Ecological Indicators 63 (2016) 240–248

Contents lists available at ScienceDirect

Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind

Quantifying the relative importance of natural variables, human

disturbance and spatial processes in ecological status indicators of
boreal lakes
Janne Alahuhta a,∗ , Jukka Aroviita b
a
Department of Geography, University of Oulu, P.O. Box 3000, FI-90014, Finland
b
Finnish Environment Institute, Freshwater Centre, P.O. Box 413, FI-90014 Oulu, Finland

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

Article history: Spatial processes are increasingly associated with species distributions in freshwaters. However, these
Received 8 June 2015 processes are usually neglected in bioassessment techniques, which may introduce uncontrolled vari-
Received in revised form 2 December 2015 ation in ecological indicators used to express human disturbance. We used partial linear regression to
Accepted 3 December 2015
quantify the relative importance of natural variables, human disturbance and spatial variables in struc-
Available online 30 December 2015
turing variation in boreal lake status indicators based on six biological indicator groups (phytoplankton,
macrophytes, diatoms, littoral and profundal macroinvertebrates and fish). We found that, of the pure
Keywords:
fractions, human disturbance explained most variation (7–32%) of the ecological quality ratios (EQRs)
Bioassessment
Ecological quality metrics for all groups, with the exception of littoral macroinvertebrate metric, which was most controlled by
Dispersal natural and spatial variables (15% and 16%, respectively). In addition, pure fractions of natural and spatial
Lakes variables and joint fractions of different explanatory variable groups structured all biological metrics
Metacommunity to various degrees. Phytoplankton, diatom and profundal macroinvertebrate EQRs responded purest to
Spatial processes human disturbance but only weakly to pure natural or spatial variation. Our work demonstrates that
spatial processes and spatial structuring of the environment can bias bioassessment techniques and hin-
der the detection of human impact. Thus, it is important to acknowledge spatial autocorrelation, context
of metacommunity dynamics, species dispersal traits and variable spatial extent when defining refer-
ence conditions and bioassessment techniques for different biological groups. More research is needed
to better understand the relative role of spatial processes on ecological metrics originated from different
freshwater ecosystems. To this end, our work provides an example of how sources of variation can be
identified to increase accuracy in freshwater bioassessment.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction importance of environmental conditions and spatial structuring

vary, as species sorting is primarily influenced by niche processes,
Species distributions are structured by local environmental con- whereas patch dynamics, mass-effects and neutrality emphasise
ditions, historical factors and spatial processes (i.e. dispersal and the significance of spatial processes (Leibold et al., 2004; Heino,
colonization-extinction dynamics), which strongly shape neigh- 2011; Soininen, 2015). Dispersal-related spatial processes can be
boring metapopulations and metacommunities (Hanski, 1998; particularly important in freshwater ecosystems which are distinct
Leibold et al., 2004; Heino et al., 2015). A suitable habitat for environments surrounded by hostile terrestrial matrix. Moreover,
the individual species occurs as a network of idealized habitat dispersal ability strongly varies among different freshwater groups;
patches, varying in area, degree of isolation and quality, and is e.g. aquatic plants disperse via wind, water and waterfowl, whereas
surrounded by equally unsuitable habitats (Hanski, 1998). Dis- fish rely only on hydrological connections (Shurin et al., 2009;
persal among habitats with different quality leads to colonization of Hajek et al., 2012). Multiple taxonomic groups are increasingly used
new suitable areas and extinctions of populations from unsuitable to assess status of freshwaters; however, the importance of spa-
habitats (Hanski, 1991, 1998). For metacommunities, the relative tial processes have only recently been taken into account in the
bioassessment (Gray and Arnott, 2011; Anas et al., 2014; Gallego
et al., 2014; Siqueira et al., 2014).
∗ Corresponding author. Tel.: +35 8294488061. In freshwater bioassessments, richness, abundance and compo-
E-mail address: janne.alahuhta@oulu.fi (J. Alahuhta). sition of several complementing taxonomical groups are typically

http://dx.doi.org/10.1016/j.ecolind.2015.12.003
1470-160X/© 2015 Elsevier Ltd. All rights reserved.
 J. Alahuhta, J. Aroviita / Ecological Indicators 63 (2016) 240–248 241

