How and why to achieve greater objectivity in taxonomy, exemplified by a fossil ostracod (Amplocypris abscissa) from the Miocene Lake Pannon

Abstract

A project on fossil ostracods from Hennersdorf (Vienna Basin, Middle Pannonian “E” stage) documents the non-marine ostracod Amplocypris abscissa (Reuss 1850) as a polymorphic taxon. The high morphological variability of the valves belonging to this species and its geographic distribution in the Lake Pannon point to a widely spread fossil taxon. This traditional view emerges from the fact that A. abscissa displays few homologous “landmarks” (morphological reference points) which makes it difficult to compare morphotypes within and among populations. The present contribution offers arguments for the need of objective criteria when describing ostracods with few visible morphological traits. It is demonstrated that using a composite algorithmic approach which combines information implemented in the computer programme MORPHOMATICA for different variables, measured on interval and ratio scales, is able to define morphological traits objectively. The data analysed with multivariate statistics are further used for diagnostic profiles of clearly delineated morphotypes. The potential taxonomic value of three morphotypes here identified for Amplocypris abscissa is discussed. It is argued that this taxon represents a phylogenetic lineage within which a cluster of species with discrete morphological traits exists. Finally, it is proposed to apply similar algorithms for the necessary revision of the whole group of Amplocypris species from Lake Pannon.

1. Introduction

1.1. Objectivity

Objectivity is a term generally understood as the ability of describing things without reference to individual (idiosyncratic) interpretations. This concept is commonly perceived as controversial. This is due to the fact that within human experience and implicitly in the act of perception and those of critical interpretation of the information a more-or-less implicit relationship between subject and object, respectively between knower and known exists (von Foerster 2005). In the present context we use the term objectivity as the act of achieving precise descriptions of our objects of study following well established protocols and/or algorithms. These latter should be easily understood and, when necessary, repeated for verification and further critical interpretation. Only in this way can our data become a matter for serious discussion which should allow conceptual and/or practical advances within our field of research. With other words we express here what the American philosopher Dewey (1929: 28) wrote: “The scientific investigator convinces others that not by plausibility of his definitions and the cogency of his dialectic, but by placing before them the specified course of searchings, doings and arrivals, in consequence of which certain things have been found. His appeal is for others to traverse a similar course, so as to see how what they find corresponds with his report”. It is important to underline that in the present project the selected procedures for study and their usage and interpretation are the product of extended experience and practice we have developed in recent years. It should be understood as a kind of “personal knowledge” as conceived by Polanyi (1958). This author defends the epistemic concept of verifiability of scientific ideas. This point of view differs markedly from the way the well known palaeontologist G.G. Simpson (1961) defined the concepts of objectivity versus those of subjectivity when dealing with animal taxa. For Simpson (1961: 114) organisms are real, objective, entities while “the concept of a taxon […] is invariably subjective”. The present contribution tries to demonstrate that it is possible to use instead of subjective views for conceptual problems, as those occurring in animal taxonomy, “transparent” arguments which help to accept within a scientific discourse the objectivity of ideas.

1.2. The specific problem related to the taxonomic status of the fossil ostracod Amplocypris abscissa (Reuss)

Taxonomists are commonly confronted with incomplete descriptions. The situation is exacerbated for fossil taxa which display few homologous ‘landmarks’ (morphological reference points) useful for defining morphological traits. These latter are necessary for building clear taxonomic diagnosis, for phylogenetic analysis or for projects dealing with (palaeo)ecology, (palaeo)biogeography or (bio)stratigraphy.

Amplocypris abscissa was briefly described by A.E. Reuss (1850) from several outcrops near Vienna and Sopron, respectively from the Vienna and the Eisenstadt-Sopron basins (Late Miocene, Pannonian). Subsequently, it was mentioned (cf. inter alia Mehes 1907; Pokorný 1952; Olteanu 1971; 1986; Sokač 1972; Krstić 1973; Jiriček 1985; Zelenka 1989, 1992; Vasiliev et al. 2010) all around the palaeo-Lake Pannon. This latter was a giant limnic system which was extended during the Late Miocene from the western sites now in Austria, Hungary, Slovakia, parts of Czech Republic, to eastern and southern ones, respectively Romania, Serbia, Croatia, Slovenia (cf. Magyar et al. 1999a and here Fig. 5). The wide geographic distribution of one Amplocypris species contrasts with the high number of fossil species belonging to this genus, considered to be endemic to Lake Pannon (Krstić 1973). Therefore the question we ask is if Amplocypris abscissa is really a species extensively distributed in Lake Pannon or if it represents a cluster of species, each with discrete morphologic traits and more-or-less restricted geographic distribution.

Fig. 5.

Palaeogeographic reconstruction of Lake Pannon, c. 9.5 Ma BP (from Magyar et al. 1999a); HD – position of the Hennersdorf sampling site.

1.3. The A.E. Reuss’ description of Amplocypris abscissa in “Die fossilen Entomostraceen des Österreichischen Tertiärbeckens”

The monograph of A.E. Reuss “Die fossilen Entomostraceen des Österreichischen Tertiärbeckens” (1850) is a milestone for the documentation of the Late Miocene age ostracods of the Vienna Basin during the Pannonian (cf. Zorn 1999). Reuss described and illustrated 90 species from which 80 were new to science. The illustration is obviously poor and the systematics simplistically resolved.

Reuss (1850) described the right valve of A. abscissa as being large (1.5 mm long), with the posterior third elongated and postero-ventrally angular shaped, from where also the name originates (Fig. 1A, B).

Fig. 1.

Details of original figures of the Reuss’ monograph; A, B – Amplocypris abscissa (Reuss); C, D – Amplocypris recta (Reuss). Note that the orientation of valves in the original plate 8 of Reuss for A (Fig. 3 in Reuss) and C (Fig. 12 in Reuss) are with the posterior towards the top extremity up-set. Here the correct orientation of the right valves in B (Fig. 2a in Reuss) and D (Fig. 11a in Reuss) follows the present-day practice standards.

The first sampling site mentioned in the description is an area located between Atzgersdorf and Altmannsdorf, in the southern part of today’s Vienna. The next locality mentioned by Reuss is Sopron in Hungary (at the time named Ödenburg).

A part of the original sediment material of Reuss’ collection still exists at the Natural History Museum Vienna (NHMW), Geological–Palaeontological Department. We were able to recover several valves of this species from the material of Sopron (Ödenburg; cf. Annex 1, Tab. 1), which are much similar to the Lectotype here presented (see 1.4.).

