Figure 8: Panel A: Conceptualized prediction for Sr/Ca vs. Mg/Ca relationships of drip waters inside HRC as a function of rainfall amount (wet vs. dry) and mixing between two geochemical endmembers: HRC limestone (Sr/Mg = 0.022 mMol Mol-1) and Quincy,

1
Calcite Farming in Florida Caves Isotope (d18O
and d13C) and trace element paleoproxy
calibrations Darrel Tremaine1
(tremaine_at_magnet.fsu.edu), B. P. Kilgore1, P. N.
Froelich2 (pfroelich_at_comcast.net) 1 Department of
Oceanography, NHMFL - Geochemistry, 1800 E Paul
Dirac Drive, Tallahassee, Florida, 32310. United
States 2 Froelich Educational Services, 3402
Cameron Chase Drive, Tallahassee, Florida 32309.
United States
Abstract
Density Driven Ventilation and Resulting Cave Air
Isotopic Composition
Hydrochemistry X/CaH2O and X/CaCaCO3
Figure 4 Colored solid circles denote
atmospheric samples above cave (free air under
tree canopy). Colored solid squares represent
seasonal soil gas samples. Triangles denote
in-cave air transect snapshots. Each endmember
average (atmosphere and soil gas) is
circumscribed by a 2s (range) oval. The
best-fit line includes soil gas, atmospheric and
in-cave samples from the current study. The
equation describing the best-fit line (d13C
5697 / CO2 - 22.4, R2 0.94) indicates that the
isotopic composition of HRC air is a function of
mixing in a two-endmember system (soil gas and
atmosphere) 5, 6, 7, 8.
Cave formations, or speleothems, are important
for their high-resolution records of continental
paleoclimate 1, 2, 3, 4. Site-specific
calibrations between speleothem proxies and
environmental parameters are often missing so
that interpretations as changes in rainfall,
temperature, or vegetation cover are only
qualitative. Here we report the results of a
20-month paleoproxy calibration in Hollow Ridge
Cave (HRC), Marianna, FL, the second phase of an
investigation into the effects of cave-air
ventilation on speleothem growth rates,
depositional timing, and isotopic and chemical
composition. High-resolution collections of
rainfall, drip waters and modern seasonally
farmed calcite coupled with continuously
monitored cave-air CO2, Radon-222, temperature,
humidity, barometric pressure, drip rates, and
airflow velocity and direction has allowed
characterization of the in situ relationships
among (1) cave temperature, drip d18O, and
calcite d18O (2) drip water d13CDIC, ventilation
rate, and calcite d13C and (3) net rainfall
amount, drip water chemistry, and resulting
calcite chemistry. Farmed cave calcite slowly
precipitated in oxygen isotopic equilibrium with
cave drip water at a known temperature is 0.82
heavier in d18O than predicted from both theory
and laboratory-based calibrations i.e., the
equilibrium H2O-CaCO3 fractionation factor (a) is
larger than generally accepted (Fig. 5) 8,
10-15. Combining our d18O-calcite-H2O-temperature
data with that of other well-constrained cave
studies results in a new calcite temperature
calibration of 1000ln(a) 16.1(103T-1) - 24.6,
which exhibits a ?d18O / ?T slope of -0.177,
lower than the accepted -0.206 / C. Farmed
calcite also exhibits a d13CCaCO3 enrichment near
cave entrances that decreases linearly with
distance into the cave interior (a -7 gradient),
suggesting that d13CCaCO3 is controlled by carbon
isotope equilibration between drip waters and
cave air, which we show to be a function of
outside air ventilation. A whole cave Hendy
test 16 illustrates that ventilation has no
observable effect on modern d18OCaCO3, but drives
d13CCaCO3 heavy by 1.9 VPDB (Fig. 7). Major
cations, anions, and selected trace element
measurements in rainfall, drip waters, and farmed
calcite reveal that farmed calcite X/Ca ratios (X
Li, Na, Mg, Si, K, Mn, Sr, Ba, U) track
short-term variation in drip water chemistry and
can be site-specific (drip site) calibrated to
net rainfall, yet appear random on a whole cave
basis. Mixing of the Mg/Ca and Sr/Ca signals
between pure rainwater and limestone dissolution
