Introduction

The structure, composition and trophic interactions of aquatic invertebrate communities are affected by a variety of factors, which include water chemistry (Daufresne et al. 2003, 2009), macrophyte community structure (Cheruvelil et al. 2000, 2002; Grenouillet et al. 2002) and biotic interactions such as predation (Diehl 1992; Batzer et al. 2000; Zimmer et al. 2001). Water chemistry affects epiphytic macroinvertebrates through the direct physiological effects of variables such as temperature (Sweeney and Vannote 1986; Hawkins et al. 1997) and pH (France 1990).

In most habitats, plant communities determine the physical structure of the environment and have a substantial influence on the distributions and interactions of animal species (McCoy and Bell 1991; Tews et al. 2004). The presence of macrophytes in water bodies generally enhances the complexity of the habitat and provides a wider range of niches, which sustain more diverse communities than simpler unvegetated areas (Grenouillet et al. 2002). Different species of macrophytes may differ in physical structure, morphology or architecture (e.g. number, size, orientation and arrangement) of leaves, branches and stems (Chick and McIvor 1994). These differences in plant structure contribute to habitat complexity and so affect animal assemblages (Cattaneo et al. 1998; Diehl and Kornijow 1998; Cheruvelil et al. 2000, 2002). Epiphytic macroinvertebrate abundances tend to be greater on plants with complex than those with simple morphologies (Cattaneo et al. 1998; Cheruvelil et al. 2000, 2002). This is thought to be because morphologically complex plants have more attachment and better refuge sites than morphologically simple plants (Thomaz et al. 2008).

Fish predation has strong effects on invertebrate community structure (Morin 1984a, b; Diehl 1992; Diehl and Kornijow 1998; Nyström et al. 2001; Åbjörnsson et al. 2002). Fish predators are size-selective and so large-bodied invertebrates, particularly predatory taxa, tend to be more adversely affected than small-bodied taxa (Morin 1984a, b; Wellborn et al. 1996). In fishless freshwater habitats, medium- to large-sized predatory invertebrate taxa generally dominate the invertebrate assemblage, replacing fish as the main predators (Wissinger and McGrady 1993), and strongly affect the abundances of small nonpredatory invertebrates (McPeek 1990; Nyström and Åbjörnsson 2000; Åbjörnsson et al. 2002). Invertebrate taxa that usually occur in fish-free waters are usually more active and grow much faster than those in waters with fish (Steiner et al. 2000).

Lagarosiphon ilicifolius and Vallisneria aethiopica are submerged macrophytes that are common and abundant in Lake Kariba, a large reservoir in southern Africa. Vallisneria has basal rosettes of flexible ribbon-like leaves that can occupy the full height of the water column in shallow waters. Lagarosiphon, which also can occupy the full height of the water column in shallow waters, has filiform stems that are circular in transverse section, with small sessile and mostly alternate leaves. Thus, Lagarosiphon is structurally different and probably more complex than Vallisneria. Both plants can occur in extensive monospecific underwater meadows. A study on macroinvertebrates associated with Lagarosiphon and Vallisneria in Lake Kariba, which lies between latitudes 16°28′S and 18°06′S, and longitudes 26°40′E and 29°03′E, found that the same macroinvertebrate taxa occurred on the two macrophytes, but most of the taxa were much more abundant on Lagarosiphon than on Vallisneria (Phiri 2010). This was attributed to the greater refuge from fish predation and possibly greater quantities of food provided by Lagarosiphon to macroinvertebrates than is provided by Vallisneria. But it is not known whether other factors play an equally important role as fish predation.

In this study, we set out to determine whether the aquatic insect assemblage associated with Lagarosiphon and Vallisneria differs in fishless ponds. Thus, the study explored differences in epiphytic insect assemblages that in the absence of fish predation got established on the plants, when the plants grew separately or alongside each other. Generally, structural complexity has positive influence on abundance and taxonomic richness of animals in both terrestrial and aquatic environments (Bell et al. 1991). We therefore expected that a more diverse and more abundant insect assemblage would be established on Lagarosiphon than on Vallisneria in fishless ponds. We also assessed whether odonate naiad size class distributions differed between the two plant species.

Materials and methods

Description of ponds

The study was carried out in six small artificial ponds at the University of Zimbabwe Lake Kariba Research Station. The circular ponds with a diameter of 2.2 m and a depth range of between 0.7 and 0.9 m are made of iron sheeting. The bottom substrate of the ponds comprised a mixture of sand and mud obtained from the shores of Lake Kariba. Every 2 weeks, the ponds were filled to overflowing with tap water that originated from the lake and had been treated using sand filtration and chlorination by the municipality of Kariba.

