Observations of the Middle Island Sinkhole in Lake Huron – a Unique Hydrogeologic and Glacial Creation of 400 Million Years

Home , Alcove (landform)

P A P E R Observations of the Middle Island Sinkhole in Lake Huron – A Unique Hydrogeologic and Glacial Creation of 400 Million Years

A U T H ORS ABSTRACT Steven A. Ruberg In the northern Great Lakes region, limestone sediments deposited some 400 NOAA-Great Lakes Environmental million ybp during the Devonian era have experienced erosion, creating karst features Research Laboratory such as caves and sinkholes. The groundwater chemical constituents of the shal- Scott T. Kendall low seas that produced these rock formations now contribute to the formation of Bopaiah A. Biddanda a unique physical (sharp density gradients), chemical (dissolved oxygen-depleted, Grand Valley State University sulfate-rich) and biological (microbe-dominated) environment in a submerged sinkhole near Middle Island in freshwater Lake Huron. A variety of methods including Tyrone Black aerial photography, physico-chemical mapping, time series measurements, remotely Michigan Office of Geological Survey operated vehicle (ROV) survey, diver observations and bathymetric mapping were Stephen C. Nold employed to obtain a preliminary understanding of sinkhole features and to observe University of Wisconsin – Stout physical interactions of the system’s groundwater with Lake Huron. High conductivity ground water of relatively constant temperature hugs the sinkhole floor creating a Wayne R. Lusardi distinct sub-ecosystem within this Great Lakes ecosystem. Extensive photosynthetic Russ Green purple cyanobacterial benthic mats that characterize the benthos of this shallow Thunder Bay National Marine sinkhole were strictly limited to the zone of ground water influence. Sanctuary Tane Casserley Introduction tions of chloride, and 100-fold higher NOAA NMS Maritime Heritage ubmerged karst features (sink- concentrations of sulfate (Ruberg et al., Program Sholes) were discovered in Lake Huron 2005). A variety of non-photosynthetic Elliott Smith (Michigan) during a 2001 side scan benthic microbial mats were observed Noble Odyssey Foundation sonar survey expedition conducted in this deepwater aphotic sinkhole by the Thunder Bay National Marine system. Water samples collected from T. Garrison Sanders Sanctuary (TBNMS) and the Institute the sinkhole plume contained bacterial Grand Valley State University for Exploration (Coleman, 2002). Our concentrations (~9x109 cells l-1) an or- Gregory A. Lang initial understanding of the composi- der of magnitude higher than ambient 9 NOAA-Great Lakes Environmental tion of Lake Huron sinkhole systems lake concentrations (~1x10 cells l-1), Research Laboratory is based on a preliminary 2003 ROV and showed evidence for the occur- exploration of the Isolated Sinkhole lo- rence of significant chemosynthesis in Stephen A. Constant cated 10 miles offshore in the TBNMS this lightless deep water environment University of Michigan at a depth of 93 m (Figure 1), reveal- (Biddanda et al., 2006). These rates of ing a visible cloudy layer attributed to chemosynthesis occurring in the Lake groundwater. When compared to am- Huron Isolated sinkhole were com- bient lake water with a temperature of parable to those measured in thermal 3.9 °C and specific conductivity of 140 vents in Yellowstone Lake (Cuhel et µS/cm, groundwater in these systems al., 2002). was observed to be several degrees high- Understanding the nature of the er in temperature, with 10-fold higher groundwater emerging in Lake Huron’s conductivity, 10-fold higher concentra- sinkholes requires an introduction

12 Marine Technology Society Journal FIGURE 1

Map of the North American Laurentian Great Lakes Basin showing regions of Silurian-Devonian aquifers potentially having karst formations (left, modiﬁed from Grannemann et al., 2000), and regions of above ground karst formations in Alpena County, MI and submerged sinkholes in the Thunder Bay National Marine Sanctuary (TBNMS), Lake Huron (right, after Coleman, 2002; Biddanda et al., 2006).

