Science of the Total Environment 639 (2018) 100–109

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Groundwater contamination and land drainage induce divergent

responses in boreal spring ecosystems
Kaisa Lehosmaa a,⁎, Jussi Jyväsjärvi a, Jari Ilmonen b, Pekka M. Rossi c, Lauri Paasivirta d, Timo Muotka a,e
a
University of Oulu, Department of Ecology and Genetics, P.O. Box 3000, FI-90014 Oulu, Finland
b
Metsähallitus, P.O. Box 94, FI-01301 Vantaa, Finland
c
Water Resources and Environmental Engineering Research Group, University of Oulu, P.O. Box 3000, FI-90014, Finland
d
Tahkonkatu 12 as, 9, 24100 Salo, Finland
e
Finnish Environment Institute, Natural Environment Centre, FI-90014 Oulu, Finland

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

• Anthropogenic pressure altered the

structure and functioning of boreal
springs.
• Groundwater contamination reduced
diversity of spring invertebrates and
bacteria.
• Primary production exhibited a subsidy-
stress response to groundwater con-
tamination.
• Ecosystem processes were inhibited by
drainage-induced groundwater
brownification.
• Ecosystem functions should be used in
bioassessment of groundwater-
dependent ecosystems.

a r t i c l e i n f o a b s t r a c t

Article history: Degradation of freshwater ecosystems has engendered legislative mandates for the protection and management
Received 27 February 2018 of surface waters while groundwater-dependent ecosystems (GDEs) have received much less attention. This is so
Received in revised form 30 April 2018 despite biodiversity and functioning of GDEs are currently threatened by several anthropogenic stressors, partic-
Accepted 9 May 2018
ularly intensified land use and groundwater contamination. We assessed the impacts of land drainage (increased
Available online xxxx
input of dissolved organic carbon, DOC, from peatland drainage) and impaired groundwater chemical quality
Editor: José Virgílio Cruz (NO− 3 -N enrichment from agricultural or urban land use) on biodiversity and ecosystem functioning in 20 south-
ern Finnish cold-water springs using several taxonomic and functional measures. Groundwater contamination
Keywords: decreased macroinvertebrate and bacterial diversity and altered their community composition. Changes in mac-
Biodiversity roinvertebrate and bacterial communities along the gradient of water-quality impairment were caused by the re-
Brownification placement of native with new taxa rather than by mere disappearance of some of the original taxa. Also species
Groundwater-dependent ecosystem richness of habitat specialist (but not headwater generalist) bryophytes decreased due to impaired groundwater
Leaf decomposition quality. Periphyton accrual rate showed a subsidy-stress response to elevated nitrate concentrations, with peak
Nitrate enrichment
values at around 2500 μg L−1, while drainage-induced spring water brownification (increased DOC) reduced
Primary production
both periphyton accrual and leaf decomposition rates already at very low concentrations. Our results highlight
the underutilized potential of ecosystem-level functional measures in GDE bioassessment as they seem

⁎ Corresponding author.
E-mail address: kaisa.lehosmaa@oulu.fi (K. Lehosmaa).

https://doi.org/10.1016/j.scitotenv.2018.05.126
0048-9697/© 2018 Elsevier B.V. All rights reserved.
 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109 101

to respond to the first signs of spring ecosystem impairment, at least for the anthropogenic stressors studied by
us.
© 2018 Elsevier B.V. All rights reserved.

1. Introduction from those in near-pristine reference conditions. More specifically, we

expected (i) land drainage to have the strongest impact on macroinver-
Freshwaters are globally among the most threatened ecosystems tebrates (Ilmonen et al., 2009), while autotrophs (Virtanen et al., 2009)
and freshwater biodiversity loss is disproportionately high (Strayer and microorganisms (Stein et al., 2010; Korbel et al., 2013) should re-
and Dudgeon, 2010). Agricultural intensification and urbanization spond more to changes in groundwater quality. We further expected
rank highest among anthropogenic threats to freshwater ecosystems, (ii) ecosystem processes to show a subsidy-stress response (Odum
causing not only nutrient enrichment but also organic pollution (Malaj et al., 1979; Wagenhoff et al., 2011) to groundwater contamination. At
et al., 2014) and salinization (Cañedo-Argüellus et al., 2013, 2016). low-to-moderate doses, nutrient (nitrogen) enrichment should acceler-
The ongoing degradation of freshwater ecosystems has engendered ate primary productivity and decomposition rates, while at high con-
legislative mandates for the protection and management of surface wa- centrations, nutrients (and associated chemicals) cause groundwater
ters (e.g. European Water Framework Directive) while groundwater- contamination, reducing ecosystem process rates (see Woodward
dependent ecosystems (GDEs) have received substantially less et al., 2015). Alternatively, any positive effects of nutrient enrichment
attention. could be largely cancelled by other contaminants associated with agri-
Cold-water springs are fed by a continuous flux of thermally stable culture or urbanization, resulting in no net change to ecosystem process
groundwater, creating unique environmental setting for the develop- rates. We also expected (iii) ecosystem processes to show divergent re-
ment of highly specialized biota (Cantonati et al., 2012). Like most sponses to the two stressors, groundwater contamination resulting in
GDEs, also springs are threatened by land use and impaired water qual- the stress-subsidy type of response while land drainage might suppress
ity (Barquín and Scarsbrook, 2008; Lehosmaa et al., 2017). Since the ecosystem functions already at low stressor levels, due mainly to altered
nineteenth century, peatland drainage has greatly modified northern thermal stability and increased brownification caused by humic sub-
European forest landscapes (Holden et al., 2004), the goal being en- stances (Lehosmaa et al., 2017).
hancement of forest growth by channeling surplus water to recipient
streams and lakes. In Finland, drainage was practiced until 1997 when 2. Materials and methods
the Finnish Forest Act declined further forest drainage, and currently
only maintenance of old drainage networks is permitted. Forest drain- 2.1. Selection of study sites
age has deteriorated fluvial ecosystems by, for example, increasing sed-
iment loads and nutrient concentrations (Jyväsjärvi et al., 2014; We collected biological and environmental data from 20 southern
Nieminen et al., 2017). Little is known about drainage impacts on Finnish (60.13°–62.16° N; 22.4°–26.9° E) springs. Reference sites and
GDEs, although some recent studies have documented increased surface sites disturbed by land drainage or groundwater contamination were
water inflow, lower thermal stability and elevated concentrations of spatially interspersed (see Appendix 1). The study sites were compara-
dissolved organic carbon (DOC) in drainage-impacted springs ble in terms of size and habitat structure with a distinct spring pool and
(Lehosmaa et al., 2017), with major consequences on spring biodiver- an outflowing stream. Spring habitat integrity was assessed using a
sity (Lehosmaa et al., 2017). four-tiered classification by Heino et al. (2005) whereby sites with
Groundwater resources are globally compromised by several harm- drainage ditches flowing directly into the spring pool were assigned de-
ful agents. Nitrate (NO−3 ) is the most ubiquitous contaminant of ground- graded (hereafter “Drainage”; class 0). Pristine or near-pristine springs
water (Spalding and Exner, 1993) and is derived from both diffuse with no visible human impact in the vicinity of the spring (within at
sources (Mahvi et al., 2005) and a mixture of urban and agricultural least 100 m from a spring) were assigned “Natural” (class 3). In situ
point sources (Wakida and Lerner, 2005). Nitrate typically covaries water quality (pH, electrical conductivity) and water chemistry data of
with other groundwater contaminants (Katz et al., 2009; Nolan and the parent aquifers (obtained from national HERTTA database by Finn-
Weber, 2015) and can thus be considered a proxy of groundwater ish Environment Institute, SYKE) were used to confirm that springs in
chemical degradation. Increasing concentrations of dissolved inorganic these two groups showed no major alteration of groundwater quality.
salts (Cañedo-Argüellus et al., 2016) impose an additional threat to To include sites with degraded water quality (WQ), we screened,
groundwater resources. In northern regions, increased salinity origi- using SYKE groundwater monitoring database, all southern Finnish
nates mainly from road salt (NaCl) which is easily transported to aquifers with disproportionately high nitrate (N2000 μg L−1 NO− 3 -N)
groundwater (Williams et al., 2000; Thunqvist, 2004). concentrations compared to national mean values (see Rossi et al.,
Despite increasing awareness of the degradation of groundwater 2015). Map surveys were used to locate springs connected to contami-
and groundwater-dependent ecosystems, comparative studies on the nated aquifers and these sites were visited to confirm that their habitat
diversity and community composition of GDE biota and ecosystem pro- structure did not deviate from the reference condition. This site selec-
cesses along anthropogenic stressor gradients are largely lacking. We tion procedure resulted in a data set containing 1) ‘Natural’ springs un-
assessed the impacts of two such stressors, land drainage and ground- disturbed by any anthropogenic stressors (n = 7); 2) ‘Drainage’ springs,
water contamination, on structural (benthic macroinvertebrates, i.e. springs directly impacted by drainage but with negligible changes in
aquatic bryophytes, leaf-decomposing fungi and groundwater bacteria) groundwater quality (n = 5), and 3) ‘WQ’ springs with impaired
and functional (primary productivity and organic matter decomposi- groundwater quality but no drainage impacts (n = 8).
tion) characteristics in 20 boreal springs. Springs unaffected by drainage
and with unaltered water chemistry were used as reference sites 2.2. Environmental data
whereas the impacted sites were characterized by either drainage-
induced thermal instability and brownification of the spring water (in- Area (m2) of different habitat types (i.e. spring pools, helocrenes,
creased DOC content) or by strongly elevated nitrate concentrations and spring brooks with either minerogenic or organogenic substrate)
from agricultural or urban land use. We expected that both stressor was measured during sampling visits. Depth of the spring pool and
types should cause communities and ecosystem functions to deviate depth and width of the outflowing stream were measured at five
102 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109

