Methods
Overview of timeline, system, and sampling site designation
The dam on the Maple River was originally part of an old hydroelectric plant and consisted of an earthen embankment, a concrete spillway, and old concrete housing for the electric turbines. The earthen embankment was 366 meters long and approximately 4.5 meters high. The top drawing spillway created the 42.4-acre17.2 ha Lake Kathleen with a maximum depth of 3.7 meters.
The University of Michigan Biological Station (UMBS) began a long-term standardized sampling protocol on the Maple River in 2012. The Maple River is part of the Cheboygan River Watershed and is 42.49 km long. The Maple River contains two branches: the West Branch and the East Branch (Figure 1). The West Branch Maple River begins at the Pleasantview Swamp (45.53°N, -84.92°W), while the East Branch begins as water flows outward from Douglas Lake (44.92°N, -84.45°W) (Godby 2010). The Main Stem of the Maple River is the area where the branches converge, located at the Maple River Dam at Lake Kathleen in Emmet County, Michigan (Godby 2010) (Figure 2). The Maple River was sampled once annually during researchers’ field seasons in either May orand June from 2012 until 2021. Dam removal began in August of 2018 and was completed by the following spring. Movement of onshore sediment and sand continued through the summer of 2020.
Five standardized sampling sites were chosen based on locations where a priori ideas of possible increased sediment and discharge from the dam removal would impact the river ecology (Figure 1). Two of the selected sites were upstream (ranging from 700 to 1000 m upstream from the dam) of the deltas into Lake Kathleen (45.53°N, -84.77°W), representing the East and West branches' most downstream points under dammed conditions. These points, East Branch (EB 200) and West Branch (WB 100), were the transition points between the river and lake habitats and were indicative of increased sedimentation from changes in river velocity. Three of the sampling sites were located on the Main Stem of the Maple River (referred to here as MB 7, MB 14, and MB 31) downstream of Lake Kathleen. In total, 62 sampling points were identified including 8 locations on the East Branch, 5 locations on the West Branch, 6 locations on Lake Kathleen, and 43 locations on the Main Branch. Only the five targeted locations (EB 200, WB 100, MB 7, 14, and 31) were sampled in each yearcontinuously for all variables over the 11-year period. Site numbers are used for identification only and not indicative of any measurement taken. The GPS coordinates for all locations are located in the Supplementary Data.
Main Branch sites 7 and 14 offered the most optimal information on how dam removal impacts various habitat types because of the close proximity of these sites to the previous dam. MB 7 was closest to the previous dam site and occurred at the first bend with both an erosional and depositional zone. This site offers context of dam removal for shallower habitat types as riffles and sandbars were prominent here. Next, MB 14 was located in the first run of the mainstem and offers a direct contrast to MB 7 in terms of water depth and water velocity due to differences in substrate type and composition. Further downstream, MB 31 was beyond the dam removal's direct effects and was sampled as a control site. Thus, these sites allowed for the collection of data with variable habitat types and dam proximity.
After the determination of the sample locations, pre-sampling data, in the form of a site description, was collected for each site every time an area was sampled. First, a geographical description of each location was recorded. The geographical description included notes on the direction to the site in enough detail that allowed researchers to return to each specific site. Similarly, the surrounding land use for all sites was also recorded.
Abiotic measurements
Abiotic measurements consisted of two different categories of measurements. The first category included physical characteristics of the stream, flow, instream water chemistry, and physiographic characteristics (bankfull width, discharge, flow velocity, pH, turbidity, conductivity, and light penetration). A second category of abiotic measurements included analytical water chemistry measuring for various forms of nitrogen and phosphorus as well as chlorophyll a.
Physical Characteristics
The physical characteristics of the stream at each sample site were taken. To calculate the discharge of water, the wetted width of the stream was subdivided into different cells (ranging from 7 to 10 cells) of equivalent sizes. The velocity of the water was recorded at 60% of the total depth within each individual cell using a digital flow meter (Hach FH950). Discharge (l/s) of the stream was calculated by multiplying the velocity (cm/s) of the water, the width of the cell, and the depth of the cell. The quotient produced from this formula was the discharge for that specific cell. The discharge of all of the cells were added together to produce total discharge at that point in the stream.