1.4. The documentation of Amplocypris abscissa by K. Turnovsky and A. Papp

During 1946 and 1950 A. Papp and E. Thenius (cf. their 1954 monograph) produced a micro- and macro-palaeontological study on the Late Miocene environment in the Vienna area, especially at Vösendorf. Papp & K. Turnovsky (1950) identified A. abscissa of Reuss at Leobersdorf and attributed it to the genus Herpetocypris Brady & Norman.

Figures of their material were drawn by A. Papp and circulated in an unpublished report (1949) of Turnovsky: “Die marinen und sarmatischen Ostracoden des Wiener Beckens; Revision der Reussischen Arbeit” (deposited in the library of the Geological Survey in Vienna, Document A 00508 R 1981) and later on published by Turnovsky, in Grill & Küpper (1954). A sample containing specimens of A. abscissa from REUSS’ material collected around Atzgersdorf was chosen by Turnovsky and deposited at the Natural History Museum Vienna (Geological–Palaeontological Department) under the catalogue number 1949/III/16. The specimens are now registered as Lectotype REUSS 75 Atzgersdorf (NHMW - 2011/0158/0001-003). We present here (Fig. 2A) the Papp’s drawing from Turnovsky (1954) and compare it to the chosen lectotype specimen (Fig. 2B, C). The inner view of the valve especially is illustrative of the acuminate shape of A. abscissa and resembles the specimens from the Reuss material sampled at Sopron (Fig. 3).

Fig. 2.

Amplocypris abscissa, right valve; A, B – inner side; C – outer side; A – figure reproduced from Turnovsky (1954, Pl. 14, Fig. 13b, det. A. PAPP); B, C – Lectotype NHMW-2011/0158/0003 (Atzgersdorf); scale bar = 0.5 mm.

Fig. 3.

Amplocypris abscissa, Reuss material from Sopron; A–C – right valve (RV), D – left valve (LV); A, B – external view, C, D internal view, A – valve ÖD 7444 in TLM view, B – valve ÖD 10Ra, SEM view, C – valve ÖD 12Ri inner view, D – valve ÖD 6; Li inner view; numbers pertain to the photos of the material deposited in the ostracod collection at the Universalmuseum Joanneum, Department of Geology and Palaeontology, Graz; scale bars = 0.5 mm.

We will take this information as evidence for the definition of A. abscissa (Reuss), sensu stricto, when we will redefine (below) this species with morphometric and multivariate statistic procedures.

1.5. Definition of Amplocypris Zalanyi, 1944 and comments on Amplocypris recta (Reuss, 1850)

Zalanyi (1944) created the genus Amplocypris (type species A. sinuosa Zalanyi, 1944) based mainly on the special shape of the dorsal area of the valves. Zalanyi (1959) expanded the diagnosis of Amplocypris mentioning that the radial pore canals on the anterior and antero-ventral side are long, present in high number and many of them are fused at the base (cf. here Fig. 8B). These latter characteristics are key diagnostic traits for this genus (cf. below).

Fig. 8.

Amplocypris abscissa, from Hennersdorf, TLM view (reversed position of specimen HD 7316); A – right valve (scale bar = 0.5 mm); B – the antero-ventral side showing radial canals (see text), detail from A.

Mehes (1907) described from Soceni (Banat, Romania) and Sopron (Hungary) a new species called Herpetocypris strigata. This species was later on classified by Pokorný (1952) as Herpetocypris recta (Reuss). The extensive studies of ostracods related to the Reuss’ species by Krstić (1966, 1973) documented the validity of the genus Amplocypris. Species, like A. abscissa, do not have the characteristic traits of the genus Herpetocypris Brady & Norman (Herpetocypridinae Kaufmann 1900). Amplocypris species lack anterior septa and the inner list is undeveloped. We have to mention also that the high density of simple pores (i.e., the normal pores, cf. Meisch 2000) on the whole surface of the valves and the inflated shape of the carapace, are characteristic traits for the Amplocypris species. Additionally these species do not display “vacuole voids” within the calcified lamella of the valves, a characteristic of Herpetocypris (cf. Danielopol et al. 1986).

All these peculiar traits of Amplocypris point to the assignment of this genus to the Eucypridinae Bronshtein 1947 and not to the Herpetocypridinae Kaufmann 1900. Amplocypris appears closely related to the genus Tonnacypris DIebel & PIetrzeniuk, recently revised by Van der Meeren et al. (2009) and also assigned to the Eucypridinae. Species of this latter genus (we examined T. tonnensis DIebel & PIetrzeniuk and the information in Van der Meeren et al. 2009) display on the antero-ventral margin of the valves shorter radial canals with few fused branches, as compared to the valves of Amplocypris species.

Reuss (1850) described and illustrated (cf. here Fig. 1C, D) a species related to Amplocypris abscissa he called recta. Reuss (1850) mentioned that this latter species displays three different “varieties” that he numbered with a, b and c. Pokorný (1952), as first reviser of what it is now named Amplocypris recta, decided that the morphological traits of the “b variety” should be diagnostic for this taxon and the figure 11a in the Reuss’ monograph (cf. here Fig. 1D) should be considered the lectotype of A. recta (it was not stated where this lectotype is deposited!). Reuss (1850) pointed out that the “b variety”, as compared to the “a variety”, has the ratio height / length of less than 50 % and the antero- and the postero-dorsal sections are more or less angular. We identified at St. Margarethen in Burgenland (Lower Pannonian stage) valves which fit the descriptions of Reuss (1850) and of Pokorný (1952). We superimposed on the outline of Reuss’ recta mentioned above (cf. Fig. 1D) the outline of St. Margarethen (cf. our specimen No. 3471 SM16; Annex 1, Tab. 1; Fig. 4A) and their close similarity becomes apparent (cf. here Fig. 4B). Therefore we will use here this species for comparative purposes in order to better define Amplocypris abscissa.

Fig. 11.

Right valves of Amplocypris species (reversed view), diversity of shapes; A – A. recta, specimen SM16 - 3471; B to F – valves of A. abscissa; B - HD 3829, C - HD 7354, D - HD 7373, E - HD 7942, F - HD 7356; scale bars = 0.1 mm.

Fig. 4.

Amplocypris recta, right valve; A – specimen SM16 - 3471, St. Margarethen; B – superimposition of Amplocypris outlines for RV (reversed) normalised to equal surface; 0 – outline of the original A. recta figured by Reuss (cf. here Fig. 1D), 1 – outline of the specimen SM16 - 3471; length of vector lines 2.5 enlarged.

