Sensitivity of macroinvertebrate indicator taxa to metal gradients in mining areas in Northern Spain

1 Sensitivity of macroinvertebrate indicator taxa to metal gradients in mining areas in 1 Northern Spain 2 3 4 5 N. COSTAS1, I. PARDO1,*, L. MÉNDEZ-FERNÁNDEZ2, M. MARTÍNEZ-MADRID3 and P. RODRÍGUEZ2 6 7 8 9 1Dpt. Ecology and Animal Biology. University of Vigo. 36310 Vigo, Spain. 10 2Dpt. Zoology and Animal Cell Biology. University of the Basque Country, Apdo. 644. 48080 11 Bilbao, Spain. 12 3Dpt. Genetics, Physical Anthropology and Animal Physiology. University of the Basque Country 13 Apdo. 644. 48080 Bilbao, Spain 14 15 16 17 18 19 *Corresponding author: Tel.: +34 986812585; fax: +34 986812556; e-mail:20 [email protected] 21 22 23 Declarations of interest: none 24 This is the accepted manuscript of the article that appeared in final form in Ecological Indicators 93 : 207-218 (2018), which has been published in final form at https://doi.org/10.1016/j.ecolind.2018.04.059. © 2018 Elsevier under CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) 2 25 Abstract 26 The study of macroinvertebrate communities using a Reference Condition Approach 27 (RCA) is a powerful tool for detecting the impacts of mining activities due to heavy metal 28 contamination in rivers. However, studies on this topic based on an RCA fully satisfying the 29 Water Framework Directive (WFD) criteria for reference site selection are still scarce. Following 30 a WFD-compliant RCA design, a total of 14 reference and 15 test sites were intensively sampled 31 during July 2015 in the Nalón River basin (Asturias, Northern Spain). Macroinvertebrate samples, 32 water physicochemistry, metal concentrations in sediments and habitat quality were evaluated 33 at each selected site. To determine whether increasing heavy metal levels in sediments have 34 significant ecological impacts in the structure of macroinvertebrate communities in the study 35 area, we applied a combination of non-parametric multivariate and univariate procedures, as 36 well as gradient analysis and Threshold Indicator Taxa ANalysis (TITAN) over the biotic and 37 environmental datasets. We found significant alterations in the macroinvertebrate community 38 structure with respect to the reference condition (REF group) for those test sites within mercury 39 (HG group) and gold (AU group) mining areas, with Ephemeroptera, Plecoptera and Trichoptera 40 (EPT) richness and abundance being the community descriptors showing the strongest 41 alterations in relation to mining pressures in the studied area. Metal concentrations in 42 sediments were significantly correlated to the metrics of community structure related to 43 abundance, richness and biodiversity, with As and Hg being the two metals showing higher 44 biological effects on macroinvertebrate assemblages. TITAN also allowed us to establish a set of 45 macroinvertebrate families specifically sensitive to metal concentrations in sediments, most of 46 which corresponded to EPT families. Our results prove that determined sensitive taxa could be 47 considered as reliable indicators of metal pollution in the mining areas of Northern Spain. From 48 a WFD perspective, our study clearly indicates that the responsive community descriptors found 49 in this study are actually key parameters in the evaluation of the ecological status in the rivers 50 of Northern Spain. These data are highly relevant for the future development of environmental 51 quality standards for river sediments in Spain. 52 53 54 55 56 57 Key-Words: Heavy metals; benthic macroinvertebrates; metrics; community structure; mines 58 59 3 60 1. Introduction 61 Freshwater ecosystems are one of the most endangered ecosystems worldwide, mainly 62 due to human dependence on them (Sala et al. 2000; Dudgeon et al. 2006). Increasing anthropic 63 pressure on running waters has led to severe levels of environmental degradation in these 64 ecosystems. Indeed, industrialisation, urbanisation, land-use changes and watercourse 65 alterations are the main drivers threatening running waters (Malmqvist & Rundle 2002). Among 66 the existing pressures, mining activities may well be one of the most persistent impacts on rivers 67 and streams (Marqués et al. 2001a), and metal pollution has become a major concern regarding 68 the preservation of aquatic systems in and downstream of mining areas (Luoma et al. 2010). In 69 general, together with suspended particulate material, sediments represent the main storing 70 site for heavy metals in rivers (mainly in the finer fraction, < 63 µm) (ERMITE-Consortium 2004), 71 and both sediment deposition and resuspension may represent potential sources of pollution at 72 different spatial scales (Martin 2000). 73 Assessing the risk of the ecological impacts from human activities is the core of an 74 effective management of freshwater ecosystems, and in this regard, the evaluation of the 75 relationships between biological responses and environmental pressures is crucial. The Water 76 Framework Directive (WFD, EC 2000) approach is focused on the development of ecological 77 status classification systems for water bodies based on assessments of the degree of deviation 78 in biological condition at particular study sites from the so-called Reference Condition (EC 2000). 79 In Europe, this concept is defined by the use of minimally disturbed sites (sensu Stoddard et al. 80 2006) characterised by type-specific biological conditions and the accompanying 81 physicochemistry and hydromorphology, in which the absence of significant human 82 disturbances has been assessed (Pardo et al. 2012). Within this context, the use of different 83 biological quality elements (i.e., macrophytes, phytobenthos, phytoplankton, benthic 84 macroinvertebrates and fish) is promoted by the WFD as an effective assessment tool for 85 detecting impacts derived from anthropogenic pressures, such as eutrophication or organic 86 pollution (Sánchez-Montoya et al. 2010; Pardo et al. 2014). Among them, metrics based on 87 macroinvertebrate community structure are probably the most widely used indicators of water 88 quality (Metcalfe 1989; Reynoldson & Metcalfe-Smith 1992; Birk et al. 2012; García et al. 2014). 