random locations. Current velocity (m/s) was measured at a single loca- sterilized 0.2 μm MO BIO filter membranes and then extracted with
tion at the spring outlet. Water samples (1 L) were taken from the PowerWater DNA Isolation Kit (MO BIO laboratories, Carlsbad, Califor-
spring pool in early June and they were analyzed for dissolved organic nia, USA). Fungal DNA was extracted from 0.07 g of leaf powder using
carbon (DOC), total phosphorus (TP), nitrate nitrogen (NO− 3 -N) and PowerSoil DNA Isolation Kit. DNA content of bacteria and fungi ex-
chloride (Cl−) using Finnish national standards (National Board of tracted from samples was diluted to 5 ng μL−1. The rRNA coding for
Waters, 1981). In addition, electrical conductivity, pH and water tem- fungi was amplified using ITS primers ITS1F (Gardes and Bruns, 1993)
perature were measured from the spring pool at each site visit (April and 58A2R (Martin and Rygiewicz, 2005) and for bacteria the 16S
and June) with a field meter (WTW Multi 350i meter). Water tempera- rRNA region using 519F and R926 primers (Lane, 1991). The amplicons
ture variation in each spring was measured with temperature loggers were sequenced using the Ion Torrent PGM sequencer at Biocenter Oulu
(iButton, Thermochron; Maxim Integrated, San Jose, CA, USA) set to Sequencing Center (University of Oulu, Finland) with Ion HiQ chemistry
measure temperature at 60-min intervals in April–June for 49 days. Per- and 316 chips (see Appendix 2 for molecular analyses). The sequences
centage of different land cover (forests, open mires) and land use types were analyzed using Quantitative Insights into Microbial Ecology
(agriculture, built areas, logged areas) within a 1-km buffer around each (QIIME) pipeline (Caporaso et al., 2010). The sequence library was
study site were estimated with ArcMap 10.1 Desktop (ESRI, 2011) using split by samples and quality-filtered based on the quality scores for
Corine 2012 Land Cover data. every sequence. Quality scores below 25 were removed, and minimum
and maximum sequence lengths were 200 bp and 1000 bp, respectively.
2.3. Functional measures Sequences with ambiguity and more than two mismatches in the
primer were also removed. Sequences were clustered into operational
We measured two key ecosystem processes, primary productivity taxonomic units (OTUs) using the Usearch61 algorithm, clustering
and leaf litter decomposition, in each spring. Periphyton accrual rate, a OTUs at 97% identity (Edgar, 2010). Chimeras were detected using
surrogate of primary productivity, was measured by incubating three UCHIME method (Edgar et al., 2011), and OTU composition was deter-
unglazed ceramic tiles (5 × 5 cm; attached on top of a house brick) in mined using BLAST (basic local alignment search tool) against Unite
each spring pool for 49 days. After the incubation period, periphyton ITS database (Kõljalg et al., 2013) and against Greengenes 16S database
biomass (μg cm−2) was measured from moist tiles with a phytobenthos (DeSantis et al., 2006). As sequence numbers varied among samples,
fluorescence measurement instrument (Benthotorch®, bbe Moldaenke, OTU data were rarefied to the lowest shared sample size (fungi n =
Germany) (see Kahlert and McKie, 2014). 2542, bacteria n = 5282) and singletons were discarded prior to statis-
Leaf breakdown assays were used to measure organic matter de- tical analyses. Two natural sites for bacterial analyses and one WQ site
composition. Alder (Alnus incana) leaves were collected in August for both fungal and bacterial analyses were discarded due to sample
2014 prior to abscission and they were air-dried for seven days. Alder contamination.
leaf packs (15 × 15 cm, n = 3 per site) containing 4 g of leaves were
placed in spring pools in late April 2015 and leaf decomposition rates 2.5. Statistical analyses
were measured after 49 days. Leaf packs of two different mesh sizes
were used to allow shredding invertebrates to enter the bags (8 mm, 2.5.1. Environmental data
to measure total decomposition rate) or to exclude them (0.2 mm, here- Principal component analysis (PCA) was used to characterize envi-
after microbial decomposition rate). After incubation, the remaining leaf ronmental variation of the study sites and to visualize differences in en-
material was carefully cleaned of invertebrates and other material in the vironmental settings among the three site groups (Natural, Drainage
laboratory. A subsample of 1 g of leaf material was taken from each fine and WQ). Prior to analysis, all variables were standardized to zero
mesh bag for fungal DNA extraction and analyses (see below). The re- mean and unit variance. The final number of principal components
maining material was dried at 60 °C for 48 h and ashed for four hours retained was determined using the broken-stick model whereby eigen-
at 550 °C to determine ash-free dry mass (AFDM). Leaf breakdown values from the PCA were compared to values given by the broken stick
rate (k) was calculated using the negative exponential decay model. distribution (Jackson, 1993). A component is retained if its eigenvalue is
larger than the value given by the broken stick model. Among-group
2.4. Taxonomic measures differences in environmental variables of particular interest were tested
with one-way ANOVA, followed by Tukey's pairwise comparisons. Test
Biological sampling was conducted mainly in May 2015, with some assumptions were checked prior to ANOVAs, and if assumptions were
additional macroinvertebrate and bryophyte data being extracted not satisfied, log10-transformation was applied to the response variable.
from Ilmonen et al. (2009, 2012). Benthic invertebrates were sampled
using a 20-cm wide D-frame hand net (mesh size 500 μm). Five 20-s 2.5.2. Diversity measures
subsamples representing all available habitat types (spring pools, We used Generalized Linear Models (GLMs) with a poisson or
helocrenes, spring-fed streams with minerogenic or organogenic sub- quasipoisson (in case of overdispersion) link function to test the differ-
strate) were taken in 2-m intervals within the first 10 m from the spring ences in taxonomic richness among the site groups (exception bryo-
source by kicking the bottom and sweeping and pressing mossy or phyte richness). As our study springs harbor many generalist
muddy substrates and collecting loose material into the net (see macroinvertebrate and bryophyte species typical of ponds and headwa-
Ilmonen et al., 2012). Animals were preserved in 70% ethanol in the ter streams (Ilmonen et al., 2009), we conducted separate analyses for
field and later sorted and identified (including chironomids) to the low- generalists and for species known to prefer springs and spring-fed
est feasible level (usually species). streams as their main habitat (i.e. crenophiles; see Ulvinen et al.,
Bryophytes were sampled at 1-m intervals from the point of ground- 2002; Ilmonen et al., 2009). Taxon richness of specialist and generalist
water discharge along the main course of the flow in seven 0.5 × 0.5 m macroinvertebrates were rarefied to 200 individuals prior to GLM
quadrats. In the smallest springs (b10 m2), only five quadrats were sam- analyses.
pled. For each plot, we identified all bryophyte species, including semi- Linear mixed effects-models (LME) were used to test differences in
aquatic taxa, and estimated their percentage cover (%) visually. periphyton accrual and leaf breakdown rates and bryophyte species
Fungal and bacterial assemblages were determined using DNA se- richness among the spring groups. LME models were constructed
quencing techniques. Subsamples of frozen leaf material were freeze- using lme4 (Bates et al., 2014) and lmerTest (Kuznetsova, 2016) pack-
dried and pulverized for the extraction of fungal DNA, while bacterial ages in R (R Core Team, 2014) and included site groups as fixed effects
samples were extracted from a 100-mL water sample taken from the and individual leaf bags or periphyton tiles nested within sites, or in
upmost 30 cm of the spring pool. Water samples were filtered with case of bryophytes a plot, as random effects.
 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109 103

