2021 - Paradise Pond Sediment Redistribution Project Report.pdf
Facilities Management
Northampton, Massachusetts 01063
October 18, 2022
Department of Environmental Protection
Water Quality Certification Program
One Winter Street, 5th Floor, Boston, MA 02108
Attn: Stephanie Moura, Director, Division of Wetlands and Waterways
Photo of Paradise Pond sediment redistribution project site during dozer work ‐ Tuesday, Dec. 21, 2021. (view to Northeast)
Mill River channel on left has been filled with sediment. River is flowing through canal to the right, not in photo.
Re: Paradise Pond Sediment Redistribution Project – 2021 Report
Pertaining to: Combined Permit – Chapter 91 Permit and 401 Water Quality Certification
At: Smith College
Paradise Pond Sediment Management Project
Northampton, MA 01063
Referencing: 401 WQC Transmittal No. X281564, Wetlands File No. 246‐0725, Chapter 91 Permit No. 15001
EOEEA Certificate No. 15282, ACoE Application No. NAE‐2012‐02550
Dear Ms. Moura,
The attached reports describe in detail the work of the Paradise Pond Sediment Redistribution Project and the potential
impacts of that work to the Mill River downstream of Paradise Pond in Northampton Massachusetts. This is the 7th report
submitted to State regulators as required by the permits referenced above. The 2021 Report contains 3 reports, from 3 authors.
“Paradise Pond Sediment Management Report 2021”, Robert M. Newton, Geoscience Solutions LLC – pdf page
“Freshwater Mussel Monitoring in the Mill River” 2016 ‐ 2021, Ethan Nadeau, Biodrawversity – pdf page
“Impact of Sediment Redistribution on Macroinvertebrates in the Mill River” 2018 ‐ 2021, Marney Pratt, Smith Bioscience – pdf
page (this report was included verbatim in the 2020 report)
The 2021 sediment redistribution activity was started on Dec. 6, 2021 (partial drawdown) and completed on Dec. 22, 2021. A
small excavator was employed to excavate a shallow canal along the right bank of the Pond to redirect flow into areas where
the sediment is soft and did not support a bull dozer in previous years. Once a partial canal was cut, the bull dozer entered the
Pond and created a large stockpile of sediment near the upstream river channel entrance into the Pond. The sediment
stockpile was then used to plug the river flow through the natural channel, redirecting the water to the partial canal. The flow
of redirected river completed cutting of the downstream soft sediments within 20 minutes. Please see the drawdown log at the
end of this report.
Observations:
o The diversion canal worked well to remove accumulated sediments in a portion of the Pond that was not possible to
excavate with heavy equipment.
o The diversion canal also lowered the river stage on the East side of the island which we hoped would help to dry soft
sediments there and permit use of heavy equipment.
o Redistribution of the soft sediments using a dozer continued to be difficult.
o One precipitation event (12/18 into 12/19) filled the Pond, but did not cut the temporary sediment dike constructed
to divert the river.
o Sluiced sediment falls out quickly under low flow conditions, collecting in the plunge pool and further downstream.
o High flow events remove accumulated downstream sediment consistently and quickly.
Conclusions:
A dozer is the most appropriate piece of equipment to redistribute course sediment within the Pond. Less so with
finer/ softer organic sediment.
Sediment released through the sluice gate fell out and accumulated in the plunge pool, and in the river beyond
Lamont Bridge.
The sediment remained in place until higher flows mobilized it and transported it further downstream.
Respectfully submitted,
Gary Hartwell, Project Manager
Smith College, Facilities Management
126 West St., Northampton, MA 01063
Ecc: David Wong and Susan You, MassDEP Div. Wetlands & Waterways, Boston; David Cameron and Mark Stinson,
MassDEP WERO; Paul Sneeringer and Barbara Newman, USACoE; Misty‐Anne Marold, and Thomas W. French, NHESP,
Mass. Div. F&W; Sarah LaValley and Kevin Lake, Northampton Conservation Commission; David Veleta, PE, and
Kristine Baker, PE, Northampton DPW; Robert M. Newton, Smith College Geosciences; Marney Pratt, Smith College
Biological Sciences; Ethan Nadeau, Biodrawversity; Peter Gagnon, Scott Richards, Smith College Facilities Management
Sarah Pierce, PARE Corp
2021 – Paradise Pond Sediment Redistribution Project Photos
12/10/2021 – Diversion canal, right bank of Pond 12/13/2021 – Plugging river with Pond sediment
12/13/2021 – Canal filling with River water 12/13/2012 – Canal cutting into downstream Pond channel
12/13/2021 – Full river flow through canal 12/13/2021 – Canal at 4:22 PM
12/22/2021 – Complete sediment redistribution, Pond filling
Paradise Pond Sediment Management Report
2021
Robert M Newton
Geoscience Solutions LLC
March 6, 2022
Figure 1. Drone image of sediment redistribution operations on December 14, 2021. Note the diversion channel
(right side of photo).
2
Sediment redistribution in 2021 included the construction of a diversion channel located along
the south shore of the pond running eastward from the pond inlet to the lower section of the
pond just upstream from the dam (figure 1). This channel was constructed at the beginning of
sediment redistribution operations and was used to divert the channel of the Mill River away
from the north side of the pond so that sediment could be placed in the old channel to reduce
bank erosion in that area. Flow through the diversion channel stimulated erosion and sediment
transport as the gradient of this channel was significantly steeper than the natural channel.
This resulted in the release of approximately 3,500m3 of sediment from the sediment
redistribution area.
2021 Hydrologic Summary
2021 was a relatively wet year with a total of 47.74in of precipitation as measured by the dam
station tipping bucket rain gage. If anything, this is an underrepresentation of the total
precipitation as the tipping bucket is not heated. Storms during this period caused 7 flow
events with peak discharge over 1000cfs (figure 2).
Figure 2. Hydrograph for discharge at the Lamont Bridge station for 2021. There were 7 flow events with
discharge greater than 1000cfs.
The highest flow (2,750cfs) occurred on July 18, continuing the trend of maximum flows
occurring in the summer and fall rather than in the spring.
December Redistribution Operations
Pond level was lowered beginning at 7am on December 6, 2021 and was raised again on
December 22 (figure 3). For most of the period, water level was 5-7ft below the dam crest,
however, a rain event (0.72in) on December 18/19 briefly caused the pond to refill as the inflow
3
briefly exceeded the flow capacity of the sluice gate (figure 3). Also, on December 13 the pond
was briefly fully drained (-11.6ft) during the opening of the diversion channel.
Figure 3. Height of water in the pond relative to the elevation of the dam crest during the December
drawdown period.
There were 5 precipitation driven hydrologic events during the drawdown period (figure 4).
Inflow to the pond varied from a low of 75cfs to a high of 275cfs, but only the highest flow
Figure 4. Hydrograph showing the variations in inflow to the pond as measured upstream at the
Clement St Bridge USGS Mill River Northampton gage station.
4
caused the pond to briefly refill. Maintenance of a fairly steady drawdown level during periods
of widely varying inflow was due to excellent management of the sluice gate.
Immediately following the lowering of the pond, a drone mission was flown (12/7)to establish
the baseline conditions prior to any sediment redistribution. A total of 10 control points were
established using an EOS Arrow Gold RTK GPS system. These control points are critical for
processing and calibrating the Digital Scene Model (DSM) that is created from the drone-based
air photos. This is done using Pix4D Mapper software that is based on Structure-from-Motion
photogrammetry techniques. The resulting DSM (figure 5) has a vertical resolution of better
than 2cm.
Figure 5. Digital Scene Model (DSM) created from drone-based air photos taken on December 7, 2021. The
red dots show the location of high precision control points used in the analysis.
5
Diversion Channel Construction
After several days of drying, the pond bottom was firm enough to support the weight of an
excavator and bulldozer. The excavator was used to construct a diversion channel to carry the
inflow from the Mill River along the right bank of the pond. Without the diversion channel,
flow is along the left bank. There were several reasons for diverting the flow.
• Because the diversion channel formed a more direct path to the dam, it had a steeper
gradient and was therefore capable of transporting more sediment and the sediment
moved could be coarser grained.
• The shallow constructed channel would undergo considerable downward and lateral
erosion that would remove sediment from the pond without the need for excavation
and transport by heavy equipment.
• Sediment from the pond bottom could be used to fill and stabilize the abandoned left
bank channel. Lateral erosion of the left bank channel had destabilized the pond bank
and pondside trail. Filling that channel should stabilize the bank.
The diversion channel was completed and inflow to the pond bottom was diverted on
December 13 at 12:55pm. Inflow was approximately 100cfs at this time.
Figure 6. Bulldozer cutting off flow to the left bank channel at 12:55pm on December 13. This caused the inflow
water to move through the diversion channel that lies to the right in this photo.
Downstream turbidity, as measured at the Lamont Bridge station, increased to approximately
190NTU with the opening of the sluice gate on December 6 (figure 8). This was due to the
6
release of sediment that had accumulated just upstream of the sluice gate since the last time it
had been opened. Turbidity quickly dropped to less than 10 NTU by early on December 7. A
rain event beginning on December 11 (figure 4) caused turbidity levels to briefly rise as high as
90 NTU.
Figure 8. Outflow turbidity as measured at the Lamont Bridge station during the December sediment
redistribution operations.
Figure 7. Aerial view of diversion channel taken on December 22, just before the
pond was refilled.
7
The highest turbidity (295NTU) measured during the sediment redistribution occurred on
December 13 (figure 8) and was associated with a brief complete draining of the pond when the
diversion channel was opened. The brief draining occurred because there was no downstream
flow to the residual pond during the time it took to fill the empty diversion channel. At the
time of diversion, the abandoned channel also stopped providing water to the residual pond
and thus the water level dropped to the level of the sluice gate. As soon as the diversion
channel filled and provided flow to the downstream system, the residual pond refilled. The
initial spike to 290 NTU in the turbidity graph was due to erosion occurring in the pond bottom
in the area immediately upstream from the sluice gate. This ceased as soon as the residual
impoundment was re-established. Continued high values in turbidity after this time (figure 8)
were due to erosion in the diversion channel which was quickly deepened as a knickpoint
migrated upstream from the lower end of the diversion channel. This knickpoint took several
hours to migrate all the way to the pond inlet. By December 15, turbidity values had stabilized
at values just under 15 NTU.
High turbidity values after December 19 were due to precipitation events. It is interesting to
note that there was not a lot of turbidity associated with the highest flow event. This was
because pond bottom erosion was minimized when the pond filled. It wasn’t until after the
pond bottom re-emerged that erosion in the diversion channel increased due to concentrated
inflow being confined within the channel. This coupled with sediment redistribution operations
explains the late variations in turbidity.
Migration of the knickpoint
December 13, 1:35pm
December 13 2:10pm
Figure 9. Two views of the lower end of the diversion channel showing the migration of the knickpoint over a 35
minute period.
Erosion in the diversion channel began as soon as water started flowing through its full length
and a knickpoint rapidly developed at the point where the diversion channel entered the lower
end of the pond. A knickpoint is a steep section of channel where erosion is concentrated.
Concentrated erosion on this steep channel slope causes the knickpoint to migrate upstream.
Migration of the knickpoint was monitored using the webcam on the roof of Sabin Reed (figure
8
10). It moved the length of the diversion channel in about 24 hours. As it moved, the gradient
of the knickpoint decreased.
December 13, 1:18pm December 13, 1:38pm
December 13 2:13pm December 13, 2:38pm
Figure 10. Migration of the knickpoint as indicated by the webcam on the roof of Sabin Reed.
After the migration of the knickpoint, the longitudinal profile of the diversion channel reached a
dynamic equilibrium with the flow conditions and there was no more downward erosion. From
that point onward, changes in the channel were mainly due to lateral erosion. This resulted in
some widening of the channel together with some lateral channel migration. The diversion
channel was only active for about 10 days, but even so by the end, there was some evidence of
initial meander development.
Figure 11. Bank erosion and initial meander formation in the diversion channel
on December 22.
9
Sediment Redistribution
Sediment redistribution operations were started immediately following the opening of the
diversion channel. A significant part of this operation involved filling portions of the now
abandoned channel that ran along the left bank of the pond (figure 12). In all previous
operations this channel carried the inflow across the bottom of the pond following the historic
path of the pre-dam Mill River. During high flow events, when the pond is full, this drowned
channel experiences relatively high velocities and bank erosion has become a concern. A slope
failure that occurred along the path on this side of the pond when the pond was drained
emphasized this concern. Therefore, part of the sediment redistribution project involved filling
parts of the old channel to help stabilize the slope.
Air photo of sediment moved into the old channel at
the pond inlet, December 14.
Air photo showing sediment redistribution operations
filling the old channel on the left bank of the pond,
December 14.
Completed fill of left bank channel on December 22. Sediment redistribution into old channel across from
the boathouse, December 14. Note diversion channel
in upper right and bulldozer "grave" in lower left.
Figure 12. Sediment redistribution operations.
It is likely that future high flow events will re-excavate this channel. This would not be a bad
result, as a major goal of this project is to release as much sediment as possible during high
flow events and the scouring of a new channel does release more sediment. We can then
10
anticipate that in the next sediment redistribution more sediment can be moved into the new
channel and the cycle will repeat itself.
Sediment Redistribution Results
Drone flights were flown on December 7, December 14, and December 22. The December 7
and December 22 flights were designed to collect air photo data to create DSMs while the
Figure 13. Digital Scene Model derived from drone photography collected on December 22.
11
December 14 flight just collected observational photos and video. Unfortunately, the initial
December 22 mission ended with the drone in the top of a large Sycamore tree, but fortunately
a second drone was deployed and successfully completed the photo mission. There were also
problems surveying the 10 control points with RTK GPS due to problems accessing the campus
wireless network. In the end, there were only 4 control points with good enough data to create
the DSM. There were not enough points for a thorough error analysis but it appears that the
error was comparable to the December 7 DSM.
The DSMs were converted to Digital Elevation Models (DEMs). The DSM data from both the
December 7 and December 22 flights were first clipped to the area of the pond to eliminate
most vegetation effects. It was then necessary to further filter the data to eliminate
measurements associated with vegetation that overhangs the edge of the pond. This was
particularly a problem with some of the larger trees at the edge of the pond (e.g. the large
sycamore that was removed during the project). The normal pool elevation (41.5m) was used to
screen the elevation data such that higher elevations were nulled. The result was a dataset
that only represented the land surface (DEM)
A series of topographic profiles were established to determine how sediment redistribution
changed the configuration of the pond bottom. These were constructed using elevations
extracted from the two DEMs along the lines of profile shown in figure 14. Comparison of the
pre- and post- redistribution lines (figure 15) show where sediment was removed and where it
was accumulated. It should be noted, that elevations determined by photogrammetry are
unreliable in water covered channel areas. Although the bottom is often visible, light refraction
causes a non-parallactic displacement that results in elevation errors. In other water covered
areas the photogrammetric analysis measured the elevation of the water surface which was
different at the times of the two drone missions.
Line 1 (figure 15) clearly shows how just under 0.5m of sediment was removed from the pond
bottom and used to fill the old left bank channel. The left side of the graph also shows how the
sediment was banked up above normal pond level to protect that area from future bank
erosion. There has been approximately 1m of erosion by the diversion channel on the right side
of the graph. This represents material that did not have to be moved by the bulldozer.
Line 2 (figure 15) shows over 0.5m lowering of the bottom and a significant filling of the old
channel. Integrating the areas under the two profiles shows a slight gain (3.5m3/m) in sediment
volume as more sediment was added to the old channel than was removed from the adjacent
pond bottom. This likely came from redistributed sediment moved in from the southwest as
shown by bulldozer tracks on the orthophoto.
Line 3 (figure 15) shows a net loss of sediment along the line of profile. Measurements of the
infilling of the channel on the right side of the figure are not reliable as this portion of the old
channel held water throughout the redistribution operations.
Line 4 (figure 15) marks the maximum extent of redistribution operations on the northeast side
of the island. Most of the upstream part of the point bar was removed but not much was done
12
on the downstream end as the soft sediment proved difficult for the bulldozer and it got stuck
crossing the lower part of the bar (figure 12).
Figure 14. Red lines show where topographic profiles have been constructed comparing pre and post sediment
redistribution bathymetry. Base map established using the December 7 data. Colors reflect bottom
elevations and the grey areas are based on topographic slope.
Line 5 (figure 15) is a relatively undisturbed profile where, aside from a couple of bulldozer
tracks, there was no sediment redistribution. It serves as a reference to check on the
reproducibility of the DEM results. The “before” and “after” redistribution lines are almost
coincident except for the two clear sets of bulldozer tracks that emphasize how soft the
sediment was in this area. The fact that the pre-distribution line consistently lies above the
post line, shows that a significant amount of drying related compaction occurred during the
time the pond bottom was exposed
13
Figure 15. Topographic profiles constructed along the lines shown in figure 14. Note that all profiles have the
same vertical scale but different horizontal scales. Elevations in water filled areas are unreliable due
in part to refraction displacement.
Line 6 (figure 15) runs from the island to the athletic field side of the pond. Only about 30m of
the pond bottom was redistributed to a depth of less than 0.5m at this location. The athletic
field side of this line was untouched except for the incision of the diversion channel. There was
likely at least 1m of erosion in the channel that grew to a width of approximately 10m. The
undisturbed section of pond bottom experience varying amounts of compaction due to drying,
reaching a maximum of about 15cm.
Line 7 (figure 15) was not in the sediment redistribution area and had a lot of water. It clearly
shows the close correlation between the “before” and “after” profiles. There was little
evidence for compaction as the sediment here was sandy with little organic material.
14
Line 8 (figure 15) was the longest profile line and shows up to 0.5m of sediment removal from
the pond bottom and the filling of the old channel to a depth of about 1m.
Pond Depth
The water depth in the pond was calculated by subtracting the DEM from the mean water level
(41.5m). The resulting raster yields water depths for every square meter of the pond. These
were calculated for both “before” and “after” sediment redistribution. The rasters were then
queried to determine how much of the pond was greater than 2ft deep (figure 16).
