Oecologia (2002) 133:402–411 DOI 10.1007/s00442-002-1001-x
COMMUNITY ECOLOGY
G. Mancinelli · M. L. Costantini · L. Rossi
Cascading effects of predatory fish exclusion on the detritus-based food web of a lake littoral zone (Lake Vico, central Italy)
Received: 6 February 2002 / Accepted: 15 June 2002 / Published online: 5 September 2002 © Springer-Verlag 2002
Abstract An exclosure experiment was carried out in the reed-dominated littoral zone of a volcanic lake (Lake Vico, central Italy) to test whether the impact of predatory fish on benthic invertebrates cascades on fungal colonisation and breakdown of leaf detritus. The abundance, biomass, and Shannon diversity index of the invertebrate assemblage colonising Phragmites australis leaf packs placed inside: (1) full-exclosure cages, (2) cages allowing access only to small-sized fish predators, and (3) cageless controls, were monitored over a 45-day period together with the mass loss and associated fungal biomass of leaf packs. The species composition of the fungal assemblage was further assessed at the end of the manipulation. In general, invertebrate predators did not show any significant response to fish exclusion, either on a trophic guild or on a single taxon level. In contrast, the exclusion of large predatory fish induced a diverse spectrum of changes in the abundance and population sizestructure of dominant detritivore taxa, ultimately increasing the biomass and Shannon diversity index of the whole detritivorous guild. These changes corresponded with significant variations in leaf detritus decay rates as well as in the biomass and assemblage structure of associated fungal colonisers. Our experimental findings provide evidence that in Lake Vico effects of fish predators on invertebrate detritivores influence the fungal conditioning and breakdown of the detrital substrate. We conclude that in lacustrine littoral zones predator-driven constraints may structure lower trophic levels of detritusbased food webs and affect the decomposition of leaf detritus originated from the riparian vegetation. Keywords Detritivores · Fish predation · Fungi · Leaf detritus decomposition · Trophic cascade
G. Mancinelli · M.L. Costantini · L. Rossi (✉) Department of Genetics and Molecular Biology, Ecology Area, University of Rome “La Sapienza”, Via dei Sardi 70, pI-00185 Rome, Italy e-mail:
[email protected]
Introduction The long-lasting argument over the relative importance of top-down vs. bottom-up constraints on food web structure has been partially resolved by acknowledging the dynamic nature of food web interactions (e.g. Hunter and Price 1992; Polis et al. 1996 and literature cited). However, aquatic food webs have provided unquestionable evidence of strong, community-wide effects determined by predator-driven constraints [see reviews by Persson (1999) and Polis (1999)]. In lakes, in particular, trophic cascades (sensu Carpenter et al. 1985) have been extensively reported in pelagic systems [e.g. Brett and Goldman (1996) and literature cited; Carpenter et al. (2001)], providing a successful conceptual framework integrating biogeochemical processes with the linear food chain theory [i.e. Hairston et al. (1960); Fretwell (1987); HSS theory hereafter]. Predator-driven interactions have been unequivocally demonstrated also in benthic systems for specialised predators (e.g. molluscivores: Brönmark and Vermaat 1998 and literature cited; McCollum et al. 1998), generalist foragers (e.g. Brönmark 1994), and even omnivores (Nyström et al. 1999). To date, however, most of the investigations have focused on algal-based food webs. The understanding of top-down effects in detrital food webs is scant, despite the fact that detritus generates the major flow of energy in aquatic ecosystems (Polis and Strong 1996). Only stream studies have provided evidence of predator-driven interactions on leaf detritus breakdown (Oberndorfer et al. 1984; Reice 1991; Malmqvist 1993; Usio 2000; Konishi et al. 2001); moreover, the effects on microbial decomposers, known to play a central functional role in freshwater benthic systems [e.g. Irons (1994); Sabetta et al. (2000); Komínková et al. (2000); Bohman and Tranvik (2001) and literature cited; see also Graça (2001) for a review] are still largely unexplored. To evaluate the impact of fish predators on a detrital food web, we carried out an exclosure manipulation in the reed-dominated littoral zone of a volcanic lake (Lake Vico, central Italy). Full-exclosure cages and cageless
403
controls were used to: (1) examine the effects of predatory fish exclusion on the build-up of a macroinvertebrate assemblage colonising Phragmites australis leaf packs, and (2) assess whether treatment-induced changes in the leaf pack community influence the fungal conditioning and breakdown of the detrital substrate. Following observations that full cages excluded a diverse, highly sizedifferentiated guild of benthivorous fish, cages covered with coarse netting (10-mm mesh) were further used to evaluate the specific impact of small-sized vs. largesized fish predators.