endmembers provides an elemental tracer for karst
recharge and throughput. This work provides
a multivariate calibration of Holocene-aged
speleothems in HRC and will result in a more
precise understanding of the effects of northern
Gulf of Mexico monsoonal intensity on the
isotopic and chemical composition of HRC
speleothems in subtropical north Florida.
8
3
4
3
Figure 8 Panel A Conceptualized prediction for
Sr/Ca vs. Mg/Ca relationships of drip waters
inside HRC as a function of rainfall amount (wet
vs. dry) and mixing between two geochemical
endmembers HRC limestone (Sr/Mg 0.022 mMol
Mol-1) and Quincy, Florida rainfall (Sr/Mg
0.0019 mMol Mol-1). The X,Y origin of the plot
represents very wet conditions (diluted Sr and
Mg), while the distal ends of both axes represent
dry conditions (evaporation enrichment of
dissolved limestone and rainfall Sr and Mg). High
rainfall (Wet Conditions) dissolves more
limestone, increasing Ca2 (aq) and decreasing
Sr/Ca and Mg/Ca ratios. Increased hydraulic head
results in shorter response times and less
evapotranspiration (i.e. higher (P-E)/P). Low
rainfall (Dry Conditions) promotes high
evapotranspiration (lower (P-E)/P), less
hydraulic head, increased residence time, and
increased Sr/Ca and Mg/Ca concentration ratios.
Panel B Measured Sr/Ca vs. Mg/Ca (mMol Mol-1) in
HRC drip waters and ground water. Also plotted
are Quincy rainfall (Sr/Mg 0.0019 mMol Mol-1)
and HRC limestone (Sr/Mg 0.022 mMol Mol-1)
(dashed lines). Aqueous cation chemistry is
distinct at each drip site. Dry period (2008)
drip waters have higher Sr/Ca and Mg/Ca ratios,
while 2009-2010 drips have steadily decreasing
Sr/Ca and Mg/Ca ratios 9.
Figure 3 A three-dimensional cartoon of HRC
displaying seasonal ventilation patterns. Dense
(cold and dry) winter air (Blue Lines) flows into
Entrance A and is convectively warmed and
moistened as it travels upward through the
Ballroom, Sump, and Smith and Jones Room and out
of the top of the cave near the Fissure Crack.
Air also travels through the Tube and is cycled
through the Retort Room, back through the
Entrance Room and into the Ballroom. Buoyant
(warm and wet) summer air (Orange Lines) flows in
the top of the cave and is convectively cooled as
it travels downward in two paths (1) through the
Fissure Crack, Retort Room and out through
Entrance A, and (2) through the Smith and Jones
Room, Sump, Ballroom and the Throat leaving
Entrance A. Calcite growth rates at Smith and
Jones farming locations also indicate that
atmospheric air is entrained through cracks in
the eastern edge of the cave. Figure presented
with permission from D. Nof, FSU 8.
Site Description Hollow Ridge Cave
9A
B
Calcite d18O Nature vs. Laboratory
Calcite d13C Ventilation Effects
5
6
CS2
CS1
Figure 9 Mg/Ca and Sr/Ca in HRC drip waters and
farmed calcite. Panel A Sr/Ca vs. Mg/Ca in HRC
drip and ground waters, plotted with limestone
and rainfall endmember ratios (solid lines).
Panel B Sr/Ca vs. Mg/Ca in HRC calcite plotted
with best-fit lines through each growth site
cluster. Line segments are denoted by
drip/farming site (1) Ballroom and Duece (2)
Richard, Lucky, Smith Jones A-1 (3) Smith
Jones A-2 and Smith Jones B 9.
Conclusions
Figure 1 Figure 1 Plan view of HRC, Marianna,
FL. Continuous time series instrument groups are
shown as colored circles. Calcite farming
locations are denoted by colored stars and
triangles. Blue and red squares denote sites of
drip and groundwater sampling for trace elements,
dD and d18O, major cations and anions, dissolved
inorganic carbon (TCO2 DIC HCO3- 2CO32-),
and pH. Entrance D serves as the main entrance
and is elevated 7 meters above the other
entrances. Shaded gray areas indicate the
general locations of formations. Map adapted
from Boyer (1975). Cave owned and managed by
Southeastern Cave Conservancy Incorporated, P.O.
Box 71857, Chattanooga, TN 37407-0857, Chairman
Brian Krebs Cave Steward Allen Mosler.

