Similar shells are not necessarily a reliable guide to phylogeny: Rissoa guerinii Recluz, 1843, and Rissoa lia (Monterosato, 1884) (Caenogastropoda: Rissoidae): a case study

2010·8.2 MB·English

by  Francesco Criscione

THE NAUTILUS 124(3):117-128, 2010 Page 1 17
Similar shells are not necessarily a reliable guide to phylogeny:
Rissoa guerinii Reeluz, 1843, and Rissoa lia (Monterosato,
1884) (Caenogastropoda: Rissoidae): a case study
Francesco Criscione1
Francesco Paolo Patti-
Functional andEvolutionaryEcologyLaboratory
StazioneZoologica“Anton Dohrn”
P.ta S. Pietro, 1, 80077
Ischia(NA), ITALY
ABSTRACT tal mode, whereas a larger paucispiral protoconch is
directly linked to non-planktotrophy.
According to the recent theory of speeiation through loss of There are several examples of nortehastern Atlantic
planktotrophy, the pairs of northeastern Atlantic species ol caenogastropod pairs ofrelated taxa that possess identi-
caenogastropodswouldbe the resultolacladogeneticevent in cal teleoconchs and differ exclusively in protoconch
whichaplanktotrophicancestorgave risetotwogeographically characters. Oliverio (1996) provided a list of 28 ol these
separated species with different modes ol larval development, P/NP pairs and indicated that their specific status is still
butretainedvirtuallyidenticalteleoeonchcharacters. This idea
a matterofdebate.
was proposed as the working hypothesis for the origin ol the
pair of supposed sibling species Rissoa guerinii Reeluz, 1843 Verduin (1986) maintained that in species of Rissoa
and R. lia (Monterosato, 1884). The present study shows that (Rissoidae) the occurrence of this phenomenon is of
claims ofa close morphological resemblance between the two remarkable extent and nine pairs of Oliverios list were
speciesareunjustifiedasaconsiderabledivergenceemergedin species ofthis genus. Inhis revision of the genus,Verduin
shell geometric morphometric analysis and in 16S rRNA mito- (1976, 1982, 1985, 1986) considered differences in devel-
chondrialgene. Weconclude thatR. guerinii and II lia cannot opmental mode to definitively separate species, thus
beregardedassisterorcrypticspecies andshouldnolongerbe regarding each member of a P/NP pair as a distinct spe-
consideredaplanktotrophie/non-planktotrophiepair. cies. His opinion was commonly accepted (e.g. Bouchet,
Additional keywords: Gastropoda, siblingspecies, larval devel- 1989), although some authors have more recently
opmentstrategy,mitochondrial DNA, geometricmorphometry suggested the possibility of intraspecific variability in lar-
val development (Waren, 1996; Cadee, 1998) or that this
variation indicated incipient speeiation (Rehfeldt, 1968;
Panico and Patti, 2005).
Oliverio (1996) suggested a mechanism of speeiation
explaining the origin of P/NP pairs in the NE Atlantic
INTRODUCTION
region. This involved the modification of larval devel-
Many authors (e.g. Jablonski and Lutz, 1983) identified opment, with one species abandoning planktotro-
two main categories ol developmental strategies in phic feeding, thus giving rise to another species. From
marine invertebrates: planktrotrophy (P), with larvae the observation that in the Mediterranean the non-
feeding on plankton, and non-planktotrophy (NP), with planktotrophic mode is more dominant in the East-
planktonic larvae feeding only on their yolk supply ern basin, Oliverio (1996) suggested a paleogeographic
(leeithotrophy), orwith direct development. model to explain the reasons of speeiation through the
According to the so-called “shell-apex rule” (Thorson, loss of planktotrophy. During glacial periods sea level
1950), larval development in marine gastropods can lows caused isolation between the Mediterranean and
be inferred from observations of the protoconch. A the Atlantic, and between the Eastern and Western
multispiral protoconch and a smaller dimension of the Mediterranean basins. The resultingconditions gave rise
initial whorl accounts for a planktotrophic developmen- to factors thought to select against planktotrophic larvae
(fluctuations in the energy input, restricted areas, and
1 Current address: Malacology Section, Australian Museum, 6 higherpredationpressure; Strathmann, 1978a, b). These
College Street, 2010 Sydney, NSW Australia, franceseo. factors may have caused the shift from planktotrophy to
[email protected] lecitotrophy resulting in speeiation and origin of a P/NP
”[email protected]; both authors arecorrespondingauthors pair.