measured by indices (e.g. Hering et al., 2006; Alahuhta et al., conditions for the indicators are established. In undisturbed refer-
2009; Birk et al., 2012). Observed index values are compared to ence conditions, species should be able to disperse freely among
expected near-natural reference conditions to estimate anthro- suitable habitats, whereas impacted areas may limit dispersal
pogenic impairment (Stoddard et al., 2006) using ecological quality among habitats or, species may disperse from reference areas to
ratios (EQRs), as demanded e.g. by the European Union Water impacted ones, forming sink-source systems (Shurin et al., 2009;
Framework Directive (WFD, European Commission, 2000). The dis- Heino, 2013; Siqueira et al., 2014). Moreover, the influence of spa-
tinction of the signal due to human impairment from the natural tial processes may vary at different scales, as dispersal limitation
background variation inherent to ecosystems is thus an essen- often increases at broader scales (Heino, 2011; Alahuhta and Heino,
tial property of well performing bioassessment techniques. Ideally, 2013; Meynard et al., 2013). In an optimal bioassessment, the effect
the index EQRs should indicate only variation due to anthro- of spatial processes on the multiple taxonomic groups with differ-
pogenic disturbance, such as water quality, whereas other variation ing dispersal abilities would be acknowledged; however, they are
related to e.g. natural characteristics or spatial processes should be often neglected as assessments are carried out across large regional
accounted for in reference conditions (Hawkins, 2006). However, entities, weakening the performance of the assessment technique
this is often not the case. Residual effects of natural characteristics (Frimpong et al., 2005; Aroviita et al., 2009; Heino, 2013; Alahuhta
may, for example, mask the variation of anthropogenic disturb- et al., 2013).
ance (Aroviita et al., 2009; Jyväsjärvi et al., 2009; Ozkundakci In this study, we investigated how different indicators of lake
et al., 2014). Moreover, natural background and anthropogenic- ecological status respond to environmental gradients and spatial
originated variation in water quality forms a complex interplay, processes. Specifically, we aimed to quantify the relative roles of
because both natural (e.g. bedrock) and man-made (e.g. land use) natural environmental variation, human disturbance and spatial
sources contribute to these environmental gradients (Elser et al., processes on six status indicators (phytoplankton, macrophytes,
2007). Furthermore, natural template of the landscape (e.g. geol- diatoms, littoral macroinvertebrates, profundal macroinverte-
ogy) often determines the type of land use (e.g. agriculture, Yates brates and fishes) used in WFD bioassessment. For the purpose, we
and Bailey, 2006) which may create confounding inferences about used national bioassessment data and metrics across one large river
the true ecological status and therefore challenges the separation basin in Finland. We distinguished natural background (i.e. geol-
of natural and anthropogenic variation in bioassessment metrics. ogy and soil) and anthropogenic (i.e. land use) variation in water
Bioassessment techniques in lakes are typically based on quality, in addition to measures of space, to quantify the relative
phytoplankton, diatoms, macrophyte, macroinvertebrate and fish importance of these sources of variation on the status indicators.
communities that vary in their response to natural and anthro- We hypothesised that spatial processes and natural variables have
pogenic factors (Birk et al., 2012). Species richness and diversity a small influence on the status indicators, because they should be
are often controlled by ecosystem size so that in larger lakes the accounted for in the metric values, whereas, when present, human
higher number of habitats results to higher total species rich- disturbance (pure effect or geographically structured) should con-
ness for all biological groups (Dobson et al., 2000; Johnson and tribute highest to the metric values. We also expected, based on
Goedkoop, 2002; Rask et al., 2010; Gottschalk and Kahlert, 2012; general dispersal abilities of different biological groups studied
Alahuhta et al., 2013). Similarly, larger depth gradient increases (Soininen, 2007; Heino et al., 2012; Alahuhta et al., 2013), that
the number of habitats and further species richness and diver- spatial processes probably have some importance for diatoms,
sity of phytoplankton, diatom, macrophyte and fish (Chambers and macroinvertebrates and fish.
Kalff, 1985; Moss et al., 1997; Jyväsjärvi et al., 2009; Olin et al.,
2013). Lake volume and catchment size also primarily reflect habi-
tat availability (e.g. more lakes are present in larger catchments) 2. Material and methods
for primary producers and fish (Alahuhta et al., 2015). In addition,
higher visibility and lower colour value in water are associated with 2.1. Study area and status indicators
greater species richness and diversity for majority of taxonomical
groups, as photosynthesis is decreased in humic and deeper water Our study area spanned Vuoksi river basin in south-east Finland.
layers (Chambers and Kalff, 1985; Heino et al., 2013). The main Vuoksi is the largest river basin in Finland, covering 68,500 km2 ,
anthropogenic pressures degrading the ecological status of lakes and extends to Russia. We studied lakes in the Vuoksi river basin,
are eutrophication (nutrient pollution), acidification and hydro- because the different biological groups could be expected to dis-
morphological alterations to which the different taxonomic groups perse within a river basin and good coverage of monitoring data
may respond differently. For example, the primary producers (phy- was available from the river basin. The average flow at outlet is
toplankton, diatoms and macrophytes) typically show a strong 631 m3 s−1 and precipitation is on average 643 mm per year. Fresh-
response to increased nutrient concentrations in water (Dobson waters cover 19.2% of total area of the Vuoksi. The total number of
et al., 2000; Jeppesen et al., 2000), whereas the direct response of lakes in the Vuoksi is circa 12,600, of which 95% are small (<1 km2 ).
primary (macroinvertebrate) The lakes are typical boreal lakes, ranging from large clear-water
and secondary (fish) consumers to nutrients is often weaker lakes to shallow, polyhumic lakes (Räike et al., 2003). Other land
(Johnson and Hering, 2009), although discernible (e.g. Tolonen cover types are urban (3.5%) and agricultural areas (6.4%), forests
et al., 2001). Also location within a lake can affect how biologi- (67.3%) and wetlands (3.3%).
cal assemblages respond to anthropogenic pressures. Assemblages We used national biomonitoring data of ecological status from
in the littoral zone are often directly exposed and have strong the 2nd cycle of WFD river basin management planning in this
response to alterations in the hydrological regime (e.g. Aroviita and study. For the purpose, we gathered recent estimates of lake sta-
Hämäläinen, 2008a; Sutela et al., 2013) or local morphological con- tus as indicators (i.e. ecological quality ratios (EQR), see below) of
ditions (e.g. Brauns et al., 2007; Pilotto et al., 2015) which unlikely pelagic phytoplankton, aquatic macrophyte, littoral diatom, littoral
directly affect e.g. pelagial phytoplankton communities. macroinvertebrate, profundal macroinvertebrate and fish commu-
In addition to natural and anthropogenic characteristics, the nities of assessed lakes in Vuoksi river basin from the national
influence of spatial processes on the performance of status indices public database OIVA (http://www.ymparisto.fi/oiva) maintained
should be evaluated (Heino, 2013; Siqueira et al., 2014). The by the Finnish Environment Institute (Syke). The total number of
consideration of spatial processes in the assessment techniques assessed lakes was 134 for phytoplankton, 84 for macrophytes, 64
depends on how they are taken into account when the reference for diatoms, 61 for littoral macroinvertebrates, 102 for profundal
 242
Table 1
Descriptive statistics of EQRs and explanatory variables among the studied lake sets in Vuoksi river basin in Finland.
Phytoplankton SD Macrophytes SD Diatoms (n = 64) SD Littoral SD Profundal SD Fish SD
(n = 134) (n = 84) macroinvertebrates macroinvertebrates (n = 66)
(n = 61) (n = 102)

Mean Min Max Mean Min Max Mean Min Max Mean Min Max Mean Min Max Mean Min Max
EQR 0.8 0.2 1.0 0.2 0.8 0.4 1.0 0.1 0.7 0.1 1.0 0.2 0.8 0.3 1.0 0.2 0.8 0.3 1.0 0.2 0.8 0.3 1.0 0.2
Natural variables
TP natural 16.4 8.0 30.0 7.4 16.5 8.0 30.0 7.5 16.0 8.0 30.0 7.5 16.8 8.0 30.0 7.7 15.0 8.0 30.0 6.7 16.1 8.0 30.0 7.3
(␮g l−1 )

J. Alahuhta, J. Aroviita / Ecological Indicators 63 (2016) 240–248

TN natural 448.3 320.0 610.0 94.7 449.5 320.0 610.0 96.5 443.9 320.0 610.0 94.5 452.8 320.0 610.0 98.0 431.4 320.0 610.0 87.7 445.5 320.0 610.0 91.8
(␮g l−1 )
Color 62.6 30.0 90.0 23.7 61.8 30.0 90.0 22.8 61.4 30.0 90.0 24.0 62.0 30.0 90.0 24.2 60.6 30.0 90.0 23.8 62.7 30.0 90.0 24.3
natural
(mgPT l−1 )
Altitude 111.5 69.8 211.0 33.4 109.4 75.7 197.1 32.3 116.6 75.8 211.0 37.1 121.9 75.7 211.0 36.7 111.0 69.8 203.6 33.8 115.5 75.8 187.2 35.0
(m.a.s.l.)
Surface area 43.9 0.2 894.2 123.1 45.5 0.3 601.3 113.9 68.9 0.2 894.2 163.9 66.0 0.4 894.2 166.7 50.7 0.2 894.2 134.1 67.1 0.2 894.2 160.5
(km2 )
Mean depth 5.3 0.7 23.0 3.2 5.2 0.5 16.6 2.9 5.5 1.0 23.0 3.5 5.1 0.5 11.3 2.7 5.6 2.1 16.6 2.7 5.3 1.2 11.3 2.7
(m)
Lake 176.7 2.6 3628.7 508.5 154.6 3.0 2355.3 357.5 196.3 2.6 2355.3 428.5 178.4 4.2 2355.3 421.7 161.1 2.6 2355.3 357.3 198.2 2.6 2355.3 424.6
perimeter
(km)
Volume 462703.7 521.6 8998590.0 1421062.6 406385.8 892.0 5506690.0 1054066.1 613086.2 521.6 8998590.0 1575331.7 588757.3 976.0 8998590.0 1596763.9 451921.2 892.0 8998590.0 1289015.9 613078.0 892.0 8998590.0 1548950.5
(m3 )
Catchment 2411.7 1.1 74500.5 8731.2 2882.8 2.5 74500.5 10191.5 2096.2 1.1 49516.9 7026.4 2731.9 3.4 74500.5 11357.0 1969.2 1.1 49516.9 7065.6 1979.8 1.1 49516.9 6747.7
size (km2 )
Lakes within 9.5 0.0 43.5 8.4 10.4 0.0 43.5 9.3 13.3 0.6 43.5 9.4 14.0 0.0 43.5 9.6 10.7 0.2 43.5 8.9 11.5 0.3 43.5 9.3
catchment (%)
Anthropogenic variables
TP 5.8 0.0 70.4 11.9 7.5 0.0 70.4 14.4 4.5 0.0 61.0 11.0 3.4 0.0 42.0 7.7 4.6 0.0 43.0 8.7 4.1 0.0 43.0 9.9
man-made
−1
(␮g l )
TN 91.3 0.0 1000.7 162.3 108.1 0.0 1000.7 194.2 78.7 0.0 984.0 172.1 51.6 0.0 490.0 106.6 76.5 0.0 984.0 139.8 73.1 0.0 984.0 157.3
man-made
−1
(␮g l )
Color 29.0 0.0 410.0 51.9 29.6 0.0 410.0 60.5 25.4 0.0 180.0 35.1 31.2 0.0 180.0 42.6 23.2 0.0 180.0 34.9 28.2 0.0 310.0 50.3
man-made
(mgPT l−1 )
Urban 3.1 0.0 13.6 2.4 3.4 0.0 12.2 2.3 3.1 0.0 8.9 2.1 2.8 0.0 12.2 2.3 3.3 0.0 14.4 2.6 2.7 0.0 8.9 2.1
development
(%)
Agriculture 5.1 0.0 30.2 5.4 5.9 0.0 32.0 6.1 5.1 0.0 32.0 6.0 3.9 0.0 22.8 4.4 5.3 0.0 32.0 5.3 4.1 0.0 32.0 5.1
(%)
Peat mining 1.5 0.0 9.7 2.7 0.5 0.0 8.6 1.4 0.3 0.0 8.9 1.4 0.4 0.0 6.8 1.3 0.2 0.0 6.6 0.7 0.3 0.0 8.9 1.1
(%)
 J. Alahuhta, J. Aroviita / Ecological Indicators 63 (2016) 240–248 243