• Analyses of farmed calcite d18O and drip water
  d18O coupled with in situ temperature reveal that
  natural cave calcite is heavier than predicted
  by either theory or lab experiments by 0.82
  0.24 VPDB. Comparison with other modern cave
  data suggests a new empirical paleotemperature
  equation 1000 ln a 16.1 (103 T-1) 24.6

• Analyses of farmed calcite d13C and drip water
  d13CDIC reveals that ventilation drives calcite
  heavier than predicted from equilibrium
  precipitation with drip water DIC. We use a
  whole cave Hendy test to illustrate a
  ventilation-induced shift in calcite d13C of
  1.9 VPDB. Speleothem d13C is often regarded as
  a paleovegetation proxy. The implication of this
  work is that since all caves must breath to
  precipitate calcite, speleothem d13C may actually
  be a robust ventilation proxy.

Figure 6 Y-axis Predicted calcite d13C values
(Black Line) based on 5050 stoichiometric
contributions of Oligocene limestone (d13C -2
) and soil gas (d13C -20.7 ) to drip water
DIC plotted with measured drip water DIC (Filled
Blue Circles) and farmed calcite average d13C
(Open Squares) for each season. X-axis Fraction
ventilated () at each farming location LA
(Larry), LU (Lucky), D (Duece), BR (Ballroom),
SJB (Smith Jones Room B), SJA2 (Smith Jones Room
A2), SJA1 (Smith Jones Room A1), and R (Richard).
Fraction ventilated was calculated by isotopic
mass balance of cave air CO2 grab samples. Shaded
area indicates the observed 0.66 (1s) range
in soil gas d13C. During winter, spring, and
summer Smith and Jones A1 and Richard calcite
d13C values are consistent with predicted drip
13C. During spring and summer Duece calcite is
also consistent with predicted drip 13C. All
other sites exhibit a significant enrichment in
13C suggesting either rapid CO2 degassing (Hendy,
1971) or equilibrium precipitation with some
percentage of cave air-derived DIC. In both
cases, a shift toward higher calcite d13C values
indicates ventilation effects on calcite isotopic
composition 8.
Materials and Methods

• Air temperature, ?CO2, relative humidity,
  barometric pressure, and 222Rn activity are
  continuously monitored at both Cave Station 1
  (CS1) and Cave Station 2 (CS2). An acoustic
  anemometer records air flow velocity and
  direction through Entrance A, allowing
  characterization of cave ventilation dynamics
  either breathing in, breathing out, or stagnant.
  Drip rate is continuously monitored in the
  Ballroom (Figure 1).

• High resolution drip water collection and
  calcite farming have revealed hydrochemical
  clustering that is likely controlled by
  stratigraphic chemical inhomogeneities in host
  limestone. Moreover, aqueous Mg/Ca and Sr/Ca
  ratios are shown to decrease during very wet
  periods, and the signal is captured in modern
  calcite. This work provides modern ground
  truthing to a long-standing hypothesis that
  speleothem Mg/Ca and Sr/Ca are accurate
  paleohydrologic proxies, but must be calibrated
  at each drip site.

• Modern calcite is farmed on glass microscope
  slides (2 x 1) installed atop actively growing
  stalagmites under drip sites throughout HRC
  (Figure 2). Each plate is attached to a
  stalagmite with a flexible wire mesh. Samples
  are analyzed on an Agilent Quadrupole 7500cs
  ICP-MS for major and trace element X/Ca ratios,
  and for isotopes (d13C and d18O) on a Finnigan
  DELTA plus XP IR-MS.

Figure 5 1000 ln a vs. 103 T-1 for calcite where
the right Y-Axis is HRC calcite d18O (VSMOW), and
the top X-Axis is Temperature (C) plotted for
Hollow Ridge Cave calcite (Black Open Squares).
HRC data shown are average values for each
farming site by season. Other modern cave
studies (Solid and Crossed Squares),
laboratory-based inorganic precipitation values
(Circles), and theoretical calculations (Lines)
are reported from those studies that reported
calcite d18O H2O data. The red line is the
best-fit line from Kim and ONeil (1997) as
modified by Kim et al. (2007) 5 mMol L-1 Ca2
(Red Filled Circles) where 1000 ln a 18.03 (103
T-1) - 32.17. Hollow Ridge Cave drip samples
ranged from 0.6 1.8 mMol L-1 Ca2 and average
drip d18OH2O was -3.75 0.33 (2s range)
(VSMOW). The light-blue dashed line through cave
data is the linear best-fit line 1000 ln a
16.1 0.65 (103 T-1) 24.6 2.2 (98 C.I.)
(R2 0.94) for all cave deposits. This result is
significantly different from Kim and ONeil
(1997) 5 mMol data line at the 98 confidence
interval. Analytical uncertainty for single
measurements during the Hollow Ridge Cave study
was less than 0.14 (1s) 8. 1000 ln a
1000ln (1000d18OCalcite) / (1000d18ODrip
H2O) where for HRC d18ODrip H2O -3.75 0.33
(VSMOW).
7
Future Research Goals

