Ecological Indicators 60 (2016) 970–979

Contents lists available at ScienceDirect

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

Aligning indicators of community composition and biogeochemical

function in stream monitoring and ecological assessments
Thomas B. Parr a,∗ , Christopher S. Cronan a , Thomas J. Danielson b , Leonidas Tsomides b ,
Kevin S. Simon c
a
School of Biology and Ecology, University of Maine, Orono, ME 04469-5722, USA
b
Maine Department of Environmental Protection, 17 State House Station, Augusta, ME 04333-0017, USA
c
School of Environment, University of Auckland, Auckland, New Zealand

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

Article history: Reliable and inexpensive indicators of ecosystem function are essential for accurately monitoring and
Received 25 April 2015 describing ecosystem integrity. Currently, most state and federal assessments of aquatic ecological
Received in revised form 23 August 2015 integrity rely on structural indicators and assume tight coupling of structure and function. We used
Accepted 27 August 2015
fluorescent composition of dissolved organic matter as a metric for certain ecosystem functions and
Available online 25 September 2015
compared the resulting index of autochthonous microbial dissolved organic matter (DOM) to macroin-
vertebrate indicators and classifications of water quality attainment reported by the Maine Department
Keywords:
of Environmental Protection (Maine DEP) at 142 stream sites. We observed that metrics of sensitive insect
Stream
Macroinvertebrate
orders such as relative Plecoptera generic richness, relative Ephemeroptera abundance, and generic rich-
Biomonitoring ness of EPT (Ephemeroptera, Plecoptera, and Trichoptera) were negatively correlated with higher values
Dissolved organic matter of metrics based on autochthonous microbial DOM sources. At the same time we observed an increase in
Parallel factor analysis the Hilsenhoff Biotic Index with increasing microbial DOM. We compared the abundance of this microbial
Urbanization DOM component to Maine DEP measured attainment classes and found that microbial DOM generally
separated sites with high biological integrity from sites where the biotic community was highly degraded.
This highlights that measures of biogeochemical ecosystem function complement measures of structure
in biological assessments.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction
In many countries, legislation has been enacted to protect water
bodies for a variety of human and natural uses. The ultimate goal
of such legislation is to sustain ecosystem services that provide
benefits such as clean water and consumable biomass production
by protecting both ecosystem structure and function (e.g. United
States Clean Water Act, CWA; European Union Water Framework
Directive, WFD). Many regulatory tools relying on chemical and
biological indicators have been developed to help determine the
ecological status of these water bodies (Dolédec and Statzner, 2010;
Birk et al., 2012; Carter and Resh, 2013). However, monitoring
agencies generally rely heavily on structural indicators such as
fish, macroinvertebrate, or algal community composition metrics
∗ Corresponding author. Present address: Department of Plant and Soil Sciences,
531 South College Avenue, Newark, DE 19716, USA. Tel.: +1 2182600286.
E-mail address: tbparr@udel.edu (T.B. Parr).
to determine if a water body is meeting its water quality goals (Birk
et al., 2012; Carter and Resh, 2013).
Macroinvertebrates, fish, and algae have long been used as
structural indicators of water quality because they integrate envi-
ronmental conditions and stressors over long periods (Hilsenhoff,
1987). Taxonomic composition of the community reflects both
acute and chronic exposure to anthropogenic stressors such as
‘flashy’ discharge, elevated pollutants, and resource availability
(Walsh et al., 2005; Wenger et al., 2009). The widespread use of
ecosystem structural metrics to assess ecosystem integrity and to
protect ecosystem services presupposes that ecosystem function
matches ecosystem structure (Davies and Jackson, 2006; Sandin
and Solimini, 2009; Woodward et al., 2012). This assumption may
be particularly problematic in aquatic ecosystems where measures
of biodiversity structure may poorly reflect ecosystems services
like water purification (Cardinale et al., 2012). This may occur
because community composition does not always reflect function
in disturbed ecosystems (Sandin and Solimini, 2009; Young and
Collier, 2009; Woodward et al., 2012), especially if the commu-
nity assayed does not directly drive the function of interest. For