to the major geologic forces shaping collapse in the recent one million year “exposed” at the edges of the Michi- sediments and lake morphometry in geologic history of Michigan’s northern gan Basin by glacial scouring. Water the Michigan Basin. Most noticeable lower peninsula. easily entered the partially developed are collapse features of the past 10,000 The large-scale loss of evaporites paleokarst (fossil karst) systems and years since the glaciers retreated from and collapse of the Detroit River Group enhanced the karst as dissolution of the the area. One of these, the Middle Is- and formations above it were caused evaporites continued. As the evaporite land sinkhole/resurgence, is the surface by water flow. Surface and ground volumes were diminished, the weight expression of a collapse pipe extending waters ﬂowed into hydraulically open of overlying formations collapsed the down through the Rogers City and tectonic faults to deeply buried layers of group and caused structural adjust- Dundee limestone formations into the the group. Surface and ground waters ment faulting in the formations above, top of the Detroit River Group karst also penetrated the group where it is allowing more water to penetrate. For- system, approximately 200 feet below FIGURE 2 (Figure 2). Some inland sinkholes have collapsed through up to 800 feet of Schematic diagram of water sources and ﬂow along subsurface pathways to discharge into Lake additional overlying formations not Huron. Inland recharge areas exert water pressure on aquifers, forcing water through the karst system to Lake Huron. present at Middle Island (Black, 1983). Gypsum, anhydrite, and salts in the Detroit River Group formation were deposited in the Michigan Basin in the early Middle Devonian Period, approximately 400 million ybp (Gardner, 1974). Toward the end of this deposi- tion there was a brief uplift and dissolution of some of the evaporites. Basin subsidence and new sediments buried and “fossilized” the karst features that had developed. This originally shallow and minor karst development would set the stage for large-scale dissolution and

Winter 2008/2009 Volume 42, Number 4 13 mations were weaker and were more sinkhole (45° 11.911’ N, 83° 19.662’ perature averaging approximately 5 °C. likely to develop sinkholes where faults W; Figure 1) is located approximately Visibility underwater varied depending crossed each other. 2 miles offshore at a depth of 23 m on the season, ranging from only 6 Erosion had exposed the edges of in NW Lake Huron. In the present meters at times during the mid-to-late formations before the last glacial period. study, a variety of methods including summer, to 20 meters or better in the Glaciation scoured the exposed surface physico-chemical mapping, time-series spring and early summer. Diver tasks of collapsed formations, sometimes fill- measurements, diver observations and included mapping of the sinkhole ing in and plugging already developed bathymetric mapping were employed “alcove” (a deep hole that fills with sinkholes. As the glacier retreated, to obtain a preliminary understanding groundwater and spills over a sill into it left behind a cover of diamictite of sinkhole features and to observe the wider sinkhole area) and “arena” (glacial till) from zero to 700 feet that some of the physical aspects of the sink- (the wider bowl-shaped sinkhole area mantled bedrock and sinkholes, it left hole system’s groundwater interaction that opens out to Lake Huron) areas, water available to surface streams and with Lake Huron. deploying light scientific equipment, groundwater, and formations stressed sample collection, and collection of still by the removal of the glacial weight and and video imagery (Figures 3, 4 and 5). crustal rebound. Some of the stress was Survey Procedures Plan and profile views (Figure 5) of the relieved by collapse of windows into (Methods) sinkhole alcove were produced by diver the karst system. If these windows were The project utilized divers from the observation and cursory measurements above the static water level (controlled NOAA TBNMS, NOAA Maritime utilizing hand-held tape measures, a by the Great Lakes levels of the time), Heritage Program, Noble Odyssey dive computer recording depth of wa- water would enter the system at that Foundation (NOF), State of Michigan, ter, and a compass to orient magnetic point (Figure 2). If the window was and East Carolina University. Their north. Observations were recorded below the static level, water was able reports and drawings provided observa- on Mylar attached to a submersible to exit the system under hydrostatic tions and measurements unobtainable plastic slate. pressure into the Great Lakes system, from imagery, surface mounted acous- Bathymetric survey operations, bringing with it the dissolved minerals tics or moored instrumentation. Dives time-series mooring deployments, and and nutrients accumulated during its were conducted from the TBNMS physical and chemical mapping were transit through the formations. The 41-foot R/V Huron Explorer, 28-foot conducted from the NOF’s 28-foot latter formations resulted in the sub- R/V Sweetwater, and the NOF R/V R/V Sounder, NOAA’s R/V Huron merged sinkhole systems that are the Pride of Michigan. Trained chiefly in Explorer and smaller vessels. The NOF subject of our present study. underwater archaeology, the dive team bathymetric mapping system con- Sinkholes have been the source of adapted readily to the range of skills sisted of an Innerspace 455 single-beam scientific curiosity for many years in required for this multi-disciplinary echosounder with a 200 kHz 3° stain- the northeast region of the lower pe- fieldwork. The result was an efficient, less steel transducer pole-mounted to ninsula of Michigan. Very near shore, cost effective and safe acquisition of the side of the R/V Sounder. The a submerged sinkhole on Lake Huron data. Prior to the dives, a SeaBotix differential GPS unit linked to the in Misery Bay, northeast of Alpena, was LBV150SE ROV equipped with video echosounder was a Trimble DSM 232 first described in the latter part of the and sensors providing observations of with nominal +/- 1-5 meter accuracy. nineteenth century to be one or two depth, conductivity and temperature HYPACK® software was used to set up hundred feet in diameter and 79 feet was used to survey the layout and the survey grid and log and process the deep (Boulton, 1876). Farther offshore hydrological conditions of the entire data. Sounding data were collected over lies Middle Island sinkhole, a destina- sinkhole ecosystem. the sinkhole area along east-west, north- tion for recreational divers for at least Depending on assigned workloads, south grid lines 5-10 meters apart. Lake the latter part of the twentieth century divers used either a single 80 ft.3 tank conditions during bathymetric data (Harrington, 1990). Remarkably, this or twin 112 ft.3 tanks, resulting in collection were moderate with light is the first multi-disciplinary scientific average bottom times of 40 minutes. winds and 0.3 meter waves. Maps of study undertaken of this unique un- Maximum depths ranged from 9 the sounding data were generated using derwater ecosystem. The Middle Island meters to 22 meters, with water tem- interpolation and visualization methods