• Water samples were collected bi-weekly for 20
  months. Samples were analyzed for major cations
  by quadrupole ICP-MS and major anions by Ion
  Chromatography. Samples were also analyzed for
  isotopes (dD and d18O) on a Finnigan DELTA plus
  XP IR-MS.

• This research has laid the analytical
  foundation from which to interpret ancient HRC
  speleothems. The calibrations provided by modern
  data will be used to fine tune our interpretation
  of the northern Gulf of Mexico paleoclimate as
  recorded by two Holocene-aged HRC speleothems
  Lucky and SJB-2. These two speleothems will be
  compared to other north Florida paleoclimate
  records in an effort to provide high-resolution
  data to further our understanding of the
  teleconnections between high latitude
  insolation-forced climate change, and low
  latitude climate patterns.

References
Figure 7 Farmed calcite d13C vs. d18O (Hendy
test). The top X-Axis is estimated cave
temperature (C) at each location. Open symbols
(summer green squares fall blue squares
winter purple squares and spring red squares)
represent samples that precipitated in
ventilation flow path, while closed shapes with
the same color designations represent samples
precipitated out of ventilation flow path. Open
circles represent sites that experience
intermediate ventilation and are not included in
the linear regressions. Vertical dotted arrow
represents the predicted geochemical vector for
fast drip CO2 degassing. Horizontal dotted arrow
represents predicted vector for fast water
evaporation. The diagonal dotted vector
indicates equilibrium precipitation with a
mixture of soil gas and limestone derived DIC
(d13C -11.35 ) and cave air CO2-derived DIC
(d13C - 2 ) over a range of in situ
temperatures. The dashed black line with slope ?
d13C / ? d18O 0.44 indicates the predicted
temperature dependent equilibrium slope. Solid
lines through each subset of data are linear
best-fit lines. Line A d13C 1.7 (d18O) 2.1
Line B d13C 1.9 (d18O) 4.0 . Farming
sites on Line A (Summer Lucky, Ballroom, and
Smith and Jones B Fall Lucky, Ballroom, and
Richard Winter Larry, Lucky, Ballroom, Smith
and Jones B, Smith and Jones A1 and A2 Spring
Larry, Smith and Jones B, and Smith and Jones A2
experience ventilation-driven CO2 degassing as
indicated by a 1.9 0.96 (1s) shift in d13C
CaCO3. Unlike d13C, the 2.2 range in d18O is
due to the 9.5 C difference in growth
temperatures. 8
13 Jimenez-Lopez et al. (2001) 14 Tarutani et
al. (1969) 15 Horita and Clayton (2007) 16
Hendy (1971) 17 Banner et al. (2007)
7 Tremaine (2010) 8 Tremaine et al.
(2011) 9 Tremaine and Froelich (2011) 10 Kim
and ONeil (1997) 11 ONeil et al. (1969) 12
Chacko and Deines (2008)
1 Wang, Y. et al. (2008) 2 Cruz et al.
(2009) 3 Bar-Mathews et al. (1996) 4 Baldini
et al. (2002) 5 Kowalczk (2009) 6 Kowalzck
and Froelich (2010)
This research is conducted in collaboration with
Southeastern Cave Conservancy, PO Box 71857,
Chattanooga, TN 37407-0857, Chairman Brian
Krebs Cave Steward Allen Mosler with financial
support from Dawn Baker, Plum Creek Timber,
Seattle, WA.
Figure 2 Modern calcite farming plates mounted
atop actively growing stalagmites by a flexible
wire mesh similar to methods of Banner et al.
(2007) 17.