The relationships between the functional measures (periphyton ac- 2.5.4. Indicator OTUs for different spring groups
crual and leaf decomposition rates) and NO− 3 -N or DOC content were To identify indicator macroinvertebrate and bacterial taxa for each
assessed with linear regressions. Quadratic terms were included in the site group, indicator species analysis was applied using indicator func-
model if they improved the model fit (based on inspection of residual tion of the indicspecies package (De Cáceres, 2015) in R. This analysis
plots, adjusted r2-values and associated P-values). Explanatory and re- combines information of species abundance and occurrence in each
sponse variables were log10-transformed if needed to linearize the rela- group, the strongest indicators being both abundant and present in
tionships and to improve data distribution. To assess the potential only a single group, with a high indicator value and low permuted P-
contribution of grazing and shredding macroinvertebrate abundance value for that particular group (Dufrêne and Legendre, 1997).
on periphyton accrual and leaf decomposition rates, respectively, we
calculated total abundances of grazers and shredders by weighing the 3. Results
abundance of each taxon with its shredding and grazing propensity
(0–1). The feeding type propensities were obtained from the 3.1. Environmental data
European trait database (Schmidt-Kloiber and Hering, 2015).
The first three principal components explained 49.1% of total vari-
2.5.3. Community composition and community-environment relationships ance in the environmental data. The first PC axis represented a gradient
Among-group differences in macroinvertebrate, bryophyte, fungal in groundwater quality and related changes in land use (22.2% of varia-
and bacterial species composition were demonstrated visually with tion explained). PC1 correlated positively with nitrate nitrogen and
non-metric multidimensional scaling (NMDS) using metaMDS function chloride concentrations, as well as percentages of built areas and agri-
of vegan package (Oksanen et al., 2015) of the R program. Statistical dif- cultural fields within a 1-km buffer (Fig. 1). The second PC axis repre-
ferences in species composition among the site groups were tested sented a gradient in land drainage (15.3% explained) with positive
using one-way nonparametric permutational multivariate analysis of correlation with spring habitat integrity and negative correlation with
variance (PERMANOVA; Anderson, 2001) with adonis function in DOC concentration, thermal variation of spring water (temperature
vegan package (Oksanen et al., 2015). PERMANOVAs were run using CV) and area of logged forests within the buffer (Fig. 1). The three site
the Bray-Curtis similarity coefficient on abundance data, and statistical groups were distinctly separated in the ordination space, the WQ sites
significance was estimated based on 9999 permutations. A significant being located on the upper right-hand side, drained sites in the lower
global test (i.e., macroinvertebrates and bacteria; see Results) was en- middle and natural sites on the upper left-hand side of the ordination
sued by pairwise PERMANOVAs. In these two cases, we also assessed space (Fig. 1b).
the relative importance of the components of beta diversity (i.e. turn- Environmental conditions differed strongly among the site groups
over and nestedness) with bray.part function in betapart package (PERMANOVA: F2,17 = 3.57, P = 0.001). Both WQ (F1,13 = 23.17, P b
(Baselga et al., 2017) using species abundance data (see Barwell et al., 0.001) and drainage-impacted (F1,10 = 4.40, P = 0.005) springs differed
2015). This analysis was conducted to test if community variability from the natural springs, and there was also a significant difference be-
along the environmental gradients was caused by the low-diversity tween the two disturbed site groups (F1,11 = 50.7, P b 0.001).
sites being nested subsets of high-diversity sites (species loss) or by Nitrate nitrogen and chloride concentrations in the WQ sites
true turnover (species replacement) along the gradient. Beta diversity were on average 30 and 15 times higher, respectively, than in
component matrices (turnover, nestedness) were correlated with envi- natural springs (Fig. 2a and b; Appendix 3), while drainage-
ronmental matrix (nitrate concentration; strongly charged to PC1) impacted sites did not differ from natural sites (Fig. 2). However,
using partial Mantel tests, with the potential confounding effect of geo- drainage sites had higher DOC concentrations than did the natu-
graphical distances between sites being partialled out. Finally, the ral springs (Fig. 2c; Appendix 3) and also a substantially different
among-group differences in environmental conditions were also tested seasonal trajectory of water temperature. The influx of early-
with PERMANOVA, based on a Euclidean distance matrix. spring snowmelt waters resulted in lower spring water

a) b)
2

0.5 Habitat integrity

Principal component 2 (15%)

Helocrenes
% Agricultural fields Natural
pH
NO3-N
Organic substrates Cl- % Built areas WQ
TP
0

0.0

% Open mires
% Forests
−1

DOC
−0.5 % Sparse canopy Drainage

Temperature CV
−2

−1.0
−3

−1.0 −0.5 0.0 0.5 1.0 −3 −2 −1 0 1 2

Principal component 1 (22%)

Fig. 1. PCA ordination of the environmental variables of the study springs. Arrows in panel a) indicate the direction and strength of the correlations between the principal components and
each environmental variable. Polygons in the panel b) enclose all sites within a site group.
104 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109

Fig. 2. Mean (±95% CI) values for (a) nitrate nitrogen (NO− −
3 -N), (b) chloride (Cl ) and (c) dissolved organic carbon (DOC) concentration in different spring groups. Groups sharing a letter
do not differ significantly (one-way ANOVA, followed by Tukey's test, P b 0.05).

