Northeastern Naturalist 420 M.K. Shank 22001199 NORTHEASTERN NATURALIST 2V6(o2l). :2462,0 N–4o4. 52 Physicochemical Controls on Spatiotemporal Distribution and Benthic Mat Severity of Didymosphenia geminata in Pine Creek, an Unregulated Watershed in Northern Pennsylvania Matthew K. Shank* Abstract - Didymosphenia geminata (Didymo) is a benthic freshwater diatom that has been globally expanding its range and extracellular stalk production in freshwater ecosystems. Didymo has been observed in reaches downstream of hypolimnetic reservoir releases in the northeastern US since 2007. This study focused on a newly observed (2013) Didymo occurrence in Pine Creek, a highly forested and unregulated watershed in north-central Pennsylvania. Study objectives included comparing contemporary distribution with historical data to provide insight on historical occurrence, quantifying physicochemical controls on Didymo distribution and benthic mat severity, and examining historical changes in water chemistry that might affect habitat suitability. At present, Didymo cellular distribution is limited to upper reaches of Pine Creek where median soluble reactive phosphorus (SRP) is 2.7 μg/L; median SRP was 4.8 μg/L at sites where Didymo was absent. At the epicenter of distribution in Pine Creek where SRP was consistently less than 2 μg/L, increased streamflow flashiness and water temperature were associated with decreased benthic mat severity. My results suggest SRP thresholds for Didymo proliferation may vary depending on whether streams are regulated by reservoirs with hypolimnetic releases. Mann–Kendall trends tests of a ~20-y water chemistry dataset show that orthophosphate and sulfate concentrations decreased while pH increased within Pine Creek, which may have implications for Didymo habitat suitability. Further research is warranted to determine whether improving water quality following the industrial era may facilitate Didymo colonization. Introduction Didymosphenia geminata (Lyngbye) M.Schmidt (Didymo) is a benthic freshwater diatom that was historically restricted to relatively pristine oligotrophic lakes, streams, and rivers in circumboreal regions. Recently, however, Didymo has expanded both its range and ecological tolerances, which has resulted in production of extracellular stalks (i.e., benthic mats) where it was previously undocumented or existed in low abundance (Blanco and Ector 2009, Lavery et al. 2014, Spaulding and Elwell 2007). Didymo has received significant attention in contemporary scientific literature due to its dramatic range expansion, increased frequency of nuisance mat-formation events, and its unusual ability to produce thick benthic mats in low-nutrient environments (Bothwell and Kilroy 2011). Cool, clear, stable flows further facilitate Didymo mat proliferation in riverine habitats (e.g., Bray et *Monitoring and Protection Program, Susquehanna River Basin Commission, 4423 North Front Street, Harrisburg, PA 17110; mshank@srbc.net. Manuscript Editor: Hunter Carrick Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 421 al. 2016, Cullis et al. 2012, Kirkwood et al. 2009), which, in addition to degrading the experience of recreationists, has the potential to alter physical and biological conditions in aquatic ecosystems and negatively impact local economies (Beville et al. 2012, Spaulding and Elwell 2007, Whitton et al. 2009). Didymo appears to have been introduced to New Zealand (Kilroy and Novis 2018, Kilroy and Unwin 2011), but other locations closer to the historic range of this species (i.e., the northeastern US) lack conclusive data concerning the origin of this organism. Didymo has existed on the North American continent for at least 11,000 y, as evidenced by a fossilized specimen collected in British Columbia (Letham et al. 2016). Didymo has been present and stable in abundance in southern Alaska for ~800 y (Pite et al. 2009). Records from the northeastern US include subfossil evidence of Didymo presence on Long Island, NY (Lohman 1939), from the Delaware River near Philadelphia, PA (Boyer 1916, 1927), and from an unknown watershed in Virginia in the late 20th century (Patrick and Reimer 1975). However, Didymo mats were not visually observed in surface waters in the northeastern US until around 2007, and it has since colonized reaches influenced by hypolimnetic reservoir releases in the Delaware River, Youghiogheny River, Savage River, and Gunpowder Falls, among others (Keller et al. 2017, Klauda and Hanna 2016, Shank et al. 2016). For a time, the presence of Didymo seemed limited to settings with hydrologic and thermal regulation provided by reservoirs. Then, in June 2013, Didymo was detected in the middle portion of the mainstem of Pine Creek in Lycoming County near Hamilton Bottom, PA. This initial observation consisted of Didymo cells collected using a plankton-tow net in the water column. In October 2013, Didymo was found attached to the substrate 80 km further upstream on West Branch Pine Creek in Potter County near Galeton, PA (Shank et al. 2016). Pine Creek is an unregulated watershed that is largely forested and considered recreationally and ecologically valuable. The absence of dams in the watershed, and the resulting lack of hydrologic and thermal controls, makes the Didymo colony present in Pine Creek unique in the context of Didymo proliferation in the northeastern US. Didymo is capable of producing nuisance benthic mats up to 20 cm thick characterized by exorbitant growth of extracellular stalks, which may comprise up to 90% of Didymo biomass (Whitton et al. 2009). Recent research has led to key insights into physicochemical variables that govern abundance and distribution of Didymo, including streamflow regulation, geomorphic characteristics, watershed disturbance, soil type, water chemistry, geology, and turbidity. A full synthesis of the associations of Didymo with these variables is beyond the scope of this manuscript, but has been completed elsewhere (i.e., Blanco and Ector 2009, Bray et al. 2016, Cullis et al. 2012, Whitton et al. 2009). Due to the regional nature of this work, I focus on a limited number of physicochemical variables. Foremost is soluble reactive phosphorus (SRP), which at concentrations less than 2 μg/L results in decreased Didymo cell division but stimulates stalk production (Bothwell and Kilroy 2011, Bothwell et al. 2014). James et al. (2015) demonstrated that adding phosphorus (P) significantly Northeastern Naturalist 422 M.K. Shank 2019 Vol. 26, No. 2 decreased Didymo mat biomass in Rapid Creek, SD. The combination of low SRP and cool, clear, stable flows has been shown to increase Didymo mat severity (e.g., Bray et al. 2016; Cullis et al. 2012; Kilroy and Bothwell 2012, 2014; Kirkwood et al. 2009). Although mechanisms for Didymo mat development have been identified, thresholds other than less than 2 μg/L SRP have not been thoroughly quantified and may be regionally specific (Cullis et al. 2012, Kunza et al. 2018). Further, the role of changing environmental conditions has been posited as a possible reason for global Didymo proliferation. Deposition of nitrogen (N) from burning fossil fuels, which in concert with earlier growing seasons and hydrologic shifts due to climate change, could be causing P limitation that favors Didymo over broad scales (Bothwell et al. 2014, Taylor and Bothwell 2014). The improvement of water quality in the northeastern US following the end of industrial and resource extraction eras, coupled with air pollution regulations that reduce stream acidification (Stets et al. 2012, Stoddard et. al 1999) could also be factors in creating suitable habitat for Didymo, which has gone unstudied to date. In this study, I focused on 4 objectives that are central to the ecological understanding of Didymo on a regional scale. The first objective was to compare historical algal data in Pine Creek with contemporary Didymo cellular distribution data to provide insight to the ongoing debate related to the native or non-native status of Didymo. Second, I analyzed SRP concentrations at sites with and without Didymo cellular presence to determine if the less than 2-ug/L threshold for mat proliferation is appropriate in the northeastern US. Third, I developed a novel approach using continuous monitoring of stream temperature, turbidity, and flow at the epicenter of Didymo in Pine Creek to quantify the associations between the measured variables and Didymo mat