Seventh Biennual LIS Research Conference Proceedings, 2005, p. 27-32

The late Pleistocene-Holocene History of Long Island Sound

¹ Earth & Environmental Sciences, Wesleyan University, Middletown CT 06459

² Department of Geology and Geophysics, Yale University, New Haven CT  06511

Introduction

The early history of  Long Island Sound (LIS) reflects the interplay between rising global sea levels, rising of the land as a result of glacial rebound, and fill-in by sedimentation of a depression formed after the retreat of the Wisconsinan Ice Sheet. The main events in this history are, in sequential order, retreat of the ice sheet from Long Island onto the Connecticut mainland, establishment of Glacial Lake Connecticut at the site of modern LIS and southern CT, establishment of Glacial Lake Hitchcock in central Connecticut, Massachusetts, New Hampshire and Vermont, drainage of Glacial Lake Connecticut and subsequent erosion/fluvial dissection of its lake bottom, early transgression of the sea into LIS, drainage of Lake Hitchcock, and fill-in of LIS by the sea and marine sedimentation (Lewis and Stone, 1991, Lewis and DiGiacomo-Cohen, 2000, Stone et al., 2005). During post-glacial times, loess was deposited in Connecticut and on Long Island (Kundic and Hanson, 2003).

The age of these events has been established with radiocarbon dates of peat and detrital organic matter in sediments on land and in LIS, in combination with counting varves deposited in the glacial lakes (see Ridge, 2003 and Stone et al., 2005 for listing and references). Many authors give ages in radiocarbon years, summarized and calibrated into calendar years by Kundic and Hanson (2003), and updated by Ridge (2003). The Harbour Hill-Fishers Island-Charlestown moraine indicates the position of the ice margin on the northern edge of Long Island at 21,300 calendar years BP (Ridge, 2003). The Captain Island-Norwalk Islands-Old Saybrook-Wolf Rocks moraine marks the limit of the ice sheet in southern Connecticut and LIS at about 20,400 calendar years BP (Ridge, 2003). Glacial Lake Connecticut started to form between these two moraines, between 21,300 and 20,400 calendar years BP. Varved sediments were deposited in this lake coeval with the formation of glacio-fluvial deltas along its northern shore (Lewis and Stone, 1991). Glacial Lake Connecticut may have drained around 18,000 calendar years BP (~15.5 radiocarbon years, Lewis and Stone 1991), although exact ages are not available.

Lake Hitchcock started to form when the ice had retreated further to the north, and a morainal dam formed close to Rocky Hill, CT at about 19,100 calendar years BP (Ridge, 2003). Rittenour et al. (2000) studied the lake development between varve years 2868 and 6900, assigning calendar ages of 17,500 to 13,500 years for this interval. The same varves were dated at 18,300 to 14,100 calendar years by Ridge (2003). Lake Hitchcock persisted for at least 4100 years according to the New England varve chronology (Antevs, 1922), draining around 13,500 calendar years BP (Rittenour et al., 2000), updated to 14,100 calendar years BP by Brigham-Grette et al. (2001) and Ridge (2003). In contrast, Stone et al. (2005) suggest an age of 13,500 radiocarbon years BP (16,500 calendar years BP) for the end of the stable phase and draining of Lake Hitchcock.

The age of the first marine invasion in LIS was estimated by Lewis and Stone (1991) and Kundic and Hanson (2003) at 14,300 calendar years BP (12,455 radiocarbon years BP), but Stone et al. (2005) argue (p. 12) that the early transgression (flooding of the existing channels) may have started between 15,000 and 16,000 years BP (probably calendar years but not specified), coinciding with the early part of melt-water pulse 1a of Fairbanks (1989) and Fairbanks et al. (1992). After the marine invasion of the fluvial channels, the main marine unconformity cut into the upper lake beds and estuarine channel fill deposits (Lewis and Stone, 1991). A marine deltaic morpho-sequence was built over this unconformity between 13,000 and 9500 radiocarbon years BP (Stone et al., 2005; corresponding to 15,600-11,300 calendar years BP) to the west of the mouth of the Connecticut River, with its abundant sediment possibly provided by erosion of the now dry Lake Hitchcock beds. Stone et al. (2005) argued that sea level in LIS did not rise much during the delta building phase because the rates of crustal uplift and absolute sea level rise were similar. Sea level in LIS then rose sharply during melt water pulse 1b (Fairbanks, 1989) around 9500 radiocarbon years BP.

Disagreements about timing of events (e.g., draining of Lake Hitchcock by Stone et al. 2005 versus Rittenour et al., 2000) and assumptions about timing of glacial rebound suggest that the time frame of the post-glacial events is not yet rigorously defined. To address this, we dated the onset of the marine transgression in LIS with targeted radiocarbon dates on material from LIS sediment core LISAT12.

