PII: S0043-1354(01)00081-1

Wat. Res. Vol. 35, No. 15, pp. 3625–3634, 2001 # 2001 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/01/$ - see front matter

ARE MEIOFAUNA TRANSIENT OR RESIDENT IN SAND FILTERS OF MARINE AQUARIUMS? SERGE PARENT1,2*, ANTOINE MORIN1 and DANIEL GAGNON2,3 1

Department of Biology, Ottawa-Carleton Institute of Biology, University of Ottawa, 30 Marie-Curie, Ottawa, Canada K1N 6N5; 2 Research and Development Division, Biodoˆme de Montre´al, 4777 Pierre-De Coubertin, Montre´al, Canada H1V 1B3 and 3 Groupe de recherche en e´cologie forestie`re interuniversitaire (GREFi), Universite´ du Que´bec a` Montre´al, C.P. 8888, succ. Centre-Ville, Montre´al, Canada H3C 3P8 (First received 1 April 2000; accepted in revised form 1 October 2000)

Abstract}A paradoxical situation was found in the sand filters of a cold marine mesocosm: meiofaunal masses which were large enough to inhibit the mineralization and nitrification processes coexisted with nitrogen cycling bacteria. To test whether the copepod-dominated meiofauna were resident and actively feeding or transient and carried passively through the sand filters, residence times (RTs) were measured for various meiofaunal groups in a newly started filter and in a long established one. Most meiofauna colonized the newly started filter in less than 6 h, but their RTs were less than 24 h. In contrast, RTs were 147 d for halacarids, 291 d for harpacticoid copepods and 1228 d for nematodes in the long established filter. Mesocosm periphyton, which occupied a large fraction of the mesocosm surface area and was characterized by high meiofaunal densities, was probably the main source of meiofauna in the sand filters. Pool sediments, consisting of gravel or sand, were second to periphyton and contributed hydrozoans and mesopsammic species to the filters. The small copepod Pseudonychocamptus proximus progressively replaced the large Tisbe furcata in sand filters during the fall of 1995 and was responsible for the large increase in meiofaunal biomass observed after spring 1996. This replacement was presumably facilitated by the copepod size selection process operated by the filters. Large copepods were retained by the surface layer of sand or brought up by the backwash water and then exit the mesocosm through the drain. High meiofaunal populations did not significantly affect nitrogen cycling bacteria in sand filters probably because meiofauna also fed on other abundant food sources which were carried in by the water flow. # 2001 Elsevier Science Ltd. All rights reserved Key words}meiofauna, marine aquarium, sand filters

INTRODUCTION

Meiofaunal organisms are a natural component of water treatment systems (Andersson et al., 1994) and of life support systems such as marine aquariums, microcosms and mesocosms (Adey and Loveland, 1998; Boucher and Chamroux, 1976). They have been found in sediments, on various filtration media and even on glass surfaces (e.g. Placozoans; Ruppert and Barnes, 1994). Much attention has been given to the role of meiofauna in the nitrifying filters of water treatment systems. Filter fly larvae (Psychodidae) and snails (Physa sp.), both meiofaunic in size, were found to completely eliminate nitrification at times in pilot tertiary trickling filters (Parker et al., 1989). Rotifers and nematodes were also shown to reduce nitrifica-

*Author to whom all correspondence should be addressed. Biodoˆme de Montre´al, 4777 Pierre-De Coubertin, Montre´al, Canada H1V 1B3. E-mail: sergeparent@ville. montreal.qc.ca

tion by more than half in a suspended carrier biofilm reactor (Lee and Welander, 1994). However, there has been little study of meiofaunal activity in other artificial habitats. Meiofaunal populations of up to 40 g m 2 were recorded in the sand filters of the St. Lawrence mesocosm (SLM) at the Montreal Biodome (Parent and Morin, 1999a). These populations consisted of 22 species and were largely dominated in both number (91%) and biomass (98%) by harpacticoid copepods (Table 1). Experiments performed in heterotrophic microcosms replicating the SLM sand filters showed that meiofauna significantly affect the nitrogen cycle. Meiofauna reduced the apparent mineralization rate of detritus at rates of 9–36% g 1 dry mass of meiofauna m 2. Also, at low levels of particulate organic nitrogen, meiofauna lowered the apparent nitration rate at a rate of 20% g 1 dry mass of meiofauna m 2. Given the high meiofauna biomasses found in the SLM sand filters, little or no mineralization or nitrification would take place if meiofauna were as active in the sand filters as they

3625

3626

Serge Parent et al.