1.6. Amplocypris abscissa from Hennersdorf and additional comparative material

The ostracod material here discussed originates mainly from the Hennersdorf site (Fig. 5) studied by Harzhauser & Mandic (2004), Harzhauser et al. (2004, 2008). The investigated area is a clay pit about 10 km south–east from Vienna. Currently, the section is about 14 m thick and comprises blue-grey clay with in situ clusters of Congeria subglobosa Partsch in the base. This is topped by a 4 cm-thick bivalve coquina, termed He 4 in Harzhauser & Mandic (2004). This marker bed underlies an approximately 8 m-thick unit of green-blue-grey clay and silty-clay intercalated by numerous thin layers with loose dreissenid coquinas. The top of the Hennersdorf section is represented by about 7–8 m yellow-grey clay and silt which is practically barren of macrofossils except for a marker coquina bed (He 1 in Harzhauser & Mandic 2004) in its basal part. The succession is dated as Middle Pannonian and corresponds to the later stage, Pannonian E (Harzhauser et al. 2004). The magnetostratigraphic dating of the section Hennersdorf allowed its correlation with the chron C5n (Magyar et al. 1999b).

Our Amplocypris material from Hennersdorf (cf. Annex 1, Tab. 1) comes from bulk samples collected in 2006 in a series of layers, labelled in figure 6 on the left of the lithological profile. Additionally 11 short cores of 0.35 to 0.5 m length, sliced in half centimetre layers offered high resolution samples (cf. for their approximate position, Fig. 6). Especially core No. 6 (cf. Annex 1, Tab. 1) was rich in Amplocypris valves, both adult and juvenile specimens.

Fig. 6.

Lithological profile of the Hennersdorf section (Pannonian E), from Harzhauser & Mandic (2004) on which on the left side the bulk samples are indicated and on the right side the relative position of the 11 short cores (cf. text).

The ostracod assemblage at Hennersdorf is quantitatively dominated by Cyprideis obesa (Reuss) and Hemicytheria folliculosa (Reuss). We recognized only one species of Amplocypris that was initially left in open nomenclature (cf. Annex 1, Tab. 2 in Harzhauser et al. 2008). The frequency of this Amplocypris species (as number of valves in the various samples analysed) is low as compared to the former two mentioned species (MINATI, Doctoral Thesis, in preparation).

Figures 7 and 8 display specimens which closely resemble the images of REUSS, therefore we consider that we identified A. abscissa (Reuss) at Hennersdorf. However, we noticed a wide range of morphologic variability, hence at the beginning we will treat this species as A. abscissa (sensu lato). Further analysis (see below) will bring more precision to this preliminary identification, and we will separate one of the morphotypes as A. abscissa (sensu stricto) and two other morphotypes remain to be assigned to other species.

Fig. 7.

Amplocypris abscissa, left (A, B) and right (C, D) valves from Hennersdorf; A, B – specimen HD 7381; C, D – specimen HD 7380; A, C –external view in TLM, B, D internal side, SEM views; scale bars = 0.5 mm.

The acuminate valves of A. abscissa from Hennersdorf display some interesting characteristics, respectively the left valve is higher and slightly more rounded than the right valve (Fig. 7), the dorsal hinge groove is enlarged in both the anterior and posterior third sections and on the ventral side two small pegs are visible (cf. Fig. 7B).

The valves display large anterior and posterior vestibula (cf. Fig. 8A). On the anterior and antero-ventral side one notices (Fig. 8B) dense ramified radial canals and on the whole surface wall a large number of normal pores exist. The posterior vestibule is larger (Fig. 8A) than those of Amplocypris recta (Fig. 4A).

R. Pipík (Banská Bystrica) offered photographs of valves which he sampled at Sopron in 2008. The Sopron sections are dated as Pannonian E stage (Magyar et al. 1999b). A right valve of Amplocypris abscissa offered by PIPÌK will be compared with our material (see below).

Specimens of Amplocypris recta used here were sampled at St. Margarethen in Burgenland at the outcrop “Altes Zollhaus” in the layer 16 dated as Lower Pannonian (cf. Annex 1, Tab. 1; Harzhauser et al. 2002; Buttinger 2008). Additional Amplocypris material for this study comes from Soceni (Banat, Romania), an outcrop in the Turislav valley sampled during 2006 (cf. Annex 1, Tab. 1; Danielopol et al. 2009). R. Olteanu provided to one of us (D.L.D.) Amplocypris material from outcrops at the localities Groşi and Sinteşti (both in Banat, north from Soceni) dated as Early–Middle Pannonian age (Olteanu 1971, 1986). Finally we used for this project also published figures from various authors, especially from Reuss (1850), Zalanyi (1944), Pokorný (1952), Olteanu (1971) and Krstić (1973).

The material presented here is stored at the Universalmuseum Joanneum, Department of Geology & Palaeontology, Graz. The lectotype of A. abscissa is deposited at the Natural History Museum Vienna, Geological–Palaeontological Department.

2. Morphometric approaches for Amplocypris abscissa

2.1. Combining geometric with traditional morphometric techniques for comparative morphology and systematics

In recent years we have assisted in the development of techniques using multivariate statistics to the quantitative description of biological shapes. Reviews of the rapid development of this research direction were published inter alia by Rohlf (1990), Bookstein (1991, 1993), Richtsmeier et al. (1992, 2002), Macleod (1999, 2002), Zelditch et al. (2004), Adams et al. (2004), Sheets et al. (2006), Mitteroecker & Gunz (2009), Strauss (2010). Specific to ostracod research we have to mention the contributions of several distinguished specialists like R.L. Kaesler, R.H. Benson, R.A. Reyment and A. Baltanàs. Some of their results are presented in Benson et al. (1982), Foster & Kaesler (1988), Reyment (1985), Baltanàs & Geiger (1998), Baltanàs et al. (2003), Baltanàs (2008). The recent progress in ostracod research was achieved inter alia by the application of geometric morphometrics to the description of morphological diversity (disparity) of valves belonging to various species. We will mention the classic paper of Reyment & Bookstein (1993) and more recent contributions of Baltanàs et al. (2000), Elewa (2004), Iepure et al. (2007), Minati et al. (2008), Danielopol et al. (2008), Gross et al. (2008). Further information is presented in Baltanàs & Danielopol (2011).

For the systematics of poorly ornamented ostracods geometric morphometric techniques are not generally used despite the fact that algorithms for other groups were developed (cf. Macleod 1999; Zelditch et al. 2004; Sheets et al. 2006; Van Bocxlaer & Schultheiss 2010). In the present contribution we propose a composite method based on the algorithms existing in the computer packages MORPHOMATICA 1.6 (cf. Linhart et al. 2007; Brauneis et al. 2008; Neubauer & Linhart 2008) and PRIMER v6 combined with PERMANOVA+ for PRIMER (Clarke & Gorley 2006; Anderson et al. 2008) which should help the classification of morphotypes and further the improvement of systematics of various ostracod groups.

2.2. Choice of variables and their analysis with multivariate statistics

In a first step we use geometric morphometrics for outlines, based on the Linhart’s B-spline approach in the computer programme MORPHOMATICA 1.6 (cf. the web site “Methods in Ostracodology” at http://palstrat.uni-graz.at).

