DISCUSSION
Our low-cost field methods allowed us to calculate five flow metrics
corresponding with potentially ecologically relevant aspects of the flow
regime (Poff et al. 1997; Clausen and Biggs 2000). Demonstrating the
capability of using low-cost and robust techniques in flashy systems is
an important step towards monitoring and understanding the flow regime
of intermittent streams. The use of redundant methods of stream flow
measurement allowed us to collect variations in data inherent to each
method, such as when sensors sat in small, isolated pools of water at
substrate level, had residual water on the electrodes, or from high
humidity levels. Additionally, the initial dates of first noflow period
varied slightly between methods, likely for the same reasons. The sensor
data produced significant advantages: data processing time was faster
and less prone to error because no manual evaluation of stage height was
required. The sensors also provided stream temperature data, which can
be of high biotic importance, especially in intermittent streams
experiencing extreme temperatures during drying. The methods presented
in this study can be used to quantify variation in intermittent flow at
other ungauged sites.
Current modeling strategies may not accurately or fully depict the flow
regimes of small, highly variable watersheds. Predictions by the
Oklahoma HAWQS model were within the range of what we observed in the
field for some metrics but underestimated the number of no-flow days.
The number of no-flow days is a simple, but ecologically and
conceptually important metric describing stream drying (Poff et al.
1997; Olden and Poff 2003; Zipper et al. 2021). Prior calibrations of
hydrologic models have reported varying success in calculations of
hydrologic metrics, often overestimating low flow conditions (Kiesel et
al 2017; Pool et al. 2017), a direct contrast to our observations,
wherein the HAWQs models underestimated low flow conditions. Other
observations in the region report levels and durations of stream drying
consistent with our observations (Homan et al. 2005). Generally, our
streams exhibited higher numbers of no-flow days than predicted in a
localized framework developed by Leasure et al. (2016), implying that
environmental conditions may be harsher than anticipated in Ouachita
streams.
Watershed area, a measure of stream size, is often considered to be one
of the defining variables in flow patterns. Larger streams typically
have less extreme variations in flow, with lower relative extremes of
both high and low flows (Chiang et al. 2002). Slope is also frequently
regarded as an important driver of flow, given that slope plays a role
in determining the speed at which water moves through a system, with
steeper slopes more likely to shed water faster (Harr 1977; Paznekas and
Hayashi 2016). We observed some of our strongest correlations between
watershed area and slope with our flow metrics, however these
relationships were surprisingly variable. Because the gradient of stream
sizes included in our study was relatively narrow, other variables may
be exerting strong influences on streamflow patterns, such as watershed
and ecological processes (Snelder and Biggs 2002). The impact of the
relatively fine spatial scale of this study may have also influenced the
lack of significant links between land use or geological features and
flow patterns. Landscape factors typically used to predict hydrology may
not have varied enough within our study area to predict flow dynamics,
however variability and high flow conditions may be better explained at
the landscape level than drying trends. Our observed variability in
hydrologic metrics and the relatively weak observed relationships among
landscape factors and metrics highlight the challenge of predicting
ecologically meaning measures of stream drying in small streams.
However, we acknowledge the exploratory nature and small sample size in
this study limited our ability to statistically detect strong
relationships among landscape variables and hydrologic metrics. There is
a need for more field observations of stream drying to understand
drivers of variation in stream drying dynamics at intermediate spatial
scales relevant to ecological process and land management.
In addition to watershed characteristics, climate maintains a
well-established link with flow regime (Poff et al. 1997; Snelder and
Biggs 2002; D’Ambrosia et al. 2017). Within runoff driven systems, such
as the Glover River basin, local variation in precipitation likely
serves as a driver of variation in stream drying, a pattern seen in
other intermittent systems (Costigan et al. 2015). Southeastern Oklahoma
is characterized by high intensity and short duration precipitation
events, with two primary rainy seasons (Oklahoma Climatological Survey
2023). During the sampling period, we observed variation in local
precipitation between adjacent drainages. Variation in rainfall and
hydrograph responses can shift within a storm, based upon storm
direction, speed, movement patterns and watershed shape (Roberts and
Klingeman 1970; Jensen 1984; Singh 1997). Given the relatively small
area of our study extent, the scale of precipitation variability is
somewhat surprising and difficult to account for. Measuring
site-specific precipitation would be beneficial to further inform and
understand the drivers of small intermittent streams.
Climate change projections for the region predict an increase in short
duration, high intensity precipitation events, potentially less overall
annual precipitation (Karl et al. 2009; Marion et al. 2013; USGCRP
2018), and increased drought potential (Strzepek et al. 2010). These
projections will likely result in more frequent no-flow periods,
interspersed with large high-flow events. Beyond the obvious risk of
more streams completely drying in response to longer no-flow periods,
longer drying events also stress aquatic organisms where isolated pools
remain. In isolated pools, increased temperatures and decreased
dissolved oxygen levels can pass lethal thresholds for aquatic
organisms, leading to mortality for less tolerant species (Capone and
Kushlan 1991; Hopper et al. 2020). More intense precipitation events may
increase the intensity of disturbance, potentially impacting
reproductive success of resident organisms (Humphries et al. 1999) or
influencing community structure (Bernardo et al. 2003).
Long-term field observations of stream drying are needed to understand
drivers of variation in stream intermittency. This study documents the
ability to measure ecologically relevant hydrologic metrics, however
long-term data would serve to smooth the inherent year to year
variability in precipitation, both annually and between sites, and more
accurately represent between-site variability in flow patterns (Poff
1997). Without long term flow data, hydrologic dynamics can be difficult
to predict or model, limiting the applicability of environmental flow
standards (Zhang et al. 2016). Our comparison of HAWQS model predictions
to observed data support previous observations that current long-term
stream gauges in the US can provide potentially misleading or flawed
information about no-flow periods, highlighting the need for increased
data collection of streamflow data in smaller, intermittent streams
(Zimmer et al. 2020).
Both stream intermittency sensors and trail cameras can serve to provide
valuable information about stream flow regimes. We successfully
evaluated the five major aspects of the flow regime for a gradient of
intermittent streams. These techniques readily allow for classification
of flow regimes into three ecologically relevant categories, but given
the small sample size, both temporally and spatially, identification of
watershed level impacts on flow regime remains less clear. Within the
Glover River basin, our ten streams exhibited flashy intermittent flow
patterns, with variable signatures among systems. Although we found some
evidence of watershed characteristics correlating with flow patterns,
many of the other drivers of local flow regime remained elusive,
potentially due to highly variable precipitation patterns. Evaluation of
the selected hydrologic indices serves to underscore the highly variable
nature of intermittent streams in Southeastern Oklahoma, and the
difficulty in linking watershed scale factors with local flow regime.