Page 4 - slootman_et_al_2016
P. 4

A. Slootman et al. / Earth and Planetary Science Letters 444 (2016) 192–204  195

‘dune cross-bedded clinoform units’. The succession of dune cross-     prior to imaging with a Jeol JSM 7001F Scanning Electron Mi-
bedded clinoform units is repeatedly interrupted by up to 10 m         croscope (SEM). A semi-qualitative scheme was developed for the
thick, erosion-based sandy to gravelly clinoform units composed of     assessment of the petrographic state of sample preservation. Thir-
amalgamated sets of backset-bedded deposits (grey in Fig. 2), as-      teen out of the 60 samples passed the SEM screening test and were
cribed to the cyclic formation by numerous hydraulic jumps (cf.        subsequently prepared for 87Sr/86Sr measurement (see Supplemen-
Cartigny et al., 2014). Such prograding bodies are here referred to    tary Material for details). 87Sr/86Sr was measured with a Thermo
as ‘backset-bedded clinoform units’, which are best developed in       Neptune PLUS Multi-Collector ICP-MS in static mode. Long-term
slope and outer ramp settings and pinch out basinwards. They con-      (over one year) external reproducibility of the SRM987 standard
tain rare escape burrows (Ophiomorpha) and accumulations of large      on the mass spectrometer is 7 ppm (one standard deviation).
ex situ bioclasts up to 15 cm in diameter. These backset-bedded
clinoform units comprise the remaining half of the sedimentary             Each of the thirteen samples was measured multiple times;
record.                                                                hence the results consist of thirteen sample mean 87Sr/86Sr val-
                                                                       ues and accompanying sample standard deviations. The table of
    A clinoform couplet is defined as the combination of one dune       McArthur et al. (2012) gives the mean, lower limit and upper limit
cross-bedded clinoform unit and one backset-bedded clinoform           87Sr/86Sr value corresponding to a 95% confidence interval with
unit. The clinoform succession of Favignana Island contains ca. 35     time steps of 50 kyr, which were decreased to increments of 10 kyr
of such clinoform couplets, the thickest and steepest of which crop    by linear interpolation. Following the guidelines of McArthur et al.
out in the large sea cliff shown in Fig. 2. The distal termination     (2012) and assuming normal distribution of errors, each sample
of the succession in the SE reflects the final position reached by       was assigned an estimated mean age and a standard deviation. For
progradation. The clinoform succession is believed to have origi-      each sample the cumulative probability for every possible combi-
nally extended towards the inferred carbonate factory on the sub-      nation of age (Ma) and duration (i.e. time span (kyr) in which all
marine high between the islands of Favignana and Levanzo (Fig. 1).     samples were deposited) were calculated, in order to answer the
                                                                       question: What is the probability P i,T ,t that sample i, with mean
3. Methods
                                                                       age μi and standard deviation σi , was deposited within a period of
    To unravel the nature and timing of the events that built the
Favignana carbonate ramp succession, three time scales need to         T kyr beginning at time t Ma? Subsequently, the cumulative prob-
be constrained: (1) time contained within one clinoform couplet,       ability of all samples for every possible combination of T and t
(2) duration of flow events associated with the deposition of in-       was obtained by calculation of the product of P i,T ,t for i that goes
dividual backset-bedded clinoform units, and (3) cumulative time       from 1 to 13 (see Supplementary Material for details). The results
required to construct the deposits of a single dune cross-bedded       are close to but not normally distributed.
clinoform unit. The time scale for deposition of one clinoform cou-
plet was obtained from strontium isotope stratigraphy. Flow dura-      3.2. Duration of flow events
tion associated with individual clinoform units was estimated by
dividing sediment volume by their respective fluxes, which follow           A scale order for the duration of flow events was estimated by
from palaeohydraulic reconstructions.                                  dividing cross-sectional sediment volume by sediment discharge
                                                                       per unit width (i.e. specific sediment discharge), derived from
3.1. Strontium isotope stratigraphy                                    palaeohydraulic reconstructions. The flow-parallel cross-sectional
                                                                       area of the different clinoform units was measured from the cliff
    Strontium isotope stratigraphy is based on direct comparison       section in Fig. 2. Specific sediment discharge is the product of
of measured 87Sr/86Sr in a sample to the known 87Sr/86Sr vari-         average flow velocity, flow thickness and sediment concentration.
ation in sea water calibrated through time (Smalley et al., 1994;      For dune cross-bedded clinoform units, the link between sedimen-
McArthur et al., 2012). This method thus requires the measure-         tary structures and flow properties was quantified using the field-
ment of strontium isotope ratio in minerals that precipitated from     based bedform stability diagram of Van den Berg and Van Gelder
sea water, such as biogenic calcite, which have not been diage-        (1993). Associated sediment fluxes were calculated with classical
netically altered. Quaternary 87Sr/86Sr variation is characterised by  sediment transport predictors for bed-load-dominated transport
a rapid monotonous increase, making this period very suitable for      (Meyer-Peter and Mueller, 1948). For backset-bedded clinoform
strontium isotope dating (McArthur et al., 2012). Samples of red al-   units, on the other hand, the experimental relation between an-
gal composition are recognised as highly reliable in recording the     tidune length, flow velocity and flow thickness was used (Kennedy,
original sea water strontium isotope ratio, but nonetheless need to    1963; Hand, 1974; Cartigny et al., 2014). These experimental rela-
undergo a screening procedure to determine the state of sample         tions were previously validated in modern environments by e.g.
preservation (Smalley et al., 1994).                                   Alexander and Fielding (1997), who showed that the wavelength
                                                                       of gravel antidunes in a bankfull tropical river correspond to flow
    Six stratigraphically distributed sample locations were selected,  depth and velocity. Backset-bedding formed predominantly un-
comprising a succession of 20 clinoform couplets in the large sea      der suspension load transport, hence on the basis of previous
cliff and the eastern part of the island (Fig. 1A). From each lo-      experimental work (Cartigny et al., 2013), which links sediment
cation ten fossil rhodoliths (red algal spheres) with a minimum        concentration-dependent turbulence damping to the formation of
diameter of 5 cm were collected. After superficial cleaning with        ‘spaced stratification’ (sensu Hiscott, 1994), sediment concentra-
a brush and ultrasonic bath, each rhodolith was carefully frag-        tions were inferred from this type of stratification observed in
mented into 5–10 mg pieces. Subsequently, about 50 mg of each          outcrop. The experimentally derived relations can be applied on a
sample was hand-picked using a binocular microscope. Criteria for      field-scale as demonstrated by Cartigny et al. (2013) and are there-
selection were the absence of sand grains, the absence of visible      fore justifiably used here to estimate the duration of flow events.
non-primary minerals, and the absence of visible, recent organic
contamination. For each of the 60 rhodoliths, one fragment was             The transport of bed and suspended load was derived us-
mounted on a conductive aluminium support with double-sided            ing models for siliciclastic particles. Physical experiments, how-
conductive carbon tape. An ultra-thin, ca. 10 nm gold coating was      ever, have shown that biogenic carbonate sands have lower mo-
then deposited on the samples by low vacuum sputter coating            bility thresholds and settling velocities than siliciclastic grains, due
                                                                       to differences in grain shape, density and intragranular porosity
                                                                       (Prager et al., 1996; Smith and Cheung, 2004, and refs. therein).
                                                                       Initiation of motion of carbonate particles was found to occur for
   1   2   3   4   5   6   7   8   9