temperature in the beginning of the study (early May); with an 3.2. Taxonomic richness
opposite pattern by the end of the study period (early June) as
water temperature in drainage-impacted sites (due to surface Altogether 147 macroinvertebrate, 28 bryophyte, 124 fungal and
water influx) exceeded the ambient groundwater temperature 722 bacterial taxa were identified in our data. Nearly half (46.5%) of
(Appendix 4). the macroinvertebrate taxa consisted of chironomid larvae, and the

Fig. 3. Taxonomic richness (means ±95% CI) of (a) habitat specialist and (b) generalist macroinvertebrates, (c) habitat specialist and (d) generalist bryophytes, (e) leaf-decomposing fungi
and (f) groundwater bacteria in each spring group. Lines above or below error bars indicate significant differences from natural conditions according to generalized linear models (* P b
0.05, ** P b 0.01). For fungi and bacteria, taxonomic richness was defined as the number of OTUs determined using BLAST (basic local alignment search tool; see text for more information).
 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109 105

majority of the remaining taxa were composed of other dipterans (19%), OTU composition of the WQ sites differed from natural conditions
coleopterans (10%) and trichopterans (10%). The majority (78%) of (F1,10 = 5.63, P = 0.005), while land drainage did not alter bacterial as-
bryophyte flora belonged to Bryophyta, while the remaining taxa semblages (Natural vs. Drainage F1,8 = 0.71, P = 0.74).
(22%) were liverworts. Of the 124 fungal OTUs, 65% belonged to Asco- Indicator species analysis identified 38 bacterial indicator OTUs for
mycota, followed by Basidiomycota (25%), Chytridiomycota (6%), natural springs and eight and nine indicators for the drainage-
Zygomycota (2.5%) and unknown OTUs. Of the 722 groundwater bacte- impacted and WQ sites, respectively (see Appendix 5). The best bacte-
ria OTUs, 33% were Proteobacteria, while Bacteroides (9%), rial indicators (Indicator value = IV N 0.80, P b 0.01) for natural springs
Actinobacteria (8%), Firmicutes, Acidobacteria and Chloroflexi (all 6%), were the proteobacteria OTUs Beijerinckia sp., Pedomicrobium sp. and
and Cyanobacteria and Planktomycetes (both 5%), comprised the ma- the family Cystobacterineae, while OTU of the family Armatimonadaceae
jority of the remaining OTUs. with four coarser level BLAST hits also indicated natural conditions (Ap-
Rarefied taxon richness of both specialist and generalist macroinver- pendix 5). The proteobacteria Paucibacter sp., Acidovorax sp., Delftia sp.
tebrates differed among site groups, being lower in WQ than in natural and Comamonadaceae, and the bacteroides Flavobacterium sp. were
springs (P = 0.003 and P b 0.001, respectively), whereas the difference the strongest indicators for groundwater contamination (IV N 0.70, P b
between natural and drainage-impacted sites was non-significant (spe- 0.04). The proteobacteria Gallionella sp., Acidisoma sp. and
cialists: P = 0.18; generalists: P = 0.26) (Fig. 3a–b). Species richness of Telmatospirillum sp. best indicated land drainage (IV N 0.80, P b 0.03).
specialist bryophytes responded negatively to groundwater contamina- We found 17 macroinvertebrate indicator taxa for natural springs
tion (P = 0.037; Fig. 3c), while no such response was detected to land and two for the drained sites (see Appendix 5). The best indicators of
drainage (P = 0.16). Generalist bryophytes (Fig. 3d) and fungi natural conditions (IV N 0.80, P b 0.01) were the isopod Asellus aquaticus,
(Fig. 3e) showed no response to either of the stressors (both P N 0.10). the plecopteran larva Nemoura cinerea, the dipteran larva Dixa
Bacterial OTU richness was lower in WQ sites than in natural conditions submaculata, the lepidostomatid larva Crunoecia irrorata and the chiron-
(P = 0.03) whereas the difference between the drained and natural omid Micropsectra atrofasciata-agg., while the chironomid genera
sites was non-significant (P = 0.82) (Fig. 3f). Corynoneura spp. best indicated land drainage (IV N 0.80, P b 0.01). No
macroinvertebrate taxa indicated groundwater contamination.
A. aquaticus and Nemurella pictetii were the predominant grazers in
3.3. Community composition
natural springs (grazing propensity-weighted average abundances
45.8 and 34, respectively). Both species were less abundant in the
Macroinvertebrate community composition differed significantly
drained and WQ springs (2.3 and 0.9 for A. aquaticus and 17.5 and 8.1
among the spring groups (PERMANOVA F2,17 = 2.16, P b 0.001;
for N. pictetii, respectively). Being highly abundant and omnivorous,
Fig. 4a), while the differences in bryophyte and fungal communities
the same two species were also predominant shredders in the natural
only bordered at significance (F2,16 = 1.69, P = 0.06 and F2,16 = 1.48,
springs (45.8 and 69.3, respectively).
P = 0.06, respectively; Fig 4b and c). Bacterial communities also were
According to partial Mantel tests (partialling out geographical dis-
different in different spring types (F2,14 = 3.23, P = 0.002; Fig. 4d).
tances), macroinvertebrate beta diversity along the groundwater con-
Pair-wise PERMANOVAs showed that macroinvertebrate communities
tamination gradient (nitrate nitrogen) was driven by turnover (r =
differed from natural conditions in both drainage-impacted (F1,10 =
0.31, P = 0.04), while the role of nestedness was non-significant (r =
2.93, P = 0.003) and WQ sites (F1,13 = 2.49, P b 0.001) while the two
0.05, P = 0.29). Likewise, groundwater bacteria beta diversity was re-
disturbed site groups did not differ (F1,11 = 1.34, P = 0.14). Bacterial
lated to species turnover (r = 0.38, P = 0.001) with the contribution
of nestedness being negligible (r = 0.01, P = 0.99).

3.4. Functional measures

Land drainage reduced both periphyton accrual (P = 0.01; Fig. 5a)

and total leaf decomposition rates (P = 0.001) (Fig. 5b) whereas neither
of the ecosystem processes deviated from natural conditions in ground-
water contamination sites (periphyton biomass: P = 0.90; leaf decom-
position: P = 0.28; Fig. 5). Microbial decomposition did not differ from
reference conditions in either drained (P = 0.93) or WQ (P = 0.22)
springs, implying that the among-group differences in leaf litter decom-
position were primarily caused by leaf-shredding macroinvertebrates.
However, there was no indication of top-down effects on ecosystem
process rates as grazer abundance was unrelated to periphyton accrual
(r2 = 0.01, P = 0.66), as was also shredder abundance to leaf decompo-
sition rate (r2 = 0.05, P = 0.35).
Variation of leaf decomposition rates across all sites was best ex-
plained by DOC concentration (log-log model; r2 = 0.44, P = 0.001),
with reduced decomposition rates being associated with increasing
DOC (Fig. 6a) but not with nitrate concentrations (r2 = 0.04; Fig. 6b).
Periphyton productivity was negatively associated with DOC (r2 =
0.24, P = 0.03; Fig. 6c), whereas nitrate enrichment resulted in a
subsidy-stress type of response in periphyton accrual rate (r2 = 0.39,
P = 0.01; Fig. 6d). Both ecosystem functions responded strongly nega-
tively to even small increases in DOC (Fig. 6a, c).

