Part 2: Synthesis I: physics, nutrients, plankton and plants

The Hot Brook lakes of eastern Maine — part 2 of 5

← Introduction and study system · Contents · Synthesis II: fish, early life, predation, wildlife and forest →

3. Maine lake ecosystems: a synthesis

This section gathers what is known of small lakes like the Hot Brook pair: shallow (2–6 m), softwater, moderately colored, ice-covered for about five months, set in spruce–fir and northern-hardwood forest, and holding a warmwater fish community of perch, sunfish, bass, pickerel, suckers and bullheads. Maine and northern New England values come first; values from elsewhere are used where the region has none and are labeled as such; what is not known is gathered in section 6. The subsections follow the ecosystem from its physical setting (3.1) and nutrients (3.2) through the plankton (3.3) and plants (3.4) to the fish (3.5), their early life (3.6) and their predators (3.7), and the wildlife and forest around the lakes (3.8).

3.1 Climate, hydrology, ice and the physics of small shallow lakes

Almost nothing physical has been measured in the Hot Brook lakes. The record holds one pair of temperature readings, taken by a Maine Department of Inland Fisheries and Wildlife (MDIFW) crew in August 1949. There is no lake-level, ice, temperature-profile, oxygen, Secchi or color record, no Maine Department of Environmental Protection (DEP) or volunteer sampling station, and no gauge on Hot Brook (MDIFW survey sheets 1949, rev. 1965, 2004; Maine DEP Lakes Data GIS; Maine DEP & VLMP 2025). This section therefore builds the lakes' physical setting from the nearest climate stations, the regional stream gauges, Maine lakes of similar size, depth and color, and process studies from farther afield, and marks where those analogs stop. Morphometry is given in section 2.

3.1.1 Climate

Danforth, the lakes' own town, has a NOAA cooperative station (USC00171833) with full 1991–2020 normals (Table 3.1-1). Its annual mean air temperature is 4.9 °C, from −10.3 °C in January to 18.8 °C in July; the minimum falls below 0 °C on 198 days a year and the maximum exceeds 32.2 °C on 3.3 (NOAA NCEI 1991–2020 normals). Houlton International Airport, whose long daily record supplies the day-to-day statistics below, is almost identical (annual 4.8 °C). In Houlton's daily record the within-month standard deviation of daily mean temperature runs from 3.1 °C in July to 7.1 °C in January, with a lag-one autocorrelation of 0.64 (project computation from NOAA GHCN-Daily, 1991–2020), and December–March frost totals about 900 °C·d (project estimate).

Precipitation is spread evenly through the year: 1,157 mm at Danforth, every month between 84 and 118 mm, against 998 mm at Houlton (NOAA NCEI 1991–2020 normals) and 1,203–1,205 mm modeled for the Hot Brook watershed (US EPA StreamCat). Houlton's 30-year daily record shows rain arriving about every five days in one- to two-day systems: a day is wet (≥ 0.25 mm) 39 % of the time, rain follows a wet day with probability 0.50 and a dry day with 0.32, wet and dry spells average 2.0 and 3.1 days (longest dry spell 20), and the mean wet-day total is 6.9 mm (CV 1.33) (project computation from NOAA GHCN-Daily). At Houlton, days with at least 25.4 mm number 6.9 a year (NOAA NCEI normals). NOAA Atlas 14 gives 24-hour storm depths at Danforth of 57 mm (one-year return period), 91 mm (10-year) and 132 mm (100-year), and a 10-year one-hour depth of 33 mm; the median year's wettest day at Houlton, 56 mm, matches the one-year storm (project computation). Caribou averages about 2, 5 and 7 thunderstorm days in May, June and July (NWS Caribou climate summaries).

Danforth's normal snowfall is 243 cm, almost 90 % of it from December to March (NOAA NCEI normals). At Caribou, farther north, the median snow depth peaks at 53 cm on 1 March, falls to 20 cm by 15 April and is gone by 1 May, with at least 2.5 cm on the ground on 136 days a year (NOAA NCEI daily and annual normals); Houlton averages 120 such days (88–140), though from only nine complete seasons (project computation via RCC-ACIS). Mean snow-water equivalent (SWE) on 1 March exceeds 230 mm in northern Maine (Loiselle & Hodgkins 2002), but snowpacks vary widely: winter 2025–26 was Houlton's driest on record (NWS Caribou n.d., Winter 2025–26 seasonal narrative). Winds are light, especially in summer. Houlton's monthly means fall from 3.4–3.7 m s⁻¹ in January–May to 2.6 m s⁻¹ in August, daytime means are 3.8–4.7 m s⁻¹ and night means 1.7–3.0 m s⁻¹, and 17–26 % of hours are calm, most in August (NOAA NCEI hourly normals); Caribou is similar but records fewer calms (9.6–19.4 %). Calm thus prevails on roughly one summer night in five (project estimate).

Table 3.1-1. Monthly climate normals (1991–2020) near the Hot Brook lakes, and the monthly flow of the two nearest gauged streams as a ratio of mean annual flow (Mattawamkeag River near Mattawamkeag, USGS 01030500, 3,672 km², 1934–2026; Wytopitlock Stream, USGS 01030350, 127 km², 2008–2025; ratios computed by the project from USGS NWIS). Danforth's snowfall normals are zero from May to September.

Sources: NOAA NCEI 1991–2020 monthly and hourly normals (Danforth USC00171833; Houlton International Airport USW00014609); USGS NWIS.

3.1.2 Hydrology of the catchment

The catchments are small, forested and wet. Summed from the USGS NLDI network, Upper's outlet drains 65.8 km² and Lower's 77.2 km² (including Upper's), watershed-to-lake ratios of about 18:1 and 28:1 (project derivation). Upper's watershed is 73 % forest, 14.4 % woody wetland and 2.8 % developed, with no cropland (NLCD 2019, via US EPA StreamCat), and Hot Brook enters its south shore draining about 36 km² (project derivation from NHDPlus). With StreamCat's modeled runoff of 646 mm yr⁻¹, mean outflow is about 1.3 m³ s⁻¹ from Upper and 1.6 m³ s⁻¹ from Lower, residence times are roughly 100 and 55 days, and Lower receives 83–85 % of its water through the connecting brook (project estimates). The regional gauges agree: the Mattawamkeag River, which integrates the Hot Brook catchment through Baskahegan Stream, yields 638 mm yr⁻¹ and Wytopitlock Stream 712 mm yr⁻¹, about 0.6 of the precipitation at Houlton and Millinocket, which leaves roughly 360–440 mm yr⁻¹ of evapotranspiration; against Danforth's 1,157 mm the ratio is about 0.55 (project computation from USGS NWIS). The Mattawamkeag's annual flow varies with a CV of 0.24, from 0.50 to 1.62 times the mean in water years 1935–2025 (USGS NWIS).

The hydrograph is driven by snowmelt (Table 3.1-1). April flow is 3.3 times the annual mean and August–September flow 0.30–0.39 times, with a second, smaller rise in November–December from autumn rain, sometimes on snow. In the 1990–2025 mean daily hydrograph the peak falls on 15–19 April at 4.1–6.5 times the mean and the minimum on 25 August – 2 September at 0.18–0.29 times, and half the January–June flow has passed by a median 18 April at Mattawamkeag (range 1 April – 1 May) (project computation from USGS NWIS). That date has moved significantly earlier, by 1–2 weeks and mostly over the last 30 years, at all 11 snowmelt-dominated gauges in northern and mountainous Maine and New Hampshire, tracking March–April air temperature (r = −0.72) (Hodgkins & Dudley 2006). The median annual peak grows relative to mean flow as catchments shrink, from 6.5 times at Mattawamkeag to 19.6 times at 10-km² Williams Brook. Lake storage appears to delay the freshet: the two short records from regional lake outlets stay high into May–June (Clayton Stream was still 2.4 times its mean in May, 1982–84, and Cold Stream's mean peak came in May–June rather than April, 1904–06) (project computation from USGS records). Between June and October three small gauged streams rise by more than 30 % in a day 10–18 times a year, peaking after 1–2 days at a median 2.2–2.6 times the pre-storm flow and receding by a factor of 0.81–0.88 a day (project computation). By one estimate, a 25-mm storm would raise Upper by 8–13 cm, outflow peaking 1–3 days later (project estimate).

Summer low flow is the brook's season of greatest stress. Wytopitlock Stream fell below 0.018 times its mean on 5 % of days (Q95, 2009–2025; project computation), and the Mattawamkeag's 7-day, 10-year low flow is 0.026 times its mean (Dudley 2004a; ratio computed by the project). The 2025–26 drought was severe: in 2026 the Mattawamkeag fell to 0.65 m³ s⁻¹ (23 cfs) on 11–12 October, about 0.009 times its mean, and NWS Caribou reported the second-lowest flow since 1934 (NWS Caribou drought statements, autumn 2026; USGS NWIS provisional data). On 24 August 2025 the connecting brook was nearly dry, its riffles walkable and a lake shore exposed (angler observations, project field notes).

Groundwater's share of the inflow and the surficial geology beneath the lakes are unknown. In other glaciated north-eastern lakes groundwater supplies 55 % of inflow at Walden Pond, Massachusetts (Colman & Friesz 2001) and most inflow to kettle ponds without surface inlets (Walter & Masterson 2011), whereas at Lake Auburn, Maine, surface inflow supplies about 62 % (Dudley 2004b); even at one lake, Mirror Lake in New Hampshire, estimates span 47,000–133,000 m³ yr⁻¹ depending on method (Winter et al. 1989). The notes' working default for the Hot Brook lakes is 10–30 % of inflow (project estimate). Beavers dam the region's streams and outlets: on an untrapped site in south-central Maine colonies rose from 0.15 to 0.32 per km² in 1988–92 and open water by 36 % (McCall et al. 1996), and colonies favor stream gradients below 6 % (Allen 1982). The National Wetlands Inventory maps beaver-modified wetlands within 500 m of both lakes (photo-interpreted only). Neither lake has a dam in the state dam layer or the National Inventory of Dams; the notes infer that any control on the outlets is beaver-made, but no one has checked on the ground.

3.1.3 Lake level and ice push

No level record was found for any natural lake in northern Maine; the USGS lake gauges with daily records in the state are on regulated lakes (USGS NWIS). Three Massachusetts lakes without drawdown, logged every 2 h for 3–4 years, ranged in summer from 30 cm below to 102 cm above the summer pool (median +0.5 cm) (Carmignani et al. 2021), and two New York kettle lakes with natural outlets fluctuated about 0.45 m a year (Kappel, Miller & Hetcher 2001). For the Hot Brook lakes one estimate gives an annual cycle of roughly ±15–25 cm about the summer pool, highest after the freshet and lowest in August–September, with storm rises of 2–10 cm; another puts the annual range at 0.3–0.5 m (project estimates). The fall from Upper to Lower is small: 0.70 m along the brook in the 2017 lidar (USGS 3DEP), 0.6 m in DEP's lake layer. On gentle shores a 0.4-m fall would bare about 20 m of a 2 % sand flat but only 1 m of a 40 % bedrock shore (project estimate), and in the dry August of 2025 a lake shore showed a band of exposed stones grown with grass (angler observations, project field notes).

Ice works the shore too. On lakes under about 4 km across, thermal expansion of the ice rather than wind is the main cause of ice push (Gatto, Ferrick & Calkins 2001, citing Gatto 1982), while lakes under about 0.8 km across may be too small for appreciable push (Wagner 1970, via Gatto, Ferrick & Calkins 2001). At Lake Bomoseen, Vermont, ice moved 0.5–2.1 m inland in 1990 and 1998, riding up gentle shores while vertical banks resisted (Gatto, Ferrick & Calkins 2001); at Gardner Lake, Connecticut, push built a ridge 0.6–1.2 m high (Pessl 1969, known from a search summary). Both Hot Brook lakes, 3.3–3.4 km long and 1.7–1.9 km wide, fall in the size range for thermal push, and on one cobble shore that the angler photographed rooted plants are absent to about 0.6 m depth, consistent with ice and wave scour (angler observations, project field notes). Shores are treated in sections 3.2 and 3.4.

3.1.4 Ice

There is no ice record for either Hot Brook lake. The nearest long record, West Grand Lake in Washington County, has a mean ice-out of 28 April over 1878–2008 (SD 8.8 days) and 24 April over 1979–2008 (Hodgkins 2010; statistics computed by the project). Recent dates near Danforth are earlier: East Grand Lake averaged 23 April over 2015–2026 (range 12 April – 2 May) and 17–18 April over 2021–2026 (Maine Bureau of Parks and Lands; project computation). Farther north (46.6–47.0° N), mean ice-out comes 4–11 days later than at West Grand (Table 3.1-2). Ice-out advanced about 9 days between 1850 and 2000 across northern and mountainous New England, March–April air temperature explaining about half its year-to-year variance (Hodgkins, James & Huntington 2005), and of 29 lakes in a Maine DEP analysis 17 have significantly earlier ice-out and none later (Bacon 2019). The rate depends on the period: 1.8 days per decade over the 50 years to 2008 but 0.4–0.6 over 75–125 years (Hodgkins 2013, abstract), and in eight Maine lakes the 1–2-week advance since 1956 came mostly in steps (Ellis & Greene 2019, abstract). Late-February ice on the Piscataquis River thinned by about 23 cm (46 %) between 1912 and 2001 (Huntington, Hodgkins & Dudley 2003).

Table 3.1-2. Ice-out in Maine lakes. Long-term statistics computed by the project from Hodgkins (2010); East Grand and the 2015–26 West Grand mean from the Maine Bureau of Parks and Lands.

Ice-on is far less known. Maine has no ice-on trend, and the state's climate reports name ice-on as a monitoring gap (Maine Climate Council 2020, 2024, as relayed in the notes; not re-verified); Lake Stewards of Maine reports ice-in as late as January at some lakes since 2005 and in February in 2022 and 2023, trends "likely not significant" (Lake Stewards of Maine, Ice Data in Maine Lakes). A lake cannot freeze until its whole column has cooled to about 4 °C, so small, shallow basins ice over first (Maine DEP, Seasonal Magic). For the Hot Brook lakes the notes estimate ice-on in late November to mid-December and 135–150 days of ice, shallow coves freezing first and Upper's deep, windy north basin last (project estimates). Late-winter ice in Maine is typically 30–75 cm thick: 31–50 cm at 16 lakes around Bridgton in 2026, with maxima of 35–75 cm in 2018–2025 (LEA 2026), and about 61 cm on lower Moosehead Lake in spring 2026 (anglers' reports, Bangor Daily News 2026). Stefan's law with Houlton's ≈ 900 °C·d of frost and coefficients from Alaskan lakes gives 40–80 cm (project estimate; Arp et al. 2020). Much of the ice can be white: when snow pushes the ice below its freeboard, slush floods and refreezes, and at Lake Pääjärvi, Finland, snow-ice made up 10–43 % of the ice column (Leppäranta & Kosloff 2000). Beyond LEA's winter visits, no snow-on-ice series or white-ice fraction was found for any Maine lake.

Snow, not ice, controls the light beneath. At the Bridgton-area lakes in 2026, light just under the ice was about 1–12 % of surface irradiance, falling as snow deepened from none to about 20 cm, and chlorophyll often peaked within a few meters of the ice (LEA 2026). Median optical properties from 46 Northern Hemisphere lakes are an albedo of 0.55 and attenuation of 0.79 m⁻¹ for black ice, 0.60 and 4.35 m⁻¹ for white ice, and 0.94 and 8.96 m⁻¹ for snow (Donini et al. 2026; units confirmed only in a secondary summary, values varying widely between studies). Layered, these give about 33 % transmission through 40 cm of snow-free black ice, 1.8 % under 10 cm of snow and 0.3 % under 30 cm (project estimate), bracketing the Maine measurements. At Lake Balaton, Hungary, ice under 4 cm of snow passed about a tenth of surface light, and phytoplankton production was 4 mg C m⁻² h⁻¹ under snow-covered ice against 20 under snow-free ice (Herodek & Oláh 1973). No under-ice light measurement was found from northern Maine.

