GSA Abstract
Using fossil counts to determine biodiversity is critically important to quantifying the
history of biodiversity over geologic time scales. Some fossil taxa, like brachiopods,
contain easily recognizable, non-repeating morphological features that can be
reliably counted as individual specimens. Other fossil taxa, e.g., many echinoderms,
lack convenient non-repeating morphologic features for identifying individuals within
bulk samples. How many individuals do 10 crinoid columnals or echinoid spines
represent? However, even with fossil organisms with fewer moving parts, choosing
what and how to count individual specimens can affect perceived biodiversity.
Each valve of a bivalve has only one umbo, which is frequently the most durable part
of the valve in addition to being attached to hinge structures that are useful for
identifying the bivalve’s taxonomic classification. It is therefore logical to count
bivalve umbones when determining how many individual bivalve valves are present
in a bulk sample. However, there are other non-repeating elements in the valves of
individual taxa that could also be used (e.g., the adductor scar in oysters or the
pallial sinus in some burrowing clams), as long as enough taxonomic information
remained on the fossil to identify the bivalve. Limiting the sampling of countable
specimens only to those with umbones does limit the recorded species richness as
specimens that could be classified and added to the species list. Those without
umbones are not counted. To address how this issue affects measures of
biodiversity, all identifiable bivalve fragments for one bulk sample of the Ft.
Thompson Formation (Pleistocene, Florida) have been classified and counted. In
addition, a sample of a smaller size fraction (1-4 mm sediment size versus >4 mm)
of the same bulk sample was analyzed to determine what, if any, size effects were
present. The specific biases of using this technique for quantifying biodiversity will be
addressed.
Introduction
Determining what kind of fossil can be counted as a biological individual has
proved problematic, especially with two element or multielement skeletons. From
a functional standpoint, what constitutes one individual specimen must be
determined by the type analysis being attempted and what material is available.
For instance, when testing a technique that tests for clustering of living
individuals based on their fossil remains, only whole, articulated specimens in life
position can be used (Leighton and Schneider, 2004). Whole, articulated valves
may also be used simply because they are abundant, especially when working
with sub-fossil assemblages (e.g., Rollins and West, 1997; Mondal et al., 2014)
When determining predation intensity on bivalves, whole valves must be used for
both predatory boring (e.g., Kelley, 2003; Grey et al. 2006) and comparisons of
healed to fatal shell breakage (Alexander and Dietl, 2001, 2003; Zuschin et al.
2003) because the predatory marks could appear anywhere on the valve, and
thus might be missed if valve fragments were also included in the analyses.
Whole valves are also used when the morphological measurements being
collected from the fossil material functionally can only be found if the whole valve
is included (e.g., Kelley, 1983; Bush et al., 2002; Kowalewski et al., 2002) or
because fragments could not be reliably identified to the preferred taxonomic
level (e.g., Zuschin et al., 2005). Whole valves have also been preferred for
some taphonomic studies (for examples, see Kidwell et al, 2001) and to reduce
the affect of taphonomic size sorting (for discussion, see Zuschin et al., 2003)
The definition of a whole valve versus fragments can also vary from study to
study. For instance, in this study, a whole valve is defined as all valves with a
complete margin and valve morphology. Boreholes from predators and boring
bryozoans, sponges and other organisms did not prevent a valve from being
considered a whole valve. In contrast, Davies et al. (1990) defined a whole valve
as any valve in which more than 90% of the valve was present and from which
an anterior-posterior length could be measured. All of the valves considered
whole valves in this study would also be considered whole valves for those using
Davies et al.’s (1990) definition. However, the reverse is not true, e.g., if the
missing 10% of the valve includes the valve’s hinge, it would not be considered a
whole valve in this study. Similarly, Davies et al.’s (1990) subdivision of
fragments into major fragments (20% or more of the valve present) and minor
fragments (less than 20% of the valve present) differs from the definitions used in
this study, although only in part. The definition of minor fragment used in Davies
et al. (1990) also includes several other categories, but functionally excludes
hinge fragments. Therefore the definition of minor fragments used here is
essentially the same as Davies et al. (1990) minus the size requirement.
Many studies determine the number of individuals in a sample by counting the
number of some non-repeating morphological feature, usually the umbo or hinge
for bivalves. Each valve of a bivalve has only one umbone or hinge, which
frequently contain important morphological information allowing for taxonomic
classification of the specimen. Each bivalve does have two valves, so frequently
the number of valves is divided by two to determine the number of individuals in
a sample, although not always (for discussion, see Gilinsky and Bennington,
1994 and Peters, 2004). Variations on this technique include counting all
articulated specimens as single individuals and then dividing the remaining
disarticulated hinge specimens as half individuals (e.g., < Di Geronimo and
Robba, 1976>, Bernasconi and Stanley 1997; Carnevale et al., 2011,) This
definition of an individual represents the <insert summary of Gilinsky and
Bennington here).