Culture and maintenance of plants in ponds

Lagarosiphon and Vallisneria were collected from Lake Kariba for culture in the ponds. Before planting, the plants were thoroughly washed using tap water to remove invertebrates that were on them while in the lake. Using random selection, two ponds (Ponds 1 and 4) were cultivated with Vallisneria, another two (Pond 2 and 3) with Lagarosiphon, and the remaining ponds (Ponds 5 and 6) with a mixture of the two plant species. An attempt was made to ensure that as much as possible the bottom substrate of each pond was covered by vegetation. The ponds cultured with both plants were divided into two equal parts: one side cultivated with Vallisneria and the other with Lagarosiphon and were regularly maintained to ensure there was no complete mixing in the growth of the two plants. The ponds were created and aquatic vegetation grown in August/September 2007. They were maintained for 10 months while aquatic insects colonised and became established in each pond.

Macroinvertebrate sampling and processing

Over a period of 6 weeks, July to August 2008, macroinvertebrate samples associated with the plants were collected every 2 weeks from each pond using a square-shaped (0.25 by 0.25 m) sweep net with a 500-μm mesh. In the monoculture ponds, sampling for insects involved sweeping over two approximately 1-m-long stretches of vegetation, while in the ponds with both plant species, an approximately 1-m stretch of each plant was sampled. Macroinvertebrate specimens were sorted from the vegetation, preserved in 10% formalin and later identified to the lowest possible taxonomic level, which mostly was the genus-level. Length measurements of odonate naiads, which were the most common and abundant group of insects on both plants, were done using a dissecting microscope with a calibrated graticule.

Water physicochemical assessment in the ponds

Water physicochemical characteristics, temperature (°C), pH, conductivity (μS cm−1) and dissolved oxygen (DO) concentration (mgl−1) of each pond were recorded on a weekly basis using the appropriate meters. Temperature was measured using a mercury thermometer, and pH was measured using a WTW 330i (Geotech Environmental Equipment, Denver, Colo. USA) pH meter, calibrated using two-stage calibration against buffers at pH 7 and 9. Conductivity was determined by using a WTW Cond 330i conductivity meter and reported at 25°C. Dissolved oxygen was measured with a WTW Oxi 330 oxygen meter that was calibrated in water vapour-saturated air using the OxiCal®-SL calibration vessel.

Data analysis

The design employed in this study consisted of two replicates per treatment. According to Carpenter (1990), inadequate replication as was used here may be worse than no replication at all since variability of community and ecosystem variates can be quite high, such that treatment effects are not detected unless they are very large. The samples that were collected every fortnight from each pond over the 6-week period were pseudoreplicates (see Hurlbert 1984). We therefore used principal components analysis (PCA) to explore for differences in water physicochemical characteristics among ponds. Macroinvertebrate abundance was expressed as number of animals per sweep, and the nonmetric multidimensional scaling (nMDS), analysis of similarities (ANOSIM) and similarities percentages (SIMPER) procedures were used to explore for similarities and differences in insect community structure between Lagarosiphon and Vallisneria. Two univariate community assemblage measures were also calculated for each sample, the number of taxa and Shannon-Wiener (H′) diversity index using the PRIMER-e version 6.1.5 (Clarke and Gorley 2006). They were used to assess for differences in insect community assemblage associated with the two macrophyte species. The Kolmogorov–Smirnov test was used to test for differences in body-size class distribution of insects associated with the two plants. Statistical analysis was done using the Simfit statistical software package version 6.0.24 (Bardsley 2009).

Results

Water quality in ponds

Higher average conductivity values were recorded in Ponds 1 and 4, in which Vallisneria was the only plant, than in the other four ponds (Table S1). Principal component analysis (PCA) showed that the first and second axes accounted for 49.7 and 26.4% of the variation in water physicochemical aspects of the ponds. The first axis was largely correlated with pH and conductivity, while the variation in the second axis was largely due to temperature differences (Fig. 1). Comparatively, the third and fourth axes were unimportant and accounted for 17.2 and 6.7% of the variation in water physicochemical conditions. Pond 1 was most distinct, with separation largely occurring along the first axes (Fig. 1).