89 In fact, worldwide mining-derived water and sediment contamination by metals was 90 demonstrated to have a direct effect on different benthic community structural components, 91 such as abundance, diversity, richness, or metrics related to sensitive species (Hirst et al. 2002; 92 Solà et al. 2004; Doi et al. 2007; Qu et al. 2010; Poulton et al. 2010; Ruiz-García et al. 2012; Byrne 93 et al. 2013). For all that, studies were reference conditions are designated according to WFD 94 4 precepts are desirable towards creating an adequate background to allow being able to 95 efficiently use benthic macroinvertebrates as indicators of metal pollution. 96 The Nalón River basin was one of the main mining-exploited areas in Northern Spain, 97 where the Texeo copper mines (located in the Riosa district) and the mercury mines (Mieres, 98 Pola de Lena and Somiedo districts) were still active until the early 1970s (Méndez-Fernández et 99 al. 2015). Spoil heaps derived from past mining activities can still represent an important source 100 of local contamination for river sediments (Loredo et al 2006, 2010). Previous studies have 101 reported sediment pollution in rivers affected by mine spoil heaps, as well as metal 102 bioaccumulation and toxicity in the Nalón River basin (Ordoñez et al. 2013; Méndez-Fernández 103 et al. 2015). The environmental quality standards (EQS) directive (EC 2013) recognised the 104 importance of both sediments and biota matrices for water quality policies in Europe. However, 105 both sediment and biota EQSs for metals and metalloids (henceforth, metals) need to be 106 developed by member states. Different proposals have attempted to relate metal tissue 107 concentrations to metal-induced ecological impacts on different macroinvertebrate metrics 108 (Luoma et al. 2010; Schmidt et al. 2011; Rainbow et al. 2012; De Jonge et al. 2013; Bervoets et 109 al. 2016). However, the integration of metal levels in the sediment due to mining activities and 110 the ecological impacts on river macroinvertebrate communities should be the object of deeper 111 research, especially in regions like the Nalón River basin with a long history of mining activities. 112 Thus, the evaluation of metal concentrations in sediments within a Reference Condition 113 Approach (RCA) may be relevant not only in the development of useful tools for assessing the 114 risk of mining activities derived from metal pollution, but also in the development of EQSs in 115 Spain or at European level. 116 A key issue in biomonitoring is the level of taxonomic resolution used. Although recent 117 studies pointed out the benefits of using finer taxonomic levels in stream assessment (Guerold 118 2000; Lenat & Resh 2001; Svitock et al. 2014), the family level continues to be preferred as it 119 was demonstrated to provide valuable ecological information (Lenat & Barbour 1994) and is able 120 to respond significantly to different pressures (for example, see Pardo et al. 2014). Moreover, 121 different studies demonstrated similar response patterns to different stressors at both the 122 genus and family levels (i.e., Bowman & Bailey 1997; Pond et al. 2008). Considering that family 123 constitutes the level of resolution used in ecological quality assessment in the context of the 124 WFD in Northern Spain (Pardo et al. 2014; METI 2015), evaluating the ability of this taxonomic 125 level to detect significant metal contamination is of high relevance. 126 The general aim of this study is to detect and estimate the possible effects of sediment 127 metal pollution on the macroinvertebrate communities inhabiting the rivers located in mining 128 areas of the Nalón River basin in order to improve assessment, management and conservation 129 5 efforts. We hypothesise that metal pollution plays a fundamental role in structuring 130 macroinvertebrate communities in these areas. The specific objectives include 1) to describe the 131 community composition of macroinvertebrates at both reference and potentially impacted 132 sites, 2) to assess the response of a benthic community structure to a gradient of metal 133 concentration in the sediments based on the metal levels measured at each site as a ratio with 134 a pre-selected reference condition in the study area, 3) to detect benthic taxa that are sensitive 135 to metal pollution using a wide exposure range of sediment concentrations, and 4) to determine 136 whether the variation in metrics at the family level can be detected by alterations due to 137 sediment metal concentrations. Overall, our research on the macroinvertebrate communities 138 within the mining areas of the Cantabrian region using a reference condition approach may 139 facilitate the identification of target-sensitive families for the development of adequate EQSs, 140 both at regional and European scales. 141 2. Materials and methods 142 2.1. Study area 143 This study was conducted in the Nalón River basin (Asturias, Northern Spain), including 144 both the Nalón and Narcea rivers and some of their tributaries (Fig. 1). This basin is the largest 145 one within the Cantabrian water district, with a total length of 140.8 km covering a total area of 146 4907 km2. The dominant climate is oceanic, characterised by mild winters and summers, and 147 abundant rainfall throughout the year (a mean annual precipitation of 1296 mm). As a 148 consequence of its lithological characteristics, the study region also witnessed historical mining 149 activity, with more than 800 sites active between 1950 and 1975 (Águeda-Villar & Salvador-150 González 2008; Ordóñez et al. 2008). 151 A total of stream 29 stream sites were sampled once during July 2015 (Fig. 1). The 152 sampling design considered two main groups of sites: reference sites (n = 14) and test sites (n = 153 15), and covered four of the macroinvertebrate community-based river typologies (Types 1, 2, 154 4 and 5) of the Cantabrian region (Pardo et al. 2014). Reference sites (REF group) were selected 155 according to WFD criteria (EC 2000) and validated following the procedure and criteria of the 156 absence of pressures described in Pardo et al. (2012). Eight of the fourteen reference sites 157 actually belong to the spatial network used by the Cantabrian Hydrographical Confederation 158 (CHC) the Nalón River Water Management Agency (unpublished data). Test sites were selected 159 covering rivers near historic (mercury and copper mines) and active (gold mine) mining 160 industries, and thus subject to known inputs of heavy metals. Three of them were located in a 161 copper mine area (CU group), 8 in mercury mine areas (HG group) and 4 in a gold mine area (AU 162 group) (Fig. 1). 