Under the ice the column is inversely stratified, temperature rising quickly through the first meter and then slowly toward 4 °C at the bottom (LEA 2026), and meltwater forms a dilute surface layer (mean specific conductance 12 µS cm⁻¹ under the ice in 43 Maine lakes; Cowing & Scott 1980). Early in winter, heat from the sediment is the main heat source, 0.6–4.5 W m⁻² in small ice-covered lakes, and lateral transport from shallow to deep water can match the vertical fluxes (Bengtsson et al. 1996; Malm et al. 1997). Once the snow goes, sunlight through the ice drives radiative convection, deepening the mixed layer by about 0.5 m d⁻¹ and up to 3 m d⁻¹ (Bouffard & Wüest 2019); at Lake Vendyurskoe, Karelia, this phase lasted 2–7 weeks (Zdorovennova et al. 2021b, abstract). In a 3–5-m lake such convection would mix the whole column within 1–10 days (project estimate).

3.1.5 Temperature and mixing: shallow polymictic versus stratified lakes

The only temperatures on record for Hot Brook are from August 1949: a surface reading of 23.3 °C (74 °F), printed identically on both lakes' survey sheets, which also both print "75 feet – 71 °F" (21.7 °C), a depth impossible in lakes 5.5 and 4.0 m deep; if this was a near-bottom reading under a wrong depth label, the column was within 1.7 °C from top to bottom. The sheets call each lake "shallow and does not stratify during summer months" (MDIFW survey sheets 1949, rev. 2004).

Among Maine lakes, depth largely decides whether a lake stratifies. Of 35 lakes surveyed in late summer 1974–76, six were unstratified, among them Long Pond (5.5 m), Portage Lake (7.6 m), Pushaw Lake (8.5 m) and Molasses Pond (14.3 m deep but 5.2 m in mean depth over 507 ha), and four lakes 11.6–14.6 m deep stratified only 4–5 % of their volume (Cowing & Scott 1980). MDIFW survey sheets show the same gradient, from shallow ponds within 0.6–1.7 °C top to bottom and an isothermal Horseshoe Lake (6.1 m) to a weakly stratified Fifth Machias Lake (8.2 m), and recent profiles from shallow lakes near Hot Brook were uniform within about 1 °C (Table 3.1-3). Brackett Lake, 7.6–10 m deep, sits on the threshold, strongly stratified in some summers and mixed in others. The notes infer that persistent summer stratification in Maine needs a maximum depth of about 8–10 m, and that a Hot Brook-type lake is polymictic with a top-to-bottom difference of 0–3 °C (project inference).

Table 3.1-3. Late-summer temperature and dissolved oxygen in lakes near Hot Brook and in Maine survey lakes. Saturation for the EPA National Lakes Assessment (NLA) profiles was computed by the project assuming 100–150 m elevation.

The physics comes down to wind stirring against buoyancy. A 1-m, 1 °C afternoon thermocline in Upper under a 5 m s⁻¹ wind has a Wedderburn number of about 0.015, far below the value near 1 at which a thermocline upwells, so any moderate wind destroys daytime stratification (project estimate). A forested shore shelters roughly 50 canopy heights downwind (Markfort et al. 2010), and full wind stress may need about 1 km of open water to develop (Markfort et al. 2014, abstract). The Hot Brook lakes have median maximum fetches of about 2 km (project computation on the lidar shoreline); behind a 15–20-m forest the notes infer that the upwind third to half of a 2–3-km lake is sheltered (project inference). Convection can outweigh the wind: in 40 temperate lakes, convection exceeded wind shear in 18, including all 11 lakes smaller than 10 ha (Read et al. 2012, abstract). The lakes' basin-scale motions are small and fast; the fundamental surface seiche has a period of about 19 minutes and a 10 m s⁻¹ wind sets up only about 3 cm (project estimates).

Polymictic lakes do stratify in calm, hot spells, and those episodes matter for oxygen. In the polymictic Müggelsee, Germany, two heatwave summers produced stratification events of a week to two months, during which bottom-water oxygen fell strongly and soluble phosphorus rose (Wilhelm & Adrian 2008, abstract), and Maine DEP reports that shallow lakes "that have undergone little or no stratification in the past are now experiencing this phenomenon" (Range Ponds 2024, citing DEP). Dense plant beds stratify on their own: in Lake Monger (0.3–1.2 m deep), beds filling more than half the water column stratified for about 8–12 h a day (modeled), with gradients of up to 10 °C over 0.8 m, and in the model a full canopy cut depth-averaged flow by about 90 % (Vilas et al. 2017). Night cooling of shallow margins drives density currents that flush littoral water offshore; at Rotsee, Switzerland, these thermal siphons ran on 37 % of days and flushed a 1-m-deep plateau in about 10 h (Doda et al. 2022), and the same scaling gives about 5.5 h for a Hot Brook cove on a clear autumn night (project estimate). Color also shapes the heat budget: across lakes of increasing color the epilimnion thinned from 2.5 to 0.75 m (Houser 2006, abstract), and among 86 Ontario lakes colored lakes kept more stable deep-water temperatures than clear ones, which lost cold water in warm years (Snucins & Gunn 2000, abstract).

In Maine lakes, spring turnover happens near 4 °C, quickly in small, protected lakes (Schmitt n.d., popular source). Thermistor strings in 13 lakes show stratification from June to early November and full mixing in late October–November (LEA). Late-summer water in shallow lakes near Hot Brook ran 19–24 °C (Table 3.1-3), but heat waves push Maine lake surfaces much higher: Moose Pond reached 29.4 °C at the surface on 18 July 2024 (LEA 2024) and the Range Ponds 28.2–30.2 °C on 12 July 2024 (Range Ponds 2024). Maine lakes have warmed. The average annual maximum August lake temperature has risen 2.3 °C since 1977 (Maine Climate Council 2026); a preliminary DEP analysis of 92 lakes since 1980 found surfaces 3 °C warmer and thermoclines more than 1 m shallower (Maine Climate Council 2020); and Landsat skin temperatures of 3,955 northern New England lakes rose 0.8 °C per decade in 1984–2014, fastest in small lakes (Torbick et al. 2016, abstract). Large storms disrupt stratification only briefly: Tropical Storm Irene in August 2011 cut Schmidt stability within hours in nine north-eastern lakes, all but one recovered to at least 80 % of pre-storm stability within a week, and the terrestrial inflow turned them net-heterotrophic for longer (Klug et al. 2012).

3.1.6 Oxygen

In summer, shallow, wind-mixed lakes in eastern Maine are well oxygenated from top to bottom (Table 3.1-3). Skitacook Lake (5.2 m) held 8.2–9.1 mg/L at every depth in August and September (97–102 % of saturation), Webster Pond 86–89 % (EPA NLA, via the Water Quality Portal), and Baskahegan, Faulkner and North lakes 7.6–8.7 mg/L at their deepest readings (Maine DEP & VLMP 2025). Slight undersaturation is common: about 75 % of 33 southern Quebec lakes were slightly undersaturated in O₂ in summer, the balance between net autotrophy and heterotrophy falling at dissolved organic carbon (DOC) of 4–6 mg/L (Prairie, Bird & Cole 2002, abstract), a range that includes Hot Brook's expected DOC (project inference). Oxygen solubility at the lakes' elevation (129 m) is 14.4 mg/L at 0 °C, 12.9 at 4 °C and 8.3 at 24 °C (Garcia & Gordon 1992; project computation). Stratified Maine lakes are different. Among the more productive stratified lakes of the 1974–76 survey only one had hypolimnetic oxygen above 2.0 mg/L and 67 % had less than 0.5 mg/L (Cowing & Scott 1980), Brackett Lake's bottom water fell to 0.2–0.4 mg/L in stratified summers (Maine DEP & VLMP 2025), and at three New England lakes earlier ice-out brought lower late-summer bottom oxygen, explaining 32–46 % of its variation (Hodgkins 2013, abstract). Summer has its own hazard: MDIFW reports late-summer and early-autumn low-oxygen kills that "follow cloudy spells and plant dieback" (MDIFW 2020), and in shallow, densely vegetated Chara ponds surface water swung from 12–25 % saturation at sunrise to 205–235 % in the afternoon while the bottom went anoxic at midday (Andersen, Kragh & Sand-Jensen 2017).

Re-aeration sets how far respiration can pull a mixed lake below saturation. At 15-ha Mirror Lake, New Hampshire, the gas-transfer velocity k₆₀₀ was 0.64 m d⁻¹ and independent of wind over a 50-day period (Cole & Caraco 1998, abstract); a cross-lake model gives about 1.6 m d⁻¹ for the 2.8–3.6-km² Hot Brook lakes at a 2.5 m s⁻¹ wind and 3.2 m d⁻¹ at 6.5 m s⁻¹ (Vachon & Prairie 2013; project computation), though predictions are least reliable for small lakes in light wind (Klaus & Vachon 2020). Night convection alone sustains about 0.55–0.93 m d⁻¹ for heat losses of 25–200 W m⁻² (project computation after Tedford et al. 2014). A lake at 95 % saturation with k of 1–2 m d⁻¹ is therefore running a net demand of only 0.43–0.86 g O₂ m⁻² d⁻¹ (project computation). Whole-lake metabolism spans 0.1–5.0 mg O₂ L⁻¹ d⁻¹ for gross primary production (GPP) and respiration (R) across the GLEON network of lakes (Solomon et al. 2013, abstract); at Lake Sunapee, New Hampshire, annual median GPP was 0.38–0.90 mg O₂ L⁻¹ d⁻¹, net ecosystem production (NEP) near zero and diel swings "generally below 1 mg/L" (Richardson et al. 2017). Across lakes of intermediate phosphorus the bed contributes 5–80 % of primary production, depending on morphometry (Vadeboncoeur et al. 2003, abstract). For Hot Brook the notes expect summer NEP of −0.9 to +0.3 g O₂ m⁻² d⁻¹, surface water at 88–100 % of saturation and a diel swing of 0.2–0.8 mg/L (project estimate).

The sediments are a major oxygen sink in a shallow lake, and a poorly measured one. Intact winter cores from 12 Upper Mississippi backwater lakes consumed 0.04–0.44 g O₂ m⁻² d⁻¹ at 0–5 °C and 0.14–1.46 g m⁻² d⁻¹ corrected to 20 °C, highest at vegetated, low-flow sites (Perner & Kreiling 2025); lake models typically use 1–4 g m⁻² d⁻¹, and Lake Mendota's effective summer value in a model was 1–1.5 g m⁻² d⁻¹ (Ladwig et al. 2021). Uptake falls within hours when near-bed turbulence dies (Bryant et al. 2010, abstract), and littoral sediments mineralize about three times as fast as profundal ones (den Heyer & Kalff 1998, abstract). For Hot Brook in July–August the notes expect a lake-mean sediment demand of about 0.7–1.0 g O₂ m⁻² d⁻¹, rising to 1–2 g m⁻² d⁻¹ in the organic mud of Upper's north basin (project estimate).

Winter is the oxygen risk: the ice seals the surface while the sediments keep respiring. No under-ice oxygen profile exists for Hot Brook, and none was found for any shallow lake near it. At the Bridgton-area lakes in 2026, surface water under the ice was near saturation, but 5 of 16 lakes had bottom water at or below 2 mg/L on at least one visit (LEA 2026). The whole-lake depletion rates the notes found come from Arctic lakes: 0.2–0.39 g O₂ m⁻² d⁻¹ in three lakes 9–13 m deep against 0.03–0.14 in 24-m Toolik Lake, higher in smaller lakes, and falling to 20–50 % of the initial rate by late winter (Schwefel et al. 2023, abstract). Over Upper's published mean depth of 3.35 m these become 0.06–0.12 mg/L d⁻¹, and 0.2–0.4 mg/L d⁻¹ in 1-m coves; held for 145 days they would remove 9–17 mg/L from the mean column, enough for late-winter hypoxia (project estimate). Declining late-winter rates, lateral exchange and under-ice photosynthesis work against this. At Lake Vendyurskoe early-freezing winters lost more than 50 % of the column's oxygen and late-freezing ones less than 40 %, and long spring convection raised oxygen by 10–15 % before ice-out (Zdorovennova et al. 2021a, abstract). In 20 northern Alaskan lakes depth and littoral area explained most of the late-winter oxygen, longer ice meant less, and snow depth mattered in shallow lakes (Leppi, Arp & Whitman 2016). Oxygen is patchy under ice: at hypoxic Upper Red Rock Lake, Montana, the top meter and the water off stream mouths stayed well oxygenated into late winter (Davis et al. 2020).

In Maine, the documented winterkills that the notes found are in very shallow, weedy ponds. MDIFW records winterkill "during the more severe winters" in Passaconway Pond (1.2 m deep), where oxygen fell to 0.8 mg/L in February 1961; in Great Pond, Cape Elizabeth (1.5 m), "during years of heavy snowfall and thick ice cover"; and in Sherman's Mill Pond (1.8 m), "especially during years when water flows are low" (MDIFW lake survey maps). Pushaw Lake (8.5 m) showed "depressed oxygen levels during the winter" (Cowing & Scott 1980), and Minnesota DNR names small, shallow lakes, early ice-on, late ice-out and heavy snow as the risk factors (Minnesota DNR). No winterkill at Hot Brook was found in the records. The lakes' persistent white perch fishery is indirect evidence that winter oxygen stays adequate: MDIFW notes that white perch persist in Maine lakes of about 20 ha and 7.6 m (25 ft) or more "if the deeper water is not depleted of oxygen during the winter months" (Smith 2002), and the Hot Brook lakes, shallower than that rule of thumb, still hold the fishery (the inference is the notes'). Fish tolerances of low oxygen are treated in section 3.5.

3.1.7 Light and water color

No Secchi depth, color or DOC is on record for either lake. The sheets do not describe the water's color or clarity (MDIFW 1949). The nearest analog lakes are moderately colored and fairly clear (Table 3.1-4), and from them the notes infer a Secchi depth of about 3.5–4.5 m for Hot Brook, color of roughly 20–50 SPU and DOC of about 4–5 mg/L (project inference). Earlier project estimates of 30–60 SPU and Secchi 2.5–4 m predate the analog data. One of the angler's photographs shows clear, golden water over cobble on one shore (angler observations, project field notes).

Table 3.1-4. Water clarity and color in lakes near Hot Brook (DEP annual means or ranges of yearly means; EPA NLA single visits). Color is true color in SPU (Pt-Co units).

Maine's lakes are on average lightly colored: mean true color is 20.7 SPU (range 0–197), and lakes above 25 SPU are classed as colored (Lakes of Maine 2017 report guide; Lake Stewards of Maine). Above 30 SPU phosphorus "is tied up with the color producing compounds" and Secchi-based trophic indices overstate productivity (Cowing & Scott 1980), so DEP assesses such lakes from chlorophyll, oxygen and judgment (Maine DEP 2024; trophic state is treated in section 3.2). Color and algae both cut clarity: in 1,151 paired August samples Secchi depth correlated with chlorophyll at Spearman r = −0.74 in colored lakes and −0.65 in clear ones (Maine DEP 2009, Chapter 583 nutrient-criteria description), and Maine lakes are clearer in droughts, summer Secchi falling with color and with January–June precipitation (Bacon & Williams, via Range Ponds 2024).

Whether Maine's lakes are getting murkier is disputed. Landsat imagery of 547 lakes shows mean Secchi falling from 4.94 m in 1995 to 4.38 m in 2010, and in north-eastern Maine from 5.22 to 4.36 m (McCullough, Loftin & Sader 2013, via a University of Maine summary), whereas volunteer Secchi records show clarity stable or slightly improving over 1976–2013 (Canfield et al. 2016, per a search summary; source not opened) and DEP reports trends that "appear to remain stable" (Maine DEP 2024). Where water has darkened, the browning is linked to recovery from acid deposition (Maine DEP 2024; Monteith et al. 2007, abstract), with more precipitation and runoff and longer growing seasons also cited (Maine Climate Council 2020): all 29 high-elevation Maine lakes in one set had positive DOC trends over 1986–2015, 19 significant, while low-elevation lakes were mixed (Gavin et al. 2018, abstract), and in Acadia lakes transparency fell by about 0.3 m a year in clear lakes and 0.1 m a year in brown ones with chlorophyll steady (Strock et al. 2017, abstract). DOC arrives in pulses: up to 86 % of annual DOC export from forested catchments occurs in storms and snowmelt (Raymond & Saiers 2010, as cited by Klug et al. 2012), and six Maine drinking-water lakes answered severe storms with a short DOC spike, a sustained rise, or, in the largest, no change (Warner & Saros 2019, via a news summary).

