Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/357831190

Urban buried streams: Abrupt transitions in habitat and biodiversity

Article in The Science of The Total Environment · January 2022


DOI: 10.1016/j.scitotenv.2022.153050

CITATIONS READS

3 210

5 authors, including:

Chelsea Hintz Michael Booth


University of Cincinnati University of Cincinnati
1 PUBLICATION 3 CITATIONS 19 PUBLICATIONS 160 CITATIONS

SEE PROFILE SEE PROFILE

Tammy Newcomer Johnson Ken Fritz


United States Environmental Protection Agency United States Environmental Protection Agency
52 PUBLICATIONS 1,560 CITATIONS 91 PUBLICATIONS 3,682 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Chelsea Hintz on 02 February 2022.

The user has requested enhancement of the downloaded file.


Science of the Total Environment 819 (2022) 153050

Contents lists available at ScienceDirect

Science of the Total Environment


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

Urban buried streams: Abrupt transitions in habitat and biodiversity



Chelsea L. Hintz a, , Michael T. Booth a, Tamara A. Newcomer-Johnson b, Ken M. Fritz b, Ishi Buffam a,c
a
Department of Biological Sciences, University of Cincinnati, Cincinnati, OH 45221, United States of America
b
United States Environmental Protection Agency, Office of Research and Development, Cincinnati, OH 45268, United States of America
c
Department of Landscape Architecture and Management, Swedish University of Agricultural Sciences, Alnarp, Sweden

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Stream burial is widespread in industrial-


ized and densely populated urban areas.
• Culverts have dramatic impacts on the
structure and function of streams.
• The buried segments of streams had
smaller substrate and deeper water
depths.
• Buried segments also had depleted periph-
yton and macroinvertebrate populations.
• Transitions in many abiotic and biotic pa-
rameters were abrupt rather than gradual.

A R T I C L E I N F O A B S T R A C T

Article history: Stream burial, the rerouting of streams into underground culverts, is common in industrialized and densely populated
Received 11 August 2021 urban areas. While stream burial is common in urban environments, direct characterization of the within-culvert en-
Received in revised form 3 January 2022 vironment is rare and it is unclear if buried reaches reflect neighboring open reaches regarding habitat, biota, and
Accepted 7 January 2022
water chemistry. Additionally, for a buried stream, the entrance and exit of the culvert are abrupt habitat transitions
Available online 14 January 2022
within the stream channel, and it is unknown if these transitions lead to similarly abrupt responses in biotic and abiotic
Editor: Sergi Sabater characteristics or if responses are gradual. Quantifying the within-culvert environment and transitions upon entering/
exiting the culvert has rarely been done but can help inform management practices regarding how these systems are
Keywords: impacted and establish a baseline for evaluating daylighting or stream restoration projects. To understand how cul-
Urban streams verts affect longitudinal biotic and abiotic characteristics of urban streams, we evaluated longitudinal patterns of phys-
Buried streams ical habitat characteristics, stream water physiochemistry, periphyton biomass, and macroinvertebrate density and
Transitions diversity in two urban streams that included long (>100 m) culvert reaches. Abrupt transitions in a suite of abiotic
Macroinvertebrates and biotic variables were observed at the entrances and exits of the culverts whereas some variables showed no re-
Substrate
sponse to the culvert presence. Periphyton biomass and macroinvertebrate density were reduced by 98% and 92%, re-
Culverts
spectively, by culverts in the two streams. Within the culverts, we observed greater water depths (average of 10 cm
outside vs 26 cm within the culvert), finer benthic substrate, and diversity of macroinvertebrates was reduced by
50%. Nutrient concentrations, in contrast, showed no response to the presence of a culvert. Within 60–90 m down-
stream of the culvert exits, most of the measured parameters returned to levels similar to those observed upstream
of the culvert, suggesting that the ecosystem impacts of urban culverts, though dramatic, may be spatially constrained.
Published by Elsevier B.V.

⁎ Corresponding author.
E-mail address: hintzca@mail.uc.edu (C.L. Hintz).

http://dx.doi.org/10.1016/j.scitotenv.2022.153050
0048-9697/Published by Elsevier B.V.
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

1. Introduction (Macpherson et al., 2012; Wellman et al., 2000). The impact of culverts on
substrate size is not consistent across streams, however, and depends on a
The effects of urbanization on stream ecosystems have been coined the host of factors including channel slope, type of hydraulic control, substrate
“urban stream syndrome” because urbanization induces fundamental envi- embeddedness, and culvert shape, length, and material (Howley, 2004).
ronmental modifications which can change stream ecosystem habitat and Previous work has demonstrated that culverts can contain larger substrate
function (Walsh et al., 2005). These changes in habitat and function can re- (U.S. Department of Agriculture, 2009), smaller substrate (Macpherson
sult from the increase in impervious cover within the watershed and the as- et al., 2012; Wellman et al., 2000), or show no differences in substrate
sociated increase in runoff volume. Changes to sediment regime, thermal size (Hope et al., 2014) when compared to adjacent open stream reaches.
regime, and pollutant loading can also cause changes to stream ecosystem Additionally, downstream of culverts there is often bed scouring leading
habitat and function (Kaushal et al., 2015; Wenger et al., 2009). Urbaniza- to pool formation and subsequent deposition of fine substrates (Wellman
tion can also lead to extreme physical modifications within the stream et al., 2000). Stream burial simplifies and fragments habitat, decreasing
channel, and perhaps the most extreme impact of urban development on habitat availability (Elmore and Kaushal, 2008), and modifying ecosystem
streams is their wholesale containment and burial within culverts or pipes processes such as nutrient retention (Beaulieu et al., 2014).
(Elmore and Kaushal, 2008; Julian et al., 2015). Given the physical changes to the streambed created by burial, changes
Stream burial is widespread in heavily industrialized and densely popu- in water chemistry might be expected within culverts. However, few stud-
lated urban areas (Elmore and Kaushal, 2008). For example, within the U.S. ies have directly observed changes in water chemistry due to stream burial.
cities of Baltimore and Detroit, an estimated 66% and 85% of all streams In one detailed study of a single urban New Hampshire stream, no differ-
have been buried, respectively (Elmore and Kaushal, 2008; Napieralski ences were found between buried and open reaches in a variety of solute
and Carvalhaes, 2016). A broader survey of U.S. Census Bureau Urban concentrations and water chemistry parameters (Hope et al., 2014). Nutri-
Areas in 11 megaregions classified 11,490 km2 of metropolitan areas as ent uptake is slow or undetectable within culverts (Beaulieu et al., 2014,
“urban stream deserts,” where stream channels have likely been removed 2015; Pennino et al., 2014), suggesting that culverts primarily transport nu-
or buried (Napieralski and Carvalhaes, 2016). In Europe, extensive stream trients between open stream reaches with minimal in-stream retention or
burial has been documented in London (Napieralski and Carvalhaes, removal within the culvert.
2016), and country-wide extent of stream burial are 15% and 20% in Research regarding the impact of burial on stream biota has focused on
Denmark and Sweden, respectively (Wild et al., 2011). Burial happens to how a culvert impacts animal (primarily fish and some macroinvertebrates)
streams of various sizes and flow regimes, and culverts can range in length movement (Blakely et al., 2006; Macpherson et al., 2012) and the distribu-
from tens of meters to several hundred meters (Beaulieu et al., 2014; Julian tion of basal resources within streams (Neale and Moffett, 2016). Informa-
et al., 2015; Olson et al., 2017). tion on algae is limited but biomass (as determined by concentrations of
Stream burial abruptly changes key physical drivers of stream ecosys- chlorophyll a) is typically low in culverts and gross primary production is
tems. These changes include the absence of light, the elimination of subsur- not supported (Beaulieu et al., 2014; Hope et al., 2014) due to the lack of
face interaction, and reduced riparian inputs (Fig. 1). Other physical effects light. In contrast, ecosystem respiration is observed in culverts indicating
from stream burial and the presence of a culvert can include increased the presence of heterotrophic organisms and organic matter (Beaulieu
water velocities at the culvert exit (Elmore and Kaushal, 2008; Hope et al., 2014; Hope et al., 2014). Shifts in macroinvertebrate functional feed-
et al., 2014), constricted stream channel width (Beaulieu et al., 2014), lim- ing groups have been observed in culverts relative to the surrounding
ited sediment depth (Hope et al., 2014), and altered substrate composition stream, attributed to changes in the distribution of basal resources (Neale

Fig. 1. Conceptual diagram describing hypothesized changes resulting from stream burial based on literature review. Colored boxes (grey and yellow) represent parameters
directly quantified in our study (with the exception of subsurface interactions and velocity). Grey boxes are primary physical responses and could influence changes in the
yellow boxes which represent secondary effects. Blue arrows between boxes represent the directional influence of one variable on another variable. Black arrows within boxes
represent the anticipated effect on a variable in response to burial (“↑” indicates an increase, “↓” indicates a decrease, and “↑↓” indicates inconsistent change within the
literature).