Materials and methods Study site and fish predators Lake Vico is a volcanic basin located in central Italy (42°19′N, 12°10′E) with a surface of 12.1 km2 and a maximum depth of 48.5 m [see Sabetta et al. (2000) and literature cited for complete morphometrical and chemical data]. The common reed Phragmites australis (Cav.) Trin. ex Steud. extends along >60% of the coastline to a depth of about 1 m. In the northern area of the lake, in particular, P. australis dominates the riparian vegetation and represents the primary contributor to leaf detritus inputs to the littoral zone (L. Sabetta, personal communication). The manipulation started on 4 June 1999 in an embayment located in the northern area of the lake. The study site was characterised by a dense reed belt extending 10–15 m offshore, interspersed with a number of shallow (50–60 cm water depth), openwater pools. Visual observations of fish present at the study location started 10 days before the beginning of the experiment and continued throughout the manipulation. Two broad categories of benthivorous fish were observed: (1) small fish (n=172), mainly represented by the eurialine gobiid Knipowitschia panizzae Verga [generally >10 mm girth range, n=84), i.e. adult Lepomis gibbosus L. (100–150 mm TL, n=43), juvenile Perca fluviatilis L. (90–110 mm TL, n=21) and adult, large-sized Tinca tinca L. (usually >250 mm TL, n=18). One week before the start of the manipulation, a spatial abundance of 15–28 individuals m–2 was assessed for K. panizzae by fencing open-water pools (n=4; 3 mm mesh size) of known area and enumerating all captured gobiids. For L. gibbosus, P. fluviatilis, and T. tinca an abundance of 2–4, 0.1–0.2, and 0.1–0.3 individuals m–2, respectively, was roughly estimated in open-water pools by visual observations. Other large predatory fish common in Lake Vico include Esox lucius L. and Anguilla anguilla L., but they were never observed at the study location. Cage treatments The manipulation was carried out using cages. They consisted of tetrahedric plastic frames 1 m high with a square (0.5×0.5 m) ballasted bottom covered with a 20-mm plastic mesh. Senescing leaves of Phragmites australis were collected during winter 1998–1999 and air dried. Aliquots of leaf fragments were weighed after drying at 60°C for at least 72 h (leaf packs: 3.000±0.001 g initial dry mass) and tied together to facilitate their manipulation and retrieval. Four leaf packs were loosely fixed to the bottom of each cage. Fish predators access to leaf packs was manipulated in three cage treatments: (1) open frames without mesh cover, allowing free access to predators (open treatment in the following text), (2) frames with walls and bottom covered with square 3-mm mesh netting, precluding access to all vertebrate predators (3-mm treatment), and (3) frames with walls and bottom covered with square 10-mm mesh netting (10-mm treatment). Procedural controls were
used to examine potential artefacts due to the cages themselves. They were similar to fully covered cages but the lower 25 cm of the wall mesh was removed to allow access to fish predators. A total of 54 cages (i.e. 15 cages and three procedural controls per treatment) were randomly placed in a ~60-m2 wide, open-water pool within the reed stand; water depth was chosen in order to keep the ballasted bottom of the cages in contact with the substrate and the top edge 30–40 cm from the water surface. Sampling and laboratory procedures Leaf packs from three replicated cages per treatment were sampled on days 3, 10, 20, 30, and 45 from the start of the experiment. Packs from procedural controls were collected on day 45. Each leaf pack was carefully enclosed underwater in an UV-sterilised plastic bag, untied from the bottom of the cage, and transferred to the laboratory at 4°C in the dark. Simultaneously, field instruments were used to determine dissolved oxygen, temperature (Mettler-Toledo M0128), and pH (Delta-OHM HD8705) of the water body enclosed within three randomly chosen cages per treatment. In procedural controls, water parameters were determined at the last sampling time