163 6 Fig. 1. Location of the Nalón River basin (Northern Spain) and the 29 sampling sites (with their 164 corresponding identification codes) included in this study. ● = Reference sites (REF), ○ = sites within 165 mercury-mining areas (HG), = sites within gold-mining areas (AU), = sites within copper-mining areas 166 (CU). 167 168 169 2.2. Data collection and laboratory processing 170 2.2.1. Biological data 171 Collection and processing of macroinvertebrate samples was done according to the 172 Spanish official protocol ML-Rv-I-2013 (see Pardo et al. 2014). The sampling protocol followed a 173 multihabitat procedure adapted from Barbour et al. (1999). At each study site, a total of twenty 174 ‘sample units’ of 0.125 m2 each (distributed proportionately in the main habitats existing along 175 a 100 m reach) were collected using a kick-net (500 µm mesh size), and combined into a single 176 mixed sample (a total sampled area of 2.5 m2). After collection, the samples were preserved 177 with 70% ethanol and stored. All were washed and separated into three fractions using a series 178 of sieves of different mesh sizes (5, 1 and 0.5 mm). The invertebrates were sorted in each 179 fraction and subsamples taken in the 1 and 0.5 mm fractions only when necessary (Wrona et al. 180 1982). Taxon richness and abundance were evaluated at the family level (except for aquatic 181 oligochaetes and Hydracarina) under 57x magnification (Olympus SZX9). 182 2.2.2. Environmental data 183 7 Water samples from each site were collected in 100 ml polypropylene bottles and kept 184 frozen until their analysis at CACTI (Centro de Apoio Ciéntifico e Tecnolóxico á Investigación, 185 University of Vigo, Spain). Nine variables were measured: nitrites (N-NO2−), nitrates (N-NO3−), 186 phosphates (P-PO4-3), ammonium (N-NH4+) and chlorides (Cl−) by means of a continuous-flow 187 analyser (Auto-Analyser AA3, Bran+Luebbe, Germany); cations of calcium (Ca+2), magnesium 188 (Mg+2), potassium (K+) and sodium (Na+) were analysed by ICP-OES (Optima 4300 DV, Perkin-189 Elmer). Several physicochemical variables were measured in situ using portable devices: water 190 temperature, pH and dissolved oxygen (Orion 5-star, ThermoScientific), and electric conductivity 191 (Orion 3-Star, ThermoScientific). 192 Sediment samples from the upper 5-10 cm layer of fine sediments were collected along 193 a 25 m reach of the river bank at each study site using a stainless-steel spade and stored at 4°C 194 in the dark. Particle-size distributions (expressed as a percentage of the dry weight), total organic 195 carbon percentages (%TOC: Bryan et al. 1985; USEPA 1990) and metal concentrations (Al, As, 196 Cd, Cr, Cu, Fe, Hg, Mn, Ni, Pb, Se, and Zn, reported as dry weight, µg g-1 dw) were measured in 197 the sediment samples according to methods described in Méndez-Fernández et al. (2017); for 198 Al, Fe and Mn, their corresponding limits of quantification (LOQ) were 0.05 µg l-1, 0.03 µg l-1 and 199 0.1 µg l-1, respectively. Recovery rates were within certified values (Al has no certified values) in 200 the standard sediments (Buffalo River sediment: RM8704, USA and Sewage Sludge-3: CRM 031-201 040, UK), used as quality controls. 202 The Fluvial Habitat Index IHF (Pardo et al. 2002) was calculated at each site. This index 203 is a measurement composed of several aspects of habitat heterogeneity divided into seven 204 blocks: substrate inclusion in riffles and sedimentation in pools, frequency of riffles, substrate 205 diversity, flow velocity and depth regime, shade cover, heterogeneity in both physical habitat 206 diversity and food resources, and aquatic vegetation coverage and composition. 207 2.3. Statistical analysis 208 2.3.1. Multivariate analysis of macroinvertebrate communities 209 The entire biological dataset consisted of a log(x+1)-transformed ‘taxon x site’ matrix of 210 abundances. The effects of metal levels in the sediments and other chemical and physical 211 environmental factors on the benthic community structure were examined using a 212 permutational multivariate analysis of variance (PERMANOVA) (Anderson 2001). One-way 213 models (fixed factors = ‘Mining activity’), using 9999 maximum unrestricted permutations, 214 followed by pairwise comparisons (at 999 permutations) were performed at a significance level 215 of α < 0.05. The Bray-Curtis index was chosen as a suitable multivariate distance measure for 216 taxon composition and structure (Bray & Curtis 1957). Due to the lower number of possible 217 permutations for the pairwise tests, a Monte-Carlo sampling procedure was applied to obtain 218 8 more reliable P values (Anderson et al. 2008). To help clarify the nature of the PERMANOVA 219 results, a permutational analysis of multivariate dispersions (PERMDISP) (at α < 0.05) was also 220 applied to test for homogeneity of dispersions between the mining activity groups of sites, in 221 order to determine if any significant PERMANOVA result may be in fact due the effect of the 222 considered factor or if a dispersion effect is present and further data exploration is desirable. 223 Similarity percentage analyses (SIMPER) (Clarke & Gorley 2006) were applied to determine 224 which taxa contributed most to the within-group similarities and dissimilarities between groups. 225 All analyses were conducted using PRIMER v.6 software (Clarke & Gorley 2006). 226 2.3.2. Univariate descriptors of macroinvertebrate community structure 227 In the search for specific descriptors of alterations in community structure, 228 macroinvertebrate data were used to calculate a set of abundance and diversity metrics, mainly 229 selected on the basis of previous studies assessing mining impacts (Gray & Delaney 2008; Qu et 230 al. 2010; Poulton et al. 2010; Byrne et al. 2013): total abundance, abundance of EPT 231 (Ephemeroptera, Plecoptera and Trichoptera) taxa, % of 3 dominant taxa, % of oligochaetes, 232 taxon richness, richness of EPT taxa, and the Margalef richness and Shannon-Wiener diversity 233 indices. Kruskal-Wallis non-parametric tests followed by multiple comparisons with the Dunn-234 Bonferroni post-hoc test, were used to assess for significant differences in the selected indices 235 for the predefined mining activity treatments (REF, HG, CU and AU site groups). Comparisons 236 were conducted using the IBM® SPSS Statistics 22 software. 