Effects of multiple stressors on hyporheic invertebrates in a lotic system

Effects of multiple stressors on hyporheic invertebrates in a lotic system

Ecological Indicators 3 (2003) 65–79 Effects of multiple stressors on hyporheic invertebrates in a lotic system S.M. Nelson∗ , R.A. Roline Ecological...

219KB Sizes 14 Downloads 41 Views

Ecological Indicators 3 (2003) 65–79

Effects of multiple stressors on hyporheic invertebrates in a lotic system S.M. Nelson∗ , R.A. Roline Ecological Research and Investigations Group, Technical Services Center, Bureau of Reclamation, Denver, CO 80225, USA

Abstract The impacts of co-occurring stressors (river regulation by a dam and mine drainage) on hyporheic macroinvertebrate assemblages and environmental variables were investigated in a mountainous area of central Colorado at seven sampling sites in Lake Fork during two different seasons. Specific aquatic macroinvertebrate assemblages were associated with river regulation and trace element impacts. Paraleuctra was primarily associated with mining impacted sites, while Polycelis, Hydra, and Simulium were most abundant at sites nearest the dam. There were sometimes large differences in toxicity between hyporheic and surface water samples. Occasionally toxicity was present in the surface water, but absent in hyporheic water from the same site, while on one date, toxicity was detected in the hyporheic but undetected in surface water. The distinct taxa found in the hyporheic, and differences between hyporheic and surface water quality suggest that understanding the hyporheic zone is important in the study of impacted aquatic systems. © 2003 Elsevier Science Ltd. All rights reserved. Keywords: Hydra; Hyporheic; Macroinvertebrate; Manganese; Metals; Mining; Multiple stressors; Paraleuctra; Polycelis; River regulation; Water quality

1. Introduction Anthropogenic activities cause a diversity of stresses to aquatic systems that may affect biota at different spatial and temporal scales. Acute and chronic stresses from the same agent may impact biotic communities differentially, as may introduction of contaminants to geographically close, but spatially different areas such as subsurface versus surface water. Despite a growing awareness that interactions between stressors need to be understood (Folt et al.,

∗ Corresponding author. Tel.: +1-303-445-2225; fax: +1-303-445-6328. E-mail address: [email protected] (S.M. Nelson).

1999; Jooste, 2000), ecologists have generally focused on single perturbations. Habitats are often impacted by the presence of several stressors. Stressors frequently co-occur in lotic systems such as streams and rivers. Stream communities may be disparately stressed as water travels from mining influenced headwaters towards lowland areas affected by municipal and industrial effluents. Along with chemical impacts, hydrological and/or physical stresses caused by impounding lotic systems, may occur. Studies that focus on single stressors can be inadvertently confounded by other stressors that have occurred some distance upstream of the studied impact. A variety of potential effects may be elicited from multiple stressors and prevent normal recovery patterns from taking place.

1470-160X/03/$ – see front matter © 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S1470-160X(03)00012-8

66

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

Lake Fork, a stream in the mountains of central Colorado, presents an opportunity for study of invertebrate community responses to multiple stressors. Over a relatively short distance, this stream is impacted by both river regulation and mine drainage inputs. Because plans were developed for treatment of mine drainage entering this stream, a test for measurement of biotic response to changes in metal impacts was needed. However, the study of metal impacts in this stream is confounded by the stress of water regulation from a dam just upstream of the mine drainage inflow. Two conflicting recovery gradients exist in this situation. Recovery from river regulation typically results in increased invertebrate diversity downstream away from the dam, while increased metal concentrations downstream would cause decreased diversity. The input of metals effects onto the gradient of recovery from regulation may change the path trajectory; therefore, identification of taxa useful for bioassessment of metals impacts in the face of regulation impacts was also a study objective. Our goal was to develop biomonitoring tools specific to the type of impact so that monitoring of invertebrate recovery from mining impacts could be separated from regulation impacts. We collected invertebrate samples from the shallow, in-stream hyporheic for studying these impacts. The hyporheic zone is generally considered the subsurface portion of streams or rivers where ground water and surface water mix (Vallett et al., 1993). This zone is typically considered to extend to a few meters in depth within the interstitial habitat, but can be up to 2 km from the river channel (Stanford and Ward, 1988). Biological productivity, in some cases, may be very high in the hyporheic zone. The hyporheic community may be especially sensitive to stressors present in this study. Changes in substrate are often associated with river regulation (Taylor, 1978) and Richards and Bacon (1994) found hyporheic communities to be particularly sensitive to changes in substrate. In addition, hyporheic macroinvertebrate communities are responsive to metals effects from mine drainage (Nelson and Roline, 1999). The shallow hyporheic zone often plays a critical role in regulating metal fluxes between surface and groundwater environments (von Gunten et al., 1991; Gibert et al., 1995). This present study provides information on the relatively new field of hyporheic pollution ecology.