prior to leaf pack collection. In the laboratory, three leaf packs were haphazardly chosen from each cage and washed with deionised water on a 0.5-mm sieve to remove invertebrates. Leaf fragments were subsequently dried (60°C for at least 72 h) and weighed to the nearest milligram. The ash free dry mass (AFDM) was estimated as the mass loss after ignition (550°C, 6 h). Invertebrates were identified under a dissection microscope to the lowest possible taxonomic level and enumerated. Taxa were included into two broad trophic guilds – predators and detritivores – according to the available literature. Detritivores were further categorised into tropho-functional groups (i.e. shredder, scraper, and collector; see Appendix 1) following Cummins and Klug (1979). The dry mass (60°C, ~72 h) and biomass (mass loss after ignition at 550°C, 6 h) was determined for each individual to the nearest 0.01 mg; the biomass of chironomids, oligochaetes, and triclads was assessed cumulatively for specimens sampled on the same leaf pack. Data were expressed as number of individuals g–1 and mg g–1 leaf pack AFDM, respectively. Species richness (=number of identified taxa, S) and Shannon diversity index (H′) (=ΣpiLnpi, where pi is the proportional abundance/biomass of the ith taxon) were determined for each trophic guild. To avoid pseudoreplication, only the mean of leaf pack AFDM and invertebrate abundance/biomass estimated on the three replicate leaf packs from each cage was used; moreover, invertebrate data determined at day 3 were discarded from further analyses, since only eight specimens of the amphipod Echinogammarus veneris were collected in total. Fungal biomass was estimated on the fourth leaf pack from each cage by the extraction and quantification of ergosterol. Two series of leaf disks (9.0 mm internal diameter, n=10) were taken within 6 h from collection with a sterilised cork borer. The first disk series was used to determine sample AFDM (±0.01 mg) according to the aforementioned procedures; the ergosterol was extracted from the second disk series by reflux in methanol and consequent partitioning into pentane according to the procedure described in Gessner et al. (1991). Ergosterol was detected at 280.5 nm using a Waters HPLC system (600 pump, 996 photodiode array detector, 0.39×15 cm µBondapak C18 reversed-phase column) and was quantified against a calibration curve of pure ergosterol extracts. Fungal biomass was estimated assuming a conversion factor of 5.4945 µg ergosterol mg–1 fungal dry mass (Gessner and Chauvet 1993) and normalised per milligram detritus AFDM. At the last sampling time, fungal strains were isolated from leaf packs prior to ergosterol extraction using the dilution plating technique [see Sabetta et al. (2000) for a detailed description of the technique]. In brief, leaf packs were stirred at 100 r.p.m. in membrane-filtered (0.2 µm) lake water for 20 min. Tests carried out on leaf packs from previous samplings excluded any significant bias of the stirring procedure on subsequent ergosterol determinations. Consequently, duplicate dilutions of leaf
404 pack suspensions (0.5 ml) were inoculated on Petri dishes containing malt-extract agar (Difco), and incubated at 25°C for 10 days. Fungal strains grown during the incubation period were isolated axenically and classified to the lowest taxonomic level possible. Data analysis Values in the text are expressed as means±1 SE with n=3 replicates. For univariate analyses, the assumption of homogeneity of variances was checked using Cochran’s C-test, and transformations were used if necessary. When required, post-hoc comparisons of means were performed by Student Newman Keuls’ (SNK) tests. Statistical significance was evaluated at α=0.05; sequential Hochberg correction (Hochberg 1988) was used for multiple tests to adjust α and reduce the risk of a type-I error. Data from open cages determined at day 45 were preventively compared to procedural controls to test for cage artefacts using univariate and multivariate statistical procedures (see further in this section). In addition, two-way ANOVAs were performed