2
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

and Moffett, 2016). Downstream drift of aquatic invertebrates can be lower observed in open upstream reaches, except at the culvert exit where pool
downstream of culverts and the abundance and diversity of EPT (Ephem- formation and deposition of fine sediments is often observed (Wellman
eroptera-Plecoptera-Trichoptera) invertebrate taxa (taxa indicative of bet- et al., 2000). We expected a downstream spatial lag at the culvert exit,
ter water quality) can be reduced when compared to nearby streams i.e., the presence of a pool and smaller substrate would be present some dis-
without culverts (Meyer et al., 2005). Dispersal can be negatively impacted tance downstream before the stream reach recovered its upstream qualities
because culverts act as upstream flight barriers for adult insects, reducing and characteristics. Water temperature was expected to decrease in the cul-
upstream recruitment (Blakely et al., 2006). The upstream movement of verts due to the lack of solar radiation and gradually recover downstream of
some fish can be negatively impacted by culverts, with culvert slope (> the culvert exit. We expected an increase in inorganic nutrient (N and
3–5%) and hang height (drop to stream) being the most important factors P) concentrations as distance into the culvert increased (due to mineraliza-
impacting upstream fish movement (Burford et al., 2009; Favaro et al., tion rates > uptake rates), and a decrease in inorganic nutrient concentra-
2014; Macpherson et al., 2012; Poplar-Jeffers et al., 2009). tions downstream of the culvert as the periphyton community and
Streams are naturally heterogenous (Allan, 2004) and it is unclear associated primary production recover.
whether the properties of buried stream reaches generally reflect adjacent Regarding biotic parameters, we expected that periphyton biomass
open reaches (regarding habitat, biota, and water chemistry) or if buried within the culvert would be minimal due to the elimination of light
reaches fundamentally differ in structure and function, leading to habitat (Beaulieu et al., 2014). We expected an abrupt decline in periphyton bio-
fragmentation (Beaulieu et al., 2014; Blakely et al., 2006; Elmore and mass upon entering the culvert and an abrupt increase upon exiting the cul-
Kaushal, 2008; Macpherson et al., 2012). To date, research has examined vert to levels observed upstream. We also expected the density and
the impacts of culverts by comparing reaches up- and downstream of cul- diversity of macroinvertebrates to be lower within the culverts compared
verts or by only evaluating a single location within a buried reach. Because to upstream reaches because of the elimination of resources (e.g., algae,
few studies have taken a fine-scale approach in the characterization of bur- leaf litter subsidies; Fuller et al., 2004; Kiffney et al., 2004), and to recover
ied reaches, little is known about the spatial variation of physical, chemical, downstream of the culvert, but with a longer spatial lag than that of periph-
and biological parameters as channels transition from open to buried to yton.
open reaches.
For a buried stream, the entrance and exit of the culvert are abrupt phys-
ical transitions within the stream channel, and it is unknown if these tran- 2. Methods
sitions lead to abrupt changes in biotic and abiotic characteristics or if
changes observed in the stream are more gradual. Abrupt changes in 2.1. Stream description and study design
land-use can impact biota via edge effects by direct and indirect means
(Murica, 1995), but there can be a spatial lag between riparian change We monitored two streams whose buried reaches are >100 m in length
and in-stream conditions (Feijó-Lima et al., 2018). For example, transitions – Congress Run and an unnamed tributary (here and throughout referred to
in land-use (e.g. forest to pasture) have resulted in responses on different as Princeton) from July 2017 – August 2018. Congress Run and Princeton
spatial scales from 10s of m (temperature and sediment size) to 100s of m are first-order tributaries in the Mill Creek Watershed, an urban watershed
(macroinvertebrate diversity and abundance) (Feijó-Lima et al., 2018; in Cincinnati, OH, USA (29.3% impervious cover). Cincinnati has a temper-
Goss et al., 2014). We expect that when a buried stream enters and exits a ate climate with no defined dry and wet seasons and receives an average of
culvert, different ecosystem components will respond in varying ways, 1143 mm of precipitation (primarily rainfall) annually (NOAA, https://
both in the extent of the response and in the spatial lag. www.weather.gov). The Congress Run watershed is 9.9 km2 with 73.6%
Urbanization is expected to increase (US Census Bureau, 2016) leading developed land (mixed industrial and residential use) and 18.8% impervi-
to further stream burial in developing areas (Weitzell et al., 2016), so it will ous surface cover (US Geological Survey, 2016), but retains a riparian cor-
be pertinent to know whether burial leads to additional downstream effects ridor with an average of 75% canopy cover (summer). The stream is
(i.e., a spatial lag in community characteristics) as well as the localized eco- intermittent within the study reach and has a 183 m long buried reach
logical impact of burying the stream. Determining the ecological impact of (Shifflett et al., 2019). Stormwater outfalls and a combined sewer overflow
stream burial – and the spatial extent of this impact – could help inform de- are present along Congress Run. The Congress Run culvert (approximately
cisions involving stormwater management, habitat, nutrient processing, 2 m tall and 3 m wide) is constructed of corrugated metal and runs beneath
flood control, and the monitoring of restoration projects (Newcomer- a golf course (Fig. 2). The study reaches have slopes of 0.89% (upstream),
Johnson et al., 2014; Newcomer Johnson et al., 2016). 0.22% (culvert), and 1.16% (downstream). The Princeton watershed
We aimed to (1) directly characterize the longitudinal spatial variation (8.8 km2) is 95.3% developed land and 38.2% impervious surface cover
in biotic and abiotic ecosystem parameters within buried stream reaches, (US Geological Survey, 2016). Princeton is surrounded by commercial
(2) determine the magnitude of change for a given parameter upon enter- land use but retains an intact riparian corridor (large trees, brush within
ing/exiting a culvert, and (3) evaluate spatial lags in the response of differ- the understory) of 2–3 m on each bank within the study reach (CH, personal
ent parameters to the abrupt transitions at the entrance and exit of a culvert. observation). Princeton is a perennial stream with stormwater outfalls pres-
We evaluated two urban streams that included long (>100 m) culvert ent and includes a 120 m long buried reach. The culvert is constructed of
reaches by carrying out longitudinal transects of physical habitat character- corrugated metal with a poured concrete bottom and runs beneath a high-
istics, stream water physiochemistry, periphyton biomass, and macroinver- way interchange (Fig. 2). The Princeton culvert is approximately 4.4 m tall
tebrate density and diversity. These parameters are often impacted in urban and 4 m wide. The reaches have slopes of 1.02% (upstream), 0.19% (cul-
streams and can be further altered by stream burial. We hypothesized vert), and 0.66% (downstream).
(Fig. 1) that buried stream reaches represent altered ecosystems when com- Our study design included longitudinal surveys including stream
pared to open upstream and downstream reaches and that we would ob- reaches upstream of the culvert, within the culvert, and downstream of
serve clear transitions in biotic parameters (such as periphyton biomass the culvert to characterize the abiotic and biotic spatial variation within
and macroinvertebrates) and some abiotic parameters (such as tempera- reaches and across transitions between open and buried reaches (Fig. 2).
ture) downstream of the entrance and exit of the culvert. Periphyton and physical habitat (such as light, depth, substrate size) were
Our specific expectations regarding abiotic parameters were as follows: collected at 13 evenly spaced sites per stream: 3 sites upstream of the cul-
At the culvert entrance we hypothesized no change in substrate size or vert (upstream), 7 within-culvert sites (culvert), and 3 sites downstream
water depth, given the variable findings of previous research and host of of the culvert (downstream). The distance between sites was 30 m in Con-
factors that influence these parameters (Howley, 2004; Macpherson et al., gress Run and 20 m in Princeton. All measurements described were con-
2012; Wellman et al., 2000). Within the culverts, we expected that the dis- ducted at baseflow as identified by a local Mill Creek USGS stream gage
tribution of substrate size and water depth would be similar to what was (#03259000).

3
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

Fig. 2. a) Map of Congress Run. Distance between sites is 30 m. b) Congress Run, downstream opening of culvert. c) Map of Princeton. Distance between sites is 20 m.
d) Princeton, downstream opening of culvert.

2.2. Physical habitat (light, substrate, water depth) culvert sites corresponded to the entrance, mid-point, and exit of the cul-
vert. Instantaneous measurements of temperature, dissolved oxygen, pH,
The distribution of substrate size was determined at each site (N = and conductivity were collected with a calibrated YSI ProDSS (Yellow
13 sites per stream) in June 2018. At each site, ≥100 pieces of sub- Springs, OH). The YSI ProDSS was calibrated before each day of sampling
strate were randomly selected, and their size (determined by the inter- according to the manufacturer's instructions. Water samples (1 L) were col-
mediate axis) was determined using a gravelometer (Wildco, Yulee, lected in a downstream to upstream direction to determine nutrient concen-
FL). The D18%, D50%, and D84% (the particle size at which 18%, trations and were kept on ice until returned to the lab for filtering. Samples
50%, and 84% of the substrate was smaller and represents the smaller, were filtered (0.45 μm) within 24 h of collection (Millipore MFTM mem-
median, a coarser fraction of substrate; Olsen et al., 2005) were deter- brane filter; Millipore, Billerica, MA) and frozen until analysis. Samples
mined. For culverted sites, the bottom of the culvert was assigned its were analyzed for reactive phosphorus using the ascorbic acid method
own category and was not included in the calculations of D18%, (Murphy and Riley, 1962), for NH+ 4 using the phenol-hypochlorite reaction
D50%, and D84%. Water depth (cm) and an instantaneous measure- (Weatherburn, 1967), and for NO− 3 using a spectrophotometric method
ment of photosynthetically active radiation (PAR, expressed as photo- (Doane and Horwath, 2003). All methods for determining nutrient concen-
synthetic photon flux density, PPFD) were recorded during trations were adapted for a microplate reader (Biotek Synergy H1 Hybrid
periphyton collection (every other month) at each of our 13 sites per Microplate reader; Biotek, Winooski, VT). Dissolved organic carbon
stream. Light readings were collected with a LI-COR LI-192 (LI-COR (DOC) and total dissolved nitrogen (TN) were determined using a coupled
Biosciences, Lincoln, NE) and were collected under varying cloud Shimadzu TOC-VCPH-TNM-1 analyzer. Nutrient concentrations are reported
cover and time of day. in units of mg L−1 of the target element, C, N or P. Appropriate calibration
curves (regarding the range of concentrations expected) were used for each
nutrient and an independent QC check was used in each analysis. If samples
2.3. Water chemistry fell outside the standard curve, samples were rerun with a new calibration
curve. Additionally, seasonal deployments of calibrated MiniDOT tempera-
Stream water quality parameters were measured every other month at 5 ture and oxygen loggers (PME, Vista, CA) were deployed for several days at
locations within each stream corresponding to 1 upstream site, 3 culvert 5 locations within the streams corresponding to 1 upstream, 3 culvert sites,
sites, and 1 downstream site (Fig. 2). The upstream and downstream sites and 1 downstream site (Fig. 2). Logging intervals for these continuous de-
were at the beginning and end of each study reach, respectively. The 3 ployments was set to 10 min.