237 2.3.3. Environmental gradients 238 The habitat matrix consisted of all measured environmental data that might potentially 239 affect the distribution of macroinvertebrate assemblages along the study area (29 samples x 30 240 variables). Multivariate gradient analyses were employed to explore the variability in the habitat 241 characteristics and visualise major trends within the dataset. Spatial ordination of 242 environmental variables (previously log-transformed and standardised) was examined through 243 Principal Coordinate Analysis (PCoA) in PRIMER v.6 software (Clarke & Gorley 2006), and the 244 contribution of each variable to the ordination was calculated by Pearson correlation analysis to 245 identify the stressor gradients that better characterised the study area. Prior to PCoA, redundant 246 variables were eliminated by means of a Spearman’s correlation matrix (at α < 0.05), establishing 247 r = 0.7 as the cut-off level. 248 To assess metal levels in the sediments, we first estimated background metal levels as 249 the concentration range in sediments measured in pre-established reference sites, i.e., in the 250 absence of identified anthropogenic pressures (although diffuse pollution could not be 251 discounted). Threshold levels were defined as the high percentile of the background data 252 distribution in reference sites of the Nalón River basin, and interpreted as no-effect 253 9 concentrations. Metal threshold levels were calculated as the 90th percentile (P90) (with their 254 95% confidence intervals) from the distribution of the log-transformed data at the reference 255 sites, calculated after bootstrapping (with 1000 restarts) using IBM® SPSS Statistics 22 software, 256 and then anti-logged to transform them back. 257 2.3.4. Pressure-response relationships 258 To ascertain whether increasing levels of metals in the sediments were in fact associated 259 with the possible de-structuring effects detected in the macroinvertebrate communities through 260 multivariate and univariate comparisons, two approximations were considered: a) Pearson 261 correlation coefficients between the log-transformed (when necessary) values of each 262 univariate metric and the concentrations of heavy metals and SedPoll scores, calculated with 263 IBM® SPSS Statistics 22 software; b) individual invertebrate responses to stress gradients 264 evaluated by means of Threshold Indicator Taxa ANalysis (TITAN) (Baker & King 2010) (run with 265 the TITAN 2.0 package in R.3.3.2.). The TITAN method uses indicator value (IndVal) scores 266 obtained from indicator species analysis (Dufrêne & Legendre 1997) and standardised to z-267 scores to detect both the location of taxon-specific change points and the response direction 268 along an environmental gradient (Baker & King 2010). Depending on the taxa associated with 269 either side of the change point, they are designated as declining (z-) or increasing (z+), and are 270 used to trace cumulative responses: sum(z-) and sum(z+) (Baker & King 2010). Large values of 271 sum(z) scores imply that many taxa show strong responses at a similar value of the 272 environmental gradient. Synchronous significant changes among taxa are considered as 273 evidence of a community-level threshold (Baker & King 2010). By means of bootstrapping, TITAN 274 identifies reliable and pure indicator taxa as well as the uncertainty around obtained change 275 points (both at the taxon and community levels). We performed TITAN on the log(x+1)-276 transformed abundances of those macroinvertebrate taxa occurring in at least 3 samples (Baker 277 & King 2013). 278 Taxon responses were evaluated along the general gradients obtained from the PCoA 279 and along a metal pollution gradient in the sediment defined by the degree of difference of the 280 test sites to the reference condition. At each site, every single metal concentration in the 281 sediment was divided by the corresponding P90 value calculated in the reference sites to get a 282 metal quotient (P90-Q). The metal concentration in the sediment was then classified into five 283 potential categories on the basis of the metal quotients as follows: 1) Similar-to-reference sites 284 where P90-Q ≤ 1.0, 2) Low metal concentration where P90-Q = 1.1-2.0, 3) Medium metal 285 concentration where P90-Q = 2.1-10.0, 4) High metal concentration where P90-Q = 10.1-50.0, 286 and 5) Very high metal concentration where P90-Q ≥ 50.1. Next, at each site the relevant metals 287 were selected (see the Results section) and P90-Q values summed and standardised by the 288 16 3.3. Pressure-response relationships 426 The significant differences in community structure previously shown in section 3.1, 427 appeared to be clearly associated with metal levels in the sediment according to the pressure-428 response analyses. 429 Correlation analyses between metal concentrations in the sediments and the metric 430 descriptors of community structure showed a significant correlation (Table 5). As and Hg were 431 the two metals showing higher effects on the macroinvertebrate community abundance, 432 richness and biodiversity, with 6 and 5 metrics affected, respectively (Table 5); for Cd, Cu and Se 433 only 1 to 3 metrics were altered, and Pb was not significantly correlated with any of the 434 evaluated metrics. The strongest observed associations were EPT abundance with As (r = -0.800, 435 P < 0.001), and EPT richness with As and Hg (r = -0.646 and r = -0.659 respectively, P < 0.01) 436 (Table 5). Indeed, abundance metrics were more frequently affected by several metals (4), 437 followed by total richness (3 metals), suggesting that these metrics might be useful in detecting 438 metal mixture pressures in field situations. Results that are reinforced by the significant 439 correlations observed also between metrics and the SedPoll scores (Table 5). 440 441 Table 5. Pearson correlation coefficients of the metal concentrations in sediments (used as relevant for 442 TITAN) and the 9 selected descriptors of benthic community structure. Significant correlations are 443 indicated by * P < 0.05; ** P < 0.01 and *** P < 0.001. 