Towards stressor-specific macroinvertebrate indices

Science of the Total Environment 619–620 (2018) 144–154

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Towards stressor-specific macroinvertebrate indices: Which traits and taxonomic groups are associated with vulnerable and tolerant taxa? Elisabeth Berger a,b,d,⁎, Peter Haase a,c, Ralf B. Schäfer d, Andrea Sundermann a,b a

Senckenberg Research Institute and Natural History Museum Frankfurt, Department of River Ecology and Conservation, Gelnhausen, Germany Goethe University Frankfurt am Main, Faculty of Biological Sciences, Department Aquatic Ecotoxicology, Frankfurt am Main, Germany University of Duisburg-Essen, Faculty of Biology, Department of River and Floodplain Ecology, Essen, Germany d University Koblenz-Landau, Institute for Environmental Sciences, Department of Quantitative Landscape Ecology, Landau, Germany b c

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Macroinvertebrate abundances changed along water quality gradients. • Taxa responses showed little specificity towards different gradients. • Traits associated with decreasing and increasing taxa were determined. • Assessment of traits improved mechanistic understanding of taxa responses.

a r t i c l e

i n f o

Article history: Received 29 August 2017 Received in revised form 2 November 2017 Accepted 2 November 2017 Available online xxxx Editor: D. Barcelo Keywords: Trait-based biomonitoring Micropollutants Wastewater Ecotoxicology Multiple stressors

a b s t r a c t Monitoring of macroinvertebrate communities is frequently used to define the ecological health status of rivers. Ideally, biomonitoring should also give an indication on the major stressors acting on the macroinvertebrate communities supporting the selection of appropriate management measures. However, most indices are affected by more than one stressor. Biological traits (e.g. size, generation time, reproduction) could potentially lead to more stressor-specific indices. However, such an approach has rarely been tested. In this study we classify 324 macroinvertebrate taxa as vulnerable (decreasing abundances) or tolerant (increasing abundances) along 21 environmental gradients (i.e. nutrients, major ions, oxygen and micropollutants) from 422 monitoring sites in Germany using Threshold Indicator Taxa Analysis (TITAN). Subsequently, we investigate which biological traits and taxonomic groups are associated with taxa classified as vulnerable or tolerant with regard to specific gradients. The response of most taxa towards different gradients was similar and especially high for correlated gradients. Traits associated with vulnerable taxa across most gradients included: larval aquatic life stages, isolated cemented eggs, reproductive cycle per year b1, scrapers, aerial and aquatic active dispersal and plastron respiration. Traits associated with tolerant taxa included: adult aquatic life stages, polyvoltinism, ovoviviparity or egg clutches in vegetation, food preference for dead animals or living microinvertebrates, substrate preference for macrophytes, microphytes, silt or mud and a body size N2–4 cm. Our results question whether stressor-specific indices based on macroinvertebrate assemblages can be achieved using single traits, because we observed that similar taxa responded to different gradients and also similar traits

⁎ Corresponding author at: Senckenberg Research Institute and Natural History Museum Frankfurt, Department of River Ecology and Conservation, Gelnhausen, Germany. E-mail addresses: [email protected], [email protected] (E. Berger).

https://doi.org/10.1016/j.scitotenv.2017.11.022 0048-9697/© 2017 Elsevier B.V. All rights reserved.

E. Berger et al. / Science of the Total Environment 619–620 (2018) 144–154

145

were associated with vulnerable and tolerant taxa across a variety of water quality gradients. Future studies should examine whether combinations of traits focusing on specific taxonomic groups achieve higher stressor specificity. © 2017 Elsevier B.V. All rights reserved.

1. Introduction Biomonitoring of freshwater invertebrates is widely used to gauge and track changes in the environment and to define the ecological state or health of a biological system (Cairns and Pratt, 1993; Friberg et al., 2011). The major advantage is that the biological state is measured directly, instead of inferring perturbations based on chemical measures or other potential pressures (Extence and Ferguson, 1989). Thus, ecological health assessment based on biomonitoring forms the basis for environmental management decisions in Europe (European Parliament and Council, 2000). The ecological health status of rivers is mainly derived from indices that compare the occurrence, abundance, community composition or richness of – for example all or specific macroinvertebrate taxa – to some sort of expected or reference condition (Armitage et al., 1983; Birk et al., 2012; Böhmer et al., 2004b; Extence and Ferguson, 1989). Although these taxonomy-based indices provide good evidence of general degradation, they are often limited in their ability to indicate which stressor is causing the degradation (Böhmer et al., 2004a; Schäfer et al., 2011). Multiple stressors acting on a single freshwater community is the prevalent situation in Europe (Schäfer et al., 2016). Therefore, applied freshwater scientists and managers are interested in obtaining more stressor-specific indices in order to enable identification of the causative stressor whether that is toxicants, nutrient enrichment or poor habitat quality (Baird et al., 2008; Boxall et al., 2012; Culp et al., 2011; Rubach et al., 2011). This knowledge would allow the selection of the most (cost-)effective water management measure including chemical regulatory decisions. Trait-based biomonitoring approaches have been suggested to achieve this aim (Menezes et al., 2010; Statzner and Bêche, 2010). Traits are “well-defined, measurable properties of organisms usually measured at the individual level and used comparatively across species” (McGill et al., 2006). They can represent adaptations to local habitat conditions (Southwood, 1977) such as body form, respiration strategy and locomotion type (Townsend and Hildrew, 1994; Tullos et al., 2009; Verberk et al., 2008). Therefore, traits may improve a mechanistic understanding of cause-effect relationships by integrating ecological theory (i.e. the habitat templet concept) into biomonitoring (Bonada et al., 2007; Statzner and Bêche, 2010) and thus indicating the stressor(s) responsible for the biological impairment (Dolédec and Statzner, 2008; Mondy et al., 2016; Mondy and Usseglio-Polatera, 2013; Rubach et al., 2011). For example, the SPEcies At Risk (SPEAR) index was developed following this approach and incorporates physiological (relative pesticide sensitivity) and biological traits that are assumed to increase the vulnerability of taxa to pesticides (generation time ≥ 0.5 per year, low migration ability, and presence of aquatic stages during the time of maximum exposure to pesticides, Liess and von der Ohe, 2005). This index successfully discriminated between reference and pesticide-contaminated sites in different European biogeographic regions and was shown to react specific to pesticides exposure in small agricultural streams (Schäfer et al., 2007). However, other studies reported that the index also responded to habitat quality and other stressors (Mondy et al., 2012; Rasmussen et al., 2012). Shared sensitivity of taxa to a variety of stressors and the fact that many stressors co-occur in the environment (Schäfer et al., 2016) can explain why such indices respond to more than one environmental gradient. Thus, the development of truly stressor-specific indices based on the assessment of invertebrate communities is a major challenge regardless whether the trait-profile or the taxonomic composition is assessed.