macroinvertebrates and 66 for fishes (Table 1). Twenty-three same summer (June–September) concentrations from epilimnion layer
lakes were assessed by all six groups. (0–2 m) from period 2000 to 2010 (data from the OIVA database).
The data were based on field surveys conducted in 2000–2010 Lake morphometry were obtained from OIVA database.
(mainly from 2006 to 2012) using standard methods by trained per- For human disturbance (anthropogenic water chemistry and
sonnel, and the assessments were made according to the current land use) variables, we used anthropogenic total phosphorus
Finnish WFD bioassessment guidance (Vuori et al., 2009; Aroviita (which was the concentration that exceeded the reference value,
et al., 2012). The technique comprises in each biological group com- i.e. observed value – reference value, ␮g l−1 ), anthropogenic total
prises of 2–5 taxonomic metrics that aim to portray the different nitrogen (concentration that exceeded the reference value, ␮g l−1 ),
community features that are normatively defined by the WFD to anthropogenic colour (value that exceeded the reference value,
be used for the ecological status classification. In all the groups, the mg Pt l−1 ) and proportions of land use for agriculture (%), urban
assessment is primarily based on correspondence of two features, development (%) and peat mining (%). We thus assumed that the
taxonomic composition and abundance, and to their undisturbed values above the reference in each lake type represent anthro-
reference conditions. The phytoplankton communities were sam- pogenic influence. If observed water chemistry was lower than the
pled from surface water (0–2 m) with a tube sampler (Järvinen reference value, we set it to zero for data analyses. The land use
et al., 2013) and were assessed with chlorophyll a concentration, data were compiled from CORINE land cover database (year 2006).
phytoplankton biomass, proportion of harmful blue-green algae The land use variables were prepared using ArcGIS 10.1 (Esri Corp.,
(Vuori et al., 2009) and Swedish trophic plankton index (Willen, Redlands, CA, USA).
2007). The aquatic macrophytes included hydrophytes, helophytes Spatial variables (i.e. spatial structure across the lakes within
and bryophytes and were surveyed using a main belt transect each study area) were derived from the analysis of Moran’s eigen-
method (Kanninen et al., 2013). Macrophytes were assessed with vector maps (MEM; Borcard and Legendre, 2002). Geographical
proportion of type-specific reference taxa, compositional similarity distances cannot be analyzed as locations, but they are changed
to the reference community (percent model affinity, PMA, Novak into MEMs that map neighborhood relationships at different scales.
and Bode, 1992) and a macrophyte trophic index (Alahuhta et al., Spatial variables originated from MEMs are orthogonal (linearly
2012). Littoral diatoms were sampled from stony shores with a independent) and they are obtained from spectral decomposi-
toothbrush (Gottschalk and Kahlert, 2012) and were assessed by tion of a truncated distance matrix of the spatial relationships
occurrence of type-specific reference taxa (TT, Aroviita et al., 2008b) among sampling locations. MEMs can be applied to any set of
and PMA (Aroviita et al., 2012). Littoral macroinvertebrates were locations providing a good coverage of a given geographic land-
sampled with a kick-net method from stony shores (Johnson and scape (Griffith and Peres-Neto, 2006). Significant spatial variation
Goedkoop, 2002) and were assessed with TT and PMA (Aroviita manifested in MEMs may be a consequence of missing impor-
et al., 2012). Profundal macroinvertebrates were sampled from the tant explanatory variables which show strong spatial structure,
deepest point of each lake using an Ekman sampler (Jyväsjärvi et al., dispersal limitation or historical effects on aquatic species (Dray
2009) and were assessed with Profundal Invertebrate Community et al., 2012). MEMs are able to address complex patterns of spa-
Metric (PICM) and PMA (Jyväsjärvi et al., 2012, 2014). Pelagic fish tial variation: the first eigenvectors often represent large-scale
communities were surveyed using stratified random sampling with variation, whereas the ones with small eigenvalues should rep-
Nordic gillnets and assessed with total biomass, total number of resent finer-scale variation (Diniz-Filho and Bini, 2005). In this
individuals, cyprinid biomass % and occurrence of indicator species study, we used geographical coordinates of lake centres to calculate
(Olin et al., 2013). Euclidean distances between lakes and only positive eigenvec-
For each taxonomic group and lake, the EQR was a mean value tors based on minimum truncation distances were employed. In
calculated over normalized EQRs of the corresponding metrics fol- boreal region with glaciation-originated freshwaters and relatively
lowing the current national guidelines (Vuori et al., 2009; Aroviita stable environmental conditions, MEMs using Euclidean distances
et al., 2012). The EQR ranges from 0 (bad status) to 1 (high) and rep- have proved to capture complex spatial patterns equally well com-
resents observed/expected reference value demanded by the WFD pared MEMs based on watercourse distances (Heino et al., 2012;
to indicate anthropogenic impairment of the biota. Ideally, the only Grönroos et al., 2013; Alahuhta et al., 2015). MEM analyses were
variation in EQRs should be related to anthropogenic disturbance run separately for each lake sets (i.e. for phytoplankton, macro-
or sampling error (Hawkins, 2006). However, uncontrolled varia- phyte, diatom, littoral and profundal macroinvertebrate and fish
tion due to natural factors or spatial processes may result in EQR study lakes). db-MEMs were constructed using the pcnm package
variation that is not due to anthropogenic activities and may lead in the R environment (Dray et al., 2012).
in misinterpretation of the ecological status. We specifically exam-
ined the extent to which natural and anthropogenic environmental 2.3. Statistical analyses
variation and space influence the EQR variation (see below).
Because the number of lakes varied among the biological groups,
2.2. Explanatory variables we first investigated whether the average environmental con-
ditions and heterogeneity of environmental conditions differed
Three sets of explanatory variables were used: (i) natural vari- among the study lake sets. Average differences are mean dif-
ables (i.e. lake morphometry and estimated natural background ferences in environmental conditions among the study lake sets
variation in water chemistry), (ii) human disturbances (i.e. land in multivariate space, whereas heterogeneity differences refer to
use and water chemistry estimated to be altered by human activity, variation in environmental conditions within each study lake set
Table 1) and (iii) spatial variables. Natural variables included natu- and how this variation differs among the lake sets (Heino et al.,
ral total phosphorus concentration (␮g l−1 ), natural total nitrogen 2013). If the environmental heterogeneity varies strongly among
concentration (␮g l−1 ), natural water colour (mg Pt−1 ), lake surface the study lake sets, then direct comparisons of results among the
area (km2 ), mean depth (m), lake perimeter (km), volume (m3 ), lake sets may be inappropriate. However, if the heterogeneity is
catchment size (km2 ) and proportion of lakes within the catch- similar among the study lake sets, the comparison of results is
ment (%). For the estimates of natural background concentrations more meaningful. Average environmental conditions were inves-
of total phosphorus, total nitrogen and water colour, we used lake tigated using Constrained Analysis of Principal Coordinates (CAP;
type specific expected reference values from the current national Anderson and Willis, 2003) with Euclidean distance and standard-
guidelines (Vuori et al., 2009). All water chemistry was median ized environmental variables. We ran an ANOVA-like permutation
244 J. Alahuhta, J. Aroviita / Ecological Indicators 63 (2016) 240–248

