Spatial distribution of benthic macroinvertebrates in the Erhai basin of southwestern China

Abstract

The macroinvertebrate communities and relationships with environmental variables were characterized in the Erhai basin by sampling at 44 sites. Among the 76 taxa observed, Baetis sp., Tubificidae, Simulium sp., Eukiefferiella sp., Gammarus sp. and Parakiefferiella sp. were the dominant taxa. Canonical correspondence analysis indicated that seven environmental variables, namely altitude, Ca²⁺, chemical oxygen demand, NO₃–N, total phosphorus, streambed width, and a qualitative habitat evaluation index (QHEI), were significantly related to the distribution of benthic macroinvertebrates. Two-way indicator species analysis divided total sampling sites into four groups. Weighted-averaging regression and cross-calibration produced strong models for predicting NO₃–N, altitude, and QHEI, which enabled the selection of the following benthic taxa as potentially sensitive indicators of certain levels of NO₃–N, altitude, and QHEI: Aracina, Atrichops morimotoi, Ceratopsyche sp., Guttilelopia sp., and Nemoura sp. for NO₃–N; Heptagenia sp. and Parakiefferiella sp. for altitude; and Aracina for QHEI.

Keywords:

• macroinvertebrates
• spatial distribution
• environmental relationships
• biomonitoring

Introduction

Biological indicators have widespread appeal to scientists, environmental managers, and the general public and are mainly used to assess the condition of the environment as early warning signals of ecological problems, and as barometers for trends in ecological resources (Niemi and McDonald 2004). Biological indicators may reflect the intensity of anthropogenic stress and have been used as tools to assess the impact of human-induced changes in freshwater ecosystem (Kamden-Toham and Teugels 1999). Benthic macroinvertebrates have long been used as indicators of stream ecosystem health, and they are the most commonly used biological indicators in running water (Rosenberg and Resh 1993).

Benthic macroinvertebrates are relatively easy to collect and identify, most of them are sensitive with short life histories and can respond rapidly to change, and they have a diversity of functional traits that provide a variety of responses to changing environmental conditions (Collier 2008). Numerous studies on the relationships between benthic macroinvertebrates and environment (e.g., Sandin 2003; Soldner et al. 2004; Mykrä et al. 2007) have aimed to find indicator species, community structure traits or multimetric indices to assess water quality and river ecosystem health (Li et al. 2010).

In China, there have been few studies on the character of benthic macroinvertebrate distribution or the selection of indicator taxa at the basin scale. Therefore, the objectives of our study were to characterize benthic macroinvertebrate community composition in the Erhai basin and identify taxa that could potentially be used as indicators of water physicochemical conditions in the basin.

Methods and materials

The Erhai Basin is located at 25°25′–26°10′N and 99°32′–100°27′E in Yunnan Province, China, and includes the 250-km² Lake Erhai (Figure 1). The region has a typical warm monsoon climate, with wet and dry seasons each year.

Figure 1. Location of sampling sites in the Erhai basin and identification of the four final groups denoted by TWINSPAN analysis.

During February and March of 2009, we sampled 44 sites on streams within the basin. At each site, when possible, we collected five randomly located benthic samples with a 30 × 30 cm² Surber sampler (420 -µm mesh). Specimens were preserved in 10% formalin and were identified according to Morse et al. (1994) and Merritt and Cummins (1996).

A Hydrolab Minisonde (Hach Environmental, Loveland, CO) was used to measure surface water pH, conductivity, salinity, total dissolved solid (TDS), dissolve oxygen (DO), and water temperature. Surface water was also collected for analysis of the following: total nitrogen (TN), ammonium (NH₄–N), nitrate (NO₃–N), total phosphorus (TP), orthophosphate (PO₄–P), calcium (Ca²⁺), magnesium (Mg²⁺), chlorine (Cl⁻), hardness, alkalinity, chemical oxygen demand (COD), and total organic carbon (TOC). All the analyses were carried out according to the standard methods (Cai 2007). Altitude and streambed width were also recorded for each site.

The steam habitat and geomorphology were characterized by a qualitative habitat evaluation index (QHEI; Rankin 1989). The QHEI assessed the substrate, riparian zone, and bank erosion and computed a summary score. QHEI scores >60 suggest high habitat quality, whereas scores <45 suggest poor habitat quality (Moerke and Lamberti 2004).

The relationships between environmental variables and benthic macroinvertebrate assemblages were explored using canonical correspondence analysis (CCA). All the biotic data were converted into relative abundance and log-transformed before analysis, and rare species were down-weighted in CCA. Forward selection and Monte Carlo permutations were used to select a minimum set of variables that had significant and independent relationships with the benthic macroinvertebrate distribution.

Two-way indicator species analysis (TWINSPAN) was used to classify the samples into groups based on macroinvertebrate data and identify the indicator taxa of the clusters (Dufrêne and Legendre 1997).