4. Discussion
Fig. 4. Nonmetric multidimensional scaling (NMDS) ordinations of (a) macroinvertebrate,
(b) bryophyte, (c) fungal and (d) bacterial assemblages. Ellipses denote 95% confidence Human actions have degraded aquatic habitats worldwide, leading
intervals around the group centroid. to biodiversity loss and change (Vörösmarty et al., 2010) which often
106 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109

Fig. 5. Mean (±95% CI) values for (a) periphyton accrual and (b) total decomposition rates in each spring group. Lines below error bars indicate significant differences from natural
conditions according to linear mixed effects models (*P b 0.05, ***P b 0.001).

translates into alteration of ecosystem functions and services (Loreau and other harmful contaminants beyond their natural levels (Lapworth
et al., 2001). Springs and other groundwater-dependent ecosystem et al., 2012; Mauffret et al., 2017).
(GDEs) are no exception to this unwanted development and, although We evaluated the effects of two anthropogenic stressors, groundwa-
they have attracted increasing interest by ecologists (see Cantonati ter contamination and land drainage, on boreal spring biodiversity and
et al., 2012), the impacts of human stressors on the structure and func- ecosystem functioning. Our cross-taxon comparison showed that, con-
tioning of GDEs remain poorly studied. Agricultural and forestry-related cordant with our hypothesis, taxonomic richness and species composi-
land uses are the predominant causes for the global deterioration of tion of groundwater bacteria, as well as diversity of habitat specialist
spring biodiversity (Barquín and Scarsbrook, 2008). Land modification, bryophytes, were compromised by impaired groundwater quality.
regardless of its purpose, alters spring habitat quality but also deterio- However, against our predictions, also macroinvertebrate communities
rates groundwater quality by elevating concentrations of salts, nutrients were impaired by groundwater contamination. Leaf-decomposing fungi

Fig. 6. Relationships between leaf decomposition rate and (a) DOC and (b) nitrate (NO−
3 -N) concentration, and periphyton accrual rate and (c) DOC and (d) nitrate concentration. The
coefficients of determination and corresponding P-values are given for each model.
 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109 107

and generalist bryophytes showed much weaker responses to both inflow remains sufficient (Ilmonen et al., 2012; Lehosmaa et al., 2017).
stressors. As predicted, groundwater contamination altered periphyton Benthic invertebrate richness was greatly reduced by groundwater con-
productivity, with a unimodal response to elevated nitrate concentra- tamination and this reduction was equally strong for both spring habitat
tions but, unexpectedly, we did not observe a similar subsidy-stress re- specialists and headwater generalist taxa. This result parallels that of
sponse in leaf decomposition rate. As hypothesized, land drainage- Beketov et al. (2013) that agricultural pesticides have caused a regional
induced brownification reduced ecosystem process rates already at biodiversity loss of stream invertebrates. Similar to bacteria, changes in
very low stressor levels and caused compositional and diversity changes community composition along the gradient of groundwater contamina-
to macroinvertebrate assemblages, while other taxonomic groups tion were caused mainly by true turnover rather than nestedness, this
seemed more tolerant of drainage impacts. being so despite IndVal analysis was unable to identify any strong inver-
For simplicity, we used nitrate concentration as a surrogate for tebrate indicators for the contaminated sites. Interestingly, some of the
groundwater contamination, noting substantial compositional differ- indicators for natural springs (A. aquaticus, N. cinerea) are considered
ences between reference and contaminated sites. This implies that the headwater generalists, while others are genuine spring specialist spe-
threshold set by EU legislation for acceptable groundwater nitrate cies (C. irrorata, L. nigra, N. pictetii) (see Ilmonen et al., 2009).
level (The Nitrates Directive: b50 mg L−1) is insufficient for protecting While land drainage did not contribute much to taxonomic richness
GDE biota. However, the degradation of spring biota in contaminated or composition of spring biota, it had a notable effect on ecosystem func-
sites is unlikely to be driven solely by elevated nitrate concentrations. tioning by lowering periphyton accrual and total leaf decomposition
Salts, heavy metals, pesticides and other contaminants often covary rates. Increased DOC concentration causes brownification of the spring
with nitrate in agricultural and urban environments, creating a ‘cocktail water, decreasing light penetration (Brett et al., 2017) and reducing
of contaminants’ with unexpectedly adverse effects on aquatic biota substrate availability for algal attachment (Wojtal and Sobczyk, 2012),
(see Relyea, 2009). Mauffret et al. (2017), for example, reported that tri- both of which are plausible explanations for reduced periphyton accrual
azine pesticide concentration increased parallel with nitrate concentra- rate in our drainage-impacted sites. Reduced leaf decomposition rate in
tion. Furthermore, Chowdhury et al. (2003) showed that groundwater drainage-impacted sites was likely caused by increased input of fine
aquifers with high nitrogen concentrations are vulnerable also to sediments and DOC from drainage ditches: leaf bags at the drained
other anthropogenic pollutants. While many of these potential agents sites were typically covered by fine organic sediments (K. Lehosmaa,
of chemical degradation were not measured in our study, we did ob- pers. obs), which may have prevented efficient utilization of leaf litter
serve a strong positive correlation (r = 0.85) between nitrate and chlo- by decomposers (Lecerf and Richardson, 2010; Sanpera-Calbet et al.,
ride concentrations, supporting our assumption that sites high in nitrate 2012). In addition, complexity of DOC increases with concentration,
are likely stressed by other contaminants as well. making allochthonous compounds less available for hydrolase break-
Groundwater contamination caused substantial changes to commu- down (Emilson et al., 2017). Differences in leaf decomposition rates be-
nity composition and OTU richness of groundwater bacteria. In contrast tween natural and drainage-impacted sites were mainly driven by leaf-
to recurrent observations about the positive relationship between shredding macroinvertebrates, while microbial decomposition rate was
groundwater contamination and bacterial diversity (Johnson et al., unaffected by land drainage. It is generally assumed that DOC fuels mi-
2004; Stein et al., 2010), we documented a notable decrease of bacterial crobial activity in groundwater (Griedler and Lueders, 2009; Chapelle
diversity in sites with elevated nitrate and chloride concentrations. et al., 2013), but our results suggest a weak relationship between micro-
However, our result parallel those reported for bacterial communities bial decomposer activity and DOC concentrations in boreal springs. Also
in forest soils after addition of nitrogen fertilizers (Liu and Greaver, Korbel et al. (2013) noted that microbial activity was uncorrelated to
2010). Partitioning of bacterial beta diversity into turnover and total DOC concentration in groundwater.
nestedness components showed that changes in community composi- Periphyton accrual rate was also sensitive to changes in groundwa-
tion with degrading groundwater quality were caused by the replace- ter quality; as hypothesized, primary productivity showed a subsidy-
ment of native bacterial OTUs with new taxa rather than by mere stress response to groundwater contamination, first increasing rapidly,
disappearance of the original taxa. Many of the non-native bacterial plateauing at around 2500 μg L−1, and then decreasing at higher con-
OTUs present at our chemically degraded sites are involved in purifying centrations. The decreasing part of the curve likely represents the “cock-
contaminated groundwater and in the breakdown of organic contami- tail of contaminants” typical of urban and agricultural streams (Relyea,
nants (Griedler and Lueders, 2009). For instance, OTUs of the genus 2009). Woodward et al. (2015) detected a similar unimodal response
Acidovorax are used in wastewater treatment plants as denitrifying bac- to nutrients for leaf decomposition, while we did not see a similarly
teria (Ehsani et al., 2015), and species of the genus Paucibacter have strong effect of nitrate enrichment on leaf decomposition rate. Several
been linked to high nitrogen levels in urban streams (Hosen et al., Central European studies have documented a strong inhibition of leaf
2017). The cosmopolitan bacterial phyla Proteobacteria, Bacteroides, decomposition (particularly the microbially-induced part of it) by agri-
Actinobacteria and Firmicutes (see Salis et al., 2017) were among the cultural pesticides (e.g. Rasmussen et al., 2012; Fernández et al., 2015),
dominating taxa also in our data, supporting the view that GDEs do but the level of disturbance by nutrients and/or pesticides in those stud-
not support entirely endemic bacterial groups (Stein et al., 2010; ies has been much higher than in our impacted springs. Interestingly,
Sirisena et al., 2013). Feckler et al. (2017) showed that when a pristine microbial community
Our results suggest that leaf-decomposing fungi in springs are rather was exposed to fungicides, their leaf-breakdown efficiency was dramat-
tolerant of anthropogenic stressors. This may relate partially to the gen- ically reduced, and this effect was enhanced by high nutrient levels,
erally low activity and diversity of fungal decomposers in the constantly while a community with a history of fungicide exposure responded pos-
cold groundwater-dominated conditions (Bärlocher et al., 2008). Alter- itively to simultaneous effects of fungicides and nutrients. It therefore
natively, the level of disturbance at out study sites may not have seems likely that the threat to aquatic ecosystems caused by increasing
exceeded the tolerance level of most fungal taxa. Spring specialist bryo- use of pesticides and fertilizers will be particularly high in areas where
phytes, but not headwater generalist bryophytes, responded to ground- the level of impact is currently modest, as in the southern Finnish
water contamination by reduced diversity, whereas they showed little springs studied by us.
response to land drainage. Such a low responsiveness to land-use dis- Springs are small and isolated aquatic islands within a terrestrial ma-
turbance was unexpected, as our previous studies have shown that trix and therefore highly vulnerable to anthropogenic pressures. For in-
land drainage may alter the abundance, diversity and community com- stance, it has been estimated that only 10% of springs in southern
position of spring-dwelling bryophytes (Lehosmaa et al., 2017, 2017). Finland have retained a near-natural status (Ilmonen et al., 2008). Our
Our results support the view that spring bryophytes may be tolerant results show that groundwater contamination has detrimental effects
of even relatively strong habitat modification as long as groundwater on spring biodiversity and primary productivity whereas drainage-
108 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109