(199 permutations) to test for the significance of the separation fish EQR (Table 2). Pure anthropogenic disturbance had the highest
of lake sets in multivariate space. Environmental heterogeneity effect on phytoplankton (32%), macrophyte (19%), diatom (13%) and
across the lakes within each study lake set was studied using aver- fish (16%) status. For littoral macroinvertebrates, pure influences
age distances to a region’s centre and the tests of homogeneity of of natural and spatial variables were equally important (16% and
multivariate dispersion (PERMDISP; Anderson, 2006). PERMDISP 15%, respectively). In addition, joint fraction of natural and spatial
utilizes the ANOVA F-statistic to compare among-group (i.e. study variables contributed strongly to phytoplankton (31%), macrophyte
lake sets) differences in the distance from observations to their (20%) and littoral macroinvertebrate (8%) status. Diatom (18%) and
group centroid and is a multivariate analogue of Levene’s test both macroinvertebrate (littoral: 10%, profundal: 10%) assessments
for homogeneity of variances. Significance of among-study lake were also affected by the joint effect of human disturbance and
set differences was tested through permutation of least-squares spatial variables.
residuals. We conducted environment variable analysis based on Of the natural variables, catchment size was the most sig-
Euclidean distances on standardized variables. In each analysis, sta- nificant variable explaining the variation in phytoplankton EQR,
tistical significance was assessed based on permutation tests with natural colour was the most important variable for macrophyte,
999 runs. In both CAP and PERMDISP analysis, we used – besides of lake perimeter for diatom and mean depth for fish EQR (Table 3). For
human disturbance and land use variables – original concentrations littoral macroinvertebrates, natural TP, mean depth, catchment size
of total phosphorus, total nitrogen and colour. CAP and PERMDISP and lake area were equally important natural variables. No natural
analysis were performed with vegan package in R (Oksanen et al., variables explained the variation in profundal macroinvertebrate
2012). EQR. Considering anthropogenic disturbance, anthropogenic TP
We used partial linear regression (pLR) to distinguish the rela- had the strongest effect on phytoplankton indicator and anthro-
tionships between the variation in the EQRs and the explanatory pogenic TN on macrophyte and fish indicator. Proportion of
variable groups and to explore the extent to which EQR values were agriculture contributed solely to diatom and both macroinver-
explained by natural, antropogenic and spatial factors. The protocol tebrate EQRs. The spatial structure varied among data sets due
of Anderson and Gribble (1998) was followed for pLRs. Total varia- to different number of lakes, for which direct comparison of
tion in the status indicators was partitioned into eight fractions: (a) individual spatial variables is not meaningful. However, various
pure effect of natural variables, (b) pure effect of human disturb- spatial variables affected phytoplankton, macrophyte and littoral
ance, (c) pure effect of space; combined variation due to the joint macroinvertebrate EQR, whereas first spatial variables represent-
effects of (d) natural variables and human disturbance, (e) natural ing large-scale spatial patterns were most important for diatom and
variables and space, (f) human disturbance and space, and (g) the profundal macroinvertebrate indicators. This suggests that spatial
three groups of explanatory variables; and finally (h) unexplained patterns operating at the whole catchment are important for these
variation. The detailed procedures to estimate these fractions are metrics, whereas patterns functioning at different spatial scales
explained elsewhere (Legendre et al., 2005; Borcard et al., 2011). structure other metrics. No spatial variables were selected for fish
Response variables were square root arcsine-transformed prior to EQR.
the analysis to improve normality.
Variation explained by each variable group was evaluated with
adjusted R2 , which provides unbiased estimates of the explained 4. Discussion
variation (Peres-Neto et al., 2006). In forward selection, type I error
can be avoided by using adjusted R2 values, which are also compara- 4.1. Response of the status indicators to the human disturbance
ble between different models as the number of explanatory variable
is taken into account (Blanchet et al., 2008). Following the proce- In an ideal bioassessment technique based on the reference
dure of Blanchet et al. (2008), the forward selection was carried condition approach, the status indices should only respond to
out only if a global test using all explanatory variables in a variable human disturbance, whereas natural background variation due
set was significant. Forward selection using the Monte Carlo per- to geographical characteristics and spatial processes ought to
mutation test (999 permutations, ˛ = 0.05) was then used to obtain be accounted for in the reference values (Hawkins, 2006). We
significant variables for further analysis. All pLRs were performed found that most of the lake status metrics of the environmentally
in the R environment with Packfor and Vegan packages (Oksanen homogenous set of lakes responded primarily to human disturb-
et al., 2012). ance, although a large share of the variation was also attributed
to natural variation or space, or geographically structured human
disturbance. The metrics were strongly negatively associated with
3. Results anthropogenic water quality (increased total phosphorus, total
nitrogen and colour) and agriculture (see Table 3), emphasizing that
CAP analysis indicated that average environmental conditions generally the separation of water quality to natural background and
did not differ among the lake sets (F = 0.949, p = 0.52) and that the anthropogenic-originated components performed relatively well.
environmental heterogeneity was not significantly different among The pure human disturbance explained most variation in phy-
the lake sets (F = 0.500, p = 0.78). Variation in anthropogenic total toplankton, macrophyte, diatom and fish metrics, suggesting that
phosphorus and nitrogen were strongly positively correlated to their basis was adequately founded in the bioassessment tech-
agriculture, whereas natural total phosphorus and nitrogen cor- niques. By contrast, especially the littoral macroinvertebrate metric
related weakly negatively to agricultural land use (Appendix S1 was influenced weakly by pure fraction of human disturbance and
in Supporting Information). Urban development was mainly nega- strongly by pure fractions of natural variables and space, which
tively correlated with natural nutrient variables and positively with indicates important implications for improving the bioassessment
anthropogenic total nitrogen and phosphorus. Water colour was technique (see below). Generally the results concerning response
poorly associated to land use, whereas peat mining was weakly of the status indicators to the human disturbance were expected, as
positively connected to all water chemistry variables across the the primary producers (phytoplankton, macrophytes and diatoms)
lake sets. respond most directly to changes in nutrients (Smith et al., 2006),
The total variation explained was 65% of the phytoplankton, 52% the main stressor for lakes in Vuoksi river basin. Furthermore,
of the macrophyte, 43% of the diatom, 50% of the littoral macroin- the fish metric is developed to indicate impacts of eutrophica-
vertebrate, 28% of the profundal macroinvertebrate and 26% of the tion on the fish communities (Rask et al., 2010), which explains
 J. Alahuhta, J. Aroviita / Ecological Indicators 63 (2016) 240–248 245

Table 2
Results of variation partitioning using adjusted R2 based on partial linear regression (pLR) with a strict forward stepwise selection criterion (Blanchet et al., 2008). Variation
in the status metrics is divided among natural variables, human disturbance and spatial variables. The significant pure fractions based on permutation-like test are in italic
(p < 0.05). MI: macroinvertebrate.

