Diversity of Late Cenozoic Actinopterygian Assemblages of the South of Eastern Europe

Abstract

In the late Cenozoic, the south-west of Eastern Europe was a region affected by extensive hydrological transformations that resulted in the retreat of the Eastern Paratethys and the emergence and further evolution of freshwater communities. In recent decades, a relatively rich fossil actinopterygian fauna has been described from this area. The present study was based on previous systematic studies and aimed to assess and trace the temporal dynamics of the diversity of fish assemblages that existed in the area from the Late Miocene until the end of the Pleistocene. Species diversity, taxonomic diversity, taxonomic complexity, and functional diversity were analysed. It was found that the diversity of the fish assemblages notably decreased during the Late Miocene, when representatives of the families Clariidae, Moronidae, Sciaenidae, and Gobiidae disappeared, and remained relatively low during the Pliocene. During the Pleistocene, however, functional diversity gradually increased, despite fluctuating species and taxonomic diversity and taxonomic richness and complexity, which suggests an increasing stability of the coenotic structure within the fish communities. The revealed temporal trends reflect the impact of the palaeoenvironmental and palaeoecological processes that characterised the region during the late Cenozoic, particularly orogenic and climatic changes, and the evolution of a typical limnophilous, lacustrine-riverine fish fauna.

1. Introduction

In the late Cenozoic, the south-western part of Eastern Europe was a region adjacent to the Eastern Paratethys, and, due to its location, it had been affected by its dynamics. The Paratethys, a large shallow inland sea that stretched from Central Europe to the Aral Sea in Western Asia, appeared at the beginning of the Oligocene (ca. 34 Mya) after separating from the Tethys Ocean [1,2]. This sea or mega-lake had periodically reconnected with the Mediterranean, which allowed for faunal exchange between the two basins [3,4,5].

The eastern part of the Paratethys was the largest, represented by the Sarmatian Sea, which existed 14–10 Mya and included the basins of the modern Black, Azov, Caspian, and Aral seas. The Sarmatian Sea became isolated and progressively desalinated [1,6,7] due to the increasing inflow of freshwaters from numerous rivers. About 10 Mya, it reconnected with the World Ocean, becoming the Maeotian Sea, which included the Black Sea and the Caspian Sea linked by the North Caucasian Seaway. The next stage in the development of the Eastern Paratethys was the formation of the desalinated Pontian Sea, which existed 8–4 Mya and was populated by freshwater fauna [1,6,7]. Beginning from the Pliocene, the area of the Eastern Paratethys gradually decreased, a process accompanied with repeated transgressions and regressions within the basin [1,2,6,7,8,9]. As a result of these basin-wide transformations, thick layers of sediments yielding numerous animal fossils have accumulated in the south of Eastern Europe [10].

The history of the formation of terrestrial vertebrate faunas of the region, including amphibians [11], reptiles [11,12,13,14], birds [15,16], and mammals [17,18,19,20,21,22,23,24,25,26,27] has been studied in detail, but the taxonomic composition of fishes abundantly represented in materials from the late Cenozoic localities of the region have only been studied relatively recently [28,29,30,31,32,33,34,35,36].

In addition to systematic studies, analysis of palaeodiversity is of key importance for tracing the evolutionary development of fish faunas and their possible paths of transformation in the near future, as well as for broader palaeoenvironmental and palaeoecological reconstructions. Biological diversity is an important ecological resource, which assembles, develops, and transforms over geological time. It is a source of stability, having a buffer role in the biosphere by combating the negative impact of various environmental factors [37].

The aim of this study is to analyse various facets of the diversity of the late Cenozoic actinopterygian assemblages of the south of Eastern Europe, including the species, taxonomic, and functional diversity from the Late Miocene until the end of the Pleistocene, and to trace their temporal dynamics. The diversity indices considered characterise the structure of coenotically interrelated functional groups of species and could reflect the carrying capacity and realised niche of the palaeocommunities that existed in the region during different geological stages.