2. Methods and materials 2.1. Sampling methods Seven sites along Lake Fork, a tributary of the upper Arkansas River in central Colorado were examined in October 1999 and May 2000. The stream originates above Turquoise Lake (elevation 3000 m) and flows into the Arkansas River south of Leadville, Colorado. Turquoise Lake is impounded by the bottom-release Sugar Loaf Dam. Flows above the reservoir during run-off in May and June (1.9 m3 /s) are higher than those below the reservoir (0.5 m3 /s) (Dash and Ortiz, 1996). Much of the flow from Turquoise Lake is diverted through the Mount Elbert conduit for hydropower production, thus bypassing Lake Fork. Sites (Fig. 1) included a reference station above the reservoir (LF-0), two stations below the reservoir (LF-1 and LF-2) and above a metals point source (Dinero Tunnel), two stations below the Dinero Tunnel point source (LF-3 and LF-4), and two stations further downstream and away from the reservoir and Dinero Tunnel metals point source (LF-5 and LF-6). We used hyporheic pot samplers to collect macroinvertebrates. Details of hyporheic pot construction and installation are described in Nelson and Roline (1999). Briefly, samplers were built of 30 cm long perforated polyvinyl chloride (PVC) pipe with an inside diameter of 10 cm. These samplers were filled with local sediment and installed vertically into the substrate. Three hyporheic pots samplers were initially placed at each station in July 1999. During sample collection (October 1999 and May 2000), contents of hyporheic pots were placed into separate containers and macroinvertebrates preserved in formalin. In the laboratory, samples were washed in a 600 ␮m mesh sieve to remove formalin, organisms were picked from the substrate under 10× magnification, and invertebrates identified to lowest practical taxon under a binocular dissecting scope. 2.2. Trace element samples Hyporheic pore water samples were collected in situ from the bottom of hyporheic samplers through a fused glass airstone (used to prevent clogging of tubing), connected by plastic tubing which led to the

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

67

Fig. 1. General location of sampling sites within the study area on Lake Fork.

surface (e.g. Nelson and Roline, 1999). A hand operated vacuum pump was used to withdraw samples. Surface water samples were also collected. Cadmium (Cd), copper (Cu), iron (Fe), manganese (Mn), and zinc (Zn) were identified with inductively coupled plasma (ICP)/emission spectroscopy (EPA, 1983; APHA, 1989). Water samples used for metals analyses were filtered (0.45 ␮m filter) and preserved with 1 ml of nitric acid. Rate of metal deposition was estimated as the accumulation of metals collected from PVC rings (ca. 2656 mm2 surface area) placed in the bottom and top of hyporheic pots for a known period (July–October and October–May). Deposition rates were assumed to be constant and were calculated as mg per unit area

per day. Metal deposits were analyzed by ICP after acid/microwave digestion to remove metals from PVC rings. 2.3. Physicochemical variables We also measured pH, conductivity, alkalinity, and hardness during the study. Surface water temperatures measured in July–May were used to calculate a temperature coefficient of variation (CV) for each site as an indicator of reservoir effects. A smaller CV would indicate that the site was more impacted by the hypolimnetic withdrawal from the reservoir while a large CV would suggest that the stream had a more natural temperature range.

68

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

2.4. Toxicity A rapid (3 h assay) toxicity test kit (MetPLATETM ) was used for detection of toxicity (enzymatic inhibition) in surface and hyporheic water samples. This test was performed according to methodology in Bitton et al. (1994). This test uses freeze–dried Escherichia coli bacteria as the test organism and is generally specific to metal toxicity (Bitton et al., 1994). Toxicity values from laboratory studies (Nelson and Roline, 1998) generally correspond to those obtained from 48 h testing with the standard test organism, Ceriodaphnia dubia. Dilutions and controls were made in reconstituted very-soft water (Peltier and Weber, 1985). Test results are measured as degree of inhibition, considering the control to represent 0% inhibition. Data were plotted as percent inhibition versus log of sample concentration and the EC50 (50% inhibition) determined from linear regression analysis of the data. For purposes of this study, samples in which there was no observed toxicity were set at an EC50 of 100% rather than >100%. The smaller the amount of sample at which an EC50 is observed, the greater the toxicity. 2.5. Habitat assessment Information on particle size of substrate material was obtained from size gradations of dried mineral samples from pot samplers. Current velocity was measured at a point 10 cm above each hyporheic pot. Piezometers, made of PVC pipe (15 mm i.d.), were installed to a substrate depth of 30 cm at each site. Hydraulic head, the difference between water height in hyporheic zone piezometers and ambient stream water surface, was measured manually with a length of chalked wire. Positive hydraulic head readings suggest hyporheic discharge or upwelling, with hyporheic water entering the stream channel. Negative values indicate downwelling or recharge from the stream channel into the hyporheic zone. 2.6. Data analysis Multivariate analysis (CANOCO 4.0 and TWINSPAN) and biotic indices of abundance and taxa richness measures were used to compare invertebrate assemblages. Ordination techniques were used to

examine patterns in the macroinvertebrate data and to identify physical and chemical variables that were most closely associated with invertebrate distributions. Initial analyses of the macroinvertebrate data sets used detrended correspondence analysis (DCA), and revealed that the data sets had relatively short gradient lengths (less than 3), suggesting that unimodal models were perhaps inappropriate for analysis. Therefore, redundancy analysis (RDA) was used for direct gradient analyses. Infrequent taxa (taxa contributing <0.5% of total numbers counted) were deleted and faunal data transformed (square root) before analysis. Environmental variables were normalized if needed with square root or Arcsin transformations for percent data and ln(x + 1) for numeric data. Forward selection of environmental variables and Monte Carlo permutations were used to determine whether variables exerted a significant effect on invertebrate distributions. If physical and chemical variables were strongly correlated (r ≥ 0.6), only a single variable was selected for use in the RDA analysis to avoid problems with multicollinearity. A single variable, therefore, might also represent several other variables in the analysis. In the ordination diagram taxa and sites are represented by points and the environmental variables by arrows. The arrows roughly orient in the direction of maximum variation of the given variable. Canonical eigenvalues were used to rank the importance of environmental variables in determining species composition (ter Braak and Verdonschot, 1995). Two-way indicator species analysis (TWINSPAN), a divisive classification method based on correspondence analysis, was used to define species assemblages and the fidelity. Cut-levels used were based on numbers derived from 75, 80, 85, 90, and 95 percentiles for abundance. 3. Results 3.1. Environmental variables Some sampled environmental variables were assumed to be either indicators of regulation or metals contamination (e.g. Tables 1 and 2, Figs. 2 and 3). Regulation indicators included particles smaller than 2 mm diameter (% sand), amount of organic material, and temperature CV, all variables that may be influenced