Page 118 THE NAUTILUS, Vol. 124, No. 3
Within this scenario, species forming a P/NP pair can attempted to crawl extending their foot completely,
be considered sibling species (sensu Knowlton, 1986), enabling the head-foot to be observed in detail. Images
being both cryptic (i.e., difficult to distinguish using the were taken and digitized using a Leica DFC 300 FX
traditional morphological characters) and sister (i.e., video camera and Leica Application Suite version 2.4.0
sharingthe same ancestor). software. Color drawings were also made to better rep-
After the examination of a large amount of museum resent the colorpattern ofthe head-foot.
material ofRissoa guerinii Recluz, 1843, and Rissoa lia
(Monterosato), 1884 (Rissoidae), Verduin (1985) Traditional Shell Morphometry: Standard teleo-
highlighted a strong resemblance in shell characters conch (L, M, W^.i, Dn-i) and protoconch parameters
between these species and concluded that they “may (d and D0) (Verduin, 1982a) were measuredon asample
often only be identified reliably by their type of apex” of 100 randomly selected shells (from 53 males and
(Verduin 1986: 14). Their distribution (Verduin, 1985) 47 females) from live-collected material ofeach species.
overlaps in the Western Mediterranean, but only The total number ofshell whorls (N) was also counted.
R. guerinii occurs in the Atlantic and is absent from the The following standard teleoconch ratios were calcu-
,
Adriatic and Aegean,where R. lia is present. lated: relative height (L/N), slenderness (L/Dn_i), rela-
In this work we provide, based on both dry and live- tive aperture height (M /L) and last-whorl height over
eolleeted material, a critical reinterpretation of the two width ratio (WN.]/ Dn_;l).
taxa and investigate theirphylogenetic relationships. A principal component analysis (PCA) and a discrimi-
nant analysis (DA) were performed on the dataset
obtained combining protoconch parameters and
MATERIALS AND METHODS
teleoconch ratios. Tests and plots were implemented by
The source of the biological material examined in this SPSS v.15 statistical software (© SPSS Inc., 2006) and
study is twofold. Snails from field sampling and empty by SYSTAT statistic software v. 12 (Wilkinson et ah,
1992).
shells belonging to the historical collection of Philippe
Dautzenberg (housed in the Royal Belgian Institute of
Geometric Shell Morphometry: Fifteen shells
Natural History (RBINS), Brussels) have been used in
(8 females and 7 males) at terminal growth were ran-
this research.
domly selected from live-collected material ofeach spe-
cies. Shells were observed using a Leica Z16 APO
Live-collected Material stereoscopic microscope, and colour images were taken
Sampling of living material was performed in the and digitized using a Leica DFC 300 FX video camera
iSnefar)a,liwthtoerrael obfotShanstpaecTieecslac,oSmicmiolnyl(Myeodcictuerrr(aSnceuadenr,i,Iopneirasn. sahnedllsLeiwceareAppalliwcaaytsiopnlaScueidteivnertshieons2a.m4e.0psoosfittwiaorne,.wTihteh
comm.), at 1—5 m depth in Dec. 2005, Apr., Jun., and the coiling axis in vertical position and the aperture on
Nov. 2006. About 0.03 m3 ofthe red alga Pterocladiella thesameplaneastheobjective(Carvajal-Rodriguezetal.,
wcaapsilcloalcleeact(eSd. bGy. SGCmeUlBinA)diSvaintneglifoereseaacnhdsHaompmlmee.rMsaatned- 210905l)a.ndUmsairnkgsth(eLsMo)ftwwearreetesptsaDbIliGs2hev.d2(.F1i0g(uRroelil1)f., 2L0M071a)i,s
rial was immersed in seawater and transferred to the the apexofthe shell; LM2, LM4 and LM6 are placedon
laboratory. Each sample was divided into 20 subsamples the right border of the profile at the beginning of the
that were individuallywashed in a tank containing 5 1 of three lastcompletewhorls. LM15, LM17, and LM19 are
50% seawater for not more than five minutes. Osmotic the corresponding landmarks on the left border of the
shock provided caused the vagile fauna to detach from profile. LM3, LM5, LAI16, andLAI18marktheinterme-
wthheeraelgatlhethsalpleucsiamenndsfawllerien tqhueicbkoltytcoomlloefcttehdeatnadnk,plfarcoemd dainadteLApoIs6i,tiLoMn1r5espaencdtiLvAelIy17b,etLwMe1e7n,LaMnd2LaMnd19LMa4l,ongLMth4e
ienallsyeadweatteerr.minLievdinugndmearteariWailldwaMsaksroorstekdopanMd42t0axsotneormeio-- lcausrtvactoumrpeleoftethwehwohrolrla;nLdML8Mi7s atmatrhekslotwheerisnuttuerremeodfiatthee
scopic microscope. Specimens belonging to Rissoa position between LM6 and LM8 along the curvature of
gmuienriendiibyancdheRc.kilniga twheerepreissoelnacteed;ofthaeiprensiesxiwnatshedertiegrh-t ptahretwohfortlh.eLouMt9erilsipt;heLmMos10taexntderLnAaIl1p2osairteiotnheinmothsetuepxtpeerr-
side of pallial cavity. Some were placed separately in nal positions respectively in the external right and left
running seawater at 18°C and provided with fresh partoftheouterlip; LM11 isthelowestpointatthebase;
Pi.n c8a0p%illeatcheaanotlh.allRie,aortehdersspewceirmeenismmdeiddinaottelsyurvpirveesemrovreed lLeMft1p4rofiisltehoefmthoestsheelxlt;erLnMal1p3oiisntthienprtohfeilleasptowihntorbletawtetehne
than 6 weeks; empty shells of dead specimens were LM12 and LM14 (closest to LM7). As described in
retained and used forobservation. Carvajal-Rodriguezetal. (2005) the matrixofrawcoordi-
nates generated by tpsDIG2 was used in tpsRelw v.1.45
Shells and Head-foot: Adult livingspecimens ofboth (Rolilf, 2007b) to compute shell size (CS), uniform
Pi. guerinii and R. lia were placed in a Petri dish with (U1 and U2) and non-uniform (several relative warps,
seawater. Shells were held with forceps and the snails RWs) shape components for each specimen. Standard
F. Criscione and F.P. Patti, 2010 Page 119
Table 1. List ofstudied material in Dautzenberg collection
(RBINS).
