, burrows: F3,140 6.542, p 0.0001, chim- neys: F ϭ 7.540, p Ͻ 0.0001; Fig. 3B). Burrows 3,192 O uptake rates were significantly less deep under all hypoxic 2 conditions than at 100% air saturation (Games– Vertical O2 concentration profiles were mea- Howell, 45% air saturation: p ϭ 0.050, 30% air sured across the sediment–water interface (Fig. ϭ saturation: p 0.003, and 15% air saturation: p 5A, B), and O2 microdistribution profiles were ϭ 0.002). Chimneys built at 15% air saturation measured around chimneys (Fig. 5C). These were significantly taller than those built at all profiles were used to calculate the oxic sediment other air saturations (Games–Howell, 30% air volume (Table 2) and the diffusive O2 uptake ϭ ϭ saturation: p 0.013, 45% air saturation: p rates by sediments and chimneys (Fig. 6). O2 0.001, and 100% air saturation: p ϭ 0.035). At penetration into the sediments decreased with 100% air saturation, burrows reached through decreasing % air saturation in the water column, the black, sulfidic sediment layer (located at a but O2 penetration was not affected by the pres- depth of 4–7 mm) and extended down to a max- ence or absence of larvae (Fig. 5A, B). The oxic imum depth of 25 mm. Under all hypoxic con- sediment volumes were 2200, 800, 800, and 300 ditions, the deepest point of the burrows gen- cm3/m2 for the 100%, 45%, 30%, and 15% air erally was in the sulfidic sediment layer, i.e., the saturation treatments, respectively. The pres- burrows never reached Ͼ7 mm into the sedi- ence of chimneys increased the oxic sediment ment. Chimneys did not persist in their size and volume substantially (Table 2). The diffusive O2 shape over the 3 wk of the experiment. Instead, uptake rates by the sediments decreased signif- some of them decreased in height or disap- icantly with decreasing O2 concentration in the peared completely only a few days after their water column (ANOVA, control treatment: F3,17 ϭ Ͻ ϭ construction. However, new chimneys were con- 19.887, p 0.0001, animal treatment: F3,75 Ͻ structed until the end of the experimental in- 108.213, p 0.0001; Fig. 6). Diffusive O2 uptake cubation. Whether the larvae destroyed or ate rates by the chimneys also were significantly in- ϭ the chimneys and what the larvae did after the fluenced by % air saturation (ANOVA, F2,20 removal of the chimneys could not be deter- 16.829, p Ͻ 0.0001; Fig. 6). Rates were the same mined. at 45% and 30% air saturation, but rates were The cumulative proportion of escaped larvae significantly lower at 15% than at 30% and 45% was independent of % air saturation (␹2 ϭ 4.555, air saturation (Scheffe´, 30% air saturation: p Ͻ 864 P. STIEF ET AL. [Volume 24

FIG. 3. Mean (ϩ1 SE) number (A) and depths and heights (B) of burrows and chimneys of Chironomus riparius larvae. Within each tube type, columns that share the same letter are not significantly different.

ϭ ϭ ϭ 0.0001, 45% air saturation: p 0.007). Moreover, 4.412, p 0.024, 30% air saturation: F2,25 Ͻ within each air saturation treatment, the diffu- 32.112, p 0.0001, and 15% air saturation: F2,38 ϭ Ͻ sive O2 uptake per unit interface area was al- 11.112, p 0.0001). ways significantly higher for chimneys than for The representative H2S measurements in the the sediment surface (control and animal treat- 100% and 15% air saturation treatments showed ϭ ments) (ANOVA, 45% air saturation: F2,22 that free H2S was not present in the sediments 2005] CHIRONOMID RESPONSE TO HYPOXIA 865

TABLE 1. Mean (Ϯ1 SE) % of larvae added to microcosms that escaped, emerged, or died in laboratory microcosms under different O2 (% air saturation) conditions. Treatments sharing the same letter within a column were not significantly different.