We produce in the Specimen module an outline based on 12 Control Points (CP) for half of the outline, (respectively, 24 CP for the whole outline) offering a very good fit to the original digitised outline (cf. Fig. 9). The position of these CP will be further used as reference directions for the vectors which will be chosen as morphological traits (see further details and Fig. 13).

Fig. 9.

Reconstruction of the outline of an Amplocypris abscissa valve (cf. right valve reversed of specimen ÖD 7445) using the Linhart’s B-spline algorithm; symbols mark the position of the control points; the cross of the orthogonal axis of inertia marks the position of the centre of gravity; the precision for the fit of the reconstructed line (in red) to the original digitised outline (in black) is left indicated.

Fig. 13.

Diagrammatic representation of an Amplocypris right valve (reversed view) inserted within a bounding box standardised for a length of 200 mm (RL = relative length); the vectors used as linear variables are: H – maximal height. Cv10 – the length (in mm) from the centre of gravity (C) to the outline on the direction of the v10 control point. Similarly to Cv10 are defined also the vectors Cm1, Cd2, Cd9, Cm2, Cv2. all measurements are done with the “tps.Dig” programme for landmarks and saved as M-variables in an additional tps file (see text).

We superimposed, in MORPHOMATICA, with the Cluster module and the standardised option (i.e. normalised for area), the digitised outline of REUSS’ figure (cf. here Fig. 1B) and those of the Hennersdorf specimen No. HD 7316 (cf. Fig. 8A). One should note in figure 10 the close fit between both valves. This is well displayed not only by the differences in the area deviation of the two superimposed valves but also expressed by the short length of the vectors.

Fig. 10.

Amplocypris abscissa, superimposition of outlines for right valves (reversed) standardised for equal surface; 0 – outline of the original Reuss figure (cf. our Fig. 1 B), 1 – outline of the Hennersdorf specimen No. HD 7316 (cf. Fig. 8A); differences in shape expressed through the length of the delta vectors (2.5 x enlarged).

We continued with the analysis of the outlines of Amplocypris material using the non-standardised approach in MORPHOMATICA. We noticed in the Hennersdorf material of Amplocypris that the form of the valves (i.e., their shape and their size) is polymorphic. Figure 11 displays such polymorphism. We compared the form of Amplocypris recta (Fig. 11A), used as a reference shape with those of several representative specimens from Hennersdorf (Fig. 11B–F). By superimposition of these outlines (Fig. 12) the differences in size and shape are clearly visible and can be spatially approximated through the dimension of the vectors at the various control points. One should note on figure 12 the postero-ventral differences around the CP v10 (see also for the location of this control point, Fig. 9). This approach is qualitatively informative during an exploratory analysis but we decided not to use it for further analyses because in cases where the vectors are orientated more-or-less parallel to the outline they do not express true differences between shapes. The problem is explained in Neubauer (2008) and in Neubauer & Linhart (2008). An additional difficulty arises with the necessary usage of a reference outline that cannot be a priori stated for general purposes.

Fig. 12.

Right valves of Amplocypris species (reversed view); superimposition of outlines for right valves showing at 12 control points differences in shape expressed as the size of the delta vectors (1.5 x enlarged); 0 – outline of the reference shape, A. recta, specimen SM16 - 3471, 1- 5 outlines of A. abscissa, specimens B–F in Fig. 11.

We decided in a first approach to use variables measured on a linear scale in millimetre units. The traits chosen as variables were defined after we standardised for length the outline in MORPHOMATICA (see details on the procedure used below). Figure 13 shows the seven linear variables we chose for an exploratory analysis. They were coded as follows: H – the maximal height of the outline measured on the y axis of the boundingbox. Cv10 – the distance between the Centre of gravity C (cf. for its definition Neubauer & Linhart 2008) and the margin of the digitised outline, using the direction of the CP v10. The other variables are defined like the Cv10 and they are named Cm1, Cd2, Cd9, Cm2 and Cv2 (cf. Fig. 13).

The seven chosen variables are measured on the MORPHOMATICA figure using the procedures implemented in the programme tps.Dig (Rohlf 2003). This is explained in the following paragraph.

From the MORPHOMATICA, Specimen module, we use the approximated outline obtained with 12 CP. We open in Specimen, Properties and activate in Graphics, Draw bounding box. We move with the mouse the left bar of the MORPHOMATICA-screen (which separates the outline image from the left side where specimen information is listed) to the right side in order to set the size of the X axis length of the bounding box to 200 mm. Further in the Specimen module we activate Export (the figure) to image (file). This latter action will save the standardised image in a bmp-file. We open the programme tps.Dig, Import source/File open (bmp-file). We continue in the Digitise landmarks mode and we fix the landmarks (cf. the black points in Fig. 13), continue setting the scale, i.e., we use in Options command Set scale fixing to 200 (the length of the X axis of the bounding box). We continue activating the key “Make linear measurements” (the comb key on the screen) and the measured lengths are transferred further with the activated right mouse key “Add as variable” in a tps-file. At the end the variables labelled M1-M7 are saved in a tps-file. This latter will be open and the data transferred in an EXCEL file with variables as columns and samples as rows (cf. Annex 1, Tab. 2). The EXCEL matrix is transferred in PRIMER+ PERMANOVA and with multivariate statistics further analysed (see below). We open in PRIMER, Sample data (samples as rows), continue with Sample matrix, activate Resemblance routine and we chose Euclidean distance. On the obtained pairwise resemblance matrix we open in Analyse the cluster programme Group average (which classifies hierarchically dissimilarities as unweighted pair-groups on arithmetic averages, the so-called UPGMA method).

We use first the data set from Hennersdorf, represented by 89 sampling units for the cluster analysis, without any prior assumption of their degree of similarities. Figure 14 shows the results expressed as a dendrogram where the Y axis shows the degree of dissimilarity of the clusters formed by the groups of sampling units.

Fig. 14.

Dendrogram relating the 89 right valves of A. abscissa from Hennersdorf using the group-average algorithm on Euclidean distances of the seven linear variables.

We select the three primary clusters and we look again in MORPHOMATICA at the outline shapes. The first group contains elements related to what we identified as abscissa morphotype. The next large group splits in two subgroups. The left-side cluster is the most abundant and covers many round-shaped outlines while the right-side cluster contains more or less rectangular shapes. We named those two clusters rotunda and quasi-abscissa morphotype-groups respectively. Figure 15 therefore proposes a first classification using the UPGMA method that we explore further through the Principal Coordinates analysis or PCO (Fig. 16). In this latter diagram one sees that the first two axis of the PCO explain 88.4 % of the total variation (for the rationale of the PCO routine, cf. Anderson et al. 2008).