2. Materials and Methods

2.1. Study Material and Geological Background

This study covers a timespan from the Late Miocene (ca. 11.6 Ma) to the Late Pleistocene (0.012 Mya) and is based on the results of a previous systematic palaeontological analysis of osteological materials coming from the alluvial and marine coastal deposits of the northern and north-western Black Sea region and the northern Azov region of Ukraine, the Lower Dniester region of Moldova, and the Lower Don region of Russia [28].

The fish remains are represented mostly by isolated skeletal elements (cranial bones, pharyngeal teeth, fin rays, and vertebrae), as well as, to a lesser extent, by skeletal imprints, otoliths, and scales, which were recovered from the fossiliferous horizons of 53 localities (Figure 1). The fossils were collected by the screen washing (1 mm mesh size) of deposits extracted from the respective fossiliferous horizons and amassed mostly during expeditions of the staff affiliated with the Department of Palaeontology of the National Museum of Natural History, National Academy of Sciences of Ukraine in 1950–2017.

In total, 11,987 specimens were identified at the species or genus level [28]. The corresponding taxonomic lists were compiled (Table S1) for further analysis of eight stratigraphic stages (ages), the order and duration of which follow the regional biostratigraphic schemes of the Neogene and Pleistocene of the south of Eastern Europe [38,39,40,41,42,43]. These ages are the late Sarmatian sensu lato (hereafter—Sarmatian), Maeotian, and Pontian of the Late Miocene; the Kimmerian of the Pliocene; the Kuyalnikian of the Late Pliocene–Early Pleistocene; and the Calabrian, Chibanian, and Tarantian of the Middle–Late Pleistocene (Figure 2). The systematics and taxonomy of the fishes used in this study follow Nelson et al. [44]. Species names given in quotation marks indicate that those species are currently considered to be valid taxa (and therefore included into the analysis), but that their taxonomic status will most likely be revised.

2.2. Estimating Palaeodiversity

The number of identified specimens (NISP) and presence/absence data from the obtained taxonomic lists were used to calculate and analyse the diversity indices of the fish assemblages for each of the eight geological stages considered. The basic operational units in these analyses were the taxa identified at the genus and species levels, whereas the specimens described in open nomenclature at family and subfamily levels were not considered. When calculating the species diversity, taxonomic diversity, taxonomic complexity, and similarity indices, species-level units identified as ‘cf.’ and ‘sp.’ were conventionally considered as independent species.

A faunal assemblage was defined as a historically formed group of species with distinct spatiotemporal features characteristic of the respective zoogeographic unit and with a certain level of evolutionary development of the group in general and of separate lineages [45]. Faunal assemblages reflect the stages of palaeogeographic transformations and are qualitative indicators of geochronological boundaries in defining biostratigraphic units [10].

Species diversity (H_sp) was calculated using the unbiased version of the Shannon index [46] based on the numbers of identified specimens:

$$H_{sp}=\left(-\sum p_i{\ln p}_i\right)-\frac{S-1}{2n}$$

where p_i is the proportion of species i, S is species richness, and n is the sample size.

Taxonomic diversity (H_tax) was considered here, based on Emelyanov et al. [37], as an index that reflects the structure of taxonomic relationships within a community. The index was calculated similarly to Shannon’s species diversity but was based on the sum and proportions of taxa of various ranks (from species to order) present in a given community or assemblage. This approach does not need the consideration of relative abundances and thus can be applied effectively to taxonomic lists containing presence/absence data. Similarly to species diversity, the unbiased version of the index was calculated as follows:

$$H_{tax}=\left(-\sum p_i{\ln p}_i\right)-\frac{T-1}{2n}$$

where p_i is the proportion of taxonomic rank i (species, genus, family, and order), T is taxonomic richness (sum of all taxonomic ranks), and n is the sample size.