Fig. 1
figure 1

PCA plot of water physiochemical attributes of the ponds. The numbers 1–6 represents the six ponds. During the sampling period, six measurements of water physicochemical aspects were obtained from each pond

Full size image

Insect assemblage structure associated with Lagarosiphon and Vallisneria

A total of 970 insects comprising sixteen taxa, eleven of which were predators, were collected during the study. It was not possible to statistically evaluate the abundance data (see methods) but the lowest average number of taxa per sweep, total number of taxa, mean insect abundance and diversity were obtained in the ponds in which Vallisneria was the only plant (Table 1).

Table 1 Insect composition and abundances (no. per sweep) (±SE) on Vallisneria and Lagarosiphon, from ponds in which the plants were cultured as single monospecific stands and the ponds in which both plants occurred alongside each other in two separate stands
Full size table

In all six ponds, the insect assemblage obtained using a 500-μm mesh was dominated by invertebrate predators, especially three anisopteran taxa, Hemicordulia, Diplacodes and Trithemis (Table 1). The three taxa were particularly dominant in ponds that were cultured solely with Lagarosiphon and were the two plants occurred together. Both Hemicordulia and Diplacodes were absent from ponds in which Vallisneria was the sole plant, but Trithemis dominated making up 68.7% of the insect assemblage (Table 1).

Analysis of similarities (ANOSIM) of the insect assemblages (Table 2) and the 2D MDS plot (Fig. 2) show that the insect assemblage on Vallisneria in single plant ponds was different from that on Vallisneria when it grew alongside Lagarosiphon. The insect community associated with Vallisneria from single plant ponds was also different from that associated with Lagarosiphon when it was the only plant in the pond or when cultured together with Vallisneria (Table 2). In ponds in which Vallisneria was the only plant, similarity in insect assemblage among samples was about 68%, with Trithemis overwhelmingly contributing 72% and Anisops 24% to average similarity (Table 3). Hemicordulia, Diplacodes and Trithemis were the most important taxa that contributed to similarity among samples in assemblage structure of insects associated with Vallisneria cultured with Lagarosiphon and with Lagarosiphon in ponds where it was the sole plant or when grown alongside Vallisneria (Table 3). The high dissimilarity of about 70% of the insect assemblage on Vallisneria in ponds where it was the only plant species compared with plants in the other ponds was largely due to differences in the abundances of Hemicordulia, Diplacodes and Trithermis (Table 4).

Table 2 Results of pairwise tests following 1-way ANOSIM giving R-statistics and P values for macroinvertebrates associated with Lagarosiphon and Vallisneria in fishless ponds (R-values range from 0 = pair indistinguishable to 1 = pair strongly differ)
Full size table
Fig. 2
figure 2

MDS 2D ordination of Ln (x + 1) transformed insect taxa abundances associated with Lagarosiphon and Vallisneria in fishless ponds. Vallisneria (mixed) infers that the sample was obtained from Vallisneria in ponds where it occurred together with Lagarosiphon

Full size image
Table 3 Results of SIMPER analysis on Ln (X + 1) transformed insect abundance data showing the major taxa contributing to similarity in community structure of insect associated with Vallisneria and Lagarosiphon in fishless ponds
Full size table
Table 4 Results of SIMPER analysis carried out on Ln (x + 1) transformed abundance data showing % contribution to average dissimilarity of insects associated with Lagarosiphon and Vallisneria in fishless ponds
Full size table

Size class distribution of dragonfly naiads associated with Lagarosiphon and Vallisneria

The frequency distribution of naiad size classes was significantly different when the two plants occurred in separate ponds (Kolmogorov–Smirnov test, D = 0.539, P = 0.013) (Fig. 3). The size class range of naiads from ponds where Vallisneria was the only plant was greatly narrowed compared with that from ponds where Lagarosiphon was the only plant, and significantly, greater numbers of smaller and larger size classes occurred on Lagarosiphon than on Vallisneria (Kolmogorov–Smirnov test, D = 0.539, P = 0.006) (Fig. 3). The naiad size class range associated with Vallisneria was from 2.49 to 4.99 mm and was dominated by one size class (3.00–3.49 mm), which was significantly more abundant than all the other size classes (Kolmogorov–Smirnov test, P < 0.05). Trithemis was the only odonate in ponds that contained Vallisneria as the sole plant, and comparison of its size class distribution in ponds where the two plants occurred singly showed no significant difference (Kolmogorov–Smirnov test, D = 0.385, P = 0.127). In ponds that comprised Lagarosiphon, the naiad size class range was from 0.99 to about 6.99 mm and the body size of most insects falling between 2.00 and 5.00 mm (Fig. 3).