Figure 16. Maps showing the portion (red) of the pond with water depths less than 2ft.
Since there will always be an area adjacent to the shoreline where water depth is less than 2ft,
only the areas in the central part of the pond were considered. This included about half of the
point bar on the east side of the island. Using this criteria, the area less than 2ft deep on
December 7 was determined to cover 14,043m2 or about 41% of the pond. After sediment
redistribution on December 22, this value had dropped to 3,153m2 or just under 10% of the
pond area. It should be noted that these percentages reflect the area of study and do not
include the extension of the pond just above the dam.
A bathymetric map was created from the December 22 DEM showing water depth contours
with a 1ft contour interval (figure 17). The deeper areas are not contoured as they were
covered in water during the surveys.
15
Figure 17. Bathymetric contour map of the pond bottom. The contour interval is 1ft.
Volume of Sediment Released
The 1m DEMs were used to calculate the total volume of sediment moved and the total volume
released from the pond. The DEM data was filtered to null all values greater than 41.5m. This
removed the overhanging vegetation from the calculations. Next areas covered in water were
removed. Since water levels were higher on December 22, the orthophoto from this date was
used to create a screening polygon that was clipped from both DEMs. The December 22 DEM
was then subtracted from the December 7 DEM. The resulting raster is made up of both
positive and negative value cells. The positive numbers mark places where sediment was
removed and the negative numbers show where it had accumulated (figure 18). Since there
was, on average, 10cm of subsidence due to compaction/drying, only areas experiencing
greater than 10cm of removal were considered as areas where sediment was removed. For
accumulation, only areas showing greater than 5cm of elevation increase were considered. The
maps (figure 18) clearly show how sediment was removed from the central pond area and was
deposited mainly in the abandoned channel on the left bank of the pond. Time ran out on the
project before all the areas south of the pond could be redistributed.
16
Figure 18. Maps showing where sediment was accumulated (green) and where it was removed (red) as a result
of sediment redistribution operations.
The total volume of sediment redistributed can be obtained by multiplying the cell values for
change in elevation in meters by the cell area in meters and then summing the results for each
area.
Accumulation Area Sediment Volume Equation
𝑉𝑜𝑙𝑢𝑚𝑒=(|𝐴𝐶!,#× 𝑤× ℎ
$
%
(Equation 1)
Where:
𝑛 = number of raster cells in the accumulation area.
𝐴𝐶!,#= individual cell within the accumulation matrix.
𝑤= width of the raster cell (1m)
ℎ= height of the raster cell (1m)
Removal Area Sediment Volume Equation
𝑉𝑜𝑙𝑢𝑚𝑒=(𝑅𝐸!,#× 𝑤× ℎ
$
%
(Equation 2)
Where:
𝑛 = number of raster cells in the removal area.
𝑅𝐸!,#= individual cell within the removal matrix.
𝑤= width of the raster cell (1m)
ℎ= height of the raster cell (1m)
17
The results of equation 2 yield the total volume of sediment that was redistributed during the
project (5,012m3) while equation 1 shows the volume that was moved to other parts of the
pond (Accumulation areas). The difference between these two numbers is the amount of
sediment that left the study area (3,388m3). These results are summarized in Table 1. It is also
possible to estimate the volume of sediment that exited the study area by simply integrating
the difference in surface elevations between the two surveys (Pixel Total). This yields a higher
value as it includes compaction/drying.
Table 1. Results of sediment redistribution of December 2021.
Volume (m3) Area (m2)
Sediment Redistributed 5,012 14,845
Sediment Accumulated 1,624 3,722
Total Removed 3,388
Total Removed (pixel) 3,725
It is likely that the total sediment removed during the project was around 3,500m3, if you
include erosion of the diversion channel. It was difficult to document the final geometry of the
diversion channel as it was filled with actively flowing water at the end of the project and
depths could not be measured by drone photogrammetry. While most of the sediment leaving
the study area likely left the pond, some may have accumulated in the area just upstream of
the dam where the pond was held at the -6ft level. To be clear, there is very little standing
water when it is at this level, but it does allow for storage of some sediment as the stream is
graded to -6ft rather than to -12ft when the gate is fully open.
Suspended Sediment
The total volume of sediment released represents both suspended and bedload materials. It is
possible to estimate how much of the total sediment released was suspended sediment by
using discharge coupled with turbidity data.
The turbidity sensor at the Lamont Bridge
collected side scan turbidity data every 15
minutes during the drawdown (figure 8). The
turbidity data (NTU) can be converted to
suspended sediment load (mg/L) using a
relationship developed by comparing these
values for samples collected during a previous
hydrologic event (figure 19). The equation for
the regression line is used to convert the NTU
values to mg/L. These values are then
multiplied by the corresponding discharge
(cfs). Since the discharge values are collected
at 15minute intervals and the discharge is
collected at 10minute intervals, it was
necessary to extrapolate the 15minute and
45minute values.
!" #"$ %#&
Figure 19. Relationship between turbidity and
suspended sediment load.
18
Units of the resulting calculations are then converted to kg/15min and all the 15minute
readings are then summed and expressed in metric tons for the December 6 - 22 period. The
total mass exported as suspended sediment over the drawdown was 164 metric tons. If you
assume a sediment density equivalent to the mineral feldspar, the volume can be estimated to
be approximately 100m3. Even though it is likely that there is some organic matter in the
sediment making the density lower and the volume somewhat larger, it is still not a significant
fraction of the 3,500 m3 total sediment released.
Downstream Monitoring
Sediment was observed collecting in the splash pool and the immediate downstream reach
(between the dam and the Lamont Bridge) during the pond drawdown. This sediment was
released through the sluice gate, especially during the high flow periods (figure 4). On
December 29, just after the sediment redistribution was completed, the bottom of the channel
upstream of Lamont Bridge was covered with a thin layer of sediment (up to 10cm) (figure 20).
Downstream of the bridge, the sediment layer was just patchy (figure 20).
The web cam located on the Lamont Bridge broadcasts a live picture of the dam and hourly,
during daylight hours, repositions to take a picture looking directly down on the channel
bottom (https://www.youtube.com/watch?v=3r52UXK-1QY). These pictures record changes in
the channel bottom over time and can be used to see how fast bedload sand is moved through
this section of the channel (Figure 20). Observations show that this sediment is mainly moved
during high discharge events.
There have been only 3 significant high flow events since the sediment redistribution
operations in December. The largest occurred on February 4 with a maximum discharge of just
over 1000cfs. Then there was a flow of 580cfs on February 18 that was followed by a flow of
760cfs on February 23 (figure 21). These flows have removed much of the sediment although
some still remains as of March 4.
19
View upstream, from the Lamont Bridge, of sediment
covering the bottom of channel on December 29.
View downstream, from the Lamont Bridge, of patchy
sediment covering the bottom of channel on December
29.
View of reference channel from Lamont Bridge camera on
December 30. Compare with photo on right.
View of reference channel from Lamont Bridge camera on
March 6. Compare with photo on left.
Figure 20. Images of sediment released as a result of the December sediment redistribution operations.
Figure 21. Hydrograph from the USGS Mill River at Northampton gage station showing discharge events
since the December sediment redistribution. Note gaps in record due to ice cover.
20
Reference Section 1
Reference Station 1 is located near the upper end of the diversion channel just downstream of
the Manhan rail trail bridge (figure 22).
Figure 22. LiDAR based DEM showing the location of Reference Station 1 located in the diversion channel of the
Manhan River. The two red markers show the location of the benchmarks that define the line of cross
section.
The last survey of this cross section was done in August of 2021, although visual inspections
have been made since the December 2021 operations. The rough nature of the rip rap that
lines the channel at this location, makes detailed surveying difficult as just a slight difference in
the position of the survey rod can result in a significant elevation difference. However, the
presence of bedload sediment is easily seen visually in the bottom of the channel and tends to
form stringers running downstream from some of the larger pieces of rip rap (figure 23). Survey
methods used to directly measure the profile at this location changed in 2019 when RTK GPS
was instead of the Total Station. The move to GPS has been driven by vandalization of the
profile marking benchmarks.
21
RTK GPS measurements at reference station 1, August
2021. Note streak of sand behind mid channel boulder
Photo on December 29, 2021 showing bedload sand
accumulating behind channel boulders at reference
station 1.
Figure 23. Photos of sediment moving through the area of reference station 1.
Although some sand sized sediment has been observed at reference station 1, it appears to
move through the section at a fairly rapid rate. Its occurrence is limited to small patches behind
the rip rap boulders that make up the bottom of the channel. Topographic cross sections at
this location (figure 24) do not show any changes through time.
Figure 24. A comparison between RTK GPS surveys done in 2019 and 2021 show no changes through time.
22
Reference Section 2
Reference site 2 is of particular interest as it lies within the Mill River diversion channel just
upstream of the Route 10 bridge and the run-of-river dam. This makes it the most likely site
where sediment released from Paradise Pond, would accumulate. Previous surveys have noted
a small sand bar on the left bank of the channel that is likely a point bar type deposit associated
with a turn to the left in the constructed channel (Figure 25).
Figure 25. Map of channel bathymetry determined using the vertical beam of a Teledyne RiverRay ADCP
in 2017. The red dots mark the location of the benchmarks that locate the ends of the
reference cross section (Ref 2).
In 2017 a detailed survey (Figure 25) was done on this part of the channel. A Leica TPS 1200
Total Station was used to collect profile data between benchmarks 10 and 11 and was used to
determine the elevation of the water in the channel. Then the vertical beam from a Teledyne
RD Instruments RiverRay Acoustic Doppler Current Profiler (ADCP) was used to survey the
bathymetry of the diversion channel from just upstream of the reference 2 profile line down to
the route 10 bridge.
Annual surveys using the Total Station at this location have proven difficult due to problems in
maintaining the benchmarks. Benchmark 11 has been lost. The benchmark disks appear to be
the targets of vandalism and at least 3 of the downstream disks have been removed. The
23
marker for benchmark 10 had to be reset in 2019 and is now located about 1m from its original
location.
Previous surveys at the Ref 2 cross section have not detected any significant changes so more
of the survey effort has shifted to monitoring the downstream bar. An RTK GPS survey was
done in July of 2019 using an EOS Arrow gold GNSS system paired with the HAMP CORS
reference station located on the roof of Sabin Reed. Data was collected both along the line
between benchmark 10 and benchmark 11, as well as in the area of the small sand bar. This
survey was redone in August 2021 using the same methodology. The advantage of this system
is that it does not require benchmarks, the disadvantage is that there is not always good
satellite geometry. The results of the 2021 survey can be compared to the river bathymetry
from the 2017 by extracting river bottom elevations from the bathymetric map at the exact
locations of the GPS survey (Figure 26 and Figure 27).
Figure 26. 2021 GPS survey points in the area of the left bank sand bar. Line 1 and Line 2, mark profile
lines for comparison of the 2021 data with the 2019 data.
1
2
24
Figure 27. Comparison of bathymetric profiles at the sand bar. The elevation difference between the 2017 and
2021 data sets is likely due to a vertical offset in the 2017 data.
The 2017 data had consistently higher bottom elevations as measured by the RiverRay
compared to the 2021 elevations measured using RTK GPS. The 2017 data was based on the
elevations of the two benchmarks which were established using differential GPS methods. The
differential GPS measurements were not very accurate but at the time, this was fine as the
surveys were designed to look simply at the changes in elevation of the channel bottom. The
problem comes in when the benchmarks are removed or altered so that we cannot go back and
measure the differences from the new RTK GPS measurements. To minimize the differences,
the 2017 data was offset such that the first points in each profile were at the same elevation
(Figure 27). While this did bring the profiles to closer agreement, the 2017 profiles were
consistently higher than the 2021 data. This could be interpreted as the result of net erosion
between 2017 and 2021, however, the uncertainties in this analysis make this an unconvincing
argument.
An RTK GPS survey was also done in 2019 and this data can be directly compared with the 2021
data (Figure 28). The problem here is that the points from the two surveys do not directly
correspond so it is difficult to make a definite conclusion other than that there has not been
much change, if any at all. To correct this deficiency in the future, a series of survey lines will be
established in that area of the bar. These lines will be entirely RTK GPS based and will not
require the placing of benchmarks that are likely to be vandalized.
25
Figure 28. Bottom elevations collected in the area of the sand bar using RTK GPS methods in 2019 and
2021.
Conclusion
Operations in December of 2021 were very effective in removing sediment from the pond. The
diversion channel proved to be particularly successful. Its steeper slope increased its capacity
to move sediment, both in terms of total sediment moved and in terms of the largest grain size
that could be transported (competence). Over time, coarse pebble sand had accumulated in
the upper parts of the pond where the Mill River enters. Prior sediment redistribution
operations had not been able to move this material out of the pond. This time, however, coarse
pebble sand sediment that clearly came from the upstream areas of the pond, was observed in
the channel near the Lamont Bridge (figure 20). Although the diversion channel was only active
for 10 days, during that time it underwent significant lateral and downward erosion. All this
sediment is material that didn’t need to be mechanically redistributed. Future projects should
take advantage of this technique and it could be improved by making the channel curved. A
channel, curved toward the island, would promote lateral migration in that direction. This
could remove even more sediment from a muddier part of the pond where the use of heavy
equipment is difficult.
26
Much of the sediment released from the pond during the redistribution operation is initially
trapped in the splash pool and the section of channel above the Lamont Bridge. This sediment
is then released downstream during higher flow events. It appears that this sediment moves
quickly through the downstream Mill River diversion channel (above Route 10). Although there
is some evidence for point bar formation on the left bank just above the Route 10 drop
structure, this appears to be a small stable feature that has not been influenced by sediment
released from the pond. The channel below the drop structure is subject to flooding from the
Connecticut River and is quite dynamic with debris dams and sand bars that come and go
through time. In this part of the system, sediment is mainly accumulating in the area of
Hulberts Pond, which lies within an old oxbow of the Connecticut River.
In terms of project methods, Digital Elevation Models (DEMs) built using Structure-from-Motion
photogrammetry from drone based aerial photographs have proven to be superior for
quantitatively analyzing sediment redistribution operations. Integrating known control points,
surveyed using RTK GPS methods, yields DEMs with elevations accurate to almost 1cm. This is
not only better than other methods used in the past, but is also much faster.
In the beginning there was some concern that it might be difficult to maintain a lowered pond
level during the fall when rain events are common. This is no longer a concern as the pond was
held stable during this project, with one exception, despite four rain events. Kudos to Gary
Hartwell for skillfully managing the sluice gate during the drawdown period. The one rain event
that filled the pond was due to the flow of the Mill River rising above the capacity of the sluice
gate. Even this did not cause a major disruption to the project. Higher flows are actually an
advantage (as long as they don’t exceed the sluice capacity) as they allow for more sediment to
be transported out of the pond.
Finally, there is some concern regarding the stability of some of the steep banks on the sides of
both the upstream and downstream river channel, as well as along the sides of the pond itself.
The filling of the old channel along the left bank of the pond should help address that issue, but
some of the other upstream and downstream channel banks are likely to fail in the immediate
future. I would recommend that, where possible, the pond monitoring system be adapted to
collect data at places where there is evidence for imminent failure (figure 29)
Figure 29. Soil crack showing potential slope failure on the right bank just south of the Lamont Bridge.
27
Acknowledgements
This project was supported by a team of dedicated individuals who were willing to
volunteer their services in a wide range of field conditions. Thanks go to the expert drone
team of Jon Caris and Tracy Tien who were willing to fly missions with short notice in
sometimes marginal conditions. Jon, without a second’s hesitation, committed a second
drone to the December 22 mission after the first drone went missing. The drone was
eventually located in a tree and this once again brought John Berryhill to the rescue. John
rescued a drone from an earlier incident at the pond several years ago. This time the
placement in the tree was much worse and he was not able to reach it directly. Although
this drone will never fly again, its removal from the tree was very much appreciated!
Thanks to Amelia Murphy, Emma Geissinger Cutchins and Greg de Wet who worked both in
the field and in the lab. They collected critical RTK GPS data on pond bottom control points
and river profiles, sometimes having to break ice to get into the river at the Route 10 bridge.
They also analyzed samples in the lab for suspended sediment concentrations. Marc
Anderson helped with sampling and instrumentation at the monitoring stations and
supervised all lab activities. Steve Davis is the projects IT expert. His assistance has been
critical in keeping the data loggers connected to the network and he is always willing to go
in the field to trouble shoot equipment. Finally, project manager Gary Hartwell is always
assisting with anything that needs to be done. He is also incredibly knowledgeable about
the Mill River system and has played a critical role in this projects success.
In Memoriam
“Pebbles”
“Pebbles” was lost on December 22 during the mission to
collect the post sediment redistribution data. It appeared to
be grabbed by a large sycamore tree near the edge of the
pond. Note the search team to the right. Although the drone
was recovered it suffered irreparable damage.
a
prepared by
prepared for
Trustees of the Smith College
Facilities Management
126 West Street
Northampton, MA 01063
April 2022
Freshwater Mussel Monitoring in the Mill River for the
Paradise Pond Sediment Management Project: 2016-2021
biodrawversity
Biodrawversity LLC
206 Pratt Corner Road
Leverett, MA 01054
REPORT
NHESP File Number: 10-27790
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
1
INTRODUCTION
From 2016 to 2021, Biodrawversity LLC collected
baseline data (2016 and 2017) and implemented an-
nual monitoring (2017 to 2021) of the mussel commu-
nity and aquatic habitat in the Mill River downstream
from Paradise Pond in Northampton, Massachusetts.
Qualitative odonate (dragonfly and damselfly) sur-
veys were also conducted in these areas in 2017. This
research was completed as part of the Paradise Pond
Sediment Management Protocol Project (“Project”) that
is being conducted by Smith College. One of the goals
of the Project is to facilitate the export of accumulated
sediment from Paradise Pond via natural streamflow
to avoid costly dredging and to potentially improve
downstream habitat. Recent management efforts
have focused on redistributing fine sediment within
Paradise Pond to put it nearer the dam, and then to
open the dam’s sluice gates during high-flow events
to allow the river to transport sediment downstream.