Methods and materials

In 1984, R.S. Lewis and co-workers collected 13 vibracores in LIS using the RV Atlantic Twin, and described the core lithologies (Thomas, 1989). Core LISAT12 was obtained near the Long Island coast to the southwest of the mouth of the Connecticut River (41^o07.70’N, 72^o28.80’W; point 25, map 2784, Stone et al., 2005) at a water depth of 37 m. We took 40 samples from core LISAT12  (curated at the Woods Hole Oceanographic Institution), and submitted material from 8 depth intervals for radiocarbon dating at NOSAMS, WHOI. We submitted mollusk (carbonate) shell fragments from all 8 samples. From 5 samples we also submitted hand-picked macrofloral remains, small twigs and leaf fragments, which are abundant in these sediments.

The level of the marine transgression in core LISAT12 is indicated by red, varved lake beds from Glacial Lake Connecticut overlain by grey sands and silts with abundant oysters (Crassostrea virginica; Szak 1987). The transgression interface is deformed as a result of coring, sloping between 540 and 562 cm depth-in-core, giving a mean depth of the marine transgressive interface of ~ 42.5 m below modern mean sea level, close to the maximum reported depth of the main transgressive interface (Stone et al., 2005). The sedimentary environment at 550-350 cm depth in the core is interpreted as intertidal to shallow subtidal, with abundant oysters, foraminifera and diatoms (Szak, 1987). Marine silt is present from 350 to 150 cm depth, and the upper 150 cm is coarser-grained, containing common eroded foraminifera and mollusk shell fragments. This upper section is interpreted as reworked, possibly sand-wave facies material (Fenster, 1995), as present in the modern environment near the LISAT12 coring site (Knebel et al., 1999).

The results for ¹⁴C measured by AMS were corrected for the measured d¹³C values. Mollusk (carbonate) data have been calibrated with the 1998 marine data set of the CALIB program, version 4.3 (Stuiver et al., 1993, 1998), using the standard 400 year marine reservoir effect. Our work on LIS ¹⁴C ages (last 150 years of LIS history) indicates that this is a reasonable approximation for LIS, although variations of + several 100 years exist (Groner, 2004). The ¹⁴C ages of macrophytes were calibrated with CALIB using the 1998 atmospheric decadal data set (Stuiver et al. 1993, 1998). The ages reported here are our best estimate means when multiple age calibrations occurred, and a detailed error analysis will be presented elsewhere. The currently accepted age of the LIS marine transgression (12,455 ¹⁴C years BP) is based on a ¹⁴C date from a depth of 487 cm in core LISAT12 (Stone et al., 2005). This sample was ‘bulk organic material including shells’ (verbatim Stone et al., 2005, p. 65; radiocarbon date GX-18094, Krueger labs), but the Krueger laboratory report states that carbonate was dissolved prior to analysis. We calibrated this datum point with the 1998 atmospheric decadal data set and will discuss it as an organic matter age.

Results

The data (Table 1) show an age profile through the marine section of core LISAT12 from 69 cm to 540 cm depth. The carbonate ages show a smooth, curvilinear relation between calendar age and depth-in-core (Figure 1). The ¹⁴C ages from terrestrial macrophytes are all substantially older than the carbonate ages from the same samples, with differences of 1400 to 3300 years, except for the sample at 239 cm, where the organic carbon and carbonate ages are very close. We fitted a polynomial to the age-depth carbonate data points and obtained a ‘virtual carbonate age’ for sample GX-18094 (Stone et al., 2005) of 9767 BP, with a corresponding calibrated ¹⁴C age of the bulk organic matter of 14,722 calendar years BP, a difference of ~5000 years.

The mollusk ages have an error of + several 100 years as a result of the uncertainty in the ¹⁴C reservoir and errors in the ¹⁴C determination (+/- 50 to 75 years) as well as the usual calibration uncertainty. Nonetheless, this total error is substantially smaller than the differences with the calibrated ¹⁴C ages from coexisting terrestrial organic matter (thousands of years). The abundant wood and leaf fragments in the marine silts are substantially older than the associated carbonate ages, and the radiocarbon age data set from bulk organic matter of several LISAT cores (Lewis, unpublished data) shows a very poor correlation between age and depth-in-core. We therefore argue that macrophyte organic carbon ages represent the time that plants died, and not of sedimentation in LIS: the wood fragments must have resided in soils or periglacial lake sediments on land for several 1000 years prior to arrival in LIS (see also Ridge, 2003). The sample at 239 cm depth is unusual with its concordant ages, and supposedly this organic matter was transported rapidly to LIS after plant die off. Bulk organic matter in the marine sediment and varved lake beds also contains abundant terrestrial plant debris and likewise gives the age of formation of the original organic matter. An event chronology based on such ¹⁴C ages provides a history that may be several 1000 years too old.