This study was designed to assess the feasibility and support the development of stressor-specific indices based on macroinvertebrates. We identify and compare taxon-specific responses to different water quality gradients and explore which traits are associated with vulnerable and tolerant taxa with respect to these different gradients. Two main questions are addressed: 1) Can we identify taxa that respond differently (with increasing or decreasing abundances) to different water quality gradients (e.g. increasing phosphorus or nitrite concentrations) in the field? 2) Which biological traits and which taxonomic groups are associated with the vulnerable and tolerant taxa and are they stressorspecific? We address the first question by applying Threshold Indicator Taxa Analysis (TITAN, Baker and King, 2010) to identify taxa that respond with an abrupt decrease (defined as vulnerable taxa) or increase in abundances (tolerant taxa) along 20 water quality gradients including gradients of major ions (e.g. sodium), physico-chemical parameters (e.g. oxygen), nutrients (e.g. nitrate) and common wastewater associated organic micropollutants (e.g. DEET). Since TITAN is a univariate approach, where the effect of co-variables cannot be assessed, we applied TITAN also to a catchment size gradient (smaller to larger streams) as an “external” co-variable. The second question is addressed by applying indicator species analysis (Dufrêne and Legendre, 1997) to trait affinity scores associated with the group of taxa identified as vulnerable or tolerant through TITAN. Based on hypotheses from previous studies (Liess and von der Ohe, 2005; Mondy et al., 2016; Mondy and Usseglio-Polatera, 2013; Statzner and Bêche, 2010), we expect to see several traits as indicative of taxa occurring at clean or impaired sites, respectively. Following the recommendation by Lange et al. (2014), we do, however, not limit the test for relationships to these a priori hypothesised associations. The primary aim was to identify taxa and traits in an explorative approach that may help to develop more stressor-specific trait-based biomonitoring tools. In addition, these taxa and traits may help to target the selection of new model organisms for chemical risk assessment experiments (Rubach et al., 2011, 2012). 2. Methods 2.1. Monitoring data We used monitoring data provided by the Saxon State Agency for Environment, Agriculture and Geology (LfULG) containing information on macroinvertebrate abundances and a complete set of 20 water quality variables from 422 sites across Saxony, Germany. Following the rationale described in Berger et al. (2016, 2017), only the most recent macroinvertebrate samples (taken 2008–2014) for a given site collected between February and July (recommended sampling season) was included in the analysis. Kick sampling was used to sample macroinvertebrates according to their relative coverage across 20 sub-habitats in a 100 m-long stream section using a hand net (25 cm × 25 cm, Haase et al., 2004). The organisms are then pooled (total sample area of 1.25 m2), counted in the laboratory and identified down to a taxonomic level as specified in the ‘Operational Taxalist for Running Water in Germany’ (Haase et al., 2006). In total, 324 taxa with occurrence frequency N 3 were identified across the 422 sites (see Supplementary information Table S1 for taxa identities). Sites covered a range of stream types with catchment areas between 1.5 and 55,590 km2 (see Table 1). The considered environmental variables included major ions (MI), nutrients (NU), micropollutants (MP) and other physicochemical conditions (PC; see Table 1 for the number of measurements and detection frequencies) as well as the catchment size as a proxy of river size. All

146


variables were measured at each site. To aggregate replicate measurements at a site, we calculated the 90th percentile and the 10th percentile in case of oxygen, reflecting the probability of encountering high chemical concentrations or low oxygen concentrations, respectively. Chemical data from the same and two previous years of the macroinvertebrate sample were included in the percentile calculation to achieve higher numbers of chemical measurements while guaranteeing close temporal proximity between biological and chemical samples. The between-site variability of the percentile concentrations is summarized in Table 1.

(group 0). Thus, 21 separate TITAN runs were carried out with different environmental gradients (see Table 1) using the same 422 sampling sites and the same 324 taxa (occurrence frequency N 3). Elaborate permutation and bootstrapping techniques embedded in TITAN enable the filtering out of taxon responses that are distinct and abrupt from those that are hidden by noise from natural fluctuations (King and Baker, 2014). TITAN is a combination of indicator species analysis (Dufrêne and Legendre, 1997) and change point analysis. For each taxon, TITAN seeks an optimum observed change point that maximizes the indicator value (IndVal score) or its standardized IndVal score (z-core; standardization with the average of 250 permuted IndVal scores and the corresponding standard deviations) along a specific environmental gradient (e.g. increasing nitrite concentrations). Depending on the relative abundance and occurrence frequency of the taxon on either side of the change point, it is assigned to a negative (group 1 = vulnerable) or positive (group 2 = tolerant) response group. The rationale is that vulnerable taxa will escape or die due to adverse environmental conditions, while tolerant taxa may increase in abundance and colonize more sites due to the disappearance of vulnerable taxa increasing available niches. For all considered water quality gradients, increasing concentrations are deemed unfavourable (potential nutrient limitation was considered negligible), except for oxygen where decreasing concentrations are unfavourable and tolerant taxa increased (group 2) with decreasing oxygen. Each analysis was repeated 500 times in a bootstrap procedure to characterize the uncertainty of the observed change points. Based on these replicates, the quality criteria “reliability” and “purity” are calculated. Reliability describes the consistency of achieving high z-scores and consequently significant (p ≤ 0.05) change points. Purity specifies the percentage of bootstrap replicates that associate a taxon with the same response group. Taxa failing to meet these quality criteria (purity ≥ 0.95 and reliability ≥ 0.95) were assigned to the group of non-responding taxa (group 0). The resulting taxon response matrix of 21 environmental variables i (20 water quality variables plus catchment area) against 324 taxa j filled with the classification values C (0 = no clear response; 1 = vulnerable; 2 = tolerant) was analysed regarding the variability in the response to the gradients. We calculated for each taxon j the number of occasions A

2.2. Trait data Trait data was obtained from Tachet et al. (2000) with some modifications made by the authors in 2003/2004 and was available for 281 of the 324 taxa included in the analysis (Section 2.3.1). Trait information was matched to the identified taxa if provided at the same or a lower taxonomic level. Conversely, trait information specified at a higher taxonomic level (e.g. genus) was not assigned to a taxon identified at lower level (e.g. family). Given that this applied only to a few taxa, they were omitted from further analysis. Twelve traits related to the morphology, life-history, mobility and ecology of the taxa were considered (Poff et al., 2006). Each trait is divided into different modalities (e.g. different size classes of the trait “size”) resulting in a total of 72 trait modalities (Table S1, Mondy and Usseglio-Polatera, 2014). An affinity score of each taxon to the different trait modalities is provided by a value between 0 (no affinity) and 3 or 5 (high affinity) from a fuzzy coding procedure (Chevenet et al., 1994). 2.3. Statistical analysis All statistical analyses were performed using R 3.3.2 (R Core Team, 2016). 2.3.1. Threshold Indicator Taxa Analysis Threshold Indicator Taxa Analysis (TITAN, R-package: TITAN2) was used to classify the response of 324 taxa towards 21 environmental gradients as vulnerable (group 1), tolerant (group 2) or non-responding

Table 1 The percentage of taxa with detectable valid change points (% res. taxa) with decreasing (% vulnerable) or increasing (% tolerant) abundances and occurrences along increasing concentrations of 20 water quality variables and catchment area. Ranges of the 90th percentile concentrations (minimum, mean, median and maximum) of these variables across 422 sites, their units, number of individual measurements (No. Meas.) and the percentage of measurements N limit of quantification (LOQ). Variable group

Variable

res. taxa

% vulnerable

% tolerant

min

Mean

Median

max

Unit

No. Meas.

% N LOQ

PC MI NU MI PC PC PC NU MI NU MI MI MP NU MP MP MP NU NU MP

Oxygena Potassium Sulphate Calcium Temperature Hardness Conductivity Ammonium Sodium Nitrite Chloride Magnesium DEET Nitrate Caffeine HHCB TCEP Phosphate Silicate TiBP Area

59.6 58.3 54.3 52.5 52.5 52.2 51.5 47.5 46.3 44.8 44.1 42.6 38.3 31.8 31.5 30.6 28.7 28.4 27.5 26.5 34.0

30.2 25.3 25.9 26.2 24.1 25.9 25.9 25.6 21.6 23.5 21.6 22.8 19.4 15.7 18.5 15.7 14.2 13.6 13.6 15.4 13.6

29.3 33.0 28.4 26.2 28.4 26.2 25.6 21.9 24.7 21.3 22.5 19.8 18.8 16.0 13.0 14.8 14.5 14.8 13.9 11.1 20.4

1.86 0.4 11 2.6 9.4 0.09 44 0.01 2.38 0.0025 2 0.6 0 0.23 0 0 0 0.005 1.7 0 1.5

8.40 6.95 120.37 65.85 15.87 2.20 577.43 0.38 29.06 0.06 47.56 13.75 26.68 6.72 268.33 39.06 17.84 0.11 6.81 25.78 1638.60

8.82 6.26 82 51.95 15.71 1.787 513.2 0.203 22.95 0.0429 39.2 12 10.625 6.68 158.7 19.65 6 0.07 6.63 4.145 22.0

10.62 48.6 1070 315 22.3 9.74 2326 8.03 240 0.474 561 58 492.2 24 4789 701 545.5 1.21 14.7 1440 55,590.2

mg/L mg/L mg/L mg/L °C mmol/l μS/cm mg/L mg/L mg/L mg/L mg/L ng/L mg/L ng/L ng/L ng/L mg/L mg/L ng/L km2