The notes found no diffuse attenuation coefficient for photosynthetically active radiation (Kd, PAR) measured in any Maine lake. Elsewhere DOC explained 87–96 % of the among-lake variation in Kd (Morris et al. 1995, abstract), and in 75 Norwegian lakes colored dissolved organic matter absorbed 56 % of PAR against 7 % for phytoplankton pigments (Thrane, Hessen & Andersen 2014). Converting Maine's many Secchi readings into light needs the product Kd × Secchi, and that is uncertain: the notes use 1.44, 1.7 (from a low-quality source) and 2.70 (the median for stained lakes; Koenings & Edmundson 1991), which at Hot Brook's likely Secchi depth of about 3.9 m give Kd of roughly 0.4 to 0.7 m⁻¹ (project estimates). In lakes this shallow, light reaches much of the bed; how color limits plant depth is treated in section 3.4. Section 5.10 records how two of these field values, Houlton's daily sequence of wet and dry days and the 1–12 % of surface light measured under Maine lake ice, replaced assumptions in earlier versions of the model.

3.2 Nutrients, shores and the land–lake link

Upper Hot Brook Lake drains about 65.8 km². Lower Hot Brook Lake drains about 77.2 km², a figure that includes Upper's catchment. The catchment-to-lake area ratios are therefore roughly 18:1 and 28:1 (NLDI basin sums; project derivation). NLCD 2019 land cover, as summarized by EPA StreamCat, is 73.0 % forest and 15.1 % wetland in Upper's catchment, and 70.4 % forest and about 15 % wetland in Lower's. Developed land covers 2.8 and 2.5 %, and hay less than 0.3 %. The main inflow, Hot Brook, drains about 36 km², roughly 55 % of Upper's catchment, and enters Upper's south shore (project derivation). The lakes flush roughly 3.5–3.7 and 6.6 times a year, and 83–85 % of Lower's inflow is Upper's outflow (project estimates; Section 3.1). No phosphorus, nitrogen, color, carbon or oxygen measurement was found for either lake or any inflow. The only water-quality data are a pair of temperature readings from August 1949 (MDIFW lake survey maps). Maine DEP lists both lakes in EPA Category 2, which presumes, without data, that their designated uses are attained (Maine DEP 2024 Integrated Report GIS). Everything below about Hot Brook's nutrients is therefore inferred from regional measurements.

3.2.1 Phosphorus export from forested and wetland catchments

Maine DEP estimates watershed phosphorus loads from land-use export coefficients (Table 3.2-1). Undeveloped forest is assigned 0.04 kg TP ha⁻¹ yr⁻¹, after "a New England regional study" (Likens et al. 1977, via Maine DEP 2003). Wetland is assigned 0.10 (0.10–0.20) and harvested forest 0.40, ten times the undeveloped rate (Maine DEP 2003). Regional SPARROW calibrations give higher forest yields, and Hubbard Brook's measured stream export is lower, half dissolved and half particulate (Table 3.2-1). The coefficients matter most for the contrasts they draw. At eutrophic Webber Pond, forest covered 47.6 % of the direct watershed but supplied only 9.3 % (49 kg) of a 526 kg yr⁻¹ external load (Maine DEP 2003).

Table 3.2-1. Phosphorus export coefficients and measured yields relevant to forested, wetland-rich Maine catchments (kg TP ha⁻¹ yr⁻¹).

Northeastern forest streams carry little phosphate (Table 3.2-2). At Bear Brook, Maine, dissolved P stayed at or below 3 µg/L even in storms, while total P rose with particulate Al and Fe (Reinhardt et al. 2004, abstract). Hubbard Brook's stream sediments buffer dissolved P toward about 2 µg/L (Meyer & Likens 1979, via the Hubbard Brook synthesis). The rest of the total P is mostly dissolved organic P (DOP) in colored, wetland-rich catchments. In 21 Finnish boreal catchments 79 % of TP was dissolved, P was exported "mainly in organic form", and TOC export explained 61–73 % of the variance in TP export (Kortelainen et al. 2006, abstract). In steep, clear catchments particulate P (PP) dominates instead. In one stream of each kind (Narraguagus River and Biscuit Brook), suspended sediment carried about 0.5 mg P per gram (project computation).

Table 3.2-2. Phosphorus concentrations, forms and concentration–discharge exponents (b in C ∝ Q^b) in regional streams. "Project computation" means fits to USGS discrete samples joined to daily flows, with non-detects treated as censored.

Particulate concentrations rise with flow while phosphate's do not, so most of the load arrives on relatively few days. Two-thirds of Hubbard Brook's annual P output leaves in ten days of high flow (Hubbard Brook synthesis, ch. 15). On three northern Maine streams of 10–127 km², the wettest 10 % of days carry 40–47 % of the water. On the two nearest Hot Brook's size (77 and 127 km²), the regional exponents put about 60–66 % of the PP and about 50–60 % of the TP on those days (project computation). Up to 86 % of annual DOC export likewise moves in storms and snowmelt (Raymond & Saiers 2010, as cited by Klug et al. 2012).

For Upper's inflow brooks, the project estimates a flow-weighted TP of about 13 µg/L (range 8–20). That is roughly 525 kg yr⁻¹ from 66 km², or about 0.08 kg ha⁻¹ yr⁻¹. This rate falls between DEP's forest and wetland coefficients and below the SPARROW forest yield. The expected split of the load is about 15 % SRP (≈ 2 µg/L), 45 % DOP (≈ 6 µg/L) and 40 % PP (≈ 5 µg/L), shifting toward DOP at baseflow and toward PP in storms (project estimates). Lower's inflow is mostly Upper's lake water. Its chemistry should therefore resemble Upper's surface water, with SRP at 1–2 µg/L or less, rather than a forest brook's (project estimate). No Hot Brook measurement underlies these numbers. No P measurement was found for any small Washington or Aroostook County forest brook; USGS holds none for the gauged Old Stream, Wytopitlock Stream or Williams Brook.

3.2.2 Bioavailability, color and nitrogen

SRP is essentially all available to algae, but much high-molecular-weight dissolved P goes unused (Boström, Persson & Broberg 1988, abstract). In algal assays on 172 Finnish samples, forest runoff ranked among the least available P sources (Ekholm & Krogerus 2003, abstract). Color adds a Maine-specific constraint. In the USGS study of 43 Maine lakes, "in lakes where color is greater than 30 units, phosphorus is tied up with the color producing compounds". Dropping those lakes raised the TP–chlorophyll r² from 0.46 to 0.69 (Cowing & Scott 1980). The project's working assumptions are that all SRP, about a fifth of forest PP and about a quarter of colored-water DOP are available. Upper's inflow would then supply about 4.5 µg/L of algal-available P out of its ~13 µg/L TP (project estimate). No algal-assay data were found for any Maine stream, and Hot Brook's color is unmeasured. Baskahegan Lake, in the same drainage, measured 46 SPU in 2004 (Maine DEP annual means 1940–2018).

Nitrogen is poorly documented. The project found no Maine forest N export coefficient. Lake Stewards of Maine state that "Most Maine lakes have significant nitrogen, so phosphorus is the nutrient most limiting". Hubbard Brook litterfall cycles 140.4 kg ha⁻¹ yr⁻¹ of eleven elements, with N the largest share (Gosz, Likens & Bormann 1972, abstract). Shoreline sources pass nitrate readily. Plumes from 24 Ontario septic systems carried 0.1–76 mg/L nitrate-N even where most P was retained (Robertson, Van Stempvoort & Schiff 2019), and Wisconsin lawn runoff carried about 230 times the nitrate of wooded catchments (Graczyk et al. 2003). How much N the Hot Brook catchment exports, and whether either lake is ever N-limited, is unknown.

3.2.3 Direct deposition, pollen and sea-run fish

Maine DEP uses 0.16 kg P ha⁻¹ yr⁻¹ (0.11–0.21), or 16 mg m⁻² yr⁻¹, for deposition onto open water (Maine DEP 2003). Hubbard Brook's bulk collectors gave about 0.04 kg ha⁻¹ yr⁻¹, which its authors consider an underestimate, and a global synthesis gives 0.27 (Tipping et al. 2014, via the Hubbard Brook synthesis). At DEP's rate, Upper receives about 59 kg yr⁻¹, roughly 10 % of its external load, and Lower about 46 kg yr⁻¹ (project estimates). Pollen adds a brief spring pulse. Lake 373 in the Experimental Lakes Area is a 27-ha lake in jack-pine forest. There, floating collectors caught 0.34–1.7 mg P m⁻² d⁻¹ for 45 days after ice-out, about 37 mg m⁻² yr⁻¹ (project arithmetic). Pollen additions raised phytoplankton, zooplankton and periphyton in littoral mesocosms (Graham, Vinebrooke & Turner 2006, abstract). No pollen deposition has been measured on a Maine lake, and Hot Brook's shores are spruce–fir and mixed forest, not jack pine.

Sea-run alewives carry marine phosphorus into lakes, and their young can carry it out again. At Bride Lake, Connecticut, an adult that dies in the lake adds about 0.68 g P, and one that survives adds about 0.18 g, two-thirds of it by excretion. Alewives supplied 23 % of the lake's annual P load there, but small restored runs may export more P than they import (West et al. 2010, abstract; see also Walters, Barnes & Post 2009). In Maine, alewife-derived P was under 5 % of mean epilimnetic P in China, Highland and Togus, but 8.7–23.9 % in Webber Pond (Thurrell 2022, thesis abstract). Sea-run alewives were first recorded at Lower Hot Brook Lake on MDIFW's 2024 depth chart. The size of the run, and whether alewives reach Upper, are unknown.

3.2.4 Retention, internal loading and the expected lake phosphorus

No measured phosphorus budget was found for any small Maine or Maritime lake with a residence time of 0.1–0.5 years (National Eutrophication Survey working papers and Maine DEP TMDLs, which may hold some, could not be reached). The project therefore computed the retention coefficient R from standard empirical models (Table 3.2-3). R is the fraction of inflowing TP that does not leave by the outlet. The model forms are as commonly cited; only the SPARROW coefficients were read in full. The central values are R ≈ 0.45 (0.34–0.59) for Upper and 0.38 (0.28–0.49) for Lower. Across 54 lakes and reservoirs, R rose with residence time, most strongly in lakes larger than 25 km² (Kõiv, Nõges & Laas 2011, abstract). Ontario's Lakeshore Capacity Model, however, cautions that its predictions are less reliable for shallow lakes (mean depth under 5 m), and both Hot Brook lakes are that shallow (Ontario Ministry of the Environment 2016).

Table 3.2-3. Phosphorus retention coefficients computed for the Hot Brook lakes (project arithmetic; Upper qs = 11.0 m/yr, τ = 0.274 yr; Lower qs = 16.9 m/yr, τ = 0.151 yr; qs is the areal water load, τ the water residence time and ρ = 1/τ the flushing rate). Formulas are as cited in the project notes; the originals were not re-read except where marked.

Particulate P is retained far more efficiently than dissolved P. The project's Stokes calculation for mineral grains at 10 °C gives settling velocities of about 235 m/d for fine sand, 1.5 m/d for fine silt and 0.06 m/d for 1-µm clay. Fine sand therefore settles within about 470 m² of the brook mouth at mean inflow and about 4,700 m² at ten times that flow (Hazen's ideal-basin area). Clay of 1 µm would take about 50 days to fall 3 m, which is comparable to Lower's residence time (project computation). The project infers that about half of the incoming PP settles near the inlet at ordinary flows and that 15–35 % passes the outlet. It puts retention at 0.75–0.9 for PP and 0.1–0.3 for DOP (project estimates). No measurement of deposition near an inlet, as against lake-wide, was found for any small northeastern lake.

With an inflow TP of about 13 µg/L and R ≈ 0.45, Upper's annual mean TP before internal loading would be about 8 µg/L including direct deposition, and Lower's about 6 µg/L (project estimates). Measured lakes in the region read similar or slightly higher. Epilimnetic TP was 9–10 µg/L in Baskahegan Lake (2004, 2011), in the same drainage, 7 µg/L in Molunkus Lake and 9–12 µg/L in Faulkner Lake (Maine DEP annual means 1940–2018). Skitacook Lake, 5.2 m deep and 47 km away, read 7.6 µg/L in August 2017 (EPA National Lakes Assessment, via the Water Quality Portal). Whether about 8 µg/L counts as oligotrophic depends on the scale used. Maine DEP's criteria, applied to long-term means, place 4.5–20 µg/L in the mesotrophic band, although for lakes colored above 30 SPU DEP judges by chlorophyll and oxygen instead (Maine DEP 2024 Integrated Report). Baskahegan's 2004 Carlson indices of 37–42 sit at the oligotrophic–mesotrophic boundary (project computation after Carlson 1977). A lake TP of 10–20 µg/L at Hot Brook would require either inflow nearer 18–25 µg/L or several µg/L from summer internal loading (project estimate).

Internal loading is well documented in Maine lakes. At eutrophic Webber Pond, sediment release (537 kg yr⁻¹) matched the whole external load (526 kg yr⁻¹) (Maine DEP 2003). In Brewer Lake, hypolimnetic TP reached 116 µg/L. Epilimnetic TP rose from 8 to 14 µg/L at autumn turnover, and an Anabaena flos-aquae bloom followed, a pattern "observed on other Maine lakes" (Cowing & Scott 1980). The Hot Brook lakes do not stratify (MDIFW lake survey maps). Comparable shallow lakes nearby were oxygenated from top to bottom in late summer: Skitacook Lake held 8.2–8.3 mg/L O₂ at every depth sampled in August 2017, and Webster Pond (profiled to 2.5 m) 7.3–7.6 mg/L in August 2012 (EPA National Lakes Assessment). Unstratified, 8.5-m-deep Pushaw Lake was well oxygenated in summer but had "depressed oxygen levels during the winter" (Cowing & Scott 1980). At Hot Brook, sediment release is therefore probably confined to warm, calm spells and to late winter under ice (project inference).

Three properties of the sediment would govern that release, and none is known at Hot Brook. The first is metal chemistry. Particulate Al, Fe and P travel together in Maine storm flows (Reinhardt et al. 2004, abstract), and sediments with molar Al:Fe > 3 and Al:P > 25 are reported not to release P under anoxia (Kopáček et al. 2005; not re-read). The second is pH. Oxic release in a Wisconsin lake rose from under 0.5 mg m⁻² d⁻¹ at pH 7 to 2–3 at pH 8.5 (James et al. 2001). The release should therefore be small at the pH of about 6.6–7.2 expected by analogy with Baskahegan Lake, except inside dense plant beds (project inference). The third is rooted plants. In mesotrophic water, sediment supplied all of the P that rooted plants took up (Carignan & Kalff 1980), and the plants return part of it to the water when they senesce (Section 3.4).

3.2.5 Terrestrial subsidies: insects, leaves and dissolved carbon

Forest streams receive enough land invertebrates to shape their fish. A Swedish outlet stream under more than 70 % conifer canopy received 39.5 ± 7.2 mg dry mass m⁻² d⁻¹ from May to October. That was about four times the input to sections covered experimentally (Erős et al. 2012). In a Japanese deciduous-forest stream, terrestrial invertebrates supplied 44.0 % of the fish assemblage's annual energy (Nakano & Murakami 2001). Lakes receive this input only along a thin edge. Airborne terrestrial particulate carbon averaged 1.88 g C per meter of shoreline per year across seven estimates, declining exponentially away from the shore (Vander Zanden & Gratton 2011). A German model concluded that a substantial fish-mediated subsidy of terrestrial P is likely only in lakes smaller than about 3 ha (Mehner et al. 2005, abstract; disputed by Wurtsbaugh 2007). Terrestrial insects reached 100 % of fish diets in some undeveloped lakes but averaged 2 % in developed ones (Francis & Schindler 2009, abstract, 24 lakes). In general, though, they are "rarely dominant in lentic fish diets" (Vander Zanden & Gratton 2011). For the connecting brook the project estimates about 12–33 mg m⁻² d⁻¹ at the summer peak, assuming a channel 3–8 m wide (unsurveyed) and scaling a per-meter-of-bank rate attributed to Mehner et al. (2005) that was seen only in a search summary. Averaged over Upper, with about 4 m of shore per 1,000 m² of water, the input is about 5–8 mg m⁻² yr⁻¹. That is under 1 % of the lake's own insect emergence (project estimates).