Once it passes downstream from the dam, the sedi-
ment will likely be widely dispersed in the Mill River
downstream to the Oxbow, or deposited on the river’s
floodplain. Smith College is studying the transport
and fate of this sediment.
The Massachusetts Natural Heritage and Endan-
gered Species Program (NHESP) required baseline
data collection and long-term monitoring of mus-
sels to determine the potential effects (beneficial or
adverse) of sediment release on state-listed mussels
living downstream from Paradise Pond, and if adverse
effects on mussels are documented, to evaluate al-
ternate approaches. State-listed mussel species that
have historically been documented in the lower Mill
River and the nearby Oxbow include Ligumia nasuta
(Eastern Pondmussel; Special Concern), Strophitus
Mussel monitoring Site 1 in the Mill River in Northampton, Massachusetts.
Ligumia nasuta (Eastern Pondmussel)
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
2
undulatus (Creeper; Special Concern), Alasmidonta
heterodon (Dwarf Wedgemussel; Endangered), and
Lampsilis cariosa (Yellow Lampmussel; Endangered).
NHESP also required a baseline qualitative survey of
odonates in the same general areas of the mussel sur-
vey to determine species composition of the odonate
community, with focus on state-listed species that
have been documented in the Mill River. Target state-
listed species included Boyeria grafiana (Ocellated
Darner), Gomphus abbreviatus (Spine-crowned Club-
tail), Gomphus ventricosus (Skillet Clubtail), and Neuro-
cordulia yamaskanensis (Stygian Shadowdragon).
This annual report, which is revised each year to
include the newest annual monitoring, summarizes
the qualitative and quantitative baseline mussel
study (2016 and 2017) and annual mussel monitoring
(2017 to 2021). The odonate data were described in
the 2017 annual report.
METHODS
The objective of the freshwater mussel study was to
obtain data on mussel presence, density, shell length
distribution, and habitat using a combination of
quantitative (quadrats) and qualitative (timed search-
es) survey methods at five locations in the Mill River
(Figure 1). Mussel survey locations overlap with the
sediment monitoring sites that are being studied by
Dr. Robert Newton of Smith College.
1. Qualitative Mussel Survey
Timed qualitative surveys were conducted at five
locations (Figure 1). Sites 1-4 were surveyed quali-
tatively in 2016, and Site 5 was partially surveyed in
2016 and 2017, and fully surveyed from 2018 to 2021.
Surveys were completed using a combination of
snorkeling, SCUBA diving, and wading. Biologists re-
corded the shell length, shell condition, location, and
habitat of state-listed species encountered during
these surveys, and also recorded counts (or general
abundance) of co-occurring species. Shell condition
refers to the degree of shell erosion (e.g., loss of perio-
stracum or other damage); for each mussel, biologists
subjectively assign a numeric score ranging from 0.0
(no shell erosion) to 1.0 (severe loss of periostracum
or other damage) and these scores are averaged for
all individuals in a sample to produce an index of shell
Mussel monitoring Site 2 in the Mill River in Northampton, Massachusetts.
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
3
N
Site 1 Mill RiverParadise P
o
n
d
OxbowSite 2
Site 3
Site 4
Site 5
Hulberts Pond
Hulberts Pond sedimentmonitoring site(not used formussel survey)
Figure 1. Qualitative mussel survey sites (5) in the Mill River from Paradise Pond
to the Oxbow.
(40 quadrats per strata;double quadrat densityin Strata 1 and 3
N
Strata 1Strata 2Strata 3
Figure 2. Site 1 for the quantitative mussel sampling, showing stratification for
quadrat sampling.
Strata 1Strata 2Strata 3
(40 quadrats per strata;double quadrat densityin Strata 1 and 3
N
Figure 3. Site 2 for the quantitative mussel sampling, showing stratification for
quadrat sampling.
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
4
condition ranging from 0 to 1. These qualitative sur-
veys determined that mussel densities in areas be-
tween Paradise Pond and Route 10 (i.e., Sites 3, 4, and
5) were considered too low for the quantitative sam-
pling described below.
2. Quantitative Mussel Survey
Quantitative sampling was initially conducted at two
locations downstream from Route 10 (Figures 2 and 3)
between September 25-27, 2016. Quantitative sam-
pling was repeated at these same two sites in 2017
(May 23-25), 2018 (June 18-20), 2019 (June 4-6), 2020
(June 3-5), and 2021 (June 8-10). Study plots spanned
the width of the river and were 50 meters in length.
A total of 120 quadrats (size = 1.0m2) were sampled
within each plot, arranged along 12 transects with
10 quadrats per transect. One exception was in 2021
when six of the quadrat locations at Site 1 were under-
neath a large impenetrable accumulation of coarse
woody debris and were therefore omitted. Qualitative
surveys at these locations showed that the primary
target species (L. nasuta) was more common in shal-
low and intermediate depths close to the shoreline,
and therefore each plot was divided into three longi-
tudinal strata and the quadrat density was doubled
in strata 1 and 3. Strata 2 occupied the middle half of
the channel, and were therefore approximately twice
as large as strata 1 and 3. Quadrat allocation included
four transects (40 quadrats) in strata 1 and 3, and four
transects (40 quadrats) in strata 2.
Biologists recorded the numbers of uncommon
mussel species (i.e., all species except for Elliptio com-
planata)) at the surface of the sediment in the entire
1.0m2 quadrat, and recorded the number of E. com-
planata at the surface of the sediment in a randomly
selected quarter of each quadrat. The different meth-
od for E. complanata was meant to reduce the survey
time, as this species can be too abundant to count ef-
ficiently. Biologists also excavated and sieved the top
10 centimeters of sediment from a randomly selected
quarter of each quadrat, and recorded the number of
individuals of each species that were found buried.
Counts for surface mussels versus buried mussels
were recorded separately. Biologists record the shell
length and shell condition of all state-listed species.
For each quadrat, biologists recorded water
depth, the presence and percent cover of different
types of substrate, substrate embeddedness, quali-
tative estimate of flow velocity, and percent cover of
submerged aquatic vegetation and woody debris.
Raw data can be provided upon request (MS Excel
file) but are omitted from this report. Substrate type
and percent cover were estimated visually using the
following categories: clay, silt, sand, gravel, cobble,
and riprap. Embeddedness was visually estimated for
quadrats that contained coarse and gravel and cob-
ble, using ratings described in Platts et al. (1983):
• <5 percent of surface covered by fine sediment
• 5-25 percent of surface covered by fine sediment
Transect and quadrat at Site 2.
Sieved materials from a quadrat along the shoreline of Site 2, with mix of trash
and three L. nasuta.
Site 5 between Route 20 and the Paradise Pond Dam.
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
5
• 25-50 percent of surface covered by fine sediment
• 50-75 percent of surface covered by fine sediment
• >75 percent of surface covered by fine sediment
3. Analysis
Mussel density (mussels/m2) and population size es-
timates (with 90% confidence intervals) of each spe-
cies were computed for each plot. The density and
population estimates were computed in two ways:
(1) all quadrats weighted equally, and (2) using the
stratification described above, where quadrats are
not weighted equally because of the double density
of quadrats in strata 1 and 3. Shell length and condi-
tion data are summarized as a means of demonstrat-
ing age structure (inferred from length-frequency dis-
tributions), recruitment, and mussel health.
RESULTS
1. Qualitative Mussel Survey
From 2016 to 2021, eight mussel species were found
during qualitative surveys (Table 2). Of the potential
state-listed species in the study area, L. nasuta were
found only at Site 1 and Site 2, and S. undulatus were
found only at Site 5. No L. cariosa or A. heterodon were
found. Other uncommon species found included
Margaritifera margaritifera at Site 5, one dead Alas-
midonta undulata at Site 2, and Pyganodon cataracta
at Sites 1 and 2. Lampsilis radiata was relatively more
common, and E. complanata was by far the most nu-
merous. It is also important to note that some species
were detected during quantitative sampling at some
sites (not during qualitative sampling), including S.
undulatus at Sites 1 and 2, M. margaritifera at Site 1,
Latin Name (abbreviation)Common Name Status
Alasmidonta undulata (AlUn)Triangle Floater
Anodonta implicata (AnIm)Alewife Floater
Elliptio complanata (ElCo)Eastern Elliptio
Lampsilis radiata (LaRa)Eastern Lampmussel
Ligumia nasuta (LiNa)Eastern Pondmussel Special Concern
Margaritifera margaritifera (MaMa)Eastern Pearlshell
Pyganodon cataracta (PyCa)Eastern Floater
Strophitus undulatus (StUn)Creeper Special Concern
Table 1. Mussel species documented in the Mill River for the Paradise Pond Sedi-
ment Management Project, from 2016 to 2021.
Site 5
Site 1 Site 2 Site 3 Site 4 2016 2017 2018 2019 2020 2021
Start Latitude 42.30068 42.30462 42.30656 42.31018 42.31326 42.31557 42.31326 42.31326 42.31326 42.31326
Longitude -72.65159 -72.64765 -72.64631 -72.64060 -72.64072 -72.64101 -72.64072 -72.64072 -72.64072 -72.64072
End Latitude 42.30170 42.30597 42.30725 42.31173 42.31599 42.31630 42.31630 42.31630 42.31630 42.31630
Longitude -72.65097 -72.64644 -72.64534 -72.63952 -72.64098 -72.64079 -72.64079 -72.64079 -72.64079 -72.64079
Survey Date 9/9/2016 9/8/2016 9/8/2016 9/8/2016 9/9/2016 6/12/2017 5/14/2018 6/1/2019 5/19/20 6/1/21
Duration (person-hrs)2.0 3.3 2.0 2.0 2.0 2.0 10.0 10.0 11.0 8.0
Mean Depth (ft)2.50 2.80 2.50 0.50 1.25 ~3-4 3.0 3.0 3.0 3.0
Max Depth (ft)5.00 6.00 4.50 1.50 3.00 ~12.0 ~12.0 ~12.0 ~12.0 ~12.0
Flow Velocity <0.1 m/s <0.1 m/s <0.1 m/s Variable Variable Variable Variable Variable Variable Variable
Dominant Substrate(s)*Cl-Si-S-G Cl-Si-S-G-W Si-RR S-G-RR Si-S-G-C S-G-C-B-BR S-G-C-B-BR S-G-C-B-BR S-G-C-B-BR S-G-C-B-BR
Species Counts**
L. nasuta 7 20 0 0 0 0 0 0 0 0
S. undulatus 0 0 0 0 1 0 5 4 2 2
M. margaritifera 0 1 0 0 1 (1)18 14 20 22 19
A. undulata 0 0 (1)0 0 0 0 0 0 0 0
P. cataracta 1 6 0 0 0 0 0 0 0 0
L. radiata 10 27 0 0 (3)10 (25)2 (1)185 (26)120 (33)73 (41)60 (24)
E. complanata 100s 100s 0 0 1 1 2 0 0 3
Table 2. Locations, survey effort, habitat, and mussel counts for qualitative mussel surveys in the Mill River, 2016 to 2021.
*Substrate Abbreviations: Cl = Clay, Si = silt, S = sand, G = gravel, C = cobble, B = boulder, BR = bedrock, RR = riprap, W = wood/detritus
**Number in parentheses indicates shell counts.
***Site 5 survey areas: 2016 = Route 20 to the walking bridge. 2017 = walking bridge to the Paradise Pond dam, 2018-2021 = Route 20 to the Paradise Pond dam.
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
6
L. nasuta from Site 1.
S. undulatus from Site 5.
A. undulata from Site 1.
A. undulata at Sites 1 and 2, and Anodonta implicata
at Site 2.
Highest densities of mussels were found at Site 1
and Site 2 (Table 2). Several hundred E. complanata
were observed at these two sites. E. complanata was
numerous along stable banks and also throughout
the channel, whereas other mussel species were
found primarily along stable banks and were uncom-
mon or absent in the middle half of the channel. No
mussels, live or dead, were found at Site 3. Only three
L. radiata shells were found at Site 4.
In 2018, 185 L. radiata were found during a longer
duration (nearly 10 person-hours) qualitative survey
at Site 5, along with 14 M. margaritifera, five S. undula-
tus, and two E. complanata (Table 2). Similar numbers
were observed during the qualitative survey in 2019:
120 L. radiata (and 33 shells), 20 M. margaritifera, and
four S. undulatus. The 2020 qualitative survey at Site 5
was generally consistent with the 2019 and 2018 re-
sults, except with fewer L. radiata (73 live, 41 shells);
22 M. margaritifera and two S. undulatus were also
found. Lastly, the 2021 qualitative survey result was
generally consistent with the 2020 result: 60 L. radiata
(+24 shells), 19 M. margaritifera, two S. undulatus, and
three E. complanata.
These qualitative survey data suggest a declining
L. radiata population at Site 5, from a high of 185 in
2018 down to 60 in 2021, and with 98 shells found
from 2019 to 2021. The few S. undulatus found at Site
5 are large and in poor condition. The M. margaritifera
population at Site 5, though small, appears to be
stable. Historic data suggest that mussels are nearly
absent within and upstream from Paradise Pond (one
live M. margaritifera was found upstream in 2016).
2. Quantitative Mussel Survey
From 2016 to 2021, eight species were detected dur-
ing quantitative sampling, including five species
at Site 1 and seven species at Site 2. Tables 3 and 4
summarize density and population size estimates for
species at each site, giving equal weight to all of the
quadrats (disregarding stratification). Tables 5 and 6
show the same statistics that were computed based
on the stratified design, with double density of quad-
rats in strata 1 and 3. Raw data can be provided upon
request. Density estimates were computed for each
strata at each site, and the overall population estimate
was the sum of the population estimates for each of
the three strata. Tables 7 and 8 more clearly contrast
the population estimates derived from these two dif-
ferent methods, for all six years. The population es-
timates based on the stratified design are probably
more accurate.
Site 1 (2016): Three species were detected. Based on
the stratified design, we estimate population sizes of
4,190 E. complanata, 158 L. radiata, and 25 L. nasuta.
E. complanata comprised 95.8 percent of the mussel
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
7
Table 4. Density and population estimates for each species at Site 2. Data ana-
lyzed without stratification. Species abbreviations as in Table 1.
Species
Statistic LaRa LiNa StUn AlUn PyCa AnIm ElCo ALL
Year: 2016
Density 0.09 0.22 0.00 0.00 0.05 0.00 1.93 2.29
St. Dev.0.53 0.84 0.00 0.00 0.39 0.00 3.11 3.16
Estimate 105 249 0 0 58 0 2,223 2,635
90% CI (Lo)13 104 0 0 -9 0 1,688 2,091
90% CI (Hi)197 394 0 0 124 0 2,759 3,180
Year: 2017
Density 0.04 0.53 0.01 0.03 0.03 0.00 3.63 4.26
St. Dev.0.24 1.43 0.09 0.16 0.16 0.00 4.58 5.01
Estimate 48 604 10 29 29 0 4,178 4,897
90% CI (Lo)7 358 -6 2 2 0 3,390 4,035
90% CI (Hi)89 849 25 56 56 0 4,967 5,759
Year: 2018
Density 0.00 0.23 0.01 0.00 0.08 0.00 3.70 4.03
St. Dev.0.00 0.87 0.09 0.00 0.31 0.00 6.98 7.04
Estimate 0 222 8 0 79 0 3,515 3,824
90% CI (Lo)0 98 -5 0 36 0 2,522 2,822
90% CI (Hi)0 345 21 0 123 0 4,508 4,825
Year: 2019
Density 0.01 0.58 0.03 0.01 0.04 0.00 14.80 15.47
St. Dev.0.09 1.56 0.18 0.09 0.20 0.00 20.58 20.79
Estimate 8 546 32 8 40 0 14,060 14,693
90% CI (Lo)-5 324 6 -5 11 0 11,132 11,736
90% CI (Hi)21 768 57 21 68 0 16,988 17,651
Year: 2020
Density 0.03 0.56 0.01 0.01 0.07 0.01 6.40 7.08
St. Dev.0.16 2.34 0.09 0.09 0.40 0.09 10.53 10.61
Estimate 24 530 8 8 63 8 6,080 6,721
90% CI (Lo)1 197 -5 -5 6 -5 4,582 5,212
90% CI (Hi)46 864 21 21 121 21 7,578 8,230
Year: 2021
Density 0.01 0.47 0.00 0.00 0.13 0.00 8.70 9.31
St. Dev.0.09 1.32 0.00 0.00 0.72 0.00 13.22 13.35
Estimate 8 443 0 0 127 0 8,265 8,843
90% CI (Lo)-5 256 0 0 24 0 6,385 6,945
90% CI (Hi)21 630 0 0 229 0 10,145 10,741
Table 3. Density and population estimates for each species at Site 1. Data ana-
lyzed without stratification. Species abbreviations as in Table 1.