induced spring water brownification reduces ecosystem process rates. Brett, M.T., Bunn, S.E., Chandra, S., Galloway, A.W.E., Guo, F., Kainz, M.J., Kankaala, P., Lau,
D.C.P., Moulton, T.P., Power, M.E., et al., 2017. How important are terrestrial organic
Our study addressed only individual stressors, however, and combina- carbon inputs for secondary production in freshwater ecosystems? Freshw. Biol. 62,
tions of these with other stressors (e.g. groundwater warming; see 833–853.
Kurylyk et al., 2014; Jyväsjärvi et al., 2015) may yield non-additive “eco- Cañedo-Argüellus, M., Hawkins, C.P., Kefford, B.J., Schäfer, R.B., Dyack, B.J., Brucet, S.,
Buchwalter, D., Dunlop, J., Frör, O., Lazorchak, J., et al., 2016. Saving freshwater from
logical surprises” (Paine et al., 1998), potentially leading to an unex- salts. Science 351, 914–916.
pectedly rapid impoverishment of spring biodiversity. Further Cañedo-Argüellus, M., Kefford, B.J., Piscart, C., Prat, N., Schäfer, R.B., Schulz, C.-J., 2013.
research is needed for a more comprehensive understanding of the indi- Salination of rivers: an urgent ecological issue. Environ. Pollut. 173, 157–167.
Cantonati, M., Füreder, L., Gerecke, R., Jüttner, I., Cox, E., 2012. Crenic habitats, hotpots for
vidual and combined effects of various anthropogenic stressors on
freshwater biodiversity conservation: toward an understanding of their ecology.
GDEs, as well as mitigation of these stressors, to secure the role of Freshw. Sci. 31, 463–480.
springs and their riparian surroundings as woodland key habitats in bo- Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K.,
Fierer, N., Gonzalez Peña, A., Goodrich, J.K., Gordon, J.I., et al., 2010. QIIME allows anal-
real forests.
ysis of high-throughput community sequencing data. Nat. Methods 7, 335–336.
Chapelle, F.H., Bradley, P.M., Journey, C.A., McMahon, P.B., 2013. Assessing the relative bio-
5. Conclusion availability of DOC in regional groundwater systems. Groundwater 51, 363–372.
Chowdhury, S.H., Kehew, A.E., Passero, R.N., 2003. Correlation between nitrate contami-
nation and ground water pollution potential. Groundwater 41, 735–745.
The present study is, to the best of our knowledge, the first one to De Cáceres, M., 2015. Package ‘Indicspecies’: Relationship between Species and Groups of
evaluate simultaneously the effects of these two anthropogenic Sites. R Foundation for Statistical Computing, Vienna, Austria.
DeSantis, T.Z., Hugenholtz, P., Larsen, N., Rojas, M., Brodie, E.L., Keller, K., Huber, T., Dalevi,
stressors on microbial, plant and macroinvertebrate assemblages, as D., Hu, P., Andersen, G.L., 2006. Greengenes, a chimera-checked 16S rRNA gene data-
well as on key ecosystem functions, of spring or, for that matter, any base and workbench compatible with ARB. Appl. Environ. Microb. 72, 5069–5072.
groundwater-dependent ecosystem. Taxonomic and functional mea- Dufrêne, M., Legendre, P., 1997. Species assemblages and indicator species: the need for a
flexible asymmetrical approach. Ecol. Monogr. 67, 345–366.
sures of boreal spring ecosystems responded differently to the stressors Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinfor-
studied by us. Groundwater contamination reduced the diversity of matics 26, 2460–2461.
spring biota and also reduced primary productivity, whereas land drain- Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensi-
tivity and speed of chimera detection. Bioinformatics 27, 2194–2200 (btr381).
age reduced both ecosystem processes already at very low levels of in-
Ehsani, E., Jauregui, R., Geffers, R., Jarek, M., Boon, N., Pieper, D.H., Vilchez-Vargas, R., 2015.
creased DOC. Our results thus emphasize the importance of including First draft genome sequence of the Acidovorax caeni sp. nov. type strain R-24608
ecosystem-level functional measures in GDE bioassessment as they (DSM 19327). ASM 3 (e01378–15).
seem to respond to the very first signs of spring ecosystem impairment. Emilson, C.E., Kreutzweiser, D.P., Gunn, J.M., Mykytczuk, N.C.S., 2017. Leaf-litter microbial
communities in boreal streams linked to forest and wetland sources of dissolved or-
As the ongoing climate change is also known to cause long-term water ganic carbon. Ecosphere 8, e01739.
brownification (Jennings et al., 2010), our results call for great caution in ESRI, 2011. ArcGIS Desktop: Release 10. Environmental Systems Research Institute,
land-use activities in the vicinity of springs and other GDEs. Finally, the Redlands, CA.
Feckler, A., Goedkoop, W., Konschak, M., Bundschuh, R., Kilian, G.J., Kenngott, K.J.G.,
use of state-of-the-art next-generation sequencing methods provided Schulz, R., Zubrod, J.P., Bundschuh, M., 2017. History matters: heterotrophic microbial
important insight into bacterial assemblages in GDEs and also revealed community structure and function adapt to multiple stressors. Glob. Chang. Biol
novel relationships between groundwater contamination and bacterial https://doi.org/10.1111/gcb.13859.
Fernández, D., Voss, K., Bundschuh, M., Zubrod, J.P., Schäfer, R.P., 2015. Effects of fungi-