444 445 On the whole, 20 families were identified as indicator taxa of the stress gradients 446 defined by the first two axes of the environmental PCoA (PCo1 and PCo2) (Fig. 4). Fourteen of 447 these taxa showed synchronous declines in response to the increasing levels of As, Cd, Cu, Se, 448 Cl-and % of the fine sediment fraction (< 63 µm) defined by PCo1 (see Fig. 3 and section 3.2). 449 Among these 14 taxa, 9 were EPT taxa, 3 were Coleoptera and 2 Diptera (Fig. 4). Regarding the 450 gradient defined by PCo2, four families (Ancylidae, Brachycentridae, Glossossomatidae and 451 Sericostomatidae) declined with increasing levels of Hg, P-PO4-3 and Cl-, while 2 families 452 (Hydrometridae and Hydracarina) seemed to be tolerant (z+) to these increases (Fig. 4). 453 However, considering that the mentioned PCo2 gradient is defined both by negative (Se and Cd) 454 and positive (Hg) correlations with metals, interpreting to what extent they are tolerant to Hg 455 17 or sensitive to Se and Cd is difficult. However, it was not possible to establish change thresholds 456 at the community level because according to the TITAN premises, a reliable community 457 threshold may be based on synchronous changes in the abundance of many indicator taxa within 458 a narrow range of the considered predictor gradient (Baker & King 2010). However, in our data, 459 there was a limited number of reliable indicator taxa, and they exhibited broad confidence 460 intervals. By considering the 95th quantile interval (QI) as indicative of the maximum resistance 461 of each sensitive indicator taxon to the mining gradients defined by the PCoA axes, we found 462 that most families showed differential sensitivities to the defined stressors both between orders 463 and within the same order (95th QI range: -0.15 to 4.33) (Table 6). In general, the dipteran family 464 Tipulidae (95th QI = -0.15) and the caddisflies Philopotamidae (95th QI = 0.44) and 465 Sericostomatidae (95th QI = 0.80) were the most sensitive taxa, while the riffle beetle family 466 Elmidae (95th QI = 4.33) showed the highest resistances among the obtained indicator taxa 467 (Table 6). 468 Using the sediment pollution scores (SedPoll) gradient, a total of 9 taxa (see Fig. 4) 469 showed synchronous decreases with the increasing metal pollution gradient. When considering 470 sum (z-) the obtained change point was 3.7, with a confidence interval ranging between 0.55 471 and 12.25. Among the indicator taxa, Baetidae appeared as the most resistant to SedPoll (95th 472 QI = 26.25), while Sericostomatidae (95th QI = 3.70), followed by Scirtidae (95th QI = 5.48) and 473 Heptageniidae (95th QI = 9.25), were less resistant (Table 6). 474 475 Fig. 4. Change points for significant macroinvertebrate indicator taxa (purity ≥ 0.95, reliability ≥ 476 0.95, P < 0.05) identified in Threshold Indicator Taxa ANalysis (TITAN), across the gradients 477 defined by the PCo1 and PCo2 axes of the environmental Principal Coordinate Analysis (PCoA) 478 and Sediment Pollution gradient (SedPoll) across the 29 sampling sites. Solid and dotted lines 479 represent the 5th-95th quantile intervals for each change point, filled symbols (●) correspond to 480 sensitive taxa (z-) and empty symbols (○) to tolerant taxa (z+) (see text in the Results section for 481 more details on this last case). 482 483 18 Table 6. TITAN indicator taxa for the environmental gradients obtained from the sediment pollution 484 gradient (SedPoll) and PCoA (two main axes: PCo1 and PCo2). The indicator value (IndVal), environmental 485 change point based on z maximum (zenv.cp) and quantile intervals (QI) of 5% and 95% are given for each 486 taxon. Response directions (zor z+) assignments are also indicated. 487 488 4. Discussion 489 Mining impacts on freshwater ecosystems have been the object of intense research 490 worldwide in recent decades, and advanced knowledge has been obtained at several 491 physicochemical and biological levels (Kelly 1988; ERMITE-Consortium 2004). In this sense, 492 multiple studies have demonstrated the negative ecological impacts of such activities on benthic 493 communities through multiple lines of evidence: bioaccumulation, toxicity, or community 494 structure changes, both at an international scale (Clements et al. 2002; Maret et al. 2003; 495 Smolders et al. 2003; Gray & Delaney 2008; Rainbow et al. 2012; Svitok et al. 2014) and at 496 regional scales, in particular, in Northern Spain (García-Criado et al. 1999; García-Criado & Aláez 497 2001; Marqués et al. 2001b, 2003). However, assessments of metal impacts on 498 macroinvertebrate communities where the considered reference condition satisfy the rigorous 499 WFD selection criteria are still scarce. The present study provides new information on the level 500 of metal pollution impairment due to mining activities in the Nalón River basin, and on the 501 effects that the increasing heavy metal levels in sediments have on the structure of 502 macroinvertebrate communities. This information constitutes a crucial database for the future 503 development of environmental quality standards (EQSs) for river sediments, since EU 504 environmental legislation on water policy highlights the relevance of the biotic components for 505 defining environmentally relevant sediment EQSs. 506 19 The application of a WFD-compliant ‘reference to test ratios’, that is an RCA design, to 507 the evaluation of sediment metal concentrations, allowed us to establish a group of metals (As, 508 Cd, Cu, Pb, Se and Hg) that significantly differ from the reference condition in the different 509 mining impacted groups of sites in the Nalón River basin. Moreover, PCoA analyses identified 5 510 (As, Cd, Cu, Se and Hg) of the mentioned metals as the main contributors to the sites ordination 511 according to environmental stressors. Although impacted sites with metal pollution also seem 512 to be affected by a slight habitat disturbance assessed by IHF according to PCoA, the fact is that 513 this result is only a reflection of the great conservation status of the selected reference sites. 