Species
Statistic LaRa LiNa AlUn PyCa MaMa ElCo JUV*ALL
Year: 2016
Density 0.18 0.03 0.00 0.00 0.00 5.13 0.00 5.35
St. Dev.0.77 0.37 0.00 0.00 0.00 11.92 0.00 12.02
Estimate 174 32 0 0 0 4,877 0 5,083
90% CI (Lo)65 -20 0 0 0 3,182 0 3,373
90% CI (Hi)283 84 0 0 0 6,572 0 6,792
Year: 2017
Density 0.14 0.20 0.00 0.00 0.00 11.83 0.00 12.18
St. Dev.0.37 1.00 0.00 0.00 0.00 24.98 0.00 25.37
Estimate 135 190 0 0 0 11,242 0 11,566
90% CI (Lo)81 48 0 0 0 7,689 0 7,958
90% CI (Hi)188 332 0 0 0 14,794 0 15,174
Year: 2018
Density 0.03 0.35 0.00 0.00 0.00 7.47 0.77 8.62
St. Dev.0.18 1.06 0.00 0.00 0.00 10.66 2.03 10.78
Estimate 32 333 0 0 0 7,093 728 8,186
90% CI (Lo)6 182 0 0 0 5,577 440 6,653
90% CI (Hi)57 483 0 0 0 8,610 1,017 9,718
Year: 2019
Density 0.02 0.33 0.01 0.00 0.01 17.83 0.00 18.19
St. Dev.0.13 1.11 0.09 0.00 0.09 23.35 0.00 23.62
Estimate 16 309 8 0 8 16,942 0 17,282
90% CI (Lo)-2 151 -5 0 -5 13,620 0 13,922
90% CI (Hi)34 466 21 0 21 20,263 0 20,642
Year: 2020
Density 0.08 0.35 0.01 0.01 0.00 9.77 0.00 10.21
St. Dev.0.43 1.33 0.09 0.09 0.00 12.43 0.00 12.67
Estimate 71 333 8 8 0 9,278 0 9,698
90% CI (Lo)10 143 -5 -5 0 7,510 0 7,896
90% CI (Hi)133 522 21 21 0 11,047 0 11,500
Year: 2021
Density 0.03 0.10 0.00 0.12 0.00 13.02 0 13.26
St. Dev.0.16 0.38 0.00 0.65 0.00 18.62 0 18.69
Estimate 25 92 0 117 0 12,367 0 12,600
90% CI (Lo)2 37 0 21 0 9,650 0 9,873
90% CI (Hi)48 146 0 212 0 15,083 0 15,327
*tiny juvenile mussels, species not determined
community at this site. All mussels were relatively less
common along the left bank (strata 1); we estimated
225 E. complanata (90% confidence limit: 87–363)
and no other species in strata 1. In contrast, we found
three species and estimated 3,338 mussels (90% con-
fidence limit = 2,161–4,514) in strata 3. Strata 2 had
intermediate densities of E. complanata and few L. ra-
diata. L. nasuta were only detected in Strata 3.
Site 1 (2017): The same three species were detected
at Site 1 in 2017. Based on the stratified design, we
estimate population sizes of 9,415 E. complanata
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
8
Table 5. Density and population estimates by strata for each species encountered at Site 1, 2016 to 2021.
Strata 1 Strata 2 Strata 3
90% CL 90% CL 90% CL All Strata
Species Year Density Estimate Lo Hi Density Estimate Lo Hi Density Estimate Lo Hi Estimate
L. nasuta 2016 0.00 0 0 0 0.00 0 0 0 0.10 25 -16 66 25
2017 0.00 0 0 0 0.40 180 3 357 0.20 50 -3 103 230
2018 0.60 150 62 238 0.13 31 -11 73 0.33 81 15 148 263
2019 0.65 163 59 266 0.00 0 0 0 0.33 146 30 263 309
2020 0.23 56 22 91 0.00 0 0 0 0.83 371 117 626 428
2021 0.18 44 11 76 0.00 0 0 0 0.11 50 1 99 94
L. radiata 2016 0.00 0 0 0 0.10 45 -29 119 0.45 113 39 186 158
2017 0.03 6 -4 17 0.10 45 10 80 0.30 75 42 108 126
2018 0.00 0 0 0 0.00 0 0 0 0.10 25 5 45 25
2019 0.00 0 0 0 0.03 6 -4 17 0.03 11 -7 30 18
2020 0.00 0 0 0 0.05 13 -2 27 0.18 79 -4 162 91
2021 0.03 6 -4 17 0.05 13 -2 28 0.00 0 0 0 19
E. complanata 2016 0.90 225 87 363 1.70 765 433 1,097 12.80 3,200 2,022 4,378 4,190
2017 2.40 600 339 861 2.70 1,215 632 1,798 30.40 7,600 5,234 9,966 9,415
2018 3.40 850 559 1,141 4.50 1,125 748 1,502 14.50 3,625 2,671 4,579 5,600
2019 18.70 4,675 3,279 6,071 10.80 2,700 2,048 3,352 24.00 10,800 7,095 14,505 18,175
2020 6.00 1,500 1,045 1,955 8.40 2,100 1,455 2,745 14.90 6,705 4,753 8,657 10,305
2021 7.80 1,950 1,393 2,507 6.63 1,658 1,219 2,096 25.56 11,500 8,123 14,877 15,108
A. undulata 2019 0.00 0 0 0 0.00 0 0 0 0.03 11 -7 30 11
2020 0.00 0 0 0 0.03 6 -4 17 0.00 0 0 0 6
P. cataracta 2020 0.00 0 0 0 0.00 0 0 0 0.03 11 -7 30 11
2021 0.20 50 -7 107 0.13 33 -11 77 0.03 13 -8 33 95
M. margaritifera 2019 0.00 0 0 0 0.03 6 -4 17 0.00 0 0 0 6
Juveniles 2018 0.90 225 87 363 1.40 350 178 522 0.00 0 0 0 575
All Species 2016 0.90 225 87 363 1.80 810 477 1,143 13.35 3,338 2,161 4,514 4,373
2017 2.43 606 342 870 3.20 1,440 821 2,059 30.90 7,725 5,318 10,132 9,771
2018 4.90 1,225 882 1,568 6.03 1,506 1,148 1,864 14.93 3,731 2,740 4,723 6,463
2019 19.35 4,838 3,452 6,223 10.85 2,713 2,059 3,366 24.38 10,969 7,191 14,746 18,519
2020 6.23 1,556 1,103 2,009 8.48 2,119 1,470 2,767 15.93 7,166 5,185 9,148 10,841
2021 8.20 2,050 1,477 2,623 6.82 1,704 1,252 2,156 25.69 11,563 8,176 14,949 15,316
(more than 2x higher than the 2016 estimate), 126 L.
radiata (slightly lower than the 2016 estimate), and
230 L. nasuta (much higher than the 2016 estimate of
25 mussels). See Table 5 for the 90% confidence in-
tervals associated with these population estimates.
E. complanata comprised 96.4 percent of the mussel
community at this site. As in 2016, all mussels were
relatively less common along the left bank (strata 1).
E. complanata were exceptionally common along the
right bank (strata 3), and more than double the den-
sity that was estimated for strata 3 in 2016. Strata 2
(center channel) had intermediate densities of E. com-
planata and few L. radiata, and unlike in 2016, L. na-
suta were detected in strata 2 (all juveniles found by
excavation, 12-15 mm in length, found in transects 5
[3] and 7 [1]).
Site 1 (2018): The same three species were detected
as in 2016 and 2017. In addition, many juvenile mus-
sels (10-18 mm) were detected that could not be reli-
ably identified, so these were treated separately in the
density and population size calculations. Based on
the stratified design, we estimated a population size
of 5,600 E. complanata, which is much lower than the
2017 estimate of 9,415 and comparable to the 2016
estimate of 4,190. We estimated 25 L. radiata, which
is much lower than the 2016 (158) and 2017 (126) es-
timates. We estimated 263 L. nasuta, which is slightly
higher than the 2017 estimate of 230, and much high-
er than the 2016 estimate of 25. We estimated 575 ju-
venile mussels that we could not identify; these were
likely mostly E. complanata based on proportions of
the three species found in this study area. See Table 5
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
9
for the 90% confidence intervals associated with these
population estimates. E. complanata comprised more
than 90 percent of the mussel community at this site.
As in previous years, mussels were relatively less com-
mon along the left bank (strata 1), except that a large
proportion of L. nasuta and juvenile mussels were
found in strata 1 for the first time. E. complanata were
exceptionally common along the right bank (strata 3).
The center channel (strata 2) had intermediate den-
sities of E. complanata, many juvenile mussels, few L.
nasuta, and no L. radiata.
Site 1 (2019): In addition to the three species that
were detected from 2016 to 2018, two additional spe-
cies were detected at Site 1 for the first time: M. mar-
garitifera (1) and A. undulata (1). Based on the strati-
fied design, we estimated a population size of 18,175
E. complanata, which is almost 2x higher than esti-
mates from any of the previous years, and almost 3.3x
higher than the 2018 estimate (Table 7). Most of this
increase was due to much higher numbers near the
shoreline (strata 1 and 3). We estimated 18 L. radiata
in 2019, which is lower but consistent with the 2018
result, and consistent with a downward trend from
2016 to 2019 (158, 126, 25, 18). We estimated 309 L.
nasuta, up from earlier estimates (263 in 2018, 230 in
2017, and 25 in 2016). See Table 5 for the 90% confi-
dence intervals for these population estimates.
E. complanata comprised more than 98 percent
of the mussel community at this site. As in previous
years, mussels were relatively less common along the
left bank (strata 1), except that a large proportion of L.
nasuta were found in strata 1 for the second consecu-
tive year. E. complanata were exceptionally common
along the right bank (strata 3). The center channel
had intermediate densities of E. complanata; just two
individuals of other species were detected in strata 2,
including one L. radiata and one M. margaritifera.
Site 1 (2020): Based on the stratified design, we es-
timated a population size of 10,305 E. complanata,
which is considerably lower than the 2019 estimate
but more consistent with the estimates in previous
years (though higher) (Table 7). We estimated 91 L.
radiata in 2020, higher than the 2018 and 2019 es-
timates and closer to the 2016 and 2017 estimates,
and opposing the downward trend observed over
the previous four years (158, 126, 25, 18, and 91 for
2016-2020, respectively). We estimated 428 L. nasuta,
up from the 2019 estimate and consistent with the
previous trend of an increasing population size (25,
230, 263, 309, and 428 for 2016 to 2020, respectively).
A. undulata were detected again in 2020 for the sec-
ond consecutive year, P. cataracta was detected at Site
1 for the first time, and M. margaritifera was not de-
tected in 2020 after it was found in 2019. See Table 5
for the 90% confidence intervals for these population
estimates. As in previous years, mussel densities were
highest along the right bank (strata 3).
Site 1 (2021): Based on the stratified design, we es-
timated a population size of 15,108 E. complanata,
which is considerably higher than the 2020 estimate
of 10,305 and more consistent with the 2019 estimate
of 18,175 (Table 7). We estimated 19 L. radiata in 2021,
sharply lower than the 2020 estimate of 91, but simi-
lar to the 2018 and 2019 estimates. We estimated 94 L.
nasuta, which is sharply lower than the 2020 estimate
of 428 and the first evidence of a decline in numbers
since monitoring began. See Table 5 for the 90% con-
fidence intervals for these population estimates. Nei-
ther A. undulata nor M. margaritifera were detected in
2021, and P. cataracta increased considerably in 2021
after it was first detected in 2020. As in previous years,
mussel densities were highest along the right bank
(strata 3). It is worth noting that six of the quadrats
at Site 1 in the center and right side of the channel
(strata 2 and 3; transects 7, 8, 10, 11, and 12) could not
be surveyed because they were under a dense accu-
mulation of coarse woody debris.
Site 2 (2016): Four species were detected. Based on
the stratified design, we estimate population sizes of
2,665 E. complanata, 108 L. radiata, 195 L. nasuta, and
51 P. cataracta. E. complanata comprised 88.3 percent
of the mussel community at this site. All mussels were
relatively less common along the right bank (strata
3), where we estimated 218 mussels (90% confidence
limit: 97–338). In contrast, we estimated 2,104 mus-
sels (90% confidence limit = 1,571–2,636) in strata
2, although these were mostly E. complanata (2,035)
and L. radiata (55). L. nasuta were not found in strata
2. Four species were found in strata 1, including rela-
tively high numbers of L. nasuta (population estimate
= 173; 90% confidence interval = 67–278) but inter-
mediate or low numbers of other species and a total
of 698 mussels (90% confidence interval = 461–934).
Site 2 (2017): Six species were detected at Site 2 in
2017, two more (S. undulatus and A. undulata) than
in 2016. Based on the stratified design, we estimate
population sizes of 4,745 E. complanata (much higher
than 2016), 108 L. radiata (lower than 2016), 498 L.
nasuta (higher than 2016), 23 P. cataracta (lower than
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
10
Table 6. Density and population estimates by strata for each species encountered at Site 2, 2016 to 2021.
Strata 1 Strata 2 Strata 3
90% CL 90% CL 90% CL All Strata
Species Year Density Estimate Lo Hi Density Estimate Lo Hi Density Estimate Lo Hi Estimate
L. nasuta 2016 0.58 173 67 278 0.00 0 0 0 0.08 23 -5 50 195
2017 1.15 345 183 507 0.10 55 -35 145 0.33 98 26 169 498
2018 0.40 100 30 170 0.03 6 -4 17 0.28 69 3 134 175
2019 1.25 375 195 555 0.23 68 -2 137 0.25 138 32 243 580
2020 1.68 503 203 802 0.00 0 0 0 0.00 0 0 0 503
2021 0.80 240 115 365 0.50 150 31 269 0.10 55 12 98 445
L. radiata 2016 0.13 38 -13 88 0.10 55 -35 145 0.05 15 -2 32 108
2017 0.05 15 -2 32 0.00 0 0 0 0.08 23 -5 50 38
2019 0.00 0 0 0 0.03 8 -5 20 0.00 0 0 0 8
2020 0.00 0 0 0 0.08 23 2 43 0.00 0 0 0 23
2021 0.00 0 0 0 0.00 0 0 0 0.03 14 -9 36 14
E. complanata 2016 1.50 450 231 669 3.70 2,035 1,496 2,574 0.60 180 67 293 2,665
2017 2.80 840 565 1,115 5.90 3,245 2,459 4,031 2.20 660 379 941 4,745
2018 2.70 675 379 971 3.80 950 629 1,271 4.60 1,150 497 1,803 2,775
2019 13.30 3,990 3,008 4,972 19.20 5,760 4,503 7,017 11.90 6,545 2,408 10,682 16,295
2020 3.70 1,110 498 1,722 13.50 4,050 3,077 5,023 2.00 1,100 187 2,013 6,260
2021 6.50 1,950 1,279 2,621 18.70 5,610 4,299 6,921 0.90 495 192 798 8,055
P. cataracta 2016 0.13 38 -13 88 0.03 14 -9 36 0.00 0 0 0 51
2017 0.08 23 2 43 0.00 0 0 0 0.00 0 0 0 23
2018 0.13 31 10 53 0.13 31 5 57 0.00 0 0 0 63
2019 0.10 30 6 54 0.03 8 -5 20 0.00 0 0 0 38
2020 0.18 53 0 105 0.03 8 -5 20 0.00 0 0 0 60
2021 0.20 60 -9 129 0.20 60 -9 129 0.00 0 0 0 120
A. undulata 2017 0.05 15 -2 32 0.00 0 0 0 0.03 8 -5 20 23
2019 0.00 0 0 0 0.03 8 -5 20 0.00 0 0 0 8
2020 0.00 0 0 0 0.03 8 -5 20 0.00 0 0 0 8
S. undulatus 2017 0.00 0 0 0 0.00 0 0 0 0.03 8 -5 20 8
2018 0.00 0 0 0 0.03 6 -4 17 0.00 0 0 0 6
2019 0.00 0 0 0 0.10 30 6 54 0.00 0 0 0 30
2020 0.00 0 0 0 0.03 8 -5 20 0.00 0 0 0 8
A. implicata 2020 0.03 8 -5 20 0.00 0 0 0 0.00 0 0 0 8
All Species 2016 2.33 698 461 934 3.83 2,104 1,571 2,636 0.73 218 97 338 3,019
2017 4.13 1,238 873 1,602 6.00 3,300 2,483 4,117 2.65 795 485 1,105 5,333
2018 3.23 806 495 1,117 3.98 994 672 1,316 4.88 1,219 563 1,874 3,019
2019 14.65 4,395 3,422 5,368 19.60 5,880 4,597 7,163 12.15 6,683 2,496 10,869 16,958
2020 5.58 1,673 1,009 2,336 13.65 4,095 3,121 5,069 2.00 1,100 187 2,013 6,868
2021 7.50 2,250 1,591 2,909 19.40 5,820 4,498 7,142 1.03 564 243 884 8,634
2016), as well as 8 S. undulatus and 23 A. undulata. See
Table 6 for the 90% confidence intervals associated
with these estimates. As in 2016, E. complanata com-
prised 89.0 percent of the mussel community at this
site. Mussels were least dense along the right bank
(strata 3), most dense in the center channel (strata 2),
and of intermediate density along the left bank (strata
1) (Table 6). Generally, more species were detected
near the banks, and the high density of mussels in
strata 2 was comprised mostly of E. complanata.
Site 2 (2018): Four species were detected during
quantitative sampling at Site 2 in 2018, two fewer
than in 2017. Neither A. undulata nor L. radiata were
detected in 2018. Based on the stratified design, we
estimate population sizes of 2,775 E. complanata, ap-
proximately ~2,000 lower than the 2017 estimate. We
estimated 3,109 mussels overall, down from the 2017
estimate of 5,333, but identical to the 2016 estimate.
See Table 6 for the 90% confidence intervals associ-
ated with these population estimates. The popula-
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
11
tion estimate of 175 L. nasuta was low compared to
the 2017 estimate of 498, but similar to the 2016 es-
timate of 195. E. complanata comprised 92.0 percent
of the mussel community at this site. Data indicate
a possible shift in mussel distribution and density in
the channel, with an increase in mussel density along
the right bank (strata 3), moderate decrease along the
left bank (strata 1), and a large decrease in the center
channel (strata 2) (Table 6). Generally, more species
were detected near the banks, and L. nasuta density
was much higher along both banks.
Site 2 (2019): Six species were detected during quan-
titative sampling at Site 2 in 2019, the same as 2017
and two higher than in 2018. Based on the stratified
design, we estimate population sizes of 16,295 E. com-
planata, almost 6x higher than the 2018 estimate and
more than 3.4x higher than the 2017 estimate (Table
6). This also resulted in a far higher estimate for the
total number of mussels (all species)—16,958, com-
pared to 3,019, 5,333, and 3,019 for 2016 to 2018, re-
spectively. E. complanata comprised 96.1 percent of
all mussels. The population estimate of 580 L. nasuta
in 2019 was higher than the 2018 estimate of 175, but
consistent with the 2017 estimate of 498. Estimates
for the other four species have been low in all four
years. Generally, more species were detected near the
banks, and L. nasuta density was much higher along
both banks. E. complanata density was high across
the entire channel.