http://dx.doi.org/10.1016/j.ecolind.2015.08.049
1470-160X/© 2015 Elsevier Ltd. All rights reserved.
 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979
example, many biogeochemical cycles are controlled by microbial
communities, while most water quality monitoring protocols focus
on macroinvertebrate and fish assemblages. This suggests that
structural metrics should be complemented by functional metrics
to accurately assess the condition of an ecosystem (Sandin and
Solimini, 2009).
Direct measures of stream ecosystem function as ecological
indicators have become more common, and frameworks have been
proposed for their use and interpretation (Young et al., 2008;
Palmer and Febria, 2012). Most of the emphasis to date has been
on decomposition of leaves and leaf analogs (Paul et al., 2006;
Simon et al., 2009; Imberger et al., 2010), microbial extracellular
enzyme activities (Harbott and Grace, 2005; Lehto and Hill, 2013),
and ecosystem metabolism (Bunn et al., 1999; Mulholland et al.,
2005; Young et al., 2008). However, the response of the latter to
urban watershed disturbances can be inconsistent across stud-
ies (Meyer et al., 2005; Bernot et al., 2010). Studies using such
functional measures have rarely related them systematically to
structural measures, but the few that have done this found that
they do not always align neatly (Young and Collier, 2009; Del Arco
et al., 2012). While the importance of directly measuring ecosystem
function is recognized by researchers and practitioners, the effort
required to obtain functional measurements remains beyond the
budgetary capacity of many agencies (Gessner and Chauvet, 2002;
Davies and Jackson, 2006).
Monitoring efforts are largely focused on small streams, as they
are important loci of ecosystem services due to their wide spa-
tial extent (>80% of river miles; Leopold et al., 1964) and intensity
of biogeochemical cycling (Alexander et al., 2000; Peterson et al.,
2001). Microbial community structure and function mediate many
ecosystem services in streams via element cycling processes. These
microbial functions are driven by the production or consumption of
organic matter (Lindeman, 1942). Dissolved organic matter (DOM)
is an important basal source of energy in stream food webs (Hall and
Meyer, 1998). In small closed-canopy streams, much of this DOM is
derived from allochthonous sources (Vannote et al., 1980), which
includes leaching from vascular plants and soils. Anthropogenic
disturbance of the vegetative land cover and soil forming processes
may influence the production and delivery of DOM to small streams
(Kominoski and Rosemond, 2012; Stanley et al., 2012; Parr et al.,
2015), affecting the structure and function of these ecosystems
(Tanentzap et al., 2014).
Whereas the concentration of aquatic DOM may respond unpre-
dictably to anthropogenic watershed disturbance (Stanley et al.,
2012), the composition, or mix of organic molecules in DOM
may change in more predictable ways (Hosen et al., 2014; Parr
et al., 2015). Undisturbed watersheds are characterized by a
high proportion of allochthonous DOM (typically humic-acids)
derived from wetland and forest covers (humic acid-like DOM).
Humic-like DOM can have a variable response to watershed
disturbance, but it tends to decrease as wetland or forest domi-
nated landscapes are urbanized (Parr et al., 2015). This response
may depend on the type of anthropogenic disturbance; in more
continuously disturbed agricultural catchments, humic-like DOM
may increase, perhaps due to soil organic matter destabilization
and export (Petrone et al., 2011; Graeber et al., 2012). Other
forms of DOM characteristic of autochthonous freshwater micro-
bial or algal sources (Williams et al., 2010; Yang et al., 2012;
Parr et al., 2015) or wastewater discharges (Baker and Inverarity,
2004; Holbrook et al., 2006) typically increase. These forms of
DOM may be more bioavailable (Hosen et al., 2014; Parr et al.,
2015) and have the potential to affect the balance of aquatic
ecosystem metabolism. As such, DOM integrates multiple water-
shed processes spanning scales from microbes to landscapes,
and can potentially serve as a useful indicator of ecosystem
function.
971
The goal of this study was to determine whether metrics of DOM
composition, which reflect ecosystem biogeochemical function
and watershed response to disturbance, align with water quality
attainment classifications based on structural macroinvertebrate
community assemblages. We hypothesized that macroinvertebrate
structural and biogeochemical functional indicators would gener-
ally align, but with some important differences because drivers of
invertebrate community structure and biogeochemical processes
only partially overlap. To evaluate this hypothesis, we compared
fluorescent DOM composition as our indicator of DOM bioavail-
ability and biogeochemical function to macroinvertebrate-based
metrics of community structure and ecosystem biological condition
collected by the Maine Department of Environmental Protection
(DEP) at sites across an urbanization gradient in the postglacial
northeastern U.S.
2. Methods
2.1. Overview
We selected 142 stream sites along a gradient of urbanization in
southern and central Maine, U.S.A. that are Maine DEP macroinver-
tebrate biomonitoring sites. In Maine, urbanization typically occurs
in a landscape matrix of second growth forest and wetlands. Most of
the watersheds in this study have been reforested for ∼50–70 years
after logging or agriculture. We excluded sites influenced primarily
by agriculture or point source discharges. At each site, we measured
a suite of physicochemical parameters and collected water samples
to measure optical DOM composition and basic water chemistry
(Table A1). Maine DEP’s water quality monitoring program assesses
the macroinvertebrate community structure of 50–60 streams per
year. Maine DEP follows a 5-year cycle and focuses sampling in 5
geographic regions on a rotating basis.
2.2. Study design and site selection
The 142 sites used were selected from the DEP database based on
three criteria. First, we selected sites for which macroinvertebrate
monitoring data was available. Second, we limited our sampling to
1st and 2nd order streams because these should be most tightly
coupled to the terrestrial environment (land cover) and to mini-
mize the effects of natural longitudinal changes on stream ecology
and physicochemical parameters (Vannote et al., 1980). Third, we
used geographic information systems (GIS) to analyze the degree of
urbanization as the percent of total watershed impervious surface
area (hard surfaces such as roofs, roads, parking lots, and side-
walks; ISA). 113 of these watersheds were visited 3 times and the
remaining 29 were visited once during the summer of 2011 to over-
lap directly with Maine DEP macroinvertebrate assessment. It is
important to note that because the Maine DEP sampling program
rotates every ∼5 years, we only have temporally matched DOM
and macroinvertebrate data for ∼63 sites. For the remaining sites,
macroinvertebrates were mostly sampled within 1–3 years of DOM
sampling. These 142 sites ranged from 0 to 60% ISA (Maine Land
Cover Database (MELCD) Impervious Cover layer, 5 m resolution,
Maine DEP, 2005) and represented the complete range of urban-
ization in the state of Maine. Average watershed slopes ranged
between 5 and 13% with high and low slopes distributed evenly
across the urbanization gradient.
2.3. Macroinvertebrate sampling and water quality attainment
determination
To comply with the Clean Water Act, the State of Maine has clas-
sified its fresh waters into 4 statutory classes: AA, A, B, and C. The
biological standards for each of these classes are summarized as
972 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979