14 Marine Technology Society Journal FIGURE 3 using three multi-parameter sondes mounted at 20.4, 21.3, and 22.3 m (a) Bathymetric contour map created from echosounder data points locating the alcove and the Instrumented Mooring used to collect time-series data in the arena (depth contours are in meters); below the surface on an anchored cable (b) shaded surface plot (scale in meters) showing the location of instrumentation used to collect time- located at 45° 11.911’ N, 83° 19.662’ series data; (c) Mooring diagram showing arrangement of thermistors and multiprobes; and (d) aerial W (Figure 3). Depth to the sinkhole photo with Middle Island in lower left and 25 ft. Whaler on right near alcove. Temperature sensors bottom was at 22.9 m. Each of the were deployed from May 19 to September 17, 2007. The YSI6920 multiprobes were used to observe variability of conductivity, temperature, pH, and dissolved oxygen from June 15 to July 24, 2007. three YSI (Yellow Springs Instruments, Inc., Yellow Springs, OH) 6920 sondes (a) (b) were equipped with a temperature/ conductivity sensor (6560), optical dissolved oxygen sensor (6150 ROX), and a pH/ORP sensor (6565). To determine the thickness of the dense groundwater layer in the alcove and throughout the arena, and obtain physico-chemical measurements, a series of multi-parameter sonde depth profiles were performed. Water mass characterized by conductivity >2x of the Lake Huron water was considered (c) (d) to be groundwater. Profiles were taken beginning over the alcove and at approximately 20 m intervals extending to the north out into Lake Huron for 230 m. Depth profiles were taken using a YSI 6600 sonde equipped with sensors for temperature/conductivity, dissolved oxygen, and pH. Data was recorded onto aYSI 650 multi-parameter display at 2 second intervals. available in IDL analysis software from 5 minutes in water. Wind speed and ITT Visual Information Solutions direction were obtained from NOAA’s (Figure 3a and 3b). The contour plot National Data Buoy Center 3 meter Results and Observations shows interpolated isobaths at 1 meter discus buoy 45003 located approxi- Diver Survey intervals. The shaded plot shows a mately 40 km northeast of the study Diver-collected imagery at a depth shaded-surface representation of the area at 45°21.02’ N 82°50.40’ W. The of 10 meters along the south edge interpolated data, rotated 200 degrees redundant R.M. Young Anemometers of the Middle Island sinkhole shows counterclockwise about the Z-axis. located 5 m above the lake surface pro- the steep drop into the alcove below Continuous time-series observa- vided hourly 8-minute averaged wind (Figure 4a). A partial wall between tions of water temperature in the Mid- speed with a resolution of 0.1 m/s and the alcove and the main arena of the dle Island sinkhole during May 19 to accuracy of +/- 1.0 m/s, and wind di- sinkhole (Figure 4b) forms a reservoir September 17, 2007 were collected at rection with a resolution of 1.0 degrees of colder, denser groundwater that is 1 hour intervals using Onset UTBI- and accuracy of +/- 10 degrees. maintained throughout the summer 001 temperature sensors attached to Continuous time-series monitoring warming period. a vertical mooring line at 1 meter and of the groundwater–lake water inter- A cloudy/shimmering layer of water 4 meters from the bottom (Figure 3c). face in the Middle Island sinkhole in at the thermo-pycno-chemocline (re- Sensor accuracy is 0.2°C over the range the arena was conducted at one-hour gion of sharp changes in temperature, of 0° to 50°C with a response time of intervals from June 15 to July 24, 2007, density and chemical composition of