Instream Water Chemistry
On-site water quality measurements included measuring the temperature (°C), pH, conductivity (μS/cm), dissolved oxygen (DO) (mg/L), and turbidity (NTU). These measurements were taken using a sonde hydrolab (Eureka Manta2 sub3) that was calibrated before heading to the field. Because some parameters may be variable throughout stream habitats, all measurements were taken in each of the cells used to measure discharge and were averaged across the measurements. All water quality measurements were taken either upstream of areas that were disturbed by sampling.
In addition to water quality parameters that were measured in the field, some parameters required that water samples be collected and sent to an analytical laboratory. These parameters were used to aid in interpretation. The parameters analyzed in the analytical laboratory included ammonia, nitrate, phosphate, total phosphorus and nitrogen, silica dioxide, and chlorophyll a. To collect samples, stream water (250 mL) was collected using acid-washed plastic sample bottles and syringes for analytical chemical analysis. All equipment was rinsed in stream water at each sample location to prevent any possible cross-contamination. Water samples were filtered (Millipore 0.45 m MCE Membrane) before collection. The assembly was removed from the syringe before filling, and a small amount of stream water exuded before reattachment to protect the integrity of the sample collecting on the micropore filter membrane. The micropore filter paper was placed in aluminum foil for chlorophyll-A analysis after sample collection. Finally, samples were taken to the Alfred H. Stockard Lakeside Laboratory at UMBS where chemical analyses were performed to quantify the above-listed parameters.
Biotic Measurements
Benthic Macroinvertebrate Community
Benthic macroinvertebrates were collected from streams using a 0.25 m2 surber sampler. Samples were collected from four different habitat sub-types (woody debris, gravel, sand, and macrophytes) along the stream at each of the sample locations. The surber sampler was placed in each habitat sub-type for a period of two minutes with the riverbed, debris, or sand being destructively sampled for that time period. After the sampling period concluded, the mesh bags were carried onto the bank, and the contents of the bags were emptied into an enamel pan. Each enamel pan was sampled for 30 person minutes total, where the sampling time was equally divided among each person (i.e., two people each sampled for 15 minutes, three people each sampled for 10 minutes, etc.). Any macroinvertebrates that were found were placed into a 250 mL bottle of 70% ethanol for preservation in order to identify specimens at a later time.
Post-processing of the specimens took place at the main campus of the University of Michigan Biological Station or at the Laboratory for Sensory Ecology at Bowling Green State University. Macroinvertebrates were counted and then placed into one of five feeding guilds (Merritt et al. 2017). The feeding guilds were shredders, collecting gatherers, filtering gatherers, scrapers, and predators, and the abundance of organisms in each guild were tallied. Finally, different ratios based upon the counts for each guild were calculated, and interpretation of these results followed Hauer and Lamberti (2017). These ratios included an autotrophy to heterotrophy index, a CPOM to FPOM index, an FPOM in transport to storage index, an index of channel stability, and an index of top-down predator control (Hauer and Lamberti 2017).
Statistical Analysis
Data conditioning and treatment followed the steps outlined in Zuur et al. (2009) PCA analysis. The first step in this process was creating Cleveland dotcharts to examine potential outliers within the dataset. The second step included a collinearity analysis performed between all of the independent variables to remove any chance of inflating p values. None of the variables were considered outliers or had significant collinearity with other dependent variables. Finally, histograms, q-q plots, and Shapiro-Wilk tests of normality were used to examine the underlying distribution of response variables. For those variables that were not normal, “BestNormalized” was run to determine which data transformation was likely to produce the best normalized data set (Peterson 2021). A principal component analysis (PCA) was performed using the FactorMineR, and factoextra functions in the R (Husson et al. 2020). All quantitative data loaded in the PCA was were standardized using z-scores (Z = (X − µ)/σ). Sampling site and sample was used as the identifying qualitative variable. Only dimensions that explained more than 10% of the variance were subsequently graphed for interpretation.