follows (Me. Rev. Stat. Ann. tit. 38, §465): Class AA waters may not
have direct discharge of pollutants and aquatic life shall be as nat-
urally occurs; Class A waters are to have natural habitat for aquatic
life and aquatic life shall be as naturally occurs; Class B waters are
to have unimpaired habitat and must be of sufficient quality to
support all aquatic species indigenous to the waterbody without
detrimental changes in the resident biological community; C class
waters may have some changes to aquatic life, but must maintain
the structure and function of the resident biological community.
The law includes definitions of key terms, such as “unimpaired”
and “without detrimental changes.” Since Class AA and A share the
same biological expectations, they will be collectively referred to
as Class A for the remainder of the paper. Streams that do not attain
any of the statutory goals are referred to as non-attaining (NA).
To assess attainment of these goals, the macroinvertebrate com-
munity is routinely sampled by the Maine DEP following standard
operating procedures (Davies and Tsomides, 2002). Briefly, three
wire rock baskets or mesh rock bags filled with 7.25 ± 0.5 kg of
clean 3.8–7.6 cm diameter cobbles are placed in the run section of
a stream for 28 ± 4 days. Bags are then collected, washed through
a sieve bucket, and the contents are then stored in 70% ethyl alco-
hol. Organisms collected are identified to the lowest level possible
which is usually at least genus. A linear discriminant model (LDA)
is then employed to classify each site based on metrics of macroin-
vertebrate community composition (Davies et al., 1995; Davies and
Tsomides, 2002). The 142 sites comprised 51 attaining Class A, 25
attaining Class B, 24 attaining Class C, and 38 NA (4 of the samples
were of ‘indeterminate’ class and excluded from further analysis).
We also calculated four common metrics for determining urban
impacts on stream community structure. The generic richness of
the Ephemeroptera, Plecoptera and Trichoptera (EPT) was mea-
sured as the total number of genera within those orders that are
typically sensitive to pollution. Relative Ephemeroptera abundance
was the total number of Ephemeroptera over the total number of
individuals sampled. Relative Plecoptera generic richness was the
number of Plecoptera genera observed divided by the total number
of genera observed. The Hilsenhoff Biotic Index (HBI) was also cal-
culated. The HBI is an indicator of organic pollution tolerance where
higher numbers indicate that a community is primarily composed
of pollution tolerant species (Hilsenhoff, 1987).
2.4. Water chemistry sampling
All sampling was conducted at base flow as determined from
nearby (<70 km) United States Geological Survey stream gages.
Each set of seasonal samples was collected within a two to three
week period. Spring sampling occurred in May prior to leaf out,
summer sampling occurred in August, and fall sampling occurred
after leaf abscission (∼October–November). At each site, we used a
Hach HQ40d multimeter to measure basic physicochemical param-
eters (temperature, dissolved oxygen, conductivity, and pH). Due
to an instrument malfunction, we only collected physicochemical
data for 23 sites in the fall. Water samples for DOM composi-
tion analysis were syringe-filtered through precombusted 0.7 ␮m
Whatman GF/F filters into acid-leached HDPE bottles and stored on
ice or at 4 ◦ C until analysis.
2.5. Optical DOM characterization
A variety of techniques may be used to characterize DOM
composition. Broadly, they may be described by bulk (elemental
composition and percent hydrophobic or hydrophilic DOM), optical
(UV/vis and fluorescence), and structural characteristics (functional
group analysis, NMR, and mass spectrometry). Among these meth-
ods DOM fluorescence has emerged as a reliable and inexpensive
method for rapidly characterizing DOM composition. Fluorescence
measures the fraction of DOM that, after absorbing light, re-emits
it at a longer wavelength. Fluorescence excitation and emission
matrices (spectra, EEMs) are composed of ∼3000 measurements
at different excitation and emission wavelength pairs. These data
can then be analyzed as the ratios of different wavelengths
(indices) or by parallel factor analysis (statistical chromatography),
which deconvolutes the spectra and identifies distinct chemical
classes.
Samples for DOM fluorescence were analyzed within 1–3 days
of collection. Prior to fluorescence analysis, samples with high
UV/Vis absorbance at 280 nm were diluted with deionized water
(Ohno, 2002) to avoid inner-filter effects. Fluorescence excitation
emission matrices (EEMs) were collected by scanning the emission
(em.) spectra (250–535 nm in 3 nm intervals) for a series of exci-
tation (ex.) wavelengths (220–535 nm in 3 nm intervals). All EEMs
were corrected for instrumental bias following the manufacturer’s
method, corrected for inner-filter effects according to Ohno (2002),
blank subtracted with DI water, and Raman normalized (Stedmon
and Bro, 2008).
The different regions of these three-dimensional spectra (exci-
tation, emission, intensity/concentration) were decomposed into
statistically similar patterns of fluorescence variation called com-
ponents using parallel factor analysis (PARAFAC, Stedmon et al.,
2003; Cory and McKnight, 2005). These components represent
chemically distinct structures within the DOM pool (Stedmon and
Bro, 2008), and have been linked to biotic and abiotic ecosys-
tem processes, including susceptibility of certain components to
UV radiation (Ishii and Boyer, 2012), transformation of ‘humic-
like’ components concomitant with denitrification (Stelzer et al.,
2014), production of ‘protein-like’ components with algal produc-
tion (Stedmon and Markager, 2005a), and DOM biodegradation
(Cory and Kaplan, 2012). PARAFAC analysis was performed in
Matlab (MathWorks, 2012) using the DOMFluor v1.7 toolbox as
described by Stedmon and Bro (2008). Due to instrument noise
at lower wavelengths, we only modeled the EEM region within
ex. 250–535 nm and em. 289–535 nm. The final model was vali-
dated with split-half validation. For each component, we report
PARAFAC results as the proportional PARAFAC Fmax score to sep-
arate the effects of composition and concentration. Proportional
scores (proportional or relative abundance) are the absolute fluo-
rescence maximum score (Fmax ; in Raman Units, RU) divided by the
sum of Fmax scores for all identified components.
2.6. Statistical analysis
We used a variety of statistical approaches to analyze trends
and relationships in the data. Ordinary least squares regression was
used to assess trends in DOM with respect to urbanization. Spear-
man rank correlation coefficients (rs ) were calculated to assess the
strength of the relationships between macroinvertebrate indicators
and DOM composition and locally weighted regressions (LOWESS)
were used to illustrate the shape of the relationship. We used a
two-way multivariate analysis of variance (MANOVA) to compare
the relative abundance of DOM composition among independent
attainment classes, seasons, and the interaction of attainment class
and season. Preliminary analysis for the MANOVA indicated that
covariance matrices were not equal among groups so we used
Pillai’s trace and the approximate F to test overall model signifi-
cance (Hand and Taylor, 1997). We then used one-way ANOVAs
with Welch’s correction for unequal variance to test for significant
differences among attainment classes and seasons (Welch, 1938).
The alpha level was adjusted for multiple comparisons using the
Bonferroni correction. We also used non-metric dimensional scal-
ing (NMDS) to assess alignment between DOM composition and
macroinvertebrate attainment classes determined by the Maine
 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979 973