Fig. 3
figure 3

Body size distribution of dragonfly nymphs associated with Lagarosiphon and Vallisneria when the two plant species were cultured in separate ponds. Trithermis was the only taxa that occurred on Vallisneria in ponds where it was the only plant. Error bars represent the standard error (SE) of the mean

Full size image

In ponds where the two plants were cultured together, the size class distribution of dragonfly naiads was similar and there were no significant differences in the proportion of insect size classes from the two plant species (Kolmogorov–Smirnov test, D = 0.188, P = 0.716) (Fig. S1). The body size of most of the naiads on both plants was between 1.50 and 6.00 mm, with a few naiads greater than 15 mm (Fig. S1).

The total insect abundance, taxa richness, insect diversity and abundances of Hemicordulia, Diplocodes and Trithemis in the ponds were also assessed with respect to the measured water parameters. Total abundance, as well as the abundances of Hemicordulia and Diplacodes, was negatively and significantly correlated with water conductivity, while Trithermis was also significantly but positively correlated with conductivity (Spearman Rank correlation, P < 0.05). This suggests that water quality, in this case conductivity, may have had a role in structuring insect community structure.

Discussion

The aquatic insect assemblage associated with two submerged plant species, Vallisneria and Lagarosiphon, in fishless ponds was studied. The study shows that the insect assemblage associated with Vallisneria when it was the only plant in ponds was different from that associated with Lagarosiphon or the assemblage on Vallisneria growing in close proximity to Lagarosiphon.

Insect assemblage structure on Lagarosiphon and Vallisneria

The total insect abundance, richness and diversity associated with Vallisneria in ponds in which it was the only plant were lower compared with ponds comprised of Lagarosiphon. This agrees with a number of findings (e.g. Kershner and Lodge 1990; Dionne and Folt 1991; Cheruvelil et al. 2002), in which total invertebrate abundance and diversity have been shown to differ among different macrophyte species. Most of these previous studies though were conducted in systems in which fish were present and fish predation affected the macroinvertebrate assemblage (e.g. Nyström et al. 2001; Åbjörnsson et al. 2002, 2004). In the current study, the component of fish predation was absent, and differences in macrophyte type, biomass and structural complexity, insect ovipositing behaviour, interactions among the insects as well as water physicochemical conditions were the main factors structuring insect assemblages on the two plant species.

The differences in insect communities associated with Vallisneria and Lagarosiphon grown in the same ponds (Pond 5 and 6) were not as distinct as compared with when the plants were grown in separate ponds. When grown together, insect taxa richness and diversity were similar between the two plant species. In one of the ponds (Pond 5), insect abundance on Lagarosiphon was much greater than on Vallisneria, while in Pond 6, the total abundance of insects did not differ between the two plants. The similarity in insect composition and diversity when the two plants occurred in the same pond was due to the close proximity of the plants, which enabled movement of insects between plant beds. This implies that in fishless environments, differences between Vallisneria and Lagarosiphon have minimal effects on insect taxa richness, composition and diversity when the two plants grow in close proximity in the same water body, but may affect insect abundances. Jeffries (1993), using artificial pondweeds, also found that an increase in structural complexity was associated with increase in abundance of a number of invertebrate species and total invertebrate abundance. In contrast, Rennie and Jackson (2005), using macrophyte weed bed density as a measure of complexity, found that in fishless lakes increased macrophyte complexity had no effect on invertebrate abundances, but in lakes with fish, invertebrate abundance was positively and strongly associated with macrophyte complexity. The current study suggests that apart from fish predation, there are other factors that may affect invertebrate community structure associated with macrophytes in aquatic environments.

Predator–prey interactions and insect assemblage structure

In the absence of predatory fish in the littoral and shallow zones of freshwater bodies, intermediate to large predatory insect taxa tend to dominate the invertebrate assemblage (Wellborn et al. 1996) and odonates, mostly dragonfly nymphs, tend to be the main predators (Wissinger and McGrady 1993; Åbjörnsson et al. 2004). In the present study, the insect community in all ponds was dominated by anisopteran nymphs. These predatory insects were probably feeding on each other, as well as other smaller invertebrates and zooplankton, which due to the use of a 500-μm mesh for sample collection were excluded from the study. Trithemis was the dominant taxon and the only odonate in ponds cultured with Vallisneria. In the other ponds, consisting solely of Lagarosiphon or both Lagarosiphon and Vallisneria, three dragonfly taxa, Hemicordulia, Diplacodes and Trithemis dominated. A fourth anisopteran, Tramea, was also collected from both plant species when they were cultured in the same pond. The occurrence of one taxon, Trithemis, in ponds in which Vallisneria was the only plant suggests greater interodonate interactions such as predation or cannibalism compared with ponds consisting of Lagarosiphon or both plants.