The Shannon–Wiener diversity (Shannon and Weaver 1949), the Pielou (1966) evenness, and taxa richness were calculated. The Kruskal–Wallis test was used to compare their differences among different groups as well as differences in environmental variables.

We used weighted-averaging regression analysis to establish indicators–inference models (Birks et al. 1990) using C2 software (Juggins 2007). Taxa optima and tolerances (i.e., the mean and standard deviation of the environmental variables over all sites where a taxon occurred) were calculated. The jackknife cross-validation procedure was used to access the predicative capability of the resulting models. The coefficient of determination (R ²) between species-inferred and observed environmental variable concentrations and the root-mean-squared error (RMSE) of prediction also were measured to access the models. Based on Kilroy et al. (2006) and Wu et al. (2009), our criteria were (1) occurrence in at least 20 of the 44 sites and (2) tolerance to the variable of interest (i.e., <0.75 × the mean tolerance for all the taxa). Untransformed taxa and environmental data were used for the weighted-averaging regressions.

In our study, CCA was carried out by CANOCO (version 4.5); TWINSPAN was used by PC-ORD (version 4.0); Kruskal–Wallis text was conducted by SPSS 16.0 and the weighted-averaging regression by C2 (version 1.5).

Results and discussion

Macroinvertebrate sampling yielded 76 taxa in 68 genera. We defined dominant taxa as those with relative abundances >5% (Bunn et al. 1986). These included Baetis sp., Tubificidae, Simulium sp., Eukiefferiella sp., Gammarus sp., and Parakiefferiella sp., and their respective relative abundances were 17.34%, 13.84%, 12.17%, 9.74%, 9.44%, 7.09%, and 6.44%. Tubificidae, Eukiefferiella sp., and Parakiefferiella sp. are tolerant taxa; their presence suggests that environmental stress in this basin is considerable. The average macroinvertebrate density was 1991 ind m⁻²; the maximum and minimum densities were, respectively, 9355 and 303 ind m⁻².

Twenty-one environmental variables, including COD, alkalinity, Ca²⁺, Mg²⁺, hardness, Cl⁻, TN, NO₃–N, NH₄–N, TP, PO₄–P, TOC, conductivity, salinity, TDS, temperature, DO, pH, streambed width, altitude, and QHEI, were employed in the CCA. The result showed that the distribution of benthic macroinvertebrate was mainly related to NO₃–N, QHEI, altitude, streambed width, COD, TP, and Ca²⁺ (Monte Carlo test p < 0.05; Figure 2).

Figure 2. CCA biplot showing significant environmental factors.

TWINSPAN separated the 44 sites into four groups with similar characteristics (Figure 1). Group A was primarily the Cangshan Mountain sites, and Group B consisted of sites located in the Miju and Bolou Rivers. The indicator taxon of Group B was Ceratopsyche sp. Group C comprised moderately disturbed sites, and Group D was seriously disturbed sites. The indicator taxon of Group C was Ceratopsyche sp., whereas the Group D indicator taxon was Tubificidae.

The Shannon–Wiener diversity, Pielou evenness and taxa richness showed significantly differences among the four groups (Kruskal–Wallis test, p < 0.05) (Table 1). Similarly, five environmental variables selected by CCA (Ca²⁺, NO₃–N, TP, altitude, and QHEI) showed significant differences among the four groups (Kruskal–Wallis test, p < 0.05).

Table 1. Community trait and environmental variables (means) of the Erhai basin streams and results of the K–W test among the four groups.

Variable	Group A	Group B	Group C	Group D	K–W test	p-value
Community trait
Shannon–Wiener	2.27	2.24	1.81	0.81	21.06	0.0001
Evenness	0.76	0.69	0.62	0.37	17.12	0.0010
Richness	20	26	19	8	22.57	<0.0001
Environmental
COD (mg L^−1)	1.57	1.96	3.09	2.70	7.28	0.0635
Ca (mg L^−1)	10.38	23.38	22.03	18.26	10.19	0.0170
NO3–N (mg L^−1)	0.11	0.21	0.78	0.45	22.43	<0.0001
TP (mg L^−1)	0.04	0.05	0.15	0.32	10.53	0.0146
Streambed width (m)	4.6	2.6	4.2	10.2	6.38	0.0947
Altitude (m)	2,385	2,199	2,043	1,891	22.02	<0.0001
QHEI	52	42	28	34	25.79	<0.0001
Note: K–W, Kruskal–Wallis.

Because the above analyses confirmed strong environmental relationships with the macroinvertebrate community, we used Ca²⁺, NO₃–N, TP, altitude, and QHEI in weighted-averaging regression analysis. The regression and calibration produced a strong model for predicting NO₃–N, altitude, and QHEI. NO₃–N, altitude, and QHEI weight-averaging taxa optima were calculated using the full dataset (n = 44), and the results are presented for the taxa with effective numbers of occurrences (i.e. >20; Table 2) Weight-averaging values of NO₃–N, altitude and QHEI optima were 0.14–0.83 mgL⁻¹, 1961–2457 m, and 29–50, respectively.