No. ofshells
Acronym contained Main originallabel
Bga About80 Rissoagaerini Bed. var. albino
R. guerinii Dautz./Rochebonne/dd. 16.IX.05
Bgb About 60 Rissoaguerini Reel.var.bipartita Dz
R. guerinii and Dur./Roehebonne/dd. 16.IX.05
Bgc About80 Rissoaguerini Reel. var. conspersa
R. guerinii Dautz. and Durouchoux/
Rochebonne/dd. 16-IX.05
Rhna ManyR. lia R. (Apicularia) lia, Benoit/typiquede
Messinal/Monterosato2.III.17
Rlmb 5 R. lia bis/R. (Apicularia) lia, Benoit/
Messina! Playa
Blp 8R. lia Rissoa lia Monts./Paulille/Buequoy
Bit manyR. lia Rissoa lia Benoit/Tanger/Pallary 1. 7.
98
Table 2. GenBank accession numbers for the sequences
used in the analysis. Accession numbers with the prefix GU
were collected for this study.
Taxon AN Analysis
R. guerinii from GUI77882to NJ, MP, ML, Ml
GU177921
R lia from GU177922to NJ, MP, ML, Mj
GU177955
Figure 1. Shell ofalive-collectedR. lia from S. Teela show- R. similis GUI77963 NJ, MP, ML, Mj
ing the placement of the 19 landmarks used for geometric R. labiosa AY676117 NJ, MP, ML, MJ
morphometricanalysis. Scalebar= 1 mm. RR.. paaurrivsacalpium AGFU441757384739 NMJJ, MP, ML, M|
R. variabilis GU177880 MJ
R. violacea GU177881 MJ
parametric tests were performed on the obtained vari-
ables usingthe SPSS/PCpackagev. 15.0.
RadularAnalysis: The shell of 10 live-collected spec- both forward and reverse strands were analysed.
imens ofeach species was removed and the bodies incu- Genbank accession numbers for sequences used in the
bated for3—4 hours at50°C in 70% KOH solution. After analysis are given in Table 2.
complete tissue dissolution, radulaewere isolated, rinsed For the sequences generated, Parsimony and Maxi-
in distilledwater, and mounted on SEM stubs. Observa- mum likelihood trees were obtained using PAUP* v.
tion and pictures were made using a Jeol JSM-6700F 4.04 (Swofford, 2003). The program Modeltest version
scanning electron microscope. The cusps of rachidian 3.06 (Posadaand Crandall, 1998) was employedto select
andlateralteethwerecounted andfrequencyhistograms HKY+I model in Maximum Likelihood. In computing
were drawn to show differences in cusps arrangement trees, the option of 1000 bootstrap replicates was
between the two species. selected. A reduced median joining network (MJ)
(Bandelt et ah, 1999) was obtained with the software
Molecular Systematics: Forty-five ethanol preserved Network v. 4.5 (http:/Avww.fluxus-teehnology.com/
specimens ofR. guerinii and 34 ofR. lia were randomly NETW4500.exe).
selected from live-collected material. The shell of each
DNA
specimen was broken in a mortar and removed. Dautzenberg Collection Material
extractionplusamplification,purificationandsequencing
of a 337 bp segment of mitochondrial 16S rRNA gene The relevant material in the Dautzenberg collection was
wereperformedas describedinCriscione etah, 2009. examined using an Olympus® SZX10 stereoscopic
Sequences obtained were aligned with CodonCode microscope. After this survey, 15 lots were selected as
Aligner v. 1.6.3 (CodonCode Corporation, Dedham, on their historical value as vouchers (Verduin, 1985)
MA) and the alignment refined by eye. For all samples. and the specific determination of the shells contained
Page 120 THE NAUTILUS, Vol. 124, No. 3
was verified according to current taxonomy. Clearly Extremely weak labial rib, often having appearance of
misclassified specimens in the lots were temporarily whitish smudge as broad as two axial ribs, starting from
removed and not further considered. Several specimens end oflast axial rib and ending before outer lip. Spiral
of each lot were observed using the stereomicroscope ridges with grooves delicately and closely cancellated as
and digital pictures were taken with an Olympus® a result of intersection with growth lines. Growth lines
CAMEDIA C-7070 WZ digital camera. Table 1 contains prosocline and running obliquely over ribs and furrows.
the list ofthe lots studied, accompanied by the data on Aperture as described by Fretter and Graham (1978).
the original label, the number ofspecimens present and Two distinct shell colour varieties detected: typical and
photographed, and their revised species determination. “var. conspersa” (Dautzenberg and Durouchoux, 1914).