The Hubbard Brook forest drops about 570 g dry m⁻² yr⁻¹ of litter, about 49 % of it leaves and about half of it in autumn (Gosz, Likens & Bormann 1972, abstract). Small forest streams run on this litter. In Bear Brook, New Hampshire, more than 99 % of energy input was allochthonous (Fisher & Likens 1973, abstract). Lakes receive leaves mainly within about 10 m of a forested shore (Vander Zanden & Gratton 2011). Airborne litter could equal at least 10 % of phytoplankton production in oligotrophic and very small eutrophic lakes (Gasith & Hasler 1976, abstract). The leachate may matter more than the leaves themselves. The leaf litter falling from forested shores of boreal Ontario lakes was estimated to supply 17.8 g of leached DOC per meter of shoreline per year, against 0.4 g after riparian clear-cutting (France et al. 1996, abstract; the paper measured the litter, and the DOC is derived from it). For Upper, that rate gives about 0.07 g C m⁻² yr⁻¹, concentrated in October–November (project estimate). Leaves break down at about 0.010 d⁻¹ for red maple and 0.0055 d⁻¹ for white pine needles in streams (Webster et al. 1999). On lake beds, exposure matters: beech decayed about six times faster on an exposed shore of Lake Constance than on a sheltered shore of Lake Stechlin, and storm waves stripped the litter away (Pabst et al. 2008, the Stechlin rate after Casper 1987).

Insects also carry nutrients from the lake to the land. Across 49 whole-lake estimates, median benthic secondary production was 2.3 g C m⁻² yr⁻¹, insects made up about 70 % of it, and 30 ± 4 % of insect production emerged (Vander Zanden & Gratton 2011). At Mývatn, Iceland, midges were about 10 % N and 1 % P by dry mass, and 70–99 % of emerging insects never returned to the water (Dreyer et al. 2015). Applied to Upper, these figures give an emergence export of roughly 7–11 mg P and 70–110 mg N m⁻² yr⁻¹. That is about half to two-thirds of the direct atmospheric P input, and roughly two orders of magnitude (about 90–220 times) more than land insects bring in (project estimate). On the lake, therefore, runoff and the atmosphere dominate the phosphorus inputs, and emerging insects dominate the exports to the land. In the brook the balance tips toward the land, which supplies about 550 g dry m⁻² yr⁻¹ of litter (project estimate).

3.2.6 The lakeshore: shore types, vegetation zones, ice push and water level

MDIFW's 1949 survey sheets, revised in 1965 and 2004, describe Upper's shoreline as "mostly rocky with some areas of sand", with "a large sandy beach in the northeast part" and wetlands on the south and west. Lower's is mostly rocky with some sand and mud, with wetlands on the north and east. Both lake bottoms are "mostly muddy" (MDIFW lake survey maps). The project classified the NLCD land cover beside each stretch of shoreline. Wetland adjoins 38 % of Upper's shoreline and 46 % of Lower's, an upper bound because NLCD's woody wetland can stand behind a stony waterline. Forest adjoins 34 % and 36 %, and forested shore that the project's model grid classes as rock a further 16 % and 11 % (project analysis). The project's range is a stony waterline along 50–65 % of the perimeter, with fen, shrub and marsh along 25–45 % (project estimate). Upper has about 14.6 km of shoreline, including a 4.2-ha island, and Lower 10.8 km. This gives shoreline development factors of about 2.2 and 1.8, or about 4 m of shore per 1,000 m² of water (project derivation).

The Maine Natural Areas Program (MNAP) treats the shoreline and its near-shore vegetation as one Lakeshore/Pondshore ecosystem of 17 communities, comprising wetlands and shore areas disturbed by ice scour, flooding or waves (MNAP, Lakeshore Ecosystem). Table 3.2-4 lists those relevant to Hot Brook, plus one stream-bank community. On a sheltered peaty margin they would form a sequence (project synthesis of the MNAP descriptions): alder thicket on the bank, sweetgale–leatherleaf fen at the waterline, pickerelweed and bulrush in the shallows, and water-lilies beyond. No Maine source measures the widths of these bands. Speckled alder, the main bank shrub, suffers about 33 % mortality if water stands at or above the root crown for 30 days or more (Fryer 2011).

Table 3.2-4. Shore and near-shore communities of Maine lakes relevant to Hot Brook (MNAP community pages; S-rank in parentheses). † seen only in a search summary.

Ice push builds the stony shores. At Gardner Lake, Connecticut, an ice-push ridge formed 0.6–1.2 m high and 1.2–3.1 m wide (Pessl 1969, as restated by Gatto, Ferrick & Calkins 2001). The thrust there came mainly from thermal expansion, not wind (Pessl 1969; relayed by a search summary). Thermal-expansion push predominates on lakes and bays less than about 4 km across, and wind push on wider ones (Gatto 1982, as cited by Gatto, Ferrick & Calkins 2001). At Lake Bomoseen, Vermont, ice moved 0.5–2.1 m inland in two events, riding up gentle shores while vertical banks resisted it (Gatto, Ferrick & Calkins 2001). The Hot Brook lakes, about 3.3–3.4 km long and 1.7–1.9 km wide, fall within the thermal-expansion range (project inference). The exposed stony band, grown over with grass, that the angler saw in the dry summer of 2025 is consistent with an ice-push lag at the winter waterline (angler observations, project field notes; project inference).

The USGS lake gauges in Maine with daily records are all on regulated lakes, and no stage record was found for any undammed Maine lake. No dam is recorded at either Hot Brook lake (state dam layer; National Inventory of Dams); any control at the outlets is probably beaver-made (project inference). Natural water-level regimes elsewhere in the Northeast are modest. In the Tully Lakes of New York, two kettle lakes with natural outlets fluctuated about 0.45 m a year, against 0.8–1.8 m in landlocked ones (Kappel, Miller & Hetcher 2001). Three Massachusetts lakes without drawdown ranged from −30 to +102 cm about their reference pools in summer, with a median of +0.5 cm (Carmignani et al. 2021). For Hot Brook the project expects a rise after the April freshet, a fall by August–September, recharge in October–November and a flat level under ice. Project estimates put the amplitude at about ±15–25 cm about the summer pool or 0.3–0.5 m a year (section 3.1.3). On a gentle shore even this matters. A 0.4-m fall bares about 20 m of a 2 % sand flat but only 2.7 m of a 15 % cobble shore (project derivation). The exposed area of Upper's north-east beach was about 1,600 m² in August 2009 and 1,300 m² in September 2023 (aerial imagery; project analysis). Level also matters to chain pickerel, which spawn over flooded marsh vegetation after ice-out (Brokaw 2001), and to loons, which nest within about 1.2 m of the water (Vermeer 1973; Section 3.8).

3.2.7 Littoral substrate and the wave base

Wave energy mainly decides where hard bottom ends and mud begins. The depth of the mud deposition boundary is set by "the several biggest storms each year". Maximum fetch or an exposure index, each combined with littoral slope, explained 89–93 % of its variance (Rowan, Kalff & Rasmussen 1992, abstract as relayed by a search summary). A relation cited from Rowan, h ≈ 1.66·√F (h in m; F, the maximum fetch, in km), reproduces the 1.68-m maximum wave base reported for Experimental Lakes Area Lake 226 (Cooley 1997). That relation was seen only via a search summary of a DFO report; the check against Lake 226 is the project's. Cooley mapped 15.9 km of shore in six Experimental Lakes Area lakes, and the results show why stony shorelines mislead. Boulder and cobble lined several kilometers of shore but covered only about 11 % of the nearshore area to 3 m depth, while detritus–sand mixtures covered about 40 %. Mixed substrate rarely extended below 2 m, "below the wave base in most areas" (Cooley 1997). The rule fails in deep, sheltered lakes, but wave action dominates in large or shallow ones (Mackay et al. 2012). Measured littoral substrate in comparable lakes is mostly fine. In 36-ha, colored (30 Pt-Co), soft-water Sanford Lake, Wisconsin, the bed from 0 to 3 m was 70.0 % silt and about 6 % gravel or coarser (Short 2001). Maine surveyed the littoral substrate of 102 lakes in 2007–2019. The published analysis gives model coefficients rather than cover values; in it, sand cover rises with shoreland disturbance (Deeds et al. 2023). In the Belgrade Lakes, sand made up about 20 % of the substrate at 0.5 m depth in front of undeveloped shores and 52 % in front of unbuffered developed ones (Bevier et al. 2024, preprint).

No substrate survey, core or diver transect exists for either Hot Brook lake. The project applied the wave-base rule to the true-scale bathymetry, taking each cell's maximum fetch over 36 directions; the median fetch was 2.0 km on Upper and 2.1 km on Lower. Critical (annual-storm) waves were taken as 77 % of the maximum, after Rowan via Cooley. On that basis the critical wave base lies at about 1.8 m, or 1.5–2.3 m depending on exposure. That leaves 23.8 % of Upper's bed and 24.0 % of Lower's above the critical base. Håkanson's dynamic ratio (Håkanson 1982, via a secondary source) gives a similar 21–24 % (project computation). Nearly the whole strip within 30 m of shore is wave-worked, 89 % on Upper and 98 % on Lower. Organic shores must therefore come from adjacent wetlands, not from shelter. Scaled by the Experimental Lakes Area and Sanford proportions, the predicted whole-lake split is 3–6 % boulder, cobble and bedrock; 4–8 % gravel and sand; 8–12 % sand–detritus mixtures; and 75–80 % mud (gyttja) and organic bottom (project estimate). This reconciles MDIFW's "mostly rocky" shoreline with its "mostly muddy" bottom. It also fits the angler's description of shores "a foot or two deep … maybe to chest height before it gets deeper, probably where the water gets darker" (angler observations, project field notes). The small hard fraction matters for smallmouth bass nests (Section 3.6) and crayfish (Section 3.7). The mud matters too: where more than 10 % of surface light reaches the bed, algae on fine sediment can hold up to 100 times the chlorophyll of those on rock and wood (Vadeboncoeur et al. 2006, abstract).

3.2.8 Coarse woody habitat

Table 3.2-5. Littoral coarse woody habitat in north-temperate lakes. Methods differ: Christensen counted logs; Marburg counted pieces crossing the 0.5-m depth contour; Merrell counted large wood from the water's edge to the 2-m contour.

* Read from a misaligned table extraction; corroborated by Weis & Sass (2011), who cite 66.3 logs/km for undeveloped northern Wisconsin lakes.

The densities in Table 3.2-5 span an order of magnitude, partly because the methods differ and because Vermont's reference sites were picked as high-quality examples (Merrell, Howe & Warren 2009). Even under one method, undeveloped lakes differed by up to 13-fold. Wood occurred at 61 % of 50-m sites, at 20–4,200 pieces/km where present. Its best predictor was downed wood on the forest floor beside the lake (partial R² 0.48–0.54), not live-tree density. Wetland-fringed and wave-exposed shores held less (Marburg, Turner & Kratz 2006). Losses persist. Christensen et al. (1996, abstract) concluded that development's losses "will affect littoral communities … for about two centuries". Large pieces can remain for several centuries (Guyette & Cole 1999, via Wolter 2012). No coarse-wood count exists for any Maine lake. Maine's 102-lake survey scored woody habitat but did not publish values (Deeds et al. 2023).

Wood is structure for fish and turtles. In one basin of Little Rock Lake, Wisconsin, about 73 % of littoral wood was removed experimentally, leaving 128 logs/km. That is comparable to lakes with 2–8 buildings/km. Within two years yellow perch, the most abundant fish, fell to very low densities. Largemouth bass shifted from a diet of 81–93 % perch to 51–55 % terrestrial prey by dry mass and grew more slowly. Benthic invertebrates did not change (Sass et al. 2006; Helmus & Sass 2008, abstract). Wood does not always limit nesting: it explained only 8–12 % of the variation in largemouth nest density along the shores of two undeveloped lakes (Weis & Sass 2011). Young smallmouth bass in Wisconsin left complex wood by August (Brown & Bozek 2010, abstract). A whole-lake wood addition produced no fish population response (Sass et al. 2012), and a multi-lake study found that added wood alone did not, on average, raise fish abundance (both relayed by search summaries). Loss of large wood also removes turtle basking sites and the vertical surfaces on which fish and amphibians attach eggs (Merrell, Howe & Warren 2009).

For Hot Brook the project expects about 150–500 logs ≥ 10 cm per km on upland spruce–fir and mixed shores. It expects the lower half of that range on wave-exposed shores, 50–150 per km or fewer on wetland shores, and 30–100 per km on developed shore. Most logs should lie within 10–15 m of the bank in water less than 1–2 m deep (project estimates). An earlier project estimate gave 300–600 per km for forested shores. None of these has been counted.

3.2.9 Shoreline development and its effects

The project reviewed aerial imagery from 2003 to 2024 alongside building footprints, E911 address points and the 2017 lidar shoreline (project analysis; confidence varies by building). It counted about 33 dwellings (camps, houses, cabins and trailer sites) within about 200 m of Upper, about half of them on the south shore, where most stand 6–30 m from the water. Upper's western shores, in timberland, have none. Lower has about 22 dwellings, mostly on its eastern shores. One stood there by 2003–05 and two more by 2009, more appeared from about 2015, and new residential lots were laid out in 2021–2024. MDIFW's 2004 statement that Lower has "no camps" therefore no longer holds. Averaged over the shoreline, this is about 2 dwellings per km on each lake (derived). That is well below the densities at which coarse wood is "often completely absent", more than 7 houses per km (Ziegler et al. 2019), or never exceeds 200 logs per km, more than 9 buildings per km (Marburg, Turner & Kratz 2006). The developed reaches themselves are denser. Seventeen docks were seen on Upper and three on Lower, 6–16 m long. They are absent from spring imagery, so they are removed each winter. Small docks are under-counted: most camps with lake access have one (angler observations, project field notes). About 11.8 km of gravel road and 1.1 km of paved Route 169 lie within 500 m of the water. Along Upper's south and east shores, Route 169 crosses 29 culverts plus a 1.5-m (60-in) culvert (MaineDOT data, via project notes). Nine mown lawns run to the water.

Development changes the littoral locally, and the studies agree on direction. Table 3.2-6 gives the only New England comparison. In Wisconsin, light under piers was a tenth of that in open water. Plant growth was 20 times greater away from piers, and invertebrates were three times fewer beneath them (Garrison et al. 2005). In 11 northern Minnesota lakes, emergent vegetation became more likely with distance from docks, and a fish index of biological integrity fell as dock density rose (Vondracek and colleagues, LCCMR final report; Dustin & Vondracek 2017). Developed shoreline carried 66 % less emergent and floating-leaf cover across 44 Minnesota lakes (Radomski & Goeman 2001, abstract). In Maine, excessive cutting of shoreland vegetation appeared to be the most frequent of 566 shoreland-zoning violations confirmed in 2002–03 (Maine DEP 2005). Fish use piers for shade but rarely to feed or nest (Engel & Pederson 1998). Where development is dense, bass recruitment and growth suffer. In five Michigan lakes with 8–22 dwellings per km, largemouth nest success fell from 0.77 to 0.45 (Wagner, Jubar & Bremigan 2006, abstract). Across 16 lakes, largemouth bass in highly developed lakes took about 1.5 more growing seasons to reach 356 mm (Gaeta et al. 2011; abstract via a search summary). No study of fish use of docks was found for Maine or New England, and none of development's effect on smallmouth bass growth.