8748 7431 8260 7431 8755 7039 8757 8256 7431 8256 8260 7431 3236 8262 3236 3231 3241 8223 8115 3231 –

100 100 100 100 100 100 100 85 100 87 100 100 46 100 77 59 32 78 100 34 –

Note: PC, Physicochemical variables; MP, wastewater associated micropollutants; NU, nutrients; MI, major ions, DEET (CAS: 134-62-3); caffeine (CAS: 58-08-2); HHCB (CAS: 1222-05-5); TCEP (CAS: 115-96-8); TiBP (CAS: 126-71-6). a Here the relationship is reversed and tolerant taxa increase with decreasing oxygen concentrations.


that it was classified as a responding taxa and a fidelity index (F) as follows: Pn Fj ¼

i¼1

C i; j

Aj

Thus, taxa with F value of 1 or 2 were exclusively vulnerable or tolerant, respectively. Moreover, the similarity of taxa response profiles across gradients was assessed using agglomerative hierarchical clustering with average linkage (as compromise between the single and complete linkage methods) on the distance matrix (simple matching, Sokal and Michener, 1958). Note that the choice of the linkage method did not influence our results as single and complete linkages lead to a similar clustering. Similar taxon responses to different stressors might, beside similar sensitivities to the stressors, also be driven by the co-occurrence of stressors. Therefore, we visualized the correlation structure of the environmental gradients using a Spearman correlation heatmap. In addition to a visual comparison between similarity of the response profiles for the different variables and the degree of correlation between variables, we quantified this relationship. First a similarity matrix of the variables (1 – simple matching distance coefficient) based on the taxon response classifications was produced that could then be compared to the correlation matrix of the environmental variables using a mantel test (999 permutations and Pearson correlations, R-package: vegan). 2.3.2. Indicator ‘trait’ analysis We performed indicator species analysis (Dufrêne and Legendre, 1997) to calculate the indicator value IndVal of biological trait modalities and taxonomic groups i for one of two groups j (vulnerable or tolerant taxa as defined through TITAN) for each variable as follows: IndValij ¼ 100 Aij Bij Aij is the proportion of the affinity scores of trait modality or taxonomic group i that are in group j. Bij is the proportion of taxa in group j that have trait modality or belong to taxonomic group i. We used 999 permutations to calculate the statistical significance of the indicator values (R package: labdsv) and only indicator values with a significance level p ≤ 0.05 are reported in the results. Given that we were interested in the indicator value of a trait (or taxonomic group) as opposed to a species, our analysis may more intuitively be referred to as indicator ‘trait’ analysis. Taxonomic groups were included to ease the interpretation of indicator traits, since certain traits can be associated with specific taxonomic groups only. We omitted the nonresponding taxa as third group in the indicator ‘trait’ analysis, because they could not be considered a homogeneous biological group (i.e. being tolerant taxa, whose abundances are not influenced by the considered environmental gradient only). Instead, there may be many reasons why TITAN was unable to classify these taxa as vulnerable or tolerant including very gradual changes in abundance along the gradient or low occurrence frequencies (rare taxa). Therefore, we did not expect specific traits to be associated with this group.

147

variables for most taxa (Fig. 1). Oxygen had the largest number of responding taxa (59.6% of the 324 analysed taxa) and the flameretardant TiBP the lowest (26.5%, Table 1). The correlation structure of these variables is shown in Fig. 2. Clustering of variables based on taxa responses showed high similarity for oxygen, conductivity, hardness, temperature, major ions, nitrite, sulphate and ammonium (Fig. 3), which were also characterized by high numbers of responding taxa (N40%). Caffeine, HHCB, DEET, TCEP, TiBP and phosphate clustered differently, however responding taxa were mostly assigned to the same response group than for the former variables (Table S1), but they had overall lower numbers of responding taxa (26.5–38.3%). Catchment area, nitrate and silicate had more dissimilar taxon response profiles compared to the other variables with several taxa responding conversely (Fig. 3, Table S1). Catchment area and silicate concentrations were also least correlated to any other environmental variable (Fig. 2). Conductivity, water hardness, major ions and oxygen were highly correlated to each other and poorly correlated with wastewater associated compounds, phosphate and ammonium. Thus, correlation structure of the environmental variables represented well the similarity of the taxon response profiles, which was also evidenced by the mantel test statistic (r = 0.56, significance b 0.01). 3.2. Traits associated with vulnerable and tolerant taxa Traits and taxonomic groups with a statistically significant indicator value for taxa identified as vulnerable or tolerant are shown in Table 2. Several traits, such as larval aquatic life stages, reproduction with isolated cemented eggs, aerial active dispersal, eggs or statoblasts as resistance forms and plastron respiration (Table 2), were consistently associated with the group of vulnerable taxa across most variables, though their indicator strength varied. Note that the IndVal method hampers the achievement of high scores for rare traits (low occurrence frequency), because the absence of that trait in a taxon is not informative. However, if rare traits achieve statistically significant IndVal scores, they must have been associated rather exclusively with one group making them potentially very good indicator traits for vulnerability/tolerance if present in the taxon. Traits that were consistently associated with tolerant taxa included adult aquatic life stages, ovoviviparity, food preference for dead animals or living microinvertebrates, piercing feeding habit and substrate preferendum for macrophytes, microphytes or mud. Other traits were associated with certain stressor-groups only; a large size N2–4 cm, surface swimmers, burrowers, substrate preference for silt and egg clutches in vegetation were associated with taxa tolerant to higher salinities and nutrient concentrations and low oxygen concentrations. By contrast, temporary attachment, aquatic and aerial passive dispersal and high

3. Results 3.1. Taxon response group classifications The TITAN response group classifications (vulnerable = 1, tolerant = 2, non-responding = 0) for each taxon and variable are provided in the Supplementary information (Table S1), while several summary statistics are provided here. 14.8% of the taxa were non-responding (failing the required quality criteria) in response to all 21 gradients. The number of classifications N0 (A) for the remaining taxa varied between 1 and 21, whereby the response direction was the same across the different

Fig. 1. Density distribution of fidelity index (F) for all taxa with at least one classification N0 (n = 276) across the 21 gradients.

148


Fig. 2. Heatmap of the Spearman correlation coefficients between 21 environmental gradients (n = 422).

numbers of reproductive cycles per year were predominantly associated with taxa tolerant to high concentrations of wastewater associated compounds. Several traits achieved significant indicator values for just one environmental variable such as the substrate preference sand for

the silicate and diapause for the ammonium gradient. However, such single values need to be interpreted with care until further studies are available. Only very few traits showed varied responses, i.e. were associated with tolerant taxa for one and vulnerable taxa for another stressor. Such traits included life cycle duration N 1 year, food preference for living microphytes and a substrate preference for twigs/roots, organic detritus/litter or flags/boulders/cobbles/pebbles. 3.3. Taxonomic groups associated with tolerant and vulnerable taxa The indicator values of taxonomic groups were generally lower than those for traits, which is due to generally lower occurrence frequency (i.e. each taxon can only belong to one taxonomic group). Plecoptera consistently indicated vulnerable taxa, and Gastropoda and Hirudinea tolerant taxa. The association of Ephemeroptera, Trichoptera and Coleoptera with the group of vulnerable taxa and the association of bivalves and Odonata with tolerant taxa was less consistent, showing large within group variability (Table 2, Table S1). 4. Discussion 4.1. Taxon-stressor specificity

Fig. 3. Agglomerative hierarchical clustering of the distances (simple matching) between 21 environmental variables (rows) based on taxon response group classifications (324 columns or taxa).