An acronym is given to enable identification oi the lot in Typical R. guerinii (Figure 2) with ribless white-brown-
the followingtext. ish or lilac whorls, youngest showing sinuous brown spi-
ral lines corresponding to furrows of ribbed whorls.
Ribless whorls with fawn or brown background colour
RESULTS
on furrows among ribs and on shell base (sometimes
Live-collected Material tinged pale lilac up to base of ribs). A checkerboard
pattern (or a series of zigzag lines) created by back-
Rissoa guerinii Shell: Shell spindle-shaped; apex ground color and a whitish color occasionally present.
solid, sharp and glossy; whorls 7, apical 2-3 nearly llat- Ribs always whitish or pale lilac. Peristome pale violet,
sided, 3-4 youngest ribbed and more convex, with pen- throatwith brown-lilac band extendingonto columella.
ultimate whorl generally bulging out and giving shell Rissoa guerinii “var. conspersa” (Figure 3) with entire
characteristic fusiform shape. Shells of females always shell covered by a uniform chess board pattern or a
bigger than those ofmales with same numberofwhorls. series of zigzag lines (as described for some typical
Shell ornamentation composed by axial ribs and inter- shells), with predominance ofa dark brown colour over
vening furrows, spiral ridges and intervening grooves, whitish colour. Ribs often encircled by a white line, nor-
and growth lines. Shells devoid ofspiral ornamentation mally interrupted in correspondence of intervening fur-
rarely found. Axial ribs limited to youngest 3-4 whorls, rows. Apex, peristome, and inner part of aperture as
11-14 perwhorl, robust andprominent, swelling in mid- described fortypicalpattern.
dle of whorl and fading toward shell base; normally
slightly opisthocline, tending to prosocline on body Rissoa guerinii Head-foot: As for shells, two colour
whorl near aperture and close to suture being slightly types detected (Figure 6, 7), viz. typical R. guerinii
flexuous; each rib equal in breadth to interspaces. and R. guerinii “var. conspersa” (Dautzenberg and
Figures 2-9. Picturesofshells anddrawings ofsoftbodypartsofR. guerinii typical (2, 6), R. guerinii “var.conspersa”(3, 7), R. lia
“var. castanea” (4, 8) and R. lia “var.ftlva" (5, 9). Scale bar= 1 mm. Drawings by D. Scuderi.
"
F. Criscione and F.P. Patti, 2010 Page 121
Durouchoux, 1914). Foot always whitish and median with a whitish spot behind their base. In R. lia “var.
part ofsole stained brown, lighter in typical R. guerinii castanea” snout dark and its distal part yellowish, in
than in “var. conspersa.” Snout light brown in R. guerinii “var.fulva” snout yellowish, often with a short brownish
typical and darker brown in “var. conspersa." Margins stripe runningalong median part.
of distal portion of snout and remaining part of head
yellowish in R. guerinii and light brown in “var. Traditional Morphometry: Among the 6 principal
conspersa.'’ Cephalic tentacles whitish, but sometimes components extracted, PCI (55%) and PC2 (26%)
dark brown in “var. conspersa.” A whitish spot behind explained most oi the variance observed (Table 3). Com-
base ofcephalic tentacles always present. ponent matrix (Table 4) and loading plot (Figure 10)
illustrate the correlation between principal components
Rwoofhiptosaerqsnluosear;irblilibwealshesoSs)hrs,,elplsolet:nh6u,elrtasSlilhrmeiaelbtqlbeuecagdolen-noisecriarrdlaie,lbdllayepasnesnoxdt(stbohtluuoilmudg,igidhon,bgltaoasutpusitwecharaoelsnrtdl2. wwcaioinrttdrhheslhdLea/ltaNlen,dvdarwLiDi/atDb0hl.nePs_M].C/2aPLnCdiIsanWnidsotaNw_sseit/iargkonDlnigyfNil.cylna5epngotsaslttiyrtioicvnvoegerlllryyyelccnaooetrrgerradeetlliwaavitteteelhddy
Sexual dimorphism as for R. guerinii. Shell ornamenta- teleoconchDratios but it is strongly positively correlated
tion of axial ribs and intervening furrows, spiral ridges with d and 0.
and intervening grooves, and growth lines. No shells No clear clusters resulted from plotting PCI value of
devoid of spiral ornamentation found. Axial ribs (when each shell against its respective PC2 value (Figure 11).
present) always 14perwhorl, strongandprominent,with The 95% confidence ellipse of R. guerinii contains
same thickness across whorl but fading towards base on almost onlypositive values ofPCI, but both positive and
bodywhorl; normally slightlyopisthocline, tending to be negative values of PC2 whereas that of R. lia encircles
prosocline; each rib slightly narrower than intervening mostlynegativevalues oi both PCI and PC2.