Fig. 15.

Dendrogram of the 89 Hennersdorf right valves of A. abscissa using the group-average algorithm on Euclidean distances as in Fig. 14 but with the three main clusters defined as abscissa (Ac), quasi-abscissa (QA) and rotunda (R).

Fig. 16.

Principal Coordinate Ordination (PCO) of 89 right valves of A. abscissa from Hennersdorf, grouped in three classes, abscissa (Ac), quasi-abscissa (QA) and rotunda (R) (cf. also Fig. 15).

We need now to confirm the preliminary classification through additional analyses. We replicate the data adding the variables measured on the outlines of the right valves of Reuss (Fig. 1 B), of Papp-Turnovsky (Fig. 2A) and of the lectotype of A. abscissa (Fig. 2C), the reference valve from St. Margarethen identified as Amplocypris recta, on six Amplocypris outlines of the original material of Reuss from Sopron and of the two outlines from Soceni from the Turislav valley (layer 5). The data are presented in Annex 1, Table 2. We further proceed with UPGMA-cluster and PCO analyses on 101 sample units. Figures 17 and 18 display the results.

Fig. 17.

Dendrogram using the group-average algorithm on Euclidean distances of 101 right valves belonging to A. abscissa from Hennersdorf (specimens labelled Ac, QA, R), from Sopron (labelled ÖAc and Ör), Soceni-Turislav valley, layer 5 (TU5), the Reuss original abscissa figure (AbscReuss, cf. Fig. 1B), the Papp-Turnovsky ’s inner view figure (Turn_i, cf. Fig. 2 A), the A. abscissa lectotype (AcLeRVev, cf. Fig. 2C) as well as Amplocypris recta from St. Margarethen (Rect, cf. Fig. 4A).

Fig. 18.

Principal Coordinate Ordination (PCO) of the valves of Amplocypris used for the Cluster Analysis displayed in Fig. 17 grouped in the main morphotype classes. The positions of the Reuss, Papp-Turnovsky and the A. abscissa lectotype valves are added (symbols as in Fig. 17).

Amplocypris recta appears the most derived item while the valves figured by Reuss, Papp-Turnovsky and the Atzgersdorf lectotype cluster with the abscissa-valves of Hennersdorf and three valves from the Reuss sample. Three other valves of the Reuss’ material cluster with the rotunda group. The two valves of Turislav valley are classified as quasi-abscissa shape. Therefore the three morphotype-categories are confirmed and we now continue with another series of analyses comparing these three main morphotypes. We use only 98 samples (i.e., we left out the St. Margarethen, the lectotype, the Reuss and the Papp-Turnovsky outlines) and we proceed with a CAP, or Canonical Analysis of Principal Coordinates (for the rationale and the procedure of this routine, cf. Anderson et al. 2008). Figure 19 displays the data on the two axes of the CAP. This latter analysis and especially a permutation test using only those two axes offers us information on how well the samples were classified.

Fig. 19.

Canonical Ordination of Principal Coordinates (CAP) of 98 valves of A. abscissa grouped within the three main morphotypes abscissa (Ac), quasi-abscissa (QA) and rotunda (R); not used are the data for the variables of Amplocypris recta and those for A. abscissa related to Reuss and Papp-Turnovsky outlines.

The total percentage of correct classified valves is 93.81 %, respectively only six valves from 98 were misclassified. They are: HD 7373, ÖD 7892, HD 8072, HD 8077, HD 8086 and HD 8090. The value of the trace statistics for the permutation test is 1.216 and that of the first squared canonical correlation is 0.67; in both cases the location of the centroids (or means) of the three groups are significantly different (P = 0.001). The misclassified shapes all originate from the rotunda group (6 from 75 samples) while the other two groups were found as 100 % well sorted. We accept the suggestion of the randomised test to change the position of the six misclassified items from the rotunda category to the quasi-abscissa. However, the present results do not answer one of the main questions of our project, namely if using quantitative methods one is able to objectively differentiate the polymorphic populations of the so-called taxon Amplocypris abscissa (sensu lato), especially the samples identified as rotunda and quasi-abscissa. Therefore we continue with the search for additional variables (see below).

2.3. The composite morphometric approach

The term composite is used here to indicate that we combine the data of variables (the previous seven ones and two additional ones) obtained from measurements on different scales, respectively on an interval scale with linear and angular values and a ratio scale. Below, we present two new variables which can be accurately evaluated. Figure 20 shows how one can measure using a graphic computer programme (e.g., we used ADOBE-PHOTOSHOP) the ratio between “h”, the lower height section (starting from the bottom to the tangential point of the posterior outline section) and “H”, the maximum height of the valve.

Fig. 20.

The pi index (h/H); h – height to the posterior tangential point, H – maximum height (cf. text).

We called this variable the “pi index” and expressed it as ratio in percentage. The next variable that we found useful is the angle formed by the lines relating the control point v10 to the adjacent ones (cf. Fig. 9). We measured it on the bounding-box figures standardised to a given length (i.e., to 200 mm length). The couple of variables Cv10, pi and angle v10 describe in a very useful way the shape of the distal section of the postero-ventral part of the outline (cf. Annex 1, Tab. 1–2). We measured the pi and the angle v10 on the 97 Amplocypris valves and continued the multivariate analysis using 9 variables, using PRIMER algorithms. For this purpose it is necessary to normalise the data; this means the values measured on different scales have to be brought to a comparable scale (Clarke & Gorley 2006). The normalisation procedure applies to PERMANOVA algorithms too (Anderson et al. 2008).

We next create in PRIMER a new matrix with nine columns (the variables) and 97 rows (the valves as sample units) that we further normalise by activating in the module “analyse-pretreatment” as mentioned above. We continue and transform this latter matrix in a new resemblance matrix, i.e., a pairwise Euclidean distance of dissimilarities. This latter will be used for further exploratory and confirmatory analyses of the morphometric data.

In the previous section the results of the CAP analysis pointed out six valves which were misclassified. As suggested by the CAP-results we changed their position from the rotunda to the quasi-abscissa group. With the new resemblance matrix we perform a CAP analysis (Fig. 21).

Fig. 21.

Canonical Ordination of Principal Coordinates (CAP) of the 97 valves of Amplocypris produced with nine variables of the right valves belonging to A. abscissa from Hennersdorf (labelled Ac, QA, R), from Sopron (ÖAc and Ör), Soceni-Turislav valley, layer 5 (TU5). To the seven scalar variables already used (cf. Tab. 2) were added the values of the angle of the control point v10 (Ang v10) as well as the pi index (cf. Tab. 3 and text). Note the position of sample ÖD 7899 from Sopron now evaluated as a quasi-abscissa morphotype and the Soceni samples located at the periphery of the sampling cloud quasi-abscissa.