Table 2. Optimum and tolerance values generated by the weighted-averaging regression analyses of benthic macroinvertebrate for NO₃–N, altitude, and QHEI in the Erhai basin.

Taxon	Number of occurrences	NO3–N (mg L^−1)		Altitude (m)		QHEI
		Optimum	Tolerance	Optimum	Tolerance	Optimum	Tolerance
Aracina	23	0.15^a	0.15	2323	218	49.8^a	5.3
Antocha bifida	32	0.22	0.20	2242	297	45.6	10.4
A. morimotoi	28	0.18^a	0.12	2258	235	47.2	8.6
Baetis sp.	41	0.36	0.34	2164	194	39.7	12.0
Ceratopsyche sp.	25	0.14^a	0.09	2310	223	48.3	7.6
Diptera	29	0.64	0.55	2047	223	34.5	12.7
Eukiefferiella sp.	40	0.34	0.34	2154	234	39.1	11.1
Gammarus sp.	33	0.51	0.46	2098	156	36.0	10.2
Guttilelopia sp.	36	0.15^a	0.17	2300	295	48.9	11.1
Heptagenia sp.	29	0.31	0.45	2158^a	107	34.5	9.1
Iron sp.	24	0.19	0.29	2238	213	42.8	11.7
Macronychus glabratus	39	0.31	0.39	2212	222	42.5	12.5
Nemoura sp.	26	0.14^a	0.08	2457	273	50.2	7.7
Parakiefferiella sp.	23	0.79	0.59	2061^a	97	29.2	7.8
Simulium sp.	42	0.57	0.51	2104	255	33.8	14.5
Stenelmis sp.	26	0.24	0.20	2261	199	47.5	7.0
Tanytarsus sp.	33	0.83	0.53	2050	189	31.9	11.8
Tubificidae	35	0.65	0.52	1961	176	30.2	11.9
Notes: Only taxa that occurred in at least 20 of 44 sites are listed.
^aIndicates a potential indicator taxon.

The best results used simple weight-averaging regression (tolerance down-weighting) with inverse deshrinking (NO₃–N: R ² = 0.72, apparent RMSE = 0.23; altitude: R ² = 0.59, apparent RMSE = 152; QHEI: R ² = 0.71, and apparent RMSE = 7). Jackknife-derived predicted NO₃–N, altitude, and QHEI values matched the measured values well (Figure 3), and the residuals plotted against predicted values (Racca and Prairie 2004) indicated no bias in the models. The models for Ca²⁺ and TP performed poorly (not shown).

Figure 3. Observed NO₃–N, altitude, and QHEI values at the 44 sites plotted against predicted values calculated from a weighted-averaging regression model.

Macroinvertebrate taxa that met the criteria specified for potential use as indicator taxa were: Aracina (0.15 mgL⁻¹), Atrichops morimotoi (0.18 mgL⁻¹), Ceratopsyche sp. (0.14 mgL⁻¹), Guttilelopia sp. (0.15 mgL⁻¹), and Nemoura sp. (0.14 mgL⁻¹) for NO₃–N; Heptagenia sp. (2158 m) and Parakiefferiella sp. (2,061 m) for altitude; and Aracina for QHEI (49.8) (Table 2).

Ordination techniques have been used successfully to interpret macroinvertebrate community structure. In this study, macroinvertebrate distributions were mainly related to NO₃–N, QHEI, altitude, streambed width, COD, TP, and Ca²⁺. In numerous studies (e.g., Miserendino 2001; Heino et al. 2003; Soldner et al. 2004), altitude and water chemistry always related to the macroinvertebrate assemblages. Altitude directly determines some environmental factors such as temperature and illumination in a region. Some studies showed that macroinvertebrate richness decreased with increased altitude (e.g., Suren 1994). Main nutrients (e.g., N and P) can affect benthic algal composition (Leland and Porter 2000), and further affect the macroinvertebrate assemblages by bottom-up control.

Our results suggest that benthic macroinvertebrate distributions can be related to NO₃–N, altitude, and QHEI. The weighted-averaging models showed significant correlation and low apparent RMSE, which indicated that the models were reliable. Numerous studies have reported the altitude (e.g., Suren 1994), NO₃–N (e.g., Richards et al. 1992), and QHEI (e.g., Beyene et al. 2009) were related to the benthic macroinvertebrate community. Environmental stress by human activity could be assessed by NO₃–N and QHEI in this basin; according to our models, this stress could also be evaluated by the presence of Aracina, A. morimotoi, Ceratopsyche sp., Guttilelopia sp., and Nemoura sp.

Acknowledgments

We thank Wen Xiao, Yanpeng Li, and Jinfa Li of Dali University for their assistance with the field sampling. This study was supported financially by the Major Science and Technology Program for Water Pollution Control and Treatment (2008ZX07526-002-07) and the National Natural Science Foundation of China (30330140).