severity. Lastly, I used Mann–Kendall trend tests to analyze a ~20-y water chemistry, discharge, and air-temperature dataset from a sentinel monitoring location within Pine Creek to determine if environmental conditions are changing, which could have implications for Didymo colonization. Field Site Description The Pine Creek watershed, which has a total drainage area of 2538 km2 and is 89% forested (USGS 2016), is located within north-central Pennsylvania (PA) in Potter, Tioga, and Lycoming counties (Fig. 1). Today, high-quality headwater streams supporting Salvelinus fontinalis (Mitchill) (Brook Trout) and Salmo trutta L. (Brown Trout) contribute to Pine Creek, which flows through the “Pennsylvania Grand Canyon” and is classified as a PA Scenic River. A history of industry and resource extraction during the colonial period caused large-scale degradation of aquatic resources. Beginning around 1800, tanneries and coal mines increased the concentrations of metals and acidity in Pine Creek, while logging reduced the prolific Pinus strobus L. (White Pine) and Tsuga canadensis (L.) Carrière (Eastern Hemlock) forests to bare ground throughout much of the watershed by the early 1900s (Detar and Kristine 2012). The health of Pine Creek has steadily increased during the post-industrial period, but the lasting effects of the millions of acreNortheastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 423 feet of logs transported to lumber mills downstream via Pine Creek is still largely responsible for the geomorphology of Pine Creek and many of its tributaries today (Detar and Kristine 2012). Figure 1. Mean soluble reactive phosphorus (SRP) concentrations in relation to D. geminata presence and cell density at monitoring locations throughout the Pine Creek watershed. The monitoring location labeled “B” is collocated with the West Branch Pine Creek (WPIN) intensive monitoring site. Northeastern Naturalist 424 M.K. Shank 2019 Vol. 26, No. 2 Methods Didymo spatial distribution in relation to soluble reactive phosphorus I completed a systematic search for periphyton/algal assemblage records in the Pine Creek watershed to serve as evidence of Didymo cell presence or absence prior to the first known observation of cells in 2013. I compile d data from floristic surveys (Potapova 2010) and diatom monitoring efforts of agencies (Susquehanna River Basin Commission [SRBC], Pennsylvania Department of Environmental Protection [PADEP], US Environmental Protection Agency [USEPA]) and academic institutions (Academy of Natural Sciences PA [ANSP]). These data were collected using various techniques including natural-substrate scrapes, tow-net deployment in the water column, and artificial tiles. I responded to the appearance of Didymo in Pine Creek with intensive surveys to determine contemporary distribution. I sampled all major tributaries in the watershed (>3rd Strahler stream order) using environmental DNA (eDNA) data collection that employed a plankton drift net with 35-μm mesh and fitted with a 250-μm-mesh pre-filter. The eDNA samples were analyzed using a quantitative real-time polymerase chain reaction procedure (Cary et al. 2014). Specific methods and results of the regional eDNA survey, which included 32 samples from 16 sites in Pine Creek during 2014–2015 are detailed by Keller et al. (2017). In addition to eDNA surveys, I collected 23 periphyton “rock scrape” samples from 15 sites within the watershed. I collected all rock scrape samples using modified Rapid Bioassessment Protocols methods (Barbour et al. 1999), where 11 pieces of natural substrate (cobbles) were removed uniformly from the entire width of the stream at each site. I used a soft brush to disturb periphyton within a 12-cm2 delimiter from each cobble, rinsed the material into a composite sample bottle, and recorded the corresponding sample volume. I used formaldehyde to preserve a 50-mL subsample of the composite sample. In the laboratory, M. Potapova created permanent slides and examined them using microscopy to determine presence or absence of Didymo cells following ANSP (2002) protocol. I collected 10 of the rock-scrape samples from 5 sites along a 39-km longitudinal section of Pine Creek in spring and fall 2015 to examine the spatial and temporal dynamics of Didymo cell density. Specifically, 300–500 valve counts were conducted to determine diatom taxa density. When Didymo was present but relative abundance was low, stratified counts were also conducted to determine Didymo density with increased accuracy. I collected water samples at 12 sites within the Pine Creek watershed (Fig. 1). I sampled each site 7–20 times during 2014–2015, for a total of 106 samples. I measured SRP and total dissovled phosphorus (TDP) concentrations (minimum detection limit = 1.1 μg/L and 7.8 μg/L, respectively) due to the association with Didymo mat formation (Bothwell and Kilroy 2011). I performed a 2-sample t-test in R to test for significant differences (α = 0.05) in SRP concentration at sites based on presence or absence of Didymo cells (R Core Team 2016). Didymo mat severity in relation to physicochemical variables I established an intensive monitoring site on the West Branch of Pine Creek (WPIN), which is the epicenter of the Didymo colony in the watershed (Fig. 1). Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 425 The high-resolution physicochemical data collected at this location were intended to elucidate relationships between physicochemical parameters and Didymo mat severity. I deployed a YSI 6600 data sonde (YSI, Inc., Yellow Springs, OH) to collect pH, specific conductance, water temperature, dissolved oxygen, and turbidity data at 15-min intervals (SRBC 2017). In addition, I characterized underwater photosynthetically active radiation (PAR) with a sensor (single LI-COR 2-pi, wiped; Licor, Lincoln, NE) fastened to the substrate using a rebar stake, and integrated into the existing data sonde. I made 16 Didymo benthic mat severity measurements (standing crop index; SCI) between April 2014 and December 2016 at WPIN. I calculated mean SCI by averaging percent coverage * mat thickness (mm) from 11 pieces of substrate that were removed from the entire width of the stream. SCI is highly correlated with periphyton ash-free dry mass (AFDM), and an SCI threshold of 220 (equivalent to 35 g/m2 AFDM) approximates a threshold for nuisance periphyton (Kilroy and Bothwell 2012). SCI is considered a surrogate for cell density, but best represents mat severity due to the large proportion of extracellular material that can be present even as cell densities remain low. For instance, Didymo cell densities did not exceed 3% of the entire diatom community in several studies, even when extracellular mats were a dominant feature on the substrate (George and Baldigo 2015, Gillis and Lavoie 2014, Spaulding and Elwell 2007). SCI is a widely used standardized metric that allows direct comparisons on local to global scales (Gillis et al. 2018, Kilroy and Bothwell 2012). Onsite streamflow monitoring at WPIN was not available; however, we used a hybrid of streamflow-record extension (Hirsch 1982) and drainage-area ratio (Hirsch 1979) methods to model average daily streamflow (ADF) at WPIN. Specifically, 35 discrete discharge measurements were collected at WPIN since 2010 that spanned a large range of hydrologic conditions. I used these discharge measurements to develop a regression equation with the most spatially correlated, unregulated US Geological Survey (USGS) stream gage for comparison of conditions at WPIN, which was determined to be gage 01544500 on Kettle Creek at Cross Fork, PA. The relationship of log10-transformed observed discharge measurements at WPIN and ADF at Kettle Creek was significant (P < 0.001; R² = 82%). To avoid the potential biases associated with extrapolating high flow outside of the range where measurements were used to derive the regression equation, I used the drainage-area ratio method (Hirsch 1979) to determine all ADF values greater than the maximum observed discharge. Once the time-series hydrograph had been generated, I calculated flow-variability metrics, including the flow coefficient of variation and Richards–Baker flashiness index. The latter measures oscillations relative to the cumulative streamflow to indicate the flashiness of streamflow during the time span (Equation 2 in Baker et al. 2004). I collected channel geometry measurements at the site using