Phytoplankton Macrophytes Diatoms Littoral MI Profundal MI Fish

Natural variables (NC) 0.01 0.02 0.04 0.16 0.00 0.09

Human disturbance (HD) 0.32 0.19 0.13 0.05 0.07 0.16
Spatial variables (SV) 0.00 0.06 0.04 0.15 0.05 0.00
NC and HD 0.00 0.02 0.00 0.00 0.00 0.01
NC and SV 0.31 0.20 0.01 0.08 0.00 0.00
HD and SV 0.00 0.00 0.18 0.10 0.10 0.00
NC and HD and SV 0.02 0.02 0.02 0.00 0.00 0.00
Unexplained 0.35 0.48 0.57 0.50 0.78 0.74

Table 3
The significant variables explaining variation in status metrics based on adjusted R2 selection method in pursuance of partial linear models (pLR). +/− in brackets indicates
the direction of relationship. SV, spatial variables derived from distance-based Moran’s eigenvector maps (db-MEMs). The first spatial variables often represent large-scale
variation, whereas the large ones should represent finer-scale variation (Diniz-Filho and Bini, 2005). nv, no variables selected.

Phytoplankton Macrophytes Diatom

2 2
Variable Adj R p Variable Adj R p Variable Adj R2 p

Natural variables Catchment size (−) 0.025 0.038 Natural colour (−) 0.065 0.009 Lake perimeter (−) 0.209 0.001
Human disturbance Anthropogenic TP (−) 0.541 0.001 Anthropogenic TP (−) 0.443 0.001 Agriculture% (−) 0.337 0.001
Agriculture% (−) 0.086 0.001 Anthropogenic TN (−) 0.027 0.027
Anthropogenic TN (−) 0.022 0.004 Anthropogenic colour (−)
Spatial variables SV6 0.158 0.001 SV17 0.082 0.005 SV2 0.202 0.003
SV19 0.061 0.001 SV16 0.056 0.025 SV6 0.056 0.018
SV5 0.042 0.003 SV6 0.046 0.025
SV9 0.036 0.006 SV5 0.037 0.027
SV8 0.025 0.015 SV12 0.035 0.030
SV14 0.031 0.040

Littoral macroinvertebrates Profundal macroinvertebrates Fish

2 2
Variable Adj R p Variable Ad j R p Variable Adj R2 p

Natural variables Natural TP (−) 0.055 0.047 nv Mean depth (+) 0.105 0.003
Mean depth (+) 0.053 0.038
Catchment area (−) 0.061 0.032
Lake area (−) 0.051 0.045
Human disturbance Agriculture% (−) 0.085 0.012 Agriculture% (−) 0.166 0.001 Anthropogenic TN (−) 0.167 0.001
Spatial variables SV12 0.144 0.001 SV3 0.083 0.003 nv
SV5 0.081 0.013 SV6 0.071 0.001
SV16 0.050 0.031
SV15 0.044 0.029

the relatively high contribution of nutrients (namely total nitro- site-specific modelling of PICM reference values (Jyväsjärvi et al.,
gen). The profundal macroinvertebrate metric was also affected by 2014). However, relatively high proportion of unexplained varia-
the human disturbance (agriculture) or spatially structured human tion for profundal macroinvertebrate, in addition to fish, metric
disturbance (Table 2), indicating that lakes with poor status are suggests that some important natural variables may be missing
situated close to each other in agriculture dominated part of the from our models.
Vuoksi river basin. Similarly, Alahuhta et al. (2013) discovered that The macrophyte and phytoplankton metrics were neither con-
macrophyte-based lake status metrics were strongly affected by siderably influenced by pure natural or spatial variation, but
anthropogenic-originated water quality variables, showing geo- were strongly influenced by spatially structured natural variation.
graphical centralizing in agricultural dominated lowland areas in Background variation of natural colour and catchment size influ-
southern and central Finland. These findings emphasize the fact enced the variation of macrophyte and phytoplankton metrics,
that the natural template of the landscape often influences the type respectively, as pLR models indicated that macrophyte-EQR had a
of land use which hinders the complete separation of natural and negative relationship with natural colour and the phytoplankton-
anthropogenic causes of variation in bioassessment metrics. EQR decreased with increasing catchment size. The macrophyte
relationship likely reflects species response to restricted light con-
4.2. Response of the status indicators to the natural and spatial ditions in lakes with high input of organic matter from peatland
variables dominated catchments, limiting their vertical habitat availability
(Chambers and Kalff, 1985). The phytoplankton-EQR decrease with
None of the status metrics performed ideally, however, as increasing catchment size is also related to habitat availability.
natural and/or spatial variables influenced all of them. Despite Number of water bodies is higher in larger catchments compared
some influence of pure space, profundal macroinvertebrate met- to smaller ones, reflecting that there are more potential freshwater
ric seemed to perform well in the sense that natural variables bodies available for diatoms in larger catchments to grow and to
did not have an independent effect on the EQR of the PICM. disperse propagules in downstream ecosystems (Vanormelingen
Natural variation of profundal macroinvertebrates assemblages et al., 2008). However, the negative response of phytoplankton
is strongly driven by basin depth (Jyväsjärvi et al., 2009), and metric to catchment size was somewhat surprising as it contra-
the unbiased performance is likely due to the fact that the dicts with the current knowledge that more habitats are present
depth-related variation is efficiently taken into account with for phytoplankton in larger lakes (Stomp et al., 2011).
246 J. Alahuhta, J. Aroviita / Ecological Indicators 63 (2016) 240–248