furrow. Nolabralriborwhitishsmudgebeforeperistome The eigenvalue of discriminant function between the
present. Spiral ornamentation as for R. guerinii. Aper- two species was 1.307, the canonical correlation 0.753
ture oval or D-shaped, peristome not showing sinuses or and the Wilks' lambda (0.433) was highly significant
slight projection of inner lip; edge very thin, beveled (p<0.001). The structure matrix (Table 5) shows the
internallyturningouttoformathough thinflange. Outer cumulativewithin-groups correlations between discrimi-
lip arising below periphery of body whorl, (somewhat natingshellvariables andthe standardizedcanonical dis-
criminant functionobtained. Table 6 shows the results of
below level at which ribs end), its curvature initially
slight or sometimes nearly straight. Columella short, the classification statistics obtained using the values ol
peristome everted over a groove, no umbilicus present. the discriminant function ofeach individualtopredictits
Two distinct color varieties detected, here named a posteriori species membership.
castanea andfulva (after Monterosato, 1884). Intermedi- Geometric Morphometry: Table 8 shows the per-
ate specimens rarely found. R lia “var. castanea” centages and a descriptive statistical summary of the
(Figure4)violet-brownishtodarkbrownwith ribs always
lighter. First two whorls white but sometimes brown to
darkbrown. Peristomeviolet-brownish. R. lia “var.fulva Table 4. Component matrix for the first two principal
components.
(Figure 5) uniformly fawn with ribs always lighter or
even whitish. First two whorls fawn or white. Peristome Component
as in “var. castanea”, but neck with a brown-lilac band
running from lowerborder oflabral rib to columella. 1 2
D -0.271 0.864
Rissoa lia Head-foot: As forR. guerinii, two different D(l -0.318 0.848
color tvpes detected remarkably distinct in colour pat- L/N 0.843 0.181
tern and corresponding to shell colour variety and thus 1VDn., 0.914 0.134
M/L -0.913 -0.151
named castanea and fulva (Figures 8, 9). Foot always W
entirely whitish and cephalic tentacles always yellowish N-i/Dn_j 0.860 0.106
Table 3. Total variance <explained by single principal components.
Initial Eigenvalues Extraction Sums of Squared Loadings
Component Total % ofVariance Cumulative % Total % ofVariance Cumulative %
1 3.294 54.901 54.9 3.294 54.901 54.9
2 1.551 25.852 80.8 1.551 25.852 80.8
3 0.432 7.193 87.9
4 0,342 5.704 93.6
5 0.235 3.909 97.6
6 0.147 2.442 100.0
Page 122 THE NAUTILUS, Vol. 124, No. 3
Table 6. Classification summaryforthe discriminantanalysis.
Species R. guerinii R. lia Total
Count R. guerinii 93 7 100
R. lia 14 86 100
%
R. guerinii 93.0 7.0 100.0
R. lia 14.0 86.0 100.0
Table 7. Multiple regression model testingallometryforthe
non-uniform shell shape variables.
Multiple regression Variables inthemodel
PCI(55%) r2 F Name Beta
0,5 20.3*** RW2 -0.648***
Figure 10. Loadingplot of the shellvariables relative to the RW9 0.282**
firsttwoprincipalcomponents.
**P < 0.05, ***P < 0.001.
relative score for CS, the two uniform components and
the first 10 RWs, explaining more than the 91% of
the overall variation. Table 7 shows the results of the
allometrie analysis for shell shape measurements
conducted by step-wise multiple regression analysis for
centroid size (as dependent variable) and two uniform
and29non-uniform measurements, as independentvari-
ables. The F test of the regression analysis was highly
significant (p<0.001) and two of the relative waqas,
RW2 and RW9, contributed significantly to the regres-
sion model on the centroid size. Centroid size, uniform
components and only the first ten relative warps were
considered in the analysis of variance (ANOVA), per-
formed to evaluate the significance ofdifferences in size
and shape variables. Shells ofR. guerinii and R. lia dif-
fered significantly in CS (p<0.001), RW1 (p<0.05) and
RW2 (p<0.05). Thecumulative results ofthe analysis are
shown in Table 8. The significance level obtained forthe
corrected analysis (ANCOVA) with centroid size as
covariate was not maintained for the difference in RW1,
but was not affected for RW2. The eigenvalue of the
PCI stepwise discriminant function between the two species,
Figure 11. Scatteqrlotofthefirsttwoprincipalcomponents. calculated for the 29 RWs, was 3.199, the canonical cor-
95% confidence ellipses are drawn for each species. Dashed relation 0.873 and Wilks’ lambda (0.238) was highly sig-
ellipse represents R. guerinii. nificant (p<0.001), indicatingagood separation between
groups. Twelve shape variables contributed to the
discriminant function (RW1, RVV2, RW4, RW5, RW6,
RW7, RW9, RW10, RW18, RW23, and RW27).