Fig. 1. Scatter plots of select macroinvertebrate community composition and pollution tolerance indices used by Maine DEP in site attainment classification against total
watershed impervious cover (%). The dark gray line represents the LOESS smooth line and is bounded by a 95% confidence interval (light gray). Ephemeroptera, Plecoptera
and Trichoptera generic richness (A), % abundance of Ephemeroptera (B), relative generic richness of Plecoptera (C), and the Hilsenhoff Biotic Index (D). Note (D) uses absolute
rather than relative (%) PARAFAC scores. A (*) indicates a significant (p < 0.005) Spearman rank correlation coefficient (rs ).

DEP. All statistical analysis was conducted in [R] (R Core Team,
2015).
3. Results
The generic richness within the EPT orders (Fig. 1A), the relative
Ephemeroptera abundance (Fig. 1B), and the relative Plecoptera
generic richness (Fig. 1C) decreased non-linearly with increasing
ISA% (rs = −0.59, −0.64, and −0.51 respectively, p < 0.005). Typically
these values declined 60–75% from their maximum values between
∼10 and 20% ISA. The HBI increased by a similar magnitude between
10 and 20% ISA (rs = 0.46), indicating more pollution tolerant species
in urban streams (Fig. 1D). Both results are consistent with previ-
ous research conducted in Maine (Morse et al., 2003). Additionally,
we note that natural variability in these same metrics was high and
decreased with increasing urbanization.
We previously developed a 9 component PARAFAC model to
explain the variation in this dataset (Parr et al., 2015). We observed
strong changes in stream DOM composition across the urbaniza-
tion gradient. Four of the components (ME2, 4, 6, and 7) were
associated with watersheds dominated by forests and wetlands
and a predominance of allochthonous DOM sources. Five of the
components (ME1, 3, 5, 8, and 9) were abundant in watersheds
with a high degree of urbanization and an increased proportion of
autochthonous DOM sources. In general, increasing urbanization
was correlated to a greater relative abundance of DOM compo-
nents associated with microbial production (e.g. ME8, Fig. 2A) and a
lower relative abundance of components associated with terrestrial
sources (e.g. ME4, Fig. 2B).
ME8 in particular was characteristic of autochthonous microbial
production and was positively correlated to urbanization (Fig. 2A,
Parr et al., 2015). ME8 is a microbial humic-like component (has
humic-like and protein-like DOM characteristics; primary and sec-
ondary excitation maxima of <250 and 295 nm, and an emission
maximum of 385 nm) that exhibits algal or microbial character-
istics. Experimental evidence indicates that this DOM component
was associated with greater DOC bioavailability (Parr et al., 2015).
To determine the extent to which DOM, as an indicator of certain
ecosystem functions, was paralleled by altered ecosystem struc-
ture, we compared DOM composition to benthic macroinvertebrate
indices and attainment classes provided by the Maine DEP.
These indicators of macroinvertebrate community composition
were correlated to varying degrees with certain DOM components
(Figs. A1 and A2). However, here we focus on component ME8
which displayed the strongest correlation with urban land use and
macroinvertebrate indices. It is important to note that these rela-
tionships are correlative and are not intended to suggest causal
linkages. Macroinvertebrate indices were correlated to ME8 to a
similar degree as ISA% and displayed a similar nonlinear pattern
of correlation. We found that increasing relative abundance of
ME8 was associated with decreasing total generic richness of EPT
orders, decreased relative abundance of Ephemeroptera individ-
uals, and declines in Plecoptera taxa in the overall community
composition (Fig. 3A–C). We also observed that the average pol-
lution tolerance of the community (HBI) increased with increasing
total abundance of ME8 (Fig. 3D). Notably, we observed greater
variability in macroinvertebrate community composition indices
at lower levels of ME8 than at higher abundances of ME8 (Fig. 3A
and D).
Using MANOVA analysis we found DOM composition was
related to attainment class (approx. F24,882 = 8.0, p < 0.005) and
season (approx. F16,586 = 22.8, p < 0.005), but not an interaction
974 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979
Fig. 2. Relationship between ISA (%) and PARAFAC components characteristic of a)
microbial and b) terrestrial sources. Vertical lines through plots indicate standard
error of measurements made in spring, summer, and fall (N = 2–3).
Data derived from Parr et al. (2015).
between attainment class and season (p > 0.05). All DOM compo-
nents except ME3 and 6 varied among attainment classes with ME2,
4, 7, and 8 showing the most significant differences in DOM com-
position among attainment classes (Fig. 4). In general, sites that
attained Class A had the highest levels of allochthonous humic-like
DOM (ME2, 4, 7; Fig. 4). ME8, which was associated with a decrease
in genera sensitive to pollution (Fig. 3), had the highest propor-
tional abundance in non-attaining sites (Fig. 4). ME8 constituted
less than 3% of the fluorescence signature for 72% of the streams
attaining Class A, whereas >95% of streams not attaining any class
(NA) typically have more than 3% of this fluorescence component.
The DOM characteristics of class B sites typically overlapped with
either Class A or C sites (e.g. ME4, Fig. 4).
Non-metric dimensional scaling of PARAFAC fluorescence com-
ponents showed that attainment classes were aligned along a
gradient of DOM composition with microbial compounds such as
ME8 most closely associated with sites of lower biotic quality and
terrestrial humic-like compounds such as ME4 associated with sites
achieving a high attainment classification (Fig. 5). Dissolved organic
matter composition did not clearly separate sites with character-
istics of slightly disturbed communities (“B” attainment classes;
Figs. 4 and 5). We also found that while most of the Maine DEP
monitoring sites in our study met or exceeded their statutory goals,
those that did not typically had a higher proportion of ME8.
4. Discussion
We hypothesized that macroinvertebrate structural and bio-
geochemical functional indicators would generally align, but
differences in the drivers of invertebrate community structure and
biogeochemical function would result in only partial overlap. Our
results support this hypothesis by showing that both structural and
functional indicators were sensitive to the degree of watershed
urbanization (ISA%), but in different ways – structural indicators
had a nonlinear response trajectory while functional indicators
had a more linear response trajectory. Consequently, macroinver-
tebrate indicators and DOM composition were correlated, but their
responses aligned most strongly at extremely high or low levels
of urbanization. In particular, PARAFAC component ME8 increased
linearly with increasing ecosystem anthropogenic disturbance and
was relatively more abundant in Class C and non-attaining streams.
A proportional abundance >10% was frequently observed in NA
sites. At the same time, this component was largely absent from
sites achieving the highest attainment class (A). Because the attain-
ment class and season did not interact to affect the mean DOM
composition, DOM composition may be a reliable tool in screening
stream attainment regardless of season.
4.1. Function, structure, and urbanization
There are many ecosystem factors that affect aquatic life and
a subset of those factors involve the production and transfor-
mation of DOM. The mechanisms of urbanization shaping the
correlations among ISA%, macroinvertebrate indices (Fig. 3), and
DOM composition (Fig. 2) likely operate at different scales within
the landscape and not all of the mechanisms impacting DOM
composition will necessarily affect macroinvertebrate community.
For example, urbanization typically increases the percent ISA in
a watershed, which commonly increases hydrologic ‘flashiness’
(Walsh et al., 2005). Percent ISA is an example of an urbanization
process with different mechanistic impacts on DOM and macroin-
vertebrates. For DOM, the percentage of ISA comes at the expense
of pre-existing land covers (wetland and forest in Maine, USA) and
reduces the abundance or connectivity of sources areas contribut-
ing to stream DOM (Parr et al., 2015). That same ISA also increases
hydrologic flashiness which physically degrades habitat for sensi-
tive taxa (Schueler, 1994; Walsh et al., 2005). At the same time,
some stressors and processes may affect macroinvertebrates with-
out altering DOM. For example, inputs or accumulations of heavy
metals in a watershed with little urbanization (e.g. mining or legacy
pollution) may impair the macroinvertebrate community without
strongly affecting the composition of DOM. The scale of processes
impacting organisms may be partially dependent on the level of
disturbance. For fish, riparian and reach scale factors may have
the most control over community structure at low levels of dis-
turbance while watershed scale factors may control community
structure at high levels of disturbance (Wang et al., 2003). In this
way, the difference in pattern of response to urban intensification
for DOM composition and macroinvertebrate structure (Fig. 3) may
be influenced by differences in factors and scales driving structure
and function.
In our analysis, structural metrics and indicators of biogeo-
chemical function (DOM composition, ME8) exhibited directionally
similar responses to urbanization, but with different trajecto-
ries. While DOM composition changed relatively linearly with
increasing urbanization (Fig. 2), the macroinvertebrate community
changed non-linearly with the fastest rates of degradation between
0 and ∼20% ISA (Fig. 1). Other studies have found varied responses
of structure and function to urban impacts and the exact relation-
ship may vary with the level of anthropogenic disturbance and the
combination of stressors driving degradation (Young et al., 2008;
Hopkins et al., 2011). Algal community indicators of structure, col-
lected across many of the same sites, displayed a range of linear
to non-linear responses as sensitive taxa decreased and tolerant
taxa increased (Danielson et al., 2011). Nutrient spiraling metrics of
ecosystem function like phosphate, nitrate, and ammonium uptake
velocity may decrease linearly if increasing watershed disturbance
(urban or agricultural) is associated with elevated nutrient inputs
 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979 975

Fig. 3. Scatter plots of select macroinvertebrate community composition and pollution tolerance indices used by Maine DEP in site attainment classification against microbial
humic-like component ME8. Error bars represent ±1 standard error around the mean of 3 seasonal sampling visits. The dark gray line represents the LOESS smooth line and
is bounded by a 95% confidence interval (light gray). Ephemeroptera, Plecoptera and Trichoptera generic richness (A), % abundance of Ephemeroptera (B), relative generic
richness of Plecoptera (C), and the Hilsenhoff Biotic Index (D). Note (D) uses absolute rather than relative (%) PARAFAC scores. A (*) indicates a significant (p < 0.005) Spearman
rank correlation coefficient (rs ).