This shows that the three main groups are largely preserved with one additional change the valve ÖD 7899 (cf. Fig. 24D) moved from rotunda to the quasi-abscissa group. This appears to be a consequence of the role played by the new variables which better characterise the postero-ventral section of the outline, respectively the margin is slightly straighter and angular than most rotunda specimens. The previous data set was further analysed with the PCO-procedure, using this time only three categories, the abscissa, rotunda and quasi-abscissa morphotypes (Fig. 22).

Fig. 24.

Shape differences between the reversed right valves HD 8090 (A) and HD 8062 (B); C – superimposition in the standardised for surface modus of the outlines HD 8062 (1) on the outline HD 8090 (0); D – the right valve ÖD 7899 (reversed view).

Fig. 22.

Principal Coordinate Ordination (PCO) of the 97 valves of Amplocypris used for the PCO with 9 variables (cf. Fig. 21), grouped within three main morphotypes abscissa (Ac), quasi-abscissa (QA) and rotunda (R).

The first two axes explain 80.5 % of the variability. The abscissa cloud of points is well separated from the two other groups. For these latter groups one notices a poorer delineation. However, even here we suspect a statistically significant separation between the centroids of the sample groups. See Anderson et al. (2008) for the way the centroid of the dispersion of the sampling points in a 2D-morphospace is calculated using as a measure of resemblance the D1 Euclidean distance.

Therefore we continue with a confirmatory CAP analysis (Fig. 23) using the first two axes (m - 2) like in the PCO. The first squared canonical correlation axis of the CAP represents 0.78 while the 2^nd one 0.60. The leave-one-out allocation of observations to groups for the choice of m - 2 shows 100 % correct classification of the samples in the groups abscissa, and quasi-abscissa and 98.53 % in the rotunda group, and in the latter from 67 valves only one item (the HD 8062 valve – on Fig 23 given as 8062 and 8062-R) was considered misclassified. The output of this analysis suggests moving this item (HD 8062) from the rotunda to the quasi-abscissa group. The results of a permutation test in CAP shows that the position of the centroids of the three clouds of sample points in the morphospace are significantly separated (P = 0.001). The trace statistics value is 1.382 (cf. information for the interest of this statistics Anderson et al. 2008).

Fig. 23.

Canonical Ordination of Principal Coordinates (CAP) for the data of the PCO analysis displayed in Fig. 22. Note the close position of the two valves 8090-QA and 8062-R, this latter considered misclassified (cf. text).

The shape differences between the valves HD 8062 and HD 8090 were further examined. Their close morphological similarity appears by comparing the figures 24 A and B and by the geometric superimposition of their outlines using the standardised sub-routine of MORPHOMATICA (cf. Fig 24C).

We were interested to see if we added a new variable we could attain a 100 % success in the statistical separation of the three morphotypes abscissa, rotunda and quasi-abscissa. Generally, for the standard description of ostracod taxa one uses also the length of adult valves. Therefore we added length to the 9 variables which characterise the valve shape (cf. Annex 1, Table 3). We now compare the “form” of the Amplocypris abscissa (sensu lato) using 10 variables. We proceeded with a new CAP for these10 variables using only two axes (m - 2). Figure 25 shows that the valves HD 8062 and HD 8090 previously poorly discriminated are now well separated.

Fig. 25.

Canonical Ordination of Principal Coordinates (CAP) of A. abscissa for the data set of 97 sampling units, grouped within the main morphotypes, abscissa (Ac), quasi-abscissa (QA) and rotunda (R) and 10 variables; the Turislav valves (TU5) are showed separately (cf. text). Note the special position in the morphospace of the samples HD 8062 and HD 8090.

This is explained by the length difference between the two valves (cf. Annex 1, Table 3), respectively 1.33 mm for HD 8090 and 1.18 mm length in the case of HD 8062. We noticed also that because the size of the two Turislav valves are very large, respectively 1.78 and 1.67 mm length (cf. Annex 1, Tab. 3), their position moved in the morphological space (cf. Fig. 25) to the left side.

In order to confirm the importance of size for the position of the valves in the morphological space we simulated the values of the Turislav valves reducing their length to 1.4 mm and 1.3 mm. Figure 26 shows that Turislav items (green ) are now placed closer to the Hennersdorf valves (blue ). The importance of this variable for the separation of the three groups of morphotypes is clear also when one uses the SIMPER routine in PRIMER. This latter procedure calculates the contribution of the variables to the similarity degree of the investigated groups (Clarke & Warwick 2001; Clarke & Gorley 2006). This type of exercise, done also with simulation for other traits, allows us to better understand the contribution of a given variable to the morphological separation of the morphotypes in a multidimensional space.

Fig. 26.

Canonical Ordination of Principal Coordinates (CAP) of A. abscissa for the data set of 97 sampling units, grouped within the main morphotypes, abscissa (Ac), quasi-abscissa (QA) and rotunda (R) and 10 variables; the position of the Turislav valves (TU5) in the morphospace is due to the simulated length variable (cf. text).

In the following, we show an example of simulation which helps to understand the separation between morphotypes using the variable Cv10. We took the outline of the valve HD 7354 which belongs to abscissa morphotype and we compared it with rotunda-shape valve HD 7370 (Fig. 27A–B).

Fig. 27.

A–D - Manipulation of shape for the reversed right valve HD 7370; A – the valve HD 7354 abscissa-type; B – the valve HD 7370 rotunda-type; C – the change of outline shape of HD 7370 through the simulated Cv10 vector’s length (cf. text); D – the new virtual HD 7370 as abscissa-type; E–F – right valves from Sopron (reversed view); E – specimen Pip_ÖD offered by R. Pipík; F – specimen ÖD 7445, coll. A.E. Reuss.

The vector Cv10 of the former valve is longer than of the rotunda one (cf. Annex 1, Tab. 2). We approximated in MORPHOMATICA their outlines using the command for 12 control points on the dorsal, respectively ventral side. Further, we changed the coordinates of the control point v10 of the valve HD 7370 with those of HD 7354 activating the key for interactive play with the control point coordinates (cf. Brauneis et al. 2008). The result of this simulation is in Fig. 27 C. The new outline was further processed with the tps.Dig and MORPHOMATICA programmes and we got a new virtual valve (Fig. 27D). This latter was further inserted into our data matrix for 9 variables and submitted to a PCO analysis. The results are in Fig. 28, pointing out the important role of the variable Cv10 for the characterisation of Amplocypris outlines.

Fig. 28.

Principal Coordinate Ordination (PCO) using 9 variables for the 97 valves of Amplocypris used in Fig. 22, the Pipík specimen (Pip_ÖD) and the virtual abscissa-type HD 7370_01.