We found that most taxa are vulnerable or tolerant towards a variety of water quality stressors, where vulnerability or tolerance means an abrupt change in abundance and occurrence along an environmental gradient. We observed that the taxa responded similarly especially to highly correlated variables. This highlights the importance to carefully interpret associations between a single stressor and a biological response in the presence of intercorrelations with


149

Table 2 Trait modalities and taxonomic groups associated with decreasing taxa (−) or increasing taxa (+) as determined with TITAN (Threshold Indicator Taxa Analysis) for 21 environmental variables. Mean IndVal ± SD is the mean indicator value ± standard deviation across the 21 variables. Trait

Trait modality

Frequency Area Oxygen* Temperature Hardness Conductivity Potassium Calcium Sodium Chloride

Size

N0.5–1 cm N2–4 cm N4–8 cm N1 year b1 1 N1 Larva Nymph Adult Ovoviviparity Isolated eggs, cem. Clutches, free Clutches, in vegetation Clutches, terrestrial Aquatic passive Aquatic active Aerial passive Aerial active Eggs, statoblasts Diapause or dormancy None Plastron Surface swimmer Full water swimmer Crawler Burrower Interstitial Temporarily attached Detritus (b1 mm) Dead plant (≥1 mm) Living microphytes Living macrophytes Dead animal (≥1 mm) Living microinvert. Vertebrates Deposit feeder Shredder Scraper Filter-feeder Piercer Parasite Flags/boulders/cobbles/pebbles Gravel Sand Silt Macrophytes Microphytes Twigs/roots Organic detritus/litter Mud Bivalvia Coleoptera Crustacea Ephemeroptera Gastropoda Hirudinea Odonata Plecoptera Trichoptera

164 82 16 156 60 262 137 258 109 92 29 72 44 15 52 245 270 148 220 95 120 208 19 25 105 277 65 85 86 179 140 214 109 61 114 8 69 153 140 65 26 7 244 208 182 111 247 50 156 197 127 8 36 9 46 20 10 9 13 72

Life cycle duration Potential no. of cycles/year

Aquatic stages

Reproduction

Dispersal

Resistance forms

Respiration Locomotion/substrate relation

Food

Feeding habits

Substrate preferendum

Taxonomic groups

+

− +

+

+

− − + + −

+

+

+

+ −

−

−

+ −

−

−

−

−

+ −

−

+ + −

+ + −

+ + −

+ + −

+ + −

+ + −

+ + −

+

+

+

+

−

− −

− + − −

−

−

−

−

−

− −

− −

− −

− −

− −

− −

+

− +

− +

+

−

−

− + −

−

−

−

−

+ + +

+ +

+ +

+

+ +

− + +

−

−

−

−

+

+

+

−

−

−

+ + −

+ + +

+ + +

+ + +

+ + +

+ + +

+ +

+

+

+

+ +

+ − + +

− + +

+

− +

+ − + − +

− −

− +

+ −

−

+

+ + +

+ +

+ − + + − −

−

+

−

+ + + − + + +

− + + + − −

other stressors. Although association does not mean causation in general, this holds especially for environmental data sets with strong correlations. This issue also applies to Threshold Indicator Taxa Analysis (TITAN). The same response of a taxon towards several stressors can be due to a shared sensitivity of the taxon towards these stressors or can be due to the correlation of these stressors and consequently represent a spurious relationship. A standard procedure to address this issue is to cluster different gradients into fewer uncorrelated gradients prior to analysis using for example PCA (principal component analysis). However, we opted for a different approach

+ + −

−

− −

+ + + − −

+

+ +

+ +

+ +

+

+

− + +

+ +

+ +

−

−

−

− −

here, analysing correlated gradients separately and then comparing the similarity of results and the correlation between different gradients retrospectively. In this way no information is lost during clustering. Much of the similarities in the response profiles of different gradients appeared to be due to their intercorrelation. However, irrespective of the cause of the same response, these results support previous studies highlighting the challenge to identify the causative or most important stressor at sites affected by multiple stressors based on the assessment of macroinvertebrate indices (Böhmer et al., 2004a).

150

Magnesium +


Sulphate − + + +

Ammonium + +

Nitrite

Nitrate

Phosphate

+

−

−

−

−

−

+ + −

+ + −

+ + −

+ + −

+

+

Silicate

DEET

Caffeine

+ + −

− −

−

−

+

+ −

+

+

− +

+ + −

+ + −

+ + −

+

+ +

− +

+

−

− −

−

−

−

−

− −

− −

− − +

− −

−

− +

− +

− +

− +

−

+ −

TCEP

−

+

+

HHCB

TIBP

− +

+

− + −

+ −

−

+ −

−

− +

− + −

−

+

+

−

+

+

− − + + +

+

− + +

+

+ +

+ + + +

+

+ +

− −

−

−

+

+ +

−

−

+ + +

+ + +

+ + +

+

−

+ + + −

+

+ + +

+ +

+ +

+ + +

+ +

+ +

+

+ +

+ +

+

+ +

+

− +

− +

+ +

+ +

+ +

+

+

+

+

+ +

+

+

−

−

−

− + + + −

− + + + −

− + + + −

+ +

+

+ +

−

−

−

Few taxa responded to only one single water quality gradient. Although TITAN applies stringent bootstrap and permutation procedures to prevent detection of false relationships between taxa and environmental gradients, we do not recommend the use or promote these taxa as truly stressor-specific indicator species without further supporting evidence. TITAN results vary between different runs and single responses may still be the result of chance or selective inference, given that the type 1 error increases when testing many univariate associations (here 324 taxa × 21 variables = 6802 associations, Taylor and Tibshirani, 2015). Moreover, although the presence of a vulnerable

−

+

−

−

Mean IndVal ± SD 0.42 ± 0.06 0.28 ± 0.03 0.08 ± 0.00 0.39 ± 0.00 0.27 ± 0.04 0.56 0.44 ± 0.04 0.52 ± 0.01 0.28 0.35 ± 0.05 0.2 ± 0.02 0.33 ± 0.02 0.17 0.09 ± 0.02 0.19 ± 0.05 0.54 ± 0.02 0.54 ± 0.01 0.45 ± 0.04 0.55 ± 0.02 0.32 ± 0.03 0.32 0.47 0.10 ± 0.01 0.13 ± 0.02 0.31 0.54 ± 0.01 0.22 ± 0.04 0.27 ± 0.01 0.33 ± 0.07 0.44 ± 0.04 0.44 0.50 ± 0.02 0.37 ± 0.03 0.25 ± 0.03 0.32 ± 0.02 0.07 ± 0.02 0.24 0.42 ± 0.03 0.38 ± 0.02 0.24 ± 0.02 0.13 ± 0.02 0.09 ± 0.02 0.56 ± 0.02 0.48 ± 0.01 0.53 0.37 ± 0.02 0.57 ± 0.02 0.23 ± 0.04 0.44 ± 0.10 0.50 ± 0.04 0.48 ± 0.10 0.06 ± 0.03 0.17 ± 0.04 0.07 ± 0.02 0.16 ± 0.08 0.14 ± 0.07 0.12 ± 0.06 0.08 ± 0.04 0.16 ± 0.03 0.22 ± 0.08

taxon may indicate the absence of a certain stressor; the absence of an individual taxon has previously been shown to be a very poor predictor for a specific problem (Leps et al., 2016).