Cumulative % % air saturation Escaped larvae Emerged larvae Dead larvae and pupae 100 7.3 Ϯ 2.0a 10.7 Ϯ 5.9a 27.3 Ϯ 8.9a 45 7.5 Ϯ 3.5a 11.8 Ϯ 7.9a 25.3 Ϯ 8.3a 30 9.4 Ϯ 5.0a 13.1 Ϯ 7.1a 37.7 Ϯ 12.7a 15 23.6 Ϯ 12.3a 10.5 Ϯ 6.0a 18.2 Ϯ 9.0a

or in the water column in animal or control 8.874, p ϭ 0.006; Fig. 7), whereas bacterial abun- treatments (data not shown). Vertical profiles of dance in the sediment increased significantly pH typically showed a decrease from pH 8.3 in with decreasing % air saturations in the animal ϭ Ͻ the water column to pH 7.5 in the sediments treatments (ANOVA, F3,35 7.668, p 0.0001). (data not shown). Percent air saturation did not affect total bacte- ϭ rial counts in chimneys (ANOVA, F2,22 2.569, ϭ Bacterial abundance p 0.099). However, total bacterial counts were significantly higher in chimneys (animal treat- Bacterial abundance in the sediment de- ments) than in sediments (control treatments) at creased significantly with decreasing % air sat- 30% and 15% air saturation (ANOVA, 30% air ϭ ϭ ϭ ϭ uration in control treatments (ANOVA, F3,8 saturation: F2,16 10.640, p 0.001, 15%: F2,18

ϳ ␮ FIG. 4. A.—Time series of O2 concentrations at the opening of a chimney in a 15% air saturation ( 40 mol

O2/L) treatment. The O2 microsensor was positioned level with the rim of the 5-mm-high chimney (Fig. 2C). At the start of each larval ventilation spurt, oxygenated water was drawn into the chimney and passed the microsensor tip; between spurts, anoxic conditions prevailed at the opening of the chimney. B.—Expansion of the temporal scale showing the first 30 min from (A). 866 P. STIEF ET AL. [Volume 24

Ϯ FIG. 5. Mean ( 1 SE) O2 concentrations across the sediment–water interface in control (A) and animal treatments (B) and toward the surfaces of chimneys in hypoxic animal treatments (C). Dotted lines in A and B indicate the sediment–water interface. Dotted line in C indicates the surface of a chimney.

9.120, p ϭ 0.002). Moreover, chimneys increased threshold % air saturation below which the be- the abundance of bacteria in oxic sediment of havior is induced (Kon and Hidaka 1983). the animal treatments (Table 2). The relative The microsensor measurements showed that ␹2 ␹2 gain was dependent on air saturation ( test, the O2 concentration at the sediment–water in- ϭ 28.39, df ϭ 3, p Ͻ 0.0001; Table 2). terface (i.e., at the level of burrow openings) was ϳ50% that of the mixed water column, regard- Discussion less of the % air saturation in the water column

(Fig. 5A, B). Conversely, the O2 concentration at Importance of chimneys for larvae the rim of actively irrigated chimneys was about Chironomus riparius larvae switched gradually the same as that of the mixed water column (Fig. from burrow to chimney construction when ex- 4). Thus, chimneys pierced through the 400- to ␮ posed to a range of decreasing % air satura- 600- m-thick diffusive boundary layer (DBL) tions. The burrows inside the sediment became and made higher concentrations of O2 available less deep, whereas the chimneys on top of the for the larvae (Kon and Hidaka 1983, Jørgensen sediments became more abundant and taller. and Revsbech 1985). The apparent O2 concen- However, small chimneys were observed even at tration at the sediment–water interface seemed 100% air saturation. Therefore, we conclude that to determine the size of the chimney to be con- C. riparius chimney-projecting behavior has no structed. However, whether the larvae respond

Ϯ TABLE 2. Mean ( 1 SE) % changes in sediment O2 dynamics and bacterial distribution associated with the presence of chironomid chimneys in laboratory microcosms. Percent air saturation treatments sharing the same letter were not significantly different.