Table 1. Relative abundances of meiofauna in the sand filters of the St. Lawrence mesocosm

MATERIALS AND METHODS

Site description Taxon

Abundance (individuals 10 specimens)

Copepoda harpacticoida Pseudonychocamptus proximus Tisbe furcata Paramphisciella (fulvofasciata?)

911

Nematoda Anticoma sp. Chomadoridae sp. Monhystera sp.

76

Rotifera Colurella (colurus?) Dicranophoridae sp.

7.0

Turbellaria Acoela sp. Trigonostominae sp. Species A Species B

2.3

Polychaeta Dinophilus gyrociliatus Nerilla sp.

0.4

Gastrotricha Chaetonotida sp.

2.0

Acarina Copidognathus biodomus nov. sp. Isobactrus setosus Thalassarachna basteri

0.3

Hydrozoa Clavidae sp. Corynidae sp. Hydractinia echinata Pandeidae sp.

1.0

3

The SLM is an indoor closed system that holds 3  106 L of cold (108C) artificial seawater (28–30%) (Fig. 1). It exhibits fish, invertebrates and seabirds from the Gulf of St. Lawrence in two display pools: the main pool (2.5  106 L) and the tidepool (25,000 L). The bottom of the main pool (475 m2) is either bare (35%) or covered with artificial rocks (32%), gravel (27%) and sand (6%). The bottom of the tidepool (68 m2) is covered with artificial rocks (69%), gravel (23%) and sand (8%). The gravel used is well sorted granite gravel of 12 mm grain size, whereas the sand is poorly sorted horticultural sand of 0.5 mm grain size. The water treatment system consists of six sand filters, a 200 m3 trickle filter filled with 9 cm Lanpac1 plastic bioballs (Lantec Products Inc., Agoura Hills, CA 91301), and an ozonation system. A water level indicator made of a 6 m long clear PVC pipe (49 mm internal diameter) runs along the side of the trickling filter. The sand filters are cylinders 3.6 m in diameter, 3 m high, and contain 65 cm of fine sand (250–300 mm effective particle size). Water flows through the filters at 6 mm s 1 and the hydraulic retention time (HRT) is 5 min. Filters are backwashed with filtered seawater every 4 d or so. Backwash flow is about 9 mm s 1 and lasts for 6– 8 min. Backwash water is sent to a 150,000 L reservoir where it is either filtered again by two filters 1.8 m in diameter and reused later, or sent to the drain. Backwash of the reservoir filters is done with tapwater and sent to the drain. Both pools and the clear PVC piping are sunlit (7–30 W m 2 direct light at the surface of the pools). Periphyton covers all hard surfaces (concrete walls and bottom, artificial and natural rocks) in the main pool (807 m2). The green algae Cladophora glomerata was the dominant and sole structuring element in 1994. Since then, two other structuring elements, the red algae Pterothamnion plumula and the yellow–green algae Vaucheria sp., have appeared and have become dominant in disturbed areas (Table 2). RTs in a newly started filter

are in the microcosms and in the nitrifying filters. However, mineralization and nitrification are believed to occur normally in the SLM: a sharp, steady increase in nitrates (9.1 mM d 1) was measured between 1992 and 1996 without any noticeable trace of ammonia during that period (Parent and Morin, 2000). SLM filters therefore seem to exhibit a paradox: how do high meiofaunal masses coexist with nitrogen cycling bacteria in the sand filters of a marine system? There are two possible explanations for this paradox. Meiofauna could simply be transient in the sand filters, carried by the water flow. Or, they could be resident and exploit another food source, absent from the microcosms, without affecting the mineralization and nitrification processes. To examine these two possibilities, we estimated the residence time (RT) of various meiofaunal groups in a newly started filter and in a long established filter. Vertical density profiles were taken for the main meiofaunal groups in a filter sand column, and the mean lengths of harpacticoid copepods were compared at different depths within the filter. Possible sources of meiofauna were also sought by sampling other habitats in the St. Lawrence mesocosm.