In order to verify the value of our new method to describe morphotypes with the algorithms offered here we considered it necessary to replicate our data. Here we show (cf. Fig. 28) that a right valve of Amplocypris abscissa from Sopron sampled in 2008 by R. PipÌk (Pip_ÖD in Fig. 27E), because of its similar shape with the Reuss’ item ÖD 7445 (Fig. 27F), when inserted in our data matrix for 9 variables and further analysed with the PCO algorithm, both occupy the same position in the morphological space. Note that the REUSS’ valve was assigned to the abscissa morphotype. Through superimposition of the two outlines obtained with the standardising for surface routine in MORPHOMATICA (Fig. 29) one can see again their close shape resemblance. We conclude that this new approach for morphological characterisation of valves with few visible character traits offers meaningful results. It gives us the impetus to continue our project in order to define the morphotypes as categorical entities.

Fig. 29.

Superimposition in standardised for surface mode of the Sopron valves ÖD 7445, Reuss’ material (0) and Pipìk’s specimen Pip_ÖD (1).

3. Diagnosis of morphotype profiles and possible applications of the morphometric approach to the systematics of the genus Amplocypris

In the previous sections we differentiated with multivariate statistical procedures within Amplocypris abscissa (sensu lato) three morphotypes that we named abscissa, quasi-abscissa and rotunda. If one intends to use these categorical entities for the systematics of Amplocypris than diagnoses are required as stated in Article 13.1 of the International Code of Zoological Nomenclature (ICZN 2000). A diagnosis following this Code is “a statement in words that purports to give those characters which differentiate the taxon from other taxa which is likely to be confused” (ICZN 2000: 103).

In the following we propose differential diagnoses using information in Table 4 (Annex 1) and in the figures accompanying this text. Table 4 shows that the three morphotypes are characterised as groups by their arithmetic mean with the 99 % confidence limits i (± CL). These latter were obtained by the bootstrap technique (which avoids parametric assumptions) implemented in the computer programme annexed to the text-book Ecological Methodology (Krebs 1999). The usage of confidence levels allows us to see how well the sample group is expected to reflect the population from which it came (Sokal & Rohlf 1995). The minimum-maximum range of each variable for the three morphotypes is also mentioned in order to show that they display slightly overlapping values between populations.

For a synthetic characterization of differences between sets of multivariate data Manly (1998) offers inter alia as a solution for the presentation of differential diagnoses, graphical profiles where the mean values for the various variables are linearly presented. Each taxon is represented by a specific trajectory.

In the present contribution we define each morphotype by a selection of traits with their statistical peculiarities which at the end produces a specific profile, in essence a general image which applies to the group.

Gould (1996) criticised the usage of the average value (the central tendency of a sample) when dealing with evolutionary trends. GOULD considered that an average value as a statistic is a mental construct which does not occur under natural conditions. Hence, he proposed to use individual variables originating from concrete entities which characterise them directly. In the following we show that this criticism can be avoided.

We produced for each morphotype a mean virtual shape using the non-standardised command in MORPHOMATICA (Fig. 30). We also looked for each morphotype in the pairwise normalised resemblance matrix of PRIMER for the case of 9 variables (i.e., the length variable not considered) and we extracted the mean pairwise Euclidean distance relating two individuals (the procedure will be demonstrated during the Workshop MIO-2, July 2011). We noticed (see below) that the outlines of these pair items resemble closely the virtual mean outlines of these morphotypes.

Fig. 30.

The virtual mean outlines of the valves belonging to the three morphotypes of A. abscissa: A – abscissa (Ac), B – quasi-abscissa (QA), C – rotunda (R) produced with the non-standardised for surface algorithm in MORPHOMATICA.

Figure 31 show the couple of the two quasi-abscissa (Fig. 31A–B) and rotunda (Fig. 31C–D) morphotypes. Further, we superimposed the outline of one of these individuals on the mean virtual outline using the standardised command and the similarity between the shapes of simulated case and the real individual appears (cf. Fig. 32 A–C). Therefore we consider that the profiles we will present below have an empirical meaning. Finally we consider that the diagnosis using both descriptive words and references to quantitative information of variables is a more informative and objective presentation of morphotypes or taxa than the traditional presentations in previous publications (cf. inter alia Krstić 1973; Jiriček 1985). In the following we will offer for the Amplocypris morphotypes differential diagnoses in a sequential way and for additional information on the Amplocypris shapes, see Annex 2.

Fig. 31.

The couple of the quasi-abscissa (A – HD 3581 and B – HD 8079) and rotunda valves (C – HD 8050 and D – HD 7339).

Fig. 32.

A–C – Superimposition on the mean virtual shape (in blue) the outline of closely similar individuals using the standardised for surface mode in MORPHOMATICA; A – the abscissa mean-shape and HD 7316; B – quasi-abscissa mean-shape and HD 3581; C – rotunda mean-shape and HD 8050; D – superimposition on quasi-abscissa HD 3581 (blue line) of the Soceni valve TU5 - 3592.

We start with Amplocypris recta, which was chosen as reference system for A. abscissa (sensu lato). The right valve of A. recta appears flat and elongated (Fig. 4A, 11A). The standardised H value is the lowest as compared to those of A. abscissa (respectively 46 % of the valve length; cf. Annex 1, Tab. 4); the postero-ventral outline section is essentially rounded. This wide postero-ventral curvature is well expressed through the low value of the length of the vector Cv10 and the high values of the angle V10 and the pi index (Annex 1, Tab. 4); it differs markedly from the other categories here discussed.

The morphotype abscissa (Figs 2, 3A–D, 7A–D, 8A, 9, 10, 11B–F, 27A, E–F, 30A, 32A; Annex 1, Tab. 4) as compared to the quasi-abscissa, can be characterised by an elongated shape with a low H value (about 48 % of the length), with an oblique postero-dorsal section of the outline (reduced length of Cm1 and Cd2 vectors) and prominent acuminate postero-ventral outline (low values of the angle V10 and pi). These latter characteristics, even if more variable, are less well expressed in the group quasi-abscissa. Clear differences exist between the abscissa and the rotunda morphotypes when one compares the posterior shape of the outline. The latter group has higher values for Ang V10 and pi. The H value of the rotunda valves is slightly higher while the length is shorter than the abscissa morphotype.

The quasi-abscissa morphotype (Fig. 11D, 30B, 31A–B, 32B & D; Annex 1, Tab. 4) has a shape which is characterised by high values for H (more than 50 % of the length), the dorso-posterior margin is generally straight and the postero-ventral section form an angle V10, frequently less than 100°. The general aspect of the shape and also the size display an intermediary position between the morphotypes abscissa and rotunda.