Variable

Sampling sites LF-0

Metals impacts pH (SU) Conductivity (␮S/cm) Alkalinity (mg/l) Hardness (mg/l) Toxicity (EC50) Regulation impacts Temperature CV % sand Organic matter (g per pot) Other variables DO (mg/l) Velocity (m/s) Hydraulic head (mm)

LF-1

LF-2

Surface

Hyporheic

Surface

Hyporheic

Surface

7.4 26 8 10.2 100

7.0 (0.0) 65 (26) 15 (4) 15.1 (3.4) 100 (0)

6.9 23 8 9.0 100

6.9 (0.0) 24 (0) 7 (1) 9.1 (0.0) 100 (0)

7.5 24 6 9.0 100

70.35 – – 9.6 0.2 (0.0) –

– 5.2 (0.6) 6.4 (0.8) 6.4 (1.0) – 19 (2)

5.27 – – 8.6 0.2 (0.1) –

– 0.4 (0.0) 0.6 (0.1) 7.4 (0.4) – 12 (4)

16.90 – – 8.6 0.3 (0.0) –

LF-3 Hyporheic 7.3 (0.0) 24 (0) 8 (2) 9.0 (0.0) 100 (0) – 0.7 (0.0) 0.6 (0.1) 7.9 (0.1) – −59 (32)

LF-4

LF-5

Hyporheic

Surface

7.3 66 6 21.8 47.2

6.9 (0.0) 66 (0.7) 5 (1) 22.2 (0.1) 48.6 (2.2)

7.3 65 7 21.9 43.9

6.9 (0.0) 64 (0.3) 5 (1) 21.9 (0.0) 67.3 (6.8)

7.2 89 15 25.8 68.6

7.0 (0.0) 90 (6.2) 12 (0) 25.5 (0.1) 88.9 (6.2)

15.97 – –

– 2.4 (1.1) 3.1 (0.8)

30.58 – –

– 1.0 (0.2) 1.8 (0.7)

12.35 – –

– 8.8 (3.9) 4.2 (1.1)

9.3 0.1 (0.0) –

8.2 (0.2) – 9 (5)

9.9 0.4 (0.1) –

Values are means with S.E. in parentheses. Additional metals impacts data are presented in Figs. 3 and 4. Cd and Cu were below detection limits.

Hyporheic

8.1 (0.2) – −109 (76)

Surface

LF-6

Surface

9.2 0.2 (0.1) –

Hyporheic

8.0 (0.2) – −5 (60)

Surface 7.0 59 13 16.8 100 17.51 – – 9.0 0.4 (0.1) –

Hyporheic 6.9 (0.0) 61 (0.0) 13 (1) 16.3 (0.0) 100 (0) – 5.9 (1.3) 2.5 (0.3) 7.5 (0.0) – −71 (86)

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

Table 1 Environmental variables associated with sampling sites in October 1999

69

70

Variable

Sampling sites LF-0

LF-1

LF-2

Surface

Hyporheic

Surface

Hyporheic

Surface

7.1 16 6 7 100

7.0 (0.1) 33 (17) 15 (9) 14 (6) 100 (0.0)

7.1 23 10 10 100

6.9 (0.1) 24 (0) 6 (1) 9 (0) 100 (0.0)

7.5 24 8 10 100

Regulation impacts Temperature CV % sand Organic matter (g per pot)

93.7 – –

– 10.6 (2.7) 16.6 (6.4)

Other variables DO (mg/l) Velocity (m/s) Hydraulic head (mm)

11.2 0.4 (0.3) –

9.0 (1.4) – 52 (2)

Metalsimpacts pH (SU) Conductivity (␮S/cm) Alkalinity (mg/l) Hardness (mg/l) Toxicity (EC50)

20.93 – – 9.3 0.5 (0.2) –

– 0.9 (0.3) 1.3 (0.2) 8.1 (0.5) – 48 (47)

22.39 – – 8.8 0.6 (0.2) –

LF-3 Hyporheic 7.0 (0.0) 24 (1) 8 (1) 9 (0) 100 (0.0) – 0.4 (0.1) 0.8 (0.1) 8.2 (0.2) – −50 (46)

LF-4

LF-5

Surface

Hyporheic

Surface

7.2 39 3 15 44.7

6.8 (0.1) 45 (4) 10 (2) 14 (1) 100 (0.0)

7.2 41 3 15 59.9

6.9 (0.1) 41 (1) 6 (0) 14 (0) 68.7 (3.5)

28.46 – –

– 1.0 (0.1) 4.7 (1.1)

27.95 – – 8.9 0.4 (0.2) –

– 5.4 (2.9) 4.1 (1.2) 7.4 (0.1) – 83 (83)

8.7 0.3 (0.1) –

Hyporheic

7.4 (0.4) – −69 (52)

Surface 5.4 79 1 20 13.4 9.61 – – 8.2 0.4 (0.1) –

LF-6 Hyporheic 5.7 (0.3) 86 (4) 4 (1) 23 (2) 38.5 (2.0) – 2.3 (0.3) 9.9 (2.8 6.4 (0.3) – −28 (27)

Surface

Hyporheic

6.6 70 12 19 100

6.7 (0.1) 78 (8) 11 (4) 20 (1) 87.0 (13.0)

12.46 – – 10.1 0.6 (0.1) –

– 4.8 (1.3) 8.4 (0.5) 6.6 (0.1) – 22 (2)

Values are means with S.E. in parentheses. Additional metals impacts data are presented in Figs. 3 and 4. Cd and Cu were below detection limits except for Cd at LF-5 [6.93 ␮g/l in surface water and 3.5 (1.5) ␮g/l in the hyporheic] and Cu at LF-5 [72.8 ␮g/l in surface water and 6.3 (2.2) ␮g/l in the hyporheic] and LF-6 [16.0 ␮g/l surface water and 6.6 (0.3) ␮g/l hyporheic].