A long established filter was shut down, drained, filled twice with tapwater and left to stand for 48 h. At startup and then at repeated time intervals, 500 L of the outcoming water were filtered in 9–12 min over a 45 mm sieve (Fig. 1). Pre-filtered water was also sampled twice in the same way 6 h after startup. Samples were sorted within 48 h and live specimens were counted. The residence time (RT) of a given taxon was set as the ratio of filter population to output per hour. Since there were no live meiofauna in the filter at the outset, meiofaunal populations were estimated using meiofaunal abundances entering and exiting the filter during different time periods. A constant inflow of meiofauna was assumed throughout the 6 h experiment. RTs in a long established filter Vertical profiles of meiofauna in the sand column of one long established (6 yr) filter were done using 35 cm3 syringes (4.44 cm2 cross-section area) and a hand-held piston-style corer 75 cm long and 1.9 cm in diameter (2.94 cm2 crosssection area). Three 66 cm long cores were taken in April 1998 and three 28 cm long cores in July 1998. Each core was cut into 4 cm long slices (11.5 cm3) that were then preserved in 5% formalin. Meiofauna were extracted by decantation (Pfannkuche and Thiel, 1988) and then collected on a 45 mm sieve. By closely examining 1 cm3 of residue of randomly chosen samples, decantation was estimated to be at least 95% effective in extracting meiofauna. Filter populations were estimated by integrating mean densities over the entire sand column. Meiofauna outputs were also repeatedly sampled from different filters in July 1998. For each sample,

Meiofauna in marine sand filters

3627

——) normal water circulation in sand Fig. 1. Hydraulic diagram of the St. Lawrence mesocosm showing (—— filters 1, 2 and 3, and ( – – – ) the path for a filter backwash in sand filters 4, 5 and 6. CP=Clear piping; F1–F6=Sand filters; G=Gravel; MP=Main pool; OT=Ozone tower; R=Rockwork; RE=Reservoir; RF=Reservoir filter; S=Sand; SP=Sampling port; TF=Trickling filter; and TP=Tidepool.

Table 2. Algae occurrences in the periphyton of the main pool of the St. Lawrence mesocosm in 1994 and 1999

Division Cyanobacteria (blue–green algae)a Lyngbya diguetti Gomont Lyngbya taylorii Drouet & Strickland Oscillatoria tenuis Agardh Spirulina subsalsa Oersted

1994

1999

Dispersed Dispersed Sub-dominant

Dominant Dispersed Dispersed Dispersed

Division Rhodophyta (red algae) Pterothamnion plumula (J. Ellis) Na¨geli Division Chlorophyta (green algae) Cladophora glomerata kuetzingianum (Grunow) Heering Sphaerocystis schroeteri Chodat Division Heterokontophyta Class Bacillariophyceae (diatoms) Amphiprora alata sp. Ku¨tz. Caloneis bacillum (Grun.) Mereschkowsky Diploneis ovalis Fragilaria intermedia Grun. Fragilariopsis pseudonana (Hasle) Hasle Gomphonema acuminatum Ehr. Navicula gracilis Ehr. Navicula radiosa tenella (de Brebisson) Grun. Nitzschia inconspicua Grun. Nitzschia tryblionella debilis Pseudo-Nitzschia seriata Cleve (Peragallo & Peragallo) Surirella spiralis Ku¨tz. (Arnott.) Mayer Synedra acus Ku¨tz. Class Xanthophyceae (yellow–green algae) Vaucheria sp. De Candolle a

Classification based on Hoek et al., 1996.

Primary structuring element Primary structuring element

Sub-dominant

Dispersed

Primary structuring element Dispersed

Rare Dominant Dispersed Dominant Dominant Rare Rare Rare Rare Dispersed Dispersed Rare Rare Secondary structuring element

3628

Serge Parent et al.

1000 L were taken. Outcoming water was filtered over a 45 mm sieve and backwashed water over a 63 mm sieve (sampling ports are shown in Fig. 1). Meiofauna were alive when sampled. They were put in 5% formalin and counted later. Subsamples of 150 copepods were taken from different levels in one filter and total length (directly from rostrum to caudal rami; Volkmann-Rocco, 1972) was recorded. Differences between samples were analyzed using one-way ANOVA followed by the Tukey means comparison test. The RT of a given meiofaunal taxon was set as the ratio of filter population to total output per day. Sampling of meiofauna in other habitats Different techniques were used to sample meiofauna in other habitats of the SLM because of the wide variety of substrates (sediments, rockwork, bioballs, etc.). Periphyton was sampled 9 times during the summer of 1999. Divers collected samples in different areas of the main pool using a 60 cm3 brushing syringe (5.53 cm2 cross-section area) connected to a collector syringe (Loeb, 1981). Samples were preserved in 5% formalin and sorted later. Thirteen samples of five bioballs each were hand collected at different depths in the tricking filter in the fall of 1996 and summer of 1997.