Winter 2008/2009 Volume 42, Number 4 15 FIGURE 4

(a) Southern edge of Middle Island sinkhole at a depth of approximately 10 m; (b) image taken from outside of the alcove, viewing rock wall (beneath diver) where groundwater contents (background) spill over into sinkhole arena (foreground).

(a) (b)

water; hereafter referred to as ‘thermo- sinkhole. Purple benthic microbial ice erosion beginning approximately cline”) was visible to divers while tran- mats were observed along the walls 6,000 ybp (Hough, 1958) when lake sitioning through surface waters into and bottom of the alcove, along the levels were a minimum of 122 m lower the alcove groundwater. The plan and wall separating the alcove and arena, than present. Lake Huron water levels profile views of the alcove and arena and throughout the southern portion began to rise from these low levels due (Figure 5) represent a synthesis of visual of the arena. Purple mats were only to rebound of the earth’s crust after the and measured dimensions resulting in observed underlying the cloudy layer weight of glacial ice was removed at the sketches that can only be created with of water at the thermocline. This was end of the Wisconsin ice age during the divers present on the underwater site. particularly noticed in the alcove, where Pleistocene epoch (Stanley, 1937). Rocky fissures on the bottom of the purple mats were not observed above alcove, at a depth of 22.5 m, provide a the thermocline. Lake waters circulate Sensor Observations major conduit for groundwater to the through the sinkhole arena via the While a thorough discussion and sinkhole alcove and arena. The alcove open northern side. This opening is modeling of Lake Huron currents is fills to a level of approximately 5 m believed to have resulted from wave and beyond the scope of this present study, before spilling over the wall separating the alcove from the arena (Figure 4b FIGURE 5 and 5b). The sustained force of the flow (a) Plan view of the alcove indicating dispersion of groundwater into sinkhole arena providing nutrients is strong enough to propel divers over to cyanobacterial mats. (b) Profile view showing the groundwater reservoir filling the alcove before the top, and they are able to surf the spilling over wall into the Arena. The interface refers to the zone of thermocline/pyncnocline where relatively denser groundwater cascade groundwater and lake water meet. (an “underwater waterfall”) over the (a) (b) sill wall down-slope onto the arena floor. Water flow can be observed in the wafting motion of long strands of whitish sulfur-oxidizing bacteria living on the rocky surface at the interface between the sulfur-rich groundwater inside the alcove and the oxygen-rich lake water. Groundwater flowing out of the alcove then fills a shallow arena depression 23 m deep (Figure 3a) and gradually disperses throughout the