Site 2 (2020): Seven species were detected during
quantitative sampling, including the same six from
2019 and, for the first time, A. implicata. Based on the
stratified design, we estimate 6,260 E. complanata, al-
most one-third the estimate of 16,295 from the previ-
ous year, but more consistent with results from 2016
to 2018 (Tables 6, 8). E. complanata comprised 91.1
percent of all mussels. The population estimate of 503
L. nasuta in 2020 was consistent with previous results.
Estimates for the other species were low and gener-
ally consistent among all five years. More species and
individuals were detected near the right bank (Strata
1) or middle of the channel (Strata 2). Only E. compla-
nata was found in Strata 3, whereas a fair proportion
of L. nasuta had been found in Strata 3 in all prior
years.
Site 2 (2021): Only four species were detected during
quantitative sampling, down from seven species in
Table 8. Comparison of unstratified versus stratified population estimates for each species encountered at Site 2, from 2016 to 2021. Differences in the statified vs.
unstratified estimates are due to the unequal weighting of quadrats in the stratified design.
Unstratified Stratified
Species 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021
L. radiata 105 48 0 8 24 8 108 38 0 8 23 14
L. nasuta 249 604 222 546 530 443 195 498 175 580 503 445
S. undulatus 0 10 8 32 8 0 0 8 6 30 8 0
A. undulata 0 29 0 8 8 0 0 23 0 8 8 0
P. cataracta 58 29 79 40 63 127 51 23 63 38 60 120
A. implicata 0 0 0 0 8 0 0 0 0 0 8 0
E. complanata 2,223 4,178 3,515 14,060 6,080 8,265 2,665 4,745 2,775 16,295 6,260 8.055
All 2,635 4,897 3,824 14,693 6,721 8,843 3,019 5,333 3,019 16,958 6,868 8,634
Table 7. Comparison of unstratified versus stratified population estimates for each species encountered at Site 1, from 2016 to 2021. Differences in the statified vs.
unstratified estimates are due to the unequal weighting of quadrats in the stratified design.
Unstratified Stratified
Species 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021
L. radiata 174 135 32 16 71 25 158 126 25 18 91 19
L. nasuta 32 190 333 309 333 92 25 230 263 309 428 94
A. undulata 0 0 0 8 8 0 0 0 0 11 6 0
P. cataracta 0 0 0 0 8 117 0 0 0 0 11 95
M. margaritifera 0 0 0 8 0 0 0 0 0 6 0 0
E. complanata 4,877 11,242 7,093 16,942 9,278 12,367 4,190 9,415 5,600 18,175 10,305 15,108
Juveniles 0 0 728 0 0 0 0 0 575 0 0 0
All 5,083 11,566 8,186 17,282 9,698 12,600 4,373 9,771 6,463 18,519 10,841 15,316
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
12
2020. The three species not detected in 2021 included
A. undulata, S. undulatus, and A. implicata; these three
species were always rare in this reach. Based on the
stratified design, we estimate 8,055 E. complanata,
which is higher than the 2020 estimate of 6,260 but
only one-half the 2019 estimate of 16,295 (Tables 6,
8). The population estimate of 445 L. nasuta in 2021
was very consistent with the 2020 estimate of 503 and
also fairly consistent with previous years. P. cataracta
may have increased over the last three years. Esti-
mates for the other species were low and generally
consistent among all six years.
Rare Species Data: Table 9 summarizes shell length
and shell condition data for the two state-listed spe-
cies, L. nasuta and S. undulatus, collected during both
quantitative and qualitative sampling. These data are
summarized for each year for L. nasuta, and for all
years combined for S. undulatus because so few were
found. For all years combined (sample size = 263) L.
nasuta ranged in length from 11.0 to 111.0 mm, with
an average length of 55.2 mm and a shell condition
index of 0.17. Considering the difficulty of detecting
juvenile L. nasuta, the high proportion of small mus-
sels suggests strong recruitment in this population,
with notable young cohorts in 2017 and 2020. L. na-
suta were found at an average depth of 1.3 ft, in a mix
of clay, silt, sand, and gravel substrates, often among
vegetation and woody debris. Too few S. undulatus
were found for a meaningful length analysis, but the
21 individuals ranged in length from 52.0 to 89.0 mm
(average = 67.5). It is likely that some of the S. undu-
latus found at Site 5 were found repeatedly (i.e., each
year) so although we report a sample size of 21, it is
likely a fewer number of individual animals. S. undula-
tus generally had moderately to highly eroded shells
(shell condition index = 0.56), particularly the individ-
uals found at Site 5.
DISCUSSION
The annual quantitative surveys conducted from
2016 to 2021 followed the same study design, and
were completed by same personnel, yet yielded vari-
able density and population size estimates, especially
for E. complanata (Tables 7 and 8). At both Sites 1 and
2, E. complanata population size estimates were far
higher in 2019 than in the other five years of sam-
pling. We are not certain why E. complanata densities
were so much higher in 2019. Results were generally
more consistent for the other species.
• L. radiata: this species may have declined at both
Sites 1 and 2 over the last six years. The qualitative
survey conducted at Site 5 documented a much
larger L. radiata population in this reach than was
previously known, but the numbers of live L. ra-
diata detected has steadily dropped from 2018 to
2021, while shell counts indicate high mortality in
this reach.
• L. nasuta: populations had appeared to be in-
Ligumia nasuta S. undulatus
(all years)Statistic 2016 2017 2018 2019 2020 2021 ALL
Total Measured 36 34 41 63 49 40 263 21
Average Shell Length (mm)68.2 46.2 52.5 61.4 47.5 53.5 55.2 67.5
Standard Deviation 19.6 23.2 18.4 18.9 22.7 19.1 21.4 11.34
Min Length (mm)15.5 12.0 23.0 25.0 11.0 24.0 11.0 52.0
Max Length (mm)99.0 91.0 88.9 111.0 107.0 91.0 111.0 89.0
Shell Condition 0.15 0.04 0.12 0.23 0.22 0.22 0.17 0.56
Length Class Counts
< 20.0 mm 2 4 0 0 12 0 18 0
20.0 - 29.9 mm 1 10 2 2 2 4 21 0
30.0 - 39.9 mm 0 0 14 5 1 5 25 0
40.0 - 49.9 mm 1 3 7 11 5 12 39 0
50.0 - 59.9 mm 3 4 4 10 14 6 41 5
60.0 - 69.9 mm 11 6 3 19 9 4 52 9
70.0 - 79.9 mm 10 6 8 4 4 3 35 3
80.0 - 89.9 mm 4 0 3 6 1 4 18 4
90.0 - 99.9 mm 4 1 0 4 0 2 11 0
> 100.0 mm 0 0 0 2 1 0 3 0
Table 9. Summary of L. nasuta and S. undulatus shell length and shell condition, for all individuals measured from 2016 to 2021.
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
13
creasing at Site 1 from 2016 to 2020, but declined
sharply in 2021. L. nasuta seems to be more stable
at Site 2.
• P. cataracta: this species may have increased at
both Sites 1 and 2.
• Other species: A. undulata, S. undulatus, A. impli-
cata are rare and only infrequently observed at
Sites 1 and 2. At Site 5, S. undulatus are scarce and
possibly declining, and M. margaritifera counts
have been low but consistent.
With six years of quantitative monitoring now com-
pleted, we are gaining a better understanding of pop-
ulation sizes and trends for each species. However,
data are rather variable. Several factors are likely to
contribute to the year-to-year differences, possibly re-
lated to timing of the surveys and survey conditions.
The 2016 survey was completed in September, and the
2017 to 2021 surveys were completed in the spring
(late May to mid-June). Time of year and streamflows
strongly influence monitoring results, and therefore
surveys have been consistent from 2017 to 2021 to
reduce variability. Quantitative surveys should occur
only from late May to mid-June when river discharge
is below 100 cfs. This should control for key variables
such as water clarity, water temperature, water depth,
and flow conditions, and also for the behavior of each
species.
The objective of long-term mussel monitoring
is to determine if, and to what extent, sediment re-
leased during sluice experiments affect mussel pop-
ulations downstream from Paradise Pond. Initially, a
sluice experiment was defined as the opening of the
sluice gate at the Paradise Pond Dam during high-
flow events to allow the river to transport accumulat-
ed sediments from the impoundment to downstream
areas. Dr. Robert Newton has been studying the ex-
port and fate of sediment released during sluice ex-
periments. The sluice experiments showed that some
sediment passed through the gate during high flow
events (see Geoscience reports) but the volume of
sediment entering the Pond during high flow events
far exceeded what would pass through the gate. In
2016, Smith College experimented with redistribut-
ing sediment within the Paradise Pond impoundment
during an extended partial drawdown to increase the
amount of sediment closer to the dam, which may
increase sediment export during subsequent sluice
experiments. Baseline data on the freshwater mussel
community was conducted in September 2016 and
again in May 2017; both of these are considered base-
line datasets because no sluice experiments were
conducted during the intervening period.
Later sluice experiments showed that the volume
of sediment released through the sluice gate during
high flow events was still minimal compared to sedi-
ments entering the Pond. After submitting findings
from the sluice experiments and sediment redistri-
bution experiment of 2016 to all regulating bodies,
a new 10 year permit, NAE-2012-2550, was issued on
Sept. 10, 2019, from the Department of Army, New
England District.
From November 1 to November 20, 2019, Smith
College conducted the first sediment redistribution
project in Paradise Pond, which began with a pond
drawdown followed by use of a bulldozer to redistrib-
ute sediment. The sluice gate was operated to control
downstream sediment release. Prior to refilling the
pond, an experiment was completed to determine
sediment release rate when the pond was completely
Accumulation of coarse woody debris at Site 1 in 2021.Extensive sand deposit at Site 5 in 2021.
Freshwater Mussel Monitoring in the Mill River for the Paradise Pond Sediment Management Project: 2016-2021
14
drained. On November 19, the pond was allowed to
completely drain, with sluice gate open, and approxi-
mately 1,120m3 of sediment was washed downstream
and accumulated in the splash pool in ~40 minutes. A
high flow event on December 14 moved most of that
accumulated sediment from the splash pool to down-
stream areas.
Smith College continued with sediment redistri-
bution three times in 2020 and early 2021. These all
began with a pond drawdown and use of a bulldozer
to redistribute sediment. These were completed dur-
ing the weeks of November 16, December 14, and
January 25, each lasting approximately five days. Two
high-flow events transported some of this material
downstream from the dam, including a November
30-December 1 storm (2+ inch rain) and a December
25 storm (1+ inch rain accompanied by snowmelt).
The latter event smashed the flashboards and clogged
the sluice gate which resulted in a prolonged partial
drawdown. The flashboards were eventually restored
in May of 2021. The sluice gate clog was cleared in Oc-
tober 2021.
Quantitative mussel surveys will be repeated at
sites 1 and 2 in 2022; this is the core element of the
long-term monitoring program. The qualitative sur-
vey will also be repeated at Site 5. These surveys will
continue to build a robust dataset that will help to
understand effects of sediment releases on mussels.
Impact of Sediment Redistribution on
Macroinvertebrates in the Mill River
(2018-2021)
Marney Pratt 2/28/22 Smith College Department of Biological Sciences
1
Contents
Executive Summary ....................................................................................................................................... 2
Introduction .................................................................................................................................................... 2
Methods .......................................................................................................................................................... 4
Environmental Conditions ......................................................................................................................... 4
Macroinvertebrate Sampling ...................................................................................................................... 4
Data analysis .............................................................................................................................................. 7
Results ............................................................................................................................................................ 8
Environmental Conditions ......................................................................................................................... 8
Sediment Types ........................................................................................................................................ 11
Macroinvertebrate Density ....................................................................................................................... 12
Functional Feeding Groups ...................................................................................................................... 14
Discussion .................................................................................................................................................... 20
Conclusion ............................................................................................................................................... 21
Literature Cited ............................................................................................................................................ 21
Acknowledgements ...................................................................................................................................... 22
APPENDIX .................................................................................................................................................. 24
2
Executive Summary
• Macroinvertebrates are commonly used as bioindicators of human impacts on freshwater
ecosystems
• The impact assessed in this report is whether redistributing sediment in the early winter (Nov-Dec)
as part of Smith College’s Sediment Management Protocol in Paradise Pond affects
macroinvertebrates in the Mill River.
• A Before-After-Control-Impact design (or BACI design) can be used to assess impacts. In this
study, an interaction between year and location is a possible indication of an impact of the
sediment redistribution that took place in the winters during 2019-2021.
• There were no interactions between year and location that would indicate an impact of winter
sediment redistribution on the macroinvertebrates downstream of Paradise Pond. Thus, it is
recommended that redistributing the sediment in the winter is the best policy to reduce impact on
the macroinvertebrates downstream of the pond.
• There was consistently greater macroinvertebrate density downstream of Paradise Pond, which is
assumed to be a general impact of the pond increasing temperature and/or nutrients just
downstream of the pond.
• Macroinvertebrates, especially gathering collectors, increased in density both upstream and
downstream of Paradise Pond in 2020 and this may be an impact of reduced human activity during
the Covid-19 pandemic lockdown.
Introduction
Aquatic macroinvertebrates are useful bioindicators of potential impacts of how Smith College is
managing the sediment in Paradise Pond because many macroinvertebrates are very sensitive to
environmental impacts (Merritt et al. 2019). A common way to asses an impact on organisms in a river is
to use a Before-After-Control-Impact design (or BACI design)(Strayer and Smith 2003). A BACI design
has at least one sample before (Before sample) and after (After sample) an impact and has sample at a site
that is likely to be affected by the impact (Impact sample) as well as for a sample that should not be
affected by the impact (Control sample). In this report, I concentrate on the potential impact from the
redistribution of sediment in Paradise Pond in the early winter (November – December) of 2019 and 2020
on the macroinvertebrates in the Mill River in Northampton, MA. The “Impact” site is the rocky riffle
habitat downstream of Paradise Pond, while the “Control” site is a rocky riffle habitat upstream of the
pond (Figure 1). Since the last sediment redistribution happened in July 2016, and previous analyses
determined that things recovered by fall of 2017 (Pratt 2020), the samples taken in 2018 and 2019 are
assumed to be fairly typical and serve as the “Before” samples while 2020 and 2021 are the “After”
samples since they happened after each round of winter sediment redistribution in November 2019 and
December 2020.
When sediment was redistributed in Paradise Pond in the summer (July 2016), there was a small, short-
lived impact on the macroinvertebrates downstream of the pond (Pratt 2020). Longer-term monitoring has
shown that the typical situation is for there to be greater overall macroinvertebrate density downstream
compared to upstream of Paradise Pond. But after moving sediment in the summer, there were fewer
macroinvertebrates downstream than normal in the fall other than one gathering collector flathead mayfly,
Stenacron interpunctatum, which increased substantially. Gathering collectors eat organic material from
softer sediments, so an influx of fine sediment from redistribution in Paradise Pond may have benefitted
3
this mayfly while many other organisms were negatively impacted. It was hypothesized that by shifting
the sediment redistribution to the winter that any fine sediment released would be cleared out by winter
storms while the macroinvertebrates are mostly in a dormant state and would therefore have less of an
impact.
Figure 1. Location of sampling sites in the Mill River in Northampton, MA. The Impact site is the site
downstream of Paradise Pond, and the Control site is the site upstream of Paradise Pond. (Image from Google
Maps)
4
Methods
Environmental Conditions
Macroinvertebrate samples were collected in late May to early June (=Summer samples) as well as mid-
September to early October (=Fall samples)(Table 1). To understand the environmental conditions
leading up to when samples were taken, I calculated some environmental variables for the 2 months
before and during samplings for each season: April 15 to June 15 for Summer, and August 15 – October
15 for Fall in each year from 2018-2021.
Daily precipitation in millimeters as well as maximum and minimum temperature in Celsius were
downloaded from the from the NOAA Global Historical Climatology Network for the Amherst, MA
station (NOAA station: USC00190120) using the rnoaa package in R (Chamberlain 2020). Total
precipitation was calculated by summing the daily precipitation values for each 2-month period in each
year. Accumulated Degree Days (ADD) was calculated as the sum of the estimated average daily temp
(mean of the minimum and maximum daily temperature in Celsius) above 0℃ for each 2-month period in
each year.
River discharge in cubic feet per second (cfs) data were downloaded from the Mill River USGS National
Water Information System station in Northampton, MA (USGS site number: 01171500) using the
dataRetrieval package in R (De Cicco et al. 2018). The typical discharge for the 2-month period was
determined using the median (since discharge values were not at all normally distributed), and the
maximum discharge over the 2-month period was used to determine the result of the strongest storms in
that 2-month period.
Macroinvertebrate Sampling
Aquatic macroinvertebrates were sampled in 0.5 m x 0.5 quadrats from riffle habitats (areas with a rocky
bottom, shallow depth, rippled water surface, and relatively fast water flow) by kick-sampling using D-
nets (LaMotte D-net, 30 cm wide base, 500 micron mesh). Nets were placed downstream with the
opening facing upstream. One to two quadrats were sampled in each of five different subsampling areas
(called microhabitats here) within each location. When two quadrats were sampled within a microhabitat
on the same date, they were combined into one sample. The percent cover of organic debris (dead leaves
and sticks), sandy sediment, and larger rocks was estimated within each quadrat. Larger rocks were
picked up and any organisms found on the rocks were picked or rubbed off and placed in a tub of river
water. Once all larger rocks were cleaned off and placed outside of the quadrat, the substratum was
disturbed by kicking the bottom for 30 seconds to 1 minute making sure that the D-net caught anything
disturbed by the kicking. The contents of the net were rinsed into the tub of water.