They were put in an ultrasonic bath with 2–3% formalin for 15 min and specimens were collected over a 45 mm sieve. Fifteen samples of the 0–2.5 cm layer of sand were collected in the main pool and in the tidepool in spring 1995 and December 1999 using 10 cm3 syringes (1.63 cm2 cross-section area). Meiofauna were extracted by decantation over a 45 mm sieve and were sorted live. The content of the clear pipe indicating the water level in the trickling filter was sampled once in August 1995. The content of the pipe was repeatedly filtered over a 45 mm sieve until no more animals were collected. Specimens were counted live. Gravel meiofauna were sampled in 1994–1996 in the main pool (13 samples) and in the tidepool (14 samples) in a semiquantitative way. Samples of the surface gravel (0–2.5 cm) were hand collected with 500 ml plastic bottles and fixed in 5% formalin. Meiofauna were later extracted by decantation and collected in a 45 mm sieve except for Hydrozoans that were individually counted on a subsample of gravel grains. To facilitate comparisons between habitats, faunal densities in the sediments and in the trickling filter were expressed as the number of individuals found in the 0– 2.5 cm layer of substrate. Also, in accordance with accepted convention (Fleeger et al., 1988), all results were standardized to a 10 cm2 area.

Fig. 2. Abundance of various meiofaunal groups in a newly started filter before filtration (left side) and after filtration (right side) at different times after startup. Copepodites include adult copepods.

Meiofauna in marine sand filters RESULTS

RTs in a newly started filter There was no progressive increase in meiofaunal abundance in filtered water after filter startup. Abundances peaked less than 10 min after startup for all taxa except rotifers which were never numerous (Fig. 2). The sampled water stank and its temperature was 168C, suggesting that incoming seawater was mixed with stagnant filter freshwater. About 20 min after startup, sampled water was close to normal temperature (128C) and salinity. Harpacticoid copepods and rotifers started residing in the filter less than 1 h after startup. No copepod nauplii were collected 20–60 min after startup (Fig. 2), which resulted in an infinite RT. Then, during the rest of the experiment, their RT was 18–23 h. The RT of copepodites and adult copepods increased progressively from 0.3 h after 38 min of operation to 6.7 h after 6 h of operation. Rotifers behaved similarly to copepod nauplii. Their RT was

3629

infinite at startup and then leveled down to 18–22 h after 5–6 h of operation. In contrast, nematodes and halacarids did not start residing in the filter soon after startup. Nematodes were always as numerous in the outflow as in the inflow (Fig. 2), and their RTs remained equal to the HRT. The RTs of halacarids also equalled the HRT during the first 2 h of operation. Their outflow abundance then dropped and RTs of 3–8 h were measured after 4–6 h of operation. RTs in a long established filter The RT for harpacticoid copepods, nematodes and halacarids in the long established filter varied between 147 and 1228 days in July 1998 (Table 3). At that point, the sand filter contained 2 million halacarids and 950 million harpacticoid copepods. The vertical profiles showed that meiofauna were found throughout the filter sand column (Fig. 3). Densities of harpacticoid copepodites, adult

Table 3. Filter population, outputs and residence time (days) of harpacticoid copepods, nematodes and halacarids in a long established filter in July 1998 (mean  SD (sample size))  106 individuals

Copepods

Nematodes

949.4  45.0 (3)

Halacarids

542.2  109.7 (3)

1.9  0.64 (3)

Returned to pools/day Backwashed to drain/day Total output/day

1.51  0.40 (10) 1.75  0.55 (12) 3.26  0.68

0.07  0.04 (11) 0.37  0.04 (4) 0.44  0.05

0.01  0.006 (10) 0.004  0.001 (4) 0.013  0.006

Residence time (RT)

291  63

1228  295

Filter population

147  86

Fig. 3. Vertical density profiles of various meiofaunal groups in a filter sand column in April and July 1998. Copepodites include adult copepods.