Similarly to phytoplankton metric, also diatom metric values traits (e.g. De Bie et al., 2012). In addition, due to different dispersal
decreased with increasing lake perimeter. Although human dis- abilities among species, they can disperse variable distances, for
turbance primarily controlled the fish metric, the variation in which spatial extent used to define reference conditions may not
natural variables (mean depth) also affected this metric. As shal- be consistent across biological groups. For example, fish disperse
low lakes suffer more often from eutrophication than deeper ones only through watercourses within a river basin, whereas macroin-
(Moss et al., 1997), the positive association between fish-EQR and vertebrates may actively fly or passively disperse by wind across
mean depth suggests that depth gradient in lakes indirectly reflects river basins and macrophyte propagules are transported long dis-
changes in trophic status. Increased depth also means more habi- tances via waterfowl (Hitt and Angermeier, 2008; Grönroos et al.,
tats for different fish species to co-exist. Thus, the results indicated 2013; Viana et al., 2013). This suggests that a river basin is a natural
that variation in natural colour for macrophyte and lake depth for spatial area for fish, whereas larger regional areas may be suitable
fish should better be taken into account in bioassessment when for other biological groups. Furthermore, natural dispersal obsta-
defining the reference values for these biological groups. cles related to stream network (e.g. beaver dams) or landscape (e.g.
Littoral macroinvertebrate EQR was strongly influenced by pure mountains) should be acknowledged when defining the reference
fractions of both natural and spatial variables, which likely resulted conditions.
to the observed low pure fraction of human disturbance and the
weak relationship with human disturbance (agriculture). The EQR
5. Conclusions
was affected by natural TP, depth, catchment size and lake area
which indicates that variation related to natural variation in trophic
Our results suggest that the basis of the lake bioassessment tech-
status and lake size is not sufficiently controlled for in the assess-
nique was well grounded for phytoplankton, diatom and profundal
ment technique. Moreover, the high independent effect of spatial
macroinvertebrate metrics, which responded purest to human dis-
variables indicates that (1) either some unmeasured spatially struc-
turbance but only weakly to pure natural or spatial variation.
tured natural or human disturbance variable(s) is missing from the
However, natural and spatial variables affected all metrics, and
models, or (2) some ecological processes related to e.g. dispersal
some were also influenced by geographically structured human
influence on the metrics (Dray et al., 2012; Siqueira et al., 2014).
disturbance, suggesting that the reference values could be more
Macroinvertebrate species generally disperse actively among suit-
accurately estimated. Despite the findings are based on a WFD
able habitats, but species with weakest aerial dispersal capabilities
bioassessment technique relying on taxonomic metrics from one
seem to be most structured by spatial-related processes (Heino
country (Finland) and from a restricted geographical area (Vuoksi
et al., 2012; Grönroos et al., 2013). In addition, macrophyte and
River Basin), the results have important general repercussions
diatom metrics were also controlled by spatial variables to some
for refining bioassessment also in other regions. Generally, the
extent. Dispersal limitation rarely affects macrophytes at a river
significance of spatial processes is recognized in population and
basin scale (O’Hare et al., 2012; Alahuhta and Heino, 2013; Alahuhta
community ecology (Hanski, 1998; Leibold et al., 2004; Heino et al.,
et al., 2014), suggesting that some spatially structured environmen-
2015) and should potentially be taken more explicitly into account
tal variable related to e.g. sediment quality or fetch is missing from
when defining expected reference values. We recommend that
our data. Instead, spatial variables have been found to considerably
the importance of spatial processes in different metrics in var-
structure diatoms at a similar geographical scale as in our study
ious freshwater ecosystems should be quantified. One first step
(Soininen, 2007; Heino et al., 2010). It may thus well be that some
would be to apply statistical proxies developed to represent dis-
diatom and macroinvertebrate species show dispersal limitation,
persal, such as MEMs, to investigate the relative role of spatial
which is not adequately accounted for in these metrics. Interest-
processes on ecological metrics originated from different fresh-
ingly, despite fish are highly dependent on streams for dispersal,
water ecosystems (see Dray et al., 2012 for more on techniques
spatial processes seem to be sufficiently accounted in their status
indicating spatial processes and their limitations). Recognizing spa-
metric. More research is needed, however, to confirm this prelim-
tial autocorrelation, context of metacommunity dynamics, species
inary finding due to relatively low overall explained variation.
dispersal-related traits and variable area size (i.e. spatial extent)
There are at least three issues related to spatial processes that
from which reference conditions are defined would form a sophis-
should be evaluated and taken into account in bioassessment tech-
ticated basis for further development of freshwater bioassessment.
niques. First, spatial autocorrelation is inherently related to all
This would also help to scrutinize the degree of true signal and bias
natural systems, meaning that lakes geographically close to each
in the assessment to better estimate levels of accuracy and confi-
other are likely more similar compared to distant ones (Legendre,
dence for the assessments (Hawkins, 2006). To this end, our work
1993). In bioassessment techniques operating across large geo-
provides an example how sources of the variation can be identified
graphical scales, the existence of spatial autocorrelation should be
and quantified in the bioassessment metrics.
acknowledged when selecting potential reference sites (see also
Shurin et al., 2009). Secondly, contemporary freshwater assessment
protocols are almost solely based on the idea of species sorting, Acknowledgements
in which environment primarily controls variation in biological
groups (Leibold et al., 2004; Heino, 2013), and have yet not taken We thank North Savo, South Savo and North Karelia Centres for
influence of dispersal into account. In a recent predictive mod- Economic Development, Transport and the Environment for col-
elling simulation, Siqueira et al. (2014) showed that mass effects lecting parts of the data and for the efforts to assess the status
may hinder detection of local environmental changes, because of lakes in the Vuoksi basin. We especially thank Antti Kanni-
species dispersal from a neighboring pristine site resulted in occur- nen for compiling part of the data. We acknowledge the efforts
rence of a pollution-intolerant species at the polluted water body. of several researchers for striving to develop lake bioassessment in
The simulation indicated that high, or oppositely low, dispersal Finland (among others; Seppo Hellsten, Heikki Hämäläinen, Jussi
can mask species sorting at polluted sites. Therefore, geographical Jyväsjärvi, Marko Järvinen, Satu Maaria Karjalainen, Minna Kuop-
arrangement of reference sites may have important implications pala, Sari Mitikka, Heikki Mykrä, Mikko Olin, Ansa Pilke, Martti
for the construction of predictive models or other approaches Rask, Kari-Matti Vuori). Parts of the data were collected in MaaMet-
when defining reference values for biological indicators. Thirdly, monitoring funded by the Finnish Ministry of Agriculture and
species abilities to disperse among suitable habitats vary within Forestry. The research was supported by the REFORM (contract No.
and between biological groups, i.e. species have different dispersal 282656) and MARS (contract No. 603378) projects funded under
 J. Alahuhta, J. Aroviita / Ecological Indicators 63 (2016) 240–248 247