The riffle sites upstream (42.319451, -72.654336) and downstream (42.315514, -72.64108) of Paradise
Pond (Figure 1) were sampled on 2-3 different days in the summer and fall (Table 1). Location
coordinates were taken with the built-in GPS sensor on a Vernier LabQuest 2 Interface (accuracy: half of
data points fall within a circle with radius of 2m). Because five microhabitats were sampled on three
different days in each location for a particular season and year, there were usually 15 samples for each
location in each year and season. One exception was in the Fall of 2018, when because of logistical and
time constraints all microhabitat samples were combined on the same date resulting in three samples from
each location. While there are fewer separate sampled for Fall of 2018, note that a similar total area was
sampled in each location (Table 1). Another exception was in the Fall of 2021, when weather conditions
made it unsafe to sample the upstream location on a third date. Thus, there were only 10 samples taken for
the upstream location in the Fall of 2021.
5
After collection, samples were taken back to lab, sieved down to 500 microns, and all macroinvertebrates
were sorted out and preserved in 70% ethyl alcohol. Most macroinvertebrates were identified using
identification keys (Peckarsky 1990; Merritt et al. 2019) down to genus or even species whenever
possible, however, some were only identified to family or occasionally just to order when identifications
were particularly difficult (such as for segmented worms and midge larvae). A complete list of organisms
found and their overall abundance in each location is given in Appendix A. Each organism was also
assigned to a functional feeding group (Stream Biomonitoring Unit Staff 2012; Merritt et al. 2019) as
indicated in Appendix A.
6
Table 1. Information for macroinvertebrates sampled in riffles upstream and downstream of Paradise
Pond in the Mill River, Northampton, MA. Each sample was from a particular microhabitat where one to
two 0.5 m x 0.5 m quadrats were used and then combined together at the microhabitat level (expect for
Fall 2018 when all microhabitat samples were combined for a particular sample date). The total area
sampled is the area of all quadrats, and the total number of organisms was the sum of all organisms
sampled for a combination of location, season, and year; and the organism density was the total number of
organisms divided by the area sampled.
Year Season Location Dates Number of Samples
Total Area Sampled (m2)
Total # Organisms
Organism Density (#/m2)
2018
Summer Downstream 05/29, 06/05, 06/07 15 7.50 2,350 313.3
Upstream 05/31, 06/06, 06/08 15 7.50 1,282 170.9
Fall Downstream 09/17, 09/25, 10/11 3 7.00 1,335 190.7
Upstream 09/24, 10/02, 10/04 3 7.75 265 34.2
2019
Summer Downstream 05/28, 06/03, 06/04 15 7.50 4,096 546.1
Upstream 05/30, 05/31, 06/05 15 7.50 1,467 195.6
Fall Downstream 09/17, 09/23, 09/26 15 7.50 1,608 214.4
Upstream 09/16, 09/24, 10/03 15 7.50 908 121.1
2020
Summer Downstream 06/06, 06/07, 06/09 15 3.75 8,926 2,380.3
Upstream 06/05, 06/07, 06/08 15 7.50 9,118 1,215.7
Fall Downstream 09/18, 09/20, 09/26 15 7.50 8,445 1,126.0
Upstream 09/19, 09/25, 09/27 15 7.50 2,868 382.4
2021
Summer Downstream 06/03, 06/08, 06/11 15 6.25 5,348 855.7
Upstream 06/02, 06/07, 06/10 15 6.25 1,710 273.6
Fall
Downstream 09/23, 09/28, 09/29 15 4.00 3,791 947.8
Upstream 09/22, 09/30 10 5.00 485 97.0
7
Data analysis
A BACI sampling design can detect a possible impact when there is an interaction between the sampling
location (in this case, Downstream site = Impact, and Upstream site = Control) and the sample that
happened before and after the impact where the impact we are most interested in was the sediment
redistribution that happened in Paradise Pond in the early winter of 2019 and 2020 (in this case, 2018 &
2019 = Before, 2020 & 2021 = After). Macroinvertebrate life cycles do vary a lot seasonally, so there is a
potentially substantial confounding seasonal effect. Thus, I generally analyzed the summer and fall data
separately. For the analyses in this report, I removed the crayfish recorded because their high mobility
makes it hard to know the true number within a quadrat. I also aggregated most of the macroinvertebrates
at the genus level but harder to identify organisms were identified to family (such as the Chironomidae
family of midges) or higher taxonomic levels (most worms were identified to phylum or class except for
the flatworm Girardia). A list of all the taxa included is in an Appendix at the end of this report
(Appendix A).
Functional Feeding Groups: Aquatic biologists categorize macroinvertebrates into functional feeding
groups (FFG) to better understand the ecology of a river ecosystem. The FFG’s in this report included:
scrapers (scrape periphyton off rocks and other surfaces), gathering collectors (gather or deposit feeds on
fine particulate organic matter in the sediment or on surface films), filtering collectors (filter feed or
suspension feed on fine particulate matter suspended in the water), shredders (chews live plant or dead
plant matter including more coarse particulate organic matter), and predators (ingest whole or partial
animals).
In addition to looking for possible impacts on each FFG separately, I also used the substrate stability ratio
which is the ratio of the scrapers and filtering collectors to the gathering collectors and shredders. Since
scrapers and filtering collectors generally prefer to be attached to rocks, while gathering collectors and
shredders prefer sand and organic debris such as decaying leaves, this ratio may indicate the relative
importance of rocky habitats to softer sediments. If sediment redistribution brings in more sediment, you
might expect the substrate stability ratio to decrease because gathering collectors and shredders would
likely do better with more sandy sediment and organic debris.
Impact on macroinvertebrates was assessed by calculating the density of all the macroinvertebrates as
well as each FFG separately. Density was calculated by counting the total number of organisms (all
macroinvertebrates or a specific FFG) in a sample and dividing that by the total sample area (m2).
Relative abundance of each FFG was also calculated by counting the total number of a particular FFG in a
sample and dividing that by the total of all macroinvertebrates in the same sample.
Statistical Analysis:
All statistical analyses were performed in R version 4.1.0 (R Core Team 2021)). In most ANOVA
models, the design was unbalanced (the number of observations was not equal in all of the interaction
cells), so I used type III sum of squares for the 2-way ANOVA full factorial models. For all ANOVA
models, post-hoc pairwise comparisons were made for terms that were significant using the Tukey HSD
adjustment. When the interaction term was significant, then interpreting main effects is problematic, so
pairwise comparisons were generally only made among the interaction pairwise groups. Main effects
were only interpreted when interaction terms were not significant. When the interaction was not
significant, a reduced model was performed using type II sum of squares and just main the effects, and
post-hoc pairwise comparisons were made for significant main effects with more than two levels using the
Tukey HSD adjustment. Anova models were performed using the lm function from the stats package (R
Core Team 2021), type II and III sum of squares were calculated with the Anova function from the car
8
package (version 3.0-12, Fox and Weisberg 2019), and post-hoc pairwise comparisons were performed
with the emmeans function from the emmeans package version 1.7.2 (Lenth 2022) using the Tukey
adjustment. In all analyses of density and substrate stability ratio, the values were log transformed before
analysis to better fit the assumption of normal residuals.
To understand the relationship between precipitation and proportion of sediment type, regression models
using the lm function from the stats package in R were performed to fit a quadratic polynomial (method =
y ~ poly(x, 2)). The value for the precipitation was the total precipitation in millimeter for the two months
before and including sampling (April 15 – June 15 for the Summer sampling, August 15 – October 15 for
the Fall sampling). Precipitation data were downloaded from the from the NOAA Global Historical
Climatology Network daily values from the Amherst, MA station (NOAA station: USC00190120) using
the rnoaa package in R (Chamberlain 2020). Since there was only one precipitation value for each
sampling time (each season in each year), I used the median percent cover of the sediment type for each
location in each season and year in the analysis. The regression was performed for each sediment type
(rock, sand, and organic debris) separately but with all seasons, years, and both locations used in one
model.
Results
Environmental Conditions
The air temperature over the 2-month period before and during summer sampling, as assessed by
Accumulated Degree Days (ADD), was within the interquartile range (IQR, which is the middle 50% of
data) in all years except 2020 where it was lower than usual (Figure 2). In the fall, the ADD was typical
in 2019 and 2020, but higher than normal in 2018 and 2021. Precipitation was within the IQR in 2018-
2020, but higher than normal in 2021 in the summer (Figure 2). In the fall, precipitation was typical in
2020, but higher than normal in 2018 and 2021 and lower than normal in 2019.
The median discharge of the Mill River for the 2-month period before and during sampling was within the
IQR in all years except for 2018 when it was slightly lower than normal in the summer (Figure 3). In the
fall, the median discharge was a little lower than the IQR in 2019 and 2020, but much higher than normal
in 2018 and 2021. The maximum river discharge in the summer was all within or just slightly above the
IQR for all years, while it was just below the IQR in 2019 and 2020 and much higher than normal in 2018
and 2021 in the fall (Figure 3).
9
Figure 2. (a) Accumulated Degree Days (ADD) and (b) total precipitation (mm) for the 2 months before and
during sampling in the late spring - early summer (April 15 to June 15, Summer) and the late summer – early fall
(August 15 – October 15, Fall). ADD was calculated as the sum of the estimated average daily temp (mean of the
minimum and maximum daily temperature in Celsius) above 0℃. Blue (precipitation) and red (ADD) points
represent the measurement for the 2-month period for each year and season, and the boxplot shows the median
(thick horizontal line), mean (X), and interquartile range (upper and lower hinge of box, IQR) for the long-term
(1938-2021). Whiskers represent the distance from the box hinge to the maximum or minimum value or no farther
than 1.5*IQR from the nearest hinge, and points outside of the whiskers are interpreted as outliers. Points to the
right of the box are the measures for each year from 2018-2021 and are labeled with the year they represent.
(a)
(b)
10
Figure 3. (a) Median and (b) maximum river discharge in cubic feet per second (cfs) for the 2 months before and
during sampling in the late spring - early summer (April 15 to June 15, Summer) and the late summer – early fall
(August 15 – October 15, Fall). Blue points represent the measurement for the 2-month period for each year in
each season and the boxplot shows the median (thick horizontal line), mean (X), and interquartile range (upper and
lower hinge of box, IQR) for the long-term (1938-2021). Whiskers represent the distance from the box hinge to
the maximum or minimum value or no farther than 1.5*IQR from the nearest hinge, and points outside of the
whiskers are interpreted as outliers. Points to the right of the box are the measures for each year from 2018-2021
and are labeled with the year they represent.
(a)
(b)
11
Sediment Types
Most of the sediment was rock for the location upstream (Median = 79%, IQR = 24%) as well as the
downstream (Median = 80%, IQR =17%) of Paradise Pond. Sand typically made up a little less than a
quarter of the substratum (upstream: Median = 17%, IQR = 20%; downstream: Median = 15%, IQR =
15%), while organic debris such as leaves and sticks generally made up less than 5% of the substratum
(upstream: Median = 2%, IQR = 5%; downstream: Median = 3%, IQR = 4%). There was some variation
in the proportion of the sediment over time between the locations and seasons (Figure 5). Notably, there
was generally more organic debris in the fall, and the overall pattern among years was similar between the
two locations with the following exceptions: (1) summer 2020 there was less rock and more sand
upstream, (2) summer 2021 there was less sand and organic debris and more rock upstream, and (3) fall
2018 there was more sand and less rock and organic debris upstream.
Since the pattern among years and seasons was reasonably similar between the two locations, it suggested
factors that would affect the whole river could show relationship with percent cover of sediment. All three
sediment types showed a non-linear relationship with the total precipitation over two months with the
highest percent cover of rock and the lowest percent cover of sand and organic debris occurring between
200-300 mm of rain over a two-month period (Figure 6). There was a significant but weak quadradic
polynomial relationship for rock (R2 = 0.51, P = 0.01), but no significant relationship between sand (R2 =
0.20, P = 0.24) or organic debris (R2 = 0.19, P = 0.26) and precipitation.
Figure 4. Proportion of different sediment types (rock, sand, and organic debris) for the riffle locations downstream
and upstream relative to Paradise Pond in the Mill River, Northampton, MA in the summer and fall from 2018-
2021. Percent cover was measured in a total of 15-30 quadrats (0.5 m x 0.5 m) in each location in each year and
season (see Table 1 for sampling dates).
12
Figure 5. Effect of two months of precipitation on percent cover of different sediment types in the Mill River,
Northampton, MA. Each point represents the median percent cover measured in 15-30 quadrats (0.5 m x 0.5 m)
sampled in each location during (Upstream and Downstream relative to Paradise Pond) in a particular year and
season (see Table 1 for sampling dates), and precipitation was summed for the 2 months before and during
sampling in the summer (April 15 to June 15) and the fall (August 15 – October 15). The curve represents a best fit
quadratic polynomial with 95% confidence interval shown in gray.
Macroinvertebrate Density
The density of macroinvertebrates (log of the number of organisms per meter squared) was not
significantly affected by the interaction between year and location in the summer (F3,112 = 1.3, P = 0.27) or
the fall (F3,83 = 2.4, P = 0.07). In both seasons, the density of macroinvertebrates was greater downstream
than upstream, and greater in 2020 than all other years (Table 2, Figure 7).
13
Table 2. Results from a full-factorial 2-way ANOVA comparing the log of macroinvertebrate density (#/m2) with
year and location as fixed effects for the summer and fall sampling season. The years included were 2018-2021,
and the locations were riffles upstream and downstream of Paradise Pond. Values shown include the type II sum of
squares (SS(II)), degrees of freedom (df), F-value, and P-value. Significant P-values are in bold red. (D =
Downstream, U = Upstream)
Season Term SS (II) df F-value P-value Post-hoc Comparisons
Summer
Year 62.3 3 48.9 <0.0001 2020 > (2018, 2019, 2021)
Location 19.8 1 46.6 <0.0001 D > U
Residuals 48.8 115
Fall
Year 45.7 3 17.5 <0.0001 2020 > (2018, 2019, 2021)
Location 33.5 1 38.4 <0.0001 D > U
Residuals 75.0 86
Figure 6. Changes in macroinvertebrate density (number of organisms per meter squared) over time in years for
the summer and fall in a riffle downstream and upstream of Paradise Pond in the Mill River, Northampton, MA.
Points represent the density from a microhabitat sample and horizontal lines represent the median of 3-15 samples
within a year for a location. The y-axis is log10 scaled. See Table 1 for sample sizes and dates and Table 2 for
statistical results.
14
Functional Feeding Groups
The overall relative abundance not taking location, year, or season into account was greatest for the
collector gatherers (45%) followed by scrapers (31%) and collector filterers (19%). Predators only made
up 3% and shredders 1.3% of the overall sample (<1% were not able to be assigned to a functional
feeding group). The pattern of functional feeding group relative abundance varied through time and with
season, but the upstream and downstream locations showed similar patterns overall (Figure 8).
Figure 7. Relative abundance of different functional feeding groups over time in years for the summer and fall in
the riffle locations upstream and downstream of Paradise Pond in the Mill River, Northampton, MA. Relative
abundance is the number of a functional feeding group divided by the total number of all organisms in the sample.
See Table 1 for sampling dates and sizes.
Substrate Stability Ratio. The log of the substrate stability ratio had a significant interaction between year
and location in the summer (Table 3), but the only difference between locations occurred in 2019 when
the ratio was greater downstream than upstream (Figure 9). The ratio was also greater in 2018 and 2019
compared to 2020 and 2021 in the summer (Figure 9). In the fall, the log of the substrate stability ratio
did not have a significant interaction between year and location (F3,83 = 0.1, P = 0.96), but there was a
main effect of location as well as year (Table 3). Overall, the substrate stability ratio was greater
downstream compared to upstream, and lower in 2020 compared to all other years in the fall (Figure 9).
15
Table 3. Results from a full-factorial 2-way ANOVA comparing the log of the substrate stability ratio with year
and location as fixed effects for the summer and fall sampling season. The years included were 2018-2021, and the
locations were riffles upstream and downstream of Paradise Pond. Values shown include the sum of squares (SS),
degrees of freedom (df), F-value, and P-value. Results with Intercept and Year x Location interaction terms are
type-III models, results with only main effects are type-II models. Significant P-values are in bold red. (D =
Downstream, U = Upstream)
Season Term SS df F-value P-value Post-hoc Comparisons
Summer
Intercept 20.3 1 43.0 <0.0001
Year 48.9 3 34.5 <0.0001
Location 0.1 1 0.3 0.62
Year x Location 4.1 3 2.9 0.04 D > U in 2019
(2018, 2019) > (2020, 2021)
Residuals 52.9 112
Fall
Year 64.8 3 11.1 <0.0001 2020 < (2018, 2019, 2021)
Location 9.3 1 4.8 0.03 D > U
Residuals 167.2 86
Figure 8. Substrate stability ratio over time in years for the summer and fall in the riffle locations upstream and
downstream of Paradise Pond in the Mill River, Northampton, MA. Points represent the substrate stability ratio
(number of scrapers and filtering collectors divided by the number of gathering collectors and shredders) from a
microhabitat sample and horizontal lines represent the median of 3-15 samples within a year for a location. The y-
axis is log10 scaled. The thicker brackets represent a significant difference among years where 2018 & 2019 had a
greater ratio than 2020 & 2021 in the summer. The thin bracket indicates a significant difference between locations
in the summer of 2019. See Table 1 for sample sizes and dates and Table 3 for additional statistical results.
16
Gathering Collectors. The interaction between year and location was not significant for the log density of
gathering collectors in the summer (F3,112 = 1.3, P = 0.27) or the fall (F3,83 = 0.4, P = 0.73). The density of
gathering collectors was greater downstream than upstream in the summer, but not quite different in the
fall (Table 4, Figure 10). There was a significant main effect of year on gathering collector density in the
summer and the fall (Table 4). In the summer, there was the greatest density of gathering collectors in
2020, intermediate density in 2021, and the lowest density in 2018 and 2019 (Figure 10). In the fall, the
density of gathering collectors was greater in 2020 than in 2018, 2019, and 2021 (Figure 10).