3630

Serge Parent et al.

copepods and nematodes were highest in the top 2.5 cm of sand. They gradually decreased in the 2.5– 10 cm layer, and were then almost constant throughout the rest of the sand column. Copepod nauplii, rotifers and halacarids had no clear distribution pattern other than having densities somewhat greater in the top 10 cm of sand. The way meiofauna exited the filter varied from taxon to taxon. Nematodes were 5 times more abundant in backwash water than in water returning to the pools (Table 3). The opposite was observed for halacarids, whereas there was no apparent difference in outputs for harpacticoid copepods. However, nauplii, which constituted 36% of the copepod outputs, were 4 times more abundant in water returning to the pools than in backwash water. They had a RT (1649 d) 8 times higher than the RT of copepodites and adult copepods (198 d), which suggests that most nauplii reached copepodite stage before leaving the filter. Size selection took place in the surface layer of the filter sand column. Harpacticoid copepodites and adult copepods exhibited similar size distributions in the 14–18 cm layer of sand and in the water returning to the pools (Fig. 4). However, both groups were significantly smaller than those entering the filter or being carried by the backwash water, which means that size selection was taking place in the surface layer of the filter sand column. A species selection process was also superimposed on the size selection

Fig. 4. Body length (mean+SE) of copepodites and adult harpacticoid copepods (upper graph) and relative abundance of Tisbe sp. (lower graph) at different levels in a long established filter. Letters indicate significant differences from a Tukey grouping (p50.05).

process. The relative abundance of Tisbe sp. was 7 times greater in the backwash water than in the water returning to the pools (Fig. 4), whereas that of Pseudonychocamptus proximus was almost the same in both samples (93% vs 99%).

Meiofauna in other mesocosm habitats Meiofauna were abundant in mesocosm habitats other than the sand filters (Table 4). Densities of up to 309 individuals per 10 cm2 were found in the periphyton which occupies the largest surface area in the mesocosm. Very high densities of hydrozoans (188 individuals per 10 cm2) were observed at the surface of the tidepool gravel. On the other hand, the trickling filter bioballs contained very little meiofauna and the polychete Dinophilus gyrodactylus was the only meiofaunal species present in the clear pipes. Large densities of copepods, nematodes, rotifers and acarians were found in the periphyton as well as in the top 2.5 cm of sediments. They dominated in these habitats as they did in the sand filters. An acarian species new to science, Copidognathus biodomus (Bartsch, 1997), was discovered in the tidepool gravel. Gastrotrichs and hydrozoans were found only in gravel, whereas turbellarians were equally abundant in periphyton, gravel and sand.

DISCUSSION

Results show that most meiofauna rapidly colonized the newly started filter. They also show that the RTs of meiofaunal taxa varied with filter age. RTs were less than 1 d in a newly started filter and 147 d or more in a long established filter. Meiofauna are therefore resident in a long established filter and do in fact feed there. Harpacticoid copepods and nematodes probably complete at least one life cycle in a long established filter since generation time in similar laboratory conditions is 29–36 d for harpacticoid copepods, and 24–54 d for nematodes (Hick, 1983; Heip et al., 1985). Halacarids, on the other hand, probably spend only part of their 180–365 d life cycle in a long established filter (Green and MacQuitty, 1987). Not all meiofaunal taxa colonized the newly started filter (e.g. nematodes) and those which did so remained in the filter for much less time than in a long established filter. The reason for this is not the presence of stagnant freshwater at startup because water salinity and temperature were back to normal less than 20 min after startup. We believe the reason is the scarcity of food in the filter. The filter was drained and filled twice with tapwater which probably cleaned out most organic matter before the experiment took place. It then took many days for the filter to fill with enough detrital organic matter, detached algae and biofilm to necessitate a backwash. We suspect that nematodes started colonizing the

Meiofauna in marine sand filters

3631

Table 4. Densities of meiofaunal taxa and species found in various habitats of the St. Lawrence mesocosm Density (individuals 10 cm 2, mean  SD) Habitat Location Surface area (m2) Sample size

Periphyton Main pool 807 9

Gravela Main pool 126 13

Sanda Main+Tidepool 35 15

Gravela Tidepool 7.5 14

Bioballsa Trickling filter 55 13

Clear pipes Piping 0.7 1

Copepoda harpacticoida (copepodites and adults) Paramphisciella (fulvofasciata?) Pseudonychocamptus proximus Tisbe (furcata)

33.2  36.9

18.8  31.2

2.9  5.1

10.8  8.9

0.07  0.10

0

X X

X X X

X X

X X X

X

Copepoda harpacticoida (nauplii) Nematoda Anticoma sp. Chomadoridae sp. Monhystera sp.

64.9  47.3 202  167 X X

20.3  25.6 10.7  27.4 X X

7.8  22.6 14.8  26.0 X X X

23.9  30.1 92.2  119.1 X X

0.002  0.004 0.022  0.015 X

0 0

Rotifera Colurella (colurus ?) Dicranophoridae sp.