the 7th Framework Programme by the European Union. Com- T., Munoz, F., Oksanen, J., Thioulouse, J., Wagner, H.H., 2012. Community ecology
ments by two anonymous reviewers considerably improved the in the age of multivariate multiscale spatial analysis. Ecol. Monogr. 82, 257–275.
Elser, J.J., Bracken, M.E.S., Cleland, E.E., Gruner, D.S., Harpone, W.S., Hillebrand, H.,
manuscript. Ngai, J.T., Seabloom, E.W., Shurin, J.B., Smith, J.E., 2007. Global analysis of nitro-
gen and phosphorus limitation of primary producers in freshwater, marine and
terrestrial ecosystems. Ecol. Lett. 10, 1135–1142.
Appendix A. Supplementary data European Commission, 2000. Directive 2000/60/EC of the European Parliament and
of the Council of 23 October 2000. Establishing a framework for Community
Supplementary data associated with this article can be found, action in the field of water policy. Off. J. Eur. Commun. 327, 1–72.
Frimpong, E.A., Sutton, T.M., Engel, B.A., Simon, T.P., 2005. Spatial-scale effects on
in the online version, at http://dx.doi.org/10.1016/j.ecolind.2015. relative importance of physical habitat predictors of stream health. Environ.
12.003. Manag. 36, 899–917.
Gallego, I., Davidson, T.A., Jeppesen, E., Perez-Martinez, C., Fuentes-Rodriguez, F.,
Juan, M., Casas, J.J., 2014. Disturbance from pond management obscures local
References and regional drivers of assemblages of primary producers. Freshw. Biol. 59,
1406–1422.
Alahuhta, J., Heino, J., 2013. Spatial extent, regional specificity and metacommunity Gottschalk, S., Kahlert, M., 2012. Shifts in taxonomical and guild composition of
structuring in lake macrophytes. J. Biogeogr. 40, 1572–1582. littoral diatom assemblages along environmental gradients. Hydrobiologia 694,
Alahuhta, J., Rääpysjärvi, J., Hellsten, S., Kuoppala, M., Aroviita, J., 2015. Species sor- 41–56.
ting drives variation of boreal lake and river macrophyte communities. Commun. Gray, D.K., Arnott, S.E., 2011. Does dispersal limitation impact the recovery of
Ecol. 16, 76–85. zooplankton communities damaged by a regional stressor. Ecol. Appl. 21,
Alahuhta, J., Johnson, L.B., Olker, J., Heino, J., 2014. Species sorting determines varia- 1241–1256.
tion in the community composition of common and rare macrophytes at various Griffith, D.A., Peres-Neto, P.R., 2006. Spatial modeling in ecology: the flexibility of
spatial extents. Ecol. Complex. 20, 61–68. eigenfunction spatial analyses. Ecology 87, 2603–2613.
Alahuhta, J., Kanninen, A., Hellsten, S., Vuori, K.-M., Kuoppala, M., Hämäläinen, H., Grönroos, M., Heino, J., Siqueira, T., Landeiro, V.L., Kotanen, J., Bini, L.M., 2013. Meta-
2013. Environmental and spatial correlates of community composition, richness community structuring in stream networks: roles of dispersal mode, distance
and status of boreal lake macrophytes. Ecol. Indic. 32, 172–181. type, and regional environmental context. Ecol. Evol. 3, 4473–4487.
Alahuhta, J., Kanninen, A., Vuori, K.-M., 2012. Response of macrophyte communities Hajek, M., Rolecek, J., Cottenie, K., Kintrova, K., Horsa, M., Poulickova, A., Hajkova,
and status metrics to natural gradients and land use in boreal lakes. Aquat. Bot. P., Frankova, M., Dite, D., 2012. Environmental and spatial controls of biotic
103, 106–114. assemblages in a discrete semi-terrestrial habitat: comparison of organisms
Alahuhta, J., Vuori, K.-M., Hellsten, S., Järvinen, M., Olin, M., Palomäki, A., Rask, M., with different dispersal abilities sampled in the same plots. J. Biogeogr. 38,
2009. Defining ecological status of small forest lakes using multiple biological 1683–1693.
quality elements and paleolimnological analysis. Fundam. Appl. Limnol. 175, Hanski, I., 1991. Single-species metapopulation dynamics: concepts, models and
203–216. observations. Biol. J. Linn. Soc. 42, 17–38.
Anas, M.U.M., Scott, K.A., Cooper, R.N., Wissel, B., 2014. Zooplankton communities Hanski, I., 1998. Metapopulation dynamics. Nature 396, 41–49.
are good indicators of potential impacts of Athabasca oil sands operations on Hawkins, C.P., 2006. Quantifying biological integrity by taxonomic completeness:
downwind boreal lakes. Can. J. Fish. Aquat. Sci. 71, 719–732. its utility in regional and global assessments. Ecol. Appl. 16, 1277–1294.
Anderson, M.J., 2006. Distance-based tests for homogeneity of multivariate disper- Heino, J., 2013. The importance of metacommunity ecology for environmental
sions. Biometrics 62, 245–253. assessment research in the freshwater realm. Biol. Rev. 88, 166–178.
Anderson, M.J., Willis, T.J., 2003. Canonical analysis of principal coordinates: a useful Heino, J., 2011. A macroecological perspective of diversity patterns in the freshwater
method of constrained ordination for ecology. Ecology 84, 511–525. realm. Freshw. Biol. 56, 1703–1722.
Anderson, M.J., Gribble, N.A., 1998. Partitioning the variation among spatial, tempo- Heino, J., Soininen, J., Alahuhta, J., Lappalainen, J., Virtanen, R., 2015. A compar-
ral and environmental components in a multivariate data set. Aust. J. Ecol. 23, ative analysis of metacommunity types and their ecological correlates in the
158–167. freshwater realm. Ecol. Evol. 5, 1525–1537.
Aroviita, J., Hellsten, S., Jyväsjärvi, J., Järvenpää, L., Järvinen, M., Karjalainen, S.- Heino, J., Grönroos, M., Ilmonen, J., Karhu, T., Niva, M., Paasivirta, L., 2013.
M., Kauppila, P., Keto, A., Kuoppala, M., Manni, K., Mannio, J., Mitikka, S., Olin, Environmental heterogeneity and beta diversity of stream macroinvertebrate
M., Perus, J., Pilke, A., Rask, M., Riihimäki, J., Ruuskanen, A., Siimes, K., Sutela, communities at intermediate spatial scales. Freshw. Sci. 32, 142–154.
T., Vehanen, T., Vuori, K.-M., 2012. Ohje pintavesien ekologisen ja kemiallisen Heino, J., Grönroos, M., Soininen, J., Virtanen, R., Muotka, T., 2012. Context depend-
tilan luokitteluun vuosille 2012–2013 – päivitetyt arviointiperusteet ja niiden ency and metacommunity structuring in boreal headwater streams. Oikos 121,
soveltaminen (Guidelines for the Ecological and Chemical Status Classification of 537–544.
Surface Waters for 2012–2013 – Updated Assessment Criteria and Their Appli- Heino, J., Bini, L.M., Karjalainen, S.M., Mykrä, H., Soininen, J., Vieira, L.C.G., Diniz-Filho,
cation). Environmental Administration Guidelines 7/2012), 144 pp. J.A.F., 2010. Geographical patterns of micro-organismal community struc-
Aroviita, J., Mykrä, H., Muotka, T., Hämäläinen, H., 2009. Influence of geographical ture: are diatoms ubiquitously distributed across boreal streams? Oikos 119,
extent on typology- and model-based assessments of taxonomic completeness 129–137.
of river macroinvertebrates. Freshw. Biol. 54, 1774–1787. Hering, D., Feld, C.K., Moog, O., Ofenböck, T., 2006. Cook book for the development
Aroviita, J., Hämäläinen, H., 2008a. The impact of water-level regulation on littoral of a Multimetric Index for biological condition of aquatic ecosystems: expe-
macroinvertebrate assemblages in boreal lakes. Hydrobiologia 613, 45–56. riences from the European AQEM and STAR projects and related initiatives.
Aroviita, J., Koskenniemi, E., Kotanen, J., Hämäläinen, H., 2008b. A priori typology- Hydrobiologia 566, 311–324.
based prediction of benthic macroinvertebrate fauna for ecological classification Hitt, N.P., Angermeier, P.L., 2008. Evidence for fish dispersal from spatial analysis of
of rivers. Environ. Manag. 42, 894–906. stream network topology. J. North Am. Benthol. Soc. 27, 304–320.
Birk, S., Bonne, W., Borja, A., Brucet, S., Courrat, A., Poikane, S., Solimini, A.G., van Jeppesen, E., Jensen, J.P., Søndergaard, M., Lauridsen, T., Landkildehus, F., 2000.
de Bund, W., Zampoukas, N., Hering, D., 2012. Three hundred ways to assess Trophic structure, species richness and biodiversity in Danish lakes: changes
Europe’s surface waters: an almost complete overview of biological methods to along a phosphorus gradient. Freshw. Biol. 45, 201–218.
implement the Water Framework Directive. Ecol. Indic. 18, 31–41. Johnson, R.K., Goedkoop, W., 2002. Littoral macroinvertebrate communities: spatial
Blanchet, F.G., Legendre, P., Borcard, D., 2008. Forward selection of explanatory scale and ecological relationships. Freshw. Biol. 47, 1840–1854.
variables. Ecography 89, 2623–2632. Johnson, R.K., Hering, D., 2009. Response of taxonomic groups in streams to gradients
Borcard, D., Legendre, P., 2002. All-scale spatial analysis of ecological data by means in resource and habitat characteristics. J. Appl. Ecol. 46, 175–186.
of principal coordinates of neighbour matrices. Ecol. Model. 153, 51–68. Jyväsjärvi, J., Aroviita, J., Hämäläinen, H., 2014. An extended Benthic Quality Index
Borcard, D., Gillet, F., Legendre, P., 2011. Numerical Ecology with R. Springer, New for assessment of lake profundal macroinvertebrates: addition of indicator taxa
York. by multivariate ordination and weighted averaging. Freshw. Sci. 33, 995–1007.
Brauns, M., Garcia, X.-F., Walz, N., Pusch, M.T., 2007. Effects of human shoreline Jyväsjärvi, J., Aroviita, J., Hämäläinen, H., 2012. Performance of profundal macroin-
development on littoral macroinvertebrates in lowland lakes. J. Appl. Ecol. 44, vertebrate assessment in boreal lakes depends on lake depth. Fundam. Appl.
1138–1144. Limnol. 180, 91–100.
Chambers, P., Kalff, J., 1985. Depth distribution and biomass of submerged aquatic Jyväsjärvi, J., Tolonen, K.T., Hämäläinen, H., 2009. Natural variation of profundal
macrophyte communities in relation to Secchi depth. Can. J. Fish. Aquat. Sci. 42, macroinvertebrate communities in boreal lakes is related to lake morphometry:
701–709. implications for bioassessment. Can. J. Fish. Aquat. Sci. 66, 589–601.
De Bie, T., De Meester, L., Brendonck, L., Martens, K., Goddeeris, B., Ercken, D., Hampel, Järvinen, M., Drakare, S., Free, G., Lyche-Solhem, A., Phillips, G., Skjelbred, B., Mischke,
H., Denys, L., Vanhecke, L., Van der Gucht, K., Van Wichelen, J., Vyverman, W., U., Ott, I., Poikane, S., Søndergaard, M., Pasztaleniec, A., Van Wichelen, J., Portielje,
Declerck, S.A.J., 2012. Body size and dispersal mode as key traits determining R., 2013. Phytoplankton indicator taxa for reference conditions in Northern and
metacommunity structure of aquatic organisms. Ecol. Lett. 15, 740–747. Central European lowland lakes. Hydrobiologia 704, 97–113.
Diniz-Filho, J.A.F., Bini, L.M., 2005. Modelling geographical patterns in species rich- Kanninen, A., Vallinkoski, V.-M., Leka, J., Marjomäki, T., Hellsten, S., Hämäläinen,
ness using eigenvector based spatial filters. Glob. Ecol. Biogeogr. 14, 177–185. H., 2013. A comparison of two methods for surveying aquatic macrophyte
Dobson, S.I., Arnott, S.E., Cottingham, K.L., 2000. The relationship in lake communi- communities in boreal lakes: implications for bioassessment. Aquat. Bot. 104,
ties between primary productivity and species richness. Ecology 81, 2662–2679. 88–100.
Dray, S., Pélissier, R., Couteron, P., Fortin, M.J., Legendre, P., Peres-Neto, P.R., Bellier, Legendre, P., 1993. Spatial autocorrelation – trouble or new paradigm. Ecology 74,
E., Bivand, R., Blanchet, F.G., De Caceres, M., Dufour, A.B., Heegaard, E., Jombart, 1659–1673.
248 J. Alahuhta, J. Aroviita / Ecological Indicators 63 (2016) 240–248