(Niyogi et al., 2004; Meyer et al., 2005). Conversely, if nutrient
inputs are low but light input to streams increases by riparian
thinning the metrics could increase (Niyogi et al., 2004). Ecosys-
tem respiration and gross primary productivity which are typically
higher in urban streams (Bernot et al., 2010) may increase lin-
early (Yates et al., 2014) or logarithmically (Clapcott et al., 2012)
with urbanization. Other functions, like leaf litter degradation may
display U shaped (non-monotonic) responses peaking at interme-
diate levels of urbanization (Bierschenk et al., 2012). The often
differing trajectories relating structure and function to watershed
disturbance suggests that there is value to measuring multiple
structural and functional endpoints to better understand overall
stream condition. These results support the idea that structural and
functional indicators should be used together to provide a more
holistic assessment of ecosystem integrity (Clapcott et al., 2012).
4.2. Distinguishing ecosystem integrity by DOM composition
While DOM composition effectively distinguished the attain-
ment classes in pristine (A) and more heavily degraded (C and NA)
sites (Figs. 4, 5), it was not able to separate sites with macroin-
vertebrate community structures which were subject to moderate
levels of disturbance or impairment (Class B). This may be because
ecosystem structure and DOM related function may not be strictly
coupled in these systems and low levels of urbanization may
affect ecosystem structure well before the watershed functional
processes shaping organic matter composition are affected (as dis-
cussed above). In other words, within the narrow range where
structure changes rapidly, DOM related ecosystem functions show
a much slower change and are consequently less sensitive in that
range.
DOM may participate in fueling biotic ecosystem function either
as a source of energy or as a growth substrate for microbes.
As such, different components of DOM may be involved in dif-
ferent aquatic ecosystem functions. C cycling potential may be
indicated by the biodegradability of the DOC pool. DOM fluores-
cence has shown that different components degrade at different
rates with protein-like components and some humic-like compo-
nents degrading most rapidly (Baker and Inverarity, 2004; Cory
and Kaplan, 2012; Parr et al., 2015). For N cycling, Stelzer et al.
(2014) found that denitrification transformed humic-like DOM
components while protein-like components remained relatively
unchanged, likely due to redox transformation. Nutrient spiraling
studies have shown that the addition of organic N may displace
demand for humic-like DOM from terrestrial source (Lutz et al.,
2012). Thus, understanding temporal change in the abundance of
key components within sites may make a useful indicator of ele-
ment cycling. DOM composition has also been correlated to whole
stream metabolism with higher GPP and P/R ratio indicated by
increasing abundance of a microbial humic-like peak (Halbedel
et al., 2013). Transformation of humic-like components has been
observed as DOM passes through an anoxic hyporheic zone with
active denitrification (Stelzer et al., 2014). Further investigation
tracking such transformations temporally may yield quantitative
indicators of denitrification or microbial redox in streams. Because
DOM composition is affected by human land use in a watershed and
is involved in many aquatic ecosystem functions, it may serve as a
vital tool for monitoring changes in ecosystem structure, function,
and connectivity.
976 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979

Fig. 4. Boxplot summarizing the differences in proportions of PARAFAC components ME1–ME9 with respect to Maine DEP attainment class using DOM data from all seasons.
Boxes represent the 1st and 3rd quartiles, the central line the median, and whiskers reflect 1.5× the inner quartile range. Lower case letters above boxes indicate means that
are significantly different by ANOVA and Tukey’s HSD test.

While DOM composition, as a metric of ecosystem function, and reproducible manner. In biomonitoring with macroinverte-
may offer an approach to enhance our understanding of ecosys- brates, the specific collection methods employed are often not
tem integrity, it also offers a means by which we can assess standardized across governmental boundaries (Friberg et al., 2006;
ecosystem integrity across broad spatial extents in a cost–effective Carter and Resh, 2013) and reflect both the preferences of the

Fig. 5. Non-metric multidimensional scaling (NMDS) ordination using a Bray–Curtis distance of DOM composition data collected at Maine DEP biomonitoring sites (p < 0.005,
stress = 0.07). Individual sites are labeled with their most recent Maine DEP macroinvertebrate attainment classification. “AA/A” class sites are highest attaining followed by
“B”, and then “C”. Highly degraded sites not attaining any of those classes are labeled “NA.”.
 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979 977

Table 1
Description of PARAFAC components (this study = ME) and comparison with previous studies. (U+/−) indicates that a component was significantly positively or negatively
correlated with impervious cover in this study.

This study Description from previous studies assessing anthropogenic disturbance Similar components

ME1 (U+) Terrestrial humic-like component observed in a variety disturbed and undisturbed ecosystems including SM1, LU5
agricultural, urban, wetland, and forested streams.

ME2 (U−) Humic-like fluorophore of terrestrial origins ubiquitous in freshwaters. Not related to time since forest B2, CW1, SM2, YG2, H1, LU2
harvest in a Tasmanian chronosequence. Positively correlated with increasing forage land cover in Quebec, CA.
Negatively correlated to indicators of anthropogenic activity Cl− and SO4 2− along an urbanization gradient
and higher in forested streams compared to urbanizing streams in Virginia studies.

ME3 Humic-like component of autochthonous microbial origin also attributed to terrestrial origins Not related to B1, Y1, SM3, CW2, GR1, YG1,
time since forest harvest in a Tasmanian chronosequence, but was higher in unharvested reference H2, LU3, CK1
watersheds than in white pine plantations and clear cut sites in North Carolina, USA. Increased with
increasing arable land and forage covers in Quebec, CA. Higher in human modified streams compared to
forested streams in Virginia, USA. Generally not bioavailable in agricultural streams in Pennsylvania.

ME4 (U−) Humic-like component, negatively correlated to indicators of anthropogenic activity Cl− and SO4 2− . H1, HP1
Abundance increased with increasing forest cover in Maryland, USA.

ME5 (U+) Observed in other studies (Cory and McKnight, 2005), but seldom reported in land use change studies.

ME6 (U−) Terrestrial humic-like. Higher at reference sites than disturbed white pine plantations in North Carolina, USA. Y2, CK3
Generally not bioavailable in agricultural streams in Pennsylvania.

ME7 (U−) Widespread humic-like; common in wetlands and forest streams. Higher at reference sites than disturbed CW6, HP2,Y3
white pine plantations in North Carolina, USA. Positively correlated with forage and monoculture, negatively
correlated with wetlands in Quebec, CA. Increases with forest cover in Maryland, USA.

ME8 (U+) Characteristics of protein-like and microbial humic-like fluorescence, may be produced autochthonously by YG3, HP3, LU3
autotrophs or derive from anthropogenic sources. Higher in human modified streams in Virginia, USA. Higher
in streams draining agro-urban watersheds in southeast China. Contributed to bioavailability and increased
with urbanization in Maryland, USA.

ME9 (U+) Protein-like component. Was equally abundant in reference, clear cut, and white pine plantations in North SM7, YG3, H3, LU4, CK4
Carolina, USA. Positively correlated to indicators of anthropogenic activity Cl− and SO4 2− and highest in
human dominated streams compared to forested streams in Virginia. Semi-bioavailable in agricultural
streams in Pennsylvania.