4
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

2.4. Periphyton biomass (chlorophyll a and ash-free dry mass) (2018). For a given variable, a transition zone is the distance through
which a variable shifts from an upstream stable condition in response to
Deployments of artificial substrate (two 10 cm × 10 cm unglazed ce- an abrupt change, and how far the buffering effects of upstream conditions
ramic tiles attached to a concrete paver, Hauer and Lamberti, 2007) were can be detected (Table 1). Given that our stream reaches included two tran-
used to determine periphyton biomass (N = 13 sites per stream, Fig. 2). Ar- sitions (the entrance/presence of the culvert and then the exit/absence of
tificial substrate was used to control for the variability of substrate between the culvert), we separated our data into two groups: (1) the upstream-
open and culverted reaches (Beaulieu et al., 2014). From July 2017 to Au- culvert transition and (2) the culvert-downstream transition. Transition dis-
gust 2018, artificial substrate was deployed every two months and collected tances were estimated for each group and variable separately.
after 4-week incubations. Samples were processed for chlorophyll a and To estimate transition distances and potential spatial lags, scatterplots
biomass (ash-free dry mass, AFDM). In the field, 2 replicate periphyton of all response variables were first created with longitudinal stream dis-
samples were collected from each paver (i.e., 1 replicate per tile). An area tance as the predictor variable. If a variable showed a monotonic response
of 9–11 cm2 per tile was scraped and the resultant slurry was transferred to the presence/absence of the culvert, the variable was regressed against
to a 500 mL HDPE bottle. Samples were then diluted with deionized distance using a sigmoidal model (Table 1) using the Levenberg-
water to a known volume and a known volume of subsample was filtered Marquardt estimation method in the minpack.lm R package (Elzhov et al.,
onto two 25-mm diameter, 0.45-μm Whatman GF/F filters (one for chloro- 2016). If a variable did not show a monotonic response, transition distances
phyll a and one for AFDM; Wallace et al., 2006). AFDM filters were pre- were not estimated. A monotonic response was assessed qualitatively. As
combusted and pre-weighed prior to filtering (Wallace et al., 2006). After our objectives were to (1) evaluate spatial variation within buried reaches,
filtering, filters (both for chlorophyll and AFDM) were placed into labeled (2) determine the magnitude of change upon entering and exiting a culvert,
aluminum foil packets and placed on ice for further processing. and (3) to evaluate spatial lags in the response of various stream parameters
Chlorophyll a samples were extracted in 95% ethanol for 24 h in dark to the abrupt transitions of entering and exiting the culvert – a sigmoidal
conditions (Nusch, 1980). After extraction, filters were removed, and the model allowed us to address these objectives. For variables that were col-
samples were centrifuged at 1000g for 5 min (Arar, 1997) and 2 mL of liq- lected bi-monthly (water depth, periphyton and macroinvertebrates), the
uid was then used for spectrophotometric chlorophyll analysis. Periphyton mean for each site was used in the regression. Variables were log-
samples were read at 750 and 664 nm, then acidified with 0.1 mL of 0.1 N transformed when necessary to meet the assumptions of the model (homo-
hydrochloric acid and reread at 750 and 665 nm (APHA, 2005) to account geneity of variances and normality). The sigmoidal model was tested
for pheophytin. Chlorophyll a calculations were made using an ethanol ab- against the null model (y ~ 1) using a log-likelihood test and model as-
sorption coefficient (Hansson, 1988). To estimate the amount of AFDM, fil- sumptions were tested using the nlsTools R package (Baty et al., 2015).
ters were dried at 50 °C for 24 h, weighed for dry weight, then combusted The sigmoidal model allowed us to estimate a stable upstream condition
for 2 h at 500 °C to determine final ash weight (Hauer and Lamberti, (yU), stable downstream condition (yD), the location of the midpoint of
2007). For each sampling date, a blank filter was processed following the the change (x0, in m) and the slope of that change (k; Table 1). The maxi-
same protocol for chlorophyll and AFDM to serve as a QC. mum allowable slope (k) in the model was set to correspond to a complete
(90%) transition within the interval from one site to the next (i.e., over the
2.5. Macroinvertebrates (density and diversity) course of either 20 m or 30 m depending on the stream). Allowing values of
k above this level did not measurably change the model fit but increased un-
Macroinvertebrates were collected with a D-net (dimensions: 12″ x 10″, certainty in the k and x0 parameters spuriously. The resulting parameter es-
250-μm mesh) at 13 sites per stream once in May 2018. Since Congress Run timates and their associated uncertainty were used as the input in the
is seasonally intermittent, a spring collection date was chosen to ensure we inverse estimation methodology (described below) to calculate the length
could collect invertebrates from both streams. The D-net was placed ap- of transition distances.
proximately 5 cm into the sediment and the sediment was disturbed for Transition distances were estimated separately for the upstream-culvert
0.5 m upstream of the net for 60 s, giving an equal sampling effort at and culvert-downstream transition. Transition distances were estimated
each site (Resh and Jackson, 1993). Samples were transferred into 1-L bot- using inverse estimation in R with the investR package (Greenwell and
tles with distilled water, placed on ice, and returned to the lab. Within 24 h Schubert Kabban, 2014). Inverse estimation infers the possible values of
upon returning to the lab, samples were washed onto a 250-μm sieve and an explanatory variable (e.g., distance) given values of a response variable
material >250 μm in size was preserved with 80% ethanol (Resh and based on a model's confidence interval (Feijó-Lima et al., 2018). For each
Hauer, 2017). Samples were subsequently sorted under a dissecting micro- transition, the beginning and end of the transition were estimated. yU ±
scope, and macroinvertebrates were identified to family (Cummins et al., SE and yD ± SE from the sigmoidal model were used as the starting values
2008). in the inverse estimation to determine the beginning and the end of the
transition, respectively. Bootstrap simulations (1000) were run and the me-
2.6. Statistical analyses dian value of the bootstrap estimate was used to determine xU (the distance
where the modeled response variable started to deviate from modeled sta-
To address the spatial extent of responses from stream burial, transition ble upstream condition, yU) and xD (the distance where the modeled re-
zones were estimated following the methods described in Feijó-Lima et al. sponse variable started to reach a stable downstream condition, yD).

Table 1
Model used to evaluate the dependence of a given ecosystem parameter on longitudinal distance, to determine transition responses at the entrance or exit of a
culvert. yD represents the stable downstream condition, yU, the stable upstream condition, d, is distance, x0 is the midpoint of the sigmoid in meters, and k is
the slope. The red boxes represent the transition zone for a given variable.
Model Function Rise Decay