Chimney-related gain (%) Bacterial abundance in oxic sediment Sediment–water Oxic sediment (relative to sediment bacteria % air saturation interface area volume in animal treatment) 100 2.3 Ϯ 0.3 0.5 Ϯ 0.1 0.0 Ϯ 0.0a 45 11.8 Ϯ 2.2 7.4 Ϯ 1.4 7.4 Ϯ 1.2b 30 8.3 Ϯ 3.0 11.4 Ϯ 1.9 14.6 Ϯ 1.8b 15 31.5 Ϯ 7.7 52.4 Ϯ 12.8 57.5 Ϯ 5.0c 2005] CHIRONOMID RESPONSE TO HYPOXIA 867

Ϯ FIG. 6. Mean ( 1 SE) O2 uptake rates of sediments and chimneys in laboratory hypoxia experiments. Rates were calculated from the concentration profiles presented in Fig. 5. Within each sample type, columns that share the same letter are not significantly different.

to the actual O2 concentration or to a proxy of stagnant conditions (Gray et al. 2002), and low hypoxic conditions (e.g., CO2 or H2S) is unclear. flow velocities lead to the expansion of the DBL ␮ Ͼ O2 depletion may favor the production of H2S from a few hundred mto 1 mm (Jørgensen close to the sediment surface (Holmer and and Revsbech 1985). Thus, chironomid larvae

Storkholm 2001), and H2S could make the larvae may somehow associate the perceived % air sat- shift their burrows from within the sediment to uration with a certain thickness of DBL to be above the sediment surface, but the microsensor penetrated. If so, the inherent response by the measurements showed that H2S was not present larvae to natural DBL dimensions was meaning- in our sediments (data not shown). CO2 also can less in our experiments because the DBL always be ruled out as a proximate factor influencing had the same thickness. chimney construction in our study because hyp- Chimneys also may provide larvae with food. oxic conditions were achieved artificially by Larvae might elongate their burrows above the blowing O2 out with N2 gas. Thus, the low O2 sediment surface if they were in need of bacte- concentrations should not have been associated rial biomass that could be scraped from the bur- with elevated CO2 concentrations. row walls (Olafsson and Paterson 2004). Three One might ask why the chimneys were much lines of evidence support this hypothesis. 1) Ex- higher (up to 8 mm) than the thickness of the trapolation of the O2 microprofiles indicated that DBL (0.4–0.6 mm), and why their height in- the 500-␮m-thick chimney walls were not com- creased with decreasing % air saturation. The pletely anoxic (Fig. 5C), and the lumens of the smallest chimneys found were tall enough to chimneys were flushed periodically with oxy- pierce through the DBL. One reason could be genated water because of larval undulation (Fig. that the thickness of the DBL is inversely cor- 4). Chimneys might allow the attachment and related with the % air saturation of the water growth of aerobic bacteria rather than the an- column under natural conditions. Hypoxia near aerobic bacteria that would grow in less oxy- the sediment surface typically develops under genated burrows. Aerobic bacteria have shorter 868 P. STIEF ET AL. [Volume 24