30.1  29.9

0.5  0.6 X

1.2  2.5 X

1.6  4.5 X

00

0

Turbellaria Acoela sp. Microstomum sp. Species A Species B Trigonostominae sp.

4.0  3.7 X

5.8  11.7 X

5.3  9.8

0.4  0.8 X

00

X

X

Polychaeta Dinophilus gyrociliatus Nerilla sp.

00

2.2  5.0 X X

2.5  5.6

Gastrotricha Chaetonotida sp.

00

00

Acarina Copidognathus biodomus nov. sp. Isobactrus setosus Thalassarachna basteri

5  7.3

0.1  0.4 X

X

X

Hydrozoa Clavidae sp. Corynidae sp. Hydractinia echinata Pandeidae sp.

00

2.8  5.1

a

X

X

X X

X X X 00

00

5.5 X

00

0.03  0.1 X

00

0

00

1.1  2.8 X X X

0.001  0.004

0

00

188  142 X X X X

00

X

X 0

In the top 2.5 cm.

filter when their preferred food items were sufficiently abundant in the filter. On the other hand, some meiofaunal species never colonized the filter. These species, such as colonial hydrozoans, do not live in fine sand (Thiel, 1988). They thrive in other habitats of the SLM and are carried into the sand filters by water flow or through vacuum cleaning of the pools (Parent and Morin, 2000). Meiofauna first established themselves in habitats of the SLM mesocosm other than the sand filters. They then colonized and bred in the sand filters when conditions became favourable. Periphyton was probably the main source of meiofauna to the sand filters since it occupies a large surface area}4 times the total area of sediments and trickling filter areas}and was characterized by high meiofaunal densities (Table 4). This would explain why filter meiofauna were mainly composed of epibenthic species such as the prehensile harpacticoid Pseudonychocamptus proximus and the strongly adhesive nematode Anticoma (Table 1). Pool sediments, consisting of gravel or sand, are second to periphyton in surface area and meiofaunal densities. They were the only source of

hydrozoans}all of which are epibenthic}to the filters, as well as of the mesopsammic species, such as the harpacticoid Paramphisciella, the gastrotrich Chaetonotida sp. and the turbellarian Microstomum sp. Clear pipes are too small a habitat, and trickling filter bioballs contained too low meiofaunal densities to significantly contribute to the mesocosm meiofauna. Moreover, we suspect that trickling filter meiofauna were only transient between the sand filters and the pools. This situation is very different from freshwater trickling filters where large meiofaunal populations are resident and greatly inhibit nitrification (Andersson et al., 1994; Parker et al., 1989). It is probable that marine taxa do not resist the harsh living conditions of trickling filters as well as freshwater taxa do. Time series data on filter meiofauna were collected for one SLM filter between 1994 and 1998 using sets of 0–2.5 cm sand samples taken with 35 cm3 syringes (Fig. 5). The data revealed that Pseudonychocamptus proximus started replacing Tisbe furcata in the fall of 1995, and was then responsible for the large increase in meiofaunal biomass observed starting in spring 1996. The copepod size selection observed in the

3632

Serge Parent et al.

Fig. 5. Meiofaunal mass (upper graph, mean+1 SD) and copepod species composition (lower graph) in the 0–2.5 cm sand layer of the SLM filters between July 1994 and July 1998. Tisbe are shown in black, Paramphiascella in grey and Pseudonychocamptus in white.

surface layer of the sand column probably facilitated and accelerated that replacement. As shown in Fig. 4, the selection was clearly exerted against Tisbe sp., which reached a larger maximum size (11 mg DW, 1.44 mm TL) in the SLM than Pseudonychocamptus proximus (4.0 mg DW, 0.70 mm TL; S. Parent, unpublished data). Sampling meiofauna in the surface layer of a filter sand column was an easy and useful way of obtaining a rapid estimate of a filter’s total meiofauna. It was subject to a relatively small variance (Table 4) because of the well sorted substrate, the uniform flow of water through the sand column and the regular backwashing of the sand. The high RT of harpacticoid copepods in a long established filter seems to contradict the low Production : Biomass (P : B) ratio measured by Boyer (1999) in the same sand filters. Because no increase in total length was observable in the deeper part of the filter sand column (Fig. 4), Boyer assumed that harpacticoid copepods did not grow significantly in sand filters. However, results presented here suggest that in fact copepods are productive in all parts of the filter sand column. As shown in Fig. 4, the filter operates a size selection on harpacticoid copepods: large copepods are removed by the backwash, while small copepods are returned to the pools. The backwash action of the filter constantly brings up the large copepods so that only small copepods are buried deep in the sand when the backwash is completed. Also, because of their size, it may be easier for large copepods to stay in the surface layer of the sand as shown by their abundance in the