Taxonomic complexity (C) is a diversity index also proposed by Emelyanov et al. [37] that combines species diversity and taxonomic diversity in order to estimate the complexity of the structural organisation of a given community. Taxonomic complexity is a function that combines taxonomic diversity (H_tax) and “hierarchical” diversity [47,48]. The latter is understood as the diversity of different ranks (species, genera, families, and orders). Thus, taxonomic complexity was calculated as

$$C=\sqrt{H_{tax}\times \frac{1}{N}\times \sum H_i}$$

where H_tax is taxonomic diversity, H_i is the hierarchical diversity of taxonomic rank i (i.e., ln i), and N is the number of taxonomic levels considered.

The Jaccard (J) and Sørensen–Dice indices (SD) were used to analyse the similarity between the species composition of fish assemblages of various geological stages according to the formulas $J=\frac{100\times C}{A+B-C}$ and $SD=\frac{100\times 2C}{A+B}$, respectively, where A and B are the numbers of species in the two assemblages compared, and C is the number of common species.

Functional diversity (FD) was estimated using the dendrogram-based model [49,50], which allows for visualising functional groups of species within the assemblage. In addition, this approach is insensitive to species splitting [51], which could be useful in analysing palaeontological datasets containing taxa identified as ‘cf.’ and ‘sp.’ Eight functional components were analysed that described the biological, ecological, and anatomical features of each species, and which were related to various niche dimensions (Tables S2 and S3). These features were analysed according to data from the Handbook of European Freshwater Fishes [52], and those characteristic for the adult stages of individual development were considered.

When assessing extinct species, the functional traits of their closest extant relatives were considered [52,53]. This approach is widely used in palaeoecological studies, although it gives a somewhat simplified picture of the state of fish palaeocommunities (not all their representatives are equally or at all represented in the fossil record). In cases of several extinct species in the same period, the traits of the morphologically more similar close species were used (e.g., Rutilus robustus is morphologically closer to R. frisii, while Rutilus tungurukensis is closer to R. rutilus).

In cases of taxa identified as ‘cf.’, the traits of the species the specimens were compared to were considered. In cases of genus-level identifications (sp.), three approaches of character coding were applied. When the taxon ‘sp.’ was the only representative of the genus or when there was another species of the genus in another period, the traits of the type species of the genus were considered, which is usually a widely distributed and an abundant species. When the taxon ‘sp.’ co-occurred with another species of the genus in the same period, the traits of the identified species were considered. Lastly, when the taxon ‘sp.’ co-occurred with several species of the genus, the traits of the morphologically closer, identified species were considered.

2.3. Data Analysis

Data analysis and visualisation were carried out in PAST 4.17 [54] and Microsoft Excel 2019. Bootstrapping (N = 9999) and the diversity t-test in PAST 4.17 were applied for statistical estimates of the obtained values of species and taxonomic diversities. Non-parametric Spearman’s rank-order correlation (r_s) analysis was used to explore dependencies between sample size (number of identified specimens) and the indices of species diversity, taxonomic diversity, and taxonomic complexity.

Gower’s index was used to build the distance matrices and was used in the hierarchical cluster analyses carried out with the UPGMA (unweighted pair group method with arithmetic mean) algorithm. Bootstrapping (N = 9999) was applied to obtain measuring estimates. The nexus files of the hierarchical cluster analysis used in the calculations of dendrogram lengths are presented in Table S4.

3. Results

3.1. Species and Taxonomic Diversity and Taxonomic Complexity

The complete list of fishes of the Late Miocene–Pleistocene of the south of Eastern Europe comprises 63 species-level taxa (i.e., including ‘cf.’ and ‘sp.’), representing 34 genera, 12 families, and 10 orders. Taxonomic richness (T) is the highest during the Sarmatian (Figure 3a), when the fish assemblage comprises thirty-eight species, representing twenty-five genera, nine families, and eight orders. In the Maeotian, both species richness and taxonomic richness already notably drop and by the Tarantian remain practically on the same level, with only a slight decrease in the Kuyalnikian (Table S5). Taxonomic diversity (H_tax), however, shows a slightly increasing trend, although its values change in a narrow range (1.17–1.26, Δ = 0.09), and the differences between the various stages are insignificant (p > 0.05, Table S6).