514 However, it doesn´t imply that the level of habitat alteration in the test sites reach such a high 515 degree to alter also benthic communities to a great extent. In general, according to Pardo et al. 516 (2002) IHF values over 50 can properly sustain a diverse macroinvertebrate community in 517 Mediterranean rivers. Besides, although there is no official index assessment for the Cantabrian 518 area, rivers located in other Northern Spain areas with same river typologies as those of the 519 present study, established an IHF value of 66 as the reference condition and 60 as the limit 520 between high/good status (RD 1/2016). Only 6 of our study sites showed values under 60, from 521 which only 2 were under 50. Thus, indicating that hydromorphological alterations may not be a 522 key stressor for benthic communities and that metal pollution represents the main 523 environmental stressor within our study area. 524 Results also proved the existence of significant responses of macroinvertebrate 525 assemblages to the specific mining conditions existing within the Nalón River basin where, as 526 said, one kind of stressor dominates, i.e., metal pollution. In our study, univariate tests were not 527 totally efficient in detecting mining impacts over macroinvertebrate communities, when 528 analyses were performed at the group level (i.e., REF, CU, HG and AU mining activity). Most 529 generic metrics (i.e., general richness, diversity indices, total abundance) seemed not to be 530 sensitive to community changes that occur under metal stress, when considering this approach. 531 These results are in accordance with previous studies indicating failure in the application of 532 univariate tests to some traditional community indices for detecting mining ecological impacts 533 (see for example Byrne et al. 2013). However, in our case it should be considered that this 534 absence of response could be attributed to the unbalanced sites design (REF n= 14, HG n=8, AU 535 n=4, CU n=3) preventing the achievement of sufficient statistical power to detect the 536 differences, or perhaps due to the existence of some uncontaminated sites in the HG and CU 537 groups masking the effects. In fact, when gradient analyses were applied, as in other studies (Qu 538 et al. 2010), significant associations were found between the mentioned community descriptors 539 and the heavy metal levels in sediments (both considering the metals at individual level and the 540 additive effect through SedPoll). Similarly, multivariate modelling in combination with TITAN 541 20 was efficient in detecting sediment metal pollution as a relevant stressor for biotic communities 542 in the study area, even in the presence of other environmental stressors (Burton & Johnston 543 2010). In general, significant responses both at a global level of change in the macroinvertebrate 544 community structure (general composition, abundance and diversity), alterations in different 545 sensitive taxa metrics (i.e., EPT richness and abundance) and regarding the individual 546 sensitivities of specific taxa (mainly EPT families) were found. These results indicate that 547 sediment metal pollution due mining activities clearly impact macroinvertebrate communities, 548 although this fact is highly masked through group analyses. This fact highlights the need for using 549 multiple lines of evidence to identify reliable changes in macroinvertebrate community 550 structures within mining areas. 551 Identifying reliable indicator taxa and their responses to mining pressure is of major 552 concern for the development of management tools in water quality policy. In this study, the 553 statistical treatment based on indicator species (IndVal) performed through TITAN was 554 satisfactory for detecting the sensitivity levels of indicator taxa to the considered metal 555 gradients. Although it was not possible to establish change thresholds at the community level, 556 TITAN allowed us to determine a group of macroinvertebrate families specifically sensitive to 557 metal concentrations in the sediments measured as SedPoll scores. Nine taxa (seven EPT plus 558 two Coleoptera) followed a sensitivity gradient, from the Sericostomatidae with maximum 559 sensitivity to the Baetidae with the lowest. In general, these results were in accord with many 560 other studies showing most of the mentioned taxa as sensitive to metal pollution (Hirst et al. 561 2002; Solà et al. 2004; Pond et al. 2008), except for two cases that were reported as tolerant in 562 the prior literature, such as the Elmidae (Clements et al. 2000; Qu et al. 2010) and 563 Hydropsychidae (Clements et al. 2000; Luoma et al. 2010). On the other hand, the mentioned 564 TITAN results for the sediment pollution scores (SedPoll) were highly coincident with those 565 reported for the PCoA gradients in terms of the sensitive indicators detected (Fig. 4), suggesting 566 that the drivers of the effects detected by TITAN for the PCoA gradients were probably due to 567 the metal concentrations in the sediments, rather than the other environmental variables 568 loading both PCo1 and PCo2. These results support the hypothesis that metal pollution plays a 569 fundamental role in structuring macroinvertebrate communities in the Nalón mining areas and 570 propose a set of indicator taxa suitable for future studies on field toxicity and bioaccumulation. 571 The presence of resistant taxa should be further explored as a possible factor preventing the 572 observation of greater impacts within the macroinvertebrate communities in this area, since the 573 rapid acquisition of resistance to metal contamination in sediments has been documented (e.g., 574 for aquatic oligochaetes, by Klerks & Levinton 1989). Although indicator taxa results do not allow 575 clear conclusions on this sense, it should be noted that 2 of the 3 taxa considered positively 576 21 associated with metal stress in TITAN analyses, are surface dwelling Hemiptera that may not be 577 exposed to metal pollution to the same extent than other taxa in close contact with the 578 contaminated sediments. In fact, surface-dwelling Hemiptera usually tended to be classified as 579 metal-tolerant (Gerhardti et al. 2004). On the other hand, the use of individual taxa change 580 points and their QI in elaborating preventive thresholds for metals with the potential to impact 581 structural taxa at different levels (see Pardo & García 2016) can be the subject of deeper 582 research to improve conservation policies. 