Table 3.2-6. Littoral habitat at unbuffered developed sites (n = 151) versus reference sites (n = 234) in 40 Vermont lakes, 2005–08; 10-m transects at 0.5 m depth (Merrell, Howe & Warren 2009).

Septic systems on non-calcareous soils retain most phosphorus. Removal averaged 97 % at 13 non-calcareous Ontario sites, and plumes reached only 1–4 m (Robertson, Van Stempvoort & Schiff 2019). Maine phosphorus control plans assign shoreline lots about 0.15–0.2 kg P a year (after Dennis et al. 1992). At heavily developed ponds, shoreline development on about 1 % of the area supplied 16–23 % of the direct watershed load. At Long Pond, Belgrade, for example, the septic systems of 267 dwellings at 17.5 per shore-mile supplied an estimated 20 % of the north basin's load (Maine DEP 2008). Gravel camp roads are among the most intense sources per hectare. A Maine gravel road serving summer cottages exported 9.0 kg TP ha⁻¹ yr⁻¹, 21.6 % of it algal-available, together with 41 t ha⁻¹ yr⁻¹ of sediment (Dudley, Olson & Handley 1997). Haunted Lake, New Hampshire, is a close analog: 69 ha, 2.4 m mean depth, flushing 4.3 times a year. There the septic systems of 21 houses contributed 1.2 kg yr⁻¹, 0.6 % of a 195-kg load (NH DES 2019). For Hot Brook, the project estimates about 0.4 kg yr⁻¹ of algal-available P per seasonal camp, about 0.7 kg as TP. At Upper's flushing rate, some 73 kg yr⁻¹ of added load would be needed to raise lake TP by 1 µg/L. Twenty to forty camps would add only about 0.1–0.2 µg/L as algal-available P, or 0.2–0.4 µg/L as TP (project estimates). Camps cannot raise lake TP from about 8 to 10–20 µg/L. If lake TP is that high, the cause lies in wetland inflow or internal loading. The camps' effects are local: nearshore SRP in front of occupied camps, nitrate through septic soils, and pulses of fine sediment and particulate P at road culverts in storms and snowmelt (project inference). None of these has been measured at any Maine camp shore.

3.3 Plankton and the microbial loop

The plankton of the Hot Brook lakes has never been sampled. The public record holds no phytoplankton, zooplankton, chlorophyll, nutrient, Secchi or oxygen value for either lake; the only water measurements are the August 1949 MDIFW survey's temperatures (surface 23.3 °C, 74 °F), and neither lake has a DEP or Lake Stewards sampling station or a Water Quality Portal record (MDIFW 1949 survey sheets; Bacon & Vaux 2019; Maine DEP & VLMP 2025; a combined 2025 Lake Stewards–DEP dataset could not be checked). Everything below comes from named analog waters: the nearest DEP-monitored lakes (Baskahegan, in the same drainage; Faulkner and North, in neighboring ones), 23 eastern Maine lakes sampled by the Passamaquoddy Tribe, subsets of the EPA National Lakes Assessment (NLA) and EMAP-Northeast surveys chosen for Hot Brook-like depth, color and phosphorus, LakePulse lakes within 250 km, and a few well-studied lakes elsewhere. What matters in applying them is that both Hot Brook lakes are shallow (maximum 5.5 m in Upper, 4.0 m in Lower on the MDIFW soundings), do not stratify in summer (MDIFW), have mostly muddy beds, and probably lie near the oligotrophic–mesotrophic boundary in soft, moderately colored water (section 3.2).

3.3.1 Trophic state and phytoplankton

Maine DEP classes lakes with long-term mean Secchi 4–8 m, chlorophyll a 1.5–7 µg/L and total phosphorus (TP) 4.5–20 µg/L as mesotrophic, and above 30 SPU of color judges them by chlorophyll and oxygen rather than Secchi and TP (Maine DEP 2024 Integrated Report), because in colored Maine lakes "phosphorus is tied up with the color producing compounds" (Cowing & Scott 1980). Nearby, chlorophyll was 3.1–3.3 µg/L in Baskahegan Lake (TP 9–10 µg/L, color 46 SPU) and 3.6–5.5 µg/L in Faulkner Lake (TP 9–12, color 18) (Bacon & Vaux 2019; Maine DEP & VLMP 2025); Baskahegan's 2004 Carlson indices (37–42) sit at the oligotrophic–mesotrophic boundary (project estimate). In the Passamaquoddy record (1,932 values, 23 lakes, 2006–2023), the seven lakes most like Hot Brook (TP 9–11 µg/L, Secchi 4.0–4.9 m) had summer median chlorophyll of 2.2–3.4 µg/L (Water Quality Portal; project analysis). About 2–5 µg/L is thus the regional expectation for Hot Brook (project estimate).

No phytoplankton species list exists for Hot Brook or for any lake in Washington or Aroostook County. The nearest seasonal series, five Maine lakes in 1975, showed spring and autumn maxima of phytoplankton cell area with a midsummer minimum in three of them: Pleasant Pond fell 7.5-fold from June to July, Forest Lake 6-fold from late April to July (its spring peak 90 % Ochromonas; diatoms 86 % of cell area in June, 14 % in July), and Brewer Lake 5-fold. The authors ascribed the minima to silica depletion and diatom settling after stratification, not grazing (Cowing & Scott 1980), and Lake Stewards of Maine describe silica-driven diatom blooms after spring and autumn turnover in most Maine lakes. Whether this stratified-lake sequence holds in lakes that never stratify is unknown. The project's notes expect chrysophytes (Dinobryon, Mallomonas, Synura, Uroglena), spring and autumn diatoms, year-round cryptophytes and summer dinoflagellates and desmids, but these expectations are from general literature cited without verification; Gonyostomum semen is probably rare, its blooms needing high DOC, low pH and stratification (Johansson et al. 2013).

Most of the algae in such lakes are hard for grazers to eat. In the 48 NLA 2012 visits that best match Hot Brook (Northern Appalachian and Upper Midwest lakes 1.5–8 m deep, TP 5–25 µg/L, Ca ≤ 8 mg/L), edible algae (flagellates, small diatoms, non-filamentous greens) were a median 21–26 % of biovolume and inedible forms (filamentous and spiny algae, large dinoflagellates, large desmids, colonial cyanobacteria) 58 %; cyanobacteria were 3–5 %. Total biovolume was 0.9–1.0 mg/L fresh weight, the edible part 0.2 mg/L, and the leading genera Peridinium (19 %), Mougeotia (9 %), Staurastrum (8 %), Dinobryon (6 %), Anabaena (5 %) and Cryptomonas (3.5 %) (EPA NLA 2012; project analysis and classification). Small edible algae "do not vary systematically with phosphorus" across north-temperate lakes, while the inedible fraction does (Watson, McCauley & Downing 1992; abstract only), and in a small Vermont lake spring food was "mostly highly edible flagellates" while gelatinous greens dominated late summer (DeMott 1983; abstract only). Under ice, phytoplankton biovolume averaged 15.8 % and chlorophyll 43.2 % of summer values across 101 lakes (Hampton et al. 2017); in 16 lakes near Bridgton, Maine, winter chlorophyll was 0.4–7.2 µg/L and often peaked within a few meters of the ice, with 1–12 % of surface light reaching the water just beneath it (LEA 2026).

Cyanobacterial blooms are a growing Maine concern but not a recorded feature of lakes like Hot Brook. In a DEP probability survey of lakes over 150 acres, which excluded most of the Hot Brook region, about 11 % had detectable microcystin and 6.8 % exceeded an EPA advisory, with open-water peaks in late August and early September (Maine DEP 2024 Integrated Report). Gloeotrichia echinulata occurred in 27 of 37 low-nutrient lakes in Maine, New Hampshire, New York and Vermont, usually at fewer than 5 colonies/L (Carey et al. 2012; abstract only); Maine peaks of 58–192 colonies/L are reported (LEA 2016; relayed by a search summary; unverified). Its colonies recruit from sediment akinetes in early summer, carry sediment phosphorus into the water and are too large to graze (Carey, Weathers & Cottingham 2008), and it "appears to have been on the increase" in New England lakes (Lake Stewards of Maine). Hot Brook's broad, shallow, muddy beds probably suit that life cycle (project inference); its presence there is unrecorded.

3.3.2 Phosphorus kinetics, sinking and other losses

Dissolved phosphate in the region's clear soft-water lakes is far scarcer than routine chemistry suggests. In 17 Acadia National Park lakes (June and August 2006–2011), 59 % of low-detection orthophosphate samples from ≤ 3 m were below 0.33 µg P/L, detected values had a median of 0.62 µg P/L, and the median orthophosphate:TP ratio was 0.03; the standard ascorbic-acid "SRP" on comparable samples read a median 3.1 µg/L, a method overestimate (Water Quality Portal, NPS; project analysis). True phosphate lies "orders of magnitude" below spectrophotometric SRP (Hudson, Taylor & Schindler 2000; abstract only). This is the range of algal half-saturation constants for P-limited growth: across 61 freshwater measurements the median was 1.08 µg P/L (IQR 0.45–7.7) — diatoms 0.87 (Asterionella formosa 0.19–2.2), small greens about 0.5, Cryptomonas 0.43, cyanobacteria 5.9 (Anabaena 10–40), chrysophytes unmeasured — while half-saturation for uptake was about 20 times higher (24.8 µg P/L) and maximum P-limited growth a median 0.63/d (Edwards, Klausmeier & Litchman 2015; project analysis of their archive). Seston C:P exceeds Redfield in oligotrophic Shield lakes, and small lakes are more P-deficient than large ones (Hecky, Campbell & Hendzel 1993; abstract only). With phosphate at or below the constants, growth depends on stored cell phosphorus; a quota (Droop) model reproduced Lake Constance's vertical P depletion where a Monod model failed (Frassl, Rothhaupt & Rinke 2014; abstract only).

In Lake Erken the fast-sinking fraction settled at 1.9 m/d when diatoms and coccal cyanobacteria dominated and 0.5 m/d under cryptophytes and chrysophytes, the slow fraction at 0.02–0.04 m/d (Rodrigo et al. 1998; abstract only); nutrient-depleted cells sank about four times faster than growing ones (Titman & Kilham 1976; abstract only), and a review puts flagellates and cyanobacteria below 1 m/d or buoyant and small diatoms near 0.3 m/d (Zhu et al. 2026). In an unstratified 3–5-m lake that implies a net diatom loss of about 0.1–0.3/d in calm spells and much less under wind mixing (project estimate). Other losses can be large: in oligotrophic Castle Lake specific losses exceeded 0.80/d in May and "cannot be attributed to water transport, sinking, or grazing" (Jassby & Goldman 1974; abstract only), and chytrid epidemics can reshape succession (Ibelings et al. 2004). The bed can also reseed the water — 32 % of algal taxa in four low-productivity basins recruited from the sediment, adding 10–50 %/d, mostly where the bed was shallower than 8 m (Hansson 1996; abstract only) — and Hot Brook's whole bed probably lies within the photic zone (project inference).

3.3.3 The spring bloom and the clear-water phase

In the classic lakes Daphnia grazes the spring bloom down into a clear-water phase (CWP) in late May or June: in Lake Constance in 1993 Secchi rose from about 2–4 m to 12–15 m in early to mid-June, after which Daphnia fell from 5.6 to 0.6 × 10⁵/m² in three weeks (Straile 2002); Schöhsee grazing reached 170 % of the lake volume per day (Lampert et al. 1986; abstract only). The clear cases come from Daphnia-dominated lakes, and no CWP driven by copepods was found in the literature searched, copepods having "a much slower numerical response", and rising fish predation shifts lakes from two clearing phases to one and then none (Sommer et al. 2012; Jeppesen et al. 1997, via Sommer et al. 2012). Regional records show a much weaker dip (Table 3.3-1). In the eastern Maine medians the CWP is invisible — late-May–early-June chlorophyll only 20 % below early May's and no Secchi peak — and lake by lake the spring maximum was a median 1.5 times the following minimum; Vermont's inland lakes dipped more often, the large lakes most. Cowing & Scott's (1980) 5–7.5-fold Maine summer minima are not CWPs, being ascribed to silica and settling. Across 562 summer NLA visits to northeastern and Upper Midwest lakes, median chlorophyll was 5.1 µg/L where Daphnia was below 1 µg dry weight (DW)/L and still 3.6–4.2 µg/L where it exceeded 10 µg/L, and the lowest chlorophyll at more than 200 µg/L of Daphnia was 0.75 µg/L (EPA NLA; project analysis). For Hot Brook this points to a weak or absent CWP: a 1.5–3-fold dip for 2–4 weeks around early to mid-June in some years, none in others (project estimate).

Table 3.3-1. Spring chlorophyll maximum and the following minimum (Water Quality Portal records; project analysis). Lake-years qualified with ≥ 5 samples between day 105 and 205, a first sample by day 160 (Maine) or 152 (Vermont), and no gap over 28 d (Maine) or 21 d (Vermont). Sampling every 2–3 weeks can miss a 1–2-week CWP; and because the ratio also counts dips not caused by grazing, the Maine values bound from above the grazing-driven dips that the sampling resolves.

3.3.4 Zooplankton biomass and composition

Summer crustacean zooplankton in small northeastern soft-water lakes is dominated by copepods, and Daphnia is scarce or absent (Table 3.3-2). In 52 Maine NLA visits copepods were a median 77 % of crustacean biomass and Daphnia 1.1 µg DW/L, absent from 37 % of visits. Depth matters: in the 2012 NLA analogs, lakes under 4 m had a median 0.6 µg/L of Daphnia and never exceeded 100 µg/L, while 13 % of visits to lakes 4–8 m deep exceeded 200 µg/L (EPA NLA; project analyses). In LakePulse lakes within 250 km of Danforth, the leading taxa by pooled biomass were Leptodiaptomus minutus (28.3 %), Mesocyclops edax (18.9 %), Daphnia catawba (11.5 %), D. galeata mendotae (9.1 %), Bosmina longirostris (5.0 %), Diaphanosoma brachyurum (4.0 %), Epischura lacustris (3.5 %) and Holopedium glacialis (2.7 %) (Paquette, Gregory-Eaves & Beisner 2021; project analysis). D. catawba is the usual summer Daphnia of the NLA analogs (in 18 of 48 lakes, with by far the largest mean biomass, 37 µg/L), M. edax occurred in 34 of 48 and Holopedium in 27 (EPA NLA 2012; project analysis). In 15 brown Kejimkujik lakes, L. minutus (mean 10.2/L) and Holopedium gibberum (6.1/L) were in every lake, D. catawba in 53 % and only at pH ≥ 5.5 (Nussbaumer, Burgess & Weeber 2014; relayed by a search summary; unverified).

Table 3.3-2. Summer zooplankton in analog lakes (medians unless stated; DW = dry weight). NLA used one vertical tow at the deepest point with 150-µm (crustaceans) and 50-µm (rotifers, nauplii) nets; LakePulse one 100-µm tow; EMAP a 202/48-µm bongo. None is from Hot Brook.

The surveys differ two- to three-fold in level — NLA medians 110–150 µg/L against about 50 µg/L for LakePulse and EMAP — probably reflecting different nets and perhaps method drift (NLA medians rose from 79 to 111 to 129 µg/L across 2012, 2017 and 2022; project analysis). Across 342 large Norwegian soft-water lakes crustacean biomass was only weakly related to productivity and fish, pH and calcium deciding which herbivore dominated in the poorer lakes (Hessen, Andersen & Faafeng 1995; abstract only), whereas across 81 shallow European lakes zooplankton tracked TP and benthic and plant-associated taxa were "typically almost half" of the biomass (Gyllström et al. 2005). The analogs support a summer crustacean biomass of roughly 30–150 µg DW/L, at least half copepods, with Daphnia near 1 µg/L (project estimate). Individual dry masses in northeastern NLA lakes were 0.025 µg for a rotifer, 0.066 µg for a nauplius, 0.64–0.82 µg for Bosmina, 2.29 µg for L. minutus, 8.0 µg for M. edax and 10.9 µg for D. catawba; the median of per-visit mean lengths of summer Daphnia was 0.97 mm (IQR 0.78–1.21), and only a quarter of visits had a mean above 1.2 mm (EPA NLA; project analysis).