In contrast, species diversity (H_sp) tends to generally decrease from the Sarmatian to the Tarantian; it fluctuates in a wider range of values (1.97–2.75, Δ = 0.78), and there are significant differences between particular stages (p < 0.05, Table S6). Similarly to taxonomic richness, species diversity declines from the Sarmatian to the Maeotian, then recovers in the Pontian, followed by a further decrease until the Chibanian, when it recovers again (Figure 3b). Accordingly, only the differences between stages having similar diversity values are insignificant (p > 0.05), as in the case of the Sarmatian and Pontian; Maeotian, Kuyalnikian, and Chibanian; Kimmerian, Kuyalnikian, and Tarantian; and Kuyalnikian and Tarantian (Table S6).

No significant correlations (p < 0.05) have been found between the number of specimens (NISP) and the analysed diversity indices, which allows us to assume that the latter are not biassed by sample size. The temporal trends in diversity indices are not linear (R² = 0.15–0.45), though they reflect faunal dynamics accompanied with changes in the entire taxonomic structure of fish assemblages, not only on the species level but also on the level of higher taxonomic ranks. The index of taxonomic complexity (C) follows a similar trend to species diversity from the Sarmatian to the Kuyalnikian, but they have contrasting dynamics during the Pleistocene (Figure 3c). Although the values of this index also vary within narrow limits (1.67–1.84, Δ = 0.17) and no significant differences have been revealed between the taxonomic complexity indices of various stages, we can see that a decrease in species diversity is accompanied with an increase in taxonomic complexity, which can be explained by the contrasting dynamics of hierarchical diversity—hierarchical species diversity decreases, generic diversity increases, while family- and order-level diversities fluctuate (Table S7).

This implies that a notable rearrangement took place in the taxonomic structure of Pleistocene fish assemblages within the southern part of Eastern Europe, most likely triggered and influenced by palaeogeographic and hydrological rearrangements due to the retreat of the environments of the Euxinian basin [7]. Taxonomic complexity and taxonomic diversity show contrasting general trends but basically have the same dynamic during the entire timespan considered, with only a slight contrast in the Calabrian, when taxonomic diversity decreases, but taxonomic complexity remains almost on the same level (Figure 3d). The taxonomically most complex is the Sarmatian fish assemblage, whereas the Tarantian one is characterised by high values of both taxonomic diversity and complexity.

When analysing the qualitative changes in the taxonomic composition of the fish assemblages, the highest level of similarity has been revealed between the Chibanian and Tarantian stages (J = 72.4, SD = 84.0), suggesting little changes in species composition from the Middle Pleistocene. The second highest degree of similarity characterises the fish assemblages of the Kimmerian and Kuyalnikian stages (J = 66.7, SD = 80), indicating a relatively stable species composition of the Pliocene fish fauna. As for the Late Miocene, the fish assemblages of the Sarmatian and Maeotian stages have the third highest similarity indices (J = 61.4, SD = 76.1), whereas the Pontian assemblage is far less similar to that of the Maeotian and Kimmerian, reflecting considerable changes in species composition of the regional fish fauna before the end of the Late Miocene (

Table 1. Values of the Jaccard (J) and Sørensen–Dice (SD) indices of similarity between the fish assemblages of the various geological stages.

J\SD	Sarmatian	Maeotian	Pontian	Kimmerian	Kuyalnikian	Calabrian	Chibanian	Tarantian
Sarmatian	—	76.1	44.1	55.6	46.2	52.2	50.7	49.2
Maeotian	61.4	—	45.6	55.7	48.1	44.8	46.4	44.4
Pontian	28.3	29.5	—	55.2	54.9	40.0	26.4	27.5
Kimmerian	38.5	38.6	38.1	—	80.0	57.6	52.6	47.3
Kuyalnikian	30.0	31.7	37.8	66.7	—	69.2	56.0	54.2
Calabrian	35.3	28.9	25.0	40.5	52.9	—	70.4	73.1
Chibanian	34.0	30.2	15.2	35.7	38.9	54.3	—	84.0
Tarantian	32.7	28.6	15.9	31.0	37.1	57.6	72.4	—

).