16 Marine Technology Society Journal NDBC buoy 45003 located 40 km NE the temperature sensors show that the above (2.4 m hab sonde). Overall, the of the study site provides some indica- groundwater at 1m hab resisted the sinkhole arena groundwater layer, rela- tion of the influence of wind speed and thermal influence of warmer waters tive to lake water, was lower in tempera- direction on the interaction of sinkhole above (4 m hab), averaging 9.94 °C. ture (typical), pH (7.3 vs 8.3), and DO groundwater with Lake Huron waters During the period JD 170 to JD 260 (4 vs >11 mg/L), while higher in spe- (Figure 6a). In general, winds with a (June 19 to September 17), water tem- cific conductivity (1.7 vs 0.3 mS/cm). southerly component tend to push peratures at 4 m averaged 11.75 °C. Throughout the monitoring period, warmer surface waters offshore through The influence of colder water during groundwater characteristics were very the open north end of the system which Lake Huron upwellings on JD 167, consistent for most of the param- are then replaced with colder subsur- 170, 185, 192, 204, 208, 220, 222, eters except temperature which varied face lake water. This wind-driven inter- and 234 was attenuated by the ground- throughout the monitoring period. action between sinkhole groundwater water influence at 1 m. Brief periods of Temperatures fluctuated in all sondes; and Lake Huron water is evident in the groundwater mixing with warmer wa- however, groundwater temperatures data obtained from temperature sensors ters above do occur as can be seen from were often 4 ºC colder than overlying suspended 1 m and 4 m height above the temperature response at 1m hab to lake water. Some sudden fluctuations bottom (hab) in the sinkhole arena the wind event lasting from JD 226 to in parameters may be explained by (Figure 6b). Temperatures observed JD 232 (August 14 to August 20). wind induced waves and currents in- over the time period showed the warm- Continuous time-series data from teracting with overlying lake waters. ing influence of the groundwater on three multiparameter sondes measur- Some significant events are observed the sinkhole benthic region from Julian ing temperature, DO, pH, and specific on JD 170, 173, 179, 190, and 200 Day (JD) 139 to JD 150 (May 19 to conductivity deployed from June 15 to when higher winds with a southerly May 30) while Lake Huron was still July 24, 2007 are presented in Figure component (Figure 6a, starred) result in the post-winter isothermal phase. 7. The placement of these instruments in sudden, but temporary, changes in During the warmer summer months, accurately depicted the layering of DO, conductivity, temperature and pH the groundwater tended to provide a the groundwater at this location. The (Figure 7). These events result in higher cooling influence, but there appears to sonde at 1.5 m hab was located in the oxygen and lower conductivity levels have been some periodic mixing likely transition zone between groundwater indicating a decrease in the thickness associated with wave action. Data from below (0.6 m hab sonde) and lake water of the groundwater layer. However, ob-

FIGURE 6

(a) Wind speed and direction (6 hour averages) for the period May 19 to September 17, 2007 from NDBC Buoy 45003 located approximately 40 km northeast of the study area in northern Lake Huron; (b) Temperature observations obtained from Onset UTBI-001 sensors located 1 m and 4 m height above bottom (hab) at depths of 21.86 m and 18.86 m, respectively. Mooring was placed in the arena at 23 m depth during May 19 to September 17, 2007. An initial warming inﬂuence of groundwater during Lake Huron isothermal conditions can be observed JD139 to JD150 (May 19 to May 30). Vertical lines indicate period of sonde deployment.

(a) (b)

Winter 2008/2009 Volume 42, Number 4 17 FIGURE 7

Dissolved Oxygen (a), Conductivity (b), Temperature (c), and pH (d) taken from sensors located in the arena at 0.6 m hab, 1.5 m hab, and 2.4 m hab during Middle Island Long-Term Monitoring from June 15 to July 24, 2007. Bottom at 22.8 m (hab=height above bottom).