3.3.5 Seasonal dynamics, winter and dormancy

The Plankton Ecology Group (PEG) sequence — a spring Daphnia peak, a midsummer decline under fish predation, an autumn recovery (Sommer et al. 1986, via a course summary; Sommer et al. 2012) — appears in the analogs only as a cross-section: in the 2012 NLA analogs median Daphnia was 3.0 µg/L in June, 0.7 in July, 0.5 in August and 6.9 in September, while crustacean biomass rose from 31 to 174 µg/L (EPA NLA 2012; project analysis; different lakes each month). In Lake Opinicon, cyclopoids, nauplii and Bosmina "fell rapidly in May – early June, and did not increase again until August" (Keast 1980; abstract only). In Lake Lacawac, Pennsylvania, D. catawba was abundant in spring and late summer and Holopedium in early and midsummer (Tessier 1986; abstract only); a re-analysis gives D. catawba spring and autumn peaks and attributes its summer suppression to two cohorts of young yellow perch (Kasparson & Polishchuk 2023). The dominant calanoid, L. minutus, breeds "intensely late in winter under ice or early in spring": in unstratified Bluff Lake, Nova Scotia, adults overwintered, subitaneous eggs appeared from February and hatched in March, the first generation matured from mid-August, and from mid-October every egg was a diapausing egg (Cooley 1973; OCR-degraded thesis). In shallow Lake Simoncouche, Québec (83 ha, mean depth 2.2 m), annual mean densities were L. minutus 12.9/L, Daphnia 4.9/L, Cyclops scutifer 3.3/L and M. edax 1.0/L, the last absent from November to May (Grosbois et al. 2017).

Under ice, crustacean density averaged 27.8 ind/L against 110 ind/L in summer across 36 stations, copepods making up about 80 % of winter and 63 % of summer density (Hampton et al. 2017; shares read from a figure). At Lake Simoncouche L. minutus lipid rose from 18 % of dry weight in September to 76 % in January, and starved adults at 4 °C survived as well as fed ones for 66 days before collapsing (Grosbois et al. 2017). The lake's under-ice biomass in 2020–21 was 35 % L. minutus, 15 % C. scutifer, 12 % the rotifer Kellicottia, 11 % Daphnia and 8 % Bosmina; M. edax, Diaphanosoma, Holopedium and Leptodora were absent; nauplii were about 34 % of biomass at melt; and egg-bearing Daphnia went from none in late winter to 21 % at overturn (Boulianne et al. 2025; values read from figures). For a Hot Brook-like lake these imply about 10–20 µg DW/L of crustaceans under ice, at least 70–80 % copepods, with 1–3 µg/L of Daphnia (project estimate).

Resting eggs matter less to Daphnia than the animals that overwinter. Oneida Lake's sediments held more than 2.5 × 10⁴ (D. galeata mendotae) and 8 × 10⁴ (D. pulicaria) eggs/m², but only 0–25 animals/m² emerged a year (about 1 % of the bank), in early spring and ending nine weeks after ice-out, while 0–2.5 Daphnia/L overwintered in the water (Cáceres 1998); a 0.25-cm layer of sediment can stop hatching (Radzikowski, Sikora & Ślusarczyk 2016; abstract only). Calanoids also bank diapausing eggs: in Oneida Lake L. minutus and Skistodiaptomus oregonensis switch to them in October, about 10⁵ eggs/m² lie in each centimeter of the top 5 cm, and L. minutus eggs survive at least two years (Hairston & Van Brunt 1994; abstract only); in Bluff Lake they hatched in March after a 4 °C chill, but there the overwintering adults, not the eggs, carried the population (Cooley 1973). Cyclopoids rest as late copepodites in the sediment — Mesocyclops leuckarti for about seven months from early October (Wærvågen & Nilssen 2010) — and Epischura lacustris as resting eggs that autumn warming above 15 °C could hatch early (Chen & Folt 1996; abstract only).

3.3.6 Feeding, food thresholds and growth

Table 3.3-3. Feeding and growth parameters for the region's zooplankton groups. Values marked † were computed in the project notes from the published equation or by unit conversion.

Daphnia "has no feeding threshold or reduced filtering activity at low concentrations" (Porter, Gerritsen & Orcutt 1982; abstract only). Calanoids saturate at about one-fourteenth of the cladocerans' food level with a type-3 response in the field (Chow-Fraser & Sprules 1992), and per unit biomass clear about 0.4 times (0.3–1 times) as much as Daphnia at low food (project estimate). That conflicts with the often-cited finding that calanoid maximum clearance exceeds filter-feeding cladocerans' "by a factor of 10" (Hansen, Bjørnsen & Hansen 1997; abstract only), which is per body volume in laboratory compilations on preferred, often motile prey — a different metric. The calanoid threshold on algae is not lower than large Daphnia's (46–79 vs 40–58 µg C/L), and cyclopoid nauplii need four to ten times Daphnia's threshold food (project comparisons); cyclopoids reproduce poorly on algae alone (Hopp, Maier & Bleher 1997; abstract only), and M. leuckarti needed rotifers to complete copepodite development (Hansen & Santer 1995; abstract only). Field growth is far below laboratory growth: in three Shield-lake mesocosm experiments (chlorophyll 1.4–4.3 µg/L) mean per-capita growth across 18–20 taxa was 0.009–0.034/d, and D. pulex/pulicaria declined at −0.32/d, against laboratory rates of 0.14–0.58/d at about 20 °C (Arnott, Azan & Ross 2017; laboratory rates from Lynch 1980, via that paper).

3.3.7 What limits Daphnia in summer, and how copepods coexist with it

Calcium, the first suspect in soft water, is probably not controlling at Hot Brook. It has never been measured there; NLA 2012 lakes in Maine, New Hampshire and Vermont with alkalinity bracketing Hot Brook's (ANC 70–180 µeq/L) had a median of 2.0 mg Ca/L (IQR 1.7–2.6, n = 22) (EPA NLA 2012; project analysis), nearby Pleasant Lake 2.1 and Skitacook 3.4–4.0 mg/L, and DEP reports calcium depletion "Down East" (Maine DEP 2024 Integrated Report). D. pulex's growth rate "decreased dramatically between 1.5 and 1 mg Ca L⁻¹", warm water and low food raising the threshold to 1.5 mg/L (Ashforth & Yan 2008; abstract only); field mesocosms gave a community threshold of 0.9–1.2 mg/L and no effect between 1.2 and 2.5 mg/L (Arnott, Azan & Ross 2017). In 43 Muskoka lakes, daphniids declined in 60 % of lakes below 1.5 mg/L and 67 % at 1.5–2.0 mg/L, and rose in every lake above 2.5 mg/L (Jeziorski et al. 2008). Holopedium needs about a tenth of daphniids' calcium (Jeziorski et al. 2015). In the NLA analogs, Daphnia showed no consistent trend across calcium bands from below 1.5 to 20 mg/L (small, confounded samples; project analysis).

Food is the stronger candidate. Live phytoplankton "rarely exceed 40 % of total POC" in 109 temperate lakes; judged on algae alone Daphnia "would face severe food limitation in a majority of lakes", but not if total seston counts (Hessen et al. 2003; abstract only). The analogs' edible algae, about 0.2 mg/L fresh weight, convert to roughly 0.02–0.03 mg C/L, below the thresholds in Table 3.3-3 though light microscopy misses picoplankton (project estimate). Small Michigan lakes showed "strong energy limitation and weak and sporadic P limitation" (DeMott, Edington & Tessier 2004; abstract only); in western Lake Erie edible algae below about 4 g/m³ fresh weight suppressed Daphnia births and began the midsummer decline (Wu & Culver 1994); and across 27 Finnish lakes (DOC 2.9–36.9 mg/L) the herbivore share fell and the copepod share rose with DOC (Estlander & Horppila 2023). Predators and depth add to this. Chaoborus punctipennis was numerically abundant in all but the deepest (> 40 m) lakes of a 43-lake Maine survey (Cowing & Scott 1980); Leptodora could account for all late-June–early-July D. galeata mendotae mortality in Lake Mendota (Lunte & Luecke 1990; abstract only); M. edax may take "fifty or more prey per predator per day" (Williamson 1984; abstract only); and Shield lakes shallower than 8 m with fish held fewer daphnids and more L. minutus than deeper lakes (Keller & Conlon 1994; abstract only). A 3–5-m lake offers no deep, dark daytime refuge (project inference). As Gliwicz (2000; abstract only) put it, "food availability merely sets the rate of population increase, while the actual population density … reflects the species' vulnerability to predation".

Copepods coexist with Daphnia chiefly because fish catch them less easily and because they eat different food. In simulated suction feeding fish caught 76–96 % of cladocerans but 7–28 % of copepods (Drenner, Strickler & O'Brien 1978; abstract only). In Schöhsee mesocosms copepods suppressed large algae and freed nanoplankton while Daphnia did the reverse (Sommer et al. 2003); copepods selectively removed 20–40-µm ciliates (Zöllner et al. 2003; abstract only), ciliates "contributed substantially to the diet of female D. minutus" (Burns & Gilbert 1993; abstract only), and calanoids do relatively better than Daphnia on cyanobacteria-rich food (Richman & Dodson 1983; abstract only). Calanoid abundance varies independently of trophic state, unlike that of cladocerans and cyclopoids (Pace 1986; abstract only).

3.3.8 Rotifers, bacteria and the microbial loop

Rotifers persist through every summer month in the lakes most like Hot Brook: monthly medians of 270–469/L (6.4–8.8 µg DW/L) in the NLA analogs, 427/L in Maine lakes ≤ 6 m with TP ≥ 8 µg/L, and 453/L in the shallowest (≤ 2 m) northeastern and Upper Midwest lakes (EPA NLA; project analysis); Cooper (1941) found about 250/L in southern Maine in August 1940. Keratella (30–46 %), Polyarthra (11–27 %) and Conochilus (9–42 %) dominate summer counts (EPA NLA; project analysis), while Kellicottia and K. cochlearis are prominent under the ice at Lake Simoncouche (Boulianne et al. 2025). Daphnia roughly halves rotifer numbers without removing them: July–August medians were 450/L where Daphnia was below 1 µg/L and 178–293/L at 1–100 µg/L, and the June–August median fell to 126/L only where Daphnia exceeded 100 µg/L at a mean length above 1.2 mm (EPA NLA; project analysis). Rotifers "cannot become abundant" beside Daphnia above about 1.2 mm, while smaller cladocerans "often co-occur with them at high densities" (Gilbert 1988; abstract only); large Daphnia kill Keratella swept into the branchial chamber (Gilbert & Stemberger 1985; abstract only), whereas in Storrs Pond, New Hampshire, D. galeata mendotae "had no effect on any of the rotifers" (Wickham & Gilbert 1991; abstract only). Rotifers also subsist on less food (K. cochlearis about 0.03 mg C/L; Table 3.3-3), take 10–40 % of their food as heterotrophic microbes (Arndt 1993; abstract only), and in Keratella and Conochilus ingest particles down to 0.5 µm (Ooms-Wilms et al. 1995; abstract only). At Mirror Lake they were 16 % of zooplankton biomass but 39.8 % of production (Makarewicz & Likens 1979). An oligotrophic, acidic Norwegian lake (Gjerstadvann; TP ~8 µg/L, color 10–30 mg Pt/L) held only about 150–160/L (Wærvågen & Andersen 2017; total a project sum).

The bacterial side of the food web is known only from other lakes. Bacterial abundance follows log₁₀ cells/mL = 5.867 + 0.776·log₁₀ chl (Bird & Kalff 1984; abstract only), which at 33.1 fg C per cell (Tulonen 1993) and 3–5 µg/L chlorophyll gives about 57–85 µg C/L (project estimate). In mesohumic Lake Pääjärvi bacterial production was 1.8–13.2 µg C/L/d in the growing season and 0.2–2.9 in winter, 19–33 % of primary production (Tulonen 1993; abstract only); across systems it averages about 20 % (Cole et al. 1988; abstract only). Colored lakes run partly on land carbon: terrestrial carbon met about 90 % of bacterial demand in a humic lake (Hessen 1992), bacterial biomass was 35–70 % allochthonous in small forested lakes (Kritzberg et al. 2004), detritus supplied 46–82 % of zooplankton body carbon (Hessen, Andersen & Lyche 1990), and mean allochthony in 18 Québec lakes was 0.31 in cladocerans, 0.18 in cyclopoids and 0.16 in calanoids (Berggren et al. 2014; all abstracts only). How bacterial carbon moves up depends on the grazers: in Upton Lake D. galeata "consumed most of the bacterial production" except in winter and autumn, and ciliates and rotifers "were never important bacterial grazers" (Pace, McManus & Findlay 1990; abstract only), whereas in eutrophic Lake Oglethorpe heterotrophic nanoflagellates did 49–81 % of bacterivory (Sanders et al. 1989; abstract only). During Lake Constance's spring bloom more than half of primary production passed through the microbial loop (Weisse et al. 1990; abstract only), and ciliates there dipped 2.5–4-fold at the CWP (Gaedke & Wickham 2004). Parasitic fungi open a further route to zooplankton (Kagami et al. 2014), and methane carbon reaches Daphnia (Jones & Grey 2011). In two small shallow German lakes, bacterial production (294–386 g C/m²/yr) exceeded the net primary production of phytoplankton, macrophytes and epiphyton (122.9–208.7; reeds added a further 92–214), while crustacean zooplankton produced 3.47–12.77 and fish 0.023–0.047 g C/m²/yr (Mehner et al. 2022).

3.3.9 Fish and zooplankton

The best-quantified fish–Daphnia threshold comes from Oneida Lake, where D. pulex "could tolerate predation by 10 kg of young fish·hectare⁻¹" but could not compensate once young-of-year (mostly yellow perch) exceeded 20–40 kg/ha; in 1975 and 1977 their consumption exceeded Daphnia production and the daphnids disappeared, while in 1976 it stayed below 27 % of production and Daphnia persisted (Mills & Forney 1983; abstract only). An EPA report quotes the threshold as 14,000 age-0 yellow perch/ha in August (Mills et al. 1987; original not read). In Lake St. George limnocorrals Daphnia collapses coincided with young perch at 30–50 kg/ha (McQueen & Post 1988), and in western Lake Erie young yellow perch accounted for all D. galeata mendotae mortality from 4 to 11 July (Wu & Culver 1994). Larvae barely affect small zooplankton, but 0+ juveniles depress large cladocerans and copepods, "especially during late summer and autumn" (Mehner & Thiel 1999; abstract only). Size follows predation: after blueback herring entered Crystal Lake, Connecticut, the modal crustacean length fell from 0.785 to 0.285 mm (Brooks & Dodson 1965), and a mean crustacean length near 0.8 mm (153-µm net) is used as a sign of a balanced fish community (EPA SOLEC 2007). When bass returned to Wintergreen Lake, Michigan, planktivores fell two orders of magnitude, zooplankton rose about tenfold and large Daphnia returned (Mittelbach et al. 1995; abstract only), but phytoplankton responses to such cascades are strong in only about a third of experiments (Brett & Goldman 1996; abstract only).