yU−yD
Sigmoid logistic f ðdÞ ¼ yU−  −k
d
1þ x0

5
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

There are limitations to this method and transition locations (xU and xD) 3.3. Water depth
cannot be estimated when there are wide SEs surrounding yU and/or yD.
Lastly, a Spearman rank correlation analysis was performed to determine In both streams, water depth increased as distance into the culverts in-
the correlation between some of our biological parameters. All statistical creased (Fig. 3b). At the culvert exits, deep pools were present that were
analyses were performed in R v. 3.6.2 (R Core Team, 2019). Data from not observed anywhere else within our study reaches. In Congress Run,
field datasheets were checked for data entry errors before analysis. mean depth upstream of the culvert was shallow (7 ± 2 cm; mean ± SE)
whereas within the culvert, mean depth was much greater (35 ± 2 cm)
(Table 2). Downstream of the culvert, mean depth became shallow again
3. Results
(15 ± 3 cm). Habitat outside of the culvert (both up- and downstream)
consisted of riffles and runs, transitioning to pool habitat within the culvert.
3.1. Light
For the upstream-culvert transition in Congress Run, a transition in depth
was detected within the first 30 m of culvert length and a new stable condi-
Light (as PAR) was present upstream and downstream of the culvert
tion was detected within 60 m into the culvert (Fig. 4, Supplementary
while PAR levels dropped to 0 within the culvert (Table 2). Within Congress
Table 1). The transition zone at the culvert exit for depth was detected
Run, instantaneous measurements of light averaged 55 ± 20 μMol
within 30 m downstream of the culvert exit and a new stable condition
m−2 s−1 (mean ± SE) in the upstream reach and averaged 72 ± 22
was detected within 30 m of the exit (Fig. 4, Supplementary Table 1).
μMol m−2 s−1 in the downstream reach across sampling months. In
In Princeton, mean depth upstream of the culvert was shallow (7 ±
Princeton, instantaneous measurements of light averaged 113 ± 32 μMol
1 cm) whereas within the culvert mean depth was greater (17 ± 3 cm)
m−2 s−1 in the upstream reach and averaged 122 ± 36 μMol m−2 s−1 in
(Table 2). Downstream of the culvert, mean depth decreased but only
the downstream reach across sampling months. Our study design did not
slightly (13 ± 2 cm). Habitat upstream of the culvert consisted of riffles
allow us to quantitatively assess transitions in light (sampling sites were
and runs, transitioning to pool habitat within the culvert. Downstream of
too widely separated, streamside canopy varied, and sampling time of
the culvert, habitat consisted of pools, riffles, and runs. Depths within the
day was not standardized), but within 20 m (Princeton) or 30 m (Congress
culvert were more variable across sampling months in Princeton when com-
Run) into the culvert PAR levels had dropped to <0.1 μMol m−2 s−1.
pared to Congress Run (Fig. 3b). At the culvert entrance in Princeton, a tran-
sition zone for depth was detected within the first 20 m of the culvert
3.2. Distribution of substrate size entrance and a new stable condition near the middle of the culvert
(Fig. 4, Supplementary Table 1). Upon exiting the culvert in Princeton, no
In both streams, substrate size, as represented by D84%, was smaller transition in depth could be detected within our 60 m downstream study
within the respective culverts than outside (Fig. 3a). In Congress Run, sub- reach (p = 0.4253, log-likelihood test between null model and the sigmoi-
strate was predominantly cobble outside of the culvert and fine gravel dal model; Table 3).
within the culvert (Table 2). The transition zone at the entrance of the cul-
vert was detected near the culvert entrance (Table 3) and the transition in 3.4. Water chemistry
substrate at the culvert exit was detected within 30 m downstream of the
exit and extended beyond the end of the 90 m downstream study reach Nutrient levels within Congress and Princeton (Table 2, Supplementary
(Fig. 4, Supplementary Table 1). In Princeton, substrate was predominantly Fig. 1) showed no spatial variation or relationship to culvert presence. In-
cobble outside of the culvert (both up- and downstream of the culvert) and stantaneous measurements of temperature, dissolved oxygen, pH, and con-
coarse gravel within the culvert (Table 2). The first transition zone in ductivity did not vary with culvert presence (Table 2) and, therefore, no
Princeton was detected within the first 20 m of the culvert, and the begin- transition zones were estimated for any water physiochemistry variables.
ning of the second transition was detected approximately 20 m downstream In Congress Run, dissolved oxygen ranged from 1.35–11.76 mg L−1, spe-
of the culvert exit (Fig. 4, Supplementary Table 1). cific conductance ranged from 343 to 1084 μs cm−1, and stream pH ranged

Table 2
Summary table of raw data. Values are mean ± SE. Values are averaged across sampling dates and the entrances and exits of the culverts have been excluded. Temperature
refers to instantaneous point measurements while Diel Temperature Range refers to data from continuous short term deployments. Density and family richness refer to mac-
roinvertebrates.
Congress Run Princeton

Variable Upstream Culvert Downstream Upstream Culvert Downstream

Water Depth (cm) [max] 7 ± 2 [24] 35 ± 2 [60] 15 ± 3 [39] 7± 1 [15] 17 ± 3 [48] 13 ± 2 [30]
Substrate D18% (mm) 14 ± 5 4±1 7±1 12 ±2 8±3 7±2
Substrate D50% (mm) 36 ± 4 8±2 24 ± 3 41 ±4 15 ± 3 24 ± 11
Substrate D84% (mm) 73 ± 9 16 ± 4 85 ± 21 115 ± 13 39 ± 15 56 ± 18
Temperature (°C) (instantaneous) 18 ± 3 17 ± 3 15 ± 2 16 ±4 17 ± 4 15 ± 3
Diel Temperature Range (max - min °C) (continuous)
Fall 1.97 ± 0.182 0.67 ± 0.210 6.18 ± 3.43 4.30 ± 0.31 2.16 ± 0.30 2.59 ± 0.14
Spring 4.06 ± 1.05 – – 5.64 ± 0.65 5.32 ± 0.62 5.08 ± 0.60
Summer 0.79 ± 0.12 0.33 ± 0.13 6.39 ± 0.19 5.81 ± 1.07 4.82 ± 0.48 4.30 ± 0.04
pH 7.9 ± 0.1 8.0 ± 0.1 7.7 ± 0.1 8.2 ± 0.1 8.2 ± 0.1 8.1 ± 0.1
Conductivity (SPC) 653 ± 97 607 ± 108 693 ± 96 1104 ± 95 1079 ± 99 1163 ± 90
PAR (μMol m−2 s−1) 55 ± 20 0.0 ± 0.00 72 ± 22 113 ± 32 0.01 ± 0.01 122 ± 36
−1
NH+4 (mg-N L ) 0.03 ± 0.01 0.03 ± 0.02 0.03 ± 0.01 0.01 ± 0.01 0.01 ± 0.01 0.02 ± 0.01
NO−3 (mg-N L
−1
) 0.33 ± 0.09 0.42 ± 0.06 0.27 ± 0.08 0.20 ± 0.12 0.21 ± 0.11 0.22 ± 0.12
TN (mg L−1) 0.58 ± 0.13 0.69 ± 0.10 0.48 ± 0.12 0.42 ± 0.09 0.43 ± 0.10 0.47 ± 0.10
PO3−
4 (mg-P L−1) 0.07 ± 0.01 0.06 ± 0.01 0.06 ± 0.01 0.03 ± 0.01 0.03 ± 0.01 0.02 ± 0.01
DOC (mg L−1) 2.8 ± 0.4 3.1 ± 0.4 2.4 ± 0.6 3.1 ± 0.2 3.1 ± 0.2 3.1 ± 0.2
Chlorophyll (mg m−2) 9±3 0.4 ± 0.1 6±2 10 ± 3 0.2 ± 0.3 10 ± 3
AFDM (g m−2) 7±2 1 ± 0.2 8±1 7±1 1 ± 0.1 5±1
Density (individual m−2) 136 ± 71 5±3 229 ± 142 260 ± 42 31 ± 10 274 ± 165
Family Richness 1.7 ± 0.6 0.5 ± 0.3 3.2 ± 0.7 3.4 ± 0.6 1.8 ± 0.2 3.9 ± 0.9

6
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

Fig. 3. Substrate and water depth for Congress Run and Princeton. In both panels, the culverts are represented by the grey boxes, and vertical dashed lines are the culvert
entrance/exit. Flow direction is from left to right. a) Substrate size for D18% (circles), D50% (triangle), and D84% (square) percentiles b) Site water depth for each
stream and sampling date. The black line represents the mean water depth across all sampling dates. Note: the y-axis is reversed in panel b.

from 7.46–9.10 across all sampling months and sites. In Princeton, dis- 3.5. Periphyton biomass – chlorophyll a
solved oxygen ranged from 7.29–12.75 mg L−1, specific conductance
ranged from 975 to 1357 μs cm−1, and pH ranged from 7.77–8.68 across In both streams, chlorophyll collected from artificial substrates was
all sampling months and sites. Culvert sites had a smaller diel temperature much lower within the culvert than in both the upstream and downstream
range during fall than the open upstream and downstream sites (Table 2). reaches (Fig. 5a, Table 2). In Congress Run chlorophyll decreased from 9 ±

Table 3
Parameter estimates for the stable upstream condition (yU), stable downstream condition (yD), the midpoint of the sigmoid (x0), and the slope of the sigmoid (k). P-values
represent the result of a log-likelihood test between the null model and sigmoidal model. If the sigmoidal model did not explain more variation than the null model, then no
sigmoidal model was fit. ND = stable condition not detected within our study reach. * = upper slope (k) limit was reached.
Variable yU ± SE yD ± SE x0 ± SE k ± SE p

Congress Run: Upstream - Culvert


Macroinvertebrate Family Richness – – – – 0.0919
Macroinvertebrate Density (individual m−2) – – – – 0.0502
AFDM (g m−2) 6.1 ± 0.8 1.1 ± 0.5 75 ± 56 21 ± 82.7* < 0.0005
Chlorophyll a (mg m−2) 11.9 ± 0.9 0.4 ± 0.6 57 ± 340 21 ± 3347.5* < 0.0005
Substrate D84% 82.2 ± 7.9 0 ± 16.1 85 ± 20 2.3 ± 1.1 < 0.0005
Water Depth (cm) 6.4 ± 3.4 41.3 ± 4.5 113 ± 11 10.5 ± 7.1 < 0.0005

Congress Run: Culvert - Downstream


Macroinvertebrate Family Richness 0.4 ± 0.27 4.9 ± 1.9 316 ± 26 13.5 ± 9.6 < 0.0005
Macroinvertebrate Density (individual m−2) 2±2 1795 ± 9296 314 ± 57 10.7 ± 11.9 0.0015
AFDM (g m−2) 1.1 ± 0.6 9.9 ± 1.5 285 ± 13 13.4 ± 6.8 < 0.0005
Chlorophyll a (mg m−2) 1.3 ± 0.42 ND ND 4.9 ± 8.3 0.0006
Substrate D84% 12.6 ± 8.7 306 ± 945 377 ± 220 9.6 ± 10.5 0.0001
Water Depth (cm) 34.7 ± 4.4 11.1 ± 8.1 298 ± 34 56 ± 984* 0.0289