4.2. Variable importance The different number of taxa responding to the 21 variables may indicate variable importance, although this was beyond the focus of this study. Nevertheless, comparisons are particularly valid because all


analyses were based on the same data. The high number of responding taxa across the water quality variable group of increased temperature, related reduced oxygen, increases in major ions and salinity, suggest rather pronounced changes and taxon turnover for these gradients. By contrast, nitrate and silicate, which formed distinctively different gradients, appeared less important. Also, less taxa responded to wastewater associated compounds. However, waste water associated compounds were measured less frequently and showed lower detection frequencies, which can influence TITAN results (Berger et al., 2016). It is also possible that the gradients in our study were insufficiently wide to cover the concentration ranges where biological effects occur for all of the taxa. Pronounced changes in taxa abundances along the oxygen and conductivity gradient were observed, even though the measured concentrations were rather low and mostly below current environmental quality standards. This may demand reconsideration of these standards to lower levels (see also Berger et al., 2017; Sundermann et al., 2015). However, looking at the traits that were associated with vulnerable taxa (i.e. interstitial taxa and a substrate preference for flags/boulders/cobbles/pebbles) and tolerant taxa (i.e. surface swimmers and burrowers having a substrate preference for silt, macrophytes, microphytes and mud) along these gradients, may hint towards clogging of the interstitial and accumulating mud as a consequence of reduced flow (that can be correlated to the water quality gradients) to be the real cause of species turn-over (Villeneuve et al., 2017). This shows that – as expected – investigating traits in routine biomonitoring samples can provide new information and enhance mechanistic understanding of taxa responses allowing more in-depth interpretations of biological changes. Water quality gradients were relatively independent (i.e. poor correlation) from the catchment size gradient and taxa responded stronger to the water quality gradients than to the catchment area gradient. We therefore conclude that the natural stream size gradient was of minor importance in the study area. However, it has to be noted that the nature of the dataset may have impeded the detection of an effect as only few larger streams were included. 4.3. Indicator traits Given that mostly the same taxa responded to the different gradients, we expected similar traits associated with the related vulnerable or tolerant taxa. Indeed, traits were generally associated with multiple water quality gradients. Although some traits appeared to be more strongly associated with the stressor group of oxygen, temperature, major ions, nitrite and salinity and other traits with the group of wastewater associated compounds and phosphate; the differences were small and including fewer taxa in the latter group may have caused an artefact. We considered univariate associations – the relationship between one trait and one stressor – only. However, Mondy et al. (2016) recently reported to have achieved stressor-specific indices through a combination of traits in complex models (Mondy et al., 2016). Therefore, the subtle differences that we reported may encourage further attempts to develop stressor-specific indices through a combination of traits. However, we focus the discussion on the general traits associated with vulnerable and tolerant taxa, because differences between variables were small. Furthermore, of the 72 trait modalities, 51 achieved a statistically significant indicator value. Possibly higher or more indicator values could have been achieved, if the taxa groups (1 and 2) on which the indicator trait analysis was based could have been further separated into weak and strong increasers/decreasers, respectively. However, such a separation would have been rather arbitrary and lead to groups too small to allow for statistical analysis. 4.3.1. Trait intercorrelations The testing for univariate associations between gradients and traits, as done in this study can be influenced by so called ‘trait syndromes’.

151

Traits can be intercorrelated across taxa because of phylogenetic constraints (Pilière et al., 2016). In other words, traits may not be adaptive (Gould and Lewontin, 1979) or may be adaptive only as part of a life history strategy, but not independently from other traits. Hence, the response of one trait to a stressor may not be causal but an indirect effect the stressor has on an intercorrelated response trait (Poff et al., 2006; Statzner and Bêche, 2010). For example, contrary to our expectation that plastron respiration should be associated with taxa tolerant to nutrients and low oxygen (Mondy et al., 2016), we found that plastron respiration consistently indicated taxa vulnerable towards these stressors. Similarly, Piscart et al. (2006) observed decreased proportion of plastron respiration in invertebrate communities with increased salinity (Piscart et al., 2006). In our study plastron respiration was limited to beetle species, which tended to be associated with vulnerable taxa. A potential explanation could be that for example elmid beetles obtain their oxygen bubble after nymphosis and then breath dissolved oxygen with the plastron working as a physical gill (Elliott, 2008). If the plastron is not replenished with atmospheric oxygen through resurfacing, the plastron would provide no advantage in low oxygen waters. Alternatively, induced surfacing could make them more prone to predation. However, there may also have been other traits related to beetles that may have caused the association. Usseglio-Polatera et al. (2000) suggested that ecological traits are possibly more independent from phylogenetic constraints (i.e. labile traits) than biological traits such as morphology (including respiration) and life-history. If this applies we would expect to see ecological traits as having higher indicator values for generally vulnerable or tolerant taxa. However, we did not observe such a pattern and the question regarding the degree of lability of traits for freshwater invertebrates is still a matter of debate (Poff et al., 2006). Therefore, we here account for potentially spurious relationships by considering ecological plausibility in the following compilation of traits associated with vulnerable and tolerant taxa. 4.3.2. Trait responses confounded by taxonomy We observed that a larger body size was associated with tolerant taxa as also found in previous studies (Lange et al., 2014; Rico and Van den Brink, 2015). Thus, a larger body size could be interpreted as an advantage for low water quality. However, this may also be a spurious relationship, because gastropods are generally large and were generally identified as tolerant. Similarly, active aerial dispersal was associated with vulnerable taxa and this trait was also observed to decrease in invertebrate communities with increasing pesticide contamination (Mondy and Usseglio-Polatera, 2013). However, we suggest that this association is possibly spurious, because high dispersal ability is typically associated with enhanced recolonization potential and thus increased resilience in unfavourable environments (Li et al., 2016; Winking et al., 2014). The potentially spurious relationship may be explained by aerial active dispersal being found in sensitive insect taxa as opposed to Gastropoda and Hirudinea, which were exclusively identified as tolerant taxa with the exception of the freshwater limpit Ancylus fluviatilis. The insect taxa Ephemeroptera (E), Plecoptera (P), Trichoptera (T) and Coleoptera (C) are widely acknowledged to be among the most sensitive taxonomic groups of macroinvertebrates. Thus, the relative abundance of EPT or EPTC taxa is often used in ecological health assessments of rivers. Although all Plecopterans were exclusively identified as vulnerable, taxa within the Ephemeroptera, Trichoptera as well as Coleoptera were divided into non-responding, vulnerable and tolerant taxa. Dipterans were dominated by tolerant taxa; however, some species for example of the genera Simulium sp. were identified as vulnerable, whereas others of the same genera as tolerant (see Supplementary Information Table S1). This shows that the variability in sensitivities can be rather high within a taxonomic group and even within genera, stressing the importance of taxonomic identification to low, favourably species level (Lenat and Resh, 2001). This also suggests that species within a genus have different traits that allow some to thrive and others to perish. Trait-based methods have been suggested

152


to be potentially more time- and cost-effective than taxonomic methods, because taxonomic identification at a coarser level would be sufficient to describe trait diversity. However, our results suggest that trait-based approaches are likely to equally require identification of taxa to low taxonomic levels and in addition knowledge on traits at low taxonomic levels. A considerable step forward would be the identification of traits that confer these differences within a group or genera. Unfortunately, trait information, for example for the dipterans, is currently available only at the family level for most taxa, hampering such identifications (but see Serra et al., 2016). Pilière et al. (2016) showed that some taxa with the same trait profile as determined by selforganizing maps (SOM), but belonging to different taxonomic groups showed different responses along an environmental gradient, suggesting some important differences that were not captured by available trait information. Conversely, Ephemeroptera that differed in their trait profiles (one group containing mainly mobile taxa and the other group mostly clinging herbivores) differed in their response to an agricultural land-use gradient. In accordance with these results, we confirm that large taxonomic groups already have an indicator value for environmental impact (i.e. EPT taxa), but that responses are confounded by significant differences within groups. Therefore, we suggest that trait-based biomonitoring approaches should not rely solely on traits, but also take phylogeny into account by for example grouping coarse taxonomic information (Ephemeroptera, Diptera etc.) with traits in a combined index. Machine-learning approaches (SOM, boosted regression trees) as applied by Pilière et al. (2016) could be used to detect such combinations (e.g. Ephemeroptera that are also clinging herbivores). 4.3.3. Traits with plausible cause-effect link Several associations that were found in our study may be interpreted as having a direct cause-effect link. For instance adult aquatic life stages were consistently associated with tolerant taxa and larval aquatic stages with vulnerable taxa. This association was also a priori predicted and confirmed by Mondy and Usseglio-Polatera (2013) in response to pesticide contamination. They observed a strong linear relationship between a pesticide gradient and the community-weighted proportion of these traits (Mondy and Usseglio-Polatera, 2013). The relationship is rather intuitive considering that juvenile stages are generally more vulnerable. Polyvoltinism was associated with tolerant taxa, which is in line with ecological theory (high recolonization potential after disturbance, more generations allowing faster adaptations) and it is a trait that has been found to respond to a high number of anthropogenic pressures (Mondy et al., 2012). Ovoviviparity consistently indicated tolerant taxa and increased frequency of ovoviviparous taxa in invertebrate assemblages were previously associated with disturbed environments (Archaimbault et al., 2010; Dolédec et al., 1999; Lange et al., 2014; Piscart et al., 2006) although contrasting results have also been obtained (Ieromina et al., 2016). Ovoviviparity may be a parental care strategy for unfavourable conditions: the sensitive egg stages are protected inside the mother's body (Díaz et al., 2008). Taxa with ovoviviparity in our study included several bivalves (Sphaerium sp. and Pisidium sp.), the snail Potamopyrgus antipodarum and crustaceans (Gammarus sp., Asellus sp.) and a mayfly (Cloeon sp.). By contrast, production of isolated eggs was associated with vulnerable taxa, which is in agreement with previous results (Lange et al., 2014). It has been suggested that the greater surface area compared to clutches increases the permeability for toxicants (Díaz et al., 2008). A food preference for dead animals and living macroinvertebrates was associated with tolerant taxa. Potentially, these traits may confer tolerance through an energy rich diet (Schäfer et al., 2011). 5. Conclusions Trait-based indices still face several challenges, especially with respect to true stressor-specificity. To date, they may not be more cost-