Two Hot Brook fish make this coupling specific: white perch are a principal fishery with a 25-fish daily bag (MDIFW lake survey sheets), and sea-run alewife enter Lower (MDIFW 2024; section 3.5). In Oneida, young white perch took 2–12 times more prey biomass per gram than young yellow perch in 1980, and their growth tracked late-summer D. pulex (Prout, Mills & Forney 1990; abstract only). Adult white perch in Lake Champlain's Missisquoi Bay ate Daphnia heavily only when it exceeded 20/L and half the zooplankton (Couture & Watzin 2008; known second-hand, original not reached), and in southern Maine in 1940 plankton crustaceans were about 6 % of the food volume of white perch ≥ 15 cm, though over 100,000 water fleas were counted in their stomachs and the share ranged 0–98 % between lakes (Cooper 1941). In the Belgrade Lakes, young white perch selected Daphnia; North Pond's mean cladoceran length fell from 0.72 mm in June to 0.41 mm by late summer, which the author attributed to alewives (Trinko 2008; abstract only). Alewives stocked into Lake George, Maine, at 6 adults per surface acre cut cladoceran density and raised rotifers, both recovering by the third year after stocking stopped (Kircheis et al. 2002); in Damariscotta Lake daphniids "dropped sharply after juvenile alewives appeared" (Blair-Glantz, Wilson & Willis 2009; abstract only); and across eight Maine lakes young alewives reduced cladoceran and copepod size (Demi et al. 2015; relayed by a search summary; unverified). Whether white perch or the alewife run shape Hot Brook's zooplankton cannot be judged without fish densities and plankton samples from the lakes; the analogs' scarce Daphnia and small cladocerans fit strong planktivory, but equally shallow depth, soft water and poor food. Section 5 takes up how the model's plankton compares with these field values.

3.4 Aquatic plants and the littoral zone

Both Hot Brook lakes are shallow enough that, in clear water, rooted plants could reach almost all of their beds (on the model's 0.25-m waterline: Upper mean 3.0 m, maximum 5.5 m; Lower mean 2.6 m, maximum 4.1 m). How much of the bed is vegetated is therefore set less by depth than by water color, waves, ice and substrate. The plants are also among the least documented parts of the lakes, so this section relies on soft-water, colored lakes of Maine and the north-east, and states for each topic what is known, or only estimated, at Hot Brook.

3.4.1 What is recorded at Hot Brook

The only plant specimens ever collected from either lake are two stoneworts, Nitella flexilis (aff.) and N. tenuissima (aff.), taken on 15 July 2015 about 10 m from shore at Upper's south end during the regional charophyte survey of Sleith et al. (2018). The "aff." marks both identifications as tentative (herbarium specimens via GBIF). Nothing has been digitized for Lower, the connecting brook or the outlet. The MDIFW lake-survey sheets (1949, revised 1965 and 2004) record no plants, but describe a shoreline "mostly rocky with some areas of sand … and mud", a bottom "mostly muddy", and wetlands on Upper's south and west sides and Lower's north and east (MDIFW 1949/1965/2004). Both lakes are listed with no known invasive aquatic plant (Lakes of Maine, search summary; Maine DEP Lake Vulnerability Analysis); the nearest infestation is variable water-milfoil, found in 2019 in Big Lake ~54 km to the south (Lake Stewards of Maine n.d.; UMaine Cooperative Extension 2020). With no water measurement since August 1949, the Secchi depth and color on which every depth limit below depends are known only from nearby analogs: Secchi ≈ 3.5–4.5 m (mean ~3.9 m), color ~18–46 PCU and alkalinity 5.5–8.4 mg/L CaCO₃ in Faulkner, North and Baskahegan lakes (Maine DEP & VLMP 2025).

Near-infrared NAIP aerial photographs (flight year not recorded) show surface-reaching vegetation, floating leaves and emergents together, on 2.3 ha (0.6 %) of Upper and 0.84 ha (0.3 %) of Lower. The vegetated cells hug the shore (median 15 m out; 90th percentile 45 m in Upper and 15 m in Lower) at model depths of 0.3–1.0 m (10th–90th percentile), the largest cluster at Upper's south end beside the inlet wetlands and smaller ones in Upper's western coves and on Lower's shores (project analysis of NAIP imagery). Submerged beds are invisible to this method, and the National Wetlands Inventory (a 1980s map whose report is dated August 1983), which maps no aquatic beds anywhere in the region, adds nothing about the beds themselves. Angler observations (project field notes; identified from photographs by eye) add bulrush (Schoenoplectus, possibly Juncus) in front of a sedge fringe; ribbon-like leaves (Potamogeton epihydrus, Sparganium or possibly Vallisneria) over sand in the brook's slow runs; fine-leaved shoots (bladderwort or native milfoil) in the brook and, at a brook mouth, a tuft on a riffle stone that may be Fontinalis or lodged drift; and one cobble-and-gravel shore with no rooted plants out to ~0.6 m.

Herbarium and iNaturalist records within ~40 km include the floating-leaved Nuphar variegata, Nymphaea odorata, Brasenia schreberi, Nymphoides cordata and P. epihydrus; five bladderworts (Utricularia); the isoetids Eriocaulon aquaticum, Isoetes lacustris and I. tuckermanii; and emergents including Schoenoplectus, Pontederia cordata and Sagittaria. Pickerelweed was the most widespread aquatic plant in the wider region (22 of 151 sites, effort-biased), and no invasive milfoil was recorded (project compilation of GBIF and iNaturalist records). Within ~50 km, water lobelia (Lobelia dortmanna) and P. robbinsii have only three sites each (e.g. water lobelia at Topsfield in 1986, P. robbinsii at Lambert Lake in 1908) and P. amplifolius and Vallisneria one each; all four are judged likely or possible at Hot Brook but are unrecorded there (project inference).

3.4.2 The soft-water flora and its growth forms

Water chemistry selects the flora. Across 321 New England sites (Hellquist 1980), the softest group of pondweeds (0–18.3 mg HCO₃⁻/L) held P. confervoides, P. bicupulatus, P. oakesianus and P. pulcher (medians 4.3–10.4 mg/L), and ribbon-leaf pondweed (P. epihydrus, median 13.4, n = 154) sat at the soft end of the next (Hellquist's group assignments do not follow the medians alone). Robbins' (P. robbinsii, 25.6) and large-leaf pondweed (P. amplifolius, 28.1) had higher medians but ranged down to 3.7–4.3 mg/L, while P. crispus (93) and P. pectinatus (113) sat far above. Baskahegan Lake, in Hot Brook's drainage, has 5.5 mg/L CaCO₃ (≈ 6.7 mg HCO₃⁻/L), within the softest group (Maine DEP & VLMP 2025; project conversion). The mechanism is carbon supply: in soft water plants that use CO₂ alone, above all the isoetids, are favored over bicarbonate users. Water lobelia takes 46–98 % of its CO₂ from the sediment (Winkel & Borum 2009).

Maine's soft-water lakes carry a few recurring communities. In 30 relatively undisturbed Maine lakes Cameron (2000) found a pipewort–water lobelia bed on mineral soil shallower than 1 m; a water-lily bed on organic soil at 0.5–1.5 m (Nymphaea odorata with Utricularia purpurea, Nuphar variegata and Eleocharis robbinsii); a pickerelweed bed; and a circumneutral suite of Vallisneria, Elodea and pondweeds, with 3–26 species per lake. The Maine Natural Areas Program describes these beds as small, often linear patches, with pickerelweed marsh typically above 50 % cover and bulrush beds often at inlets and outlets (MNAP n.d., community fact sheets; depths in Table 3.4-1). In Highland Lake, southern Maine (85 plots, 2001), P. epihydrus occurred in 81 % of plots, Eriocaulon 77 %, Brasenia 31 %, water lobelia 24 % and Nuphar 20 % (MABP Table 6.2.2). Older surveys stress sparseness: across 43 Maine lakes, "highly colored lakes had plant communities primarily composed of Scirpus and floating-leaved plants" and rocky, sandy littorals only sparse Eriocaulon (Cowing & Scott 1980), and in 53 southern Maine lakes vegetation was "generally very scanty", mostly emergent and around inlets and outlets (Cooper 1941).

The deep, low meadow expected in these lakes is Nitella and aquatic moss, not Chara (project inference). In a survey of 740 New York and New England water bodies (2014–15), Nitella flexilis was the commonest charophyte (147 of 363 sites with charophytes) and "a generalist", whereas Chara contraria was "not detected across thousands of square kilometers in interior portions of Maine, where softer water occurs" (Sleith et al. 2018). The survey's eastern-Maine specimens show Nitella at almost every lake visited in July 2015 and Chara at only four lakes, two of them in central Maine (GBIF records of that survey; project compilation). Table 3.4-1 summarizes the five growth forms.

Table 3.4-1. Growth forms of soft-water Maine lakes. Depth maxima are from Wisconsin boxplots (±10 cm) unless stated.

Sources: Nichols 1999; Sheldon & Boylen 1977; Cameron 2000; MNAP n.d., community fact sheets; Heisey & Damman 1982 (abstract); Esser 1995; Klok & van der Velde 2017; Kunii 1999; Duffy & Jackson 1994; Clesceri et al. 1975; Go Botany n.d.; Sand-Jensen et al. 2005; Sand-Jensen & Søndergaard 1979; Chappuis et al. 2016; charophytes, Alirangues Nuñez et al. 2023.

3.4.3 Depth limits: light, color and Secchi depth

The maximum depth of colonization (Zc) scales with clarity. Across 27 Finnish, 26 Florida and 55 Wisconsin lakes, Canfield et al. (1985) fitted log Zc = 0.61 log SD + 0.26 (r² 0.56; 95 % limits 46–236 % of the prediction), but the 27 mostly humic Finnish lakes followed a lower line, log Zc = 0.51 log SD + 0.18 (r² 0.82). Chambers & Kalff (1985; abstract) put the angiosperm limit at ~21 % of surface light and the charophyte limit at ~11 %. The closest regional analog is Duarte's (1987) thesis on 25 southern Québec and north-eastern US lakes: Zmc = 1.9 + 0.63·SD (r² 0.76), and the depth of maximum biomass Zmb = 1.1 + 0.4·SD (r² 0.79). Rosettes have the shallowest limit of any growth form (Middelboe & Markager 1997; abstract): water lobelia in 25 Polish soft-water lakes reached a median of only 0.8 m (0.1–2.2 m), about SD/3, at ~43 % of subsurface light (Borowiak et al. 2017).

Color moves all of these relations shallower. "For any Secchi depth, aquatic angiosperms colonized to greater depths in lakes with low colour" (Chambers & Prepas 1988; abstract), because in stained water more light is absorbed below the Secchi depth: in Alaskan lakes Kd·SD has a median of 2.70 (1.81–3.83) in stained lakes against 1.86 in clear ones (Koenings & Edmundson 1991). In Kejimkujik, Nova Scotia, clear Beaverskin Lake is vegetated across most of its bottom to 6.5 m, while humic Pebbleloggitch Lake has only a fringe shallower than ~1.8 m (abstract, search summary), and across 12 clear and 13 humic Polish lakes biomass per square meter of population area of Isoetes lacustris fell from 85.4 to < 0.001 g DW m⁻² while water lobelia held (8.4 → 6.2) (Szmeja & Bociąg 2004; abstract). No colonization-depth survey of a humic lake in Maine, New England or Québec was found.

At the analog Secchi depths of 3.5–4.5 m, the clear-water regressions put the angiosperm limit at 3.9–5.8 m, deep enough to reach most of Upper and all of Lower. The humic Finnish line gives only 2.9–3.3 m, and the depth of 21 % surface light in stained water only 2.0–2.6 m (Table 3.4-2). The spread matters most for Lower, whose bed is a saucer mostly 2.7–3.7 m deep (§2.2) at the angiosperm margin: a ±0.7 m change in the limit moves its vegetated area between about one-third and three-quarters of the bed (project estimate). For water darker than the analog mean (Secchi ~3 m, color ~45 PCU), the project's working limits are a pondweed canopy to ~2.0–2.7 m, a Nitella and moss meadow to ~2.8–3.5 m, water lobelia and pipewort to 0.9–1.2 m, and floating leaves to ~2.5 m (project estimate). Measured maximum plant depths in two Midwestern lakes were 3.4 m in Lake Vermilion, Minnesota (Secchi ~2.6 m), and 4.0 m in Rock Lake, Wisconsin (Table 3.4-3).

Table 3.4-2. Predicted maximum colonization depth (m) for Hot Brook at the analog Secchi depths, and the share of each lake's bed shallower than the limit at SD 3.9 m. Project arithmetic from the published equations; bed shares from the model's bed (1949 MDIFW soundings, the lidar shoreline and shallows derived from aerial photographs).

3.4.4 Shallow limits: waves, ice and substrate

Plants also stop short of the shore. In Duarte's 25 lakes the shallowest submerged plants grew at 0.75–1.25 m regardless of slope, exposure or transparency, which Duarte (1987; published as Duarte & Kalff 1990) attributed to ice scour; Chambers (1987; abstract), also in southern Québec, instead found the minimum depth equal to the depth of wave mixing (r = 0.86). Both plausibly act at Hot Brook, which carries ice for some 135–150 days (section 3.1) and has maximum fetches of ~1.5–2.8 km over most of its bed (project analysis). Wave exposure explained 86 % of the variance in cover in a large New Zealand lake (Riis & Hawes 2003; abstract), and in the Great Lakes cover rose from 0 % to 30 % to 78 % from coarse, exposed sites to fine, sheltered ones (Randall et al. 2004). Isoetids are the exception, increasing with exposure (Keddy 1983, 1985; abstract), though heavier ice disturbance favors small isoetid species over large (Hellsten 2000; abstract).

Slope and sediment complete the picture. In Lake Memphremagog biomass fell with slope as slope^−0.81, and in 25 lakes no plants grew on slopes steeper than ~14.8 % (Duarte & Kalff 1986; Duarte 1987). Growth falls 10- to 20-fold as sediment organic matter rises to ~20 %, and clean sands are also poor (Barko & Smart 1986). Fine mud settles only below a deposition boundary set by the year's largest storms, 2.291·F^0.549 m for a maximum fetch F in km (Rowan et al. 1992, 54 lakes), which is 1.6–4.2 m for fetches of 0.5–3 km and ~3.4 m at 2 km (project arithmetic). A per-cell analysis of the model grid, using instead a wave-theory form (storm waves at ~77 % of the maximum) and each cell's maximum fetch (median ~2 km), puts the boundary much shallower, at ~1.5–2.3 m (project estimate); the two estimates have not been reconciled. With the "mostly rocky" MDIFW shoreline, the angler's description of shores "only a foot or two deep" out to a shelf (angler observations, project field notes), and the bare cobble shore, this suggests a sparse band along most shores to ~0.75–1.25 m, rosettes on the sand and gravel of the shelf, pondweeds and lilies in sheltered coves and on mud below the deposition boundary, and the Nitella meadow deeper still (project inference). No substrate survey, core or diver transect exists for either lake (section 3.2).

3.4.5 Areal cover and patchiness

In a compilation of 139 mapped lakes, mostly in Europe and North America, macrophytes covered on average 44.1 % of lake area (n = 104), submerged plants 35.0 % (n = 82) and emergents 12.3 % (n = 77), with underwater light "the best descriptor of the cover and biomass of submerged plants" (Duarte, Kalff & Peters 1986). Applied to Hot Brook their equations disagree widely, giving about 20 % of each lake from lake area alone but 60–75 % from mean irradiance (project arithmetic).

Measured cover in comparable lakes is lower. In the EPA National Lakes Assessment, plots 10 m offshore at 10 stations per lake in 16 shallow natural Maine lakes had a median nearshore cover of 0.19 (interquartile range 0.12–0.50) in 2012, and 14 lakes in 2007 also 0.19 (0.09–0.30). Twenty-five colored, soft-water analogs in the Northern Appalachians and Upper Midwest had a median of 0.39 with plants present at almost every station, and their rocky stations averaged 15 % cover against 39 % on silt or organic bottoms (project analysis of NLA 2007 and 2012 data). Whole-lake surveys of soft, colored Adirondack lakes mapped beds on 3–39 % of lake area (AWI 2013). Beds are patchy and heavy-tailed: Little Tupper Lake's 27 beds ranged from 40 m² to 5.8 ha, and across lakes the one or two largest beds held 42–74 % of the mapped bed area (AWI 2013; project tabulation). Point-intercept surveys measure frequency (presence) rather than area, and presence overstates cover more the coarser the sampling grain: at the NLA plot scale presence exceeded mean cover 2.5–5-fold, whereas at rake scale, taking a mean cover of ~0.4–0.6 inside occupied rake footprints, a 40 % frequency corresponds to roughly 15–25 % areal cover (project estimate).