FIG. 7. Mean (Ϯ1 SE) bacterial abundance at the sediment surface (top 0–2 mm) and in chimneys in labo- ratory hypoxia experiments. Within each sample type, columns that share the same letter are not significantly different. doubling times than anaerobic bacteria. 2) Bac- the increased undulation activity required to ob- terial abundance per unit sediment volume was tain an adequate O2 supply (Leuchs 1986), and generally higher in chimneys than in the feeding larvae require energy for foraging and burrow layer of the sediment (i.e., upper 0–2 mm; Stief construction. Energy might be conserved if lar- and de Beer 2002, Olafsson and Paterson 2004). vae were to use the chimney itself as a food re- 3) Some of the chimneys disappeared a few source rather than expend additional energy days after they had been constructed. These collecting food further away from the burrows. data and observations suggest that chimneys might play a role in the diets of chironomid lar- Importance of chimneys for the benthic microbial vae. Construction of a bacterial colonization site community with distinctly different conditions (i.e., O2 sup- ply) than the sediments and harvesting bacterial Chironomid chimneys were a residual oxic biomass after incubation could be considered a sediment compartment even under severe hyp- gardening strategy (Hershey et al. 1988, Plagan- oxia in our experiment. Chimneys made up 50% yi and Branch 2000). Direct observation of larval of the oxic sediment volume in our experimental feeding and tracing of fluorescent particles aquaria at 15% air saturation, they were densely

(from chimney to larval gut) could provide ev- colonized with bacteria, and they had high O2 idence for this new hypothesis. uptake rates. Therefore, aerobic microbial path- If C. riparius do use chimneys for dietary pur- ways (e.g., nitrification or sulfide oxidation) poses, then one must ask why the larvae apply probably were more important in sediments this strategy only under hypoxic conditions. when chimneys were present than when they

One possible reason could be that low O2 con- were absent. Our study did not address possible centrations increase the energy requirements of aerobic microbial pathways, but we did quantify the larvae (Bairlein 1989, Penttinen and Holo- the diffusive O2 uptake by the sediment surface painen 1995). New metabolic costs arise from and the chimneys. This uptake is a composite 2005] CHIRONOMID RESPONSE TO HYPOXIA 869

of overall microbial O2 consumption and the rial abundance in the sediment in hypoxic con- chemical re-oxidation of reduced compounds ditions, probably because they showed reduced

(Canfield et al. 1993). Diffusive O2 uptake by the grazing at the sediment surface. If reduced flat sediment surface decreased with decreasing grazing in hypoxic conditions were the only rea- % air saturation in the water column regardless son for similar bacterial abundances in inhabit- of the presence of larvae. This positive correla- ed hypoxic sediment and uninhabited normoxic tion probably was caused by the decreasing vol- sediment, then bacterial abundances should ume of sediment in which oxidative respiration have been similar in inhabited and uninhabited and chemical oxidation could take place (Fig. hypoxic sediments. However, under hypoxic 5A, B). However, chimneys had significantly conditions, bacterial abundance was lower in higher diffusive O2 uptake rates than the sedi- uninhabited sediment than in inhabited sedi- ment surface at all % air saturations. The erect ment. Therefore, we assume that physical shape, hollow structure, and periodic irrigation changes in the sediment were induced by the of the chimneys made these biogenic structures larvae and contributed to the observed pattern accessible to dissolved O2 in the water column. of bacterial abundance. Hypoxia causes other Moreover, the chimneys are likely to be enriched sediment-dwelling macrofauna to favor advec- in easily degradable organics from larval saliva tive over diffusive water exchange with the wa- excretions used for stabilization of the chimney ter column because of compensatory bioirriga- (Leuchs and Neumann 1990) and from trapped tion (Forster et al. 1995). This shift also should suspended organic particles (Soltwedel and Vo- occur for chironomids that increase their venti- pel 2001). The dense bacterial colonization on lation activity when exposed to low O2 concen- chimneys might have been both a consequence trations (Leuchs 1986). Advective transport of of O2 and organic substrate availability and a O2 and nutrients stimulates the metabolic activ- cause of the significantly higher diffusive O2 up- ity of bacteria that are otherwise diffusion lim- take rates of the chimneys than sediment (Eck- ited (Ziebis et al. 1996). Thus, inhabited sedi- man 1985). In any case, chimneys contribute to ment may have been a favorable site for bacte- the patchy distribution of bacterial biomass and rial colonization because of intense irrigation in small-scale O2 dynamics in the sediment (Eck- the absence of significant grazing effects. man 1985, Soltwedel and Vopel 2001). Altered behavior of C. riparius larvae under Larvae changed their feeding behavior when hypoxic conditions causes spatial rearrange- exposed to hypoxic conditions. Deposit-feeding ment of sediment bacteria and diffusive O2 up- at the sediment surface was visibly reduced at take by the sediment. Chimney-projecting be- lower % air saturations. In normoxic conditions, havior creates distinct biogenic structures with