diversity. cides on decomposer communities and litter decomposition in vineyard streams.
Sci. Total Environ. 533, 40–48.
Gardes, M., Bruns, T.D., 1993. ITS primers with enhanced specificity for basidiomycetes –
Acknowledgements
application to the identification of mycorrhizae and rusts. Mol. Ecol. 2, 113–118.
Griedler, C., Lueders, T., 2009. Microbial biodiversity in groundwater ecosystems. Freshw.
We acknowledge the two anonymous referees for their constructive Biol. 54, 649–677.
comments on the previous draft of the manuscript. We thank Marko Heino, J., Virtanen, R., Vuori, K.M., Saastamoinen, J., Ohtonen, A., Muotka, T., 2005. Spring
bryophytes in forested landscapes: land use effects on bryophyte species richness,
Suokas and Saija Ahonen for helping in molecular analyses and bioinfor- community structure and persistence. Biol. Conserv. 124, 539–545.
matics, and Risto Virtanen for helping with bryophyte identifications. Holden, J., Chapman, P.J., Labadz, J.C., 2004. Artificial drainage of peatlands: hydro-
Maria Rajakallio and Dimitrios Rados aided with sample preparation logical and hydrochemical process and wetland restoration. Prog. Phys. Geogr.
28, 95–123.
in the laboratory. The study was supported by Maj and Tor Nessling Hosen, J.D., Febria, C.M., Crump, B.C., Palmer, M.A., 2017. Watershed urbanization linked
Foundation (Project Number 201500223), University of Oulu, Maa- ja to differences in stream bacterial community composition. Front. Microbiol. 8,
Vesitekniikan tuki ry, and Academy of Finland (AKVA Program, Project 1452. https://doi.org/10.3389/fmicb.2017.01452.
Ilmonen, J., Leka, J., Kokko, A., Lammi, A., Lampolahti, J., Muotka, T., Rintanen, T., Sojakka,
Number 128377). P., Teppo, A., Toivonen, H., Urho, L., Vuori, K.-M., Vuoristo, H., 2008. Sisävedet ja
rannat. Julk.: Raunio, A., Schulman, A. & Kontula, T. (toim.). Suomen luontotyyppien
Appendix A. Supplementary data uhanalaisuus – Osa 1: Tulokset ja arvioinnin perusteet. Suomen ympäristökeskus,
Helsinki. Suomen ympäristö 8/2008. S. 75–109.
Ilmonen, J., Mykrä, H., Virtanen, R., Paasivirta, L., Muotka, T., 2012. Responses of spring
Supplementary data to this article can be found online at https://doi. macroinvertebrate and bryophyte communities to habitat modification: community
org/10.1016/j.scitotenv.2018.05.126. composition, species richness and red-listed species. Freshw. Sci. 31, 657–667.
Ilmonen, J., Paasivirta, L., Virtanen, R., Muotka, T., 2009. Regional and local drivers of mac-
roinvertebrate assemblages in boreal springs. J. Biogeogr. 36, 822–834.
References Jackson, D.A., 1993. Stopping rules in principal components analysis: a comparison of
heuristical and statistical approaches. Ecology 74, 2204–2214.
Anderson, M.J., 2001. A new method for non-parametric multivariate analysis of variance. Jennings, E., Järvinen, M., Allott, N., Arvola, L., Moore, K., Naden, P., Aonghusa, C.N., Nõges,
Aust. Ecol. 26, 32–46. T., Weyhenmeyer, G.A., 2010. Impacts of Climate on the Flux of Dissolved Organic
Bärlocher, F., Seena, S., Wilson, K.P., Dudley Williams, D., 2008. Raised water temperature Carbon from Catchments, p. 199-220. In: Glen, G. (Ed.), The Impact of Climate Change
lowers diversity of hyporheic aquatic hyphomycetes. Freshw. Biol. 53, 368–379. on European Lakes. Springer Netherlands.
Barquín, J., Scarsbrook, M., 2008. Management and conservation strategies for coldwater Johnson, A., Llewellyn, N., Smith, J., van der Gast, J., Lilley, A., Singer, A., Thompson, I., 2004.
springs. Aquat. Conserv. Mar. Freshw. Ecosyst. 18, 580–591. The role of microbial community composition and groundwater chemistry in deter-
Barwell, L.J., Isaac, N.J.B., Kunin, W.E., 2015. Measuring β-diversity with species abundance mining isoproturon degradation potential in UK aquifers. FEMS Microbiol. Ecol. 49,
data. J. Anim. Ecol. 84, 1112–1122. 71–82.
Baselga, A., Orme, D., Villerger, S., De Portoli, J., Leprieur, F., 2017. Package ‘Betapart’, Jyväsjärvi, J., Marttila, H., Rossi, P.M., Ala-Aho, P., Olofsson, B., Nisell, J., Backman, B.,
Partitioning Beta Diversity into Turnover and Nestedness Components. R Foundation Ilmonen, J., Virtanen, R., Paasivirta, L., et al., 2015. Climate-induced warming imposes
for Statistical Computing, Vienna, Austria. a threat to north European spring ecosystems. Glob. Chang. Biol. 21, 4561–4569.
Bates, D., Maechler, M., Bolker, B., Walker, S., Christensen, R.H.B., Singmann, H., Dai, B., Jyväsjärvi, J., Suurkuukka, H., Virtanen, R., Aroviita, J., Muotka, T., 2014. Does the taxo-
Eigen, C., Rcpp, L., 2014. Package “lme4”. R Foundation for Statistical Computing, Vi- nomic completeness of headwater stream assemblages reflect the conservation sta-
enna, Austria. tus of the riparian forest? For. Ecol. Manag. 334, 293–300.
Beketov, M.A., Kefford, B.J., Schäfer, R.B., Liess, M., 2013. Pesticides reduce regional biodi- Kahlert, M., McKie, B.G., 2014. Comparing new and conventional methods to estimate
versity of stream invertebrates. PNAS 110, 11039–11043. benthic algal biomass and composition in freshwaters. Environ. Sci. 16, 2627–2634.
 K. Lehosmaa et al. / Science of the Total Environment 639 (2018) 100–109 109