583 From an applied perspective our findings are valuable as they represent a first step in 584 the development of specific tools to assess the impact of metal stress in the area, not only by 585 giving a set of indicator taxa for sediment metal pollution, but also by providing a reliable 586 measure of metal stress in the area through a WFD compliant RCA evaluation (i.e., SedPoll). In 587 other hand from a WFD perspective, our results are consistent with the current ecological status 588 classification systems used in rivers of Northern Spain, the multimetric index METI (METI 2015) 589 and the predictive model NORTI (Pardo et al., 2014). As the main community descriptors 590 affected by metal pollution in our study (i.e, composition, abundance and EPT metrics) represent 591 actually key parameters within these classification systems. This fact, may be indicative that the 592 mentioned classification systems could be already sensitive to community alterations due to 593 metal stress. Indeed, the evaluation of this aspect represents the next step of our research. In 594 addition, the selected taxonomic resolution level (i.e., the family level in most instances) used 595 in our study was very useful in detecting ecological impairments related to mining pressure 596 within the Nalón River basin, in accordance with the results of other studies (Clements et al. 597 2000; Marqués et al. 2003; Pond et al. 2008; Wright & Ryan 2016). Aspect that is also relevant, 598 as the mentioned classification systems for Northern Spain work with this level of resolution. 599 General conclusions 600 In our study, we prove that part of the macroinvertebrate communities (sensitive taxa) 601 could be considered as reliable indicators of metal pollution in Cu, Hg and Au-mining areas in 602 Northern Spain. As the community as a whole did not show a clear threshold response to mining 603 pressure, specific metrics based on sensitive taxa were successfully responsive to the observed 604 mining gradients. The identification of those specific taxa providing a reliable signal of the overall 605 metal stress measured through SedPoll scores (calculated within the framework of a WFD-606 compliant ‘reference-test sites’ design), emerges as a first valuable step in order to develop 607 specific tools for evaluating impact of metal stress in the context of the WFD directive. Thus, the 608 applied methodology and results are relevant for determining the importance of heavy metal 609 contaminated sediments as stressors for biotic communities in the studied area. 610 22 The discussed relationships between the communities’ structure, sensitive metrics and 611 individual taxa with existing metal gradients in the study area can be used by water authorities 612 to assess the level of metal impacts within their monitoring schedule. Similarly, it can help to 613 establish either the levels of degradation or the recovery levels to be achieved, helping to 614 support the decisions to be taken and the programme of measures to be implemented in the 615 water management of mining areas. 616 Acknowledgements 617 This investigation was supported by the research project CGL2013-44655-R, sponsored by the 618 Spanish Government, Ministry of Economy and Competitiveness (MINECO). Dr. Méndez-619 Fernández was financed by a postdoctoral position at the University of the Basque Country 620 (UPV/EHU, Spain). We greatly appreciate the support provided by Amanda Miranda and the staff 621 of the Cantabrian Hydrographical Confederation, who assisted in the selection and sampling of 622 the sites in the Nalón River basin. The authors thank Claudio Padilla, Odei Barredo and Iñigo 623 Moreno for the intense days of help during the field surveys. Liliana García is also acknowledged 624 for statistical advice. 625 References 626 Águeda-Villar JA, Salvador-González CI (2008). Mineralizaciones de Plomo-Zinc y Hierro del 627 Urgoniano de la cuenca Vasco-Cantábrica. In: García-Cortes A, Agueda Villar JA, Palacio 628 Suárez-Valgrande J, Salvador González CI (eds.), Contextos Geológicos españoles. Una 629 aproximación al patrimonio geológico español de relevancia internacional. IGME, Madrid. 630 Anderson MJ (2001) A new method for non-parametric multivariate analysis of variance. Austral 631 Ecol 26: 32-46 632 Anderson MJ, Gorley RN, Clarke KR (2008) PERMANOVA+ for PRIMER: Guide to Software and 633 Statistical Methods. PRIMER-E Ltd., Plymouth, UK. 634 Baker ME, King RS (2010) A new method for identifying and interpreting biodiversity and 635 ecological community thresholds. Methods Ecol Evol 1: 25-37 636 Baker ME, King RS (2013) Of TITAN and straw men: an appeal for greater understanding of 637 community data. Freshw Sci 32 (2): 489-506 638 Barbour MT, Gerritsen J, Snyder BD, Stribling JB (1999) Rapid Bioassessment Protocols for Use 639 in Streams and Wadeable Rivers: Periphyton, Benthic Macroinvertebrates and Fish, 2nd ed. 640 EPA 841-B-99-002, U.S. Environmental Protection Agency, Office of Water, Washington, DC. 641 Bervoets L, De Jonge M, Blust (2016) Identification of threshold body burdens of metals for the 642 protection of the aquatic ecological status using two benthic invertebrates. Environ Poll 210: 643 76-84 644 23 Birk S, Bonne W, Borja A, Brucet S, Courrat A, Poikane S, Solimini A, van de Bund W, Zampoukas 645 N, Hering D (2012) Three hundred ways to assess Europe’ surface waters: an almost complete 646 overview of biological methods to implement the Water Framework Directive. Ecol Indic 18: 647 31-41 648 Bowman MF, Bailey RC (1997) Does taxonomic resolution affect the multivariate description of 649 the structure of freshwater benthic macroinvertebrate communities? Can J Fish Aquat Sci 54: 650 1802-1807 651 Bray RJ, Curtis JT (1957) An ordination of the upland forest communities of southern Wisconsin. 652 Ecol Monogr 27: 325-349 653 Bryan GW, Langston WJ, Hummer-Stone LG, Burt GR (1985) A guide to the assessment of heavy 654 metal contamination in estuaries using biological indicators. J Mar Biol Ass UK, Occasional 655 Publication No. 4. 656 Burton GA, Johnston EL (2010) Assessing contaminated sediments in the context of multiple 657 stressors. Environ Toxicol Chem 29 (12): 2625-2643 658 Byrne P, Reid I, Wood PJ (2013) Changes in macroinvertebrate community structure provide 659 evidence of neutral mine drainage impacts. Environ Sci: Processes Impacts 15: 393-404 660 Clarke KR, Warwick RM (2001) Change in Marine Communities: an Approach to CLARKE KR, 661 GORLEY RN 2006. Primer v6: user manual/tutorial. Primer-E, Plymouth, United Kingdom. 