Table 5. Stnletnre matrix showingthe loadings of each shell
variable on the discriminant function. The variables are The standardized coefficient matrix (Table 9) shows
ordered by absolute size of correlation within discriminant the relative importance of the independent variables in
function. determining the standardized canonical discriminant
function. Using the individualvalues ofthe discriminant
Discriminant Function functions to predict a posteriori species memberships,
M/L 0.717 26 (86.7%) individuals out of 30, were assigned to the
IVN -0.699 correct species, leaving only4 (13.3%) thatwere errone-
wL/Dn_i -0.672 ously classified. Looking at species statistics (Table 10),
n_,/dn. , -0.487 13 (86.7%) specimens of R. guerinii were correctly
d 0.382 assigned to this species and only two (13.3%) were
D0 0.307 assigned to R. lia. The same percentages of correctly/
F. Criscione and F.P. Patti, 2010 Page 123
Table 8. Descriptivestatisticalsummaryandresultsof ANOVAandANCOVAforthe main shell sizeandshapevariablesbetween
R. guerinii 1and R. lia. **P < 0.05, *** P < 0.001, ns = non significant.
Measure CS Uf U2 RW1 RW2 RW3 RVV4 RW5 RW6 RW7 RW8 RW9 RW10
Varianceex]plained 32.2% 19.4% 14,5% 8.2% 4.2% 3.6% 3.2% 2,5% 2.0% 1.9%
Mean 1020 -0.001 0.001 0.012 -0.01 -0.002 -0.004 -0.003 0.003 -0.003 0.000 0.002 -0.001
R. guerinii SD 122 0.014 0.010 0.038 0.0161 0.025 0.018 0.012 0.010 0.010 0.010 0.009 0.008
n Mean 815 0.001 -0.001 -0.012 0.011 0.002 0.004 0.003 -0.003 0.003 0.000 -0.002 0.001
11/i SD 128 0.012 0.009 0.016 0.026 0.017 0.012 0.010 0.010 0.009 0.008 0.006 0.007
ANOVA 20 10*** 0.11 ns 0.45 ns 4 yg** 7.97** 0.32ns 2.05 ns 2.42 ns 1.96 "s 2.16 ”s 0.06ns 3.43 ,1S 0.75 ,,s
ANCOVA 1.06ns 0.46ns 3.21 ns 9.96** 0.20ns 1.00115 1,38ns 1.19 ns 1,57ns 0.15 ,,s 1.79ns 2.01 "s
Table9. Standardizedcoefficientmatrixshowingtherelative
importance of the shape variables.
Discriminantfunction
RW2
1.501
RW1 -1.276
RW9 -1.127
RW5 0.978
RW7 0.932
RW4
0.911
RW6 -0.893
RW23 -0.789
RVV27 0.716
RW18 -0.685
Table 10. Summary of classification results obtained em-
ploying the discriminant unction.
1
Species R. guerinii R. lia Total
Count R. guerinii 13 2 15
R. lia 2 13 15 Figure 12. ThinplatesplinerepresentationsforRW2, show-
%
R. guerinii 86.7 13.3 100.0 ing the deformation of the grid for the average values of
R. lia 86.7 13.3 100.0 R. guerinii (left) andR. lia (right).
erroneouslyclassified specimens ofR. lia were observed. Molecular Systematic^: The topologies ofthe P and
ML
In Figure 12 the thin plate spline representation allows trees (Figure 16and 17) werecomparable: two main
interpretationingeometricterms thepositive (character- clades, corresponding to R. guerinii and R. lia clearly
,
istic of R. lia) and negative deviations (characteristic of separated from each other and from the outgroups by
R. guerinii) values for the most significant non uniform high bootstrap values.
shapevariable, RVV2, between the two species. Two individual sequences of R. guerinii formed a
weakly supported clade closely related to the main one
Radular Analysis: No relevant differences emerged ofR. guerinii. The M network (Figure 18) showed two
[
between the radulae of R. guerinii (Figure 13) and consistently distinct (IS differences) clusters of haplo-
R. lia (Figure 14) in relation to the shape of teeth and types (corresponding to R. guerinii and to R. lia). R.
cusps. Rachidian with one median cusp, two pairs of guerinii displayed higher genetic structure than R. lia
,
lateral cusps and two pairs ofbasal cusps. Lateral tooth witli one haplotype occurring in 29 specimens, 2 occur-
with a median primary bigger cusp with respectively ringin2 specimens and 11 unique haplotypes. No single
2-3 inner cusps and 4-5 outer cusps in R. guerinii or haplotype of R. lia appeared to be dominant: the most
2 inner cusps and 3-5 outer cusps in Pi. lia. Figure 15 common ones occurred in 11,8, 5, 5 and 2 specimens; 3
represents the frequency distribution of tire number were unique. Both clusters were consistently distant
of cusps of the lateral tooth observed in the sample from 5 out of 6 outgroups, with R. lia closer than
studied and shows higher variability in R. guerinii for R. guerinii. UnexpectedlyR. auriscalpium revealed veiy
this character. little difference (1-2) from some R. guerinii haplotypes.
.
Page 124 THE NAUTILUS, Vol. 124, No. 3
2 3 3 4 5
Figure 15. Frequency histograms of distribution of inner
andoutercusps in thelateraltoothofR. guerinii andR. lia.