Katz, B.G., Griffin, D.W., Davis, J.H., 2009. Groundwater quality impacts from the land ap- R Core Team, 2014. Language and Environment for Statistical Computing. R Foundation
plication of treated municipal wastewater in a large karstic spring basin: chemical for Statistical Computing, Vienna, Austria.
and microbiological indicators. Sci. Total Environ. 407, 2872–2886. Rasmussen, J.J., Wiberg-Larsen, P., Baattrup-Pedersen, A., Monberg, R.J., Kronvang, B.,
Kõljalg, U., Nilsson, R.H., Abarenkov, K., Tedersoo, L., Taylor, A.F.S., Bahram, M., Bates, S.T., 2012. Impacts of pesticides and natural stressors on leaf litter decomposition in agri-
Bruns, T.D., Bengtsson-Palme, J., Callaghan, T.M., et al., 2013. Towards a unified para- cultural streams. Sci. Total Environ. 416, 148–155.
digm for sequence-based identification of Fungi. Mol. Ecol. 22, 5271–5277. Relyea, R.A., 2009. A cocktail of contaminants: how mixtures of pesticides at low concen-
Korbel, K.L., Hancock, P.J., Serov, P., Lim, R.P., Hose, G.C., 2013. Groundwater ecosystems trations affect aquatic communities. Oecologia 159, 363–376.
vary with land use across a mixed agricultural landscape. J. Environ. Qual. 42, Rossi, P., Marttila, H., Jyväsjärvi, J., Ala-aho, P., Isokangas, E., Muotka, T., Kløve, B., 2015. En-
380–390. vironmental conditions of boreal springs explained by capture zone characteristics.
Kurylyk, B.L., MacQuarrie, K.T.B., McKenzie, J.M., 2014. Climate change impacts on J. Hydrol. 531, 992–1002.
groundwater and soil temperatures in cold and temperate regions: implications, Salis, R.K., Bruder, A., Piggott, J.J., Summerfield, T.C., Matthaei, C.D., 2017. High-throughput
mathematical theory, and emerging simulation tools. Earth Sci. Rev. 138, 313–334. amplicon sequencing and stream benthic bacteria: identifying the best taxonomic
Kuznetsova, A., 2016. Package lmerTest: Tests in Linear Mixed Effects Models. R Founda- level for multiple-stressor research. Sci. Rep. 7, 44657.
tion for Statistical Computing, Vienna, Austria. Sanpera-Calbet, I., Chauvet, E., Richardson, J.S., 2012. Fine sediment on leaves: shredder
Lane, D.J., 1991. 16S/23S rRNA sequencing. In: Stackebrandt, E., Goodfellow, M. (Eds.), removal of sediment does not enhance fungal colonization. Aquat. Sci. 74, 527–538.
Nucleic Acid Techniques in Bacterial Systematics. John Wiley & Sons, Chichester, Schmidt-Kloiber, A., Hering, D., 2015. www.freshwaterecology.info - an online tool that
United Kingdom, pp. 115–175. unifies, standardises and codifies more than 20,000 European freshwater organisms
Lapworth, D.J., Baran, N., Stuart, M.E., Ward, R.S., 2012. Emerging organic contaminants in and their ecological preferences. Ecol. Indic. 53, 271–282.
groundwater: a review of sources, fate and occurrence. Environ. Pollut. 163, 287–303. Sirisena, K.A., Daughney, C.J., Moreau-Fournier, M., Ryan, K.G., Chambers, G.K., 2013. Na-
Lecerf, A., Richardson, J.S., 2010. Litter decomposition can detect effects of high and mod- tional survey of molecular bacterial diversity of New Zealand groundwater: relation-
erate levels of forest disturbance on stream condition. For. Ecol. Manag. 259, ships between biodiversity, groundwater chemistry and aquifer characteristics. FEMS
2433–2443. Microbiol. Ecol. 86, 490–504.
Lehosmaa, K., Jyväsjärvi, J., Virtanen, R., Ilmonen, J., Saastamoinen, J., Muotka, T., 2017. An- Spalding, R.F., Exner, M.E., 1993. Occurrence of nitrate in groundwater—a review.
thropogenic habitat disturbance induces a major biodiversity change in habitat spe- J. Environ. Qual. 22, 392–402.
cialist bryophytes of boreal springs. Biol. Conserv. 215, 169–178. Stein, H., Kellermann, C., Schmidt, S.I., Brielmann, H., Steube, C., Berkhoff, S.E., Fuchs, A.,
Lehosmaa, K., Jyväsjärvi, J., Virtanen, R., Rossi, P.M., Rados, D., Chuzhekova, T., Markkola, Hahn, H.J., Thulin, B., Griebler, C., 2010. The potential use of fauna and bacteria as eco-
A., Ilmonen, J., Muotka, T., 2017. Does habitat restoration enhance spring biodiversity logical indicators for the assessment of groundwater quality. J. Environ. Monit. 12,
and ecosystem functions? Hydrobiologia 793, 161–173. 242–254.
Liu, L., Greaver, T.L., 2010. A global perspective on belowground carbon dynamics under Strayer, D.L., Dudgeon, D., 2010. Freshwater biodiversity conservation: recent progress
nitrogen enrichment. Ecol. Lett. 13, 819–828. and future challenges. J. N. Am. Benthol. Soc. 29, 344–358.
Loreau, M., Naeem, S., Inchausti, P., Bengtsson, J., Grime, J.P., Hector, A., Hooper, D.U., Thunqvist, E.-L., 2004. Regional increase of mean chloride concentration in water due to
Huston, M.A., Raffaelli, D., Schmid, B., et al., 2001. Biodiversity and ecosystem func- the application of the deicing salt. Sci. Total Environ. 325, 29–37.
tioning: current knowledge and future challenges. Science 294, 804–808. Ulvinen, T., Syrjänen, K., Anttila, S., 2002. Bryophytes of Finland: Distribution, Ecology and
Mahvi, A.H., Nouri, J., Babaei, A.A., Nabizadeh, R., 2005. Agricultural activities impact on Red List Status. vol. 560. Finnish Environment Institute, pp. 313–354.
groundwater nitrate pollution. IJEST 2, 41–47. Virtanen, R., Ilmonen, J., Paasivirta, L., Muotka, T., 2009. Community concordance between
Malaj, E., von der Ohe, P.C., Grote, M., Kühne, R., Mondy, C.P., Usseglio-Polatera, P., Brack, bryophyte and insect assemblages in boreal springs: a broad-scale study in isolated
W., Schäfer, R.B., 2014. Organic chemicals jeopardize the health of freshwater ecosys- habitats. Freshw. Biol. 54, 1651–1662.
tems on the continental scale. PNAS 111, 9549–9554. Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P.,
Martin, K.J., Rygiewicz, P.T., 2005. Fungal-specific PCR primers developed for analysis of Glidden, S., Bunn, S.E., Sullivan, C.A., Davies, P.M., 2010. Global threats to human
the ITS region of environmental DNA extracts. BMC Microbiol. 5, 3–11. water security and river biodiversity. Nature 467, 555–561.
Mauffret, A., Baran, N., Joulian, C., 2017. Effects of pesticides and metabolites on ground- Wagenhoff, A., Townsend, C.R., Phillips, N., Matthaei, C.D., 2011. Subsidy-stress and
water bacterial community. Sci. Total Environ. 576, 879–887. multiple-stressor effects along gradients of deposited fine sediment and dissolved
National Board of Waters, 1981. The Analytical Methods Used by National Board of Wa- nutrients in a regional set of streams and rivers. Freshw. Biol. 56, 1916–1936.
ters. Report 213. National Broad of Waters, Helsinki, Finland. Wakida, F.T., Lerner, D.N., 2005. Non-agricultural sources of groundwater nitrate: a re-
Nieminen, M., Sallantaus, T., Ukonmaanaho, L., Nieminen, T.M., Sarkkola, S., 2017. Nitro- view and case study. Water Res. 39, 3–16.
gen and phosphorus concentrations in discharge from drained peatland forests are Williams, D.D., Williams, N.E., Cao, Y., 2000. Road salt contamination of groundwater in a
increasing. Sci. Total Environ. 609, 974–981. major metropolitan area and development of a biological index to monitor its impact.
Nolan, J., Weber, K.A., 2015. Natural uranium contamination in major US aquifers linked Water Res. 34, 127–138.
to nitrate. Environ. Sci. Technol. Lett. 2, 215–220. Wojtal, A.Z., Sobczyk, Ł., 2012. The influence of substrates and physicochemical factors on
Odum, E.P., Finn, J.T., Franz, E.H., 1979. Perturbation theory and the subsidy-stress gradi- the composition of diatom assemblages in karst springs and their applicability in
ent. Bioscience 29, 349–352. water-quality assessment. Hydrobiologia 695, 97–108.
Oksanen, J., Blanchet, R., Kindt, P., Legendre, P.R., Minchin, R., O'Hara, R.B., Simpson, G.L., Woodward, G., Gessner, M.O., Giller, P.S., Gulis, V., Hlady, S., Lecerf, A., Malmqvist, B.,
Solymos, P., Stevens, M.P.P., Wagner, H., 2015. Vegan: Community Ecology Package McKie, B.G., Tiegs, S.D., Caris, H., et al., 2015. Continental-scale effects of nutrient pol-
Version 2.3–0. R Foundation for Statistical Computing, Vienna, Austria. lution on stream ecosystem functioning. Science 336, 1438–1440.
Paine, R.T., Tegner, M.J., Johnson, E.A., 1998. Ecological surprises. Ecosystems 1, 535–545.