1. science
2. medicine
3. surgery

Predatory scars in the shells of a Recent
lingulid brachiopod: Paleontological
and ecological implications
MICHAŁ KOWALEWSKI, KARL W. FLEssA, andJONATHAN D. MARCOT
Kowalewski, M., Flessa, K.W., & Marcot, J.D. 1997. Predatory scars in the shells of
a Recent lingulid brachiopod: Paleontological and ecological implications. - Acta
P alaeontologica P olonica 42, 4, 497-53f .
The paper presents the detailed quantitative study of predatory scars in the shells of an
inarticulate brachiopod: the lingulid Glottidia palmeri Dall, 1870. The scars include four
morphological types: u-shaped,pocket, crack, and miscellaneous scars. They concentrate
and open up toward the anterior shell edge. They commonly consist of a pair of scars on
the opposite valves. The analysis of 820 specimens live-collected from two intertidal
localities inthe northern Gulf of Californiaindicates that (1) 23. Vospecimens bearrepair
scars; (2) the scars vary in size from 1.5 to 24 mrrt (mean = 2.5 mr#) and all scar types
have similar size-frequency distributions; (3) the spatial distribution of scars on the shell
is non-random; (4) the anterior-posteriordistribution of scars is strongly multimodal and
suggests seasonal predation in the late fall and winter months; and (5) the frequency of
scarred specimens increases with brachiopod size and differs between the two sampled
localities, but does not vary among brachiopod patches from the same locality. The repair
scars record unsuccessful attacks by epifaunal intertidal predators with a scissors-type
weapon (birds or crabs). The high frequency of attacks, seasonal winter predation, and
previous ecological research suggest that scars were made by winteńng shorebirds
(willets or/and curlews). However, crabs cannot be entirely excluded as a possible
predator. Because repair scars representunsuccessful predation, many of the quantitative
interpretations are ambiguous. Nevertheless, the study suggests the existęnce of strong
seasonal interactions between inarticulate brachiopods and their predators. Because
shorebirds, crabs, and lingulids may have co-existed in intertidal ecosystems since the
late Mesozoic, predatory scars in lingulid shells may have potentially a 100 million year
long fossil record.
Key
predation, lingulids, brachiopods, Glottidia palmeri, shorebirds,
words:
Recent, B aja California.
Michat Kow alew ski I michael.kowalew ski@ uni-tuebingen.de], Ins Ętut P aleobiolo gii
PAN, ul. Twarda 5l/55, PL-00-8l8Warszawa, Poland.
Karl W. Flessa [[email protected]], Department of Geosciences, University of
Arizona, Tucson, AZ 85721, USA.
Jonathan D. Marcot Idinosaur@ geo.ariTona.edu], Department of Geosciences, [.lniversity of Arizona, Tucson, AZ 85721, USA.
498
Predatoryscars: KOWALEWSKI, FLESSA, & MARCOT
Introduction
Predatory attacksby durophagousorganisms(i.e.,those that damagethe prey's exoskeleton) create marks on the hard skeletons of their prey. In present-daybenthic
marine ecosystemstraces of predationare very diverse (seeVermeij 1987 for details
and references)and include drillholes made by snails, octopods,and flatworms (e.9.,
Bromley 1981; Kabat 1990; Kowalewski 1993); irregular punch holes made by
stomatopodcrabs (e.g.,Geary et al. I99I); multĘle holes causedby fish (e.g.,Norton
1988);and fracture and breakageinflicted by crabs, birds, and other predators(e.g.,
Schoener1979;Vermeij 1987, 199f; Cadóe 1968, 1989;Vale & Rex 1988, 1989;
Kropp 1992;Walker & Yamada 1993;Cadóe et al. in press).
Predatory traces provide quantifiable data on prey-predator interactions, and
thus,theyhavebeenintenselystudiedby biologistsinterestedin thebehaviorof prey
and predators,foragingstrategies,populationbiology and ecosystemdynamics(e.9.,
Berg 1976;Wiltse 1980;Garrity & Levings 1981;Fairweather1985;Vale & Rex
1988, 1989).Moreover, predatorytracesare commonly preservedin the geological
record. Indeed, they have a diverse and impressive fossil record and have been
documented,in a greatvariety,from all the periodsof the Phanerozoic(e.g.,Vermeij
1987); and even from the Late Precambrian (Bengtson & Zhao 1992). Such fossilized traces of predation (Praedichnia;see Ekdale 1985; Bromley 1996) offer
paleontologistsinvaluable insights into ancient ecosystems(e.g.,Hoffman et ąI.
t974; Thomas t976; Robba & ostinelli 1975 Sheehan& Lesperancę1976;Kitchell
et aI. 1981; Schindel et al. 1982;Hoffman & Martinell 1984; Kowalewski 1990;
Andersen et al. 199|), into the bęhavior of extinct species (e.g.,Berg & Nishenko
1975;Kitchell 1986;Kelley 1989),and into macroevolutionaryand coevolutionary
processes(e.g.,Vermeij1977,1987;Vermeijet al. 1980,1981,1982;Tayloret al.
1980;Ftirsich & Jablonski 1984;Allmon et al. 1990;Kelley 1989, 1991;Kelley &
Hansen 1996).
Here we presenta detailed quantitativeanalysis of the repair scars found on the
shells of the Recent lingulid (family Lingulidae) brachiopod Glottidia palmeri Dall,
1870, a living fossil that inhabits the present-daymacrotidal flats of the northeastern
Baja California, Mexico. In sharp contrast to repair scars in extant mollusks and
articulatebrachiopods,which have been intenselyresearched(e.g.,Vale & Rex 1988,
1989;Thayer & Allmon l99I; Alexander 1992;Kropp 1992;Walker & Yamadal993;
Walker & Voigt 1994;Cadóe et aI. inpress),repair scarsin extantlingulid brachiopods
have not been investigated.In fact, ecological interactionsbetweenRecentinarticulate
brachiopodsand their predatorshaverarely beeninvestigatedin a rigorousfashion (see
James et aI. 1992;Emig in press;and below for references).
This study has both biological and paleontologicalgoals. We documentthe morphological types of repair scars and attemptto identify the most likely predators.By
doing so, not only do we hope to improve the ecological knowledge of present-day
inarticulatebrachiopodsand their predators,but also to provide an actualisticdescription thatmay be helpful in identifying predatoryffacesencounteredin the fossil record
of lingulid brachiopodsand othershelly organisms.Also, this studyillustratesdifficulties inherentto the interpretationof repair scars.Repair scars representunsuccessful
predation events, and thus, provide ambiguous data on the ecology and behavior of
ACTA PALAEONTOLOGTCA POLONTCA 42 (4)
499
prey-predatorinteractions,both Recent and fossil (see also Shoener 1979;Schindel
et al. I98f; Vermeij 1983;Walker & Voigt 1994).
We use repair scars to evaluate issues such as the size and site selectivity of
predatoryattacks,variation in frequencyof predatoryattacksthroughtime and among
different brachiopodpopulations,and seasonalityof predation.Our quantitativeanalysis servesalso to illustrate some statisticalmethodsuseful in analyzing repair sca.rs,
both modernand fossil. In particular,we discussquantitativetechniquesthatallow for
the assessmentof randomnessin the spatialdistributionof scarsand for therecognition
of seasonalityof predationin modern and ancientecosystems.
Previous research on durophagous predation
on lingulid brachiopods
In strikingcontrastto the Recentmollusks (e.g.,Vermeij1987,1992; Kabat 1990),the
Recent lingulids has been poorly researchedfor tracesof predation.Marginal remarks
and a few rigorous observationsscatteredin the literaturesuggestthatshells of the two
Recent lingulid genera' Glottidią and Lingula, occassionally bear predatory marks.
Drillholes has been relatively best documented.Paine (1963),in his monographon
G. pyramidata, reportedthat drilled specimenswere commonly found (I4%oof shells
containeddrillholes) on a sandbarnearthe mouthof Stum Pass (westcoastof Florida).
A predatory drillhole, most likely caused by ą muricid gastropod' was found in
G. palmerl (Kowalewski & Flessa 1994).Drillholes were also documentedin fossil
lingulids from the Tertiaryof SeymourIslandSMiedmanet aL.1988;Bitner 1996)and
the easternUSA (Cooper 1988).Among other inarticulatebrachiopods,predatory
drillholes were reportedin Recent craniids (Emig in press) and in fossil acrotretids
(Miller & Sundberg 1984;Chatterton& Whitehead 1987).
Breakage and repair scars are poorly studiedin lingulids. The only more detailed
observationwas reportedby Paine (1963) who investigatedholes made by willets
(Catoptrophorussemipalmatus)and found many partially consumed specimens of
G. pyramidata: 'In most instances,the pedicle and one valve were left, although in
othersthe anteriorends of both valves had been broken off. If shell and otherdamage
is not too extensive, the brachiopod can regenerateits lost portions' (Paine 1963:
pp.I9f-193). Repair scars,most likely causedby crabs,were also found in shells of
extantLingula (C.C.Emig writtencommunication1997).Breakageand repair scars,
caused by durophagouspredators, Ta much better documented among articulate
brachiopods,both Recent and fossil (e.9.,Brunton 1966; Thayer 1985;Alexander
1986a,1986b,1990,199f;Thayer& Allmon I99I; RuggieroI99I; Jameset al. 1992:
Baliński 1993).
To our knowledge, not only traces of predation but also predatorsthat prey on
lingulids have rarely been investigatedrigorously.Nevertheless,previous researchers
of lingulid brachiopods(e.g.,Paine 1963;Worecester1968;Emig et ąI. I978; Emig
I98f ,1983)havelistedprosobranchgastropods,crabs,fish, andshorebirdsaspotential
predators.In themost comprehensivereview so far,Emig (in press)provided a detailed
list of predatorsincluding:(1) crustaceans(hermitcrabs,stonęcrabs,portunidcrabs,
crangonids, stomatopods,shrimps, amphipods);(2) echinoderms (asteroids,op-
500
Predatory scars: KOWALEWSKI,
FLESSA,
& MARCOT
hiuroids, echinoids);(3) gastropods(muricidsand naticids);(4) fishes (demersalfishes
and sturgeons);(5) shorebirds;and (6) humans.
Gut contentstudiesprovide unquestionableevidencefor successfulpredation.The
gutcontentof CatoptrophorussemipalmatusfromtheGulfof Mexico (Paine 1963)as
well as the gut contentsof Limnodoromusgrieus,and C. semipalmatusfromthe Pacific
coast of Costa Rica (Pereira 1990; Emig & Vargas 1990) indicate that those birds
commonly prey on Glottidia. Also, the asteroid Amphipolis germinata frequently
(437o) contains shells of Glottidia pyramidata (Emtg in press). Fish may also be
importantpredatorsof lingulids. In the Gulf of Mexico, G. pyramidatą is preyedupon
by estuarinesturgeons(Mason & Clugston 1993) and tonguefish(Cooper 1973).As
reported by Campbell & Campbell (manuscript),one specimen of sturgeonin the
Florida Museum collections contained 500 specimens of G. pyramidata. Lingula
anatina was commonly found in the stomachsof demersalfishes (Emig in press).
In sum, this brief review indicates that, in present-dayecosystems,a variety of
predatorsprey upon lingulid brachiopodsand can potentiallyleave predatorytraceson
shells of their victims.
Material and methods
Study Area. - The study area is the intertidal mudflat of the southernmostdelta
plain of the Colorado River, north of San Felipe, Baja California, Mexico (Fig. 1).The
tidal flat experiencesextremelyhigh semi-diurnaltides with tidal rangesreachingup
to 10 meters(Sykes 1937 Roden 1964;Thompson 1968;Bray & Robles 1991).The
intertidalflat reachesup to severalkilometersin width. This exceptionalmacrointertidal environmentis inhabitedby an abundantmacroinvertebratefauna (seeThompson
1968; Batten et ąl. 1994) that includes the infaunal lingulid brachiopod Glottidia
palmeri Dall, 1870(Dall 1920;Thompson1968;Kowalewski 1996a).The climateof
the area is hot and arid with a maximum summer temperatureof 50'C and less than
60 mm precipitationper year (Ezcurra & Rodrigues 1986;Thomson 1993).The water
of the northern Gulf of California has a salinity of 3G39%o (Brusca 1980). Mean
monthly watertemperaturereachesabout30"C in the summerand drops to 15'C in the
winter (Thomson 1993).
The tidal flat is a progradationalfeaturecomposedof fine-grainedclastic sediments
consisting predominatelyof silts and fine-grainędsands.Several seńes of longitudinally oriented,shell-rich beachridges occur in the area(Thompson1968;Kowalewski
et aI. 1994).The facies developmentof the tidal flat, discussedin detail by Thompson
(1968), has largely been controlled by the activity of the Colorado River (see also
Kowalewskj et ąl. 1994;Kowalewski & Flessa 1995).
On Glnttidia palmeri Dall, 1870.- This species is endemic to the Gulf of CaliforniaandthePacificcoastofCalifornia(Datl1870,l87t,l9f0:Lowe l933;Hertlein
& GrantL944;Thompson1968;Emig 1983;Kowalewski1996a).Itwasdescribedfirst
by Dall (1870)from the intertidalmudflatsof northeasternBaja California, where it is
patchilydistributed,but locally very abundant(Thompson1968;Kowalewski 1996a).
Like all other Recent lingulids (e.9.,Paine 1963; Emig et al. 1978; E-ig 1982;
ACTA PALAEONTOLOGTCA POLONTCA 42 (4)
501
Fig. 1. Map of the study area showing the location of the two sampling localities. A. Regional map (Baja
California and the Gulf of California). B. The northernGulf of California and the Colorado River Delta
(Fig. 1B modified from Thompson 1968;Kowalewski et al. 1994).