effective and stressor specific than taxonomy based approaches per se. However, we showed that analysis of traits increased a mechanistic understanding of cause-effect relationships. Moreover, applicability of trait-based approaches across biogeographical regions appears feasible, because traits associated to vulnerable and tolerant taxa in our study (Germany) matched those identified in previous studies (i.e. France, New Zealand and Netherlands, Ieromina et al., 2016; Lange et al., 2014; Mondy and Usseglio-Polatera, 2013), even though we used very different methods than those previous studies for trait identification. Large-scale applicability would allow comparison of ecological status across countries and the expansion of biomonitoring to so far unmonitored regions (Friberg et al., 2011; Statzner and Bêche, 2010). Large scale applicability and increased mechanistic understanding lead us to conclude that it is worthwhile to incorporate traits into biomonitoring, even though stressor-specificity requires further scrutiny. However, analysis of European wide or even globally available macroinvertebrate biomonitoring data coupled to traits and environmental variables may allow to uncover stressor-specific trait-taxonomy combinations through machine-learning approaches in the future (Pilière et al., 2016). Moreover, understanding of trait interrelationships, trait lability and eco-evolutionary processes could be improved. Since information on invertebrate traits is still basic compared to plant trait data bases (Kattge et al., 2011), improving available trait information including also physiological traits and at lower taxonomic levels could also allow the detection of meaningful and specific relationships to environmental conditions. Thus, further studies should investigate the fundamentals of traitinterrelationships and their spatial distributions over larger areas with independent gradients to potentially achieve higher stressor-specificity. Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2017.11.022.

Acknowledgements We thank the Saxon State Agency for Environment, Agriculture and Geology (LfULG) and the Saxon State Company for Environment and Agriculture (BfUL) for the collection and provision of all chemical and macroinvertebrate data used in this study. Elisabeth Berger gratefully acknowledges funding by a non-profitable publishing group (FAZITSTIFTUNG Gemeinnützige Verlagsgesellschaft mbH, Frankfurt). The foundation took no influence on the content of the manuscript. References Archaimbault, V., Usseglio-Polatera, P., Garric, J., Wasson, J.-G., Babut, M., 2010. Assessing pollution of toxic sediment in streams using bio-ecological traits of benthic macroinvertebrates. Freshw. Biol. 55:1430–1446. https://doi.org/10.1111/j.13652427.2009.02281.x. Armitage, P.D., Moss, D., Wright, J.F., Furse, M.T., 1983. The performance of a new biological water quality score system based on macroinvertebrates over a wide range of unpolluted running-water sites. Water Res. 17:333–347. https://doi.org/10.1016/00431354(83)90188-4. Baird, D.J., Rubach, M.N., Van den Brink, P.J., 2008. Trait-based ecological risk assessment (TERA): the new frontier? Integr. Environ. Assess. Manag. 4, 2–3. Baker, M.E., King, R.S., 2010. A new method for detecting and interpreting biodiversity and ecological community thresholds. Methods Ecol. Evol. 1:25–37. https://doi.org/ 10.1111/j.2041-210X.2009.00007.x. Berger, E., Haase, P., Oetken, M., Sundermann, A., 2016. Field data reveal low critical chemical concentrations for river benthic invertebrates. Sci. Total Environ. 544: 864–873. https://doi.org/10.1016/j.scitotenv.2015.12.006. Berger, E., Haase, P., Kuemmerlen, M., Leps, M., Schäfer, R.B., Sundermann, A., 2017. Water quality variables and pollution sources shaping stream macroinvertebrate communities. Sci. Total Environ. 587–588:1–10. https://doi.org/10.1016/ j.scitotenv.2017.02.031. Birk, S., Bonne, W., Borja, A., Brucet, S., Courrat, A., Poikane, S., Solimini, A., van de Bund, W., Zampoukas, N., Hering, D., 2012. Three hundred ways to assess Europe's surface waters: an almost complete overview of biological methods to implement the Water Framework Directive. Ecol. Indic. 18:31–41. https://doi.org/10.1016/ j.ecolind.2011.10.009. Böhmer, J., Rawer-Jost, C., Zenker, A., 2004a. Multimetric assessment of data provided by water managers from Germany: assessment of several different types of stressors with macrozoobenthos communities. Hydrobiologia 516:215–228. https://doi.org/ 10.1023/B:HYDR.0000025267.58196.5f.