modified USGS National Water Quality Assessment Program protocols (Fitzpatrick et al. 1998). I modeled average channel-bed shear stress for a streamflow event that would result in mobility of coarse gravel (32-mm particle size) and defined the streamflow value that corresponded to this level of shear stress. Northeastern Naturalist 426 M.K. Shank 2019 Vol. 26, No. 2 Water chemistry sampling at WPIN resulted in 19 observations of SRP between April 2014 and January 2016. I used these discrete measurements to predict continuous concentrations. Although total P concentrations in streams are known to correlate with streamflow (Ruzycki et al. 2014), it is less clear whether SRP concentrations are likewise correlated. Streamflow was significantly positively correlated with SRP, which and had the strongest relationship with streamflow of all continuous variables collected (Pearson r = 0.648, P = 0.007). I employed the composite method (function loadComp) in the loadflex package (Appling et. al 2015, Lorenz et al. 2015) in the R software environment (R Core Team 2016) to estimate daily SRP concentrations based on empirical observations and streamflow. Once I completed data collection and generated time-series datasets, I summarized onditions prior to each Didymo mat severity sample for temporal periods of 7 d, 15 d, 30 d, 60 d, 90 d, and 120 d. These antecedent conditions summarized flow variability metrics and number of days where flow exceeded critical shear stress values. I also calculated mean water temperature and turbidity. I selected the antecedent period of each variable that explained the greatest amount of variation (adjusted R2) to serve as predictor variables in a series of linear regression models to describe variation in mat severity at WPIN. Subsequently, I employed the information theoretic approach (Burnham and Anderson 2002) to determine if streamflow, stream temperature, turbidity, or all variables best explained variation in mat severity. These candidate models reflected a priori hypotheses that cool, clear, stable streamflows facilitate Didymo mat proliferation (e.g., Bray et al. 2016, Cullis et al. 2012, Kirkwood et al. 2009). I chose Akiake’s information criterion corrected for small sample size (AICc) to determine the level of support for the 4 competing models. The model with the lowest AICc value had the most support, while competing models had a Δi (AICi - AICmin) ~ 2. Mat severity (SCI) was square-root transformed in all models to ensure that homoscedasticity was attained, residuals were approximately normally distributed, and individual observations did not have undue influence on the relationship. All analyses were performed in R (R Core Team 2016). Changes in water chemistry I obtained water chemistry data from a long-term monitoring site on the mainstem of Pine Creek near Waterville, PA (PADEP Water Quality Network site 410), where data had been collected approximately bimonthly from 1998 to 2016. This site is located on Pine Creek 62 km downstream from the nearest location where Didymo has been observed on the substrate (Fig. 1); however, this site represents a sentinel monitoring station where conditions are representative of the entire drainage and where the longest-term dataset is available. I retrieved and cleaned data from STORET (USEPA 2017), and selected parameters for analysis that have been shown to correlate with Didymo cell presence and/or mat proliferation. Total aluminum, iron, N, orthophosphate, P, sulfate, and pH contained relatively complete records, as did other surrogates for water quality, including total dissolved solids and specific conductance. Spearman tests indicated that all of these Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 427 parameters were significantly correlated with average daily stream flow (α = 0.05), which I obtained from the USGS streamflow gage 01549700 on Pine Creek in Waterville, PA (Fig. 1). Flow-adjusted Mann-Kendall trend tests were completed on these parameters to determine direction and significance of monotonic trends during the study period. Also, I aggregated average daily streamflow and air temperature for Williamsport, PA (37 km away from WQN 410; Pennsylvania State Climatologist 2017) to monthly values and examined that data using correlated seasonally adjusted Mann–Kendall tests. I considered trends significant at α = 0.05; all analyses of trends were completed using the trend package in R (Pohlert 2016, R Core Team 2016). Results Didymo spatial distribution in relation to soluble reactive phosphorus I compiled and/or examined a total of 112 algal records that pre-dated the first observation of Didymo in Pine Creek, which were collected between September 2008 and October 2012. None of these samples contained any evidence of the presence of Didymo cells or mats in the Pine Creek watershed prior to June 2013, including 3 archived samples from June and July 2012 that were within the reach where Didymo was observed attached to the substrate in 2013 (Fig. 2a). The intensive eDNA and microscopy monitoring effort enacted in response to the first visual observation of Didymo resulted in the collection of an additional 88 samples from June 2013 Figure 2. Didymosphenia geminata cellular presence and absence (a) before and (b) after the first observation in the Pine Creek watershed in June 2013. Northeastern Naturalist 428 M.K. Shank 2019 Vol. 26, No. 2 to December 2016. This effort revealed that Didymo cells were present beginning in West Branch Pine Creek and, from there, directly downstream throughout Pine Creek. There was no evidence that Didymo cells were present in any of the tributaries where sampling was conducted. Didymo cells were detected in the water column up to 101 km downstream from the upstream-most observation, while cells were present on the substrate for 39 km (Fig. 2b). I visually observed Didymo mats on the substrate for ~6 km downstream of the most upstream colony at WPIN. Notably, Didymo cells were not present in 7 samples (2 eDNA and 5 microscopy) that were collected at the most upstream site on West Branch Pine Creek between November 2013 and May 2015. Didymo mats were then observed attached to the substrate at this site in November 2015, indicating possible upstream range expansion. Didymo mat severity is highest in upstream reaches, including West Branch Pine Creek and upper Pine Creek. This finding is consistent with Didymo cell density, which peaks near the mouth of West Branch Pine Creek and then falls precipitously as tributaries with higher SRP concentrations and waste-water treatment discharges enter Pine Creek (Fig. 1). SRP had nearly tripled (from 2.0 ± 0.9 to 5.7 ± 2.1 μg/L; mean ± 1 standard deviation) 39 km downstream from the site with the highest Didymo cell density, while cell density decreased from 3130 ± 3730 to 6.6 ± 9.4 cells/cm2 (Figs. 1, 3a). Similarly, the presence or absence of Didymo cells appears to be associated with the concentration of SRP in tributaries throughout the Pine Creek watershed. Didymo has colonized West Branch Pine Creek, which contains the lowest SRP concentration of all tributaries in the watershed. All other tributaries, where Didymo was absent, have mean SRP concentrations >4 μg/L except for Little Pine Creek (Fig. 1). SRP is significantly lower at sites where Didymo was present (3.8 ± 4.7 μg/L, n = 46) compared with sites where Didymo was absent (6.6 ± 6.1 μg/L, n = 60) (2-sample t-test, P = 0.004). The SRP means for both groups were inflated due to high concentrations in a tributary site with no Didymo (Marsh Creek) and at a site on Pine Creek containing Didymo directly downstream of a waste-water treatment discharge during summer low-flow conditions (Fig. 3b). Therefore, median values, which were 2.7 μg/L and 4.8 μg/L at sites where Didymo cells were absent and present, respectively, may be more representative of the association between SRP and Didymo presence. Didymo mat severity in relation to physicochemical variables At the epicenter of Didymo colonization (WPIN), where I conducted intensive data collection, West Branch Pine Creek is a 4th order stream with a highly forested watershed and slightly acidic pH, which represents a typical montane stream in north-central Pennsylvania. There is a small (17.4 ha) impoundment on an upstream tributary (Lyman Run; Fig. 1) that impounds ~20% of the watershed. This impoundment is a simple spill-over design that does not have flood-storage