on water parameters determined in 3-mm, 10-mm, and open cages during the manipulation. In the absence of any significant interpretable cage artefacts, a second analysis compared cages with different mesh size with open cages to test for effects of predators exclusion on invertebrate abundance, biomass, species richness and H′ as well as on the fungal biomass associated with leaf packs. Effects on the abundance and biomass of invertebrates were tested on a trophic guild and a single taxon level for dominant taxa (i.e. comprising ≥0.5% of total invertebrate biomass). For the latter, individuals collected in replicate cages were pooled for each treatment/sampling time and allocated to a geometric (×2) individual biomass scale; data from consecutive samplings were cumulated when the number of individuals was 0.05). In addition, no multivariate effects were detected on the species composition of the fungal assemblage at day 45 (oneway ANOSIM, R always >0.05). Negligible cage effects occurred also on oxygen concentration, temperature and pH of the water body enclosed in cages (one-way ANOVAs, max F2,6=2.32, P=0.17 for oxygen concentration). A further comparison of data determined in 3-mm, 10-mm, and open cages during the manipulation confirmed the lack of any detectable device-induced bias on water parameters (2-way ANOVAs with Treatment and Time as orthogonal, fixed factors; Factor=Treatment and Factor=Treatment×Time, max F2,30=1.06, P=0.36 and F8,30=0.42, P=0.89, respectively, for oxygen concentration). Invertebrate biomass and abundance Leaf packs were dominated by detritivore taxa, that comprised 82.5% of total invertebrate biomass (95.6% of the total number of sampled specimens). In particular, shredder crustaceans (i.e. the amphipod Echinogammarus veneris and the isopod Proasellus coxalis) accounted for 64.9–54.4% of total invertebrate biomass and abundance, respectively, followed by scraper gastropods, mainly represented by the pulmonate snails Physa acuta and Bythinia tentaculata (12.8% in biomass and 8.1% in number). The relative contribution of other scraper gastropods (Valvata piscinalis, Lymnaea auricularia, and Acroloxus lacustris) was negligible (Appendix 1). Collectors (i.e. Chironomidae and Oligogochaeta belonging to the genus Tubifex), in spite of their considerable abundance (33.8%), contributed for only 4.1% of the total invertebrate biomass. Among invertebrate predators, Odonata larval stages and the leech Erpobdella octoculata were the most abundant taxa; the triclad Dugesia tigrina, insects of the genus Ditiscus and Notonecta, and the omnivorous decapod Palaemonetes antennarius were sampled episodically. No treatment-induced variations were observed for invertebrate predators, either on a trophic guild-level or on the abundance, biomass, and mean individual biomass of dominant taxa (i.e. odonata larval stages and Erpobdella actaculata; Table 1). In contrast, the total abundance and biomass of detritivores and, correspondingly, the H′ of the guild (Fig. 1a, b) were significantly higher for both 10-mm and 3-mm cages compared to open cage
405 Table 1 Summary of two-way ANOVAs followed by a posteriori Student Newman Keuls’ (SNK) tests testing for overall treatment effects on (1) abundance and biomass of dominant invertebrate taxa, (2) species diversity (Shannon diversity index; H′), and (3)
species richness of main trophic guilds. Sequential Hochberg correction of P-values was applied to ten comparisons of invertebrate taxa. NS Not significant (P>0.05)
Abundance
Effects on invertebrate taxa Detritivores Shredders Echinogammarus veneris Proasellus coxalis Scrapers Physa acutab Bithynia tentaculatab Collectors Chironomus plumosus Predators Odonata spp. (larval stages) Erpobdella octoculata Effects on species diversity, H′′ Detritivores Predators Effects on species richness, S Detritivores Predators a Inequalities indicate b Data transformed as
Biomass
F2,24
P
SNK resultsa
F2,24
P
SNK resultsa
13.81 1.08 0.25 5.91 12.65 29.21 1.18 15.54 16.93 1.26 0.69 1.01
0.0001 NS NS 0.0492 0.0012 Open – – 3>10=Open 3=10>Open 3=10>Open – 3>10>Open 3>10>Open – – –