Treuman& Wong 1987),it is an infaunal suspension-feederthat lives in deepvertical
burrows and is anchoredby a long pedicle (seealso Kowalewski & Demko 1996).Its
shell is thin (0.2--0.3mm), spatulate,and linguliform in outline, and reaches up to
45 mm in length. The shell has low internal septa- a diagnostic featureof the genus
Glottidią (Rowell 1965; Emig 1983).The ventral,or pedicle, valve has two septa
diverging from the beak. The dorsal, or brachial, valve has one median septumand is
slightly shorterthan its ventral counterpart.The shell consists of calcium phosphate
and containsas much as 50Voorganicmatter(G. Goodfriendpersonalcommunication).
Shells display yellow or occasionally greenor dark brown color banding. Concentric
growth lines are often visible.
Previous work on G. palmert concentratedprimarily on its taxonomyand morphology (e.g.,Dall 1920;Hatai 1938;Hertlein & Grant 1944;Emig 1983).Arecent study
by Kowalewski (I996a) indicated thatG. palmeri occurs in denselyinhabitedpatches
dominatędby single age-cohorts(see below for more details).Little is known about
causes of mortality in G. palmeri. Predation and seasonalmortality during winter
502
Predatory scars: KOWALEWSKI,
FLESSA,
& MARCOT
months- when stormserodeand redepositintertidalsedimentsor when watertemperaturesdrop beyondthe brachiopod'stolerance- are most likely factors (Kowalewski &
Flessa 1994;Kowalewski 1996a).
Field,laboratory and analytical method
Ttris analysis is basedon the SźImples collectedby our researchgroup to investigatethe taphonomy,populationbiology,
biometry,and naturalhistory of G. palmeri (seeKowalewski & Flessa 1994;Batten &
Kowalewski 1995;Smith et al. 1995;Kowalewski 1996a;Kowalewski e/ al. 1997;
Anand et al. submitted).The samples were collected from two localities on the
macrotidalflat of the lower Colorado Delta (Fig. 1).
Six patchesfrom thetwo localities (Fig. 1)were sampledin March L993,Novęmber
1993,and February 7994.This analysis includes the live-collected specimens,which
were previously measured(Kowalewski I996a),but not analyzedfor scars (n = 599),
as well as additional live-collected specimens not included in the previous study
(n - ZfI). All additional specimenswere measuredto the nearest0.1 mm using
electronic or dial calipers. Size of the shell was measuredas the length of the longer,
ventral valve. This procedureis consistentwith the one employed previously (Kowalewski 1996a).
The population data,basedon 820 specimens,are summarizedin Tablę 1. Pręvious
quantitativeanalysis (Kowalewski I996a) indicated that patchesfrom Locality One
representedsingle age-cohortsfrom single spatfalls and consisted of mature,slowly
growing specimens that were 34 years old at the time of collection. The new
population data,with an increasedsample size (numberof specimens),are consistent
with the previous work. Note that the mean specimensize in Locality One increased
by f .4 mm from March 1993 to November 1993, but did not increase in size form
November 1993 to February 1994 (Table 1). This is consistent with growth ring
analyses(Batten& Kowalewski 1995;Anand et al. submitted)which indicatesthatG.
palmeri ceasesor slows its growth during winter months.Locality Two, in the north,
includedyoungerpatchesof variousage,representingI2 yearold cohorts(Table1).
Table 1. Size datafor the Glottidia palmeri patchessampledin March 1993,November 1993,and February
1994.This analysis,basedon 820 specimens,updatesthe previous analysis,based on 599 specimens(see
Kowalewski I996a: table 1).
Variable
March 1993
November 1993
February1994
154
36.8
3f5
39.2
1.86
44.8
33.s
r72
39.1
1.53
45.8
35.4
45
28.2
4.58
36.3
fr.6
rf4
24.7
2.00
f8.4
19.1
Locality One
Sample size
Mean (mm)
Standard deviation (mm)
Maximum size (mm)
Minimum size (mm)
r.4z
39.s
32.f
Locality Two
Sample size
Mean (mm)
Standard deviation (mm)
Maximum size (mm)
Minimum size (mm)
ACTA PALAEONTOLOGTCA POLONTCA 42 (4)
s03
Each specimenwas carefully examinedunder magnificationfor any marks (scars,
cracks, growth disruptions,etc.). Marks were classified according to their presumed
ońgin as: (1) reliable marks that were, in our opinion, predatory in origin; and (2)
questionablemarks which were fresh-looking,unnaturally straight,much larger than
otherscars,and cut randomly acrossthe shells at various angles(mostlikely thesewere
caused by our shovels when we were digging for specimens or during subsequent
transport_ see also Flessa et ąI. 1992).We confined our analysis primarily to the
reliable scars.
The size and location of each scar was measuredusing a ffansparentgrid. A sector
in the grid covered approximately 3 mmz. The scar size was estimatedvisually by
comparing the area covered by the scar to the grid. For example,the size of the scar
that covered areacoffespondingto 1.5 sectorwas recordedas 4.5 Ń.
The precision
of the estimatewas ł1.5 mmz (i.e.,t0.5 sector).To estimatethe positionof the scar,
the grid was placed over each specimenwith the lateral sector '1' placed at the anterior
shelledge(Fig. 2A). Note thatscarlocationcanbe describedconsistentlyusinga single
grid systemonly for specimensthat are similar in size and were collected at the same
time (theapparentposition of scars in relation to the grid 'shifts' with shell growth).
To estimatethe relative age of the scar,we measured,to the nearest0.1 mm, the
distance (Ą from the youngest shell growth ring disrupted by the scar to the anterior
(growing) edge of the shell. Assuming that scars were made at the growing edge of the
shell, d canbe used to estimatethe relative ageof the scar' ŹNwell as to calculatethe size
of the brachiopodat the time of affack (S') (seeFig. 2B). The assumptionthat scars were
madeat theanteriorshellmarginseemsreasonablein view of Paine's (1963)observations
on predatorydamagetnG. pyramidata(seeabovefor details).Nevertheless,we will note
some alternativeinterpretationsfor our data,when this assumptionis relaxed. Note that
both, d and,Sa,estimatethe relative age of the scar (Fig. 2B). However, dis a more
appropriatemeasurebecause
itminimizes thenoiseinffoducedbyvariationinbrachiopod
growthrate:modernlingulids show an asymptoticgrowthrate(e.g.,Chuang l96l;Paine
1963;Worcester1969;Kenchington& Hammond 1978),and thus,the variationinfroduced by differencesin the growth rate among specimensshould substantiallydecrease
towardthe anteriorshell edge.In the caseof complementary(paired)scars,the distance
was measuredfor the scar closer to the anteriorshell margin.
Note that the distance analysis is appropriateonly for Locality One, because all
specimens in that locality belong to the same agelsize cohort (i.e., d is directly
comparableamong the specimens).Because the brachiopodpopulationgrew f .4 mm
betweenMarch 1993 and November 1993 (see above and Table 1), we conected d
values for shell growth for the March 1993 sample (i.e., d - d + 2.4). Because the
specimensdid not grow significantly betweenNovember 1993andFebruary 1994,the
November samples were not corrected.After the correction is applied, all distance
measurementscan be treatedas if they were taken at thę Samętime in February |994.
The morphology of each scar was describedand a schematicdrawing of the scars
outline was made. Most of the scars could be classified into three morphological
categorieshereafterreferredto as: 'u-shapedscars', 'pocket scars', and 'crack scars'
(see Fig. 3), scars with other morphologieswere classified as 'miscellaneous'.The
scars were also classified as either healed (complete repair of all shell layers) or
unhealed (incompleterepair with external shell layers still disrupted).A substantial
504
Predatory scars: KOWALEWSKI, FLESSA, & MARCOT
A
Anteriorshelledge
\
ź
\
tĘ
95
o
o
oo
a
c
Ń8
.E
o
E9
Sa, Sa,
10
11
12
13
't4
15
16
't7
I
\
\
\
/
f
/
-3-2-1123
Verticalsectors
Posteriorshellend
, 5 mm ,
Fig. f , A. Grid systememployed in this study to estimatethe position of a scar on the brachiopod shell in
plan view. A hypotheticalscar is locatedin sector- 1,3. one sectorof a gńd representsa square=3 mmz in
area. B. Schematic illustating the measurementmethod used to estimated the relative age of the scar and
thebrachiopodsize at the time of theattack.Two hypotheticalscarsare usedas examples.Scar 2 is younger
thanScar 1 (i.e.,dt > dz) and was inflicted on a largerspecimen(Sat < Saz). Symbols:dt - distanceof Scar 1
from the anterior shell edge, dz - dtstanceof Scar 2 from the anterior shell edge, Sat - the size of the
brachiopod at the time of attack recordedby Scar 1, Saz _ the size of thę brachiopodat the time of attack
recordedby Scar 2, L - the anterior-posteriorlength of the shell.
proportionof the scarsconsistedof two complementarydisruptionson oppositevalves.
Such paired scarsweretreatedin our analysisas single observationsandmeasurements
were taken only for one of the two disruptions;whichever was more appropriatefor
a given measurement(seethe example of distancemeasurementabove).
All statisticalanalysis were performedusing StatisticalAnalysis System (SAS) on
the University of Arizona interactive VAX system. The statistical analyses were
performed using the SAS/STAT procedures(SAS Institute 1989, 1990a).Statistical
randomizations and simulations were written in the SAS and SAS-IML languages
(sAs Institute1990a,1990b,1991).
ACTA PALAEONTOLOGTCA POLONTCA 42 (4)
505
Results
containedat
Out of 820 live-collectedspecimensof Glottidia palmeri,267 (32.6%o)
leastone disruption(Thble2). Among them,75 specimens(9.|Voof all specimens)bear
'questionabledisruptions'.Thus, we classified only f3.47o of collected specimens
(n = 192) as having 'scars' of unquestionablybiogenic origin. The quantitativeanalysis, presentedbelow deals primarily with the unquestionablescars.
Because some specimensbear multiple scars,the total numberof scarsexceedsthe
total numberof specimens(Table2). For example,thereare2I3 unquestionablescars,
but only 192 specimensthatbear them.Similarly, 75 specimensbearonly questionable
disruptions,but thereis a total of 85 questionabledisruptions.Thus, the sample sizes
and results vary dependingon the objective of analysis. Note that it is importantto
distinguish betweenthe analysis that targetsscars and the analysis that targetsspecimens. In general,the analysis of scars provides information about the predatorand
analysesof specimensprovides information aboutthe prey. For example,65 (33.9Vo)
specimensbear healed scarsbut there are total of 70 (32.97o)healed scars.Obviously,
33.97oof the attackedbrachiopodsweręable to heal their shells completelyby thetime
of collection,and not 3f.9Vo.Similarly, we identifiedtotal of 84 u-shapedscarsbut
there are only 78 specimens bearing them. When studying the site-selectivity of
predation,one should analyze all 84 scars, even thoughthere are only 78 specimens
that contain them. We explicitly distinguish between the two types of analysis by
presentingthem in the two separatesęctions.Although this distinction may appearto
be trivial, not distinguishing between 'traces' and 'specimens with traces' would
seriously underminethe clarity of any quantitativeanalysis.
TYpes of scars. - Although scars vary in shape,most of them can be categorized
into one of the three main morphological types (Fig. 3) referredto here as 'u-shaped
scars', 'pocket scars', ffid'crack scars'. The remainingscars (17.8%o)
could not be
assignedto any of these morphological types and were classified into the 'miscellaneous'category.
Ninety disruptionswere identified as u-shapeddisruptions.Only six of them were
questionabledisruptionsand 84 were identified as true scars (39.4Voof all scars).The
scars have a regular u-shapedoutline (Fig. 3,Ą) with their two arms typically open
directly toward the anterioredge of the shell (i.e.,the scar axis is parallel to the shell
axis). The ratio betweenthe length of the arms and the width of the 'u' ranges from
2:1 to 3:1.In many cases,u-shapedscarsconsistof two complementarydisruptionsof
similar size located on the oppositevalves and situatedat a similar distancefrom the
anterioredge of the shell.
Fifty seven disruptionswere identified as pocket disruptions.Only seven of them
were questionabledisruptionsand 50 were identified as ffue scars (23.5%o
of all scars).
Pocket scars (Fig. 3B) have a much more complicatedoutline than u-shapedscarsand
can be describedas deeppocketsthatform irregulardisruptionsin the shell. They also
open toward the anteriorbut they are much more elongatedthan u-shapedscars with
theratio betweenthelengthof the arms andthe width of the scarrangingtypically from
5:1 to 10:1.Unlike the u-shapedscars,the majorityof the pocketscarshave their axis
at an acute angle (10-20') to the shell axis. In many cases,the pocket scars consist of
506
Predatory scars: KOWALEWSKI,
FLESSA,
& MARCOT
Tablef . Summary of quantitativedata on predatoryscarsin shells of Glottidiapalmeri.