662 Statistical Analysis and Interpretation, 2nd edn. Plymouth: PRIMER-E. 663 Clarke KR, Gorley RN (2006) PRIMER v6: User Manual/Tutorial. PRIMER-E Ltd., Plymouth, UK 664 Clements WH, Carlisle DM, Lazorchack JM, Johson PC (2000) Heavy metals structure benthic 665 communities in Colorado Mountain streams. Ecol Appl 10: 626-638 666 Clements WH, Carlisle DM, Courtney LA. Harrahy EA (2002) Integrating observation and 667 experimental approaches to demonstrate causation in stream biomonitoring studies. Environ 668 Toxicol Chem 21 (6): 1138-1146 669 De Jonge M, Tipping E, Lofts S, Bervoets L, Blust R (2013) The use of invertebrate body burdens 670 to predict ecological effects of metal mixtures in mining-impacted waters. Aquat Toxicol 142-671 143: 294-302 672 Doi H, Takagi A, Kikuchi E (2007) Stream macroinvertebrate community affected by point-source 673 metal pollution. Internat Rev Hydrobiol 92 (3): 258-266 674 Dudgeon D, Arthington AH, Gessner MO, Kawabata A-I., Knowler DJ, Lévêque C, Naiman,RJ, 675 Prieur-Richard A-H, Soto D, Stiassny MLJ, Sullivan CA (2006) Freshwater biodiversity: 676 importance, threats, status and conservation challenges. Biol Rev 81: 163-182 677 Dufrêne M, Legendre P (1997) Species assemblages and indicator species: the need for a flexible 678 asymmetrical approach. Ecol Monogr 67: 345-366 679 24 ERMITE-Consortium (2004). Mining impacts on the fresh water environment: Technical and 680 managerial guidelines for catchment scale management. Younger PL, Wolkersdorfer C (eds) 681 Mine Water and the Environment 23: S2-S80. 682 EC, European Commission (2000) Directive 2000/60/EC. Establishing a framework for 683 community action in the field of water policy. Off J Eur Commun L 327: 1-71 684 EC, European Commission (2013) Directive 2013/39/EU of the European Parliament and of the 685 Council as regards priority substances in the field of water policy. Offic J Eur Union L226, 17 686 pp. 687 García L, Pardo I, Delgado C (2014) Macroinvertebrate indicators of ecological status in 688 Mediterranean temporary stream types of the Balearic Islands. Ecol Indic 45: 650-663 689 García-Criado F, Fernández-Aláez C, Fernández-Aláez M (1999). Environmental variables 690 influencing the distribution of Hydraenidae and Elmidae assemblages (Coleoptera) in a 691 moderately-polluted river basin in north-western Spain. Eur J Entomol 96: 37-44 692 García-Criado F, Fernández Aláez M (2001) Hydraenidae and Elmidae assemblages (Coleoptera) 693 from Spanish river basin: good indicators of coal mining pollution? Arch Hydrobiol 150: 641-694 660 695 Gerhardt A, Janssens de Bisthoven L, Soares AMVM (2004) Macroinvertebrate response to acid 696 mine drainage: community metrics and on-line behavioural toxicity bioassay. Environ Poll 697 130: 263-274 698 Guerold F (2000) Influence of taxonomic determination level on several community indices. 699 Water Res 34: 487-492 700 Gray NF, Delaney E (2008) Comparison of benthic macroinvertebrate indices for the assessment 701 of the impact of acid mine drainage on an Irish river below an abandoned Cu-S mine. Environ 702 Poll 155: 31-40 703 Hirst H, Jüttner I, Ormerod SJ (2002) Comparing the responses of diatoms and 704 macroinvertebrates to metals in upland streams of Wales and Cornwall. Freshwater Biol 705 47:1752-1765 706 Kelly M (1988) Mining and the Freshwater Environment, Elsevier, New York 707 Klerks PL, Levinton JS (1989). Rapid evolution of metal resistance in a benthic oligochaete 708 inhabiting a metal-polluted site. Biol Bull 176: 135-141 709 Lenat DR, Barbour MT (1994) ‘Using Benthic Macroinvertebrate Community Structure for Rapid, 710 Cost-Effective, Water Quality Monitoring: Rapid Bioassessment’, in Loeb SL and Spacie A 711 (eds), Monitoring Aquatic Systems, Lewis, Boca Raton, pp. 187–215 712 Lenat DR, Resh VH (2001) Taxonomy and stream ecology—the benefits of genus-and species-713 level identifications. J N Am Benthol Soc 20:287–298 714 25 Loredo J, Ordoñez A, Álvarez R (2006) Environmental impact of toxic metals and metalloids from 715 the Muñón Cimero mercury-mining area (Asturias, Spain). J Hazard Mater 136: 455-467 716 Loredo J, Petit-Domínguez MD, Ordoñez A, Galán MP, Fernández-Martínez R, Alvarez R, 717 Rucandio MI (2010) Surface water monitoring in the mercury mining district of Asturias 718 (Spain). J Hazard Mater 176: 323-332 719 Luoma SN, Cain DJ, Rainbow PS (2010) Calibrating biomonitors to ecological disturbance: a new 720 technique for explaining metal effects in natural waters. Integr Environ Assess Manag 6 (2): 721 199-209 722 Malmqvist B, Rundle S (2002) Threats to the running water ecosystems of the world. Environ 723 Conserv 29: 134-153 724 Maret TR, Cain DJ, MacCoy DE, Short TM (2003) Response of benthic invertebrate assemblages 725 to metal exposure and bioaccumulation associated with hard-rock mining in northwestern 726 streams, USA. J N Am Benthol Soc 22 (4): 598-620 727 Marqués MJ, Martínez-Conde E, Rovira JV, Ordóñez S (2001a) Heavy metals pollution of aquatic 728 ecosystems in the vicinity of a recently closed underground lead-zinc mine (Basque Country, 729 Spain). Environ Geol 40: 1125-1137 730 Marqués MJ, Martínez-Conde E, Rovira JV (2001b). Los Macroinvertebrados Como Índices de 731 Evaluación Rápida de Ecosistemas Acuáticos Contaminados Por Metales Pesados. Ecotox 732 Environ Restor 4 (1): 25-31 733 Marqués MJ, Martínez-Conde E, Rovira JV (2003). Effects of zinc and lead mining on the benthic 734 macroinvertebrates of a fluvial ecosystem. Water Air Soil Poll 148: 363-388 735 Martin CW (2000) Heavy metal trends in floodplain sediments and valley fill, River Lahn, 736 Germany. Catena 39: 53-68 737 Méndez-Fernández L, Rodríguez P, Martínez-Madrid M (2015). Sediment toxicity and 738 biaccumulation assessment in abandoned copper and mercury mining áreas of the Nalón 739 River basin (Spain). Arch Environ Contam Toxicol 68 (1): 107-123 740 Méndez-Fernández L, Martinez-Madrid M, Pardo I, Rodriguez P (2017) Baseline tissue 741 concentrations of metal in aquatic oligochaetes: Field and laboratory approaches. Environ 742 Poll 223: 636-643 743 Metcalfe JL (1989). Biological water quality assessment of running waters based on 744 macroinvertebrate communities: History and present status in Europe. Environ Poll 60 (1-2): 745 101-139 746 METI (2015) “Protocolo de cálculo del índice multimétrico específico del tipo de invertebrados 747 bentónicos en ríos”. Spanish Government, Ministry of Agriculture, Food and Environment 748 (ed), 14pp 749