DISCUSSION
Shell and Head-foot: While in the early malacologi-
cal literature satisfactory descriptions ol the shell of
R. guerinii are available (Jeffreys, 1869; Fretter and
Graham, 1978), descriptive data on R. lia are limited to
its essential original description (Monterosato, 1884;
Figures 13-14. SEM photographs or the radular ribbon of page 139).
R. guerinii (13) andR. lia (14). Scalebar= 10 pm. Verduin (1985) provided a reinterpretation ofthe two
taxa, stressing an extremely close morphological resem-
blance between them. However he based his diagnosis
only on museum chy material, often in a poor state
Dautzenberg Collection Material ofpreservation. As an example, the character “punctate
spiral striae” (Verduin, 1985, pages 112 and 114),
Shell VisualObservations:Rissociguerinii: The shells reported forthe shells ofboth species, should be consid-
ofthe lot Rgawereconsiderablyworn,youngerwhorls in ered the result ofthe deterioration ofthe subtle reticu-
most uniformly whitish, in some uniformly yellowish; latepattern, formedbygrowth lines andspiralridges and
older 2—4 whorls always pale violet. Most ofthe shells of easilyvisible in fresh shells.
Rgb showed worn whitish/yellowish lowermost whorls The redescriptions provided here are based on fresh
and remaining whorls with the pigmentation described shells at terminal growth, which maintain the peculiar
for R. guerinii typical. Shells of Rgc were better pre- characters ol ornamentation and pigmentation ol each
served and showed the chessboard (or zigzag) pattern species, but at the same time are readily matched with
describedabove forR. guerinii “var. conspersa” original museum material.
Dautzenberg and Durouchoux (1914), on the base of
Rissoa lia: Shells of Rlma and Rlmb were topotypes, the material ofthe lots here named Rga, Rgb, and Rgc,
the former being also from Monterosato collection. described four colour varieties for R. guerinii typical
:
Although these shells were rather worn, the two colour (white with the intervals among ribs brown), conspersa
varieties, described above forR. lia, were still detectable (brownbackgroundwithachessboardpattern formed by
on the younger whorls but had completely disappeared very small spots), albina (totally white) and bipartita
from thewhitish oldest whorls. (with the first five or four whorls dark violet and the
Shells of Rip and Rlt were worn but a different pat- remaining entirely white). Based on the same material,
tern, with a whitish spiral band running in the middle ol Verduin (1985) even suggested the presence ofa north-
the youngestwhorls on a darkbrown background. ern R. guerinii subspecies (showing the latter two color
F. Criscione and F.P. Patti, 2010 Page 125
R.guerinii
10 R.guerinii 100
Figures 16-17. Parsimony (16) and Maximum Likelihood (17) Trees drawn afterthe analysis ofIBS sequences of R. guerinii and
R. lia. Circle radius isproportionaltosequences frequency. Bootstrapvalues aregiven at the leftsideofeach branch.
R.viohicea
R.guerinii(45)
O
R.lia(34)
outgroups
medianvector
16
- ©
R.variabilis
• • • •
8
Figure IS. Median Joining Network drawn on the basis of 16S sequences. Number of sequences employed in brackets. Circles
radiusproportionaltohaplotypefrequency. Dashedlines represent mutationaldistancehigherthan4 (values reportednearby).
varieties) and a southern one (showingthe former ones). and colour patterns can be discerned, among which die
We cannot see any justification for these claims as the colourpatternconspersaknownin II guerinii."
varieties albina andbipartita are the result ofthe deteri- This observation apparently suggested to Verduin
oration of the typical pigmentation, as shown by the (1985) the idea to move R. lia from the subgenus
worn shells formingthe lots Rga and Rgb. Apicularia Monterosato, 1884, to Goniostcmm Villa, 1884
Althoughhis interpretationofR. lia (basedon dieobser- (the same subgenus ofR. guerinii). We regard his obser-
vation ofRlmaand Rlmb lots) is generallycorrect, Verduin vation as incorrect and in disagreement with the original
(1985: 114) added to its species diagnosis the observation description of Monterosato (1884) and with our observa-
diat“in asample fromTrapani, Sicily(fig. 25) manycolours tions on the material in this study. We found no evidence
,, , , , ,
Page 126 THE NAUTILUS, Vol. 124, No. 3
that the color pattern conspersa belongs to R. lia. andlessreliable, duetothelargeoverlapshown. Thisidea
Although we could notexamine the sample mentioned by is in contrastwith theviewofthe discriminatingpowerof
Verduin (1985), anexamination of the figures heprovided protoconch dimensions (Verduin, 1986).
(25a-h), allowed us to assess that these figures are com-
posedbyamixtureofshellsofR. lia andR. guerinii shells, Some Methodological Considerations: Cadee
associatedbecause oftheirsimilarapexdimensions.