capability and thus does not result in the dampening of high flows during runoff events or in lowflow augmentation. SRP at WPIN is among the lowest in the region (1.94 ± 0.95 μg/L), with the highest values (max = 4.0 μg/L) observed in the summer months Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 429 during low-flow conditions. SRP modeling revealed consistently low values, with concentrations less than 2.0 μg/L and less than 4.0 μg/L during 50% and 96% of days, respectively (Fig. 4). Other water chemistry parameters are well within ranges supporting Didymo presence elsewhere (Table 1; Whitton et al. 2009). A total of 16 observations of Didymo mat severity (SCI) were made at WPIN between April 2014 and December 2016. The SCI values varied from 0.0 to 283.5, with only 1 observation above the nuisance periphyton threshold of 220 (Kilroy and Bothwell 2012). Streamflow showed typical patterns of high seasonal flow events in Figure 3. (a) Scatter plot of mean soluble reactive phosphorus (SRP) concentration and Didymosphenia geminata (Didymo) cell density at 5 sites on Pine Creek and 7 tributary sites. (b) Box plots of SRP concentration at sites where Didymo cells were present or absent. Black triangles indicates mean values, bold lines indicate medians, lower and upper extent of boxes indicate 1st and 3rd quartiles, respectively. Dashed line indicates SRP detection limit of 1.1 μg/L. Continuous axes are square-root transformed to allow for clear visualization of SRP and cell density at low values. Northeastern Naturalist 430 M.K. Shank 2019 Vol. 26, No. 2 Figure 4. Frequency distribution of soluble reactive phosphorus concentrations at West Branch Pine Creek (WPIN) where Didymosphenia geminata mat formation was intensively monitored. Shaded polygons indicate 25th, 50th, and 75th percentiles. Table 1. Physical and chemical characteristics of a continuously monitored site on West Branch Pine Creek (WPIN). Values are shown as mean ± 1 standard deviation. Variable Value Units Didymosphenia geminata standing crop index 83.7 ± 87.5 - Physical Drainage size 182.1 km2 Proportion impounded 20.8 % Forested area 97.0 % Slope 0.6 % Average daily streamflow 3.3 ± 3.4 m3/s Substrate size 159.7 ± 171.7 mm Width 14.0 ± 2.8 m Annual water temperature 8.6 ± 6.5 °C Summer water temperature 17.5 ± 2.8 °C Dissolved oxygen 12.6 ± 2.1 mg/L Chemical Soluble reactive phosphorus 1.94 ± 0.95 μg/L Total dissolved phosphorus* 7.85 ± 0.20 μg/L Total phosphorus 18.62 ± 19.06 μg/L Nitrate 0.50 ± 0.16 mg/L Sulfate 6.36 ± 0.78 mg/L Calcium 4.64 ± 0.59 mg/L Total organic carbon 1.15 ± 0.82 mg/L Turbidity 3.62 ± 5.45 NTU Specific conductance 0.043 ± 0.005 ms/cm pH 6.82 ± 0.35 SU *Most observations are below the detection limit of 7.80 μg/l. Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 431 winter and spring, with prolonged low flows in summer and fall in 2014 and 2016. Flows were flashier throughout the summer of 2015, with numerous precipitation runoff events resulting in a flashier hydrograph. Average daily water temperature showed a seasonal pattern, with temperatures near 0 °C during winter months, and occasionally exceeding 20 °C in the summer months (Fig. 5D). Stream turbidity was low throughout the study period, with periodic peaks coinciding with high intensity precipitation events that resulted in elevated stream flows (Fig. 5). Figure 5. Plot of (a) Didymosphenia geminata observed standing crop index (SCI) values. (b) average daily soluble reactive phosphorus, (c) streamflow, (d) water temperature, and (e) stream turbidity, from January 2014 to December 2016 at the intensive monitoring site on West Branch Pine Creek (WPIN). Northeastern Naturalist 432 M.K. Shank 2019 Vol. 26, No. 2 Richards–Baker flashiness index, mean water temperature, and mean turbidity in 7-d, 30-d, and 15-d antecedent windows described 25.2%, 14.1%, and 8.9% of variation in mat severity, respectively. These variables in their respective antecedent windows had the highest adjusted R2 values when compared to other time periods of each respective variable. ΔAICc values indicated that the streamflow flashiness model received the most support, while the stream temperature model was considered competing and there was limited support for the stream turbidity model and the full model containing all variables (Table 2). As streamflow flashiness, temperature, and turbidity increased, mat severity decreased (Fig. 6). Changes in water chemistry Water chemistry within Pine Creek has changed significantly since 1998. Orthophosphate and sulfate show decreasing trends, while pH and specific conductance have increased (Table 3, Fig. 7). I observed insignificant trends in the remainder of the water chemistry parameters examined. Streamflow and air temperature showed no overall trend, but air temperature showed significant increasing trends in the months of July and October. Discussion Didymo spatial distribution in relation to soluble reactive phosphorus The intensive data collection that followed the first observation of Didymo in Pine Creek coupled with the historic data available in the watershed combine to provide insight into which physicochemical variables influence the spatiotemporal Figure 6. Relationship between Didymosphenia geminata mat severity and the 3 antecedent variables representing a priori hypotheses: (a) Richards–Baker flashiness index, (b) mean stream temperature, and (c) mean stream turbidity. Length of antecedent window is shown in square brackets under x axis label. Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 433 Table 2. Results of regression model-selection process reflecting a prior hypotheses on the effect of streamflow flashiness, temperature, and turbidity on the square-root transformed standing crop index (SCI) of D. geminata at the intensive monitoring site on West Branch Pine Creek (WPIN). Length of antecedent window is shown in square brackets in the parameter estimates columns. Parameter estimates ± standard error are shown. Parameter estimates Richards–Baker Mean water Mean turbidity Intercept flashiness temp. (ºC) (NTU) Model Type df P Multiple R2 Adjusted R2 AICc ΔAICc (sqrt SCI) index (7 days) (30 days) (15 days) Streamflow 14 0.03 30.2% 25.2% 100.07 0.00 13.06 ± 2.63 -34.69 ± 14.11 Stream temperature 14 0.08 19.9% 14.1% 102.27 2.20 11.88 ± 2.78 -0.48 ± 0.26 Stream turbidity 14 0.14 15.0% 8.9% 103.22 3.15 8.21 ± 1.41 -0.017 ± 0.011 Full model 12 0.11 38.3% 22.9% 107.22 7.15 13.76 ± 2.91 -20.55 ± 18.98 -0.25 ± 0.31 -0.012 ± 0.011 Northeastern Naturalist 434 M.K. Shank 2019 Vol. 26, No. 2 Table 3. Results of Mann–Kendall trend tests for water chemistry variables, streamflow, and air temperature at the sentinel monitoring location on Pine Creek (WQN 410). The date range, mean, standard deviation (SD), and number of ob servations (n) are included for each parameter. Trend test type/ Parameter Water chemistryparameter Covariate Z P Trend direction Date range mean ± SD n Mann-Kendall Flow Adjusted Aluminum Total (mg/L) Flow (m3/s) 1.275 0.202 Positive 10/13/98–9/19/16 0.28 ± 0.75 180 Iron total (mg/L) Flow (m3/s) 1.593 0.111 Positive 10/13/98–9/19/16 0.43 ± 1.20 180 Nitrogen total (mg/L) Flow (m3/s) -0.867 0.386 Negative 4/10/02–9/19/16 0.37 ± 0.25 157 Orthophosphate total (mg/L) Flow (m3/s) -2.810 0.005 Negative** 4/10/02–9/19/16 0.007 ± 0.005 158 Phosphorus total (mg/L) Flow (m3/s) -0.705 0.481 Negative 10/13/98–9/19/16 0.025 ± 0.037 133 pH Flow (m3/s) 4.827 0.000 Positive** 10/13/98–9/19/16 7.41 ± 0.35 181 Specific conductance (ms/cm) Flow (m3/s) 2.051 0.040 Positive* 10/13/98–9/19/16 0.096 ± 00.30 180 Sulfate total (mg/L) Flow (m3/s) -2.290 0.022 Negative* 10/13/98–9/19/16 15.05 ± 5.24 179 Total dissolved solids (mg/L) Flow (m3/s) 0.369 0.713 Positive 10/13/98–9/19/16 70.97 ± 24.10 174 Correlated seasonal Mann–Kendall Streamflow (m3/s) Season (month) 0.200 0.860 Positive 1/1/98–12/31/16 38.27 ± 54.76 6940 Air temperature (°C) Season (month) 1.000 0.332 Positive 1/1/98–12/31/16 10.73 ± 9.76 6893 *Statistically significant trends at the 5% significance level. **Statistically significant trends at the 1% significance level. Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 435 distribution of Didymo throughout the watershed. Data indicate that SRP constrains the spatial distribution, as only streams with long-term SRP median concentrations of ≤2.7 ug/L are currently colonized by Didymo cells, which is near the threshold suggested by Bothwell et al. (2014). Similarly, SRP concentrations appear to limit Didymo cell density in West Branch Pine Creek and the mainstem of Pine Creek. Tributaries and point-source discharges increase the concentration of SRP in Pine Creek as it flows downstream, which coincides a decrease in Didymo cell density (Fig. 1). My observation of sites where SRP concentrations were low but Didymo was not present suggests that other physicochemical parameters may play a role in habitat suitability. For instance, Didymo was absent in Little Pine Creek where mean SRP was 2.2 ± 1.7 μg/L (Fig. 1). However, this site is downstream of a riverine impoundment with an epilimnetic release, which in combination with its large Figure 7. Scatter plots of physicochemical variables at the sentinel monitoring location on Pine Creek (WQN 410) that were included in trend analysis. Northeastern Naturalist 436 M.K. Shank 2019 Vol. 26, No. 2 watershed, causes warmer summer temperatures (20.3 ± 3.1 °C) than preferred by Didymo (Lindstrøm and Skulberg 2008, Whitton et al. 2009). Also, Little Pine Creek has elevated sulfate concentrations (28.4 ± 13.1 mg/L), which reflect a legacy of coal mining in the watershed. Lindstrøm and Skulberg (2008) suggested that the sulfate concentration must be >2.5 mg/L to constitute suitable habitat for Didymo. Additionally, there is a positive association with sulfate concentrations and Didymo presence in Sierra Nevada streams (Rost et al. 2011). This evidence indicates that minimum thresholds of sulfate are necessary to support Didymo stalk production; however, no maximum preferred limit could be found in the literature. Sulfate is a water chemistry constituent associated with abandoned mine discharge in Pennsylvania, which also often accompanies high levels of acidity and controls solubility of other metals (Cravotta 2008). It is possible that there is a sulfate concentration, which, if exceeded, causes deleterious effects to Didymo. Data suggest the combination of elevated temperature and sulfate may cause conditions in Little Pine Creek to be unsuitable for Didymo. Didymo mat severity in relation to physicochemical variables Didymo mat proliferation occurred at WPIN, where long-term SRP concentrations averaged 1.94 ± 0.95 μg/L. At this site where SRP was consistently low, concentrations preceding sampling events did not effectively describe variation in benthic mats. Instead, an a priori model-selection approach indicated that streamflow flashiness and temperature in 7-d and 30-d antecedent windows, respectively, were the best variables to predict mat severity. Notably, Didymo mats were not present at WPIN when the 7-d Richards–Baker flashiness index values exceeded 0.25 and when 30-d mean stream temperature exceeded 15 ºC. These variables have previously been associated with conceptual models for Didymo mat proliferation and persistence (Cullis et al. 2012); however, determining the appropriate time period of each variable and quantifying relationships with mat severity are novel contributions of this research. Although previous studies have documented associations between physicochemical variables and Didymo mat severity when conducting instantaneous sample collections (e.g., Jackson et al. 2016), results herein indicate the temporally variable nature of Didymo mats makes repeated sampling efforts and/or continuous monitoring advantageous. The Richards–Baker flashiness index outperformed other measures of flashiness (streamflow coefficient of variation) and streamflow magnitude (i.e., number of flow events with shear stress capable of mobilizing 32-mm diameter sediment particles) in describing variation of Didymo mat severity. This finding suggests that variation in streamflow with concomitant variation in depth and velocity may be more influential in controlling benthic mats compared with high-magnitude discharge events that result in bed mobility and occur less frequently, which corroborates observations from streams in the Catskill Mountains of New York (Richardson et al. 2014). Typical Didymo habitat downstream of hypolimnetic reservoir releases, where streamflows and temperatures are stable, are often managed as sport fisheries for salmonids. Consequently, concerns surround manipulation of dam outflows as a Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 437 tool for Didymo management because they may conflict with successful salmonid egg incubation in gravel and/or young-of-year recruitment and survival (e.g., Klauda and Hanna 2016). However, the results herein indicate streamflow magnitude is less important than streamflow variation in controlling benthic mats. A nuanced approach of varying streamflow releases from dams, while staying within bounds of acceptable salmonid habitat, may represent a potential approach for maintaining salmonid populations while mitigating negative ecological and recreational effects of nuisance Didymo mats. However, consideration must be given to condition of Didymo mats and sediment content of streamflows because high streamflow events downstream of large reservoirs often lack suspended sediments, which accelerate degradation of stalked diatom mats (Cullis et al. 2013, Walling and Fang 2003). My results from north-central Pennsylvania indicate that in unregulated streams that experience seasonal flashy flows and water temperature variation, SRP is the main factor controlling the distribution of Didymo cells. Other studies examining similar variables in regulated streams have suggested that thresholds supporting Didymo mats may differ between regulated and unregulated reaches. For instance, in Rost and Fritsen’s (2014) study of 2 streams in the Sierra Nevadas, phosphate concentrations were similar but Didymo only colonized sites downstream of an impoundment with a hypolimnetic release. Further evidence is present from Didymo proliferation in New York and Maryland tailwater streams, where streamflows were regulated and temperatures rarely exceeded 16 ºC, but median SRP varied from 4 μg/L to 10 μg/L (Klauda and Hanna 2016, Shank et al. 2016, Silldorff and Swann 2013). This finding suggests that stable discharges and cool temperatures may allow Didymo proliferation at higher SRP concentrations. Regulated rivers, specifically below hypolimnetic release reservoirs, represent late-successional habitat (on the r–K continuum) that is preferred by Didymo due to the stability in temperature and flow (Floder and Kilroy 2009, Kirkwood et al. 2009). The stasis of temperature and flow does not result in removal of Didymo mats from the substrate, which may allow Didymo to proliferate even when SRP concentrations are not below the threshold of 2 μg/L suggested by Bothwell et al. (2014). Keller et al. (2017) found no association between Didymo cell presence and water chemistry in the mid-Atlantic region; however, 4 out of 7 sites where Didymo was present were situated downstream of hypolimnetic releases, which, in combination with instantaneous sample collection methods, may complicate threshold determination. Conversely, flashy flow environments result in higher and more variable shear stress and bed mobility, which thoroughly disturb stream habitats. These environments more frequently remove Didymo-dominated biofilms from the substrate and simultaneously open up niches for opportunistic algal species to colonize, thereby requiring extremely low SRP concentrations that favor Didymo. More research is necessary to explore the potential for different SRP thresholds in regulated versus unregulated reaches. Pine Creek is unregulated, flashy, variable in temperature, and has quickly increasing SRP concentrations moving downstream. As a result, late-successional habitat is only present during natural periods of hydrologic stability and temperature Northeastern Naturalist 438 M.K. Shank 2019 Vol. 26, No. 2 stasis. Thus, Pine Creek represents a habitat with marginal suitability for Didymo, where mat formation is only episodic due to frequent disturbances in the form of scouring and elevated water temperatures. Didymo mat formation in Pine Creek is similar to Esopus Creek in New York, where stable flows and cool temperatures were positively associated with Didymo cell density. However, the stabilizing influence of an inter-basin aqueduct (Shandaken Portal) promotes Didymo proliferation for a distance downstream in Esopus Creek (George and Baldigo 2015). Pine Creek, however, is completely unregulated, and