Variable
Total
number
Locality One
percent
number
percent
Locality Two
number
percent
100.0
61.0
39.0
28.1
10.9
4.0
2.4
0.3
0.3
169
r56
9
I
0
2
64
66
83
34
100.0
8.1
1.1
4r.5
f3.0
18.0
r7.5
65.0
35.0
36.r
45.4
18.6
100.0
28.5
7r.5
284
81
203
100.0
28.5
7r.5
t4
f13
r43
70
84
50
4l
38
81
94
38
100.0
67.1
3f .9
39.4
f3.5
19.f
17.8
38.0
44.r
r7.8
f03
134
69
82
46
39
36
78
89
36
100.0
66.0
34.0
40.4
ff.7
19.f
1'7.8
38.4
43.8
r7.8
l0
9
I
2
4
f
2
5
f
100.0
90.0
10.0
f0.0
40.0
f0.0
f0.0
30.0
s0.0
50.0
U-shaped disruptions
questionable disruptions
scars
90
6
84
100.0
7.1
92.9
88
6
82
100.0
6.8
93.f
f
0
f
100.0
0.0
100.0
Pocket disruptions
questionabledisruptions
57
7
50
100.0
r f. 3
8'7.7
53
7
46
100.0
13.f
86.8
4
0
4
r00.0
0.0
100.0
8f
4I
4l
100.0
50.0
50.0
78
39
39
100.0
50.0
s0.0
4
2
100.0
50.0
50.0
69
100.0
44.9
55.1
65
29
36
100.0
44.6
55.4
Specimens
without any disruptions
with disruptions
with scars
with questionabledisruptions
withf disruptions
with 2 scars
with 3 disruptions
with 3 scars
820
553
f67
r9f
15
27
I7
2
2
100.0
67.4
3f .6
Specimens with scars
with 2 scars
with 3 scars
with u-shaped scars
with pocket scars
with crack scars
with miscallaneous scars
with healed scars
with unhealed scars
with a scar on a ventral valve
with a scar on a dorsal valve
with a paired scar on both valve
r92
77
183
I6
2
76
4f
65
rf1
68
88
36
r00.0
8.8
1.0
40.6
23.4
18.2
r1.7
33.9
66.r
35.4
45.8
18.8
Disruptions
questionable disruptions
scars
f98
85
213
Scars
unhealed scars
healed scars
u-shapedscars
pocket scars
crack scars
miscellaneous scars
scars on ventral valve
scars on dorsal valves
paired scars on both valves
SCATS
Crack disruptions
questionabledisruptions
SCATS
Miscellaneous disruptions
questionable disruptions
SCATS
L
78
45
35
3+
a 1
J I
38
23.+
9.r
J.J
f.r
0.f
0.f
651
391
254
183
71
f6
16
a
L
z
JJ
JL
n9
1 a
IJ
9
4
I
1
0
0
100.0
92.3
1.1
5.3
L.A
0.6
0.6
0.0
0.0
100.0
11.1
0.0
ff.f
J
JJ.J
f
22.2
L
8
1
L
5
f
+
l0
J
Z
L
f
zz.f
88.9
11.1
22.2
55.6
22.2
r00.0
28.6
1r.4
100.0
50.0
50.0
ACTA PALAEONTOLOGICA POLONICA 42 G\
507
Fig. 3. The threemain morphological types of scars
(anowed) identified in shells of the inarticulate
lingulid brachiopod Glottidia palmeri, Locality
One. Detailed descriptionsare provided in the text.
A. U-shaped scar.B. Pocket scar.C. Crack scar.All
specimens reposited with the lingulid brachiopod
collection of the InvertebratePaleontologyLaboratory (Departmentof Geosciences'UniversĘ of Arizona,Tucson,USA).
two complementaryscars of similar size located on opposite valves and situatedat
a similar distancefrom the anterioredge of the shell. Thanks to all thosedifferences,
and the factthat very few scars displayed intermediatemorphological characteristics,
the pocket scars were easy to distinguishfrom the u-shapedscars in our samples.
Eighty two disruptions were identified as crack disruptions. Half of them were
questionabledisruptionsand only 41 were identified as true scars(I9.27oof all scars).
Crack scars are linear disruptions(Fig. 3C). In many cases,the scars consist of a pair
of cracks that form a v-shapeddisruption which invańably openstoward thę anterior
edge of the shell (Fig. 3C). In many cases,the cracks consist of two complementary
scars of similar size located on the oppositevalves and situatedat a similar distance
from thę anterioredge of the shell.
Sixty nine disruptions could not be classified in any of the three morphological
categoriesand were,therefore,classified as 'miscellaneous'.Almost half of them were
questionabledisruptions,and only 38 wereclassifiedas truescars(I7 .8%o
of all scars).
Some of the miscellaneousscars also open toward the anteńor edge of the shell. ln
508
Predatory scars: KOWALEWSKI,
FLESSA,
& MARCOT
140
120
n -203
tTt- 2.5
S=2.3
I(s 1 0 0
o
u)80
o
b60
-o
E40
L
E
=t
z
5
zio
20
0
fi=82
rfi= 2.1
S= 1.1
a
b40
o
u,
b30
o
-o 20
1 1.5 2
2.5 3
1.5 2
3.5 4 4.5
Scar size (mm')
Ę20
(u
o
fi=46
tTt= 2.5
S= 1.5
O
o15
o
-o
E10
L
2.5 3
3.5 4
Scar size (mm')
I
830
fl=39
t T=
I 1.7
S=0.8
a
b25
bzo
-o
Ę15
z10
z
J
5
0.5 1
1.5 2
2.5 3 3.5 4 4.5
Scar size (mm')
0.5 1
1.5 2
2.5 3 3.5 4 4.5
Scar size (mm')
Fig. 4. A-D. Size-frequencydistribution of scars from Locality One. A. Pooled data for all scar types.
B. U-shaped scars. C. Pocket scars. D. Crack scars. Symbols: n - sample size, m - mean, s - standard
deviation.
many cases, the scars consist of two complementary scars of similar size located on the
opposite valves and situated at a similar distance from the anterior edge of the shell.
Quantitative analysis of scars. - The overwhel-ing majority of disruptions
(95.3Vo)were found in specimens collected from Locality One. The quantitative
analysis can, therefore,be confined to Locality One without substantiallyreducing
sample sizes (see Table 2). This allows us to eliminate any potential uncontrollable
variationin the datacausedby biotic and/orenvironmentaldifferencesbetweenthetwo
localities. The analysis was performedboth for all scar types pooled together,as well
as, separatelyfor each of the threemorphological types of scars (miscellaneousscars
were not analyzed separately).
The size of scarsvariesfrom 1.5mmz to 24 ffi and averages 2.5 Ń
(Fig. aA).
The variation in scar size is relatively small, with standarddeviation of 2.3 mmz and
coefficient of variation, cv = 9tVo. The size-frequencydistribution is highly rightskewed(skewness= 5.4) and very peaked(kurtosis= 41.8)with themajorityof scars
ACTA PALAEONTOLOGTCA POLONTCA 42 (4)
509
r= 0.32
P =0.0035
tl=82
(u-shapedscarsonly)
aoaa aa
Oa (f
0510152025
30 35 40 45 50
Brachiopodsize at thetimeof theattack[Sa](mm)
Fig. 5. Scatter plot of ,Sa(size of the brachiopod at the time of the attack) vs. scar size. Data for u-shaped
scars from Locality One (the other two scar types do not show any significant correlation).Symbols: r Spearmanrank correlationcoefficient,p - significance of the correlationcoefficient,n - sample size.
being small. The analysis by scar type reveals a very similar patternin all threecases.
All distributionsare strongly right-skewed,vary in a similar size range (Fig. 4B-D),
and are statisticallyindistinguishablefrom one anotherin their overall shape(pairwise
Kolmogorov-Smirnovtest,p > 0.05 in all threecases).Moreover,the u-shapedscars
and the pocket scars do not differ significantly from one anotherin their median size
(Wilcoxon test,p = 0.06).The only exceptionis the significantly smaller size of crack
scars in comparison with the two other scar types (Wilcoxon test,p <<0.05 in both
cases). Howeveą given that the scar size, as defined here, is an area measure,the
relatively smaller size of the crack scars should not be surprising.
As explained above,the distanceof the scar to the anteriorshell edge can be used
to calculate the brachiopod size at the time of the attack (So)(seeFig. 2B). Spearman
rank correlationanalysis indicatesthat only in the case of the u-shapedscars,does the
scar size (S) and the brachiopod size at the time of the attack (So)show a significant
positive correlation(p = 0.0035,Fig. 5). However, althoughsignificant,the correlation
is not very strong(Spearmanrank coefficient,r = 0.32,Fig. 5).
Out of 203 scarsfrom Locality One, I7 .87oof the scars occurredon both valves of
the brachiopod(i.e.,scars were representedby a pair of disruptions).Out of 167single
scars, 46.7Vowere located on ventral valves and 53.3Voon dorsal valves. This proportion does not differ significantly from 50:50, p - 0.28 (binomial test with normal
approximation,seeZar 1984).The dorso-ventraldistributionof scars is very uniform
among the three scar types: in all cases, the paired scars representless than 207o
(Fig. 6). The proportion of single scars on dorsal vs. ventral valves does not differ
significantly from 50:50 (in all three casesp >>0.05, binomial test).Also, the homo-
Predatoryscars: KOWALEWSKI, FLESSA, & MARCOT
510
G =5.05
P = 0.282
n=167
E60
o
g
o
tL ło
Proponion of scars
| located on dorsal valves
tilllllTl Prooortionof scars
]łi:ii:ii:l
|ocdtedon ventra|va|ves
l-l
|
A'
tuo3l",.
#:'
rYPe
3
!
I
Proportionof paired scars
located on both valves
Fig. 6. Dorso-ventral distribution of scars shown for the pooled data and separatelyfor each of the three
main scar types.Data for Locality One only. The numbersabove each bar representthe sample size of that
bar. Symbols: G - log-likelihood ratio (G-statistic),p - significance of G, df - number of degrees of
freedom,n - sample size.
geneity G-test (Fig. 6) fails to indicate any statistically significant variation in dorsoventral distributionof scars among differentscar types.
Analysis of the spatial distribution of scars in the plan view was limited to
November 93 and February 94 samples from Locality One. We excluded the March
1993samplebecause,due to the shells' growth,the apparentposition of scars oshifted',
on averageby2.4 mm (seeabove).Thus, we analyzedtotal of 182 scars.For all three
scar types, the spatial distribution of scars appears highly non-random with the
majority of scars located in the anteriorpart of the shell and concentratedalong the
shell's median axis (Fig. 7A-C). Note that even though there are potentially 17 horizontal sectors(seeFig. 2A), only sectors1 through If contńn scars.The only obvious
differenceamongthethreescartypesis in their distributionparallel to the anteriorshell
edge (i.e., along lateral sectors).U-shaped scars concentratealong the median shell
axis, whereaspocket scars and crack scars are more evenly distributed and include
many scars located along the lateral margins.
The apparentnon-randomnessin scardistribution(Fig. 7) canbe analyzedrigorously using a Poisson model. In this paper,we proposean approachwhich uses a Monte
Carlo simulationbasedon the Poisson probability function (for detailsseeAppendix 1;
see also Reyment l97t). The simulationsindicate that the spatial distributionof scars
is significantly differentfrom random for each of the threescar types,p < 0.0001 (the
sameresult is obtainedfor the pooled data).
Out of 203 scarsfrom Locality One, only 33Voarecompletelyhealed(Fig. 8).Therb
is a statistically significant variation in the proportion of healed scars among the
differentscar types (see Fig. 8). U-shaped scars show the highestrate (over 52Vo),
whereas only =25Eoof crack and=l3Vo of pocket scars are healed.Healed scars and
unhealed scźLrs
differ dramatically in their distribution away from the anterior shell
edge. The mean distance d is 5.9 mm (median = 3.3 mm) for unhealed scars, and
ACTA PALAEONTOLOGICA POLONICA 42 (4)
511
10
8
6
4
2
NUMBER
OF SHELLS
0
^ł'il
s6crob
1'x._
NUMBER 12
OF SHELLS 10
12
10
I
6
4
2
0
8
6
4
2
0
6
3
4
3
2
NUMBER 6
oF st-{ELLS 5
1
0
4
3
2
1
0
Fig. 7. Spatial distribution of scars on brachiopod shells in plan vięw. Data for the November 1993 and
February 1994 samplesfrom Locality One only. Note thatlateralsector '1' correspondsto the anteriorshell
edge (to relate the charts to the brachiopod shell see also Fig. 2A). Symbols: p - probability that spatial
distribution of scars is random (basedon Monte Carlo simulations,see text and Appendix 1), n - sample
size.A. U-shapedscars.B. Pocket scars.C. Crack scars.
13.8mm (median= 16.1 mm) for healed scars.This differenceis highly significant
statistically(p <<0.05,Wilcoxon test).The increasein the proportionof healedscars
away from the anterior shell edge is not surprising consideringthe fact that the more
posteriorscars are older, and thus, more time was available for their healing.
512
Predatoryscars: KOWALEWSKI, FLESSA, & MARCOT
G = 23.4
P = 0.001
n=167
E60
o
I
o
(L40
20
1-1 Proportion
|
| of unhealed
scars
163 Proportion
li:i:i::ii::l
of healed scars
0
All
Type 1
Type?
Scar type
Type 3
Fig. 8. The proportionofhealed to unhealędscars shown for the pooled data and separatelyfor each of the
threemain scar Ępes. Data for LocaliĘ one only. The numbersabove each bar representthe sample size
of thatbar.Symbols: G - log-likelihood ratio (G-statistic),p - significanceof G, d.f- numberof degreesof
freedom,n - sample size.
$H
go
$
E
=.-i.-
F0.036 F0.005
t
^
E
i
F0.065
c..
o)
o)
o
'=
e.0.402
=
25
(U
o
a
o20
o
L
€
=1s
z
0 1 2 3 4 5 6 7I91011
1 2 1 3 1 41 5 1 6 1 7 1 81 9 2 0 2 12 2 2 9 2 4 2 5 2 6 2 7 2 8 2 9 3 0 3 1
Distancefromthe anteriorshell edge (mm)
Fig. 9. The anterior-posteriordistribution of scars on the brachiopod shell (i.e., distance-frequency
distribution).Data for Locality One only. Four major modes may represenrseasonal(winter) predation.