As long as the individual abundance determined by the number of individuals of
each taxon within a sample is used in the limited manner prescribed in the study
for which the specimens were collected, differences in determining what
constitutes an individual should matter little. However, if the taxonomic data are
then compared across studies with different definitions of what constitutes an
individual, as with large databases, these differences could create diversity
artifacts, just as other methodological differences like the chosen sieve size can
affect the results of taphonomic (e.g., Kidwell et al., 2001) and biodiversity
analyses (e.g., <get references>). In this study, the affect of including whole
valves, hinge fragments and minor fragments on diversity measurements will be
analyzed to determine how including fragments affects bivalve diversity from a
Pleistocene shell bed.
Whole valve > 90% of specimen present, able to measure shell length; hinge
fragments had shell beak; minor fragments were all others (Staff and Powell
1990; Davies et al., 1990, Zuschin et al., 2003)
Whole valves only because of difficulty identifying to species level (Zuschin et al
2005)
Articulated shell= 1 individual, disarticulated valves including umbo and/or hinge
divided by two (mathematically the same as dividing all valves by 2 – Di
Geronimo and Robba 1976, Carnevale et al 2011, Bernasconi and Stanley 1997)
Tabulation of many studies and how they defined an individual (Kidwell et al
2001) – that study used whole and fragments because they studied taphonomic
alteration
Whole valves and fragments of brachiopods, taphonomy of color (Kolbe et al
2011)
All fragments measured (studying predation regurgitation of shells by
durophagous predators – Zaton and Salamon 2008)
Fragments – measuring trends in fragmentation (Salamon et al 2014)
Individual = 1 bivalve/brachiopod valve or 1 articulated specimen. Discusses the
problem of doing this (Peters 2004)
Considered only articulated specimens – necessary for testing technique that
measured original paleocommunity clustering (Leighton and Schneider 2004).
Whole valves used to determine predation parameters such as boring frequency
(Kelley 2003) and shell breakage (Alexander and Dietl 2003).
Specimen sufficiently complete to describe taxonomy, original dimensions,
paleoecological and taphonomic characters – usually only true of “nearly
complete” valves (Kowalewski et al. 2002)
Vertebrate techniques
Field techniques for collecting high quality bone survey (Behrensmeyer and Barry
2005);
Used Gilinksy and Bennington’s MNI and the maximum number represented
(White et al 1998)
Used several techniques to quantify number of mammals (Badgley 1986)
Materials and Methods
Specimens used in these analyses were derived from 12 sediment bulk samples
of the Ft. Thompson Formation collected from a single outcrop at the Caloosa
Shell Quarry in Hillsborough County, Florida. These samples had been
previously wet washed and sieved. Bivalve hinge fragments (including whole
valves) and specimens of other taxa (mostly gastropods) were sorted from the >4
mm sediment fraction. All twelve Ft. Thompson samples were processed in the
same way, yielding bivalve hinge fragments and other taxa; this material has
been used in previous studies (Daley xxx Daley xxx). For this study, the >4 mm
sediment fraction of one of the Ft. Thompson samples (CSQ4+0B) was sorted
again to extract other identifiable bivalve shell material.
The sample was sorted under a magnifying desk lamp to attain all bivalve
fragments. Morphological features like shell ornamentation, shell microstructure
and valve shape were used to narrow down the taxonomic classification of each
fragment. Reference samples of all previously identified Ft. Thompson bivalves
were used to aid in identifying the fragments. Only fragments that could be
confidently assigned to a species were used in further analysis.
All fragments were divided into three categories, whole valves (W), hinge
fragments (H) and minor fragments (M). Whole valves were complete valves with
a complete margin. Hinge fragments contained the beak of the valve and the
central hinge structure (e.g., cardinal teeth and chondrophores, as appropriate).
Hinge fragments did not necessarily contain lateral hinge structures like lateral
teeth to be assigned to this category. Minor fragments (sensu Staff et al. xxxx)
were all other identifiable fragments of valves. Specimens were assigned to
these categories regardless of taphonomic damage such as boring and abrasion.
Specimens that could not be identified to the species level due to such damage
were not included in this analysis.
Rarefied diversity and other diversity measures were determined using the
rarefaction analysis in the Past software package (Hammer et al. 2001).
Results
Two new species recovered from minor fragments (Spisula soilidissima similis
and Trachycardium egmontianum) – both found in other samples, both large,
fragile shells
Discussion
Specific biases:
Over represented as M: easily fragmented taxa (Ostrea, Brachidontes,
Argopecten, Dosinia)
Over represented as H and W: small species P. nassula, Mulinia, P. multilineata
+ species that are hard to identify without hinge fragments
More W than H: P. amianta (small and robust), Abra aequalis (small, fragile and
smooth shelled), Transenella stimpsoni (small), Corbula contracta (small), Tellina
aequistriata (small, fragile, smooth shelled),
More H than W: Tagelus divisus (small and fragile), Tellina sp. A (small and
fragile), Anodontia alba (large and fragile), Diplodonta punctata (small and
fragile),
Fragmentation %
small
medium
large
Fragile
Average Robust
70.3
11.6
25.1
98.5
92.3
94.4
99.1 64.8*
82.4*