Didymo is only able to affect large reaches when climatic conditions are favorable. For instance, in the spring of 2012, visible Didymo mats occupied a 160-km section of the Delaware River (DRBC 2014). Aside from occasions when all conditions remain suitable for an extended period, we do not expect Didymo to pervade the substrate of Pine Creek, which raises questions about the extent of impacts to the functioning of the aquatic ecosystem. Studies have shown evidence of alterations to algal, macroinvertebrate, and fish communities due to Didymo mat proliferation (i.e., Gillis and Lavoie 2014, James and Chipps 2016, Kilroy et al. 2009). However, studies on Esopus Creek in New York did not detect any severe impacts to algal, macroinvertebrate, or fish assemblages (George and Baldigo 2015). Research should be conducted to investigate whether infrequent, episodic Didymo mat coverage has negatively affected the stream ecology of Pine Creek. Changes in water chemistry We recognize that conventional sampling protocols for microorganisms, such as Didymo, are considered inadequate to definitively determine which taxa are truly absent at any given point in time (Taylor and Bothwell 2014). Consequently, it is difficult to determine if Didymo was historically absent based on previously collected samples and/or data (Lavery et al. 2014, Spaulding and Elwell 2007). It is perplexing, however, that when similar methods were employed to collect samples after the first observation of Didymo in 2013 in the Pine Creek watershed, detection of cells was easily accomplished. Data collected herein indicate range expansion in an upstream direction on West Pine Creek during the 2014–2016 period. Similarly, George and Baldigo (2015) found evidence of upstream range expansion in Esopus Creek during 2009–2010. This rapid appearance and variable distribution may suggest that Didymo populations are not in equilibrium and could undergo additional distributional changes (Keller et al. 2017). Based on the absence of contemporary evidence of Didymo presence in Pine Creek despite a systematic search, the most likely explanation for Didymo colonization is that Didymo is a native invader that may be recolonizing its former range (Richardson et al. 2014). Regional fossilized evidence of historical Didymo presence (i.e., Boyer 1916, 1927; Lohman 1939; Patrick and Reimer 1975) indicates that Didymo inhabited adjacent watersheds historically, but Didymo mats were not reported in the northeastern US prior to the early 2000s (Spaulding and Elwell 2007). It seems plausible that changing water chemistry may have facilitated colonization of Didymo in Pine Creek. Genetic analysis of Didymo from the Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 439 mid-Atlantic indicates close relation to lineages from across the globe (Keller et al. 2017). Degradation of water quality accompanied resource extraction and industrial activities in the watershed in the 19th and early 20th centuries. Robust datasets to quantify this degradation do not exist from within the Pine Creek watershed, but historic accounts describe acidic conditions from tanneries and mines and sedimentation from widespread deforestation (Detar and Kristine 2012, PADER 1977, Ross 1991). Widespread water quality degradation was documented throughout the Ohio River and Lake Erie drainages in western Pennsylvania in the late 19th and early 20th centuries (Lewis 1906, Ortmann 1909). Although somewhat qualitative, these inventories present reliable evidence of acute levels of pollution in surface waters sufficient to extirpate many aquatic taxa. The same historical industrial activities and poor sanitation that caused degradation in western Pennsylvania drainages also occurred in the Pine Creek watershed and can be assumed to have had similar effects. Quantitative datasets from Pine Creek that span more recent times and strengthen statistical analyses show that water chemistry parameters with implications for Didymo habitat suitability are indeed changing. Since 1998, sulfate concentrations declined while pH increased, which is consistent with observations from surface waters across North America following implementation of the Clean Air Act, indicating recovery from acidification resulting from acid deposition (Stoddard et al. 1999). Orthophosphate shows a decreasing trend in Pine Creek, which is consistent with broader trends in the northeastern US, as increased deposition of N and decreased inputs of P have increased N:P ratios since 1970 (Hale et al. 2013). The continued burning of fossil fuels and modern agricultural practices are expected to increase N concentrations, while climate-induced shifts in the timing of snowmelt and growing season decrease P inputs to rivers, which could potentially be responsible for further changes in water chemistry (Bothwell et al. 2014). These converging conditions may be creating suitable Didymo habitat in streams and rivers like Pine Creek, where it did not exist decades ago. Current limitations and future research The ecological paradox of Didymo occurrence—formation is restricted to nutrient poor, often pristine settings—presents a unique paradigm for the management of freshwater ecosystems. Moreover, the current debate surrounding its native or non-native status further confounds our understanding of how policy makers should respond appropriately (Elwell et al. 2014). Maryland has opted to follow the “precautionary principle” and ban felt-sole waders and act to prevent the spread of this microorganism (Klauda and Hanna 2016). Taylor and Bothwell (2014) appropriately note that decontamination programs intended to prevent the spread of microorganisms are mostly ineffective, and Vermont has repealed its ban on felt soles (VTFWD 2016). The possible reasons for the increased nuisance behavior of Didymo, including global climate change and anthropogenic alterations to nutrient ratios (Bothwell et al. 2014), have not been fully supported by empirical studies (Bergey and Spaulding 2015, Keller et al. 2017, Kunza et al. 2018). Until this issue is resolved, the lack of clarity will make the implementation of proper management actions problematic. Additional paleolimnological studies to place the Northeastern Naturalist 440 M.K. Shank 2019 Vol. 26, No. 2 proliferation of Didymo into environmental context, similar to that of Lavery et al. (2014), and regional studies to assess the distribution and association of Didymo with physicochemical variables (e.g., Keller et al. 2017) should be encouraged to provide clarity. Bergey and Spaulding (2015) are likely right: Didymo proliferation and its environmental context is complicated. There may be many elements at play, including those supported by Bothwell et al. (2014) regarding global climate change and anthropogenic alteration of nutrients and nutrient ratios. Additional factors that are improving water quality conditions, such as mine-drainage abatement and enhanced industrial mitigation practices, may also affect Didymo proliferation. Considerations must also include the possibility for human transport of this microorganism. Future research should attempt to provide a holistic view of Didymo and its human and environmental context so that resource managers are able to make informed decisions. Conclusion This study highlights the influence of SRP on Didymo distribution. The longterm median SRP concentration of sites that were colonized by Didymo cells was 2.7 μg/L, whereas sites in which those cells were absent had with a median SRP concentration of 4.8 μg/L. Compared to mean values, long-term median values of SRP may be more representative of central tendency due to the reduced effect of high values observed during summer low flows. At WPIN, where mean SRP was consistently less than 2μg/L, streamflow flashiness and temperature were associated with Didymo mat severity. The examination of mat severity temporally using continuously monitored physicochemical variables served as a novel approach that effectively elucidated dynamics of Didymo mat proliferation in an unregulated stream. Future research should further study the SRP thresholds that support Didymo colonization and mat formation in unregulated streams compared to late-successional habitat present below hypolimnetic releases, examine the ecological implications for episodic Didymo mat coverage in unregulated streams with marginal habitat, and investigate the historic