Princeton: Upstream - Culvert


Macroinvertebrate Family Richness 3.4 ± 0.4 1.9 ± 0.4 60 ± 8 15.8 ± 155.2 0.0214
Macroinvertebrate Density (individual m−2) 260 ± 30 31 ± 24 58 ± 66 21 ± 676.7* < 0.0005
AFDM (g m−2) 7.4 ± 1.2 0.7 ± 1.2 75 ± 31 21 ± 124.8* 0.0028
Chlorophyll a (mg m−2) 10.4 ± 2.5 0 ± 2.6 64 ± 17 5.5 ± 8.1 0.0111
Substrate D84% 117.6 ± 17.6 22.6 ± 3.0 75 ± 16 21 ± 70.7* < 0.0005
Water Depth (cm) 7.4 ± 0.5 21.5 ± 1.2 102 ± 5 7.2 ± 2.0 < 0.0005

Princeton: Culvert - Downstream


Macroinvertebrate Family Richness – – – – 0.0592
Macroinvertebrate Density (individual m−2) 20 ± 9 182 ± 106 161 ± 28 12.6 ± 16.1 0.0076
AFDM (g m−2) 1.0 ± 0.8 4.9 ± 1.1 181 ± 26 56 ± 1102* 0.014
Chlorophyll a (mg m−2) 0.3 ± 1.3 9.8 ± 1.5 179 ± 22 56 ± 965* 0.0009
Substrate D84% 24.4 ± 6.5 67.4 ± 15.8 204 ± 29 56 ± 339* 0.0151
Water Depth (cm) – – – – 0.4253

7
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

Fig. 4. Transition zones detected for biotic and abiotic parameters in Congress Run and Princeton. In both panels, grey boxes represent the culvert and vertical dashed lines
are the culvert entrance/exit. Flow direction is from left to right. The start and end of the transition zones represent the medians from the inverse estimation method. Green
brackets represent the upstream-culvert transition while dark grey brackets represent the culvert-downstream transition. Horizontal dashed lines represent transitions where
the end of the transition could not be detected.

3 mg m−2 (mean ± SE) upstream of the culvert to 0.4 ± 0.1 mg m−2 In Princeton, a chlorophyll transition was detected at the culvert en-
within the culvert and recovered to 6 ± 2 mg m−2 downstream of the cul- trance and a new stable condition was detected before the next sampling
vert. In Princeton, chlorophyll decreased from 10 ± 3 mg m−2 upstream of point 20 m downstream (Fig. 4, Supplementary Table 1). For the culvert-
the culvert to 0.2 ± 0.3 mg m−2 within the culvert and recovered to 10 ± downstream transition in Princeton, a chlorophyll transition was detected
3 mg m−2 downstream of the culvert. Chlorophyll remained consistently at the culvert exit (Fig. 4) and a new stable condition was detected by the
low throughout the culverts as distance into the culverts increased. Chloro- next sampling site (20 m downstream of the culvert exit).
phyll upstream and downstream of the culverts showed considerable sea-
sonal variation in Congress Run and Princeton (Fig. 5a) while this 3.6. Periphyton biomass - ash-free dry mass
seasonal pattern was absent within the culverts.
Across sampling dates, in Congress Run, a decrease in chlorophyll In both streams, periphyton biomass (as AFDM) was present on incu-
and a new stable condition were detected upstream of the culvert en- bated artificial substrates at lower levels within the culverts than within
trance (approximately 30–60 m upstream of the culvert entrance; the open reaches up- and downstream of the culverts (Fig. 5b, Table 3). In
Fig. 4, Supplementary Table 1). This indicates a new stable condition Congress Run, periphyton biomass decreased from 7 ± 2 g m−2 (mean ±
was detected upstream of the culvert and then was maintained for SE) upstream of the culvert to 1 ± 0.2 g m−2 within the culvert and recov-
most of the culvert length. The cause of the transition occurring up- ered to 8 ± 1 g m−2 downstream of the culvert. In Princeton, periphyton
stream of the culvert is not known, but it could result from the inter- biomass decreased from 7 ± 1 g m−2 upstream of the culvert to 1 ±
mittent flow regime of Congress Run as some of the upstream sites 0.1 g m−2 within the culvert and recovered to 5 ± 1 g m−2 downstream
were dry at the time of sampling or, could represent a limitation of of the culvert. Periphyton biomass displayed no seasonal variation within
our study given that we had set distances between sampling points the culvert, whereas upstream and downstream reaches for both culverts
and it could be by chance that this particular site had lower periphy- showed seasonal variation (Fig. 5).
ton biomass. At the culvert exit, chlorophyll began to increase but a In Congress Run, periphyton biomass began to decline 30 m upstream of
downstream stable condition was not detected downstream within the culvert entrance and a new stable condition was detected as the stream
our 90 m study reach (Table 3). entered the culvert (Fig. 4). At the culvert-downstream transition in

8
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

Fig. 5. Concentrations of chlorophyll a (mg m−2; panel a) and ash-free dry mass (AFDM g m−2; panel b) on artificial substrate for each sampling date and location within the
study streams. In both panels, the culverts are represented by the grey boxes, the vertical dashed lines are the culvert entrance/exit, and the thick black line represents the
mean across all sampling dates. Flow direction is from left to right.

Congress Run, periphyton biomass began to increase upstream of the cul- culvert entrance and 20 m downstream of the exit; Fig. 4, Supplementary
vert exit (within 30 m) and the transition ended 30–60 m downstream of Table 1).
the culvert exit (Fig. 4, Supplementary Table 1). In Princeton, transitions
in periphyton biomass were detected at the culvert entrance (a decrease 3.7. Macroinvertebrate density
in AFDM) and exit (an increase in AFDM). Both of these transitions were
relatively short and stable conditions were detected before the next down- In both streams, macroinvertebrates had lower densities inside the
stream sampling location for each transition (20 m downstream of the cuvlert than upstream and downstream of the culverted reach (Fig. 6a;

Taxonomic Order

Diptera

Ephemeroptera

Oligochaetes

Other

Plecoptera

Trichoptera

Fig. 6. a) Mean macroinvertebrate density, b) Mean taxonomic family richness, and c) Proportion of sampled macroinvertebrate taxonomic orders a given site and stream.
Error bars in panels a) and b) represent the standard error. In all panels, the culverts are represented by the grey boxes and vertical dashed lines are the culvert entrance/exit.
Flow direction is from left to right.