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

Table 2 Environmental variables associated with sampling sites in May 2000

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

71

Fig. 2. Dissolved metals found in surface and hyporheic samples at Lake Fork sites. The first pair of bars represent mean values in surface and hyporheic samples in October 1999 and the second pair are from May 2000.

by the reservoir. Variables related to metal contamination included metal concentrations in water, metals deposited upon PVC substrate, pH, conductivity, alkalinity, hardness, and toxicity. Other variables assumed to be important in community distribution were dissolved oxygen (DO), current velocity, and hydraulic head. Metal concentrations were consistently low at LF-1 and LF-2 (Figs. 2 and 3). Dissolved Zn concentrations were also relatively low at sites LF-0, LF-5 (in October), and LF-6 (in October). Dissolved Cd and Cu

were below detection levels in October. Patterns of deposited metals between sites were variable. Iron deposition patterns for the sites (LF-3 and LF-4) immediately below the Dinero Tunnel inflow were similar to those found at the largely metals-unimpacted sites below the reservoir (LF-1 and LF-2) (Fig. 3). Samples from sites LF-0, LF-5, and LF-6 tended to have the highest concentrations of Fe deposits. Highest Mn and Zn deposits were generally found at the furthest downstream sites (LF-3 to LF-6) in May. Surface variables and hyporheic variables often differed. Conductivity

72

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

3.2. Macroinvertebrate fauna

Fig. 3. Deposited metals found in surface and hyporheic samples at Lake Fork sites. The first pair of bars represent mean values in surface and hyporheic samples in October 1999 and the second pair are from May 2000.

was often higher and pH lower in the hyporheic than found at the surface (Tables 1 and 2). There appeared to be site differences in metal concentrations between surface and hyporheic zones. For example, hyporheic concentrations of dissolved Fe at LF-0 were always higher than surface concentrations, while this pattern was reversed at LF-5 (Fig. 2). Manganese concentrations increased from October to May. Hyporheic concentrations of dissolved Mn at LF-3, LF-5, and LF-6 in May were higher than in October and were higher than those found at the surface (Fig. 2). It appeared that the rate of hyporheic Mn deposition increased at LF-3 and LF-4 during this time, but decreased at LF-5 and LF-6 (Fig. 3).

A total of 77 taxa were identified during the study. Environmental variables initially used in the October RDA model included, toxicity, conductivity, hydraulic head, DO, % sand, pH, Fe deposition, and temperature CV. Of these, conductivity, hydraulic head, DO and Fe deposition were dropped from the model because they did not show significance (P > 0.05). The % sand and toxicity were the most important variables on the first two axes (Table 3), while pH and hydraulic head were important on axes three and four which explained very little of the variation (Table 3). RDA analysis of October data (first and succeeding axes were significant at P < 0.002, eigenvalues of 0.443 and 0.108 for the first two axes) suggested three groups (Fig. 4a). Group 1 contained samples from LF-0, LF-5, and LF-6. Group 2 contained samples from LF-3 and LF-4, while Group 3 contained samples from LF-1 and LF-2. Toxicity decreased from Group 2 to Groups 1 and 3 (Fig. 4 and Table 1). Group 1 differed from Group 3 in having increased variation in water temperature, increased % sand, and lower pH. Group 1 contained samples relatively unimpacted by either metals toxicity or hydrology, while Group 2 was impacted by metals toxicity and Group 3 largely impacted by reservoir effects. TWINSPAN results formed similar groups, except that LF-6 was split from LF-5 and LF-0 (Fig. 4a). Toxicity and % sand were also important parameters in the May RDA (Fig. 5). Temperature variation and hydraulic head were not significant parameters in May, however, both dissolved and deposited Mn were significant (P < 0.05) components of the RDA. RDA analysis of May data (first and succeeding axes were significant at P = 0.005, eigenvalues of 0.622 and 0.051 for the first two axes) suggested three groups (Fig. 5a). Group 1 contained samples from LF-1 and LF-2 and was characterized by low toxicity and low % sand (Fig. 5, Table 2). Hydra, Polycelis, and Simulium were associated with regulated impacted sites and Paraleuctra associated with the RDA mine impacted groups in both October and May. Toxicity and substrate Mn were generally greater in samples from Group 2. LF-5, unlike in October, tended towards being grouped with LF-3 and LF-4 in May. There appeared to be more ordination scatter in May for this group than in October. Group 3 contained samples from LF-0 and LF-6 and was characterized

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

73

Table 3 Weighted correlation matrix showing relationship between species axes and environmental variables Variable

Axis 1 October

% sand Toxicity Temperature CV pH Substrate Mn Dissolved Mn Axes eigenvalues % variance of species–environment relationship