C. riparius larvae collect and ingest organic par- high bacterial abundance and high diffusive O2 ticles deposited on the sediment surface (Ras- uptake. In contrast, reduced feeding activity un- mussen 1984). Larvae reduce the abundance and der hypoxic conditions leads to high bacterial metabolic activity of particle-attached bacteria abundance, but low diffusive O2 uptake in the within the feeding layer, and they reduce the surficial feeding layer. We hypothesize that the diffusive O2 uptake by the sediment (Stief and particular geometry of chironomid chimneys de Beer 2002, Altmann et al. 2004). At 100% air (i.e., an erect and hollow structure that is peri- saturation in our study, bacterial abundance in odically irrigated with oxygenated water) favors the sediment was lower when larvae were pre- the growth of metabolically active aerobic bac- sent than when they were absent (Fig. 7), but teria. Thus, chimneys may not only facilitate the the diffusive O2 uptake rate did not differ be- larval acquisition of both O2 for respiration, but tween inhabited and uninhabited sediments. also of microbial biomass for food. The addition of organic food particles to the sediments in both inhabited and uninhabited Acknowledgements sediments may have obscured larval effects on the metabolic activity of sediment bacteria. Gaby Eickert and Ines Schro¨der (Max Planck Bacterial abundances in inhabited sediments Institute, Bremen, Germany) kindly provided us in hypoxic conditions were similar to bacterial with microsensors. We thank Armin Kureck abundances in uninhabited sediments in nor- (University of Cologne, Germany) who lent us moxic conditions. Larvae did not reduce bacte- technical equipment and Raphaela Schoon and 870 P. STIEF ET AL. [Volume 24

