Lake Scugog is a large (68 km²), shallow (mean depth = 1.4 m) headwater reservoir located in southern Ontario, Canada (Figure 1). Twelve sampling sites were monitored from May—September over 3 years (2017–2019) with site depths spanning 1–3 m across the two basins. Sampling sites spanned the two lake basins with equal coverage. Sampling locations were intentionally selected to reflect historical fish spawning locations in order to determine if N. obtusa establishment may impact fish habitat quality. Sites 5 and 9 were inaccessible each year of sampling during September due to low water levels. DO measurements were taken at 0.5 m depth intervals at each of 12 sites using a YSI 6 series multiparameter probe (YSI Inc., Yellowsprings, Ohio, United States). Benthic DO concentrations were measured 0.5 m above the sediment-water interface to minimize sediment disturbance during the reading. The sonde was suspended at each depth until the reading stabilized, before recording the value.

After DO readings were completed, sediment cores were taken at each site using an NLA Gravity Corer (Hoskin Scientific, Burlington, Canada). The top 10 cm of each sediment core were sectioned into acid washed tubes and stored on ice until returning to the lab on the same day. Tubes were acid washed to remove residual nutrients and were well rinsed with Milli-Q water to remove residual acid. To extract pore-water, sediment samples were centrifuged at 10,000 g for 10 min. Samples were stored at −20°C and analysed using the modified method of Murphy and Riley (1962), developed by the Ontario Ministry of Environment (1983). Nitellopsis obtusa was collected following the rake method of Ginn (2011). Briefly, collection comprised three rake tosses that extended to the sediment at each site. Samples were transported back to the laboratory, washed in reverse osmosis water, and identified following charophytes of North America (Wood, 1967), and status and strategy for Starry Stonewort [Nitellopsis obtusa (Desv. In Loisel.) J. Groves] management (Hackett et al., 2014).

Statistical analyses were conducted in R version 4.0.0 (R Core Team, 2019) using the packages ggpubr (Kassambara, 2020), and ggplot2 (Wickham, 2016). To ensure visual accessibility of our oxygen profiles, an accessible palette was applied from the RColorBrewer package (Neuwirth, 2014). DO profiles were averaged for each sampling date and separated by years for sites with and without N. obtusa presence. To assess differences in benthic dissolved oxygen between sites with and without N. obtusa, a Welch’s t-test was used due to unequal sample sizes. Linear regressions were performed with N. obtusa biomass (dry-weight in grams) as the independent predictor variable of benthic oxygen concentrations (mg L⁻¹) and sediment pore-water SRP (μg L⁻¹).

Studies documenting connections between aquatic macrophytes and sediment anoxia are limited (Atapaththu et al., 2018). However, dense stands of macrophytes, including those of Characeae and Myriophyllum spicatum, are often described to regulate dissolved oxygen profiles (Frodge et al., 1990; Cardinale et al., 1997; Unmuth et al., 2000; Kufel and Kufel, 2002). Consistently over the 3-year study, sites with N. obtusa had lower DO concentrations throughout the water column (Figures 2A,C,E). Generally, sites without N. obtusa, frequently dominated by M. spicatum, had DO concentrations in the supersaturated range, and maintained higher concentrations throughout the water column than sites with N. obtusa (Figures 2B,D,F). Benthic DO concentrations were significantly different (Welch’s t-test p-value < 0.001) between sites with and without N. obtusa (Figure 3).

Not only was low benthic DO associated with N. obtusa presence, but N. obtusa biomass was a negative explanatory variable for benthic DO (R ² = 0.59) (Figure 4A) and positive explanatory variable for sediment pore-water SRP (R ² = 0.90) (Figure 4B). With phosphorus loading from the watershed apparently decreasing (Kawartha Conservation, 2010), our results are likely a reflection of the mobilization of legacy phosphorus within the ecosystem. These results clearly show a distinct water quality profile associated with N. obtusa in Lake Scugog, where N. obtusa appears to be a driver of benthic hypoxia-anoxia; a necessary condition for internal loading of bioavailable phosphorus from sediments.