Discussion

Our carbonate ¹⁴C ages provide a new look at the Holocene history of LIS, indicating that the marine transgression at the site of core LISAT12 occurred only at about 10,000 calendar years BP, much later than previously proposed (14,000 - 16,000 calendar years BP, Stone et al., 2005). The ages of several other post-glacial events are also not well constrained (e.g., the draining of Glacial Lake Connecticut, the timing of isostatic rebound, the draining of Lake Hitchcock), and the late Pleistocene-Holocene history of LIS may have to be reconsidered (Figure 2).

We assume that Glacial Lake Connecticut started to form at about 20,000 calendar years BP and Lake Hitchcock started to form at 19,100 calendar years BP (Ridge, 2003). Varve counts in sediments recovered in vibracores from LIS (Szak, 1987: Reimer, 1986) combined with the thickness of lake beds (> 150 m, Stone et al., 2005) suggest that Glacial Lake Connecticut persisted for 3500-6500 years, depending on varve thickness variations between cores. The lake thus must have drained between 16,500 and 13,500 calendar years BP, a range overlapping with the age of draining of Lake Hitchcock (14,100 calendar years, Ridge, 2003). We speculate that, if Lake Hitchcock drained catastrophically, the flood may have led to the failure of the barrier retaining Glacial Lake Connecticut, with both lakes draining into the Atlantic Ocean almost simultaneously (around 14,000 calendar years BP). LIS was then dry for about 4000 years, which period included the cold and windy Younger Dryas interval (12,800 to 11,500 calendar years BP). During the Younger Dryas and till the time of inundation, the dry bottom of LIS may have been one of the source areas for e.g., the Windwood loess deposits on Long Island (Kundic and Hanson, 2003) that cover an age of 14,000 to 8500 calendar years BP.

The ocean invaded LIS starting at ~10,000 calendar years BP, possibly coinciding with melt water pulse 1b of Fairbanks et al. (1992), rather than with the older melt water pulse 1a as proposed by Stone et al. (2005). The sedimentary record of core LISAT12 indicates that the core site remained in the inter-tidal to shallow sub-tidal flat sedimentary environment for about 1000 years (Szak, 1987), with the sedimentation rate roughly equal to the rate of relative sea level rise (about 0.5 cm/yr). The sediment lithology in the middle to higher part of the marine core section suggests a larger water dept, and the rate of relative sea level rise was thus higher than the estimated sedimentation rate.

Conclusions

Over the last 20,000 years, the glacial ice sheet retreated, and lakes, including Glacial Lake Connecticut formed, followed by draining of the lake and the formation of a dry basin, followed by marine transgression and formation of modern LIS. We argue that the timing of these consecutive events should be revised, based on the new ¹⁴C dates on carbonate from core LISAT12 in eastern LIS, because the ¹⁴C ages derived from terrestrial organic matter in marine LIS sediments are up to 5000 years too old. We explain these discrepancies as a result of intermediate-term storage of terrestrial organic material in lake deposits or soils, and subsequent re-sedimentation in LIS. We speculate that the sudden drainage of Lake Hitchcock around 14,100 calendar years BP might have triggered the drainage of Glacial Lake Connecticut. The LIS basin was then dry for a 4000 year period that included the Younger Dryas, and may have served as a sediment source for loess deposits in Connecticut and Long Island. We date the marine incursion at 10,000 calendar years BP, which most likely coincided with the later stages of melt-water pulse 1b of Fairbanks et al. (1992).