Note the increasein spacingbetweenthe modes away from the anteriorshell edge.The two anteriormodes
are statisticallysignificantaccordingto 10,000-iterationbootstrap(seeAppendix 2). Symbol:p - bootstrap
estimateof local sisńficance of a mode.
ACTA PALAEONTOLOGTCA POLONTCA 4f (4)
513
Scartype:
miscellaneous
925
(U
o
a
o20
o
-o
crack scars
pocket scars
u-shaped scars
=1s
z
01
23
4
5 6 7I910
1 1 1 2 1 3 1 4 1 51 6 1 7 1 8 1 9 2 0 2 12 2 2 3 2 4 2 5 2 6 2 7 2 8 2 9 3 0 3 1
Distancefromtheanteriorshelledge(mm)
Fig. 10. The anteńor-posteriordistribution of scars on the brachiopod shell (i.e., distance-frequency
distribution).Data the sameas for Fig. 9, re-plottedhere to show the variation amongthe scar types in their
anterior-posteriordistńbution.
Scars vary in their distanceto the anteriorshell edgefrom 0 to 31 mm. This means,
given the average size of specimens(see Table 1), that scars are absentin the most
posteriorpart of the shell (6-9 mm; seealso Fig. 7). The distributionof scarsaway from
the anterior shell edge is strikingly multimodal (Fig. 9), with four well defined modes.
The first mode is locatedat the anteriorshell edge (0-1 mm distanceclass),the second
and largest mode is located 3 mm toward the posteńor from the first mode (3-4 mm
distanceclass).In addition,therearetwo smallermodestowardtheposteriorof the shell:
8-9 mm and 17-18 mm distanceclassesrespectively.Note thatthe spacingof the modes
increasesaway from the anteńor edge (Ftg. 9). We used a bootstrapprocedureto test
whetherthe modes are statistically significant (for details seeAppendix 2). The analysis
suggeststhat two anteriormodes are statisticallysignificant,whereasthe two smaller
posterior modes are not (see Fig. 9). The two posterior modes became significant,
however,whęnchartresolutionis decreasedto 2-mmbins (notshownhere).
The three main scar types differ notably in their anterior-posteriordistribution
(Fig. 10). Crack scars and pocket scars concentratemostly in the anteriorpart of the
shell (0-11 mm distanceclasses),whereasu-shapedscars dominatefartheraway from
theanterioredge(12-31 mm distanceclasses).Given thatthefrequencyof healedscars
increasesaway from the anterioredge,themoreposteriordistributionof u-shapedscars
is consistentwith their significantly higher frequencyof healing (Fig. 8). The u-shaped
scars differ statisticallyin their anterior-posteriordistributionfrom the two other scar
types (in both comparisonsp<<0.05,pairwiseKolmogorov-Smirnovtest).
Quantitative analysis of brachiopod shells. - Out of 820 brachiopodspecimens,
23.4%o
bear repair scars (seeTable Zfor quantitativesummary).Out of 192 specimens
514
Predatorvscars: KOWALEWSKI. FLESSA. & MARCOT
r= O.75
P =0.0003
8so
fi=18
I(u
o
g40
o
C)
5
=s o
o
E
o20
(s
&10
15
Shell length(mm)
Fig. 11. Scatter diagram showing frequency of scars in a given size-class of brachiopodsplotted against
brachiopod size-classes.Only the size-classeswith n > 5 arc considered.Data pooled for Localities One
and Two. Symbols: r- Spearmanrankcorrelationcoefficient,p - significanceof the correlationcoefficient,
n - sample size.
with unquestionablescars, 9O.2Vospecimensbear single scars, 8.8Vobear two scars,
and only 2 (IVo)bear three scars. Specimensbearing more than three scars were not
identified among our samplęs.
The frequencyof attackedspecimensincreaseswith brachiopodsize: whereasshells
with scarsare rare in the smaller size classes,they becameincreasinglyfrequentamong
largerspecimens.lndeed,thereis a very strongpositive correlationbetweenthefrequency
of attackedspecimensand their size (Fig. 11).Also, the frequencyof specimenswith
multiple Scars increases with the brachiopod size. The mean shell sŁe of specimens
bearing multiple scarsis significantly greaterthanthemean size of specimenswith single
scars(Z = 2.51,p =0 .Ul2,Wilcoxon test).The specimenswith multiplescarsareconfined
to larger brachiopodsize-classes.Smallest specimenwith multĘle scals has length of
35.6 mm, whereasspecimensas small as20 mm containsingle scars.
The frequencyof attackędindividuals of G. palmeri varies at thethreecomparative
levels that are available given our sampling design: (1) betweenlocalities; (2) within
Locality One amongbrachiopodpatches;and (3) within Locality One throughtime.
1. Variation between the two localities. The most dramatic variation can be observedbetweenthetwo sampledlocalities(Fig.tf; Table2). Specimenswith scarsare
over threetimes more frequentin Locality One thanin Locality Two. In both localities,
the frequencyof attackedspecimensincreasesin thelargersize-classes.However,only
in thęcaseof Locality one is this patternstatisticallysignificant(seeFig. 12).The lack
of significancefor datafrom Locality Two may be a consequenceof the smaller sample
size (only 9 specimenswith scars).
2. Yariation within Locality One. Because three patchesfrom Locality One were
sampledsimultaneouslyin November 1993,it is possible to analyzelocal (within-lo-
ACTA PALAEONTOLOGTCA POLONICA 4f (4)
515
tlt= 468,Dz=183
M , = 3 8 . 4 ,M z = 3 8 . 9
Z = 3 . 3 9 ,p ( Z ) = 0 . 0 0 0 7
D = 0.144,p(D)= 0.008
(/, 100
o
-c
U'
b80
o
-o
L
Ę60
z
tl.,=
1 6 0 ,f r z = 9
M , = 2 5 . 2M
, z=25.5
Z = 1 . 1 6p
, (Z)=0.247
D= 0.35,p(D)=0.245
U'
-c
a
o
L
I
E
=
z
'r5
10
15
Shelllength(mm)
Fig. If . Size-frequencydistributions of G. palnleń; the white parts of bars representspecimenswithout
scars and the gray pańs specimens with scars. Symbols: zl _ number of specimenswithout scans,n2 _
number of specimens with scars,Mt - median size of specimens without scars,Mz - median size of
specimenswith scars,Z - statisticfor the Wilcoxon Median testwith normal approximationand continuity
correction, p(Z) - significance of Z, D - Kolmogorov-Smirnov statistic, p(D) - significance of D.
A. Locality One. B. Locality Two.
cality) variation
in frequency
of attacked individuals
among brachiopod
patches. The
apparentvariation among the three sampled patches is not statistically significant
(Fig. 13).
516
FLESSA,
Predatory scars: KOWALEWSKI,
& MARCOT
G = 4.08
P =0.130
d f= 2
n =325
c
8oo
o
IL
40
r
Brachiopods
withoutscars
Brachiopods
ffi with
scars
0-
Patch 1
Patch 3
Patch 2
Fig. 13. The variation in relative frequency of specimens with scars among three brachiopod patches
sampledin November 1993in Locality One. The numbersabove eachbar representthe sample size of that
bar. Symbols: G - log-likelihood ratio (G-statistic),p - significance of G, df - number of degreesof
freedom,n - sample size.
154
G=29.8
P = 0.001
c
o
172
180
G=2.9
P = 0.09
G = 32.3
P = 0.001
df=2
n=506
60
I
o
rL 40
r
ffi
lł::::łjił::::i::|
lliiiiiiii$l
03.1993
11.1993
Brachiopods
withoutscars
Brachiopods
with scars
02.1994
Fig. 14. The variation in relative frequencyof specimenswith scarsthroughtime (Patch 1, Locality One).
The numbers above each bar representthe sample size of that bar. Symbols: G - log-likelihood ratio
(G-statistic),p - significance of G, df - numberof degreesof freedom,n - sample size.
3. Variation in time. Patch 1 from Locality One was sampledthreetimes:in March
1993,in November 1993,and in February 1994.There is a significant variation in the
frequencyof specimenswith scars across the seasons(Fig. 14). The frequencyof
specimens with scars is substantially lower in the March 1993 sample than in the
samples from the two subsequentseasons.There is no significant difference in the
frequency of attacked specimensbetween November 1993 and February 1994 (for
statisticaldetailsseeFig. 14).
ACTA PALAEONTOLOGICA POLOMCA 42 (4)
5r7
Interpretation
The origin of scars. - Despite severalweeks of observation,we have neverbeen
able to directly observethe agentresponsiblefor scars,and thus,our interpretationof
the origin of scars is somewhatspeculative.In addition,many abiotic and non-predatory biotic factorsexist that can causedamage,and consequently,repair scarsin shelly
organisms.These include self-inflicted damage of burrowers and predators,abiotic
damagedue to impactsof wave-bornestones,and humanactivity (for more details and
referencessee Cadóe et aI. in press).Nevertheless,it seemsquite certainthat scars (1)
arebiotic in origin; (2) record attemptsat predation;and (3) were madeby an epifaunal
organismwith a scissors-typeweapon(e.g.,claws,beak).
The small localized scars,thatcut acrossthe shell growthrings, clearly indicate an
external agent of destruction.The evidence pointing to the biotic origin of scars
includes the very limited size variation among the scars (Fig. a); their non-random
distributionconcentratednearthe anteriorshell edge(Fig. 7); the fact thatit is possible
to group scars into morphological types (Fig. 3; Table 2); and the presenceof complementary scars on opposite valves (Fig. 6). The non-randomdistribution of scars
(Fig. 7), the overwhelming dominance of single scars (Table 2), and the correlation
betweenshell size and scar size (in the case of u-shapedscars;Fig. 5) all suggestthe
predatoryońgin of scars.The fact that scars concentratenear the anterior shell edge
(Fig. 7) and, regardless of the scar type, open toward the anterior edge suggest
predatoryattacksfrom the anterior side of the brachiopod.Lingulid brachiopodsare
infaunal organismsthat live in vertical burrows and when active position themselves
with their anterior-sideup and with their anteriorshell edge alignednearthe sedimentwater interface (e.9., E-ig 1982). Thus, the scars must have been produced by ar,
epifaunal organism.The fact that many scars (again,regardlessof the morphological
type of scar) consist of a pair of disruptions (Fig. 5) - that are of similar size and
situatedon oppositevalves at a similar distancefrom the anteriorshell edge- suggest
predatoryattackswith a scissors-typeweapon such as the claws of a crab or the beak
of a bird.
The identity of predator(s?). - The most likely epifaunal intertidal predators
with a scissors-typeweaponinclude crabs and shorebirds(for review of shell-crushing
predatorsseeVermelj 1987,1992).Both crabs and shorebirdsare commonin the study
area. For several reasons, we believe that shorebirds are more likely predators.In
particular,two speciesof sandpipersshould be considered:the willet, Catoptrophorus
semipalmatus,and the long-billedcurlew Numeniusamericanus.
Shorebirds, especially willets and long-billed curlews, are common in the study
area(Wilbur 1987).They can, in fact,be extremelyabundantseasonally;coastalareas,
from California to Peru are their wintering grounds (Johnsgard1981;Hayman et al.
1986;Wilbur t987; Richards 1988).Moreover,shorebirdstendto form denseaggregations when foraging (e.g.,Recher 1966), and thus, can account for the very high
frequency of scars (Table 2). Finally, both long-billed curlews and willets are armed
with long bills which makethemparticularlywell adaptedfor preying on lingulids. Bill
length ranges from 113 to f19 mm in long-billed curlęws and from 50 to 67 mm in
willets (Haymanet al. 1986).Becausewillets were observedpreyingonGlottidia and
518
Predatoryscars: KOWALEWSKI, FLESSA, & MARCOT
damagingthe anterior shell edge (Paine 1963) and because Glottidia has previously
beenfound in the gut contentof willets (Paine 1963;Pereira1990),it seemsmore likely
to us thatthe willet ratherthanthe long-billed curlew is responsiblefor the scars.There
is, however,no direct emprical evidencewhich would allow us to decidethe predator's
identity with certainty. For the same reason, crabs cannot be entirely excluded as
possible predators.In contrast,shell-crushingfish such as rays - althoughpresentin
the area,as indicatedby frequentray pits - should be excluded as possible predators.
It seemsunlikely thattheir toothedjaws would generatealocalized andcomplementary
pair of scars with regular outlines,even if we assumethatthey are able to capturefully
infaunal organismssuch as Glottidia.
Tlvo important caveats. - TWo concerns need to be stressedbefore proceeding
with the interpretationof our data.First, as stressedabove,we are uncertainwhether
the scars were all createdby one predatory species.Thus, the presenceof the three
morphological types of scars may either suggestthat thereis more than one predator,
or thatthereis a single predatorwhich producesmore thanone type of scar.[n thelatter
case, the variation may be attributableto behavioral variation in predatoryattacksor
in the response of the prey (see below). The data do not provide an unequivocal
solution. The statistical similarities among the different scar types (i.e., similar size,
distributionaroundthe anterioredge,similar dorso-ventraldistribution),may indicate
that scars were made by one predator,but may also indic atea similar behavior (and
weapon size) of two or more predators.Similarly, the anterior-posteriordifferencesin
distribution of scar types (Fig. 9), can be used to argueeitherfor the presenceof two
differentpredatorsthatpreyed on brachiopodsat differenttimes, or changesin predatory behavior with prey size.