−0.7291 0.1441 −0.3672 0.4557 – – 0.443 73.4

2 May 0.6114 −0.4835 – – 0.2378 0.5484 0.622 88.5

October −0.3193 −0.8368 −0.0409 −0.2340 – – 0.108 17.9

3

4

Environmental eigenvalues

May

October

May

October

May

October

May

−0.0400 −0.2725 – – 0.7909 0.0212 0.051 7.2

−0.0351 0.0733 0.6414 0.3853 – – 0.031 5.1

0.6052 0.3995 – – 0.1506 −0.4247 0.021 3

0.3452 −0.0739 −0.0702 0.4689 – – 0.022 3.6

0.0488 0.3769 – – 0.0883 0.3752 0.009 1.3

0.31 0.14 0.07 0.08 – –

0.28 0.23 – – 0.06 0.13

by low toxicity and increased % sand. TWINSPAN grouped samples according to RDA, but separated LF-0 and LF-6 (Fig. 5a). Toxicity and % sand explained a high percent of species composition in the RDA (Table 3), accounting for 74% in October and 72% in May of the explained variation. The highest amount of variation was explained by % sand in both October (canonical eigenvalue 0.31) and May (canonical eigenvalue 0.28), followed by toxicity (October, canonical eigenvalue 0.14; May canonical eigenvalue 0.22). Other statistically significant variables, except dissolved Mn in May, were much lower in importance in ranking environmental variables (Table 3). LF-5 changed patterns seasonally. When only slightly toxic in October, community structure was similar to LF-0 and LF-6 according to TWINSPAN (Fig. 4a). When toxic in May, structure became more similar to that at LF-3 and LF-4 (Fig. 5a). Toxicity at LF-5 was the greatest of any site; perhaps related to low pH and higher concentrations of Cd and Cu found at this site in May (Table 2). It appeared that Colorado Gulch, just upstream of LF-5, was the source of metals. Large quantities of orange sediment with low pH water (5.1) was entering Lake Fork at the mouth of Colorado Gulch. This drainage may be an intermittent source of low pH and metals, perhaps related to snowmelt. Four sample groups were identified (Figs. 4 and 5) in both October and May using TWINSPAN with classification limited to major divisions (levels

1–3). Indicators identified by TWINSPAN included Capniidae, Paraleuctra, and Ephemerella in October (Fig. 4b) and Paraleuctra, Baetis, and Ephemerella in May (Fig. 5b). Capniidae in October appeared to separate the largely unimpacted sites LF-0, LF-5, and LF-6 from sites affected by either regulation (LF-1 and LF-2) or metals (LF-3 and LF-4). Paraleuctra then separated metals impacted sites LF-3 and LF-4 from LF-1 and LF-2. At the third division, Ephemerella separated LF-6 from all the other stations. In May, Paraleuctra separated stations LF-3, LF-4, LF-5, and LF-6 from the other stations. Baetis separated LF-1 and LF-2 from LF-0, while Ephemerella separated LF-6 from all others. Taxa and richness within the orders Ephemeroptera, Plecoptera, and Trichoptera (EPT richness) decreased below the reservoir (LF-1 and LF-2), increased slightly at stations below the metal source (LF-3 and LF-4), and increased again downstream at LF-6 (Table 4). Abundance was highest at stations immediately below the reservoir (Table 4). 3.3. Correlation of indicator organisms with mining and regulation impacts Indicators identified by RDA and TWINSPAN found during both seasons of the survey included Paraleuctra, Polycelis, Hydra, Simulium and Ephemerella. Ephemerella was not important for identifying anthropogenic stressors in this study but did differentiate between the less-impacted but

74

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

Fig. 4. Biplots from data collected in October 1999 based on a RDA analysis of sites (a) and hyporheic macroinvertebrates (b) with respect to environmental variables. Samples are labeled as LF-0 (䊊), LF-1 (䊉), LF-2 (䊐), LF-3 (䉫), LF-4 (+), LF-5 (䉭), and LF-6 ( ). Environmental variables were related (P < 0.05) to community attributes as shown by arrows. Polygons (a) enclose groups as defined by TWINSPAN. Taxa in bold (b) were identified by TWINSPAN as indicator organisms.

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

75

Fig. 5. Biplots from data collected in May 2000 based on a RDA analysis of sites (a) and hyporheic macroinvertebrates (b) with respect to environmental variables. Samples are labeled as LF-0 (䊊), LF-1 (䊉), LF-2 (䊐), LF-3 (䉫), LF-4 (+), LF-5 (), and LF-6 ( ). Environmental variables were related (P < 0.05) to community attributes as shown by arrows. Polygons (a) enclose groups as defined by TWINSPAN. Taxa in bold (b) were identified by TWINSPAN as indicator organisms.

76

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

Table 4 EPT and taxa richness and abundance associated with Lake Fork sites Site

Month

Mean EPT richness per pot

Mean taxa richness per pot

LF-0

October May

4.7 (0.7) 10.0 (2.0)

15.7 (2.3) 16.0 (3.5)

102.7 (36.8) 290.3 (87.6)

LF-1

October May

2.0 (0.6) 5.0 (0.6)

11.7 (0.9) 15.3 (0.3)

340.0 (48.2) 724.0 (117.8)

LF-2

October May

3.7 (0.3) 6.3 (0.7)

13.3 (0.9) 17.3 (1.4)

659.0 (102.6) 1546.7 (329.8)

LF-3

October May

3.7 (0.3) 6.7 (0.9)

14.3 (1.4) 18.0 (1.5)

230.0 (34.3) 347.7 (22.4)

LF-4

October May

5.3 (1.2) 10.7 (0.3)

15.0 (2.6) 18.0 (0.6)