(a) (b)

servations of lower oxygen and higher 6-17 ºC but was mostly 14-16 ºC. 9.1 ºC. While this alcove profile only conductivity on JD 169, 184, 190, These temperature data (Figure 7c) extends to a depth of 18.5 m (Figure 8), 195, and 203, indicating an increase show a similar pattern as measured the depth to bottom in the alcove in the thickness of the groundwater by the Onset sensors during the same is ~22 m, making the groundwater layer, appear to have little correlation period (Figure 6b). thickness approximately 5 m. This is with wind speed or direction. In addi- To determine the thickness of slightly higher than the ~4.5 m height tion to the parameters already meas- the dense groundwater layer and to of the rock wall that separates the al- ured, future time-series investigations measure physico-chemical parameters, cove from the arena (Figure 4b and 5). must include observations of alcove depth profiles using a multi-parameter Together with diver observations groundwater flow speed, local rainfall, sonde were obtained over the alcove discussed earlier, these data show that and ambient waves, temperatures and and northward through the arena at venting groundwater fills the alcove currents external to the sinkhole to ~20 m intervals (Figures 8 and 9). The bowl and spills over into the arena adequately explain the interactions alcove profile (Figure 8) shows that (Figure 9). Groundwater thickness in of sinkhole groundwater with overly- at about 16.5 m below the surface, the arena between 20 and 60 m along ing lake waters. Overall, groundwater specific conductivity rises from 0.2 to the transect averages 1.2 m, and then temperature ranged 8-16 ºC but was 2.3 mS/cm, DO drops from 10.4 to thins out to approximately 0.6 m thick predominantly 10-12 ºC and overlying 0.4 mg/L, pH drops from 8.4 to 7.1, which extends out to the end of our lake water temperature ranged from and temperature drops from 18.3 to measurements at 229 m (Figure 9).

18 Marine Technology Society Journal FIGURE 8 Conclusion Depth profile of specific conductivity, dissolved oxygen, temperature and pH from lake surface to near The Middle Island sinkhole was bottom of alcove sinkhole on July 13, 2007. Alcove bottom at ~22.5m. formed by the collapse of the karst system roof forming a rubble pipe with a partially open top at this location. The collapse was coincidentally near the edge of the formation escarpment. The edge of the feature has been further modified by wave action and lake-edge ice-driven erosion into the escarpment face weakened by the presence of the sinkhole. This erosional process appears to have removed most of the northern wall of the sinkhole into Lake Huron before the lake reached its current elevation. The source of the groundwater discharge could be from the natural hydrologic flow through the geologic formations and interacting with: 1) formation waters buried with the formation, or waters being squeezed In the southern portion of the arena (11.6 mg/L), temperature (10.2 ºC), out by compaction and lithification; (close to the alcove), groundwater and pH (8.4), but lower specific con- 2) ground waters captured by the characteristics were different from ductivity (0.2 mS/cm). ROV/Diver hydraulically open fault and fracture the alcove mainly with higher DO observations and water mass character- systems and released to the developing (3-4 mg/L; Figure 9) and slightly istics confirm that the distribution of karst system; 3) under ice or melt water lower conductivity (1.8 mS/cm). the purple cyanobacterial benthic mats from the glaciers entering the system In the northern portion of the arena was limited exclusively to the zone of from the surficial sinkholes; and 4) and beyond to a distance of 229 m, groundwater influence. The ground- post glacial (and perhaps some pre- groundwater DO levels gradually in- water emerging at the sinkhole likely glacial) meteoric and surface waters. creased to 6.5 mg/L (Figure 9). Over- provides a unique set of conditions The source(s) of the water, flow time, lying lake water in the arena at 12 m that support the proliferation of purple and flow routes are a complex puzzle, hab was characterized by higher DO cyanobacterial mats (Figure 9, 10). the resolution of which will require the

FIGURE 9

Schematic profile of sinkhole-lake cross section showing depth to groundwater and lake floor, approximate extent of purple mat (thickness not to scale; actual mat thickness is 1-2 mm), and groundwater DO along a 230 m transect extending from the alcove and northward through the arena on July 13, 2007 at the Middle Island Sinkhole. Groundwater thickness along this transect is approximately 5 m in the alcove, 1.4 m@40 m, 1m@60m, and 0.5m@100m. While maximum groundwater specific conductivity ranged between 1.7 and 2.3 mS/cm, a value that was 2x greater than the overlying lake water (i.e., 0.2 mS/cm) was used for de- termining groundwater thickness. Purple cyanobacterial mats oc- curred in areas with DO levels as low as 0.1mg/L up to 4 mg/L.