Second,repair scars record unsuccessfulpredationevents,and thus, are inherently
difficult to interpret(see Schoener 1979;Schindel et al. 1982;Vermeij 1983;Walker
& Voight 1994;CadÓeet al. in press).Most importantly'they cannotbe usedto estimate
predationintensity:a prey population with f07o of repair scars may in fact be preyed
upon at much higher ratesby an efficient predatoror at much lower ratesby a clumsy
predator.In fact, some predatorsare known to repeatedlyattackunsuitableprey (e.g.,
Vermeij l98f), and thus, it is feasible that a'prey' with frequentrepair scars is never
subjectedto predation.Also, if a predatoris (atleastoccasionally)successful,therepair
scarsrepresentonly a subsampleof all attacks.Unless the unsuccessfuland successful
affacks are the same in their ecological and behavioral aspects(e.g.,prey size, site of
attack),a quantitativeanalyses of repair scars may provide misleading insights into
predation.We will assumein the subsequentinterpretationthatthe repair scarsprovide
representativesample for all attacks,but some alternativeinterpretationswill also be
noted.
Behavior of the predator. - If all, or mostof thescars,weremadeby one species
of predator (e.g.,the willet), the variation in scar morphology reflects the variable
foraging behavior of thatpredator.Stenzel et ąI, (1916),for example,observedwillets
to display as much as five foraging behaviorswhen searchingfor prey: peck, multiple
peck, probe, multiple probe, and theft. Changes in searching method may introduce
variation in the strengthof the attack and the angle at which an attack is attempted.
This, in turn, may affectthe morphology of a scar producedin an unsuccessfulattack.
ACTA PALAEONTOLOGICA POLONICA 4f (4)
519
Alternatively, variation in the scar morphology may be a function of the prey escape
response.G. palmeri is capableof rapid withdrawal into the deeperparts of its burrow
with a surprisingstrength(personalobservations).According to our field observations
(see also Kowalewski & Demko 1996), the G. palmeri burrows can easily exceed
50 cm. Thus, the depthof burrows exceedsnotably the length of the crab's claws and
eventhelengthofthebillsof bothwillets(upto7 cm;afterHaymanetal.1986)and
curlews(upto 22cm; afterHaymanetaL t986).It seemslikely thatboththecapturing
methodof the predatoras well as the escaperesponseof G. palmeri is likely to change
with brachiopod size. Indeed,the differencesin the anterior-posteriordistribution of
different scar types (Fig. 10) suggestthat the variation in scar morphology may be
relatedto brachiopod size.
It is also possible that variation in scar morphology reflects variation in the
crystallographyand mineralogyof the lingulid shell.The orientationof apatitecrystals,
the orientationof organicmatrix, and proportionof organicto non-organicphasestend
to vary strongly across the shell. In Lingula, for example, the length of the c-axis in
apattte,the degreeof orientationof crystals, and proportion of apatiterelative to the
organic matter all increase toward the shell center (Iijima et al. 1988; Iijima &
Moriwaki 1990).Finally, as stressedearlier,variation in scar morphology may simply
reflect the fact that thereis more than one speciesof predatorsthat attemptto prey on
lingulids.
Site-selectivedurophagouspredationcan be evaluatedby assessingrandomnessin
the spatial distribution of predatorytraces on a shell. This line of evidence has been
repeatedlyused for predatorydrillholes, both Recent and fossil (e.9.,Reyment 1971;
Negus 1975;Kowalewski 1990;Anderson 1992).In case of our data,however,the
non-randomdistributionof scars (Fig. 7) most likely reflects the fact thatthe direction
of attackis strongly constrainedspatially: a shorebird(or crab) that attemptsto pull a
brachiopodout of its burrow is most likely to seize the anteriorshell edge.In addition,
the dorso-ventral distribution of scars (Fig. 6) does not suggest a valve-selective
predation,and the distributionof scarsalong the anteriorshell edge(Fig. 7) also shows
a lack of any site-selectivetendencies.Note, thatthe lack of site-selectivityis consistent with behavior of willets, which frequently prey upon infaunal organisms by
randomly probing burrows or by detectingprey movementundęr water (e.g.,Stenzel
et aI.1976).
The correlationbetweenprey size and scar size (Fig. 5) may be used as an argument
for size-selective predation.Note that although the correlation was found only for
u-shapedscars,this most likely reflects thatfact that other scar types were all madein
large specimens (Fig. 10); i.e., the independentvariable So varies so little that any
potentialcorrelationcannotbe detected.It is, however,debatablewhetherthe scar size
is well correlated with the predator size, and we have no rigorous data on the
size-variationin shorebirdandcrab populationsin the area.A seeminglymore convincing argumentfor size-selectivity is provided by the strong correlation betweenprey
size and the frequencyof attacks@ig. 9). Alternative explanationsexist here,however
(e.g., Vermeij & Dudley 1982). First, larger/olderindividuals have potentially had
more chance to encountera predatorthan the smallerĄounger ones. Second, a larger
individual may be more likely to survive the attack than a smaller one becausethe
damagedone by the predatoris smaller relative to the brachiopod size and, possibly,
520
Predatoryscars: KOWALEWSKI, FLESSA, & MARCOT
Ś20
g
(U
o
a
b15
o
c
o
f
Sro
tL
Distancefromanteriorshelledge(mm)
Fig. 15. The anterior-posteńordistributionof scarson the brachiopodshell. Data the sameas in Fig. 9, but
presented separately for each of the three sampling seasons (rhombs - March 1993, black circles Novembęr L993, x_ February 1994)'
becausethe larger brachiopodsmay be able to withdraw their shell into their burrows
with more strength.
Seasonality of predation. - The most interesting and least ambiguous line of
evidenceregardingpredatorbehavior is provided by the anterior-posteriordistribution
of scars (Fig. 9). The significantly multimodal distribution suggestssubstantialvariation in predatoryattacksthroughtime.We believe thatthe modesreflect seasonal,late
falVwinter predationattempts.
The youngestmode is locatedat theanterioredge(Fig. 9). The distancedata,plotted
separatelyfor each of the three sampling seasons(Fig. 15), indicate thatthis mode is
presentboth in the November 1993 as well as the February 1994 samples.All scarsin
the November 1993 anteriormode must have been made,thęrefore,in late Fall 1993,
a shorttime before the collection of the samples.The scarsin the anteriormode of the
February 1994 sample, may have also been made in the late Fall 1993 (G. palmeri
ceases or slows its growth during the winter months; Batten & Kowalewski 1995;
Anand et al. submitted),or during the 199311994Winter. The second anteriormode,
both in the November and February samples,is located 3 mm from the anteriorshell
edge.The distancebetweenthe modes coffespondsvery well to the annual growth of
the G. palmeri populationof 2.4 mm (Table2). Thus, the secondmode colrespondsto
ACTA PALAEONTOLOGICA POLONICA 42 (4)
52r
the late Fall 1992 and/orWinter 1993.This is confirmed by the March 1993 sample
which containsa significant mode at the anterioredge.When the March 1993 sample
is corrected for growth, the mode coincides with the second anterior mode in the
November 1993 and February 1994 samples(Fig. 15).
Unfortunately,we do not have data on the brachiopodpopulationsprior to March
1993, and thus, we do not know how much the brachiopod populations grew in
previous seasons.However, the growth ring analysis suggeststhat the two posterior
modes coincide with seasonalgrowth rings (Batten& Kowalewski 1995;Anand et aI.
submitted).Also, in all extantlingulids, the rate of growth slows down with increasing
size/age(seeChuang 1961;Paine L963:Worcester1969;Kenchington& Hammond
1978),and consequently,the annual growth incrementsdecreasetoward the anterior
shelledge.Thus, theincreasęin the distancebetweęnthemodesaway from the anterior
shell edge (Fig. 9) strongly suggests that the two posterior modes also represent
seasonal,falVwinter predation.The thfudmode, most likely, representsthe late Fall
1991 and/orWinter 1992 and the fourth mode the late Fall 1990 and/orWinter 1991.
The seasonalpatternof predationrecordedin the anterior-posteriordistributionof
scarsis consistentwith the hypothesisthatthe wintering shorebirdsare responsiblefor
scars.Willets and curlews aremigratorybirds thatwinter on theoceaniccoastsof North
and CentralAmerica (Johnsgard1981;Hyman et ąI.1986;Richards 1988).The coasts
of Baja California and California are their major wintering grounds (Johnsgard1981;
Haymanet aI. 1986;Wilbur 1987;Richards 1988).Those seasonalpopulationscan be
extremely abundant and often become the dominant predator in the intertidal communities (Wilbur 1987).Moreover, migratory shorebirdssuch as willets are known to
occasionallyremainin their winteringgroundsfor the entireyear (Haymenet al. 1986),
which would accountfor thepresenceof somescarsin-betweenthe modes(Figs 9, 15).
Impact on Glottidia populations. - Scars in live-collected specimens(or repair
scars in fossil specimens) representunsuccessful predation attempts,and thus, as
pointedout above,the frequencyof specimenswith scarsdoesnot necessarilycorrelate
with predationintensity.Nevertheless,it is worth noting that the frequency of specimens with scars is high: comparableto the frequenciesobservedin modern mollusks
(e.9.,Vermeij et aI. 1981)and higher than those observedin articulatebrachiopods
(e.g.,Thayer& Allmon 1991).The high frequencyof brachiopodswith scarsmay mean
thatshorebirds(or crabs)are very unsuccessfulpredatorsof lingulids and eithercannot
learn from, or can afford, their frequentfailures. This 'clumsy-but-stubborn-predator
modef is, however,inconsistentwith gut contentstudies (Paine 1963;Pereira 1990),
that show that shorebtrdsare successful predatorsof lingulids. Thus, the high frequency of brachiopods with scars indicates, we think, a high intensity of predation
(unless birds are successfulpredatorsand scars record an unsuccessfulpredationby
crabs). The high predation intensity is corroboratedby the fact that specimenswith
multĘle scarsare not frequentwhich suggests,in turn,a high rateof successfulattacks.
Note also,thateven if all predationeventswere unsuccessful,the wound causedby the
attackis likely to affectthe brachiopod'schancesof survival: unhealedscars are often
found several mm away from the anterior shell margin suggestingthat brachiopods
requiremonthsor even yearsto fully repair injuries causedby predators(alternatively,
theseunhealedscarsmay have been fresh scarsmade away from the anteriormargin).
52f
Predatoryscars: KOWALEWSKI, FLESSA, & MARCOT
At least four explanationscan be offeredfor the fact that specimenswith scars are
significantly more frequent in Locality One than in Locality Two. First, the older
populationsfrom Locality One (at least 3-4 years old; see Kowalewski 1996a;Anand
et al. submitted)have had more opportunityto encountera predatorthan the younger
populationsfrom Locality Two (1-2 yearsold). Thus, even if therewere no differences
in frequency of attacks between the two localities, populations from Locality One
would be expectedto include a higher proportionof specimenswith scars.Second,it
is
. possible that predatorsare more successfulwhen preying on smaller brachiopods.
Third, when hunting predatorsmay select for larger preys (but see above). Finally,
some subtle microhabitat differences betweenthe two localities - in particular, the
substantiallyfiner granulometryof the intertidalsedimentin Locality TWo (seeKowalewski I996a)- may influence theintensityand effectivenessof foragingby predator's
(especially shorebirds,see Quammen 1982).
The increasein thefrequencyof specimenswith scarsthroughtime,betweenMarch
1993 and November 1993,most likely reflects the cumulative increasein specimens
attacked by predators.In addition, the variation in predation intensity among the
subsequentwinters and the increasedrate of failed attackswith increasedage of the
brachiopod may also have contributedto the observedpattern.The available data are
insufficient to evaluatethe relative importanceof those factors.The lack of variation
in frequencyof specimenswith scarsamongthe threepatchesthatrangein microhabitat from the uppermostinteńidal to the lower middle intertidal (Kowalewski 1996a),
suggeststhatpredatorsprey equally intensely in the upper and middle intertidal.
Implications
Biological implications. - Despite the many ambiguities that hamper the interpretationof repair sca.rs,we believe that our study indicates the existence of strong
seasonalinteractionsbetweenshorebirds(and/orcrabs?)and inarticulatebrachiopods.
Moreover, given the very high populationdensitiesof G. palmeri in the area (Kowalewski 1996a),not only may the shorebirds(and/orcrabs?)be an importantpredator
of the brachiopod,but the brachiopodmay be an importantpart of the predatorsdiet,
especially during the winter months.
The predationon Recent articulatebrachiopodsis in generalmuch less intensethat
thaton mollusks (e.g.,Thayer& Allmon 1991;Thayer 1985;Jameset al. 1992).This
relatively lower predation intensity has been attributedto poor palatability and low
energeticvalue of the articulatebrachiopods(Thayer 1985;Baliński 1993).This study
suggests that in contrast to articulate brachiopods, lingulids are not repellent to
predators.Indeed,a simple field experimentsuggeststhatthey may be a very palatable
invertebrateprey: we fed live-collected specimens of G. palmeri to seagulls and
observedthe sameindividual seagullhappily returningfor more. Our study is consistęnt with gut contentdata suggestingthatmany groupsof predatorseatlingulids (Paine
1963;Cooper 1973;Pereira I99O;Mason & Clugston 1993;Emig in press;Campbell
& Campbell manuscript).Lingulids are also eatenby humans(e.9.,Stimpson 1860;
Yatsu I90f; Emig in press),and in some areasthey form their'...favorite article of
food...'(Stimpson1860:p. aafl.