Burrows et al. (2013) = B; Stedmon and Markager (2005b) = S; Williams et al. (2010) = CW; Yamashita et al. (2011) = Y; Yang et al. (2012) = YG; Graeber et al. (2012) = GR;
Holbrook et al. (2006) = H; Hosen et al. (2014) = HP; Lu et al. (2013) = LU; Cory and Kaplan (2012) = CK.

individuals designing the program and the budgetary or envi-
ronmental constraints they face. These differences can result in
different classifications of the same water body across political
boundaries (USEPA, 2003). Methods for the collection and analy-
sis of DOM composition are relatively standardized (Stedmon and
Bro, 2008; Cory et al., 2010) and recent reviews have shown that the
environmental source and behavior of different DOM components
can be consistent across broad spatial scales (Ishii and Boyer, 2012).
In our dataset, we saw relatively little variance in low flow DOM
composition among seasons during base flow conditions. This sug-
gests that, under similar flow conditions, it could serve as a reliable
screening tool regardless of season.
4.3. DOM for diagnosis
While we are highlighting the potential for DOM composition to
be used as an indicator of stream ecosystem function with respect
to urbanization, it may be useful as an indicator for tracking spe-
cific sources of impairment and other land uses (Table 1). DOM
from anthropogenic point sources often has distinctive character-
istics. For example, paper mill effluent contains components that
allow this input to be tracked through large rivers into coastal zones
(Cawley et al., 2012). Sewage effluents, although overlapping with
protein-like DOM, often have distinctive peaks not found in natu-
ral DOM which may be used to detect the effluent’s presence and
predict its reactivity (Baker, 2001; Baker and Inverarity, 2004).
Along anthropogenic disturbance gradients, some classes, like
protein-like DOM (ME8,9) associated with in-stream microbial
production consistently increase with increasing agricultural and
urban land cover (Table 1). The response of components more
characteristic of terrestrial DOM sources can be more varied and
may depend on the type of land use. For example, increases in
components similar to ME7 may be associated with both greater
forest cover and greater agricultural land use (Table 1). This similar
response may be due to production of organic matter in forested
areas, while in agricultural areas, it may be due to destabilization
and loss of preserved soil organic matter.
While some aspects ecosystem integrity along land use gradi-
ents may be predicted with land use maps, they do have some
limitations. Namely, the land cover they represent is often five years
old when published, the classifications may overgeneralize urban
land uses thereby obfuscating their impacts on aquatic resources
(Cadenasso et al., 2007), and they are not universally available, par-
ticularly in developing countries (Kuemmerle et al., 2013). Perhaps
most importantly, they do not necessarily describe the functional
connectivity between a stream and a watershed. Thus, simple land
cover may not always predict the structural or functional status of
a stream. Understanding and managing for stream-watershed con-
nectivity is an increasingly important management goal (USEPA,
2015).
4.4. The economics of monitoring
Although macroinvertebrate biomonitoring is considered one
of the most cost effective methods for monitoring and assessing
streams, it still represents a significant investment of time and
money for agencies charged with water quality protection. Iden-
tification of a sample to the genus or species level typically costs
∼$350–400 and the total cost of assessing a site may be $600–1200
for a small stream and up to $2250 on larger rivers (Wisconsin,
USA, Department of Natural Resources, 2013). Similarly, in 2015
it cost Maine DEP an average of $1500 per stream to assess small
streams in Maine. This approach is also time intensive, taking 6–8
months to receive final identification of all samples. This prohibits
978 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979
agencies from sampling more than a few sites per year relative to
the total number of water bodies they are charged with protec-
ting. For example, Maine DEP typically only samples 50–60 sites
per year and resamples many sites every 5 years–which is typical
of many states. DOM fluorescence analysis, on the other hand could
cost as little as $8.00 per sample (McDowell, 2015) and enable less
intensive but more extensive monitoring complementing intensive
biotic assessments.
Based on our analysis, DOM fluorescence characteristics may
be useful for assessing status of ecosystems and diagnosing some
causes of impairment. It may be a rapid and inexpensive method
for indicating ecosystem function while at the same time differen-
tiating between the extremes of high and low quality in regards
to community structure. Understanding the full impact of humans
on stream ecosystems requires indicators which can provide rich
information that integrate both structural and functional metrics.
The use of combined metrics will be increasingly important as we
seek to actively manage ecosystems for both structure and function.
The USEPA’s draft report on ecosystem connectivity highlights the
central role that DOM plays in connecting streams and wetlands to
downstream rivers (USEPA, 2015). Using structural and functional
metrics in a complementary way may enable water resource man-
agers to set and monitor management goals that reflect multiple
aspects of an ecosystem’s potential (Paul et al., 2009). In particu-
lar, for assessing the impacts of urbanization, we recommend using
the relative or absolute fluorescence of autochthonous components
like ME8.
Acknowledgements
We deeply Kari Metcalf for help in the lab and field. This research
is part of Sustainability Solutions Initiative, a program of the Sen-
ator George J. Mitchell Center, which is supported by National
Science Foundation’s Experimental Program to Stimulate Compet-
itive Research (EPSCoR, award EPS-0904155) to Maine EPSCoR at
the University of Maine. Additional support was provided by the
Maine Agricultural and Forest Experiment Station (MAFES); this is
MAFES Publication No. 3439.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.ecolind.2015.
08.049.
References
Alexander, R.B., Smith, R.A., Schwarz, G.E., 2000. Effect of stream channel size on the
delivery of nitrogen to the Gulf of Mexico. Nature 403, 758–761.
Del Arco, A.I., Ferreira, V., Graça, M.A.S., 2012. The performance of biological indi-
cators in assessing the ecological state of streams with varying catchment
urbanisation levels in Coimbra, Portugal. Limnetica 31, 141–154.
Baker, A., 2001. Fluorescence excitation – emission matrix characterization of some
sewage impacted rivers. Environ. Sci. Technol. 35, 948–953.