Winter 2008/2009 Volume 42, Number 4 19 FIGURE 10

(a) Variety of benthic microbial mats inside sinkhole dominated by the purple pigmented cyanobacteria. Arrow indicates whitish strands of sulfur-oxidizing bacteria wafting in the ﬂow of groundwater. The white ﬁlamentous sulfur-oxidizing bacteria occur at the interface between ground water (left) and lake water (right) where ground water spills from the alcove into the arena. (b) Microscopic image of diverse and abundant microbial cells and some metazoa in benthic mat material recovered from Middle Island Sinkhole.

(a) (b)

application of geochemistry and isotope green. Sufficient sulfide to support cy- 2003). Conditions for sulfur-oxida- analysis to the spatial geology of the anobacterial growth is likely produced tion are present in the sinkhole habi- area. The very low levels of oxygen and by sulfate-reducing bacteria in the tat where the oxygen-rich lake water high sulfate concentrations measured in groundwater-bearing sediments and mixes with the sulfur-rich groundwater. the alcove water support the conclusion the arena sediments on the lake floor. These long filaments are only found that the Detroit River Group aquifer is Sulfate is abundant in the groundwater at the interface between groundwater the groundwater source at the Middle (Biddanda et al., 2006), and organic and lake water (Figure 4b, 10a), further Island sinkhole, as deeper aquifers tend carbon to fuel sulfate reduction is supporting these conclusions. to be more oxygen-depleted. plentiful on the groundwater-impacted Questions regarding how the domi- The filamentous cyanobacteria that lake floor. Since cyanobacteria photo- nant cyanobacterial mats manage to form the extensive microbial purple synthetically fix carbon dioxide, they disperse and colonize sinkholes that mats observed by the divers and ROV can act as primary producers and may are small and geographically distant/ are closely related to the genus Oscil- constitute the base of the food web in isolated, as well as the significance of latoria on the basis of microscopic and these unique ecosystems. Our observa- cyanobacterial mat production at the molecular studies (Figure 10). Peri- tions in this and other shallow sunlit sinkholes to the surrounding lake food phyton such as diatoms as well as oc- submerged sinkholes in the surround- web remain to be addressed. Interest- casional metazoans such as nematodes ing areas of Lake Huron confirm the ingly, the ancestors of the cyanobacteria are also observed under the microscope presence of the purple photosynthetic that currently populate Middle Island (Figure 10b). These predominant blue- cyanobacterial benthic mats whenever sinkhole first appeared on earth ap- green algae have two modes of growth; high conductivity, sulfate-rich ground proximately 3.5 billion ybp (Taylor, they can either perform oxygenic water is present (Figure 10). 1993), but are now being sustained photosynthesis similar to green plants The occasional white bacterial by dissolved ions such as sulfates and or, when sulfide is present, they can strands attached to rocks at the alcove carbonates deposited relatively recently perform anoxygenic photosynthesis outfall are likely to be sulfur-oxidizing (hundreds of millions of years ago). using sulfide as the electron donor bacteria such as Beggiatoa (Figure 10a). Another interesting feature of the sub- (Stal, 1995). When growing in the These organisms oxidize sulfide in the merged sinkholes is the abundance of presence of sulfide, they synthesize a presence of oxygen and deposit elemen- organic-rich soft sediments found on different form of chlorophyll, caus- tal sulfur within their bodies, giving site. Organic matter preservation may ing pigments to be purple rather than them a whitish appearance (Dyer, be favored in the sediments underlying