Combining the Maine NLA cover, depth-band frequencies rescaled for stained water, and the model bed's hypsography (Upper: 11.7 % of the bed shallower than 1 m, 26.5 % < 2 m, 43.9 % < 3 m; Lower: 16.0, 24.7 and 55.1 %), the project estimated that plant stands cover ~15 % (8–25 %) of Upper's bed and ~17 % (8–27 %) of Lower's. Any vegetation, including the Nitella and moss meadow, would cover ~35–40 % (25–55 %) and ~45 % (30–65 %). Confidence is low to moderate (±~8 points), and an upper bound of ~30 % for stands is held with moderate confidence (project estimate; Table 3.4-3). Two independent lines agree on the surface-reaching fraction: Maine NLA floating and emergent cover in the 0–10 m shore band implies ~0.3–0.5 % of lake area (project arithmetic), against 0.3–0.6 % on the photographs. By either measure Hot Brook's shore looks like a typical Maine nearshore, not a weedy one.

Table 3.4-3. Macrophyte cover in comparable lakes and estimates for Hot Brook.

3.4.6 Bed structure and the seasonal cycle

Standing crops in soft, nutrient-poor water are modest (Table 3.4-1); hard-water Minnesota marl lakes, at 171–637 g DW m⁻² in 0–1 m of water, give an upper bound (Carter et al. 1993). Isoetids produce only 5–425 g DW m⁻² yr⁻¹, against 300–1,500 for other submerged plants (Chappuis et al. 2016, citing a review). Stem density differs by orders of magnitude between forms, from ~20 m⁻² in water-lily beds and ~200 in watershield to hundreds or thousands in pondweed and isoetid carpets (Table 3.4-1), which matters for fish (section 3.4.8). Canopies are low: in Lake George large-leaf pondweed reached the surface in 0–3 m of water but was only 1–3 m tall at 3–5 m, and Robbins' pondweed stayed below 0.5 m (Lake George reports, RPI). Floating leaves are short-lived, 38–48 days on average in European lilies (Klok & van der Velde 2017) and 25–28 days in watershield (Kunii 1999). No north-eastern standing crop was found for Brasenia, Nymphaea odorata, Nymphoides, Eriocaulon, Isoetes or Nitella.

In Lake George large-leaf pondweed starts right after ice-out in mid-April, has floating leaves by early July and dies back in early autumn, while Vallisneria grows only from mid-June to September (Lake George reports, RPI); ten of 26 species there kept their density and productivity through winter, with maximum winter photosynthesis 10–20 % of summer's (Boylen & Sheldon 1976). Watershield in a New Hampshire pond began growing on 24–25 April and formed winter buds from 30 July; about 35 % of the buds overwintered, and "no individual plant lasts more than one season" (Adams 1969). In Maine pickerelweed beds die back below the surface at the end of the season (MNAP n.d., community fact sheets), and pickerelweed, bulrush and lilies are expected to do so in late September–October (project estimate), whereas isoetids are evergreen, water lobelia making and losing leaves all year (Farmer & Spence 1987). With ice-out in late April and ice-on around 1 December (section 3.1), Hot Brook's evergreens should photosynthesize from ice-out, floating leaves reach the surface in late May–June, biomass peak in late July–August and senescence come in October (project estimate). No Maine plant phenology was found.

3.4.7 Plant chemistry that reaches the lake

Isoetid lawns oxygenate their sediment. In a water lobelia lawn in a Swedish sandy lake, more than 97 % of photosynthetic O₂ left through the roots, root respiration used only about a fourteenth of it, and porewater at 4 cm stayed oxic day and night (Sand-Jensen et al. 2005); the oxic layer reached 4.3 cm under Lobelia against 1.6–2.2 cm under pondweeds (Raun et al. 2010; search summary). The effect reaches the water: in Lake Kvie, Denmark, where isoetid lawns cover 55 % of the lake, annual P release from vegetated sediment was only 5 % of the plants' P requirement, and the authors judged isoetids "important for keeping lakes in an oligotrophic state" (Andersen & Olsen 1994; abstract via a search summary). Lobelia lawns were also a net carbon sink where bare sand was a source (Ribaudo et al. 2017; search summary), and isoetid rhizospheres denitrified 24–30 µmol N m⁻² h⁻¹ against 1.5–5 under pondweed or seagrass (Ottosen et al. 1999; search summary). The plants are fragile: water lobelia dies or loses anchorage at about 2.4 % degradable organic matter in the sediment (Sand-Jensen et al. 2008; search summary, method unconfirmed), which confines it to mineral shelves. No measurement of root O₂ release was found for Eriocaulon or North American Isoetes.

Bladderworts are rootless carnivores. Their traps, 1–4 mm long, make up 10–55 % of plant dry mass (Adamec 2006) and last only ~30 days (Sirová et al. 2009), and most catch little: in eight Austrian populations traps averaged about one animal each among roughly ten times as many algae and pollen grains (Koller-Peroutka et al. 2015). In two northern Michigan lakes the prey were mostly littoral cladocerans and ostracods (Wallace Hoffelt & Doubek 2023; abstract via a search summary). Prey can supply much of the plant's nitrogen, 71 ± 18 % in field U. vulgaris by δ¹⁵N (Koller-Peroutka et al. 2015), but trap investment falls with phosphorus loading (Kibriya & Jones 2007; abstract) and stops in deep shade (Adamec 2015). No field comparison of zooplankton inside and outside a stand was found; the project's arithmetic spans from under 1 % to tens of percent of the local crustacean stock removed per day (project estimate).

Rooted plants also move phosphorus and oxygen. They draw P from the sediment, all of it in mesotrophic water (Carignan & Kalff 1980), and release it mainly as they decay (Barko & Smart 1980); in Lake Wingra decaying milfoil carried about half of the dissolved P and organic matter moving from the littoral to open water (Carpenter 1980). Inside dense vegetation (~50 % cover, bays 0.6–1.3 m deep) dissolved O₂ fell below 1 mg/L while open water stayed above 8 mg/L (Miranda & Hodges 2000), but such canopy stratification needs plants filling at least half the water depth (Vilas et al. 2017), which low soft-water canopies rarely do outside the shallowest coves (project inference). Beds also accumulate 2–20 times more sediment than open bottom (Benoy & Kalff 1999). Extracts of Nymphaea odorata and Brasenia strongly inhibit other plants (Elakovich & Wooten 1987, 1989; abstracts), but there is no evidence that allelopathy matters in lakes like these.

3.4.8 Weed beds as habitat and refuge

A bed's value as refuge depends on its stem density. Largemouth bass caught bluegill as easily at 50 stems m⁻² as in open water but almost never at 250 or 1,000 (Savino & Stein 1982). In Shoup & Broderius's (2018) review 0 and 50 stems m⁻² were ineffective as refuge and 250–500 effective, though in their own trials with 40-mm juvenile bass only 250 stems m⁻² cut predation and 500 did not differ from open water. Pike entered artificial structure only below 600 stems m⁻² (Bean & Winfield 1995; abstract). A meta-analysis of 95 mostly laboratory studies found that habitat complexity reduces predation (Hedges' g = −0.85, 95 % CI −1.14 to −0.55) (Chang & Todd 2023). Refuge has a price: small bluegill sheltering from bass in vegetation fed at about a third of the open-water rate and grew 27 % less (Werner et al. 1983). On these thresholds lily (~20 stems m⁻²) and watershield (~200) stands are too sparse to impede predators, their value being shade, while dense pondweed and isoetid turf offer the most refuge, the turf only at the bottom (project inference). Nearly all of these figures come from largemouth–bluegill or pike–roach experiments, often with artificial stems; none tests smallmouth bass, chain pickerel or yellow perch in soft-water lily–pondweed beds.

In the field, fish use of beds depends on bed density and on the species. Pop nets caught 9.8 fish m⁻² in intermediate mixed beds against 1.8 in dense Hydrilla (Killgore et al. 1989). Yellow perch were commonest in patchy, species-rich vegetation and smallmouth bass on bare ground (Weaver et al. 1997), while small (< 100 mm) white perch in a Virginia–North Carolina reservoir were collected only in unvegetated littoral areas (Harris 2006). Minnesota's game fish did best at 10–60 % basin cover of submerged plants, preferably 18–40 % (Valley et al. 2004). Esocids are the most plant-bound: in Dutch waters the standing stock of pike shorter than 54 cm was "tied to a maximum per unit of vegetated area", ~110 kg per hectare of vegetated surface at one site (Grimm 1984), and in Maine young chain pickerel are "largely confined to these patches of aquatic vegetation" (Brokaw 2001, updated 2008). For zooplankton, plants must fill more than 15–20 % of the water volume to give refuge (Schriver et al. 1995), perhaps 30–40 % (Perrow et al. 1999), and across 31 North American and New Zealand lakes North American Daphnia stayed pelagic (Lauridsen et al. 1999). With Hot Brook's lake-wide plant volume estimated at ~2–5 % (15–40 % inside cove beds), plant refuges for zooplankton would be local (project estimate). Grazing on the plants by moose, deer and muskrat is treated in section 3.8.

3.4.9 Fauna of floating leaves and the littoral invertebrates

Floating leaves carry a specialist fauna: the water-lily leaf beetle Galerucella nymphaeae; reed beetles (Donacia), whose larvae tap the air channels of the rhizomes; leaf-mining and petiole-boring chironomids; the waterlily aphid; and pyralid caterpillars (Parapoynx, Elophila) and Bellura borer moths, with ladybirds hunting them on the leaves (Harms & Grodowitz 2009, via Illinois Wildflowers) and fishing spiders hunting surface insects from emergent stems (Kreiter & Wise 1996). Parapoynx maculalis was "very common wherever either of the waterlilies grew" on a Massachusetts lake (Forbes 1910). Galerucella, essentially univoltine in a north-eastern US study (Tauber et al. 1996; abstract), shortens leaf life where abundant: in a Georgia river infested Nuphar leaves lasted ~17 days against more than six weeks at a beetle-free pond (Wallace & O'Hop 1985; abstract). In three Dutch waters Donacia touched 54–73 % of lily leaves but removed ≤ 2 % of their area, and wave damage hit 60–80 % of leaves at exposed sites but none at a sheltered acid pool (Klok & van der Velde 2019). Per gram, floating-leaved plants carried the fewest invertebrates of three plant architectures compared in Virginia (Rielly 2009; abstract). Sunfish suck prey from the undersides of pads (Postupaka et al. 2021; abstract), but no diet study was found that names any pad insect in a sunfish, perch, bass or pickerel, and none of this fauna has been recorded at the lakes.

On submerged plants, epiphytic invertebrates numbered 25–676 per g of plant dry mass (mostly 36–100), with gastropods over 60 % of the biomass at three of four St Lawrence and Adirondack sites (Wilson & Ricciardi 2009). Abundance rises with plant biomass, sediment organic matter and phosphorus and falls with depth (Cyr & Downing 1988a, b; abstracts). Beds multiply the fauna of the ground beneath them: pondweed (P. nodosus) beds held 34,000–44,000 invertebrates m⁻² against 1,700–3,300 on bare sediment (Miller et al. 1989), and the amphipod Hyalella reached 17,459 m⁻² on a sedge shore of Dock Lake, Ontario, but was absent below 1.5 m (Lindeman & Momot 1983). Non-snail prey on plants fell ~2.5-fold from May to August in a Michigan lake (Mittelbach 1981). The best small-lake production benchmark is oligotrophic Mirror Lake, New Hampshire, whose whole zoobenthos averaged 2.5 g DW m⁻² and produced ~14 g DW m⁻² yr⁻¹, a P/B of ~5.6 (Strayer 1985, abstract; project arithmetic). Color lowers prey quality: across 26 Finnish lakes chironomids were more abundant in humic lakes, and several taxa held less polyunsaturated fatty acid there than in clear lakes (Kesti et al. 2021; abstract).

In soft water the snails, which dominate epiphytic biomass elsewhere, are reduced. In 26 soft-water Vermont lakes "no other Gastropods were found in lakes with median pH below 6.00 and calcium levels below 1.5 mg/l" apart from Fossaria and Ferrissia, and crayfish occurred in only two lakes, both with pH well above 6 and calcium above 2.0 mg/L (Fiske 1987); a circumneutral Wisconsin weed bed, by contrast, held 473–1,765 snails m⁻² (Brown 1997). Hot Brook's calcium is unmeasured. Project estimates range from ~1.5 to ~5 mg/L, and Skitacook Lake, ~47 km away, measured 3.4–4.0 mg/L (EPA NLA 2012 and 2017 data), so a modest, small-bodied, thin-shelled snail fauna is expected (project inference). Angler observations record snails present and crayfish "not common". Apart from odonate specimens (Upper 2005, Lower 1999), a 2005 report of the freshwater jellyfish Craspedacusta sowerbii in Lower and a state record of the yellow lampmussel in Upper (section 2), no invertebrate has been recorded in either lake, and none has been sampled from the bed or the plants.

3.4.10 Fish eggs on plants versus bare bed

The two earliest spawners use plants grown the year before. Yellow perch (~7–13 °C) and chain pickerel (~6–11 °C) spawn one to three weeks after ice-out, so their substrate is last year's standing dead stems, wood and overwintering plants (project synthesis). A yellow perch lays its whole year's clutch as a single ribbon (Ward & Barnes 2020), so a bad site "may easily result in complete failure" (Čech et al. 2012). Structure roughly doubles hatching: in a hatchery trial, eyed eggs held on artificial vegetation hatched 98.6 % (SE 0.6) against 46.2 % (SE 23.7) on a bare bottom, a difference attributed to folding, suffocation, siltation and bacterial degradation (Ward & Barnes 2020), and in the field dead Eurasian perch embryos were significantly commoner on the lake bottom than on vegetation (Smith et al. 2001; abstract). Perch prefer rigid, dead structure: in a Czech lake live submerged plants were mostly avoided in favor of dead pondweed, wormwood and reed (Čech et al. 2009; abstract), and 96 sunken trees took two-thirds of all strands (Čech et al. 2012). Color can set the depth: in a low-DOC Pennsylvania lake ~92 % of perch eggs lay deeper than 3 m and UV at the surface killed all eggs, whereas in a higher-DOC lake 76 % lay shallower than 1 m (Huff, Grad & Williamson 2004; abstract).

In Maine chain pickerel spawn "shortly after ice-out when adults move in to swampy areas containing flooded vegetation" and broadcast their eggs without a nest (Brokaw 2001, updated 2008). For the surrogate northern pike, egg densities in a Wisconsin lake were greatest in dense mats of sedge and wild-celery remains (Priegel & Krohn 1975), yet in an English gravel pit pike eggs hatched better on sand and silt (11 %) than on plants (2.5 %) in 1986, even though age-0 survival was higher in a densely vegetated lake (Wright & Shoesmith 1988; abstract). Banded killifish spawn later and on live plants: in Newfoundland from late June to mid-August at 19–23 °C, each egg hanging on an adhesive thread and Potamogeton the substrate most used, in habitat whose commonest plants were water lobelia, quillwort and pipewort, with purple bladderwort and ribbon-leaf pondweed, almost the flora expected at Hot Brook; sedimentation smothering vegetation is the main threat named (COSEWIC 2014). In a Maine lake on Mount Desert Island they spawned "throughout the day in shallow (< 2m) water" (Salinas et al. 2003). Nest-building sunfish and bass do not spawn on plants but clear a patch of firm sand or gravel (project synthesis). In Yoho and Oromocto lakes, New Brunswick, most redbreast sunfish nests were dug through a dense plant mat to coarse sand or fine gravel, and in Oromocto 85 % lay within 1 m of a rock or log (Gautreau & Curry 2006, via COSEWIC 2008). Smallmouth bass nests in central Ontario were fewer on weedy shores (0.24 per 100 m) than on sand with rock (1.18) (Chu 2001). There are no Maine or Hot Brook data on where perch ribbons or pickerel eggs lie; egg survival as a whole is treated in section 3.6.

The cover estimates of section 3.4.5 were later used to re-target the extent of the model's weed beds (section 5).

← Introduction and study system · Contents · Synthesis II: fish, early life, predation, wildlife and forest →