ACTA PALAEONTOLOGICA POLONICA 4f G\
5f3
The data presentedhere suggestthat further biological research on brachiopodshorebird/crabinteractionsin the Colorado Delta may bring forth importantcontributions towardbetterunderstandingof the intertidalecosystemsof the innermostGulf of
California. This may also prove importantfor understandingthe marine ecology and
conservationbiology of nearshoremarineecosystemsin generalbecauselingulids can
be very abundantin intertidal habitats (Emig 1983; Kenchington & Hammon 1978;
Savazzi 1991;Kowalewski 1996a).
Finally, this study shows thatrepair scarsmay offer insights into the seasonalityof
predation,and thus, into the ecology of migratory animals and their impact on their
transientecosystems.
Paleontological implications. - Repair scars in lingulid shells may be preserved
in the fossil record. This study shows that predators (most likely shorebirds) are
capable of scarring the shells of a large proportion of the intertidal populations of
G. palmeri. Moreover, repair scarsin Glottidia arealso known from at least two other
localities:(1) the intertidalflats of the Gulf of Nicoya, Costa Rica (specimensof G.
ąudebąrticollectedand providedto us by J.A. Vargas1996),and(2) theFlorida coast
of the Gulf of Texas,USA (Paine 1963).Clearly, predatoryrepair scars are common in
intertidalpopulationsof lingulids.
Unquestionablelingulids (family Lingulidae) are known since the Late Devonian
(Williams 1977) whereas sandpipers(order Chadriifonnes, family Scolopacidae)a monophyletic (Jehl 1968),exclusively predatoryfamily, which includes willets and
long-billed curlews - are known from the fossil record since the Late Cretaceous
(Brodkorb1963,I967).ThegenusNumeniusappearsin thefossil recordby theMiddle
Miocene and thę genus Tringa (a close relative of Catoptrophorus)by the oligocene
(Brodkorb 1967). Clearly, predatory shorebirds and lingulid brachiopods have coexisted in intertidal ecosystems since at least the middle Tertiary, but may have
co-existedas early as the Late Cretaceous(molluscivorousbirds are known from the
fossil record since at least the Late Eocene; Cracraft L973; Vermeij 19'77).Ttre
shell-crushingability of modern predatory arthropods(e.g, stomatopods,spiny lobsters,crabs)may have evolvedin theLate Mesozoic or Paleogene(seeVermeij 19'77,
1987). Thus durophagousarthropods and lingulid brachiopods have co-existed in
benthic ecosystemssince at least the early Tertiary.
To our knowledge,however,repairScarshave neverbęenreportedin fossillingulids. This may be partly due to the fact that lingulids have very low fossilization
potential (Worcester1969; Emig 1990; Kowalewski 1996a).Indeed,post-Jurassic
lingulidshavevery poor fossil record(Kowalewski& Flessa 1996),andthus,arerarely
preservedwell enough,or in a sufficient number,to allow for the recognitionof repair
Scars.Also, and perhapsmore likely, fossil lingulids may have beenrarely examinęd
for repair scars by paleontologists.We hope that the criteria and quantitativedata
offered here will stimulatefuture paleoecologicalresearchon lingulids and will lead
to the identification of repair scars in fossil lingulids.
If repair scars in brachiopodsare predominatelymade by shorebirds,then specimens with scars may offer a useful paleoenvironmentalindicator as such brachiopods
must have lived in the intertidal zone. The confinment of scarred specimens to
intertidalpopulationsindeedseemsto be the casefor extantGlottidia.In the courseof
524
Predatoryscars: KOWALEWSKI, FLESSA, & MARCOT
a morphometricproject (Kowalewshi et aI. 1997),we obtained specimensfrom six
localities. Specimensfrom threeintertidallocalities (Localities One andTwo from Baja
California, and one locality from Costa Rica) bear repair scars, whereas those from
three subtidallocalities (musuemcollections form the subtidalof southernCalifornia,
North Carolina, and the Gulf of Mexico) lack scars.This suggeststhatrepair scarsthat
might be found in Tertiary fossil hngulids could be used as an evidence that those
brachiopods lived in the intertidal environment;especially if other lines of evidence
exist to suggestthat shorebirds were responsible for the scars. It should be noted,
however,that scars were found in subtidalpopulationsof Lingula (C.C. Emig written
communication 199"1).
Because scars in the brachiopod shell can record seasonalpredationby migratory
birds, the identificationof fossil populationsof brachiopodswith frequentrepair scars
may be particularly exciting. Such populations may help in identifying seasonal
predationin ancientmigratory birds. Because post-Jurassiclingulids have had a very
low fossilizationpotential (Kowalewski 1996a;Kowalewski & Flessa 1996),they are
less prone to time-averagingthan mollusks (Kowalewski 1996b, 1997), and thus,
Tertiary lingulid assemblagesmay offer temporalresolution sufficiently fine for the
analysis of seasonalpredation.
In summary,we hope that this paper will stimulate paleontological research on
predatory traces on lingulids - in particular, their implications for ecological and
co-evolutionaryinteractionsbetweenlingulids and their predators.
Acknowledgments
This study was supported by the NSF grant EAR-94|7054 to Karl Flessa. Michał Kowa]ewski
gratefully acknowledges support of the Alexander von Humboldt Foundation during the final stages
of this project and thanks Wolfgang Oschmann and Jimmy Nebelsick from Ttibingen Universitiit for
their hospitaliĘ. We thank Diana Hallman and Julie Anand for help at various stages of this project,
Christian Emig for many useful comments and valuable information, and Andrzej Baliński and
Gerhard Cadóe for carefuland helpful reviews of the final manuscńpt. This is publication number 26
of C.E.A.M. (Centro de Estudios de Almejas Muertas).
References
Alexander,R.R.l9S6a.Frequencyofsublethalshell-breakageinarticulatebrachiopodassemblagesthrough
geologic time. In: P.R. Racheboeuf& C.C. Emig (eds),Les BrachiopodesFossiles et Actuels. Biostratigraphiedu Paleozoique4, 159-167.
Alexander, R.R. 1986b.Resistanceto andrepair of shell breakageinducedby durophagsin Late Ordovician
brachiopods. - J ournal of P al eontology 60,273-f85 .
Alexander, R.R. 1990.Mechanical strengthof shells of selectedextantarticulatebrachiopods:implications
for Paleozoic morphologic trends.- Historical Biolo gy 3, 169-1 88.
Alexander, R.R. 1992.Survival and repair of the naturaland surgical damagein the articulatebrachiopod
Terebratulina retusa (Linnaeus).- H istoric al B ioIogy 6, 22I -23 L
Allmon, w.D., Nieh, J.C., & Nońs, R.D. 1990. Drilling and peeling of turritelline gastropodssince the
Late Cretaceous.- Palaeontology 33, 595-61 1.
Anand,J., Flessa,K.W., & Kowalewski,M. (submitted).
inthe shellof theRecent
Seasonalgrowthbanding
lingulid brachiopod Glottidia p almeri DalI.
ACTA PALAEONTOLOGTCA POLOMCA 42 (4)
5f5
Anderson,L.C. 1992.Naticid gastropodpredationon corbulid bivalves: effect of physical factors,morphological features,and statisticalartifacts.- Palaios 7,602-6f0.
Anderson,L.C.,Geary,D.H.,Nehm,R.H.,&Allmon,W.D.l99l.Acomparativestudyofnaticidgastropod
predation on Varicorbula caloosa and Chione cancellata, Plio-Pleistocene of Flońda, U.s.A. P alaeogeography, P alaeoclimatology, P alaeoecology 85, 283-f90.
Baliński, A. 1993.A recovery from sublethaldamageto the shell of a Devonian spiriferoid brachiopod.Acta PalaeontologicaPolonica 38, 111-118.
Batten,J. & Kowalewski, M. 1995.Seasonalgrowth banding and predationscarsin the shells of a Recent
lingulide brachiopod.- Geological Society of America Abstracts with Prograrns 27 (6),A445.
Batten,J., Chesley, K.B., Graybill, J.K., Marcot, J.D., Ramsey,M., Riley, B.C.D., Shinker,J.J., Smith,
O.A., & Speer, J.H. 1994. Geomorphology, sedimentarystructures,and biota of the Lower Colorado
delta, Baja California, Mexico. - University of Arizona Gesocience Symposium, Program and
Abstracts1994,6.
Bengtson, S. & Zhao, Y. 199f. Predatorial borings in Late Precambrian mineralized exoskeletons.Science257.367-369.
Berg, J.C. 1976.Ontogenyof predatorybehavior in snails (Prosobranchia:Naticidae).- Nautilus 90, 1-4.
Berg, J.C. & Nishenko, S. 1975.Stereotypyof predatoryboring behavior of Pleistocenenaticid gastropods.
- Paleobiolo gy L, f58-260.
Biernat, G. & Emig, C.C. 1993. Anatomical distinction of the Mesozoic lingulide brachiopods.- Acta
P alaeontologic a P olonica 38, t-20.
Bitner, M.A. 1996. Brachiopods from the Eocene La Meseta Formation of Seymour Island, Antarctic
Peninsula. In: A. Gaździckj (ed.), Palaeontological Results of the Polish Antarctic Expeditions. P alaeontologia P olonica 55, 65- 100.
Bray, N.A. & Robles,J.M. 1991.Physicaloceanography
of theGulf of California.łr: J.P.Dauphin& B'R.T.
Simoneit (eds), The Gulf and Peninsular Provinces of the Californias. - American Association of
P etroleumGeolo gistsBulletin 47, 5 11-553.
Brodkorb,P.1963.BirdsfromtheUpperCretaceousofWyoming.-ProceedingsoftheXllllnteruational
O rnithologic al C ongress 1,,55-70.
Brodkorb, P.1967. Catalogue of fossil birds: Part 3 (Ralliformes, Ichthyornithiformes,Charadriiformes).
- Bulletin of the Florida StateMuseum 11,,99-2f0.
Bromley, R.G. 1981. Concepts in ichnotaxonomy illustrated by small round holes in shells. - Acta
G eologic a Hisp anica 16, 55-64.
Bromley, R.G. 1996. Trace Fossils: Biology and Taphonomy.Znded. 361 pp. Unwin Hyman, London.
-Palaeontology 9,355-359.
Brunton,H.1966.PredationandshelldamageinaViseanbrachiopodfauna.
Brusca, R.C. 1980. CommonIntertidal Invertebratesof the GuIf of Califurnia.513 pp. University of
Arizona, Tucson, USA.
Cadóę, G.C. 1968. Molluscan biocoenosesand thanatocoenosesin the Ria de Arosa, Galicia' Spain. hologische VerhandelingenLeiden 95, I-LZI.
.
Cadóe, G.C. 1989. Size-selective transportof shells by birds and its palaeoecological implications. P al aeontology 32, 4f9 437 .
Cadóe, G.C., Walker, S.E., & Flessa, K.W. (in press).Gastropod shell repair in the intertidal of Bahia la
Choya. - P al aeogeog raphy,P al ąeoclimatology, P alaeoecology.
Campbell, L.D. & Campbell, C.C. (manuscript).Distribution and paleoecology of Glottidia inexpectans
Olsson (Brachiopoda:Lingulidae).
Chatterton,B.D.E. & Whitehead,H.L. 1987.Predatoryborings in inarticulatebrachiopodArtiotretafrom
the Silurian of oklahoma.- LethaiazU,67-ą4.
Chuang, S.H. 1961. Growth of the postlarval shell in Lingula unguis (L.) (Brachiopoda).- Proceedings
of the Zo ologic al Soc i ety of London 137, 299-3 lO.
Cooper, G.A. 1973.Brachiopods (Recent).- Memoirs of the Hourglass Cruises 3, part 3, l-14. Marine
ResearchLaboratory,Florida Departmentof Natural Resources.
Cooper, G.A. 1988. Some Tertiary brachiopods of the East Coast of the United States.- Smithsonian
C ontributionsto P aleobiology 64, I-54.
Cracraft, J. 1973. Continental drift, palaeoclimatology,and the evolution and biogeographyof birds. Journal of Zoolo gy, London 169, 455-545.
Dall, W.H. 1870.A revision of the Terebratulidaeand Lingulidae, with remarks and descriptionsof some
recentforms' - American Journal of Conchology ó, 88-168.
5f6
Predatory scars: KOWALEWSKI,
FLESSA,
& MARCOT
Dall, W.H. 1871. Supplement to the 'revison of the Terebratulidae', with additions, corrections and a
revision of the Craniidae and Discinidae. - Ameńcan Journal of Conchology 7,39_79.
Dall, W.H. 1920.Annotated list of the Recent brachiopodsin the collection of the United StatesNational
Museum, with description of thirty-threenew forms. - Proceedings of the United StatesNational
Museum57,f6I-377.
Diaconis, P. & Efron, B. 1983. Computer-intensivemethodsin statistics.- ScientificAmerican 248,
116-130.
Efron, B. 1979.Bootstrapmethods:anotherlook at thejackknife. - TheAnnals of Statistics7,I-26.
Ekdale, A.A. 1985. Paleoecology of the marine endobenthos.- Palaeogeography,Palaeoclimatology,
Palaeoecology 50,63-8 1.
Emig, C.C' I98f. Terrier et position des Lingules (Brachiopodes,Inarticulós). - Bulletin de la Socićtć
Zoologiquede France 107, 185-194.