218.0 (77.4) 657.7 (46.7)

LF-5

October May

4.7 (1.2) 5.0 (1.5)

14.0 (2.0) 12.0 (1.5)

96.0 (38.5) 58.7 (10.8)

LF-6

October May

5.7 (0.3) 9.5 (0.5)

25.7 (2.3) 26.0 (2.0)

297.7 (26.3) 120.0 (46.0)

longitudinally separated sites LF-0 and LF-6. Paraleuctra may be useful for separating sites based on mining impacts. Paraleuctra was common (maximum abundance was 118 individuals per hyporheic pot, 50,081 m−3 ) at toxic sites. When October and May collections were combined (n = 39), Paraleuctra was positively correlated with toxicity (r = 0.63, P < 0.0001) and uncorrelated with sand (r = 0.31, P = 0.0554). Polycelis, Hydra, and Simulium were associated in the RDA (Figs. 4 and 5) with sites below the reservoir at LF-1 and LF-2. All three of these taxa were negatively correlated with sand (Polycelis, r = −0.6449, P < 0.0001; Hydra, r = −0.5510, P = 0.0002; Simulium, r = −0.4593, P = 0.0029) but not with toxicity (P > 0.0817). Paraleuctra Hydra, and Simulium were strongly associated with hyporheic samples in Lake Fork and rarely found in surface collections (Surber sampler) in the area (Nelson, unpublished data).

4. Discussion 4.1. Indicator organisms Simple metrics such as EPT and taxa richness appeared to have little value for identifying mining impacted sites in this system also impacted by reg-

Mean abundance per pot

ulation. Mean richness values at mining impacted sites increased slightly relative to sites immediately upstream; an atypical pattern in the Rocky Mountain region (Clements et al., 2000). Indicators, such as EPT richness, derived from studies of surface benthos may not be appropriate for use in hyporheic studies. Taxa that responded to identified stressors during both October and May included Paraleuctra, Polycelis, Hydra, and Simulium. Paraleuctra appeared to be positively associated with mining impacted sites. Perhaps what appears to be a positive response to toxicity by Paraleuctra, is really a response to increased bacteria associated with metals deposition. Brunke and Fischer (1999) have noted a high correlation of numbers of hyporheic invertebrates with bacterial abundance and have proposed that bacteria may be very important as food in the hyporheic. The biomass of ferromanganese-depositing bacteria may increase in iron/manganese oxide deposition zones (Wellnitz and Sheldon, 1995) and provide food for some macroinvertebrates (Wellnitz et al., 1994). Perhaps this is the case with Paraleuctra at Lake Fork stations that are chronically exposed to inputs of Fe and Mn. Sites affected by regulation contained species such as the non-insect taxa Polycelis and Hydra that are often associated with these impacts (Voelz and Ward, 1990, 1991). Hampton (1988) in a habitat study found that Polycelis was associated with high altitude,

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

relatively constant cool water temperatures, and substrate which contained low amounts of sand and silt, with rubble harboring the greatest numbers of Polycelis. These conditions correspond to habitat present immediately below Turquoise Lake. Habitat at stations further downstream is likely less suitable for Polycelis. Hydra was found in numbers as high as 527 individuals per hyporheic pot (223,666 individuals m−3 ) at sites associated with the dam. It is unclear from the literature why Hydra are sometimes associated with regulated streams. It is possible that they are stenothermic and respond positively to the relatively constant water temperatures coming from deep-release storage impoundments (sensu Rader and Ward, 1988). Armitage (1976) suggested that a large Hydra population below Cow Green Reservoir was encouraged by zooplankton which were produced in the reservoir. This may not be the case with Hydra in Lake Fork. Dissected individuals contained chironomids for the most part. Sand appeared to be an important variable in this study and the low amounts of sand and increased substrate permeability near reservoirs may favor Hydra. Clear, sediment-free water released below dams has a high erosive potential which can entrain fine particles and increase permeability. The use of Hydra as an indicator of regulation effects in Lake Fork may be somewhat confounded by the sensitivity of this genus to metals toxicity (Rippon and Riley, 1996; Beach and Pascoe, 1998). The low numbers of Hydra at LF-3 and LF-4 may not be indicative of recovery from regulation (increased sand), but instead symptomatic of toxicity. Simulium has also been associated with hypolimnetic reservoirs in Colorado (Voelz and Ward, 1991). Voelz and Ward (1991) suggest that high densities of simuliids at sites near dams is caused by inputs of plankton and bacteria from the reservoirs and the absence of other superior competitors. The low amounts of sand in the hyporheic below Turquoise Lake may also provide increased attachment surfaces for this insect. The regulation recovery gradient appeared to be confounded to some degree with the input of metals from the Dinero Tunnel. Temperature CV and % sand tended to increase at the sites (LF-3 and LF-4) that are most impacted by chronic mining impacts. It would be fairly easy to interpret the declines of regulation indicators at these sites as primarily caused by mining