occurrence of Didymo in the northeastern US using paleolimnological methods. Acknowledgments Marina Potapova (Drexel University) performed numerous identifications of algal samples as well as the counts and provided helpful suggestions and reviews on this manuscript. Daniel Spooner, Kelly Maloney, Dale Honeyfield (USGS), James Shallenberger (SRBC), and M. Potapova helped to secure funding and establish the scope of this research. Jeffrey Zimmerman (SRBC) performed invaluable GIS work that aided analysis and clear presentation of results. Jeffrey Butt and his colleagues at PADEP are credited with first discovery of D. geminata in Pine Creek, which led to this research. Steven Keller (University of Vermont), Bob Hilderbrand and Regina Trott (University of Maryland), and Jason Cessna (MDDNR) enabled eDNA sampling. Robert Volkmar (University of Duquesne, retired) performed algal identifications. Elizabeth Costanzo Kreger (Tioga County Conservation District) provided historical water-quality data from Pine Creek. Dawn Hintz (SRBC) coordinated continuous monitoring resources, and Graham Markowitz (SRBC) provided streamflow-modeling Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 441 support. Katie Kline (University of Maryland Center for Environmental Science) facilitated laboratory analysis of much of the water chemistry data included herein. Andrew Leakey, Matthew Elsasser, Aaron Henning, Blake Maurer, Luanne Steffy, David Haklar, and John Balay (SRBC) assisted with data collection. Pennsylvania Sea Grant and the Susquehanna River Basin Commission provided funding for this study. Literature Cited Academy of Natural Sciences Pennsylvania (ANSP). 2002. Protocols for the analysis of algal samples collected as part of the US Geological Survey National Water-Quality Assessment Program. Report Number 02-06. The Patrick Center for Environmental Research–Phycology Section, Philadelphia, PA. 132 pp. Available online at http://water. usgs.gov/nawqa/protocols/algprotocol/algprotocol.pdf. Accessed 30 May 2017. Appling, A.P., M.C. Leon, and W.H. McDowell. 2015. loadflex: Watershed flux estimates. R package version 1.0.1. Available online at https://github.com/McDowellLab/loadflex. Models and Tools for. Accessed 15 June 2017. Baker, D.B., R.P. Richards, T.T. Loftus, and J.W. Kramer, 2004. A new flashiness index: Characteristics and applications to midwestern rivers and streams. Journal of the American Water Resources Association 40:503–522. Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid bioassessment protocols for use in streams and wadeable rivers: Periphyton, benthic macroinvertebrates, and fish, 2nd Edition. EPA 841-B-99-002. US Environmental Protection Agency, Office of Water, Washington, DC. 337 pp. Bergey, E.A., and S.A. Spaulding. 2015. Didymosphenia: It’s more complicated. Bioscience 65(3):225. Beville, S.T., G.N. Kerr, and K.F. Hughey. 2012. Valuing impacts of the invasive alga Didymosphenia geminata on recreational angling. Ecological Economics 82:1–10. Blanco, S., and L. Ector. 2009. Distribution, ecology, and nuisance effects of the freshwater invasive diatom Didymosphenia geminata (Lyngbye) M. Schmidt: A literature review. Nova Hedwigia 88:347–422. Bothwell, M., and C. Kilroy. 2011. Phosphorus limitation of the freshwater benthic diatom Didymosphenia geminata determined by the frequency of dividing cells. Freshwater Biology 56:565–578. Bothwell, M., B. Taylor, and C. Kilroy. 2014. The D. geminata story: The role of low dissolved phosphorus in the formation of Didymosphenia geminata blooms. Diatom Research 29(3):229–236. Boyer, C.S. 1916. The Diatomaceae of Philadelphia and Vicinity. J.B. Lippincott Company, Philadelphia, PA. 143 pp. Boyer, C.S. 1927. Synopsis of North American Diatomaceae. Part II. Naviculatae, Surirellatae. Proceedings of the Academy of Natural Sciences. Philadelphia Supplement 79:229–583. Bray, J., J.S. Harding, C. Kilroy, P. Broady, and P. Gerbeaux. 2016. Physicochemical predictors of the invasive diatom Didymosphenia geminata at multiple spatial scales in New Zealand rivers. Aquatic Ecology 50(1):1–14. Burnham, K.R., and D.R. Anderson. 2002. Model Selection and Multimodel Inference. Springer, New York, NY. 488 pp. Cary, S.C., K.J. Coyne, A. Rueckert, S.A. Wood, S. Kelly, C.E.C. Gemmill, C. Vieglais, and B.J. Hicks. 2014. Development and validation of a quantitative PCR assay for the early detection and monitoring of the invasive diatom Didymosphenia geminata. Harmful Algae 36:63–70. Northeastern Naturalist 442 M.K. Shank 2019 Vol. 26, No. 2 Cravotta, C.A., III. 2008. Dissolved metals and associated constituents in abandoned coalmine discharges, Pennsylvania, USA. Part 2: Geochemical controls on constituent concentrations. Applied Geochemistry 23:203–226. Cullis, J., C. Gillis, M. Bothwell, C. Kilroy, A. Packman, and M. Hassan. 2012. A conceptual model for the blooming behavior and persistence of the benthic mat-forming diatom Didymosphenia geminata in oligotrophic streams. Journal of Geophysical Research 117(G2):G00N03. Cullis, J.D.S., J.P. Crimaldi, and D.M. McKnight. 2013. Hydrodynamic shear removal of the nuisance stalk-forming diatom Didymosphenia geminata. Limnology and Oceanography Fluids and Environments 3:256–268. Delaware River Basin Commission (DRBC). 2014. D. geminata, aka “Rock Snot”, discovered in the non-tidal Delaware River. Available online at http://www.state.nj.us/drbc/ home/spotlight/approved/20120531_D. geminata.html. Accessed 29 March 2018. Detar, J., and D. Kristine. 2012. Pine Creek fisheries management plan. Pennsylvania Fish and Boat Commission, Pleasant Gap, PA. 59 pp. Available online at: http://www.fishandboat. com/Fish/Fisheries/PineCreek/Pages/default.aspx. Accessed 20 February 2017. Elwell, L.C., C.A. Gillis, L.A. Kunza, and M.D. Modley. 2014. Management challenges of Didymosphenia geminata. Diatom Research 29(3):303–305. Fitzpatrick, F.A., I.R. Waite, P.J. D’Arconte, M.R. Meador, M.A. Maupin, and M.E. Gurtz. 1998. Revised methods for characterizing stream habitat in the National Water-Quality Assessment Program. US Geological Survey, Open–File Report 98–4052. Raleigh, NC. 77 pp. Floder, S., and C. Kilroy. 2009. Didymosphenia geminata (Protista, Bacillariophyceae) invasion, resistance of native periphyton communities, and implications for dispersal and management. Biodiversity and Conservation 18:3809–3824. George, S.D., and B.P. Baldigo. 2015. Didymosphenia geminata in the upper Esopus Creek: Current status, variability, and controlling factors. PLoS ONE 10(7):e0130558. Gillis, C.A., and I. Lavoie. 2014. A preliminary assessment of the effects of Didymosphenia geminata nuisance growths on the structure and diversity of diatom assemblages of the Restigouche River basin, Quebec, Canada. Diatom Research 29(3): 281–292. Gillis, C.A., A.J. Dugdale, and N.E. Bergeron. 2018. Effect of discharge and habitat type on the occurrence and severity of Didymosphenia geminata mats in the Restigouche River, eastern Canada. Hydroecology 11(5):e1959. Hale, R.L., J.H. Hoover, W.M. Wollheim, and C.J. Vorosmarty. 2013. History of nutrient inputs to the northeastern United States, 1930–2000. Global Biogeochemical Cycles 27:578–591. Hirsch, R.M. 1979. An evaluation of some record-extension techniques. Water Resources Research 15(6):1781–1790. Hirsch, R.M. 1982. A comparison of four streamflow record-extension techniques. Water Resources Research 18(4):1081–1088. Jackson, L.J., L. Corbett, and G. Scrimgeour. 2016. Environmental constraints on Didymosphenia geminata occurrence and bloom formation in Canadian Rocky Mountain lotic systems. Canadian Journal of Fisheries and Aquatic Sciences 73(6):964–972. James D.A., and S.R. Chipps. 2016. Influence of Didymosphenia geminata blooms on prey composition and associated diet and growth of Brown Trout. Transactions of the American Fisheries Society 145:195–205. James, D.A., M.L. Bothwell, S.R. Chipps, and J. Carreiro. 2015. Use of phosphorus to reduce blooms of the benthic diatom Didymosphenia geminata in an oligotrophic stream. Freshwater Science 34(4):1272–1281. Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 443 Keller, S.R., R.H. Hilderbrand, M.K. Shank, and M. Potapova, 2017. Environmental DNA genetic monitoring of the nuisance freshwater diatom Didymosphenia geminata in Eastern North American streams. Diversity and Distributions 23(4):381–393. Kilroy, C., and M. Bothwell. 