Sven Schulz who counted bacteria. This study hypoxia and organic enrichment on the coastal was supported financially by Deutscher Aka- marine environment. Marine Ecology Progress demischer Austauschdienst, Bonn, Germany, Series 238:249–279. and the Max Planck Society. GROOTELAAR,E.M.M.,J.L.MAASAND, AND C. VAN DE GUCHTE. 1996. Protocol for testing of substanc- es in chronic sediment bioassays with the fresh- Literature Cited water dipteran Chironomus riparius. Working doc- ument 94.086X, Report No. WSCE 94–05. Divi- ALTMANN,D.,P.STIEF,R.AMANN, AND D. DE BEER. sion for Ecotoxicology, Institute for Inland Water 2004. Nitrification in freshwater sediments as in- Management and Waste Water Treatment, Lelys- fluenced by larvae: quantification by micro- tad, The Netherlands. sensors and fluorescence in situ hybridization. HAMBURGER, K., P. C. DALL,C.LINDEGAARD, AND I. Microbial Ecology 48:145–153. B. NILSON. 2000. Survival and energy metabolism ALTMANN,D.,P.STIEF,R.AMANN,D.DE BEER, AND in an oxygen deficient environment. Field and A. SCHRAMM. 2003. In situ distribution and activ- laboratory studies on the bottom fauna from the ity of nitrifying bacteria in freshwater sediment. profundal zone of Lake Esrom, Denmark. Hydro- Environmental Microbiology 5:798–803. biologia 432:173–188. BAIRLEIN, F. 1989. The respiration of Chironomus larvae HEINIS,F.,AND T. C ROMMENTUIJN. 1992. Behavioural (Diptera) from deep and shallow waters under responses to changing oxygen concentrations of environmental hypoxia and at different tempera- deposit feeding chironomid larvae (Diptera) of lit- tures. Archiv fu¨ r Hydrobiologie 115:523–536. toral and profundal habitats. Archiv fu¨ r Hydro- BROECKER,W.S.,AND T.-H. PENG. 1974. Gas exchange biologie 124:173–185. rates between air and sea. Tellus 26:21–35. HELLMANN, H. 1994. Load trends of selected chemical BRU¨ CHERT, V., B. B. JøRGENSEN,K.NEUMANN,D. parameters of water quality and of trace sub- RIECHMANN,M.SCHLO¨ SSER, AND H. N. SCHULZ. stances in the River Rhine between 1955 and 2003. Regulation of bacterial sulfate reduction 1988. Water Science and Technology 29:69–75. and hydrogen sulfide fluxes in the central Na- HERSHEY, A. E., A. L. HILTNER,M.A.J.HULLAR,M. mibian coastal upwelling zone. Geochimica et C. MILLER,J.R.VESTAL,M.A.LOCK,S.RUNDLE, Cosmochimica Acta 67:4505–4518. AND B. J. PETERSON. 1988. Nutrient influence on a CANFIELD, D. E., B. B. JøRGENSEN,H.FOSSING,R. stream grazer: Orthocladius microcommunities re- GLUD,J.GUNDERSEN,N.B.RAMSING,B.THAM- spond to nutrient input. Ecology 69:1383–1392. DRUP ANSEN IELSEN AND ,J.W.H ,L.P.N , P. O. J. HOLMER,M.,AND P. S TORKHOLM. 2001. Sulphate re- HALL. 1993. Pathways of organic-carbon oxida- duction and sulphur cycling in lake sediments: a tion in 3 continental-margin sediments. Marine review. Freshwater Biology 46:431–451. Geology 113:27–40. JEROSCHEWSKI, P., C. STEUCKART, AND M. KU¨ HL. 1996. CHOI, J. H., H. ROCHE, AND T. C AQUET. 2000. Effects An amperometric microsensor for the determi- of physical (hypoxia, hyperoxia) and chemical nation of H S in aquatic environments. Analytical (potassium dichromate, fenitrothion) stress on an- 2 Chemistry 68:4351–4357. tioxidant enzyme activities in Chironomus riparius JøRGENSEN,B.B.,AND N. P. REVSBECH. 1983. Colorless Mg. (Diptera, Chironomidae) larvae: potential sulfur bacteria, Beggiatoa spp. and Thiovulum spp., biomarkers. Environmental Toxicology and in O and H S microgradients. Applied and En- Chemistry 19:495–500. 2 2 vironmental Microbiology 45:1261–1270. ECKMAN, J. E. 1985. Flow disruption by an animal J RGENSEN,B.B.,AND N. P. REVSBECH. 1985. Diffusive tube mimic affects sediment bacterial coloniza- ø tion. Journal of Marine Research 43:419–435. boundary layers and the oxygen uptake of sedi- ments and detritus. Limnology and Oceanogra- ECKMAN,J.E.,AND A. R. M. NOWELL. 1984. Boundary skin friction and sediment transport about an an- phy 30:111–122. imal-tube mimic. Sedimentology 31:851–862. KON,M.,AND T. H IDAKA. 1983 Chimney projecting ECKMAN,J.E.,AND D. THISTLE. 1991. Effects of flow behaviour of chironomid larvae (Chironomus yosh- about a biologically produced structure on har- imatsui; Diptera, Chironomidae). Journal of Ethol- pactoid copepods in San Diego trough. Deep Sea ogy 1:111–113. Research Part A 38:1397–1416. LEUCHS, H. 1986. The ventilation activity of Chirono- FORSTER,S.,G.GRAF,J.KITLAR, AND M. POWILLEIT. mus larvae (Diptera) from shallow and deep lakes 1995. Effects of bioturbation in oxic and hypoxic and the resulting water circulation in correlation conditions: a microcosm experiment with a North to temperature and oxygen conditions. Archiv fu¨r Sea sediment community. Marine Ecology Pro- Hydrobiologie 108:281–299. gress Series 116:153–161. LEUCHS,H.,AND D. NEUMANN. 1990. Tube texture, GRAY, J. S., R. S. S. WU, AND Y. Y. O R. 2002. Effects of spinning and feeding behaviour of Chironomus 2005] CHIRONOMID RESPONSE TO HYPOXIA 871