77

impacts. RDA and correlation analyses suggest that this is not the case. Despite the complexity of community responses, biotic hyporheic data appeared to be useful in discriminating between types of stressors. Hyporheic taxa that appeared to respond to stressors included Paraleuctra and Hydra, which were not commonly found in surface sediments. If the Dinero Tunnel metal source is treated, there should be a decrease in Paraleuctra at Lake Fork stations immediately below the confluence with the Tunnel. If regulation impacts were decreased, Polycelis, Hydra, and Simulium may become less abundant at LF-1 and LF-2. It appeared that only a short distance was required for recovery of benthic community from impacts of both metals and regulation. The small size (channel width of 8 m at LF-1) of this stream may have aided in recovery from temperature effects of regulation because the relative low flows allowed for rapid response to variable ambient temperatures. Palmer and O’Keefe (1990) noted that small rivers recover faster from impoundment effects than do large ones. Additions of water from unregulated tributaries, such as Siwatch tunnel, Colorado Gulch, and Rock Creek diversions also helped in recovery of natural flow attributes (e.g. Munn and Brusven, 1987) and provided fine sediments. Clean water from Rock Creek and uptake of metals by sediments aided in the protection and recovery of benthic communities from metals effects. Comparisons of toxicity in surface waters and the hyporheic, suggest that the hyporheic was a refuge for organisms from highly toxic dissolved Zn concentrations. Hyporheic toxicity at LF-3 was seasonally variable despite the constancy of toxicity in surface waters at this site. This site was strongly upwelling in May when toxicity was absent, intimating that less-toxic water from another source was entering the hyporheic. 4.2. Spatial and seasonal variability of mining impacts Analysis with RDA showed that significant environmental variables associated with hyporheic communities were largely those concerning river regulation (% sand) and toxicity (metals impacts) and that these variables explained a large amount of the variation in community structure. Toxicity correlated with many other parameters including pH, hardness, conductivity, and deposited and dissolved metals, especially Zn.

78

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79

Dissolved Mn was also recognized as an important community structuring agent for RDA in May. Metals, as expected, were associated with sites downstream of the Dinero Tunnel. There was, however, also a seasonally important metals source just upstream of LF-5. Dissolved Fe and Zn increased from October to May in surface water and Mn and Zn increased in the hyporheic downstream of Colorado Gulch. It is likely that this second source of metals is related to spring snowmelt increasing discharge from Colorado Gulch and perhaps also percolating through tailings piles in the area. Deposited metals may also leach from substrate because of lowered springtime pH’s and therefore in-stream substrate could serve as an additional metals source. RDA consistently grouped macroinvertebrate samples with sites experiencing a specific impact. Macroinvertebrate assemblages from sites LF-0 and LF-6 that were largely unimpacted, were grouped together by RDA despite being separated by the greatest longitudinal distance. Assemblages from sites that would be expected to be mostly affected by regulation were also grouped together (LF-1and LF-2). Macroinvertebrate assemblages closest to the metals inputs varied between seasons. During October, these sites (LF-3 and LF-4) were spatially near each other in the RDA diagram, while in May, when toxicity decreased in the hyporheic at LF-3, there was greater scatter. LF-5 changed TWINSPAN groups between October and May. When less toxic in October it was grouped with the less impacted site LF-0, while in May when toxic, it was grouped with LF-3 and LF-4. There was close correspondence between RDA and TWINSPAN analyses except that TWINSPAN separated LF-O and LF-6 at the last division. The similarity between analyses suggests that environmental constraints in RDA had a minimal effect on the ordination. 4.3. Ranking of stressors Our analyses of community structure indicated that biotic assemblages differed between regulation and metal impacts. Regulation (% sand) was of greatest importance in explaining this RDA community structure. The other important environmental variable identified from canonical eigenvalues was toxicity (Table 3). Studies of leaf litter breakdown at these sites (Nelson, 2000; Nelson and Roline, 2000) showed differences in functioning with these two types of

impacts. Leaf breakdown was not different above and below Dinero Tunnel on Lake Fork (Nelson, 2000) but was very different from above to below the reservoir (Nelson and Roline, 2000). Macroinvertebrate community structure differed with both types of stressors, suggesting that environmental stressors are more easily detected with structure rather than changes in a functional attribute such as leaf pack breakdown, a finding previously reported by Schindler (1987). Ecosystem impacts may be greater when changes in both structure and function are detected, as opposed to the case where only changes in structure are found. These data also support the RDA analysis findings that the impacts of regulation, in this case, are greater than those associated with mine drainage. Positioning of stressors along Lake Fork, however, may have some role in degree of impact. Because the initial influence to the benthic community is from regulation, this altered, and perhaps less responsive, community is the one that is then subjected to metal contaminants. This could decrease the importance of metals in structuring the hyporheic benthos. Potential hyporheic indicator organisms below reservoirs may have little in common with those used in more typical lotic studies (e.g. taxa and EPT richness). Biological response to multiple stressors may result in anomalous communities that have no counterpart in single stressor systems. Differences between hyporheic and surface water quality, including toxicity and taxa, indicate that knowledge of this zone is important to understanding impacted aquatic systems.

Acknowledgements Richard Durfee (RD Aquatics) identified macroinvertebrates in some samples. Thanks to Colorado Outward Bound and to Sam Scott for granting permission to sample at two of the sites. We also thank Will Clements for reviewing an early draft of this paper. This work was supported by the Science and Technology Program of the Bureau of Reclamation.

References APHA, 1989. Standard Methods for the Analysis of Water and Wastewater, 17th ed.