Baker, A., Inverarity, R., 2004. Protein-like fluorescence intensity as a possible tool
for determining river water quality. Hydrol. Process. 18, 2927–2945.
Bernot, M.J., et al., 2010. Inter-regional comparison of land-use effects on stream
metabolism. Freshw. Biol. 55, 1874–1890.
Bierschenk, A.M., Savage, C., Townsend, C.R., Matthaei, C.D., 2012. Intensity of land
use in the catchment influences ecosystem functioning along a freshwater-
marine continuum. Ecosystems 15, 637–651.
Birk, S., et al., 2012. Three hundred ways to assess Europe’s surface waters: an almost
complete overview of biological methods to implement the Water Framework
Directive. Ecol. Indic. 18, 31–41.
Bunn, S.E., Davies, P.M., Mosisch, T.D., 1999. Ecosystem measures of river health and
their response to riparian and catchment degradation. Freshw. Biol. 41, 333–345.
Burrows, R.M., Fellman, J.B., Magierowski, R.H., Barmuta, L.A., 2013. Allochthonous
dissolved organic matter controls bacterial carbon production in old-growth and
clearfelled headwater streams. Freshw. Sci. 32, 821–836.
Cadenasso, M.L., Pickett, S.T.A., Schwarz, K., 2007. Spatial heterogeneity in urban
ecosystems: reconceptualizing land cover and a framework for classification.
Front. Ecol. Environ. 5, 80–88.
Cardinale, B.J., et al., 2012. Biodiversity loss and its impact on humanity. Nature 489,
326.
Carter, J.L., Resh, V.H., 2013. Analytical Approaches Used in Stream Benthic Macroin-
vertebrate Biomonitoring Programs of State Agencies in the United States
2013-1129.
Cawley, K.M., Butler, K.D., Aiken, G.R., Larsen, L.G., Huntington, T.G., McKnight, D.M.,
2012. Identifying fluorescent pulp mill effluent in the Gulf of Maine and its
watershed. Mar. Pollut. Bull. 64, 1678–1687.
Clapcott, J.E., Collier, K.J., Death, R.G., Goodwin, E.O., Harding, J.S., Kelly, D., Leathwick,
J.R., Young, R.G., 2012. Quantifying relationships between land-use gradients
and structural and functional indicators of stream ecological integrity. Freshw.
Biol. 57, 74–90.
Cory, R.M., Kaplan, L.A., 2012. Biological lability of streamwater fluorescent dissolved
organic matter. Limnol. Oceanogr. 57, 1347–1360.
Cory, R.M., McKnight, D.M., 2005. Fluorescence spectroscopy reveals ubiquitous
presence of oxidized and reduced quinones in dissolved organic matter. Environ.
Sci. Technol. 39, 8142–8149.
Cory, R.M., Miller, M.P., McKnight, D.M., Guerard, J.J., Miller, P.L., 2010. Effect of
instrument-specific response on the analysis of fulvic acid fluorescence spectra.
Limnol. Oceanogr. Methods 8, 67–78.
Danielson, T.J., Loftin, C.S., Tsomides, L., DiFranco, J.L., Connors, B., 2011. Algal
bioassessment metrics for wadeable streams and rivers of Maine USA. J. N. Am.
Benthol. Soc. 30, 1033–1048.
Davies, S.P., Jackson, S.K., 2006. The biological condition gradient: a descriptive
model for interpreting change in aquatic ecosystems. Ecol. Appl. 16, 1251–
1266.
Davies, S.P., Tsomides, L., DEPLW0387-B2002 2002. Methods for Biological Sampling
and Analysis of Maine’s Rivers and Streams.
Davies, S., Tsomides, L., Courtemanch, D.L., Drummond, F., DEP-LW108 1995. Maine
biological monitoring and biocriteria development program., pp. 1–61.
Dolédec, S., Statzner, B., 2010. Responses of freshwater biota to human disturbances:
contribution of J-NABS to developments in ecological integrity assessments. J.
N. Am. Benthol. Soc. 29, 286–311.
Friberg, N., Sandin, L., Furse, M.T., Larsen, S.E., Clarke, R.T., Haase, P., 2006. Com-
parison of macroinvertebrate sampling methods in Europe. Hydrobiologia 566,
365–378.
Gessner, M.O., Chauvet, E., 2002. A case for using litter breakdown to assess func-
tional stream integrity. Ecol. Appl. 12, 498.
Graeber, D., Gelbrecht, J., Pusch, M.T., Anlanger, C., von, D., Schiller, 2012. Agriculture
has changed the amount and composition of dissolved organic matter in Central
European headwater streams. Sci. Total Environ. 438, 435–446.
Halbedel, S., Büttner, O., Weitere, M., 2013. Linkage between the temporal and
spatial variability of dissolved organic matter and whole-stream metabolism.
Biogeosciences 10, 5555–5569.
Hall, R.O., Meyer, J.L., 1998. The trophic significance of bacteria in a detritus-based
stream food web. Ecology 79, 1995–2012.
Hand, D.J., Taylor, C.C., 1997. Multivariate Analysis of Variance and Repeated Meas-
ures: A Practical Guide, 5th ed. Chapman & Hall.
Harbott, E.L., Grace, M.R., 2005. Extracellular enzyme response to bioavailability
of dissolved organic C in streams of varying catchment urbanization. J. N. Am.
Benthol. Soc. 24, 588–601.
Hilsenhoff, W., 1987. An improved biotic index of organic stream pollution. Gt. Lakes
Entomol. 20, 31–39.
Holbrook, R.D., Yen, J.H., Grizzard, T.J., 2006. Characterizing natural organic mate-
rial from the Occoquan Watershed (Northern Virginia, US) using fluorescence
spectroscopy and PARAFAC. Sci. Total Environ. 361, 249–266.
Hopkins, J.M., Marcarelli, A.M., Bechtold, H.A., 2011. Ecosystem structure and
function are complementary measures of water quality in a polluted, spring-
influenced river. Water Air Soil Pollut. 214, 409–421.
Hosen, J.D., McDonough, O.T., Febria, C.M., Palmer, M.A., 2014. Dissolved organic
matter quality and bioavailability changes across an urbanization gradient in
headwater streams. Environ. Sci. Technol. 48, 7817–7824.
Imberger, S.J., Thompson, R.M., Grace, M.R., 2010. Searching for effective indica-
tors of ecosystem function in urban streams: assessing cellulose decomposition
potential. Freshw. Biol. 55, 2089–2106.
Ishii, S., Boyer, T., 2012. Behavior of reoccurring PARAFAC components in fluorescent
dissolved organic matter in natural and engineered systems: a critical review.
Environ. Sci. Technol. 46, 2006–2017.
Kominoski, J.S., Rosemond, A.D., 2012. Conservation from the bottom up: forecasting
effects of global change on dynamics of organic matter and management needs
for river networks. Freshw. Sci. 31, 51–68.
Kuemmerle, T., et al., 2013. Challenges and opportunities in mapping land use inten-
sity globally. Curr. Opin. Environ. Sustain. 5, 484–493.
Lehto, L.L.P., Hill, B.H., 2013. The effect of catchment urbanization on nutrient uptake
and biofilm enzyme activity in Lake Superior (USA) tributary streams. Hydrobi-
ologia 713, 35–51.
Leopold, L.B., Wolman, M.G., Miller, J.P., 1964. Fluvial processes in geomorphology.
W.H. Freeman and Company.
Lindeman, R., 1942. The trophic-dynamic aspect of ecology. Ecology 23, 399–417.
Lu, Y., Bauer, J.E., Canuel, E.A., Yamashita, Y., Chambers, R.M., Jaffé, R., 2013. Pho-
tochemical and microbial alteration of dissolved organic matter in temperate
headwater streams associated with different land use. J. Geophys. Res. Biogeosci.
118, 1–15.
 T.B. Parr et al. / Ecological Indicators 60 (2016) 970–979
Lutz, B.D., Bernhardt, E.S., Roberts, B.J., Cory, R.M., Mulholland, P.J., 2012. Distinguish-