Finally, the rotunda morphotype (Fig. 11B & E, 14, 27B, 30C, 31C & D, 32C; Annex 1, Tab. 4) can be identified by its rounded anterior and posterior margins and by its smaller size (about 1.3 mm mean length of the RV).

The aim of the composite algorithm we developed was from the very beginning intended to help clarify the systematics of ostracod groups, especially where only poorly ornamented valves exist. For the Amplocypris abscissa case, here discussed, we are confronted with two alternatives as mentioned in the introduction: either one deals with a widely polymorphic species, which could be called A. abscissa (sensu lato) or under this latter name one can delineate several distinct taxa and in this case we consider the existence of a phylogenetic lineage within which we have a cluster of satellite species, where A. abscissa is one of these taxa. Because the three morphotypes are statistically delineated using multidimensional criteria, we favour this latter alternative. Our decision is based inter alia on the arguments of Sbordoni (1993) who points out the advantages for the degree of species separation using a composite approach. In this latter case the various types of variables are projected within a multidimensional space and analysed with multivariate statistical procedures, species are then visualised in a hyperspace. We favour the bi-dimensional space as a heuristic explanatory framework.

The morphotype abscissa (cf. its diagnosis) defines in our opinion the species Amplocypris abscissa (sensu stricto) with elements which already exist in the original description of Reuss (1850) as here documented. Therefore we have to ask what is the significance of the other two morphotypes? The rotunda form appears very similar to the shape of A. pavlovici KRSTIĆ, as figured in Krstić (1973). However, a clear assignment of the rotunda morphotype to A. pavlovici is uncertain for the moment as we do not have more information on the morphology and variability of this latter species.

The morphotype quasi-abscissa could represent another species closely related to A. abscissa (sensu stricto), taking in consideration the similarity between the shapes of the valves from Hennersdorf (on the western side of the Lake Pannon) assigned to this morphotype and the valves from Soceni (Turislav valley, layer 5) in the Romanian Banat (on the eastern side of the lake). However, we have to note that the length of the Soceni valves is more important than for those of Hennersdorf. For the moment we do not know if the large size of these valves is genetically cued or they are ecophenotypes. Therefore even if the shape of Amplocypris from Soceni appears to belong to the quasi-abscissa morphotype (cf. Fig. 32D) it is necessary to continue investigations using a larger sample size in order to test for a possible vicariant separation of quasi-abscissa-like populations from the western and the eastern sites of the Lake Pannon, reflecting possibly different ages and slightly different ecological conditions.

The next question one should ask is what is the palaeo-biogeographical distribution of A. abscissa (sensu stricto), as defined above? Taking into consideration information from several localities of the eastern side of Lake Pannon in the Romanian Banat (Groşi, Sinteşti and Soceni) where Olteanu (1971, 1986) mentioned A. abscissa and A. recta, we can make the following observations:

At Groşi, we identified the morphotype rotunda but not the morphotype abscissa (sensu stricto), as here defined. At Sinteşti, 6 km south from Groşi, we identified Amplocypris specimens where the right valve is much more acuminate than the A. abscissa from Hennersdorf. The left valves of this extremely acuminate morphotype from Sinteşti resemble the A. abscissa figured in Olteanu (1971). In our opinion the acuminate morphotype from Sinteşti (which occurs also at Groşi) belongs to a different Amplocypris species of the lineage A. abscissa.

At Soceni Mehes (1907) mentioned H. strigata, a species which Pokorný (1952) identified with A. recta. We also sampled at Soceni (2006) ostracods resembling A. recta. However, our material displays slight differences as compared to the A. recta from St. Margarethen, in Burgenland. Hence, it is possible that we deal with closely related species, but more material is needed to solve this open question. Additionally one should consider the exact stratigraphical occurrence of these morphotypes. Both localities, St. Margarethen and Soceni, are dated as Pannonian zone D (our data) and are thus slightly older compared to the Hennersdorf section from where no morphotypes similar to A. recta were detected during this project. This points to an underrated biostratigraphical value for ostracods in lake Pannon deposits.

In the case of the abscissa-morphotypes, here characterised, we noticed differences in their quantitative distribution and their degree of morphological disparity. At Hennersdorf, a site dated as Middle Pannonian age, the morphotype abscissa is rarer as compared to the rotunda one. K. Turnovsky in his unpublished 1949 report (quoted above) mentions that strongly acuminate individuals are more common at the sites dated as Early Pannonian than at Middle Pannonian ones. We hypothesise that the three morphotypes here identified as possible species differ not only by their morphological peculiarities but also by biological characteristics and/or ecological preferences. Therefore, the data presented here point to the idea that A. abscissa is a phylogenetic lineage represented by a cluster of species with more-or-less restricted geographical distributions and differentiated ecological preferences.

It is worth mentioning that the proposal to treat the morphotypes here delineated as a cluster of evolving species belonging to a phylogenetic lineage is very similar to the way Faustova et al. (2010) interpreted the sympatric morphotypes of several populations of the cladoceran group Eubosmina from North European lakes. In the case of the Holocene-Recent Eubosmina group the defined morphotypes delineated with geometric morphometric techniques represent a cluster of evolving species with apparent reproductive barriers and slightly different ecological preferences, despite their coexistence in the investigated lakes.

4. Final notes

The way one can describe and further interpret differences between ostracod valves with few visible morphological traits will certainly surprise ostracodologists. We consider that our approach, even if lengthy, is useful because the computation and interpretation, being recurrent and advancing step-by-step, make it possible to understand and verify what is done and the meaning of the output of the computation performed by the various algorithms. In this way we try to construct an objective view on the morphological differences within and among Amplocypris abscissa populations. This approach can be used for other projects related to the systematics, palaeoecology and/or evolution of Ostracoda, for example the mainly smooth metacopes. Our data here presented are accessible for critical review and can be improved at any time, this reflecting our philosophical view of what Objectivity is and how it should be further used.

We already have ideas about how to advance the refinement of Amplocypris discrimination of morphotypes and/or taxa. It is necessary to describe the outlines of both valves, as in some species the left and right valves display a slight asymmetry. Additionally, the details of the inner lamella have to be documented even if these structures are not preserved in many specimens. Despite these limitations and maybe also because our treatment of this complicated project is incomplete, we suggest that our colleagues try the recipes and the algorithms we offer here.

We conclude this digression within the field of descriptive morphology and systematics with a short statement made by a well-known Austrian biologist, G. Schatz, who started his career at the University of Graz and ended it as President of the Swiss Science and Technology Council: “Scientific exploration is an expedition into the unknown – just like good art. The greater a scientific or artistic innovation, the more it surprises us.” (cf. Schatz 2006: 113). We hope that for the near future the developments presented here will awake the interest of a wider spectrum of ostracodologists for geometric morphometrics, an exciting domain of research.