Emig, c.C. 1983. Taxonomie du genre Glottidia (Brachiopodes,Inarticulós).- Bulletin de Museum
nationaled'Histoire naturalle,Paris 4 (5),sectionA(f),469489
Emig C.C. (in press).Ecology and biogeographyof the inarticulatedbrachiopods.In; R.L. Kaesler (ed.),
Treatise on InvertebratePaleontology, Part H, Brachiopoda, 1. Geological Society of America and
University of Kansas Press, Boulder Colorado, and Lawrence Kansas, USA.
Emig, C.C. & Vargas,J.A. 1990.Glottidiaaudebarti(Broderip),(Brachiopoda,Lingulidae)from the Gulf
of Nicoya, Costa Rica. - Revistade BiologiaTropical3S,25I-258.
Emig, C.C., Gall, J.C.,Pajaud,D,, &Plaziat,I,C.1978'Róflexions critiquessur1'ócologieetlasystómatique
des lingulesactuelleset fossiles.- Gćobios|t,573_609.
Ezcurra, E. & Rodrigues, V. 1986.Rainfall patternsin the Grand Desierto, Sonora,Mexico. - Journal of
Arid Environments10, 13-28.
Fairweather,P.G. 1985. Differential predation on alternativepreys, and the survival of rocky intertidal
organismsinNewSouthWales.-Journalof ExperimentalMarineBiologyandEcology
89,135-156.
Flessa,K.w., Kowalewski,M., & Walker, s.E. 1992.Post-collectiontaphonomy:shęlldestructionandthe
Chevrolet. - Palaios 7, 553-554.
Ftirsich, F.T. & Jablonski, D. 1984. Late Triassic naticid drillholes: carnivorous gastropodsgain a major
adaptationbut fail to radiate.- Science 224,78-80.
Garrity, S.D. & Levings, S.C. 1981.A predator-preyinteractionbetweentwo physically and biologically
constrained tropical rocky shore gastropods:direct, indirect and community effect. - Ecological
Mono graphs 5I, f67 -f86.
Geary,D.H., Allmon, W.D., & Reaka-Kudla,M.J. 1991.Stomatopodpredationon fossil gastropodsfrom
the Plio-Pleistoceneof Florida.- Journal of Paleontology65,355-360.
Hatai, K. 1938. on some Cenozoic Brachiopoda from the North American region. - Ameńcan Midland
N aturali st L9, 7 06-7 2f .
Hayman, P., Merchant, J., & Prater,T. 1986.Shorebirds:An IdentfficationGuide to Waders of the World.
4Lf pp. Crrom Helm, London.
Hertlein, L.G. & Grant IV, U.S. 1944.The Cenozoic Brachiopoda of westernNorth America. - University
of Califurnia Los Angeles Publications in Mathematical and Physical Sciences 3,I-f36.
Hoffman, A., Pisera, A., & Ryszkiewicz, M. 1974. Predation by muricid and naticid gastropodson the
Lower Tortonian mollusks from the Korytnica Clays. - Acta Geologica Polonica 24,f49-f60.
Hoffman, A. & Martinell, J. 1984. Prey selection by naticid gastropods in the Pliocene of Emporda
(NortheastSpain).- Neues Jahrbuchfiir Geologie und Paliiontologie, Monatshaft L984,393-399 .
Iijima, M. & Moriwaki, Y. 1990. Orientation of apatite and organic matrix in Lingula unguis shell. Calcifted Tissue International47, f31-f4f .
Iijima,M., Moriwaki, Y., Doi, Y., & Kuboki, Y. 1988.The orientationof apatitecrystalsinLingulaunguis
shell. - JapaneseJournal of Oral Biology 30,f0-30.
James,M.A., Ansell,A.D., Collins,M.J., Curry,G.B., Peck,L.S., & Rhodes,M.C.I99f . Biology of living
brachiopods.- Advances in Marine Biology 28,175-387 .
Jehl, J.R., Jr. 1968.Relationshipsin the Charadńi (Shorebirds):a taxonomic study basedon color patterns
of the downy young.- San Diego SocieĘ of Natural History Memoir 3, I_54'
Johnsgard,P.A. 1981.ThePlovers, Sandpipers,and Snipesof theWorld.49f pp.The University of Nebraska
Press,Lincoln, Nebraska.
Kabat, A.R. 1990. Predatory ecology of naticid gastropodswith a review of shell boring predation.Malac ologia 32, 155-193.
ACTA PALAEONTOLOGICA POLONICA 4f G)
521
Kelley, P.H. 1989. Evolutionary trends within bivalve prey of ChesapeakeGroup naticid gastropods.Histońcal Biolo gy 2, 139_156'
Kelley, P.H. 1991. The effects of predation intensity on rate of evolution of five Miocene bivalves. Historical Biology 5, 65-78.
Kelley, P.H. & Hansen, T.A. 1996. Naticid gastropodprey selectivity through time and the hypothesisof
escalation.- P alaios 11.540-556
Kenchington, R.A. & Hammond, L.S. 1978. Population structure,growth and distribution of Lingula
anatina (Brachiopoda)in Queensland,Australia. - Journal of Zoology, London 184, 63-81.
Kitchell, J.A. 1986.The evolution of predator-preybehavior:naticid gastropodsand their molluscanpreys.
In: M.H. Nitecki & J.A. Kitchell (eds), Evolution of Animal Behavior: Paleontological and Field
Approaches, 88-1 10. Oxford, New York.
Kitchell, J.A., Boggs, C.H., Kitchell, J.F., & Rice, J.A. 1981. Prey selectionby naticid gastropods:
experimentaltestsand application to the fossil record.- Paleobiology 7,533-552.
Kowalewski, M. 1990.A hermeneuticanalysis of the shell-drilling gastropodpredationon mollusks in the
Korytnica Clays (Middle Miocene; Holy Cross Mountains, Central Poland). - Acta Geologica
Polonica 40. I83-fI3.
Kowalewski, M. 1993.Morphometric analysis of predatorydrillholes. - Palaeogeography,Palaeoclimatolo gy,P alaeoecology L02, 69-88.
Kowalewski, M. 1996a.Taphonomyof a living fossil: thelingulide brachiopodGlottidia palmeriDall from
Baja California,Mexico. - Palaios 1I,f44-265.
Kowalęwski, M. 1996b. Time-averaging, overcompleteness,and the geological record. - Journal of
Geology 104,3I7-3f6.
Kowalewski,M. 1997. The reciprocal taphonomicmodel. - Lethaia 30, 86-88.
Kowalewski, M. & Demko, T.M. 1996.Trace-fossils and population paleoecology:comparativeanalysis
of size-frequencydistributionsderivedfrom burrows.- Lethaia29,II3-124.
Kowalewski, M. & Flessa, K.W. 1994. A predatory drillhole in Glottidia palmeri Dall (Brachiopoda;
Lingulidae) fromRecent tidal flats of northeasternBajaCalifornia,Mexico. -Journal of Paleontology
86.1403-1405.
Kowalewski, M. & Flessa, K.W. 1995. Comparative taphonomy and faunal composition of the shelly
cheniersfrom the northeasternBaja California, Mexico' - Ciencias Marinas 21',ś5_177.
Kowalewski, M. & Flessa, K.w. 1996.Improving with age:the fossilręcord of lingulide brachiopodsand
the natureof the taphonomicmegabiases.- Geology24,977-980.
Kowalewski,M., Flessa,K.W., & Aggen,J.A.1994. Taphofaciesanalysisof Recentshellycheniers(beach
ridges),northeasternBaja California, Mexico. - Facies 3L,209-f4f .
Kowalewski, M., Dyreson,E., Marcot, D., Vargas, J.A., Flessa, K.W., & Hallman, DP. 1997.Phenetic
discriminationof biometric simpletons:paleobiologicalimplications of morphospeciesinthe lingulide
brachiopod Glottidia. - P aleobiology 23, 444469.
Kropp, R.K. 1992.Repaired shell damageamong soft-bottommollusks on the continentalshelf and upper
slopenorthof Point Conception,California.-Veliger 35, 36-51.
Lowe, H.N. 1933.At the head of the Gulf of California. - Nautilus 47, 4547.
Mason, W.T., Jr. & Clugston, J.P. 1993. Food of the Gulf Sturegoenin'the SuwanneeRiver, Florida. Transactionsof theAmerican Fisheries SocieĘ 122,378_385.
Miller, R.H. & Sundberg,F.A. 1984.Boring Late Cambrianorganisms.- Lethaia17,185-190.
Neugus, M. 1975. An analysis of boreholes drilled by Natica catena (da Costa) in the valves of Donax
vittatus(da Costa).- Proceedings of theMalacological Society of London 4L,353-356.
-Science241',
Norton,s.F. 1988.Roląofthegastropodshelloperculumininhibitingpredationbyfishes.
9f-94.
Paine, R.T. 1963. Ecology of the brachiopod Glottidin pyramidata. - Ecological Monographs 33,
r87-f13.
Pereira,A.I. 1990.Ecologfa de la alimentación de Limnodromusgriseus, Catoptrophorussemipalmatusy
Numeniusphaeopus(Aves, Scolopacidae)en Chomes,Costa Rica. M.Sc. Thesis,Universidad de Costa
Rica.
Quammen, M.L. 1982. Influence of subtle substratedifferences on feeding by shorebirds on intertidal
mudflats.- Marine Biology 71,339-343.
Recher, H.F. 1966. Some aspectsof the ecology of migrant shorebird.- Ecology 47,393407.
Reyment,R.A. 1971.IntroductiontoQuantitativePaleoecology.ffT pp. Elsevier,Amsterdam.
528
Predatorv scars: KOWALEWSKI.
FLESSA.
& MARCOT
Richards,A. 1988.BirdsoftheTideline:ShorebirdsoftheNorthernHemisphere.2f4pp.DragonsWorld,
Surrey,England.
Robba, E. & Ostinelli, F. 1975.Studi paleoecologici sul Pliocene Ligure. I. Testimoniaze dipredazione sui
molluschi Pliocenici de Albenga. - Rivista Italiana di Paleontologia e StratigrafiaBL,309-372.
Roden, G.I. 1964.Oceanographicaspectsof the Gulf of California . In: T .H. van Andel & G.G. Shor (eds),
MarineGeologyof theGulf of California.-AmericanAssociationof PetroleumGeologistsMemoir
3,30-58.
Rowell, A.J. 1965. Inarticulate.In: R.C. Moore (ed.), Treatise on InvertebratePaleontology, Part H,
Brachiopoda 1,}l260-Hf96. Geological Society of America and University of Kansas Press,Boulder,
Colorado and Lawrence, Kansas, USA.
Ruggiero, E.T. 1991.A study of damageevidence in brachiopod shells. In: D.I. MacKinnon, D.E. Lee, &
J.D. Campbell(eds),BrachiopodsThroughTirne,203-210.Balkęma,Rotterdam'
SAS Institute.1989. SASISTAT User's Guide (version 6), 4th ed. 943 pp. (vol. 1), 846 pp. (vol. 2), SAS
Institute,Cary, USA.
SAS Institute.1990a. SAS Procedures Guide (version 6),3d ed. 705 pp. Sas Institute,Cary, USA.
SAS Institute.1990b.SAS/IML.235 pp. Sas Institute,Cary, USA.
SAS Institute1991. SASl-anguageGuide.755 pp. Sas Institute,Cary, USA.
Savazzi,E,199L. Burrowing in the inarticulatebrachiopodLingulaanątina, - Palaeogeography,Palaeoclimatology,P alaeoecology 85, 10 1-106.
Schindel, D.E., Vermeij, G.E., & Zipser, E. 198f. Frequencies of repaired shell fractures among the
Pennsylvaniangastropodsof North-Central Texas. - Journal of Paleontology 56,729-:140.
Schoener, T.W. 1979. Infening properties of predation and other injury-producing agents from injury
frequencies. - Ecology 60, 1110-1 115.
Sheehan,P.M. & Lesperance,P. 1978. Effects of predation on the population dynamics of a Devonian
brachiopod. - Joumal of Paleontology 52,812-817.
Smith, O.A., Hallman, D.P., & Batten, 1.1995. Comparative taphonomy of Recent lingulide brachiopod
and bivalve shells:an experimentalstudy.- Geological Society of America Abstracts with Programs
27 (6),A-135.
Stenzel, L.E., Huber, H.R., &Page, G.W. 1976. Feeding behavior and diet of the long-billed curlew and
willet. - The Willson Bulletin 88, 3 15-332.
Stimpson,W' 1860.A trĘ to Beaufort,N. Carolina. - American Journal of Science 29,2nd series,44f445 .
Sykes, G.G. 1937.The Colorado Delta. - Carnegie InstitutionWashingtonPublication 460,1-193.
Taylor, J.D., Morris, N.J., & Taylor, C.N. 1980. Food specialszationand the evolution of predatory
prosobranchgastropods.- P al aeontology 23, 375 409 .
Thayer, C.W. 1985.Brachiopods versusmussels:competition,predation,andpalatability. - Science228,
r5f7-15f8.
Thayer, C.W. & Allmon, R.A. 1991. Unpalatable thecideid brachiopods from Palau: Ecological and
evolutionaryimplications.In:D.I.MacKinnon,
D.E. Lee, &J.D. Campbell(eds\,BrachiopodsThrough
Time, 26I-264. Balkema, Rotterdam.
Thayer, C.W. & Steele-Petrovic,H.M. 1975.Burrowing of the lingulid brachiopod Glottidia pyramidata:
its ecologicand paleoecologicsignificance.- Lethaia8,f09-221.
Thomas, R.D.K. 1976.Gastropodpredationon sympatricNeogene speciesof Glycymeris (Bivalvia) from
the easternUnited States.- Journal of P aleontology 50, 488499.