Scrapers. The interaction between year and location was not significant for the log density of scrapers in
the summer (F3,112 = 2.1, P = 0.10) but was significant for the fall (Table 4). The density of scrapers was
greater downstream than upstream in the summer and the fall (Table 4, Figure 10). There was a
significant main effect of year on scraper density in the summer with greater density in 2020 compared to
all other years (Table 4). In the fall, the only significant difference between locations and years was in
2020 when scrapers were more abundant downstream (Table 4).
Filtering Collectors. The interaction between year and location was not significant for the log density of
filtering collectors in the summer (F3,112 = 1.9, P = 0.13) or the fall (F3,83 = 0.4, P = 0.73). There was a
significant main effect of location for both seasons with greater filtering collector density downstream
than upstream (Table 4, Figure 10). There was also a significant main effect of year on filtering collector
density in both seasons, but while post-hoc comparisons found that the density was greater in 2020 than in
2021 in the summer, there were no significant pairwise comparisons for years in the fall (Table 4, Figure 10). Predators. The interaction between year and location was not significant for the log density of predators
in the summer (F3,112 = 1.1, P = 0.32) or the fall (F3,83 = 1.4, P = 0.25). There was a significant main effect
of location for both seasons with greater predator density downstream than upstream (Table 4, Figure 10). There was also a significant main effect of year on predator density only in the summer, with greater
density in 2020 than in 2021 (Table 4, Figure 10).
Shredders. The interaction between year and location was not significant for the log density of shredders
in the summer (F3,112 = 1.2, P = 0.37) or the fall (F3,83 = 0.5, P = 0.70). There was a significant main effect
of location for both seasons with greater shredder density downstream than upstream (Table 4, Figure 10). There was also a significant main effect of year on shredder density, with greater density in 2020 than
in all other years in the summer and greater density in 2020 and 2021 compared to 2019 in the fall (Table
4, Figure 10).
17
Table 4. Results from full-factorial 2-way ANOVA tests comparing the log density (#/m2) of different functional
feeding groups (FFG) with year and location as fixed effects. The years included were 2016-2019, and the
locations were riffles upstream and downstream of Paradise Pond. Values shown include the sum of squares (SS),
degrees of freedom (df), F-value, and P-value. Results with Intercept and Year x Location interaction terms are
type-III models, results with only main effects are type-II models. Significant P-values are in bold red. (D =
Downstream, U = Upstream)
FFG Season Term SS df F-value P-value Post-hoc Comparisons
Gathering Collectors
Summer
Year 164.3 3 85.5 <0.0001 2020 > 2021 > (2018, 2019)
Location 8.6 1 13.4 0.0004 D > U
Residuals 73.7 115
Fall
Year 174.5 3 21.1 <0.0001 2020 > (2018, 2019, 2021)
Location 10.3 1 3.7 0.06
Residuals 237.0 86
Scrapers
Summer
Year 39.9 3 27.8 <0.0001 2020 > 2019 > (2018, 2021)
Location 16.2 1 34.0 <0.0001 D > U
Residuals 55.0 115
Fall
Intercept 46.7 1 31.8 <0.0001
Year 51.9 3 11.8 <0.0001
Location 3.3 1 2.2 0.14
Year x Location 20.5 3 4.7 0.005 D > U in 2020
D in 2020 > everything else
Residuals 121.6 83
Filtering Collectors
Summer
Year 16.8 3 3.3 0.02 2020 > 2021
Location 49.1 1 29.1 <0.0001 D > U
Residuals 193.9 115
Fall
Year 19.0 3 2.8 0.04 none
Location 73.9 1 32.6 <0.0001 D > U
Residuals 194.8 86
Predators
Summer
Year 10.1 3 3.0 0.04 2020 > 2021
Location 34.9 1 30.6 <0.0001 D > U
Residuals 131.3 115
Fall
Year 2.6 3 1.0 0.40
Location 40.5 1 45.9 <0.0001 D > U
Residuals 75.8 86
18
FFG Season Term SS df F-value P-value Post-hoc Comparisons
Shredders
Summer
Year 40.7 3 14.3 <0.0001 2020 > (2018, 2019, 2021)
Location 18.2 1 19.1 <0.0001 D > U
Residuals 109.4 115
Fall
Year 20.1 3 5.1 0.003 (2020, 2021) > 2019
Location 5.3 1 4.0 0.05 D > U
Residuals 112.5 86
19
Figure 9. The effect of location and year on the density of various functional feeding groups for the summer and
the fall in a riffle downstream and upstream of Paradise Pond in the Mill River, Northampton, MA. Points represent
the density (number individuals per meter squared) of a particular functional feeding group from a microhabitat
sample and horizontal lines represent the median of 3-15 samples within a year for a location. The y-axis is log10
scaled and note the maximum is not the same for each graph. See Table 1 for sample sizes and dates and Table 4
for additional statistical results.
20
Discussion
The most important question to answer was whether moving sediment redistribution in Paradise Pond to
the late fall/early winter (November – December) would reduce the impact on the Mill River downstream
of the pond. When sediment was redistributed in Paradise Pond in the summer (July 2016), there was a
measurable impact on the macroinvertebrates downstream of the pond (Pratt 2020). In particular, one
gathering collector, the flathead mayfly Stenacron interpunctatum, increased substantially downstream of
the pond in the fall of 2016 while many other macroinvertebrates decreased. The impact was short-lived
as things went back to the typical situation by fall of 2017, but there was a measurable impact. In this
current report, I analyzed data from 2018-2021. Both 2018 and 2019 samples were long enough after the
July 2016 sediment redistribution and occurred before the winter sediment redistributions in 2019 and
2020 to serve as good “Before” years, while 2020 and 2021 could both be considered “After”. By using
the location upstream of Paradise Pond as a “Control” site and downstream as the “Impact” site, I can test
for an interaction between year and location to help discover any potential impacts from sediment
redistribution.
Importantly, there were few to no interactions between location and year in the samples from 2018-2021
which suggests little to no impact of winter sediment redistribution. The riffle sampling area downstream
of Paradise Pond almost always had greater density of overall macroinvertebrates as well as each of most
of the functional feeding groups, but this was true in all years and suggests a general impact of the
presence of the pond rather than an impact of sediment redistribution. There is evidence from other
studies that other areas just downstream of a small dam are affected by the conditions created by the
presence of an impoundment. Small dams that create impoundments like Paradise Pond can increase
water temperatures downstream and increase the quality of food for some macroinvertebrates from an
increase in organic material that gets released into the area downstream (Singer and Gangloff 2011). In a
review of the effects of small impoundments on stream conditions and macroinvertebrates, most studies
found an increase in water temperature, a decrease in dissolved oxygen, and small increases in nutrients
such as phosphates and nitrates just downstream of small dams (Mbaka and Mwaniki 2015). They also
reported that some studies found an increase, some found a decrease, and others found no change in
macroinvertebrate abundance downstream of small dams. It is not known why macroinvertebrate density
is higher downstream of Paradise Pond, but it seems likely that it is a general effect of Paradise Pond
influencing the temperature of the water and nutrients or food abundance or quality released from the
impoundment. Detailed water quality and environmental testing including the temperature, dissolved
oxygen levels, nutrient load, and food availability need to be done over a long time to see if this is the
case.
When sediment was redistributed in the summer, the difference in macroinvertebrate density downstream
and upstream of Paradise Pond was less pronounced possibly because there was an overall negative effect
of the sediment released from the pond on the macroinvertebrates downstream of Paradise Pond (Pratt
2020). Fine sediment noticeably built up in the downstream location after the July 2016 sediment
redistribution but was mostly cleared out by the next summer sampling time. There was still a noticeable
lack of difference between the upstream and downstream macroinvertebrates in the summer of 2017, but
it may take the macroinvertebrates a summer of reproduction and population growth to recover. By the
fall of 2017 the norm of greater macroinvertebrate density downstream was restored. In contrast to
summer sediment redistribution, when sediment was redistributed in the winter, it is assumed that any
buildup of sediment downstream of the pond generally got cleared out by winter storms while the
macroinvertebrates are more dormant. Because there was no clear detectable impact of winter sediment
redistribution on the macroinvertebrates, it is recommended that moving sediment in the early winter is
the best policy to reduce impacts on the Mill River downstream of Paradise Pond.
21
Another interesting result found in the 2018-2021 samples was the large increase in macroinvertebrate
density and lower substrate stability ratio in 2020 (noting that sometimes 2021 showed a similar but
usually weaker trend). The increase in density in 2020 was also seen for most of the functional feeding
groups but was especially pronounced for the gathering collectors. About 2/3rd of the samples in the
summer of 2020 had over 1000 gathering collectors per meter squared (especially chironomid midges).
The decrease in substrate stability ratio can be explained by the relatively large increase in gathering
collectors relative to the other functional feeding groups. Because this increase in density occurred in the
riffles downstream and upstream of Paradise Pond, this indicates an impact on the Mill River in general
rather than an impact of sediment redistribution or the pond itself. The precipitation and discharge over
the two months prior to sampling in 2020 were fairly typical (Figure 2, Figure 3), but the air temperature
(as measured by Accumulated Degree Days – ADD) was a little cooler before the summer 2020 sampling
(Figure 2). It is possible that there was something about the cooler temperature that benefited the
macroinvertebrates in the river. There may also be other impacts further upstream of both sampling
locations that benefited the macroinvertebrates. Maybe there was increased organic matter as food for the
gathering collectors and other functional feeding groups, but there was similar or lower sand and organic
debris (the preferred substrates for gathering collectors and shredders) in the summer of 2020 downstream
and more of all functional feeding groups downstream. The nutrient and food levels available for each
functional feeding group was not measured directly, so it is hard to know how food availability might
have changed over time. It is important to keep in mind that there was another notable thing about 2020
which was the impact of the Covid-19 pandemic lockdown on humans. Reduced air and water pollution
occurred in 2020 while humans were in pandemic lockdown because of reduced travel and economic
activities (Rume and Islam 2020; Facciolà et al. 2021). It is not clear how changes in human activity
during the lockdown would benefit the macroinvertebrates specifically beyond some improvements in air
and water quality, but it is an interesting idea to test with further research.
Conclusion
The samples taken from upstream and downstream of Paradise Pond in 2018-2021 show no evidence of
impact from winter sediment redistribution in the pond. The results from assessing the impact of the
summer sediment redistribution led to the prediction that winter sediment redistribution should have less
impact. It is reassuring that the data from 2018-2021 support this new strategy of moving sediment in the
winter to lower the impact on macroinvertebrates. The question why the macroinvertebrate density, and
especially the density of gathering collectors, increased so much in 2020 remains. Additional monitoring
will be needed to determine if cooler late spring, early summer temperatures may cause an increase in
macroinvertebrate density in general. Measuring nutrient levels would also be beneficial in helping
interpret results of additional macroinvertebrate monitoring. If over the long-term 2020 stands out as an
exceptional year, it could suggest that the changes in human activity during the Covid-19 pandemic
lockdown may be the ultimate cause of the increased macroinvertebrate density. More research is needed
to understand what changes in human activity caused the increases in macroinvertebrate density.
Literature Cited
Chamberlain S. 2020. rnoaa: “NOAA” weather data from R. https://CRAN.R-project.org/package=rnoaa.
De Cicco LA, Lorenz D, Hirsch RM, Watkins W. 2018. dataRetrieval: R packages for discovering and retrieving
water data available from U.S. federal hydrologic web services. Reston, VA: U.S. Geological Survey.
https://code.usgs.gov/water/dataRetrieval.
22
Facciolà A, Laganà P, Caruso G. 2021. The COVID-19 pandemic and its implications on the environment. Environ
Res. 201:111648. doi:10.1016/j.envres.2021.111648. [accessed 2022 Feb 27].
https://www.sciencedirect.com/science/article/pii/S0013935121009427.
Fox J, Weisberg S. 2019. An R companion to applied regression. Third. Thousand Oaks CA: Sage.
https://socialsciences.mcmaster.ca/jfox/Books/Companion/.
Lenth R. 2022. emmeans: Estimated Marginal Means, aka Least-Squares Means. https://CRAN.R-
project.org/package=emmeans.
Mbaka JG, Mwaniki MW. 2015. A global review of the downstream effects of small impoundments on stream
habitat conditions and macroinvertebrates. Environ Rev. 23(3):257–262. doi:10.1139/er-2014-0080. [accessed 2020
Jul 30]. http://search.ebscohost.com/login.aspx?direct=true&db=asn&AN=109171946&site=ehost-live.
Merritt RW, Cummins KW, Berg MB, editors. 2019. An introduction to the aquatic insects of North America. Fifth
edition. Dubuque, IA: Kendall Hunt Publishing Company.
Peckarsky BL, editor. 1990. Freshwater macroinvertebrates of northeastern North America. Ithaca: Comstock Pub.
Associates.
Pratt M. 2020. Impact of sediment redistribution on macroinvertebrates in the Mill River (2015-2019).
Northampton, MA: Smith College.
R Core Team. 2021. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for
Statistical Computing. http://www.R-project.org/.
Rume T, Islam SMD-U. 2020. Environmental effects of COVID-19 pandemic and potential strategies of
sustainability. Heliyon. 6(9):e04965. doi:10.1016/j.heliyon.2020.e04965. [accessed 2022 Feb 27].
https://www.sciencedirect.com/science/article/pii/S2405844020318089.
Singer EE, Gangloff MM. 2011. Effects of a small dam on freshwater mussel growth in an Alabama (U.S.A.)
stream. Freshw Biol. 56(9):1904–1915. doi:10.1111/j.1365-2427.2011.02608.x. [accessed 2020 Jul 30].
https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1365-2427.2011.02608.x.
Strayer DL, Smith DR. 2003. A guide to sampling freshwater mussel populations. Bethesda, Md: American
Fisheries Society (American Fisheries Society monograph).
Stream Biomonitoring Unit Staff. 2012. Standard operating procedure: biological monitoring of surface waters in
New York State. New York State Department of Environmental Conservation Division of Water.
Acknowledgements
The funding for the research presented in this report was provided by Smith College Facilities Management, the
Department of Biological Sciences, and the Summer Undergraduate Research Fellowship Program. The data were
collected with the help of many Smith College students from the Bio 131 courses in the fall semesters and by
Summer Undergraduate Research Fellowship (SURF) students in the summer of 2018 (Renee Halloran, Lyric
Williams, and Sasha Clapp), 2019 (Sasha Clapp, Andrew Turgeon, Samikshya Dhami), and 2021 (Britney Danials,
Tess Goldmann, Kelly McKenna, Claire Jordan, Lucy Grant). Special thanks to Denise Lello who helped collect
data in her sections of Bio 131 in the fall semesters. During the covid-19 pandemic lockdown in 2020, many people
helped with outside collecting including Henry Renski, Sasha Clapp, Windy Pearman, June Ariens, Claire Jordan,
Claire Lord, Sawyer Blake, and Adriana Grow. Julie Dubreuil helped sort, count, and ID macroinvertebrates from
the summer 2020 samples.
24
APPENDIX
Appendix A. Total number of different taxa found in the Mill River in Northampton, MA for the
locations Downstream and Upstream of Paradise Pond summed for all years and seasons (2018-2021).
Most organisms were identified to genus, but some were only identified to family, order, or even phylum.
FFG = functional feeding group, scr = scraper, cf = collector filterer, cg = collector gatherer, sh =
shredder, prd = predator, prc = piercer.