The preying of odonates on other odonates, that is, mutual or intraguild predation is a key factor in structuring odonate communities in fishless systems (Benke et al. 1982; Merril and Johnson 1984; Robinson and Wellborn 1987). In the current study, mutual predation may have resulted in exclusion by Trithemis of all the other odonate taxa from ponds that were cultured solely with Vallisneria. Thus, although this study did not directly assess the structural differences between the two plants, these may have affected insect community structure, through dampening of predatory interactions on Lagarosiphon. Another possible explanation is that Hemicordulia, Diplacodes and Tramea can live on Vallisneria, but probably do not deposit their eggs on Vallisneria, thus were absent in ponds where it was the only plant.

The suggestion that structural differences between the two macrophytes may have resulted in comparatively less mutual predation among odonates on Lagarosiphon than Vallisneria was also supported by the differences in body-size class distributions of naiads. Predatory invertebrates will generally eat all prey that are within the range of sizes they can capture and handle (Peckarsky 1984; Wellborn et al. 1996). Thus, a larger odonate individual will prey on comparatively smaller sized odonate individuals (Merril and Johnson 1984). The body-size class distribution of naiads from ponds in which Vallisneria was the only plant was characterised by a narrow size class range, largely dominated by individuals of one size class. In ponds with Lagarosiphon, the body-size range of naiads was broader with a number of size classes contributing significant proportions to the odonate community. When the two plants were cultured in the same pond, body-size distribution of naiads was similar, supporting the inference that close proximity allowed for movement and exchange of insects between the plant beds. Thus, in the habitat, which consisted entirely of Vallisneria, the high levels of predation on smaller sized naiads by larger ones reduced cooccurrence of large numbers of odonate naiad individuals characterised by large differences in body size. Interestingly, the body-size class distribution of Trithemis associated with the two plants did not differ significantly when they occurred singly in separate ponds. This suggests that apart from habitat differences, other factors such as behavioural differences and habitat use may enable coexistence of different size classes of naiads.

Water quality and insect assemblage structure

The study findings also suggest that aspects of insect assemblage structure may have been influenced by water quality in the ponds. Aquatic plants can alter physical and chemical conditions of water bodies (Petticrew and Kalff 1992; Madsen et al. 2001), and macrophyte species differing in growth form and physiological capabilities have been shown to have considerably differing effects on water physicochemistry (e.g. Jaynes and Carpenter 1986; Wigand et al. 1997). Conductivity was higher in ponds in which Vallisneria was the only plant than in the other four ponds. Interestingly, the abundances of Hemicordulia and Diplacodes were strongly and negatively correlated, whilst that of Trithemis was strongly and positively correlated with water conductivity. This raises the question of whether Vallisneria and Lagarosiphon can differentially alter water physicochemical properties in small water bodies. The dominance of Trithemis and absence of Hemicordulia and Diplacodes in the ponds in which Vallisneria was the only also plant raises the issue of whether ovipositing of eggs was influenced by differences in water conductivity. Thus, although the absolute conductivity values of the six ponds were quite low and probably had no effect on insect assemblage structure, further studies are needed to determine how aquatic insects choose ovipositing sites, especially with respect to differences in water quality and different submerged macrophyte species.

Apart from a limited number of replicates, a major weakness of the study was that we did not measure plant biomass in the ponds. Invertebrate density and therefore biomass has been shown to be positively associated with macrophyte biomass aquatic environments (Hargeby et al. 1994; Palomäki and Hellsten 1996; Tolonen et al. 2005). Thus, in the current study, differences in macroinvertebrate abundances may have been partly due to differences macrophyte biomass among the ponds. The study also used a 500-μm mesh to sample for macroinvertebrates, which excluded smaller fauna from the assemblage. The mesh size of 500 μm has been adapted by the International Standardisation Organisation (ISO) as separation between microinvertebrate and macroinvertebrate fauna, a standard that generally includes a number of planktonic copepod and cladoceran species in the macrofauna (ISO 1985).