larvae. Zoologisches Jahrbuch fu¨ r Systematik 117: REVSBECH, N. P., B. B. JøRGENSEN,T.H.BLACKBURN, 31–40. AND Y. C OHEN. 1983. Microelectrode studies of

LEVIN, L. A. 2003. Oxygen minimum zone benthos: the photosynthesis and O2,H2S, and pH profiles adaptation and community response to hypoxia. of a microbial mat. Limnology and Oceanogra- Oceanography and Marine Biology 41:1–45. phy 28:1062–1074. NEUMANN, D., M. KRAMER,I.RASCHKE, AND B. GRA¨ FE. ROSENBERG, R., H. C. NILSSON, AND R. J. DIAZ. 2001. 2001. Detrimental effects of nitrite on the devel- Response of benthic fauna and changing sedi- opment of benthic Chironomus larvae, in relation ment redox profiles over a hypoxic gradient. Es- to their settlement in muddy sediments. Archiv tuarine, Coastal and Shelf Science 53:343–350. fu¨ r Hydrobiologie 153:103–128. SOLTWEDEL,T.,AND K. VOPEL. 2001. Bacterial abun- OLAFSSON,J.S.,AND D. M. PATERSON. 2004. Alteration dance and biomass in response to organism-gen- of biogenic structure and physical properties by erated habitat heterogeneity in deep-sea sedi- tube-building chironomid larvae in cohesive sed- ments. Marine Ecology Progress Series 219:291– iments. Aquatic Ecology 38:219–229. 298. PENTTINEN,O.P.,AND I. J. HOLOPAINEN. 1995. Physi- STIEF,P.,AND D. DE BEER. 2002. Bioturbation effects of ological energetics of a midge, Chironomus riparius Chironomus riparius on the benthic N-cycle as Meigen (Insecta, Diptera): normoxic heat output measured using microsensors and microbiologi- over the whole life cycle and response of larva to cal assays. Aquatic Microbial Ecology 27:175–185. hypoxia and anoxia. Oecologia (Berlin) 103:419– VON DANWITZ, B. 1983. Chironomiden und andere 424. Bodentiere eines hochbelasteten Flachsees (Poel- PINDER, L. C. V. 1986. Biology of freshwater Chiron- venn, Nettetal). Diploma Thesis, University of omidae. Annual Review of Entomology 31:1–23. Ko¨ln, Ko¨ln, Germany. PLAGANYI,E.E.,AND G. M. BRANCH. 2000. Does the WIERINGA,E.B.A.,J.OVERMANN, AND H. CYPIONKA. limpet Patella cochlear fertilize its own algal gar- 2000. Detection of abundant sulphate-reducing den? Marine Ecology Progress Series 194:113– bacteria in marine oxic sediment layers by a com- 122. bined cultivation and molecular approach. Envi- RASMUSSEN, J. B. 1984. Comparison of gut contents ronmental Microbiology 2:417–427. and assimilation efficiency of fourth instar larvae ZIEBIS, W., M. HUETTEL, AND S. FORSTER. 1996. Impact of two coexisting chironomids, Chironomus ripar- of biogenic sediment topography on oxygen flux- ius (Meigen) and Glyptotendipes paripes (Edwards). es in permeable seabeds. Marine Ecology Pro- Canadian Journal of Zoology 62:1022–1026. gress Series 140:227–237. REVSBECH, N. P. 1989. An oxygen microsensor with a guard cathode. Limnology and Oceanography 34: Received: 11 November 2004 474–478. Accepted: 2 June 2005