Taxon Downstream Upstream FFG
Chironimidae* 8,445 6,891 cg
Cheumatopsyche 3,522 523 cf
Stenonema 3,156 1,069 scr
Stenacron 2,151 833 cg
Segmented Worms 2,001 807 cg
Leucrocuta 1,843 1,153 scr
Hydropsyche 1,730 488 cf
Hydropsychidae 1,466 132 cf
Neophylax 1,412 1,165 scr
Chimarra 1,289 82 cf
Stenelmis 1,269 249 scr
Psephenus 933 466 scr
Apatania 716 740 scr
Ephemerella 646 257 cg
Isonychia 476 155 cf
Ferrissia 386 434 scr
Optioservus 366 120 scr
Agapetus 339 255 scr
Antocha 305 251 cg
Taeniopteryx 293 66 sh
Glossosoma 197 118 scr
Pycnopsyche 172 98 sh
Lebertia 142 47 prd
Perlesta 141 58 prd
25
Taxon Downstream Upstream FFG
Acroneuria 133 26 prd
Girardia 128 0 prd
Baetis 122 214 cg
Dannella 122 131 cg
Rhyacophila 121 61 prd
Acentrella 100 149 cg
Plauditus 97 104 cg
Drunella 96 40 scr
Sperchon 96 40 prd
Psychomyia 93 4 cg
Philopotamidae 84 49 cf
Dolophilodes 72 52 cf
Procloeon 71 111 cg
Ceraclea 70 1 cg
Epeorus 65 72 scr
Leuctra 63 15 sh
Caecidotea 58 21 cg
Simulium 52 8 cf
Torrenticola 49 18 prd
Nematomorpha 42 12 prd
Chironimidae* 40 14 prd
Polycentropus 35 22 cf
Sphaerium 35 6 cf
Paraleptophlebia 32 20 cg
Paragnetina 30 11 prd
Chironimidae* 28 110 cf
Goera 24 17 scr
Nigronia 24 13 prd
Sperchonopsis 23 12 prd
26
Taxon Downstream Upstream FFG
Alloperla 21 14 cg
Rhagovelia 20 0 prd
Cernotina 19 28 prd
Glossosomatidae 19 22 scr
Tipulidae 19 0
Gammarus 18 4 cg
Oecetis 18 2 prd
Polycentropodidae 18 15 cf
Hemerodromia 14 1 prd
Helicopsyche 13 4 scr
Clinocera 12 5 prd
Eurylophella 12 8 cg
Prostoma 12 1 prd
Climacia 11 0 prd
Corydalus 10 5 prd
Neureclipsis 10 5 cf
Claassenia 9 1 prd
Culicidae 9 1 cf
Hygrobates 9 3 prd
Simuliidae 9 1 cf
Teloganopsis 9 0 cg
Hemiptera 8 1 prd
Neoleptophlebia 8 2 cg
Beloneuria 7 1 prd
Bivalvia 7 0 cf
Oulimnius 7 4 scr
Attaneuria 6 1 prd
Perlodidae 6 0 prd
Torleya 6 3 cg
27
Taxon Downstream Upstream FFG
Aturus 5 0 prd
Bezzia 5 7 prd
Boyeria 5 2 prd
Mideopsis 5 3 prd
Siphlonurus 5 2 cg
Trichoptera 5 3 cf
Wormaldia 5 2 cf
Argia 4 0 prd
Fingernail Clams 4 0 cf
Hydroptilidae 4 9 scr
Leptoceridae 4 2 cg
Chironimidae* 4 1 sh
Neoperla 4 0 prd
Prosimulium 4 0 cf
Agnetina 3 0 prd
Empididae 3 1 prd
Ephydridae 3 0 cg
Pteronarcys 3 0 sh
Rhithrogena 3 34 scr
Sialis 3 2 prd
Sweltsa 3 1 prd
Testudacarus 3 0 prd
Atractides 2 0 prd
Attenella 2 5 cg
Curculionidae 2 1 sh
Gammaridae 2 1 cg
Habrophlebiodes 2 0 scr
Hansonoperla 2 1 prd
Heterocloeon 2 1 scr
28
Taxon Downstream Upstream FFG
Insecta 2 1
Microcylloepus 2 1 scr
Nematoda 2 0 cg
Ophiogomphus 2 3 prd
Perlidae 2 2 prd
Protzia 2 1 prd
Tipula 2 0 sh
Anthopotamus 1 0 cg
Baetidae 1 2 cg
Blephariceridae 1 1 scr
Carabidae 1 0 prd
Ceratopogonidae 1 2 prd
Collembola 1 0 cg
Diphetor 1 0 cg
Diptera 1 0
Hetaerina 1 0 prd
Hydra 1 0 prd
Hydroptila 1 4 scr
Isoperla 1 1 prd
Iswaeon 1 2 scr
Lara 1 0 sh
leeches 1 3 prd
Lepidoptera 1 0 sh
Leptophlebia 1 0 cg
Limnephilidae 1 0 sh
Mystacides 1 1 cg
Neohermes 1 0 prd
Petrophila 1 7 scr
Physa 1 2 cg
29
Taxon Downstream Upstream FFG
Planorbidae 1 0 scr
Platyhelminthes 1 0 prd
Plecoptera 1 0
Pseudocloeon 1 0 cg
Psychomyiidae 1 0 cg
Sciaridae 1 1
Sigara 1 0 prd
Suwallia 1 1 prd
Trombidiformes 1 2 prd
Blepharicera 0 27 scr
Neotrichia 0 4 scr
Plectrocnemia 0 4 cf
Chironimidae* 0 2 scr
Ameletus 0 1 scr
Atherix 0 1 prd
Caenis 0 1 cg
Centroptilum 0 1 cg
Cinygmula 0 1 scr
Crambidae 0 1 sh
Diploperla 0 1 prd
Liodessus 0 1 prd
Psychoda 0 1 cg
*Note that the family Chironomidae is listed more than once because some were identified lower than
family and were in different functional feeding groups. So here they are summed by family and FFG.
Paradise Pond Dam Upstream Station - USGS gauging station at Clement Street bridge - www.waterdata.usgs.gov/ma/nwis/uv?01171500
Maintenance Activity Log Downstream Station - Smith Mill River & Paradise Pond Monitoring Stations - http://pond.smith.edu
Paradise Pond Sediment Management Plan - DEP file # 246-0725 - Expires Sept. 2024 Paradise Pond, Smith College, Northampton, MA
Waterfront Resource Adaptive Management Plan - DEP file # 246-0715 - Expires 11/12/22 Paradise Pond, Smith College, Northampton, MA
Maintenance activity
1: Erosion & Sediment Controls X 7: Flashboard repair
2: Bank Maintenance & Erosion Repair 8: Exercising of Low-Level Outlet Sluice Gate
3: Rodent Control 9: Dam Inspection
4: Maintenance of existing structures X 10: Vegetation Management Will mow island brush after surveying for invasive plant species
X 5: Partial Drawdowns 11: Pond Maintenance Dredging - SPECIAL ACTIVITY
X 6: Floating Debris Removal X 12: Other Sediment redistribution within Pond (new 5 year permit)
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage
Stage NTU Observations/notes
12/6/21 6:20 AM Check Flow 81.2 101 0% 0.14 1.16 19 37F, Raining
6:28 AM Open gate 100%(initiated operation at 6:28 AM. Full stroke takes 8 minutes)
9:13 AM Check Flow 83.8 220 100% -2.79 1.56 1 38 F, light rain (0.1 in.)
9:21 AM Check Flow 83.8 217 100% -3.16 1.55 3 "
9:45 AM Operate Gate 45%
12:04 PM Check Flow 85.1 114 45% -8.36 1.21 108 39F, rain has stopped
12:20 PM Operate Gate 30%
2:30 PM Site visit Met Jess Applin (Land Stewardship), walked island to survey invasive plants
4:04 PM Operate Gate 36%
4:18 PM Check flow 94.7 114 36% -3.90 1.21 5
4:40 PM Check Flow " 117 36% -3.92 1.22 4 55F, windy
8:56 PM Check Flow 99 119 36% -3.81 1.23 1 48F, light rain
9:00 PM Operate Gate 40%
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/7/21 6:20 AM Check Flow 108 125 40% -4.70 1.25 2 36F, partly sunny (total rain yesterday, 0.15in.)
9:49 AM Check Flow 106 127 40% -4.34 1.26 0 37F, " (conducted drone survey AM)
10:51 AM Check Flow 105 127 40% -4.33 1.26 0 37F, partly sunny
1:03 PM Check Flow 103 126 40% -4.44 1.25 0 37F, overcast
2:44 PM Check Flow 100 126 40% -4.63 1.25 14 38F, mostly cloudy
4:19 PM Check Flow 99 124 40% -4.92 1.25 0 37F, mostly cloudy
5:02 PM Operate Gate 38%
6:26 PM Check Flow 96 121 38% -5.28 1.24 3 32F
7:20 PM Operate Gate 35%
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/8/21 7:26 AM Check Flow 86.4 110 35% -5.69 1.19 2 29F, flurries
8:10 AM Operate Gate 32%
10:55 AM Check Flow 85.1 102 32% -5.26 1.17 0 32F, overcast
12:44 PM Check Flow 85.1 103 32% -5.14 1.17 0 33F, overcast
2:48 AM Check Flow 82.5 103 32% -5.11 1.17 0 34F, overcast
4:38 PM Check Flow 83.8 103 32% -5.18 1.17 0 34F, overcast
7:19 PM Check Flow "" " -5.22 " 0 30F, snowing
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/9/21 6:05 AM Check Flow 83.8 103 32% -5.03 1.17 0 23F, partly cloudy
8:49 AM Check Flow 82.5 102 32% -5.11 1.17 4 28F, mostly sunny
10:12 AM Check Flow 83.8 103 32% -5.17 1.17 0 35F, mostly cloudy
11:28 AM Check Flow 82.5 103 32% -5.23 1.17 0 34F, partly sunny
3:34 PM Check Flow 81.3 101 32% -5.50 1.16 3 33F, overcast
9:00 PM Check Flow 79.9 100 32% -5.80 1.16 0 30F, overcast
9:17 PM Operate Gate 30%
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/10/21 7:20 AM Check Flow 77.4 96 30% -5.04 1.15 10 30F, overcast - Excavator arrived. Had meeting in Pond. Set up laser level, grades
8:30 AM Excavation Started digging diversion canal (East to West) along South side of Pond
11:18 AM Check Flow 77.4 96 30% -5.12 1.15 0 33F, overcast
3:22 PM Check Flow " " " -5.22 1.14 0 38F, overcast
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/11/21 8:07 AM Check Flow 78.6 96 30% -5.17 1.14 3 32F, light rain
8:42 AM " " " " " " 0 32F, light rain
10:18 AM " 79.9 97 " -5.12 1.15 0 33F, "
10:40 AM " " 100 " -5.02 1.16 0 34F, Rain
1:25 PM " 113 105 " -3.01 1.18 8 37F, light rai (0.39 in)
1:32 AM Operate Gate 36%
2:19 PM Check Flow 117 122 " -2.62 1.24 8 37F, drizzle
2:41 PM Operate Gate 100%
3:36 PM Check Flow 121 211 " -3.54 1.52 13 37F, drizzle
4:18 PM " 125 201 " -4.84 1.50 15 38F, light drizzle
4:40 PM Operate Gate 55%
6:42 PM Check Flow 160 164 " -5.70 1.38 25 40F
7:18 PM " 167 168 " -5.01 1.40 15 48F
7:40 PM Operate Gate 80%
10:31 PM Check Flow 167.2 197 " -5.45 1.48 16 60F, rain
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/12/21 6:10 AM Check Flow 184 198 80% -5.29 1.50 8 41F, clearing (rain total for 12/11 - 0.52 in)
7:41 AM " 186.1 203 " -4.60 1.51 5 40F, partly sunny
8:19 AM " 180 203 " -4.51 1.51 3 40F, mostly sunny
5:03 PM " 137 167 " -7.64 1.40 16 38F, clear
5:08 PM Operate Gate 55%
6:33 PM Check Flow 132 158 " -5.95 1.36 9 35F, clear
9:00 PM " 127 154 " -6.02 1.35 2 34F
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/13/21 5:54 AM Check Flow 111 137 55% -8.33 1.29 11 34F, overcast
6:39 AM Operate Gate 45%
7:11 AM Check Flow 111 126 " -6.82 1.25 3 32F, overcast
7:29 AM " 108 128 " -6.66 1.26 2 32F, starting to clear
8:25 AM " 106 131 " -6.23 1.27 0 37F, mostly cloudy
9:00 AM LGP dozer delivered Started shaping upstream area for river diversion
10:45 AM " 105 131 " -6.18 1.27 0 42F, mostly cloudy
12:54 PM Plug River Dozer pushed stockpiled sediment into the river channel, stopping the flow through the
channel and re-directing it to the Canal that was excavated along the South perimeter of
the Pond. After about 15 minutes the water was flowing through the new Canal and into
the river Channel downstream of the Crew House. Stage dropped significantly in the
upstream channel and at the gate until river flow was back to normal after full flow was
passing through the canal. Significant cutting during this time and after as the river flow
continued to shape the canal.
3:58 PM Check Flow 102 119 " -5.58 1.23 136 53F, Clear
4:30 PM Operate Gate 50%
6:49 PM Check Flow 100 130 " -6.27 1.27 106 44F, clear
7:25 PM " 99 101 " -5.56 1.16 69 39F, clear - gate clogging
7:40 PM Operate Gate 100%
8:42 PM Check Flow 99 113 " -606.00 1.21 77 40F, clear
9:09 AM " 99 116 " -5.70 1.22 114 36F, " Dozer operator worked until 5:30 pm
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/14/21 6:25 AM Check Flow 94.7 124 100% -4.94 1.25 145 28F, clear - Pond stage remained stable overnight - turbidity sensor may be covered in
sediment. Bob set up autosampler before dozer plugged the channel.
9:32 AM "91.9 122 " -5.21 1.24 130 44F, clear
10:18 AM "" 122 " -5.27 1.24 130 47F, clear
1:57 PM "90.5 121 " -5.64 1.23 126 51F, clear Drone flew around 1:00 pm. Captured video & still photos of Pond site
4:22 PM "89.1 119 " -5.89 1.23 130 47F, clear Dozer operator worked until 4:30
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/15/21 5:58 AM Check Flow 85.1 114 100% -6.54 1.21 119 27F, mostly cloudy
7:30 AM Clearing logs Worked on clearing logs at sluice gate
11:30 AM "
Remove biggest log at 11:30. Pond stage dropped immediately. Shortly afterward, pond
stage came back up. Just before noon I made a series of adjustments, closing the gate
until it stalled at 14% then opening to 30%, closing again to 10% (stalled) opening to
22%, closing again to 8% (stalled) then opening to 100% at 12:25 pm
12:00 PM
to 12:25 p Operate gate 8% to
100%see above
12:40 PM "40%
2:18 PM Check Flow 83.8 113 " -7.07 1.21 108 42F, overcast
3:00 PM "" 113 " -6.84 1.21 113 42F, overcast
4:19 PM "" 111 " -7.05 1.20 111 41F, overcast
8:20 PM "85.1 112 " -6.89 1.20 109 36F, rain
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/16/21 1:20 AM Check Flow 97.5 122 40% -5.60 1.24 101 39F, raining
2:32 AM "103 125 " -5.09 1.25 103 40F, "
3:00 AM Operate Gate 100%
6:17 AM Check flow 141 178 " -7.74 1.43 97 41F, overcast
4:17 PM "180 218 " -4.35 1.56 88 59F, partly sunny (0.5" rain in 24 hrs.)
7:37 PM "162 209 " -5.42 1.53 91 55F, clear
8:50 PM "158 202 " -6.13 1.51 80 52F, "
9:10 PM Operate Gate 75%
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/17/21 6:18 AM Check Flow 131 173 75% -8.82 1.42 93 53F, clear
7:20 AM Operate Gate 55%Closed gate to 4% (didn't stall) and reversed to 55%. Plunge pool mis 70% sedimented
in.
8:43 AM Check Flow 127 165 " -6.66 1.39 87 50F, clear
10:38 AM "124 160 " -7.03 1.37 89 52F, clear
11:27 AM "122 158 " -7.15 1.36 89 53F, clear
12:15 PM Operate Gate 50%
4:19 PM Check Flow 114 148 " -6.84 1.33 87 52F, thin clouds moving in
5:36 PM "113 146 " -6.98 1.32 85 47F, overcast
5:50 PM Operate Gate 45%
10:27 PM Check Flow 108 144 " -6.10 1.32 76 43F, full moon w/ hazy halo
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/18/21 7:14 AM Check Flow 100 138 45% -7.04 1.29 77 37F, overcast
9:34 AM "99 136 " -7.21 1.29 76 "
11:29 AM Operate Gate 40%
1:18 PM Check Flow 97.5 129 " -6.08 1.26 75 35F, light rain / sleet
4:17 PM "100 134 " -5.45 1.28 76 33F, "
4:47 PM "103 135 " -5.30 1.28 75 "
6:18 PM "111 139 " -4.66 1.30 76 33F, rain
8:31 PM "129 145 " -3.35 1.32 74 34F, rain (0.45 in)
8:50 PM Operate Gate 100%
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/19/21 9:15 AM Check Flow 282 268 100% -0.09 1.71 11
Rain ended at 6am (total rain 0.71 this event) Visited site between 8:00 am & 9:00 am.
Temp sand dike used to divert river into canal is still above water although Pond is full.
It continues to redirect river flow into the canal. Pond is spilling although stage reports -
0.09.
9:19 AM "274 274 " -0.08 1.73 4 36F, clearing (small patches of blue sky)
2:30 PM "229 252 " -0.16 1.67 0 31F, partly sunny
3:36 PM "222 244 " -0.19 1.64 0 "
4:19 PM "212 241 " -0.21 1.63 0 30F, mostly clear
5:18 PM "206 237 " -0.25 1.62 0 29F, fair
8:54 PM "186 234 " -0.63 1.61 0 26F, fair
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/20/21 4:22 AM Check Flow 148 215 100% -2.91 1.55 0 20F, clear
6:22 AM "141 151 " -4.53 1.51 1 19F, "
6:41 AM Operate Gate 70%Viewed canal from boat ramp. Ice on Pond bottom. Large puddle on North side.
7:23 AM "139 185 " -5.35 1.45 24 19F, clear
8:22 AM "139 178 " -6.06 1.43 25 "
8:35 AM Operate Gate 60%
9:05 AM Check Flow " 162 " -6.15 1.38 21
20F, clear - Dozer pushing sediment "dike" into the canal. Grounds removing trees at
damaged road area.
10:26 AM "" 161 " -6.41 1.37 71 26F, mostly sunny - Dozer pushing sediment into canal (high turbidity)
11:20 AM "ice? 160 " -6.46 1.37 47 27F, clear
11:52 AM Operate Gate 55%
4:26 PM Check Flow 125 152 " -6.37 1.34 18 31F, clear
8:32 PM "122 149 " -6.76 1.31 14 20F, fair
8:45 PM Operate gate 50%
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/21/21 6:54 AM Check Flow ice 142 50% -6.23 1.31 2 22F, clear
7:30 AM "" 140 " -6.45 1.30 4 22F, overcast
7:50 AM Operate Gate 45%
9:20 AM Check flow " 131 " -5.83 1.27 33 32F, mostly sunny (dozer pushing sediment into canal)
11:34 AM "141 133 " -5.59 1.28 23 41F, clear
3:02 PM "111 135 " -5.09 1.28 7 41F, partly cloudy
4:41 PM "108 136 " -4.95 1.29 4 33F, mostly cloudy
4:50 PM Operate Gate 48%
8:34 PM Check flow 106 138 " -5.96 1.29 7 28F, high thin clouds, moon visible
9:05 PM Operate Gate 46.5%
Upstream Downstm. Gate Pond Stream
Date Time Specific Activity Flow - cfs Flow - cfs Position Stage Stage NTU Observations/notes
12/22/21 6:57 AM Check Flow 108 140 46.5% -5.22 1.30 2 33F, light rain
8:05 AM Operate gate 60%
8:20 AM Check Flow 121 174 " -4.91 1.42 17 33fF, light rain
10:43 AM "137 174 " -4.93 1.42 15 35F, rain ending (0.41 in from 3:00 am to 10:30 am)
12:36 PM Operate gate 80%
2:28 PM Check flow 186 204 " -4.41 1.52 11 40F, partly sunny (post SedRed survey complete. Pond re-fill starting)
2:32 PM Operate gate 20%
4:50 PM Check Flow 203 79 " -0.52 1.08 4 36F, mostly cloudy
5:30 PM Operate gate 0%Gate closed fully without incident.