vertical profiles (Fig. 3). They would stay near the surface either for feeding reasons}fresh detritus accumulate there and form an ‘‘Aufwuchs’’, or as a result of a purely physical process}big copepods cannot bury nor fit between sand grains and remain near the top, whereas small ones are pushed deep into the sand by the water flow. Therefore, a large portion of the copepod production is constantly removed by the backwash and sent down to the drain, which means that the P : B ratio of harpacticoid copepods is higher than found by Boyer (1999). Given the favourable conditions prevailing in a sand filter} stable temperature and salinity, water saturation with dissolved oxygen, abundance of food and absence of macrofaunal predators –, the P : B ratio of harpacticoid copepods is possibly similar or even larger in sand filters than in natural habitats. Consequently, if meiofauna are resident in long established sand filters, why do they not affect mineralization and nitrification processes as much as in the experimental microcosms of Parent and Morin (1999a,b)? We propose that meiofauna in sand filters of marine aquariums have access to and thrive on food sources which were not available in experimental microcosms. Most of the meiofauna found in the sand filters feed on bacteria as well as on microalgae and diatoms: Pseudonychocamptus (Decho and Moriarty, 1990), Tisbe (Hicks, 1983), Anticoma (Wieser, 1959) and Colurella (ThaneFenchel, 1968). The SLM periphyton is an important source of diatoms and microalgae for the filters (Parent and Morin, 2000). Preliminary data (S. Parent, unpublished data) suggest that mesocosm

Meiofauna in marine sand filters

bacterioplankton and phytoplankton are also retained by the filters and therefore may serve as food for meiofauna. All these types of food were not available to meiofauna in experimental microcosms because they were kept in the dark and contained too little water (13.5 L) to sustain bacterioplankton in significant quantities. Furthermore, a large volume of water circulates through the SLM sand filters and continuously replenishes the food supply. Contrary to the situation in the SLM sand filters, benthic algae and plankton are not available to meiofaunal taxa living in freshwater trickling filters. All taxa found there are grazers and the biofilm is the only food they have available. Benthic algae, plankton and other organic particles are either retained anteriorly by sand filters or carried through the trickling filter by water flow. They are not spread onto a surface to scrape. This probably explains why the large meiofaunal populations found in trickling filters so greatly inhibit nitrification (Andersson et al., 1994; Parker et al., 1989). The coexistence of high meiofaunal biomasses with nitrogen cycling bacteria is to be found in the sand filters of any marine system, as long as it has a periphyton area or, at least, a large area of fine gravel or sand where meiofauna (inadvertently carried in with macroinvertebrates, live colonial structures or macroalgae) can establish themselves before invading the sand filters. Coexistence should also occur in the sand filters of similarly developed freshwater aquariums. Water flow therefore does not prevent meiofauna from colonizing a sand filter. At first, meiofauna are temporarily resident or only transient in sand filters. They must complete their life cycle in the mesocosm periphyton or sediments. Then, as the bacterial biofilm forms and the products of filtration accumulate on the surface of, and throughout the filter sand column, meiofauna become resident and complete their life cycle within the filter. Although more direct information is needed to assess the exact status of nitrogen cycling bacteria in sand filters, we know that they are present and that high meiofaunal populations do not affect them as much as they do in marine microcosms or freshwater trickling filters. The reason is not that meiofauna populations are transient or non-productive, but rather that they feed on other abundant sources of food which are brought in by the water flow of the life support system. Acknowledgements}This research was supported by the Montreal Biodome and by a grant to A. Morin from the Natural Sciences and Engineering Research Council of Canada. We are grateful to Dr. Judith Fournier (Polychetes), Dr. Dale R. Calder (Hydrozoans), Dr. Ilse Bartsch (Acarians), Serge Paquet (Algae) and Michel Cle´ment (Harpacticoid Copepods) for their taxonomic determinations. We also wish to thank Jean Bouvrette, Mec. Eng. and Maintenance and Operation Manager, and Normand Desrochers, Mec. Eng., for their technical help and cooperation, as well as the Biodome diving team (S. Ethier, J.-P. Genet, R. Savignac and P. Sylvestre) for collecting the

3633

pool samples. Brian Colwill, C.Tr., kindly revised the English style and grammar, and two anonymous reviewers provided helpful comments on an earlier version of this paper.