E. Berger et al. / Science of the Total Environment 619–620 (2018) 144–154 Böhmer, J., Rawer-Jost, C., Zenker, A., Meier, C., Feld, C.K., Biss, R., Hering, D., 2004b. Assessing streams in Germany with benthic invertebrates: development of a multimetric invertebrate based assessment system. Limnol. - Ecol. Manag. Inland Waters. New Methods for Assessing Freshwaters in Germany 34:pp. 416–432. https://doi.org/10.1016/S0075-9511(04)80010-0. Bonada, N., Dolédec, S., Statzner, B., 2007. Taxonomic and biological trait differences of stream macroinvertebrate communities between mediterranean and temperate regions: implications for future climatic scenarios. Glob. Chang. Biol. 13:1658–1671. https://doi.org/10.1111/j.1365-2486.2007.01375.x. Boxall, A.B.A., Rudd, M.A., Brooks, B.W., Caldwell, D.J., Choi, K., Hickmann, S., Innes, E., Ostapyk, K., Staveley, J.P., Verslycke, T., Ankley, G.T., Beazley, K.F., Belanger, S.E., Berninger, J.P., Carriquiriborde, P., Coors, A., Deleo, P.C., Dyer, S.D., Ericson, J.F., Gagné, F., Giesy, J.P., Gouin, T., Hallstrom, L., Karlsson, M.V., Larsson, D.G.J., Lazorchak, J.M., Mastrocco, F., McLaughlin, A., McMaster, M.E., Meyerhoff, R.D., Moore, R., Parrott, J.L., Snape, J.R., Murray-Smith, R., Servos, M.R., Sibley, P.K., Straub, J.O., Szabo, N.D., Topp, E., Tetreault, G.R., Trudeau, V.L., Van Der Kraak, G., 2012. Pharmaceuticals and personal care products in the environment: what are the big questions? Environ. Health Perspect. 120:1221–1229. https://doi.org/10.1289/ehp.1104477. Cairns Jr., J., Pratt, J.R., 1993. A history of biological monitoring using benthic macroinvertebrates. In: Rosenberg, D.M., Resh, V.H. (Eds.), Freshwater Biomonitoring and Benthic Macroinvertebrates. Chapman & Hall, New York, pp. 10–27. Chevenet, F., Dolédec, S., Chessel, D., 1994. A fuzzy coding approach for the analysis of long-term ecological data. Freshw. Biol. 31:295–309. https://doi.org/10.1111/j.13652427.1994.tb01742.x. Culp, J.M., Armanini, D.G., Dunbar, M.J., Orlofske, J.M., Poff, N.L., Pollard, A.I., Yates, A.G., Hose, G.C., 2011. Incorporating traits in aquatic biomonitoring to enhance causal diagnosis and prediction. Integr. Environ. Assess. Manag. 7:187–197. https://doi.org/ 10.1002/ieam.128. Díaz, A.M., Alonso, M.L.S., Gutiérrez, M.R.V.-A., 2008. Biological traits of stream macroinvertebrates from a semi-arid catchment: patterns along complex environmental gradients. Freshw. Biol. 53:1–21. https://doi.org/10.1111/j.1365-2427.2007.01854.x. Dolédec, S., Statzner, B., 2008. Invertebrate traits for the biomonitoring of large European rivers: an assessment of specific types of human impact. Freshw. Biol. 53:617–634. https://doi.org/10.1111/j.1365-2427.2007.01924.x. Dolédec, S., Statzner, B., Bournard, M., 1999. Species traits for future biomonitoring across ecoregions: patterns along a human-impacted river. Freshw. Biol. 42:737–758. https://doi.org/10.1046/j.1365-2427.1999.00509.x. Dufrêne, M., Legendre, P., 1997. Species assemblages and indicator species: the need for a flexible asymmetrical approach. Ecol. Monogr. 67:345–366. https://doi.org/10.1890/ 0012-9615(1997)067[0345:SAAIST]2.0.CO;2. Elliott, J.M., 2008. The ecology of riffle beetles (Coleoptera: Elmidae). Fr. Rev. 1:189–203. https://doi.org/10.1608/FRJ-1.2.4. European Parliament and Council, 2000. Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for community action in the field of water policy. Off. J. L 327, 1–73 (22.12.2000). Extence, C.A., Ferguson, A.J.D., 1989. Aquatic invertebrate surveys as a water quality management tool in the Anglian Water region. Regul. Rivers Res. Manag. 4:139–146. https://doi.org/10.1002/rrr.3450040206. Friberg, N., Bonada, N., Bradley, D.C., Dunbar, M.J., Edwards, F.K., Grey, J., Hayes, R.B., Hildrew, A.G., Lamouroux, N., Trimmer, M., Woodward, G., 2011. Biomonitoring of human impacts in freshwater ecosystems: the good, the bad and the ugly. Advances in Ecological Research. Elsevier:pp. 1–68 https://doi.org/10.1016/B978-0-12-3747945.00001-8. Gould, S.J., Lewontin, R.C., 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proc. R. Soc. Lond. B Biol. Sci. 205, 581–598. Haase, P., Lohse, S., Pauls, S., Schindehütte, K., Sundermann, A., Rolauffs, P., Hering, D., 2004. Assessing streams in Germany with benthic invertebrates: development of a practical standardised protocol for macroinvertebrate sampling and sorting. Limnol. - Ecol. Manag. Inland Waters. New Methods for Assessing Freshwaters in Germany 34:pp. 349–365. https://doi.org/10.1016/S0075-9511(04)80005-7. Haase, P., Schindehütte, K., Sundermann, A., 2006. Fließgewässerbewertung - Download Bestimmung [WWW Document]. Oper. Taxaliste Als Mindestanforderung Bestimm. Von Makrozoobenthosproben Aus Fließgewässern Zur Umsetzung EUWasserrahmenrichtlinie Dtschl. URL. http://www.fliessgewaesserbewertung.de/ download/bestimmung/, Accessed date: 15 July 2015. Ieromina, O., Musters, C.J.M., Bodegom, P.M., Peijnenburg, W.J.G.M., Vijver, M.G., 2016. Trait modality distribution of aquatic macrofauna communities as explained by pesticides and water chemistry. Ecotoxicol. Lond. Engl. 25:1170–1180. https://doi.org/ 10.1007/s10646-016-1671-5. Kattge, J., Díaz, S., Lavorel, S., Prentice, I.C., Leadley, P., Bönisch, G., Garnier, E., Westoby, M., Reich, P.B., Wright, I.J., Cornelissen, J.H.C., Violle, C., Harrison, S.P., Van Bodegom, P.M., Reichstein, M., Enquist, B.J., Soudzilovskaia, N.A., Ackerly, D.D., Anand, M., Atkin, O., Bahn, M., Baker, T.R., Baldocchi, D., Bekker, R., Blanco, C.C., Blonder, B., Bond, W.J., Bradstock, R., Bunker, D.E., Casanoves, F., Cavender-Bares, J., Chambers, J.Q., Chapin Iii, F.S., Chave, J., Coomes, D., Cornwell, W.K., Craine, J.M., Dobrin, B.H., Duarte, L., Durka, W., Elser, J., Esser, G., Estiarte, M., Fagan, W.F., Fang, J., Fernández-Méndez, F., Fidelis, A., Finegan, B., Flores, O., Ford, H., Frank, D., Freschet, G.T., Fyllas, N.M., Gallagher, R.V., Green, W.A., Gutierrez, A.G., Hickler, T., Higgins, S.I., Hodgson, J.G., Jalili, A., Jansen, S., Joly, C.A., Kerkhoff, A.J., Kirkup, D., Kitajima, K., Kleyer, M., Klotz, S., Knops, J.M.H., Kramer, K., Kühn, I., Kurokawa, H., Laughlin, D., Lee, T.D., Leishman, M., Lens, F., Lenz, T., Lewis, S.L., Lloyd, J., Llusià, J., Louault, F., Ma, S., Mahecha, M.D., Manning, P., Massad, T., Medlyn, B.E., Messier, J., Moles, A.T., Müller, S.C., Nadrowski, K., Naeem, S., Niinemets, Ü., Nöllert, S., Nüske, A., Ogaya, R., Oleksyn, J., Onipchenko, V.G., Onoda, Y., Ordoñez, J., Overbeck, G., Ozinga, W.A., Patiño, S., Paula, S., Pausas, J.G., Peñuelas, J., Phillips, O.L., Pillar, V., Poorter, H., Poorter, L., Poschlod, P., Prinzing, A., Proulx, R., Rammig, A., Reinsch, S., Reu, B., Sack, L., Salgado-Negret, B., Sardans, J., Shiodera, S., Shipley, B., Siefert, A., Sosinski, E., Soussana, J.-F., Swaine, E., Swenson, N., Thompson, K., Thornton, P., Waldram, M., Weiher, E., White, M., White, S., Wright, S.J., Yguel, B., Zaehle, S., Zanne, A.E., Wirth,