9
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

Table 2). In Congress Run, macroinvertebrate density decreased from burial indeed serves as an additional stressor on stream ecosystems poten-
136 ± 71 individuals m−2 (mean ± SE) upstream of the culvert to 5 ± tially leading to habitat fragmentation.
3 individuals m−2 within the culvert. At some culvert sites in Congress
Run, no macroinvertebrates were collected. Downstream of the culvert, 4.1. Spatial variation and magnitude of change within buried reaches
macroinvertebrate density increased to 229 ± 142 individuals m−2 in Con-
gress Run. The upstream-culvert transition was not modeled because our Our work highlights that even in degraded urban systems, culverts have
hypothesis (i.e. the sigmoidal model) did not explain more variation than dramatic impacts on the structure and function of streams. Urban streams
the null model (p = 0.0502, log-likelihood test between null model and are typically characterized by altered channel morphology and reduced bi-
the sigmoidal model; Table 3). Upon exiting the culvert in Congress Run, otic richness (Walsh et al., 2005) and stream burial may serve as an addi-
a transition in macroinvertebrate density was detected at the culvert exit tional stressor on urban ecosystems. We observed that substrate size,
and a stable condition was detected before the next downstream sampling periphyton biomass (chl a and AFDM), and macroinvertebrate abundances
point (30 m downstream). decreased and remained low throughout the culverts in our study.
In Princeton, macroinvertebrate density decreased from 260 ± 42 indi- In-stream habitat can serve as a control on the presence and diversity of
viduals m−2 (mean ± SE) upstream of the culvert to 31 ± 10 individuals biota and we observed that physical habitat was altered within the culverts.
m−2 within the culvert. Downstream of the culvert in Princeton, macroin- Specifically, deeper water depths and smaller substrates were observed.
vertebrate density increased to 274 ± 165 individuals m−2. In Princeton, Water depth increased with distance into the culvert and the culverts eval-
macroinvertebrate density decreased surrounding the culvert entrance uated in this study created pool-like environments at their downstream end.
and a stable culvert condition was detected before the next downstream These pools continued downstream of the culvert exits and pools of this
sampling point (20 m downstream). For the culvert-downstream transition depth were not observed anywhere else within our studied stream reaches.
in Princeton, macroinvertebrate density increased between 20 and 40 m Based on our observation that pavers at within-culvert sites washed away
upstream of the culvert exit and a downstream stable condition was de- 48% of the time during the study period compared with only 9% of pavers
tected near the culvert exit (Fig. 4, Supplementary Table 1). Macroinverte- from sites outside of the culverts, culverts had higher velocities within the
brate density in the Princeton culvert recovered to upstream levels before culverts during rain events. The physical habitat within our culverts repre-
the culvert exit (Table 2). Lastly, overall, macroinvertebrate densities in sented a different habitat type than was observed elsewhere in our study
both streams were correlated with periphyton biomass (as AFDM) (via reaches and this was also reflected in the presence and diversity of biota
Spearman rank correlation, p < 0.005, r = 0.54). within the culverts.
Biotic communities were negatively impacted by the presence of the
3.8. Macroinvertebrate family richness culverts – specifically, we observed a loss of periphyton biomass and
lower macroinvertebrate density and taxa richness in culverts than in up-
Macroinvertebrate diversity was low in both streams (Fig. 6b, Table 2) stream/downstream reaches. The lack of light is an obvious limitation on
regardless of stream reach and was primarily composed of Chironomidae periphyton biomass production, and the availability of periphyton can be
and Oligochaeta (Fig. 6c). The number of families present in the culvert a direct control on macroinvertebrates (Kiffney et al., 2004). Macroinverte-
was lower than in reaches up- and downstream of the culverts (Fig. 6b). brate communities can also be influenced by substrate size (Cummins et al.,
An average of 0.5 and 1.8 families were found within the culvert in Con- 2008). Smaller-sized substrates provide less stable habitat and are nega-
gress Run and Princeton, respectively and at some within-culvert sites in tively correlated with macroinvertebrate density and diversity (Allan and
Congress Run, no macroinvertebrates were found. Castillo, 2007; Roy et al., 2003) and our study culverts were dominated
In Congress Run, macroinvertebrate family richness was higher up- by these smaller substrates. Our upstream and downstream reaches had
stream than within the culvert but our hypothesis (i.e., the sigmoidal wider distributions of substrate size and larger predominant substrate
model) did not explain more variation than the null model (p = 0.09) so which corresponded to increased macroinvertebrate density and diversity.
the upstream-culvert transition was not modeled (Table 3). For the Similar relationships have been observed between substrate size and mac-
culvert-downstream transition in Congress Run family richness began to in- roinvertebrates in other urban streams (Roy et al., 2003; Wood and
crease within 30 m downstream of the culvert exit and a stable downstream Armitage, 1997).
level of family richness was detected by 60 m downstream of the culvert In both streams, substrate size, periphyton biomass, and macroinverte-
exit (Fig. 4, Supplementary Table 1). In Princeton, transitions in macroin- brate abundances generally recovered downstream of the culvert to levels
vertebrate diversity were detected at the culvert entrance (a decrease in di- similar to those observed upstream of the culvert. A new stable downstream
versity). This transition was short, and a stable level of macroinvertebrate condition could be detected within our study reach (within 90 m for Con-
richness was detected before the next downstream sampling location for gress Run and within 60 m for Princeton) for most variables (but see
each transition (20 m downstream of the culvert entrance) (Fig. 4, Supple- Fig. 4) suggesting the impacts from the culvert are spatially constrained
mentary Table 1). Downstream of the culvert in Princeton, macroinverte- for the measured variables and at the spatial scale we monitored. The
brate family richness increased but our hypothesis (i.e., the sigmoidal downstream recovery of macroinvertebrates that we observed is encourag-
model) did not explain more variation than the null model (p = 0.42) so ing and suggests that the impact of culverts may be effectively limited to
the culvert-downstream transition was not modeled (Table 3). within the culvert itself. However, the macroinvertebrate communities
were already impacted upstream of the culvert (i.e. low density and a com-
4. Discussion munity comprised of taxa indictive of poor water quality in these urban
streams (Alberts et al., 2018; Cummins et al., 2008; MBI, 2012) so it re-
We aimed to directly characterize spatial variation in ecosystem param- mains to be seen if a more diverse macroinvertebrate community in a less
eters within buried stream reaches, determine the magnitude of change degraded stream would recover to the same extent downstream of the bur-
upon entering/exiting a culvert, and to evaluate how different parameters ied segment.
responded spatially to the abrupt transitions of entering and exiting the cul- A suite of measured water chemistry variables (including dissolved N, P,
vert. Our data show that the within-culvert stream environment was consis- and DOC) did not change measurably upon entering the culverts or exhibit
tently altered across a suite of parameters including water depth, substrate any variation with distance into the culverts. This supports previous work
size, periphyton biomass, and macroinvertebrate density and taxa richness showing that culverts primarily serve as transporters of nutrients rather
(Figs. 3, 5, and 6). We also found that there was little spatial variation than transformers (Beaulieu et al., 2014; Hope et al., 2014; Pennino et al.,
within buried reaches, as a stable condition was detected within the cul- 2014). In areas with high levels of urbanization and stream burial, culverts
verts for most variables. Culverts have been suggested as agents of fragmen- may represent a missed opportunity for nutrient processing and thus con-
tation in lotic systems (Fuller et al., 2015) and here we demonstrate that tribute to the downstream export of nutrients (Beaulieu et al., 2015).

10
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

4.2. Transition distances For the upstream-culvert transition, we hypothesized that all transitions
would begin at or downstream of the culvert entrance. Unexpectedly, two
To the best of our knowledge, this is the first detailed spatial character- transitions were detected upstream of the culvert entrance (Congress Run,
ization of within-culvert environments. We aimed to evaluate how various AFDM and chlorophyll a). The cause of this is not known but it could result
stream parameters responded to the abrupt transitions of entering and from the intermittent flow regime of Congress Run as some of the upstream
exiting the culvert and found that individual parameters responded some- sites were dry at the time of sampling or, could represent a limitation of our
what differently. Estimated transition zone lengths from the inverse estima- study given that we had set distances between sampling points and it could
tion methodology ranged from 11 to 106 m for different parameters in be by chance that this particular site had lower periphyton biomass.
Congress Run and from 6 to 67 m in Princeton (Supplementary Table 1).
Our calculated transition zones were somewhat shorter than those previ- 4.3. Study limitations
ously reported in other stream ecosystem studies (described below), but
we are not aware of other studies that have examined transitions in re- The analysis and interpretation was practically constrained by the lim-
sponse to the entrance/exit of a culvert. ited number of measurement locations (13 per stream), and the longitudi-
The abiotic parameters that displayed transitions in our systems were nal spacing of our study sites set the minimum extent over which we
substrate size and water depth. A change in substrate size is often associated could detect transitions. The spacing was set at 20–30 m as a balance of
with riparian change in streams and has been previously detected on scales practical considerations and our expectation of transitions in the 10s to
of hundreds of meters in other regions (Chakona et al., 2009; Feijó-Lima 100 s of m distance. In fact, many of our measured parameters transitioned
et al., 2018; Goss et al., 2014). When a riparian corridor transitions from quite abruptly, often in less than the interval between two consecutive sam-
forest to pasture, a shift to smaller substrates has been detected, attributed pling locations. While this result is relevant and important and indicates
to increased bank erosion (Chakona et al., 2009; Feijó-Lima et al., 2018). that in already-urbanized streams the ecosystem effects of culverts may
The opposite pattern has been detected when land-use surrounding streams be relatively transient, we were not able to precisely estimate distances
transitions from agriculture to forest – with larger substrate detected down- over which many of the more abrupt transitions occur. Follow-up studies
stream of the transition (Goss et al., 2014). In our culverted streams, we ob- would ideally use a study design with smaller spacing between sampling
served both of these changes abruptly at the entrance and exit of the points, as well as extending further up- and down-stream to ensure that
culvert, respectively. Transitions in water depth have been observed in re- all transitions are captured fully. Finally, there were some potentially im-
sponse to riparian changes and over long distances (>500 m) (Goss et al., portant ecosystem variables that were not directly measured in our study,
2014). As with substrate size, we observed 2 abrupt transitions in water for instance stream velocity (Fig. 1), which could vary in response to culvert
depth as the stream entered and exited the culverts. presence.
We generally did not observe transitions in water chemistry and water
temperature in response to the presence/absence of a culvert. Transitions 4.4. Conclusions: implications for stream monitoring and restoration
in water chemistry have been observed in response to riparian transitions
in other settings (Chakona et al., 2009; Goss et al., 2014) and were expected Urbanization is projected to increase globally through 2050 (United
here as well due to the large reduction in periphyton biomass within the Nations, 2019) and increasing urbanization increases the likelihood of
culvert. Given that changes in stream water temperature have been de- stream burial (Napieralski and Carvalhaes, 2016). While we have evaluated
tected in less extreme conditions (e.g., entering long stretches of shade; stream burial at a reach scale, our results suggest that the impact of burial
Rutherford et al., 2004; Storey and Cowley, 1997), we were surprised for our monitored parameters at a stream network scale could be signifi-
that impacts on water temperature (via instantaneous point measurements) cant. This has also been suggested by previous work which evaluated the
were relatively minor within the culvert given the complete elimination of impact of stream burial on nutrient processing (Beaulieu et al., 2015). We
solar radiation and slow water velocity. have shown that culverts are altered habitats when compared to upstream
Generally, macroinvertebrate density, richness, and periphyton bio- and downstream reaches, but they are not devoid of biota. Biotic parame-
mass were suppressed within 20–30 m of the stream entering a buried ters (periphyton and macroinvertebrates) were reduced by the presence
reach, indicating that burial has an almost immediate effect on the biolog- of the culvert but recovered downstream of the culvert, suggesting that cul-
ical structure of urban streams. This effect persisted throughout the verts could represent promising places for restoration projects because the
length of the buried reach, demonstrating that burial serves as an addi- within-culvert habitat is greatly degraded compared to upstream and
tional stressor that can further limit an already depauperate urban stream downstream reaches. Restoration projects are often completed on a reach-
biological community. Periphyton biomass displayed abrupt transitions size scale and culverts could represent ideal locations for these projects.
(within 20–30 m, with the exception of the culvert-downstream transition The United Nations has deemed 2021–2030 the Decade on Ecosystem Res-
in Congress Run) as expected given the elimination of light in the cul- toration (United Nations, 2021) and our work provides a timely and valu-
verts. Previous work has shown that a change in macroinvertebrates can able starting point for how these types of systems could be impacted and
be detected for hundreds of meters downstream before reaching a stable where potential improvements to habitat and biodiversity could be made.
condition in response to a riparian change (Chakona et al., 2009; Goss Restoration projects in aquatic systems occur on a spectrum and restora-
et al., 2014; Storey and Cowley, 1997; Tanaka et al., 2015), but our mac- tion in buried streams can range from daylighting (i.e. culvert removal) or
roinvertebrate parameters had very short transition zones at the culvert altering culvert design to increase connectivity and passage for aquatic or-
entrances and exits (with the exception of the upstream-culvert transition ganisms (e.g. installation of baffles or the implementation of “stream simu-
in Congress Run). Our results do not match our a priori hypothesis that we lation design”). While culvert removal may not be possible in all locations,
would observe changes in chlorophyll and AFDM first and then we would improvements to culvert design could be implemented. Past work has
observe changes in macroinvertebrate parameters with a spatial lag shown that cost-sharing between transportation agencies and conservation
downstream (consistent with bottom-up control; Goss et al., 2014; groups can increase both groups' return-on-investment when the replace-
Kiffney et al., 2004). Transitions in these parameters were detected be- ment of aging, failing, and/or under-sized culverts is paired with surface
fore our second within-culvert site in each stream (i.e., 20 or 30 m into road improvements and culvert improvements (Neeson et al., 2018).
the culvert) so we were unable to determine if our a priori hypothesis While restoration projects would ideally be implemented at a
was supported at a finer spatial scale. But, overall, macroinvertebrate watershed-scale, large scale projects are often difficult to implement in
densities were correlated with periphyton biomass (as AFDM). The mac- urban environments (Bernhardt and Palmer, 2011) because of existing in-
roinvertebrate communities were dominated by Chironomidae and frastructure. Our data suggest that there is potential for measurable im-
Oligochaeta which are primarily collector-gatherers feeding on deposited provements for a suite of parameters where restoration (via culvert
fine organic matter (Cummins et al., 2008). removal, i.e., daylighting) is feasible. We recognize that restoration is