ing dynamics of dissolved organic matter components in a forested stream using
kinetic enrichments. Limnol. Oceanogr. 57, 76–89.
Maine Department of Environmental Protection (MeDEP), 2005. Maine Land Cover
Database Impervious Cover Layer.
MathWorks, 2012. Matlab R2012b.
McDowell, W.H., 2015. New Hampshire Water Resources Research Center Ana-
lytical Services & Prices, http://www.wrrc.unh.edu/analytical-services-prices
(accessed 22.08.15).
Meyer, J.L., Paul, M.J., Taulbee, W.K., 2005. Stream ecosystem function in urbanizing
landscapes. J. N. Am. Benthol. Soc. 24, 602–612.
Morse, C.C., Huryn, A.D., Cronan, C.S., 2003. Impervious surface area as a predictor
of urbanization on stream insect communities in Maine, USA. Environ. Monit.
Assess. 89, 95–127.
Mulholland, P.J., Houser, J.N., Maloney, K.O., 2005. Stream diurnal dissolved oxy-
gen profiles as indicators of in-stream metabolism and disturbance effects: Fort
Benning as a case study. Ecol. Indic. 5, 243–252.
Niyogi, D.K., Simon, K.S., Townsend, C.R., 2004. Land use and stream ecosystem func-
tioning: nutrient uptake in streams that contrast in agricultural development.
Arch. Hydrobiol. 160, 471–486.
Ohno, T., 2002. Fluorescence inner-filtering correction for determining the humifi-
cation index of dissolved organic matter. Environ. Sci. Technol. 36, 742–746.
Palmer, M.A., Febria, C.M., 2012. The heartbeat of ecosystems. Science 336,
1393–1394.
Parr, T.B., Cronan, C.S., Ohno, T., Findlay, S.E.G., Smith, S.M.C., Simon, K.S., 2015.
Urbanization changes the composition and bioavailability of dissolved organic
matter in headwater streams. Limnol. Oceanogr. 60, 885–900.
Paul, M.J., Bressler, D.W., Purcell, A.H., Barbour, M.T., Rankin, E.T., Resh, V.H., 2009.
Assessment tools for urban catchments: defining observable biological poten-
tial. J. Am. Water Resour. Assoc. 45, 320–330.
Paul, M.J., Meyer, J.L., Couch, C.a, 2006. Leaf breakdown in streams differing in catch-
ment land use. Freshw. Biol. 51, 1684–1695.
Peterson, B.J., et al., 2001. Control of nitrogen export from watersheds by headwater
streams. Science 292, 86–90.
Petrone, K.C., Fellman, J.B., Hood, E., Donn, M.J., Grierson, P.F., 2011. The origin and
function of dissolved organic matter in agro-urban coastal streams. J. Geophys.
Res. 116, G01028.
R Core Team, 2015. R: A Language and Environment for Statistical Computing.
Sandin, L., Solimini, A.G., 2009. Freshwater ecosystem structure–function relation-
ships: from theory to application. Freshw. Biol. 54, 2017–2024.
Schueler, T., 1994. The importance of imperviousness. Watershed Prot. Tech. 1,
100–111.
Simon, K.S., Simon, M.A., Benfield, E.F., 2009. Variation in ecosystem function in
Appalachian streams along an acidity gradient. Ecol. Appl. 19, 1147–1160.
Stanley, E.H., Powers, S.M., Lottig, N.R., Buffam, I., Crawford, J.T., 2012. Contemporary
changes in dissolved organic carbon (DOC) in human-dominated rivers: is there
a role for DOC management? Freshw. Biol. 57, 26–42.
Stedmon, C.A., Bro, R., 2008. Characterizing dissolved organic matter fluorescence
with parallel factor analysis: a tutorial. Limnol. Oceanogr. Methods 6, 1–6.
Stedmon, C.A., Markager, S., 2005a. Tracing the production and degradation of
autochthonous fractions of dissolved organic matter using fluorescence analysis.
Limnol. Oceanogr. 50, 1415–1426.
979
Stedmon, C.A., Markager, S., 2005b. Resolving the variability of dissolved organic
matter fluorescence in a temperate estuary and its catchment using PARAFAC
analysis. Limnol. Oceanogr. 50, 686–697.
Stedmon, C.A., Markager, S., Bro, R., 2003. Tracing dissolved organic matter in aquatic
environments using a new approach to fluorescence spectroscopy. Mar. Chem.
82, 239–254.
Stelzer, R.S., Thad Scott, J., Bartsch, L.A., Parr, T.B., 2014. Particulate organic mat-
ter quality influences nitrate retention and denitrification in stream sediments:
evidence from a carbon burial experiment. Biogeochemistry 119, 387–402.
Tanentzap, A.J., Szkokan-Emilson, E.J., Kielstra, B.W., Arts, M.T., Yan, N.D., Gunn, J.M.,
2014. Forests fuel fish growth in freshwater deltas. Nat. Commun. 5, 1–9.
USEPA, EPA 841-B-03-003 2003. Elements of a State Water Monitoring and Assess-
ment Program.
USEPA, EPA/600/R-14/475F 2015. Connectivity of Streams and Wetlands to Down-
stream Waters: A Review & Synthesis of the Scientific Evidence.
Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., Cushing, C.E., 1980. The
river continuum concept. Can. J. Fish. Aquat. Sci. 37, 130–137.
Walsh, C.J., Roy, A., Feminella, J., Cottingham, P., Groffman, P.M., Morgan, R.P., 2005.
The urban stream syndrome: current knowledge and the search for a cure. J.
North Am. Benthol. Soc. 24, 706–723.
Wang, L., Lyons, J., Rasmussen, P., Seelbach, P., Simon, T., Wiley, M., Kanehl, P., Baker,
E., Niemela, S., Stewart, P.M., 2003. Watershed, reach, and riparian influences on
stream fish assemblages in the Northern Lakes and Forest Ecoregion, U.S.A. Can.
J. Fish. Aquat. Sci. 60, 491–505.
Welch, B.L., 1938. The significance of the difference between two means when the
population variances are unequal. Biometrika 29, 350–362.
Wenger, S.J., et al., 2009. Twenty-six key research questions in urban stream ecol-
ogy: an assessment of the state of the science. J. North Am. Benthol. Soc. 28,
1080–1098.
Williams, C.J., Yamashita, Y., Wilson, H.F., Jaffé, R., Xenopoulos, M.A., 2010. Unrav-
eling the role of land use and microbial activity in shaping dissolved organic
matter characteristics in stream ecosystems. Limnol. Oceanogr. 55, 1159–
1171.
Wisconsin Department of Natural Resources, 2013. Tier I Surface Water Monitoring
Activities.
Woodward, G., et al., 2012. Continental-scale effects of nutrient pollution on stream
ecosystem functioning. Science 336, 1438–1440.
Yamashita, Y., Kloeppel, B.D., Knoepp, J., Zausen, G.L., Jaffé, R., 2011. Effects of water-
shed history on dissolved organic matter characteristics in headwater streams.
Ecosystems 14, 1110–1122.
Yang, L., Hong, H., Guo, W., Huang, J., Li, Q., Yu, X., 2012. Effects of changing land use
on dissolved organic matter in a subtropical river watershed, southeast China.
Reg. Environ. Chang. 12, 145–151.
Yates, A.G., Brua, R.B., Culp, J.M., Chambers, P.A., Wassenaar, L.I., 2014. Sensitiv-
ity of structural and functional indicators depends on type and resolution of
anthropogenic activities. Ecol. Indic. 45, 274–284.
Young, R.G., Collier, K.J., 2009. Contrasting responses to catchment modification
among a range of functional and structural indicators of river ecosystem health.
Freshw. Biol. 54, 2155–2170.
Young, R.G., Matthaei, C.D., Townsend, C.R., 2008. Organic matter breakdown and
ecosystem metabolism: functional indicators for assessing river ecosystem
health. J. North Am. Benthol. Soc. 27, 605–625.