2012. Didymosphenia geminata growth rates and bloom formation in relation to ambient dissolved phosphorus concentration. Freshwater Biology 57:641–653. Kilroy, C., and M. Bothwell. 2014. Attachment and short-term stalk development of Didymosphenia geminata: Effects of light, temperature, and nutrients. Diatom Research 29(3):237–248. Kilroy, C. and P. Novis. 2018. Is Didymosphenia geminata an introduced species in New Zealand? Evidence from trends in water chemistry, and chloroplast DNA. Ecology and Evolution 8(2):904–919. Kilroy, C., and M. Unwin. 2011. The arrival and spread of the bloom-forming, freshwater diatom, Didymosphenia geminata, in New Zealand. Aquatic Invasions 6(3):249–262. Kilroy C., S.T. Larned, and B.J.F. Biggs. 2009. The non–indigenous diatom Didymosphenia geminata alters benthic communities in New Zealand rivers. Freshwater Biology 54(9):1990–2002. Kirkwood, A., L. Jackson, and E. McCauley. 2009. Are dams hotspots for Didymosphenia geminata blooms? Freshwater Biology 54:1856–1863. Klauda, R.J., and K.V. Hanna. 2016. Didymosphenia geminata infestation in Maryland: Reactions and responses by the Maryland Department of Natural Resources 2008–2014. Maryland Department of Natural Resources, Annapolis, MD. 36 pp. Kunza, L.A. C.A. Gillis, J.Z. Haueter, J.N. Murdock, and J.M. O’Brien. 2018. Declining phosphorus as a potential driver for the onset of Didymosphenia geminata mats in North American rivers. River Research and Applications 34:1105–1110. DOI:10.1002/ rra.3332. Lavery, J., J. Kurek, K. Ruhland, C. Gillis, M. Pisaric, and J. Smol. 2014. Exploring the environmental context of recent Didymosphenia geminata proliferation in Gaspesie, Quebec, using paleolimnology. Canadian Journal of Fisheries and Aquatic Sciences 71(4):616–626. Letham, B., A. Martindale, R. Macdonald, E. Guiry, J. Jones, and K.M. Ames. 2016. Postglacial relative sea-level history of the Prince Rupert area, British Columbia, Canada. Quaternary Science Reviews 153:156–191. Lewis, S.J. 1906. Quality of water in the upper Ohio River Basin and at Erie, PA. US Geological Survey Water-Supply and Irrigation Paper 161. US Government Printing Office, Washington, DC. 114 pp. Available online at https://pubs.usgs.gov/wsp/0161/report.pdf. Lindstrøm, E.A., and O. Skulberg. 2008. Didymosphenia geminata: A native diatom species of Norwegian rivers coexisting with the Atlantic salmon. Pp. 35–40, In M.L. Bothwell and S.A. Spaulding (Eds.). Proceedings of the 2007 International Workshop on Didymosphenia geminata. Canadian Technical Report on Fisheries and Aquatic Sciences Fisheries and Oceans Canada, Nanimo, BC, Canada. 96 pp. Lohman, K.E. 1939. Pleistocene diatoms from Long Island, New York. US Geological Survey Professional Paper 189:229–235. Lorenz, D., R. Runkel, and L. De Cicco. 2015. rloadest: River Load Estimation. R package version 0.4.3. Ortmann, A.E. 1909. The destruction of fresh-water fauna in Western Pennsylvania. Proceedings of the American Philosophical Society 48(191):90–110. Northeastern Naturalist 444 M.K. Shank 2019 Vol. 26, No. 2 Patrick, R., and C.W. Reimer. 1975. Diatoms of the United States, Volume 1. Academy of Natural Sciences of Philadelphia. Monograph Number 13, Philadelphia, PA. 688 pp. Pennsylvania Department of Environmental Resources (PADER). 1977. Pine Creek watershed report on mine-drainage pollution. Bureau of Water Quality Management, Harrisburg, PA. 25 pp. Pennsylvania State Climatologist. 2017. Daily weather data for Williamsport, PA. Available online at http://climate.psu.edu/data/ida/index.php?t=3andx=faa_dailyandid=KIPT. Accessed 25 February 2017. Pite, D.P, K.A. Lane, A.K. Hermann, S.A. Spaulding, and B.P. Finney. 2009. Historical abundance and morphology of Didymosphenia species in Naknek Lake, Alaska. Acta Botanica Croatica 68(2):183–197. Pohlert, T. 2016. trend: Non-parametric trend tests and change-point detection. R package version 0.2.0. Available online at https://CRAN.R–project.org/package=trend. Accessed 17 February 2017. Potapova, M. 2010. Diatoms of north-central Pennsylvania. Final Report for Grant Agreement WRCP–08286. Academy of Natural Sciences, Philadelphia, PA. 40 pp. R Core Team, 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available at https://www.R–project.org/. Accessed 1 February 2017. Richardson, D.C., I.A. Oleksy, T.J. Hoellein, D.B. Arscott, C.A. Gibson, and S.M. Root. 2014. Habitat characteristics, temporal variability, and macroinvertebrate communities associated with a mat-forming nuisance diatom (Didymosphenia geminata) in Catskill mountain streams, New York. Aquatic Sciences 76:553–564. Ross, R.M. 1991. The effect of acid mine-drainage and related sources of pollution on aquatic macroinvertebrate life in Pine Creek. Pine Creek Headwaters Protection Group, Wellsboro, PA. 14 pp. Rost, A.L., and C.H. Fritsen. 2014. Influence of a tributary stream on benthic communities in a Didymosphenia geminata-impacted stream in the Sierra Nevada, USA. Diatom Research 29(3):249–257. Rost, A.L., C.H. Fritsen, and C.J. Davis. 2011. Distribution of freshwater diatom Didymosphenia geminata in streams in the Sierra Nevada, USA, in relation to water chemistry and bedrock geology. Hydrobiologia 665:157–167. Ruzycki, E.M., R.P. Axler, G.E. Host, J.R. Henneck, and N.R. Will. 2014. Estimating sediment and nutrient loads in four western Lake Superior Streams. Journal of the American Water Resources Association 50(5):1138–1154. Shank, M.K., M. Potapova, K. Maloney, D. Honeyfield, and D.E. Spooner. 2016. Didymosphenia geminata in Pennsylvania: An investigation of current and historic distribution, habitat suitability, and nutrient content. Susquehanna River Basin Commission, Harrisburg, PA. 58 pp. Available online at http://seagrant.psu.edu/topics/invasive-species/ research/didymosphenia-geminata-pennsylvania-investigation-current-and. Accessed 1 February 2017. Silldorff, E.L., and M.M. Swann. 2013. Observational and experimental work with Didymosphenia geminata in the lower Delaware River: A report to the Pennsylvania SeaGrant. Delaware River Basin Commission, West Trenton, NJ. 42 pp. Available online at http:// seagrant.psu.edu/topics/invasive–species/research/nutrient–thresholds–didymosphenia– geminata–blooms–and–stalk. Accessed 22 January 2017. Spaulding, S., and L. Elwell. 2007. Increase in nuisance blooms and geographic expansion of the freshwater diatom Didymosphenia geminata. Open-file Report 2007-1425. US Geological Survey, Reston, VA. 42 pp. Northeastern Naturalist Vol. 26, No. 2 M.K. Shank 2019 445 Stets, E.G., V.J. Kelly, W. Broussard, T.E. Smith, and C.G. Crawford. 2012. Century-scale perspective on water quality in selected river basins of the conterminous United States. Scientific Investigations Report 2012–5225. US Geological Survey, Reston, VA. 120 pp. Stoddard, J.L., D.S Jeffries, A. Lukewille, T.A. Clair, P.J. Dillon, C.T. Driscoll, M. Forsius, M. Johannessen, J.S. Kahl, J.H. Kellogg, A. Kemp, J. Mannio, D.T. Monteith, P.S. Murdoch, S. Patrick, A. Rebsdorf, B.L. Skjelkvale, M.P. Stainton, T. Traaen, H. van Dam, K.E. Webster, J. Wieting, and A. Wilander. 1999. Regional trends in aquatic recovery from acidification in North America and Europe. Nature 401:575–578. Susquehanna River Basin Commission (SRBC). 2017. Standard operating procedure: Remote water quality monitoring network. Susquehanna River Basin Commission, Harrisburg, PA. 61 pp. Available online at http://mdw.srbc.net/remotewaterquality/methods. htm. Accessed4 August 2017. Taylor, B.W., and M. L. Bothwell. 2014. The origin of invasive microorganisms matters for science, policy, and management: The case of Didymosphenia geminata. BioScience 64(6):531–538. US Environmental Protection Agency (USEPA). 2017. Storage and retrieval and waterquality exchange (STORET). Available online at http://www.epa.gov/storet/. Accessed 25 February 2017. US Geological Survey (USGS). 2016. The StreamStats program for Pennsylvania. Available online at https://water.usgs.gov/osw/streamstats/pennsylvania.html. Accessed 14 December 2016. Vermont Fish and Wildlife Department (VTFWD). 2016. Felt-soled wader ban is repealed. Available online at http://vtfishandwildlife.hosted.civiclive.com/cms/One.aspx?portalId =73163&pageId=4278738. Accessed 29 March 2018. Walling, D.E., and D. Fang. 2003. Recent trends in the suspended sediment loads of the world’s rivers. Global and Planetary Change 39:111–126. Whitton, B., N. Ellwood, and B. Kawecka. 2009. Biology of the freshwater diatom Didymosphenia: A review. Hydrobiologia 630:1–37.