11
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

expensive and can have various goals (Bernhardt et al., 2005) and suggest Baty, F., Ritz, C., Charles, S., Brutsche, M., Flandrois, J.-P., Delignette-Muller, M.-L., 2015. A
toolbox for nonlinear regression in R: the package nlstools. J. Stat. Softw. 66 (5), 1–21.
that restoration via daylighting should be considered a component of Beaulieu, J.J., Mayer, P.M., Kaushal, S.S., Pennino, M.J., Arango, C.P., Balz, D.A., Canfield,
watershed-scale restorations that can provide broader ecosystem services. T.J., Elonen, C.M., Fritz, K.M., Hill, B.H., Ryu, H., Santo Domingo, J.W., 2014. Effects
Further, we suggest that culverts should be included in monitoring of urban stream burial on organic matter dynamics and reach scale nitrate retention. Bio-
geochemistry 121 (1), 107–126.
plans, especially if restoration is being considered, and should be sampled Beaulieu, J.J., Golden, H.E., Knightes, C.D., Mayer, P.M., Kaushal, S.S., Pennino, M.J., Arango,
where feasible and safe. Given that a new stable condition was observed C.P., Balz, D.A., Elonen, C.M., Fritz, K.M., Hill, B.H., 2015. Urban stream burial increases
for most ecosystem parameters within the culverts, sampling at one or watershed-scale nitrate export. PLoS One 10 (7), 1–15.
Bernhardt, E.S., Palmer, M.A., 2011. River restoration : the fuzzy logic of repairing reaches to
two central locations could be sufficient to characterize the culvert environ-
reverse catchment scale degradation. Ecol. Appl. 21 (6), 1926–1931.
ment if appropriate for the study design and research questions. Note that Bernhardt, E.S., Palmer, M.A., Allan, J.D., Alexander, G., Barnas, K., Brooks, S., Carr, J.,
this may not hold true for shorter culverts where transition zones could Clayton, S., Dahm, C., Galat, D., Gloss, S., Goodwin, P., Hart, D., Hassett, B., Jenkinson,
R., Katz, S., Kondolf, G.M., Lake, P.S., Lave, R., Meyer, J.L., Donnell, T.K.O., Pagano, L.,
occur over most of a culvert's length. Transitions were apparent in most
Powell, B., Sudduth, E., 2005. Synthesizing U.S. river restoration efforts. Science 308
of our monitored variables, so managers should be aware of these potential (April), 636–637.
transition zones when choosing monitoring locations within the culvert. Blakely, T.J., Harding, J.S., Mcintosh, A.R., Winterbourn, M.J., 2006. Barriers to the recovery
of aquatic insect communities in urban streams. Freshw. Biol. 51 (9), 1634–1645.
Burford, D.D., Mcmahon, T.E., Cahoon, J.E., Blank, M., 2009. Assessment of trout passage
CRediT authorship contribution statement through culverts in a large Montana drainage during summer low flow. N. Am. J. Fish
Manag. 29 (3), 739–752.
Chakona, A., Phiri, C., Chinamaringa, T., Muller, N., 2009. Changes in biota along a dry-land
Chelsea L. Hintz: Conceptualization, Methodology, Software, Valida- river in northwestern Zimbabwe: declines and improvements in river health related to
tion, Formal analysis, Investigation, Resources, Writing – original draft, land use. Aquat. Ecol. 43 (4), 1095–1106.
Writing – review & editing, Visualization, Funding acquisition, Project ad- Cummins, K., Berg, M.B., Merritt, R., 2008. An Introduction to the Aquatic Insects of North
America. 4th ed. Kendall Hunt Publishing.
ministration. Michael T. Booth: Conceptualization, Methodology, Valida- Doane, T.A., Horwath, W., 2003. Spectrophotometric determination of nitrate with a single
tion, Resources, Writing – review & editing, Visualization. Tamara A. reagent. Anal. Lett. 36, 2713–2722.
Newcomer-Johnson: Conceptualization, Methodology, Validation, Re- Elmore, A.J., Kaushal, S.S., 2008. Disappearing headwaters: patterns of stream burial due to
urbanization. Front. Ecol. Environ. 6 (6), 308–312.
sources, Writing – review & editing. Ken M. Fritz: Conceptualization, Elzhov, T.V., Mullen, K.M., Spiess, A.N., Bolker, B., 2016. minpack.lm: R Interface to the
Methodology, Validation, Resources, Writing – review & editing. Ishi Levenberg-Marquardt Nonlinear Least-Squares Algorithm Found in MINPACK, Plus Sup-
Buffam: Conceptualization, Methodology, Validation, Formal analysis, Re- port for Bounds.
Favaro, C., Moore, J.W., Reynolds, J.D., Beakes, M.P., 2014. Potential loss and rehabilitation
sources, Writing – review & editing.
of stream longitudinal connectivity: fish populations in urban streams with culverts. Can.
J. Fish. Aquat. Sci. 71 (12), 1–12.
Feijó-Lima, R., Mcleay, S.M., Silva-Junior, E.F., Tromboni, F., Moulton, T.P., Zandonà, E.,
Thomas, S.A., 2018. Quantitatively describing the downstream effects of an abrupt land
Declaration of competing interest cover transition: buffering effects of a forest remnant on a stream impacted by cattle graz-
ing. Inland Waters 8 (3), 294–311.
Fuller, R.L., Kennedy, B.P., Nielsen, C., 2004. Macroinvertebrate responses to algal and bacte-
The authors declare that they have no known competing financial inter- rial manipulations in streams. Hydrobiologia 523, 113–126.
ests or personal relationships that could have appeared to influence the Fuller, M.R., Doyle, M.W., Strayer, D.L., 2015. Causes and consequences of habitat fragmenta-
work reported in this paper. tion in river networks. Ann. N. Y. Acad. Sci. 1355 (1), 31–51.
Goss, C.W., Goebel, P.C., Sullivan, S.M.P., 2014. Shifts in attributes along agriculture-forest
transitions of two streams in Central Ohio, USA. Agric. Ecosyst. Environ. 197, 106–117.
Acknowledgements Greenwell, B.M., Schubert Kabban, C.M., 2014. Investr: an R package for inverse estimation.
The R Journal 6 (1), 90–100.
Hansson, L.A., 1988. Chlorophyll a determination of periphyton on sediments: identification
We would like to thank the many students who helped conduct this of problems and recommendation of method. Freshw. Biol. 20 (3), 347–352.
work, especially J. Goetz, E. Sanderson, C. Tran, M. Urbanic, and N. Willse Hauer, F.R., Lamberti, G.A., 2007. Methods in Stream Ecology. Academic Press.
for their sustained help in the field and lab. We would like to thank R. Feijo- Hope, A.J., McDowell, W.H., Wollheim, W.M., 2014. Ecosystem metabolism and nutrient up-
take in an urban, piped headwater stream. Biogeochemistry 121 (1), 167–187.
Lima for guidance on the analysis. We would like to thank S. Jacobs from
Howley, C.S., 2004. The Relationships Among Culvert Characteristics and Culvert Sedimenta-
the US EPA for managing access to Congress Run and for his help in getting tion. University of Tennessee-Knoxville.
the project started. Finally, we would like to thank the US EPA-University of Julian, J.P., Wilgruber, N.A., de Beurs, K.M., Mayer, P.M., Jawarneh, R.N., 2015. Long-
Cincinnati Graduate Trainee Program, the University of Cincinnati Sigma term impacts of land cover changes on stream channel loss. Sci. Total Environ. 537,
399–410.
Xi Grant in Aid of Research, and the University of Cincinnati Wieman- Kaushal, S.S., McDowell, W.H., Wollheim, W.M., Johnson, T.A.N., Mayer, P.M., Belt, K.T.,
Benedict Award for funding. Although this work was reviewed by the Pennino, M.J., 2015. Urban evolution: the role of water. Water 7 (8), 4063–4087.
USEPA and approved for publication, it might not necessarily reflect official Kiffney, P.M., Richardson, J.S., Bull, J.P., 2004. Establishing light as a causal mechanism
structuring stream communities in response to experimental manipulation of riparian
Agency policy. Any use of trade, firm, or product names is for descriptive buffer width. J. N. Am. Benthol. Soc. 23 (3), 542–555.
purposes only and does not imply endorsement by the U.S. Government. Macpherson, L.M., Sullivan, M.G., Foote, A.L., Stevens, C.E., 2012. Effects of culverts on
stream fish assemblages in the Alberta foothills. N. Am. J. Fish Manag. 32 (3), 480–490.
Meyer, J.L., Poole, G.C., Jones, K.L., 2005. Buried alive: potential consequences of burying
Appendix A. Supplementary data headwater streams in drainage pipes. Proceedings of the 2005 Georgia Water Resources
Conference Held April 25–27, 2005, at the University of Georgia.
Supplementary data to this article can be found online at https://doi. Midwest Biological Institute (MBI), 2012. Biological and Water Quality Study of Mill Creek
and Tributaries 2011.
org/10.1016/j.scitotenv.2022.153050.
Murica, C., 1995. Edge effects in fragmented forests: implications for conservation. 1TREE 10
(2), 58–62.
Murphy, J., Riley, J.P., 1962. A modified single solution method for the determination of
References phosphate in natural waters. Anal. Chim. 27, 31–36.
Napieralski, J.A., Carvalhaes, T., 2016. Urban stream deserts: mapping a legacy of urbaniza-
Alberts, J.M., Fritz, K.M., Buffam, I., 2018. Response to basal resources by stream macroinver- tion in the United States. Appl. Geogr. 67, 129–139.
tebrates is shaped by watershed urbanization, riparian canopy cover, and season. Freshw. National Oceanic and Atmospheric Administration. NOAA's National Weather Service. U.S.
Sci. 37 (3), 640–652. Dept. of Commerce. http://www.nws.noaa.gov/.
Allan, J.D., 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems. Neale, M.W., Moffett, E.R., 2016. Re-engineering buried urban streams: daylighting results in
Annu. Rev. Ecol. Evol. Syst. 35 (2002), 257–284. rapid changes in stream invertebrate communities. Ecol. Eng. 87, 175–184.
Allan, D., Castillo, M., 2007. Stream Ecology: Structure and Function of Running Waters. Neeson, T.M., Moody, A.T., O’Hanley, J.R., Diebel, M., Doran, P.J., Ferris, M.C., Colling, T.,
APHA, 2005. Standard Methods for the Examination of Water and Waste Water. 21st ed. McIntyre, P.B., 2018. Aging infrastructure creates opportunities for cost-efficient restora-
American Public Health Association, Washington, DC. tion of aquatic ecosystem connectivity. Ecol. Appl. 0 (0), 1–9.
Arar, E.J., 1997. EPA Method 446.0: In Vitro Determination of Chlorophylls , B, C 1 + C 2 and Newcomer Johnson, T.A., Kaushal, S.S., Mayer, P.M., Smith, R.M., Sivirichi, G.M., 2016. Nu-
Pheopigments in Marine and Freshwater Algae by Visible Spectrophotometry. U.S. Envi- trient retention in restored streams and rivers: A global review and synthesis. Water 8
ronmental Protection Agency. (116), 1–28.