(1998: 91) critically revised the methodology used by
However, the statement that some R. lia show “the
Verduin (1976, 1982b, 1985, 1986) highlighting that the
oprtbherseceeesrnsevthneacltepliseoo”cnfusl(biVrmaeoarrkddeureinencbo,ornro1dwl9son8t5iso:sfhR1icR1p4oi)lsa,osnuoirdaslRiislnnpt.esaagwfnrhrdeioecmwmhietSenhantrcsdiwiroincmtliehea vmleoercataistcuiaorlnepmoofesnittthiesonloinofefdatlhoaennsgdhwelhDli0cahn“dddetaphneednsdDom0oenawrhetahmeteaaasrcubciruterrdaa.tr”ey
Cadee (1998) dealtwith the former issue byperforming
(Tyrrhenian Sea), showing this same peculiar coloration repeated measurements ofd and D on specimens ofR.
(Fasulo, pers. comm.). Although this coloration is not membranaeea, whose vertical place()ment was reiterated
typical of the type material ofR. lia further observation
ten times, and obtained a standard deviation of about
on live-collected material from these localities is needed mm
to properlyaddress the issue. 0.01 for both the variables. This value is roughlythe
same as that which separates the mean values of d and
poartRbtilesorstnocah(ie.sse).p,,epcaoilsetishtiotouengnhdanttdhoedymiammieannytsaiisonhnostwhoefinehtlereaamsdpe-enfctoisoftiacscvosalprooituasr- TDh0isofmeR.anlsiatahantdsRo.mgeueorfintihieidniftfheisrenscteusdyin(nportotsohcoownn)c.h
dimensions might be indeed the consequence of the
tion in colour intensity (Fretter and Graham, 1978). This
phenomenon has been employedtosupportspecies iden- inaccurate measurements, rather than representing real
tity (R. pawn and R. interrupta Waren, 1996; R. guerinii variation. The alternative solution of performing mea-
surements on digital photographs has been also recently
easnidzeR.repcaennhtorsmpeecnisaitsionCr(i7s1cimoenembertaaln.a,e2e00a9)tyopretAo haynpdothB-, employed (Criscione et al., 2009). The second issue is
that real Rissoa protoconchs often do not correspond to
Rehfeldt, 1968; R. auriscalpium type a and b, Colognola
et al., 1986). Tins rule is in agreement with the slight bthuetimdoeasltsohefllthaepemxafriegurraetdhebryiVrerredguulianr.(1I9n77p;roFtiogcuornec1h)s,
variation in color intensity observed in this study lor the
wtwhoosvearcieotlioersvoarfiRe.tigeusesrihnoiwiedbuttwiot ddiosetsinncottpaatptpelrynst.o R. lia tlihkeeltihneesealtohnegrewhiiscahldaragendraDng0eshofouplodssibbeilmiteiaessutroeldocaantde
the choice is largely arbitrary.
Traditional Morphometry: Although the present
morphometric analysis is restricted to populations from Geometric Morphometry: The ANOVA revealed a
a single locality, its results are not only helpful in under- significant larger size (CS) of the shells of R. guerinii
standing intrapopulation variability in shell morphology, compared to those of R. lia, confirming the results
but also in drawing more general conclusions in terms of visual observations and traditional morpliometiy.
ofspecific differentiation. The differences observed can ANOVA showed a significant difference (p<0.05)
be expressed in terms of teleoconch and protoconch between the two groups in the first (RVV1) and the sec-
variation, with the first element being by far the most ond (RW2) non-uniform shape variables. The signifi-
important. Populations of both species show a wide cance for RW2 remained unaltered when correcting the
variability in protoconch dimensions, which is larger for analysis for CS (ANCOVA), while that of RW1 was not
R. guerinii. Rissoa lia shows twice the intrapopulation significant, indicating that the shape difference ex-
variation in teleoconch characters than that observed plained by that variable was dependent on size. This
for R. guerinii. means that the two groups differ exclusively on the sec-
The values of shell parameters obtained for the two ond non-uniform shape variable (RW2), independently
species showed a considerable overlap, which is mainly from the correlation between shape and size.
dependenton the strongsimilarityinprotoconch dimen- Tlle discriminant function calculated from all the
sions and not, surprisingly, by the resemblance of the non-uniform shape variables, was successful in morpho-
teleoconehs. Most shells ofR. guerinii appear to have a metrically discriminating the two groups. RW2 was the
larger relative size, a more elongated aperture and to be most important variable in determining the distinction
more slenderthan most R. lia shells. between R. guerinii and R. lia. The mean values ofthis
Discriminant analysis showed that R. guerinii and variable for each of the two groups (positive for R. lia
R. lia populations can he distinguished mainly for the and negative for R. guerinii) were plotted in a tps repre-
more elongated aperture of R. guerinii shells, for their sentation (Figure 12). The plot showed thatthatvariable
larger relative size, and the higher slenderness. The RW2 is a reflection of the most obvious discernable
importance ofprotoconch is onlymarginal. shell shape difference. This comprised the more fusi-
In summary, our results show that teleoconch mor- form shape of R. guerinii, contributed by consistently
phometrycan be efficiently used to discriminate between narrowerwhorls than those ofR. lia (represented bythe
R. guerinii andR. lia whereasprotoconchdimensions are contraction ofthe corresponding zone of the grid) and a
less representative of the overall interspecific variation substantiallyequal penultimatewhorl.