20 Marine Technology Society Journal water masses that contain high sulfate References Hough, J.L. 1958. Geology of the Great Lakes. University of Illinois Press, Urbana. and low oxygen, as found in Lake Biddanda, B.A., Coleman, D.F., Johengen, Cadagno, Switzerland, which is fed T.H., Ruberg, S.A., Meadows, G.A., VanSum- Ruberg, S., Coleman, D., Johengen, T., by a subaquatic spring (Hebting et al., eran, H.W., Rediske, R.R., Kendall, S.T. 2006. Meadows, G., Van Sumeren H., Lang, G., 2006). Based on the known widespread Exploration of a Submerged Sinkhole Ecosys- Biddanda, B. 2005. Groundwater plume map- distribution of carbonate aquifer in the tem in Lake Huron. Ecosystems. 9:828-842. ping in a submerged sinkhole in Lake Huron. Mar Technol Soc J. 39(2):65–69. region (Black, 1983; Granemann et Black, T.J. 1983. Selected views of the tectonics, al., 2000), submerged sinkholes may structure and karst in Northern Lower Michi- Stal, L.J. 1995. Physiological ecology of be more numerous in the Great Lakes gan, Michigan. In: Kimmel, R.A., ed. Tecton- cyanobacteria in microbial mats and other basin (Biddanda et al., 2006). These ics, structure and karst in Northern Lower communities. New Phytologist. 131:1-32. ecosystems could be important sites Michigan. Michigan Basin Geological Scociety for biological production as well as Field Conference Proceedings. pp. 66-77. Stanley, G.M. 1937. Lower Algonquin Beaches of Cape Rich, Georgian Bay. Geo Soc carbon sequestration. The pursuit of Boulton, W. 1876. Complete History of Am Bull. 48:1665-1686. such ecological enquiries in the future Alpena County, Michigan. Argus Book and will be facilitated by the results of the Job Rooms, Alpena. Taylor, T.N., Taylor, E.L. 1993. The Biology first geophysical mapping of the Mid- and Evolution of Fossil Plants. New Jersey: Coleman, D.F. 2002. Underwater Archaeology dle Island sinkhole reported here. Prentice Hall. in Thunder Bay National Marine Sanctuary, Lake Huron. Mar Technol Soc J. 36(3):33-44. Acknowledgments Cuhel R.L., Aguilar, C.A., Anderson, P.D., The authors acknowledge the fund- Maki, J.S., Padock, R.W., Remsen, C.C., Klump, J.V., Lovalo, D. 2002. Underwater ing contributions made by NOAA’s domains in Yellowstone Lake: hydrothermal Office of Ocean Exploration during the vent geochemistry and bacterial chemosynthe- initial survey and mapping cruises lead- sis. In: Anderson, R.J., Harmon, D., eds. Yel- ing to this work. Wind speed and wind lowstone Lake: Hotbed of Chaos or Reservoir direction data was obtained from the of Resilience. Yellowstone National Park: Yel- National Data Buoy Center. The authors lowstone Center for Resources and the George would like to recognize the contribu- Wright Society. pp. 27–53. tions made by Dennis Donahue, marine Dyer, B.D. 2003. A field guide to bacteria. superintendent of NOAA’s Great Lakes Ithaca, NY: Cornell University Press. Environmental Research Laboratory (GLERL); Jeff Gray, TBNMS manager Gardner, W.C. 1974. Middle Devonian strati- for laboratory and logistical support; graphy and depositional environments in the Luke Clyburn, captain of the Pride Michigan Basin: MBGS, Special Paper, p. 32-41. of Michigan (NOF) for ship, survey, Grannemann, N.G., Hunt, R.J., Nicholas, and dive support; and Pat Sanders of J.R., Riley, T.E., Winter, T.C. 2000. The HYPACK for bathymetric mapping sup- importance of groundwater in the Great Lakes port. Joe Hoyt and Eric Strickler assisted region. In: USGS water resources investiga- with field work and the data collection. tions report 00-4008. Sanders was supported on NSF and Harrington, S. 1990. Divers Guide to Michi- Michigan Space Grants Consortium gan. St. Ignace, Michigan: Maritime Press. graduate fellowships, whereas Strickler Hebting, Y., Schaeffer, P., Behrens, A., Adan, was supported by an NSF internship. P., Schmitt, G., Schnecknburger, P., Bernasco- This work was funded by National ni, S., Albrecht, P. 2006. Biomarker evidence Science Foundation grants 0603944 for a major preservation pathway of sedimen- (Biddanda) and 0604158 (Nold), and is tary organic carbon. Science, 312:1627-1631. GLERL contribution number 1493.

Winter 2008/2009 Volume 42, Number 4 21