12
C.L. Hintz et al. Science of the Total Environment 819 (2022) 153050

Newcomer-Johnson, T.A., Kaushal, S.S., Mayer, P.M., Grese, M.M., 2014. Effects of Tanaka, M.O., Fernandes, J. de F., Suga, C.M., Hanai, F.Y., de Souza, A.L.T., 2015. Abrupt
stormwater management and stream restoration on watershed nitrogen retention. change of a stream ecosystem function along a sugarcane-forest transition: integrating ri-
Biogeochemisiry 121, 81–106. parian and in-stream characteristics. Agric. Ecosyst. Environ. 207, 171–177.
Nusch, E.A., 1980. Comparison of different methods for chlorophyll and phaeopigment deter- U.S. Census Bureau, 2016. Measuring America: our changing landscape. 2015 American Com-
mination. Arch. Hydrobiol. 14 (14), 36. munity Survey.
Olsen, D.S., Roper, B.B., Kershner, J.L., Henderson, R., Archer, E., 2005. Sources of variability U.S. Department of Agriculture, F. S, 2009. Culvert Scour Assessment.
in conducting pebble counts: their potential influence on the results of stream monitoring U.S. Geological Survery, 2016. The StreamStats Program.
programs. J. Am. Water Resour. Assoc. 41 (5), 1225–1236. United Nations, Department of Economic and Social Affairs, Population Division, 2019. World
Olson, J.C., Marcarelli, A.M., Timm, A.L., Eggert, S.L., Kolka, R.K., 2017. Evaluating the effects Urbanization Prospects 2018: Highlights (ST/ESA/SER.A/421).
of culvert designs on ecosystem processes in northern Wisconsin streams. River Res. Appl. United Nations, Science Task Force for the UN Decade on Ecosystem Restoration, 2021.
33 (5), 777–787. Science-based Ecosystem Restoration for the 2020s and Beyond.
Pennino, M.J., Kaushal, S.S., Beaulieu, J.J., Mayer, P.M., Arango, C.P., 2014. Effects of urban Wallace, J.B., Hutchens, J.J., Grubaugh, J.W., 2006. Transport and storage of FPOM. In:
stream burial on nitrogen uptake and ecosystem metabolism: implications for watershed Hauer, R.F., Lamberti, G.A. (Eds.), Methods in Stream Ecology, 2nd ed. Academic Press,
nitrogen and carbon fluxes. Biogeochemistry 121 (1), 247–269. New York, pp. 249–271.
Poplar-Jeffers, I.O., Petty, J.T., Anderson, J.T., Kite, S.J., Strager, M.P., Fortney, R.H., 2009. Walsh, C.J., Roy, A.H., Feminella, J.W., Cottingham, P.D., Groffman, P.M., Morgan Ii, R.P.,
Culvert replacement and stream habitat restoration: implications from brook trout man- 2005. The urban stream syndrome: current knowledge and the search for a cure. J. N.
agement in an Appalachian watershed. Restor. Ecol. 17 (3), 404–413. Am. Benthol. Soc. 24 (3), 706–723.
R Core Team, 2019. R: A Language and Environment for Statistical Computing. R Foundation Weatherburn, M.W., 1967. Phenol-hypochlorite reaction for determination of ammonia. Anal.
for Statistical Computing, Vienna, Austria. Chem. 39, 971–974.
Resh, V.H., Hauer, R.F., 2017. In: Lamberti, G.A., Hauer, F.R. (Eds.), Macroinvertebrates, 3rd Weitzell, R., Kaushal, S., Lynch, L., Guinn, S., Elmore, A., 2016. Extent of stream burial and
ed. Vol. 1. Elsevier, Boston, MA, pp. 297–319. relationships to watershed area, topography, and impervious surface area. Water 8
Resh, V.H., Jackson, J.K., 1993. Rapid assessment approaches to biomonitoring using benthic (11), 538.
macroinvertebrates. In: Resh, Vincent H., Rosenberg, D.M. (Eds.), Freshwater Biomoni- Wellman, J.C., Combs, D.L., Cook, S.B., 2000. Long-term impacts of bridge and culvert con-
toring and Benthic Macroinvertebrates. Chapman & Hall, New York, pp. 195–233. struction or replacement on fish communities and sediment characteristics of streams.
Roy, A.H., Rosemond, A.D., Paul, M.J., Leigh, D.S., Wa, J.B., Ac, L., Roy, A.H., 2003. Stream J. Freshw. Ecol. 15 (3), 317–328.
macro invertebrate response to catchment urbanization (Georgia, U.S.A.). Freshw. Biol. Wenger, S.J., Roy, A.H., Jackson, C.R., Bernhardt, E.S., Carter, T.L., Sudduth, E.B., Walsh, C.J.,
48, 329–346. 2009. Twenty-six key research questions in urban stream ecology: an assessment of the
Rutherford, J.C., Marsh, N.A., Davies, P.M., Bunn, S.E., 2004. Effects of patchy shade on state of the science. J. N. Am. Benthol. Soc. 28 (4), 1080–1098.
stream water temperature: how quickly do small streams heat and cool? Mar. Freshw. Wild, T.C., Bernet, J.F., Westling, E.L., Lerner, D.N., 2011. Deculverting: reviewing the evi-
Res. 55 (8), 737–748. dence on the “daylighting” and restoration of culverted rivers. Water Environ. J. 25
Shifflett, S.D., Newcomer-Johnson, T., Yess, T., Jacobs, S., 2019. Interdisciplinary collabora- (3), 412–421.
tion on green infrastructure for urban watershed management: an Ohio case study. Wood, P.J., Armitage, P.D., 1997. Biological effects of fine sediment in the lotic environment.
Water 11 (738), 1–19. Environ. Manag. 21 (2), 203–217.
Storey, R.G., Cowley, D.R., 1997. Recovery of three New Zealand rural streams as they pass
through native forest remnants. Hydrobiologia 353 (1–3), 63–76.

13

View publication stats

You might also like