Brainard and Schulz (2017) suggested that when dense benthic mats of N. obtusa undergo senescence, nutrients are released from sediments. Despite N. obtusa not undergoing senescence in our study, N. obtusa biomass was a very strong predictor of sediment pore-water SRP concentration. Based on the negative relationship with DO, it is likely that N. obtusa is indirectly facilitating sediment phosphorus release into the water column. Internal phosphorus loading of bioavailable phosphorus is regarded as a primary driver of cyanobacterial blooms (Bormans et al., 2016). Thus, this prominent association between N. obtusa and internal phosphorus loading in Lake Scugog infers an indirect faciliatory role in Microcystis bloom development.

Unlike other macrophytes, the life-cycle and reproductive strategies of some Characeae are dependent on hypoxic-anoxic events at the sediment water interface. Generally, Characeae that reproduce through vegetative propagules rely on viable propagule banks within the sediments (Migula, 1897). Propagules can persist in sediments for extended periods of time, however, when buried deeper than 2 cm, the potential to germinate is lost (Bonis and Grillas, 2002). This is attributed to the necessity of hypoxic-anoxic conditions at the sediment water interface to initiate germination (Bonis and Grillas, 2002). The strong association of reduced benthic DO with increasing N. obtusa abundance in our study suggests that N. obtusa may alter local habitat conditions to promote propagule bank germination. This positive-feedback system may explain why N. obtusa is initially delayed in becoming dominant in the macrophyte community because it takes time for there to be sufficient biomass to induce hypoxic-anoxic conditions for propagules to germinate.

Dispersal and establishment of N. obtusa within invaded regions is poorly understood (Larkin et al., 2018). Generally, macrophytes can be dispersed through viable propagules and vegetative fragments (Reynolds et al., 2015; Green, 2016). Charophyte propagules have been known to be dispersed through epizoochory and endozoochory (Bonis and Grillas, 2002). However, there is mounting evidence that the majority of N. obtusa dispersal throughout invaded regions is occurring through watercraft movement and deployment (Sleith et al., 2015; Midwood et al., 2016; Harrow-Lyle and Kirkwood, 2021). Given our results, which document habitat alterations conducive of propagule germination, dispersal and establishment must be areas of focus going forward. Thus, implementing management programs that target boat launches may be effective in preventing N. obtusa spread, while also allowing early detection for new populations within invaded regions.

Given the strong inference from our results that N. obtusa is altering the biogeochemical fate of oxygen and phosphorus in Lake Scugog, we propose that N. obtusa is acting like an ecosystem engineer of internal biogeochemical processes. An ecosystem engineer is defined as a non-human organism that has direct or indirect effects on ecosystem processes, resulting in significant alterations to ecosystem structure and function. With increasing biomass and dominance in aquatic weed beds, N. obtusa may be reducing water-column mixing and exchange with atmospheric oxygen. Although depleted near-bed DO is known to drive internal phosphorus loading in lakes, there is also the negative impact that water column hypoxia-anoxia can have on biota. Studies are now emerging that show the negative effects of N. obtusa on aquatic communities (e.g., Brainard and Schulz, 2017; Harrow-Lyle T. J. and Kirkwood A. E., 2020; Ginn et al., 2021), but much remains unknown about the effects of N. obtusa on the aquatic food web, especially fish. Considering the extent of hypoxia to anoxia reported here in a lake designated as polymictic (i.e., periodic complete mixing of the water column), this poses serious concerns regarding the quality of sportfish habitat.

These findings not only raise questions about habitat condition in Lake Scugog, but other lakes in the region where N. obtusa has become established. With the distribution of N. obtusa expanding across Ontario lakes (Harrow-Lyle and Kirkwood, 2021), our study provides clear observational data that raise concerns for biogeochemical cycles, benthic habitat structure, and other biota in invaded lakes. These results also support our previous work, which implicated N. obtusa as a biotic driver of Microcystis blooms in Lake Scugog (Harrow-Lyle T. and Kirkwood A. E., 2020). Here we provide for the first time evidence of a possible mechanism for promoting bloom development, whereby N. obtusa drives down benthic DO to facilitate internal-phosphorus loading from sediments as well as possibly propagule germination. Further studies should be conducted to verify if these inferred effects by N. obtusa in Lake Scugog are occurring in other invaded lake ecosystems in North America, as well as identify possible phosphorus reserves most affected upon biogeochemical cycle alteration. If such notable impacts to lake biogeochemistry are documented region-wide, there will be more certainty that N. obtusa is acting as an ecosystem engineer in invaded lakes.