153

C., 2011. TRY – a global database of plant traits. Glob. Chang. Biol. 17:2905–2935. https://doi.org/10.1111/j.1365-2486.2011.02451.x. King, R.S., Baker, M.E., 2014. Use, misuse, and limitations of Threshold Indicator Taxa Analysis (TITAN) for natural resource management. Application of Threshold Concepts in Natural Resource Decision Making. Springer, New York, NY:pp. 231–254 https://doi.org/10.1007/978-1-4899-8041-0_11. Lange, K., Townsend, C.R., Matthaei, C.D., 2014. Can biological traits of stream invertebrates help disentangle the effects of multiple stressors in an agricultural catchment? Freshw. Biol. 59:2431–2446. https://doi.org/10.1111/fwb.12437. Lenat, D.R., Resh, V.H., 2001. Taxonomy and stream ecology—the benefits of genus- and species-level identifications. J. N. Am. Benthol. Soc. 20:287–298. https://doi.org/ 10.2307/1468323. Leps, M., Leisner, S., Haase, P., Sundermann, A., 2016. Prediction of taxon occurrence: a test on taxon-specific change point values of stream benthic invertebrates. Freshw. Biol. 61:1773–1786. https://doi.org/10.1111/fwb.12817. Li, F., Stoll, S., Sundermann, A., Haase, P., 2016. Dispersal capacity shapes the early succession of benthic invertebrates in restored rivers. Sci. Total Environ. 569–570, 1570–1578. Liess, M., von der Ohe, P.C., 2005. Analyzing effects of pesticides on invertebrate communities in streams. Environ. Toxicol. Chem. 24:954–965. https://doi.org/10.1897/03652.1. McGill, B.J., Enquist, B.J., Weiher, E., Westoby, M., 2006. Rebuilding community ecology from functional traits. Trends Ecol. Evol. 21:178–185. https://doi.org/10.1016/ j.tree.2006.02.002. Menezes, S., Baird, D.J., Soares, A.M.V.M., 2010. Beyond taxonomy: a review of macroinvertebrate trait-based community descriptors as tools for freshwater biomonitoring. J. Appl. Ecol. 47:711–719. https://doi.org/10.1111/j.1365-2664.2010.01819.x. Mondy, C.P., Usseglio-Polatera, P., 2013. Using conditional tree forests and life history traits to assess specific risks of stream degradation under multiple pressure scenario. Sci. Total Environ. 461–462:750–760. https://doi.org/10.1016/ j.scitotenv.2013.05.072. Mondy, C.P., Usseglio-Polatera, P., 2014. Using fuzzy-coded traits to elucidate the non-random role of anthropogenic stress in the functional homogenisation of invertebrate assemblages. Freshw. Biol. 59:584–600. https://doi.org/10.1111/ fwb.12289. Mondy, C.P., Villeneuve, B., Archaimbault, V., Usseglio-Polatera, P., 2012. A new macroinvertebrate-based multimetric index (I2M2) to evaluate ecological quality of French wadeable streams fulfilling the WFD demands: a taxonomical and trait approach. Ecol. Indic. 18:452–467. https://doi.org/10.1016/j.ecolind.2011.12.013. Mondy, C.P., Muñoz, I., Dolédec, S., 2016. Life-history strategies constrain invertebrate community tolerance to multiple stressors: a case study in the Ebro basin. Sci. Total Environ. 572:196–206. https://doi.org/10.1016/j.scitotenv.2016.07.227. Pilière, A.F.H., Verberk, W.C.E.P., Gräwe, M., Breure, A.M., Dyer, S.D., Posthuma, L., de Zwart, D., Huijbregts, M.A.J., Schipper, A.M., 2016. On the importance of trait interrelationships for understanding environmental responses of stream macroinvertebrates. Freshw. Biol. 61:181–194. https://doi.org/10.1111/fwb.12690. Piscart, C., Usseglio-Polatera, P., Moreteau, J.-C., Beisel, J.-N., 2006. The role of salinity in the selection of biological traits of freshwater invertebrates. Arch. Hydrobiol. 166: 185–198. https://doi.org/10.1127/0003-9136/2006/0166-0185. Poff, N.L., Olden, J.D., Vieira, N.K.M., Finn, D.S., Simmons, M.P., Kondratieff, B.C., 2006. Functional trait niches of North American lotic insects: traits-based ecological applications in light of phylogenetic relationships. J. N. Am. Benthol. Soc. 25:730–755. https://doi.org/10.1899/0887-3593(2006)025[0730:FTNONA]2.0.CO;2. R Core Team, 2016. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria (URL: http://www.R-project.org/). Rasmussen, J.J., Wiberg-Larsen, P., Baattrup-Pedersen, A., Friberg, N., Kronvang, B., 2012. Stream habitat structure influences macroinvertebrate response to pesticides. Environ. Pollut. 164:142–149. https://doi.org/10.1016/j.envpol.2012.01.007. Rico, A., Van den Brink, P.J., 2015. Evaluating aquatic invertebrate vulnerability to insecticides based on intrinsic sensitivity, biological traits, and toxic mode of action. Environ. Toxicol. Chem. 34:1907–1917. https://doi.org/10.1002/etc.3008. Rubach, M.N., Ashauer, R., Buchwalter, D.B., De Lange, H., Hamer, M., Preuss, T.G., Töpke, K., Maund, S.J., 2011. Framework for traits-based assessment in ecotoxicology. Integr. Environ. Assess. Manag. 7:172–186. https://doi.org/10.1002/ieam.105. Rubach, M.N., Baird, D.J., Boerwinkel, M.-C., Maund, S.J., Roessink, I., den Brink, P.J.V., 2012. Species traits as predictors for intrinsic sensitivity of aquatic invertebrates to the insecticide chlorpyrifos. Ecotoxicology 21:2088–2101. https://doi.org/10.1007/ s10646-012-0962-8. Schäfer, R.B., Caquet, T., Siimes, K., Mueller, R., Lagadic, L., Liess, M., 2007. Effects of pesticides on community structure and ecosystem functions in agricultural streams of three biogeographical regions in Europe. Sci. Total Environ. 382:272–285. https:// doi.org/10.1016/j.scitotenv.2007.04.040. Schäfer, R.B., Kefford, B.J., Metzeling, L., Liess, M., Burgert, S., Marchant, R., Pettigrove, V., Goonan, P., Nugegoda, D., 2011. A trait database of stream invertebrates for the ecological risk assessment of single and combined effects of salinity and pesticides in South-East Australia. Sci. Total Environ. 409:2055–2063. https://doi.org/10.1016/ j.scitotenv.2011.01.053. Schäfer, R.B., Kühn, B., Malaj, E., König, A., Gergs, R., 2016. Contribution of organic toxicants to multiple stress in river ecosystems. Freshw. Biol. https://doi.org/10.1111/ fwb.12811 (n/a-n/a). Serra, S.R.Q., Cobo, F., Graça, M.A.S., Dolédec, S., Feio, M.J., 2016. Synthesising the trait information of European Chironomidae (Insecta: Diptera): towards a new database. Ecol. Indic. 61:282–292. https://doi.org/10.1016/j.ecolind.2015.09.028. Sokal, R., Michener, C., 1958. A statistical method for evaluating systematic relationships. Univ. Kans. Sci. Bull. 28, 1409–1438. Southwood, T.R.E., 1977. Habitat, the templet for ecological strategies? J. Anim. Ecol. 46: 337–365. https://doi.org/10.2307/3817. Statzner, B., Bêche, L.A., 2010. Can biological invertebrate traits resolve effects of multiple stressors on running water ecosystems? Freshw. Biol. 55:80–119. https://doi.org/ 10.1111/j.1365-2427.2009.02369.x.

154


Sundermann, A., Leps, M., Leisner, S., Haase, P., 2015. Taxon-specific physico-chemical change points for stream benthic invertebrates. Ecol. Indic. 57:314–323. https:// doi.org/10.1016/j.ecolind.2015.04.043. Tachet, H., Richoux, P., Bournaud, M., Usseglio-Polatera, P., 2000. Invertébrés d'eau douce: systématique, biologie, écologie. CNRS Editions, Paris. Taylor, J., Tibshirani, R.J., 2015. Statistical learning and selective inference. Proc. Natl. Acad. Sci. 112:7629–7634. https://doi.org/10.1073/pnas.1507583112. Townsend, C.R., Hildrew, A.G., 1994. Species traits in relation to a habitat templet for river systems. Freshw. Biol. 31:265–275. https://doi.org/10.1111/j.13652427.1994.tb01740.x. Tullos, D.D., Penrose, D.L., Jennings, G.D., Cope, W.G., 2009. Analysis of functional traits in reconfigured channels: implications for the bioassessment and disturbance of river restoration. J. N. Am. Benthol. Soc. 28:80–92. https://doi.org/10.1899/07-122.1. Usseglio-Polatera, P., Bournaud, M., Richoux, P., Tachet, H., 2000. Biological and ecological traits of benthic freshwater macroinvertebrates: relationships and definition of

groups with similar traits. Freshw. Biol. 43:175–205. https://doi.org/10.1046/j.13652427.2000.00535.x. Verberk, W.C.E.P., Siepel, H., Esselink, H., 2008. Life-history strategies in freshwater macroinvertebrates. Freshw. Biol. 53:1722–1738. https://doi.org/10.1111/j.13652427.2008.02035.x. Villeneuve, B., Piffady, J., Valette, L., Souchon, Y., Usseglio-Polatera, P., 2017. Direct and indirect effects of multiple stressors on stream invertebrates across watershed, reach and site scales: a structural equation modelling better informing on hydromorphological impacts. Sci. Total Environ. 612:660–671. https:// doi.org/10.1016/j.scitotenv.2017.08.197. Winking, C., Lorenz, A.W., Sures, B., Hering, D., 2014. Recolonisation patterns of benthic invertebrates: a field investigation of restored former sewage channels. Freshw. Biol. 59:1932–1944. https://doi.org/10.1111/fwb.12397.