Thompson, R.W. 1968. Tidal flat sedimentationon the Colorado River delta, nońhwestern Gulf of
California. - Geological Society of America Memoir L07, 1-133.
Thomson, D.A. (ed.) 1993. 1993 Tide calendar for the northern Gulf of Caltfurnia. [No page numbers].
Printing and Publishing Support Services, University of Arizona, Tucson, USA.
Treuman,E.R. &Wong, T.M. 1987.Theroleofthecoelemasahydrostaticskeletoninlingulidbrachiopods.
- Journal of Zoology, London 2L3, 22I-23f .
Vale, F.K. & Rex, M.A. 1988.Repaired shell damagein deepseaprosobranchgastropodsfrom the western
North Atlantic. - Malacologia 28, 65-79.
Vale, F.K. & Rex, M.A. 1989. Repaired shell damagein a complex of ńssoid gastropodsfrom the upper
continentalshelf southof New England. - Nautilus 103, 105-108.
Vermeij, GJ.1977. The Mesozoic marine revolution: evidence from snails, predators and grazers. P aleobiology 3, f45258.
Vermeij, G.J. 1982.Gastropodshell form, breakage,and repairin relationto predationby the crab Calappa.
- Malacologia23,l-12.
ACTA PALAEONTOLOGICA POLONICA 42 G\
529
Vermeij, G.J. 1983. Traces and trends in predation, with special reference to bivalved animals. P alaeontology 26, 455465 .
Vermeij, G.J. 1987.Evolution and Escalation.52T pp. Princeton University, Princeton.
Vermeij, GJ. 1992.A Natural History of Shell.fOT pp. Princeton University, Princeton.
Vermeij, G.J. & Dudley, E.C. 1982. Shell repair and drilling in some gastropodsfrom fupley Formation
(UpperCretaceous)of the South-eastern
U.S.A. - Q7ę76ęęgus
Research3,397403,
Vermeij, G.J., Zipser, E., & Dudley, E.C. 1980.Predationin time and space:peeling and drilling in terebrid
gastropods.- P aleobiology 6, 352-364.
Vermeij, G.J., Schindel, D.E., & Zipser, E. 1981. Predation through geological time: evidence from
gastropodshell repair.- Science 214, I0f+-I026.
Vermeij, G.J.,Zipser,E., &Zardini, R. 1982.Breakage-inducedshell repairin some gastropodsfrom the
upperTriassic ofltaly. - Journal ofPaleontology 56,233-f35.
Walker, S.E. & Yamada, S.B. 1993.Implicationsfor the gastropodfossil record of mistakencrab predation
on emptymollusc shells.- Palaeontology36,735-74L.
Walker,S.E. &Voigt,J.R. 1994.Paleoecologicandtaphonomicpotentialof
deepseagastropods.-Palaios
9,49-59.
Wiedman,L.A., Feldmann,R.L., Lee, D.E., & Zinsmeister,W.J. 1988.Brachiopodafrom the La Meseta
Formation (Eocene),Seymourlsland,Antarctica.ln: R.M. Feldmann& M.O. Woodburn (eds),Geology
and Paleontology of the Seymour Island, Antarctic Penninsula. - Geological Society of America
Memoir 169,531-539.
Wilbur, S.R. 1987.Birds of Baja Califurnia. 253 pp.University of California,Berkeley,USA.
-Journal
Wiltse,W.L l9S0.PredationbyjuvenilePolinicesduplicatus(Say)onGemmagemma(Totton).
of ExperimentalMarine Biology and Ecology 42, t87-199.
Worcester,W. 1969.On Lingula reevei.Unpublished Master Science Thesis. 49 pp.University of Hawaii,
Hawaii.
Yatsu, N. 1902.On the habitsof the JapaneseLingula. - AnnotationeszoologicaeJaponensis4,61-67 .
Zar,J.H. L984.BiostatisticalAnalysis.718 pp. Prentice,Englwood Cliffs.
Appendix
1
The Monte-Carlo model based on the Poisson distribution. The SAS program (for SAS release 6.11)
that performs the procedure described here is available upon request from M. Kowalewski.
The approach proposed here is based on the method described by Reyment (I91L) to analyze
the spatial distribution of predatory marks. In contrast to the Reyment's approach, we employ here
a non-parametric, computer-intensive technique that would not have been feasible 25 years ago,
when the Reyment's approach was published. The test evaluates whether the spatial distribution of
scars is different from random. In this approach, the scar-frequency distribution calculated for the
acfual data was compared to the distńbution expected for random data predicted by the Poisson
distribution with mean 1,.
?u= N/s
where fu- poisson distribution mean; N- number of scars; and s - number of sectors.
(1)
A given scar-frequency distribution includes s observations where N equals:
s
l,r=Dt
(f)
i=I
where y is a number of scars in an i-th sector.Consider a3x3 grid of 9 sectorsthat contains 7 scars.
Suppose, for example, that there is 1 sector with 3 scars, 4 sectors with 1 scar, and 4 sectors with
0 scars. The scar-frequencydistribution is thus based on 9 observations(3,1,1,1,1,0,0,0,0)
that sum
up to 7.
The null hypothesis that the data have a random spatial distribution can be evaluated by
comparing the sector-frequency distribution to the theoretical distribution predicted by Poisson
530
Predntory scars: KOWALEWSKI,
FLESSA,
& MARCOT
function with l" = N/s. In the above example of 9 sectors and 7 scars, the expected distribution
should be calculated for ?u= 0.7778. The significant difference between the actual distribution and
the expected distribution Ęects the null hypothesis that data came from the population with a random
spatial distibution. In Reyment's approach the two disfributions are compared directly using the X2
statistic. However, such an approach requires an arbitrary a priori lumping of the categories that
contain small percentage of observations (Reyment I97I) and also requires assumptions of )P test
which can be avoided when using a computer-intensivemethod (e.g.,Diaconis & Efron 1983).
Our approach consists of the following steps:
Step 1. 10 000 random datasetswith sample size s, are drawn from the Poisson distribution with
mean 1,,where fu is calculated from the actual data. In the simulation, we used the Poisson function
provided by SAS (see SAS 1991).
Step 2. Each of the random datasetsis compared to the expected Poisson disfribution with mean
1,. The difference between the two distribulions is estimated by G (a log likelihood ratio), the
parameter which is often considered superior to the -X2statisticr).
G=2ź,,,,'E&)
(3)
where G - likelihood ratio, f(i)o -nvntber of observation in category i, f(i)" - number of expected
observation in category i as expected from Poisson distribution.
Note, that 10 000 G-values deńved in the simulation provide an estimate of the probability
distribution for random datasets with sample size s drawn from Poisson function with a mean 1".
Step 3. A single G-value for the actual data is calculated using the sź}meexpected Poisson
distribution as used in Step f. The probability associated with the obtained G-value (p) is then
estimated using empirical dishibution of G-values obtained in Step 2. We used the percentile method
(also called nave bootstrap approach, see Efron 1979):
(4)
P=x/i
where p - is probability that data came from a Poisson-distributed population, x - z number of
randomly derived G-values greater than or equal to the G-value calculated for the actual data, and
i - number of iterations (10 000 in our case). Note that i defines the number of decimal places that
can be reportedforp; when p = 0, thenp <Ili (i.e.,0.0001in our case).
1) Note, thatthe choice of the goodness-of-fitstatisticis not cńtical in our app1oachbecausethe probability
function is built empirically through a Monte-Carlo simulation.Thus, X, or Kolmogorov-Smirnov D
could be used equally successfullyin our approach.
Appendix 2
Bootstrap procedure for local significance of modęs (employed here to analyze data presented on
Fig. 10). The program written in SAS/IMLis available upon request from M. Kowalewski.
In order to assessthe statistical significance of multiple modes in a single distńbution, we propose
here a simple bootstrap procedure. The procedure estimates the probabiliĘ of whether a given mode
is locally significant - i.e., significant in respect to the directly adjacent bins (frequency classes) or locally insignificant (i.e., could be produced by sampling of a population which lacks that mode).
The procedure is based on the resamptng of the actual data according to the following protocol.
Step 1. The modes to be evaluated are identifiedt in the original data (in ow case, 4 modes
indicated on Fig. 10).
Step 2. A sample (in our case 203 distance measurements presented on Fig. 10) is resampled
with replacement. The resulting bootstrap sample is used to calculate a new frequency distribution.
ACTA PALAEONTOLOGICA POLOMC A 4f (4)
531
Each mode is then evaluated according to the simple cńterion. If the number of observation in the
mode is greater than in any of the two adjacent frequency-classes, thę bootstrap sample is assigned
'1'. ff any of the two adjacent classes has equal or greater number of observation than the
value of
'0'.
mode, the sample is assigned value of
Step 3. Step 2 is repeated n number of times. In our analysis łz= 10,000. The null hypothesis
for a given mode is rejected2)if p < 0.05. Wherep (probabilĘ that sample came from a population
'mode' has fewer or equal number of observations than at least one of the two directly
where the
adjacent bins) is given by the following formula:
p = 1 .- L
where x - number of bootstrap samples ,l*.n
n - number of iterations.
(1)
were assigned
'f
in the procedural Step 2, and
1) Note that position and height of modes dependon how we define the bins. In case of our data,bins are
1-mmintervals(Fig. 10).
2) Although the analysis evaluates four hypothesis, the tests are dependent(i.e., based on the same
bootstrapsamples)and Bonferroni correctionshould not be applied.
współczesnego
w muszlach
Blizny drapieżnicze
ram ie nio no ga lin gulidowe go Glo ttidia p almeri
giczne i ekolo giczne
paleontolo
i ich znaczenie
MICHAŁ KowALEwsKI. KARL w. FLEssA i JoNATHAN D. MARCoT
Streszczelnie
Szczegótowa ana|izailościowabLtzndrapiezniczych w muszli ramienionogabezza.
wiasowego Z grupy lingulidów opartajest na 8f0 okazach Glottidia palmeri Dall,
1870. Okazy Zebrano z dwóch stanowisk z równi międzypływowych pótnocnowschodniĄ Zatokl Kalifornijskiej w Meksyku. Blizny drapieznicze występują w
kiesze23.4%o
osobników.Mozna je podzieLtćnaczterykategorie:blizny u-kształtne,
Niezaleznieod ich morfologii,blizny koncentĄą
niowe, spękaniowęorazpozostałe.
jest rozwartąStroną
się przy przedniejkrawędzimuszli. Większośćb|iznzorientowana
w kierunku przedniej krawędzi muszli i wiele z nich składa się z dwóch b|izn
zlokalizowanych naprzeciwlegle nabr^Bznej i grzbietowejskorupceramienionoga.
(1) blizny wahająsię w wielkościod 1,5 do 24 mmf
Analizailościowawskazuje,źLe
_f,5 mmz;iwszystkiecztery kategorieblizn*ająpodobnerozkładywielko(średnia
ści;(f) rozI7l7eszczeniebliznna powierzchni muszli nie jest losowe, podczasgdy ich
rozkład grzbieto-brzaszny wydaje się być losowy; (3) proporcja zagojonych b|izn
wzrastaw kierunku tylnej częścimuszli; (4) rozkładblizn jest udetza1ącomultimodalny i sugerujesezonalnedrapieznictwoskoncenffowanepóźnąjesieniąizimą; (5) częokazów zbliznamt
stośćbltznwzrasta wraz z wielkościąramienionoga;i (6) częstość
waha się znaczącopomiędzy dwoma badanymi stanowiskami,jak również'w obrębie
stanowiska 1 w czasie, ale nie waha się znacz4co pomiędzy różnymi miejscami
opróbowania w obrębiestanowiska1.
53f
Predatory scars: KOWALEWSKI.
FLESSA.
& MARCOT
Blizny repręZentująnieudane ataki jakiegośepifaunalnego drapieznika, wypoSazonegow nozycowy narząd chwytny (np. szczypce kraba, dzi6b ptaka). Wysoka
częstość
ataków, ich sezonalność
i poprzedniebadaniaekologiczne zgodnie sugeruj4,
żeb|iznys4 wynikiem ataków drapieznychptaków (Catoptrophorussemipalmatus|ub
Numeniusamericanrzs).
WybrzeiaPó|wyspu Kalifornijskiegosąmiejscemzimowania
drapieznych ptaków i goszczą ich liczne populacje. Nie moinajednak całkowicie
wykluczy ć,że drapieznikamibyłykraby.Poniew aŻbltznyreprezentuj4 nieudaneataki,
ilościowe ana|izy dotycząceselektywnościmiejsca ataku, selektywnościwielkości
ofiary i wpływu drapieznictwa na dynamikę populacji ramienionoga są trudne do
j ednoznacZnego zinterptetowania.Niemniej j ednak, wyniki sugerują istnienie silnej
sezonalnej za|eźzności
ekologicznej pomiędzy ptakami (krabami?) i ramienionogami.
Ana|tzailustruje metodyilościoweuiyteczne w badanichwspółczesnychi kopalnych
bltzn drapieźLniczych
i ma istotneimplikacje paleontologiczne.Drapiezneptaki, k.aby
i ramienionogi mogły współzamieszkiwać strefy mtędzypływowepocząwszy od
późnegomezozoiku. Tak więc, blizny drapiezniczew muszlachramienionogów mogą
mieć długizapis kopalny, interesuj4cydla paleontologii, szczególnie z punktu widzenia paleoekologii ewolucyjnej.PoniewaŻblizny w muszlach linguIidów pozwilujq
rozpoznać sezonalne drapieznictwo, kopalne blizny mogą potencjalnie dostarczyć
danych ekologicznych i etologicznych,które sąZazwycza1rzadkodostępnew paleontologii.