References

• Antevs, E., 1922, The recession of the last ice sheet in New England. Am. Geograph. Society Res. Series v.11, 120p.
• Brigham-Grette, J., Rittenour, T.M., and Mann, M.E., 2001, Drainage history of Glacial Lake Hitchcock and paleoclimatic implications of late quaternary sediments and terraces in the central CT valley. GSA, Abstracts with Programs, v.33, p.14.
• Fairbanks, R. G., 1989. A 17,000-year glacio-eustatic sea level record: influence of glacial melting rates on Younger Dryas events and deep ocean circulation. Nature, v. 342, p. 637-642.
• Fairbanks, R.G., Charles, C.D., and Wright, J.D., 1992, Origin of global meltwater pulses. In: Radiocarbon after four decades, ed. Taylor, R.E., p. 473-499
• Fenster, M.S., 1995, The origin and evolution of the sand sheet facies, eastern Long Island Sound. PhD thesis, Boston University, p. 325.
• Groner, M., 2004, Carbon dynamics in Long Island Sound: a study using stable and radioactive isotopes. Senior undergraduate thesis, Wesleyan University, Middletown CT, USA, p.138
• Knebel, H., Signell, R.P., Rendigs, R.R., Poppe, L.J., and List, J.H., 1999, Seafloor environments in the Long Island Sound estuarine system. Marine Geology, v. 155, p. 277-318.
• Kundic, V., and Hanson G.N., 2003, Geochemistry of Loess on Long Island. http://pbisotopes.ess.sunysb.edu/lig/Conferences/abstracts-04/kundic/kundic.htm
• Lewis, R. S., and Stone, J. R., 1991, Late Quaternary Stratigraphy and Depositional History of the Long Island Sound Basin. J. Coastal Res., Spec. Issue, v. 11, p. 1-23.
• Lewis, R.S., and DiGiacomo-Cohen, M.L., 2000, A review of the geologic framework of the Long Island Sound Basin with some observations relating to postglacial sedimentation. J. Coastal Research, v.16, p. 522-532.
• Reimer, G.E., 1986, Sedimentology and interpretation of environment of deposition of rhythmically laminated silt and clay facies of LISAT6 core, Long Island Sound. Report from Rutgers University.
• Ridge, J.C., 2003, The last deglaciation of the northeastern United States: a combined varve, paleomagnetic and calibrated ¹⁴C chronology. In: Geoarcheology of landscapes in the glacial northeast, eds. Cremeens, D.L. and Hart, J.P., New York State Museum Bulletin, v. 497, p. 15-45.
• Rittenour, T.M., Brigham-Grette, J., and Mann, M.E., 2000, El Nino-like climate teleconnections in New England during the late Pleistocene. Science, v. 288, p. 1039-1042.
• Stone, J.R., Schafer, J.P., London E.H., DiGiacomo-Cohen, M.L., Lewis, R.S., and Thompson, W.B., 2005, Quaternary Geologic Map of Connecticut and Long Island Sound Basin. Scientific Investigations MAP 2784 with explanations, p. 72.
• Stuiver, J.R. and Reimer, P.J., 1993, Extended ¹⁴C database and revised CALIB 3.9 ¹⁴C calibration program. Radiocarbon, v. 35, p. 215-230.
• Stuiver, J.R., et al., 1998, INTCAL98 radiocarbon age calibration 0-24,000 BP. Radiocarbon, v.  40, p. 1041-1083.
• Szak, C., 1987. The nature and timing of Late Quaternary events in eastern Long Island Sound. MSc Thesis, University of Rhode Island. p. 75.
• Thomas, H. F., 1989, Final Summary Report for a study of  'The geologic framework of Southern New England: Vibracoring of Long Island Sound', Minerals Management Service, department of the Interior, Subagreement MMS/AASG.

Acknowledgements

We thank Ralph Lewis for his suggestion that we study the LISAT cores in more detail and for all his help in providing information on these cores. We thank the Connecticut Sea Grant College Program for funding. We also thank Ellen Mecray (USGS, Woods Hole, MA) and Ellen Roosen at WHOI. We thank the staff at NOSAMS, WHOI, for their help, especially Susan Handwork and Kathy Elder. Wesleyan students Tracy Krueger and Polina Rabinovitz helped with sampling and sample processing.

Depth (cm)	14C(Carbonate) Age, years BP	Calibrated Age, years BP	14C (Corg) Age, years BP	Calibrated Age, years BP	DT Corg-Carb
69	7890	8351
169	7720	8168
239	7990	8414	7180	7973	-441
355	8590	9007	10250	12021	3014
414	8960	9598	10750	12874	3276
445	9220	9833	10250	12021	2188
487*			12455	14722	4955
497	9110	9804	9840	11226	1422
540	9220	9833

*sample GX-18094 (Stone et al., 2005)

Figure 1. Calibrated radiocarbon ages of samples from core LISAT12. The size of the symbols approximates the compounded errors. Sedimentation rates for the bottom and intermediate interval are shown as well as the interpolation polynomial for the carbonate ages (heavy curve and ‘model age’ expression at the top). All macrophyte ages (except for one) are substantially older than carbonate ages from the same depth interval.

Figure 2. Chronology of the events that shaped LIS. Calibrated carbonate and organic matter ages in samples from core LISAT 12 are indicated by filled circles and open squares. Lake Hitchcock and Glacial Lake Connecticut both formed prior to 16,000 calendar years BP. Lake Hitchcock drained around 14,100 calendar years BP, which may have led to the drainage of Glacial Lake Connecticut directly afterwards. LIS was then dry for close to 4000 years, including the Younger Dryas period. The sea started to invade LIS initially only in the deepest channels, with the main incursion occurring around 10,000 calendar years BP. The marine sequence built up with initially intertidal deposits, followed by marine silts and reworked coarser deposits.