Legendre, P., Borcard, D., Peres-Neto, P.R., 2005. Analyzing beta diversity: parti- Shurin, J.B., Cottenie, K., Hillebrand, H., 2009. Spatial autocorrelation and dispersal
tioning the spatial variation of community composition data. Ecol. Monogr. 75, limitation in freshwater organisms. Oecologia 159, 151–159.
435–450. Siqueira, T., Durães, L.D., de Oliveira Roque, F., 2014. Predictive modelling of
Leibold, M.A., Holyoak, M., Mouquet, N., Amarasekare, P., Chase, J.M., Hoopes, M.F., insect metacommunities in biomonitoring of aquatic networks. In: Ferreira,
Holt, R.D., Shurin, J.B., Law, R., Tilman, D., Loreau, M., Gonzalez, A., 2004. The C.P., Godoy, W.A.C. (Eds.), Ecological Modelling Applied to Entomology. Springer
metacommunity concept: a framework for multi-scale community ecology. International Publishing, New York, pp. 109–126.
Ecol. Lett. 7, 601–613. Smith, V.H., Joye, S.B., Howarth, R.W., 2006. Eutrophication of freshwater and marine
Meynard, C.N., Lavergen, S., Boulangeat, I., Garraud, L., Van Es, J., Mouquet, N., ecosystems. Limnol. Oceanogr. 51, 351–355.
Thuiller, W., 2013. Disentangling the drivers of metacommunity structure across Soininen, J., 2015. Spatial structure in ecological communities – a quantitative anal-
spatial scales. J. Biogeogr. 40, 1560–1571. ysis. Oikos, http://dx.doi.org/10.1111/oik.02241.
Moss, B., Beklioglu, M., Carvalho, L., Kilinc, S., McGowan, S., Stephen, D., 1997. Soininen, J., 2007. Environmental and spatial control of freshwater diatoms – a
Vertically-challenged limnology; contrasts between deep and shallow lakes. review. Diatom Res. 22, 473–490.
Hydrobiologia 342–343, 257–267. Stoddard, J.L., Larsen, D.P., Hawkins, C.P., Johnson, R.K., Norris, R.H., 2006. Setting
Novak, M.A., Bode, E.W., 1992. Percent model affinity: a new measure of macroin- expectations for the ecological condition of streams: the concept of reference
vertebrate community composition. North Am. Benthol. Soc. 11, 80–85. condition. Ecol. Appl. 16, 1267–1276.
O’Hare, M.T., Gunn, I.D.M., Chapman, D.S., Dudley, B.J., Purse, B.V., 2012. Impacts of Stomp, M., Huisman, J., Mittelbach, G.G., Litchman, E., Klausmeier, C.A., 2011. Large-
space, local environment and habitat connectivity on macrophyte communities scale biodiversity patterns in freshwater phytoplankton. Ecology 92, 2096–2107.
in conservation lakes. Divers. Distrib. 18, 603–614. Sutela, T., Aroviita, J., Keto, A., 2013. Assessing ecological status of regulated lakes
Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O’Hara, R.B., Simpson, with littoral macrophyte, macroinvertebrate and fish assemblages. Ecol. Indic.
G.L., Solymos, P., Stevens, M.H.H., Wagner, H., 2012. Vegan: Community Ecology 24, 185–192.
Package. R Package Version 2.0-3, http://CRAN.R-project.org/package=vegan. Tolonen, K.T., Hämäläinen, H., Holopainen, I.J., Karjalainen, J., 2001. Influences of
Olin, M., Rask, M., Ruuhijärvi, J., Tammi, J., 2013. Development and evaluation of the habitat type and environmental variables on littoral macroinvertebrate com-
Finnish fish-based lake classification method. Hydrobiologia 713, 149–166. munities in a large lake system. Fundam. appl. limnol. 152, 39–67.
Ozkundakci, D., Hamilton, D.P., Kelly, D., Schallenberg, M., de Winton, M., Verburg, Vanormelingen, P., Cottenie, K., Michels, E., Muylaert, K., Vyverman, W., De Meester,
P., Trolle, D., 2014. Ecological integrity of deep lakes in New Zealand across L., 2008. The relative importance of dispersal and local processes in structuring
anthropogenic pressure gradients. Ecol. Indic. 37, 45–57. phytoplankton communities in a set of highly interconnected ponds. Freshw.
Peres-Neto, P.R., Legendre, P., Dray, S., Borcard, D., 2006. Variation partitioning Biol. 53, 2170–2183.
of species data matrices: estimation and comparison of fractions. Ecology 87, Viana, D.S., Santamaria, L., Michot, T.C., Figuerola, J., 2013. Migratory strategies of
2614–2625. waterbirds shape the continental-scale dispersal of aquatic organisms. Ecogra-
Pilotto, F., Bazzanti, M., Di Vito, V., Frosali, D., Livretti, F., Mastrantuono, L., Pusch, phy 36, 430–438.
M.T., Sena, F., Solimini, A.G., 2015. Relative impacts of morphological alteration Vuori, K.-M., Mitikka, S., Vuoristo, H., 2009. Pintavesien ekologisen tilan luokittelu
to shorelines and eutrophication on littoral macroinvertebrates in Mediter- ((Guidance on Ecological Classification of Surface Waters in Finland. Part 1:
ranean lakes. Freshw. Sci. 34, 410–422. Reference Conditions and Classification Criteria, Part 2: Environmental Impact
Rask, M., Olin, M., Ruuhijärvi, J., 2010. Fish-based assessment of ecological status of Assessment). Environmental Administration Guidelines 3/2009), 120 pp.
Finnish lakes loaded by diffuse nutrient pollution from agriculture. Fish. Manag. Willen, E., 2007. Växtplankton i sjöar. Bedömningsgrunder, vol. 5. SLU, Institutionen
Ecol. 17, 126–133. för Miljöanalys, Rapport, 33 pp.
Räike, A., Pietiläinen, O.-P., Rekolainen, S., Kauppila, P., Pitkänen, H., Niemi, J., Raate- Yates, A.G., Bailey, R.C., 2006. The stream and its altered valley: Integrating landscape
land, A., 2003. Trends in phosphorus, nitrogen and chlorophyll a con-centrations ecology into environmental assessments of agro-ecosystems. Environ. Monit.
in Finnish rivers and lakes in 1975–2000. Sci. Total Environ. 310, 47–59. Assess. 114, 257–271.