S.M. Nelson, R.A. Roline / Ecological Indicators 3 (2003) 65–79 Armitage, P.D., 1976. A quantitative study of the invertebrate fauna of the River Tees below Cow Green Reservoir. Freshwater Biol. 6, 229–240. Beach, M.J., Pascoe, D., 1998. The role of Hydra vulgaris (Pallas) in assessing the toxicity of freshwater pollutants. Water Res. 32 (1), 101–106. Bitton, G., Jung, K., Koopman, B., 1994. Evaluation of a microplate assay specific for heavy metal toxicity. Arch. Environ. Contam. Toxicol. 2, 25–28. Brunke, M., Fischer, H., 1999. Hyporheic bacteria-relationships to environmental gradients and invertebrates in a prealpine stream. Arch. Hydrobiol. 146 (2), 189–217. Clements, W.H., Carlisle, D.M., Lazorchak, J.M., Johnson, P.C., 2000. Heavy metals structure benthic communities in Colorado mountain streams. Ecol. Appl. 10 (2), 626–638. Dash, R.G., Ortiz, R.F., 1996. Water-quality Data for the Arkansas River Basin, Southeastern Colorado, 1990–1993. Open-File Report 95-464, US Geological Survey, Denver, CO. EPA., 1983. Methods of Chemical Analysis of Water and Wastes. EPA600/4-79-020, Environmental Monitoring and Support Laboratory, Cincinnati, OH. Folt, C.L., Chen, C.Y., Moore, M.V., Burnaford, J., 1999. Synergism and antagonism among multiple stressors. Limnol. Oceanogr. 44 (3, Part 2), 864–877. Gibert, J., Marmonier, P., Vanek, V., Plénet, S., 1995. Hydrological exchange and sediment characteristics in a riverbank: relationship between heavy metals and invertebrate community structure. Can. J. Fish. Aquat. Sci. 52, 2084–2097. Hampton, A.M., 1988. Altitudinal range and habitat of triclads in streams of the Lake Tahoe basin. Am. Midland Naturalist 120 (2), 302–312. Jooste, S., 2000. A model to estimate the total ecological risk in the management of water resources subject to multiple stressors. Water SA 26 (2), 159–166. Munn, M.D., Brusven, M.A., 1987. Discontinuity of Trichopteran (Caddisfly) communities in regulated waters of the Clearwater River, Idaho, USA. Regulated Rivers Res. Manage. 1, 61–69. Nelson, S.M., 2000. Leaf pack breakdown and macroinvertebrate colonization: bioassessment tools for a high-altitude regulated system? Environ. Pollut. 110, 321–329. Nelson, S.M., Roline, R.A., 1998. Evaluation of the sensitivity of rapid toxicity tests relative to daphnid acute lethality tests. Bull. Environ. Contam. Toxicol. 60, 292–299. Nelson, S.M., Roline, R.A., 1999. Relationships between metals and hyporheic invertebrate community structure in a river recovering from metals contamination. Hydrobiologia 397, 211– 226. Nelson, S.M., Roline, R.A., 2000. Leaf litter breakdown in a mountain stream impacted by a hypolimnetic release reservoir. J. Freshwater Ecol. 15 (4), 479–490.

79

Palmer, R.W., O’Keefe, J.H., 1990. Downstream effects of a small impoundment on a turbid river. Arch. Hydrobiol. 119 (4), 457– 473. Peltier, W.H., Weber, C.I., 1985. Methods for Measuring the Acute Toxicity of Effluents to Freshwater and Marine Organisms, third ed. EPA/600/4-85/013, Environmental Monitoring and Support Laboratory, Cincinnati, OH. Rader, R.B., Ward, J.V., 1988. Influence of regulation on environmental conditions and the macroinvertebrate community in the upper Colorado river. Regulated Rivers Res. Manage. 2, 597–618. Richards, C., Bacon, K.L., 1994. Influence of fine sediment on macroinvertebrate colonization of surface and hyporheic stream substrates. Great Basin Nat. 54, 106–113. Rippon, G.D., Riley, S.J., 1996. Environmental impact assessment of tailings dispersal from a uranium mine using toxicity testing protocols. Water Resources Bull. 32 (6), 1167–1175. Schindler, D.W., 1987. Detecting ecosystem responses to anthropogenic stress. Can. J. Fish. Aquat. Sci. 44 (Suppl. 1), 6–25. Stanford, J.A., Ward, J.V., 1988. The hyporheic habitat of river ecosystems. Nature 335, 64–66. Taylor, K.V., 1978. Erosion downstream of dams. In: Binger, W.V., Buehler, J.P., Clarke, F.J., DeLuccia, E.R., Harza, R.D., Jansen, R.B., Peters, J.C., Thomas, M.F., Chadwick, W.L. (Eds.), Environmental Effects of Large Dams. American Society of Civil Engineers, NY. ter Braak, C.J.F., Verdonschot, P.F.M., 1995. Canonical correspondence analysis and related multivariate methods in aquatic ecology. Aquat. Sci. 57, 255–289. Vallett, H.M., Hakenkamp, C.C., Boulton, A.J., 1993. Perspectives on the hyporheic zone, integrating hydrology and biology, introduction. J. North Am. Benthol. Soc. 12, 40–43. Voelz, N.J., Ward, J.V., 1990. Macroinvertebrate responses along a complex regulated stream environmental gradient. Regulated Rivers Res. Manage. 5, 365–374. Voelz, N.J., Ward, J.V., 1991. Biotic response along the recovery gradient of a regulated stream. Can. J. Fish. Aquat. Sci. 48, 2477–2590. von Gunten, H.R., Karametaxas, G., Krähenbühl, U., Kuslys, M., Giovanoli, R., Hoehn, R., Keil, R., 1991. Seasonal biogeochemical cycles in riverborne groundwater. Geoch. Cosm. A 55, 3597–3609. Wellnitz, T.A., Grief, K.A., Sheldon, S.P., 1994. Response of macroinvertebrates to blooms of iron-depositing bacteria. Hydrobiologia 281, 1–17. Wellnitz, T.A., Sheldon, S.P., 1995. The effects of iron and manganese on diatom colonization in a Vermont stream. Freshwater Biol. 34, 465–470.