REFERENCES

Adey W. H. and Loveland K. (1998) Dynamic Aquaria: Building Living Ecosystems. Academic Press, Toronto, 498 pp. Andersson B., Aspegren H., Parker D. S. and Lutz M. P. (1994) High rate nitrifying trickling filters. Water Sci. Tech. 29(10–11), 47–52. Bartsch I. (1997) Copidognathus biodomus (Halacaridae: Acari), a new species from eastern Canada. Mitt. Hamb. Zool. Mus. Inst. 94, 153–159. Boucher G. and Chamroux S. (1976) Bacteria and meiofauna in an experimental sand ecosystem. I. Material and preliminary results. J. Exp. Mar. Biol. Ecol. 24, 237–249. Boyer J.-F. (1999) The estimation of the P : B ratio for harpacticoid copepods in the sand filters of a cold marine mesocosm. Honours Thesis, Dalhousie University, Halifax, 37 pp. Decho A. W. and Moriarty D. J. W. (1990) Bacterial exopolymer utilization by a harpacticoid copepod: a methodology and results. Limnol. Oceanogr. 35(5), 1039–1049. Fleeger J. W., Thistle D. and Thiel H. (1988) Sampling Equipment. In Introduction to the Study of Meiofauna, Vol. 7, eds R. P. Higgins, H. Thiel, Smithsonian Institution Press, Washington, pp. 115–125. Green J. and MacQuitty M. (1987) Halacarid Mites (Arachnida: Acari). In Synopsis of the British Fauna Vol. 36. eds D. M. Kermack and R. S. K. Barnes, Linnean Society of London, London, pp. 1–177. Heip C., Vincx M. and Vranken G. (1985) The ecology of marine nematodes. Oceanogr. Mar. Biol. Ann. Rev. 23, 399–489. Hick G. R. F. (1983) The ecology of marine meiobenthic harpacticoid copepods. Oceanogr. Mar. Biol. Ann. Rev. 21, 67–175. Hoek C. van den, Mann D. G. and Jahns H. M. (1996) Algae: An Introduction to Phycology. Cambridge University Press, Cambridge, 623 p. Lee N. M. and Welander T. (1994) Influence of predators on nitrification in aerobic biofilm processes. Water Sci. Tech. 29, 355–363. Loeb S. L. (1981) An in situ method for measuring the primary productivity and standing crop of the epilithic periphyton community in lentic systems. Limnol. Oceanogr. 26(2), 394–399. Parent S. and Morin A. (1999a) The role of copepoddominated meiofauna in the mineralization of organic matter in a cold marine mesocosm. Can. J. Fish. Aquat. Sci. 56, 1938–1948. Parent S. and Morin A. (1999b) Role of copepoddominated meiofauna in the nitrification process of a cold marine mesocosm. Can. J. Fish. Aquat. Sci. 56, 1639–1648. Parent S. and Morin A. (2000) N budget as water quality management tool in closed aquatic mesocosms. Water Res. 34(6), 1846–1856. Parker D., Lutz M., Dahl R. and Bernkopf S. (1989) Enhancing reaction rates in nitrifying trickling filters through biofilm control. J. Water Pollut. Control Fed. 61(5), 618–631. Pfannkuche O. and Thiel H. (1988) Sample Processing. In Introduction to the Study of Meiofauna, Vol. 9, eds R. P. Higgins and H. Thiel, Smithsonian Institution Press, Washington DC, pp. 134–145.

3634

Serge Parent et al.

Rupert E. E. and Barnes R. D. (1994) Invertebrate Zoology. Saunders College Publishing, Montreal. Thane-Fenchel A. (1968) Distribution and ecology of nonplanktonic brackish-water rotifers from Scandinavian waters. Ophelia 5, 273–297. Thiel H. (1988) Cnidaria. In Introduction to the Study of meiofauna, Vol, 19. eds R. P. Higgins and H. Thiel,

Smithsonian Institution Press, Washington DC, pp. 266–272. Volkmann-Rocco B. (1972) Species of Tisbe (Copepoda, Harpacticoida) from Beauport, North Carolina. Arch. Oceanogr. Limnol. 17, 223–258. Wieser W. (1959) Free-living marine nematodes}IV. General Part. Acta Univ. Lund. N.F. Adv. Ser. 2, 55, 1–111.