14.16 6.05 12.35 5.73 9.61 5.39 0.24 9.63 18.46 1.09 0.66 1.27
Open 3=10>Open 3=10>Open 3>10=Open 3=10>Open 3=10>Open – 3>10>Open 3>10>Open – – –
2.03 1.68
NS NS
– –
6.28 1.09
0.0062 NS
3=10>Open –
1.14 0.81
NS NS
– –
significant differences at P0.05) y′=y0.5
Table 2 Individual biomass (mg, mean±SE) of the shredder crustaceans Echinogammarus veneris and Proasellus coxalis collected on Phragmites australis leaf packs in frames with walls and bottom covered with square 3-mm mesh netting (3-mm cage), frames
with walls and bottom covered with square 10-mm mesh netting (10-mm cage), and open frames without mesh cover, allowing free access to predators (open cage) over the experimental period
Taxon
Time (days)
3-mm Cage
10-mm Cage
Open cage
Echinogammarus veneris
10 20 30 45 10 20 30 45
1.25±0.25 1.30±0.02 1.17±0.10 1.59±0.03 0.69±0.17 1.06±0.34 0.80±0.16 0.38±0.06
1.45±0.15 1.35±0.07 1.43±0.09 1.71±0.21 0.80±0.06 0.83±0.02 0.88±0.26 0.41±0.05
0.85±0.09 1.10±0.08 1.12±0.12 1.07±0.05 0.47±0.14 0.57±0.02 0.84±0.09 0.22±0.04
Proasellus coxalis
controls, whereas no effects were detected on the species richness (Table 1). With the only exception being Bithynia tentaculata, the biomass of the most abundant detritivores showed a significant increase in both exclusion treatments (Table 1, Fig. 2). Yet, the ultimate determinant of these changes varied considerably among taxa. The higher biomass of Echinogammarus veneris observed in 10-mm and 3-mm cages was exclusively due to a significant increase in the mean biomass of individuals (Table 2, Table 3), as no treatment effects occurred in the abundance of the gammaridean (Table 1, Fig. 2). Similarly, the positive variation of Proasellus coxalis total biomass (Table 1, Fig. 2) was due to an increase of the mean biomass of individuals collected in both exclusion treatments (Tables 2, 3), even though marginally significant changes were detected also in the abundance of the isopod (Ta-
ble 1; Time×Treatment effect: F6,24=4.08, P=0.045 after Hochberg correction). In particular, significant increases in full-exclosure cages occurred at days 30 and 45 (SNK test, P0.05). Scrapers and the dominant snail Physa acuta were characterised by significant treatment-induced increases in the abundance and biomass, but no differences were detected between 10-mm and 3-mm cages (Table 1; Fig. 2). Further, the mean individual biomass of P. acuta and of the other dominant snail Bithynia tentaculata were unaffected by fish exclusion (Table 3); negligible variations occurred also in the population size structure (crossed among-treatments χ2 test: P always >0.05). An analogous treatment-induced increase was observed in the biomass and abundance of collectors and of the dom-
406
Fig. 1 (a) Total biomass and abundance of detritivore invertebrates per unit ash-free dry mass (AFDM) of Phragmites australis leaf packs in frames with walls and bottom covered with square 3mm mesh netting (3-mm cage treatment; ●), frames with walls and bottom covered with square 10-mm mesh netting (10-mm cage treatment; ● ), and open frames without mesh cover, allowing free access to predators (open cage treatment; ▲). (b) Shannon diversity index (H′) of the detritivorous guild calculated from biomass and abundance data. Bars=±1 SE. Ind Individuals
inant collector Chironomus plumosus (Table 1, Fig. 2); significant differences in the abundance and biomass were further observed between exclusion treatments (Table 1; SNK test, P always Open – –
NS NS
– –
Predators Odonata spp. Erpobdella tentaculata
1.15 0.45
a Inequalities indicate significant differences at P0.05)
407
Fig. 3 (a) Size-frequency distributions of the shredder amphipod Echinogammarus veneris in 3-mm (●), 10-mm (● ● ), and open cages (▲) over the experimental period. Numbers in parentheses indicate sample size for each treatment. (b) Standardised χ2 residuals for comparison of large (open vs. 10-mm cage) and small (10-mm vs. 3-mm cage) predator effects on the population size structure of Echinogammarus veneris. For abbreviations, see Fig. 1
markably compared to the full exclusion treatment (min χ2=49.43, P