3.2. Functional Diversity

In general, the index of functional diversity demonstrates a decreasing trend, although this trend is not linear (R² = 0.20), as in the case of several other indices, such as species diversity, taxonomic complexity, and taxonomic richness (Figure 4). In a temporal context, functional diversity decreases quite notably during the Late Miocene, remains almost the same in the Pliocene, but increases during the Pleistocene, though not reaching values as high as in the Late Miocene. When compared with other indices, the most notable differences can be observed at the end of the Late Miocene (during the Pontian), when species diversity, taxonomic diversity, and taxonomic complexity all seem to be recovering after the Maeotian, but functional diversity continues to decline (Figure 4a–c).

A similar, but much less noticeable trend is seen during the Kuyalnikian (Late Pliocene–Early Pleistocene). The rest of the Pleistocene, however, is characterised by a different dynamic. In the Calabrian, when the functional diversity and taxonomic complexity of the fish assemblages start to rise (Figure 4c), species and taxonomic diversity are both in decline (Figure 4a,b). Later, in the Chibanian, species diversity recovers (Figure 4a), but taxonomic diversity and complexity both drop noticeably, and they only recover in the Tarantian (Figure 4b,c).

As for taxonomic richness (i.e., the sum of all taxonomic ranks represented in the assemblage), it generally follows the same temporal trend as functional diversity. This trend has been associated with the mathematics behind the dendrogram-based methods of functional diversity assessments [51], although we can see a departure from this trend in the Chibanian, when functional diversity continues to increase despite a slight decrease in taxonomic richness (Figure 4d).

When looking at the composition of the fish assemblages according to their shared functional traits, a number of functional groups can be singled out. In the Sarmatian (Figure 5a), most of the taxa are those that prefer environments with sandy substrate, further diverging by vegetated and non-vegetated habitats. Basses (Morone) and pikeperches (Sander) represented the group of large-bodied carnivores in, respectively, the estuarine and lacustrine–riverine habitats without dense aquatic vegetation. In vegetated lacustrine and riverine habitats with slow and moderate currents and sandy substrates, three functional groups can be identified: a group of large carnivores (including pikes, catfishes, and sabre carps), a group of medium-sized carni-omnivores (including cyprinids and percids like Aspius, Leuciscus, Squalius, Tinca, and Perca), and a group of small- to medium-sized phytophilous omnivore breams (Abramis, Blicca) and roaches (Rutilus cf. rutilus). This latter group in habitats with mud/silt substrates is represented by carps (Carassius, Palaeocarassius). A small group of medium- to large-sized lithophilous carni-omnivore fishes in non-vegetated riverine and estuarine habitats with fast currents and gravel/cobble bottoms is represented by salmonids (Hucho) and cyprinids (Luciobarbus, Rutilus frisii). Gobies (Gobius) appear in the latter group during the Maeotian (Figure 5b).

Functional diversity during the Late Miocene is lowest during the Pontian (FD = 3.33), and there are changes in the composition of several functional groups too. Salmonids disappear from riverine and estuarine habitats with gravel/cobble substrates, whereas sturgeons (Acipenser) and seabreams (Pagrus) appear along with pikeperches (Sander) in habitats with the same substrate, but with moderate currents (Figure 5c). Plantivorous rudds (Scardinius) and grass carps (Ctenopharyngodon) are confined to mainly lacustrine environments with mud/silt substrate. In lacustrine and riverine environments with sandy substrate and aquatic vegetation, two groups of carnivorous and one group of omnivorous fishes can be identified. The carnivorous groups are basically of the same body size and prefer the same habitat conditions (presence of aquatic vegetation, mainly moderate to fast currents), but differ in their reproductive strategies: pikes (Esox) and perches (Perca) are phytophilous, whereas latids (Lates), percids (Leobergia), and catfishes (Silurus) have different types of spawning grounds. The group of omnivorous fishes undergoes little change and includes dace (Leuciscus), bream (Abramis, Blicca), tench (Tinca), and roach (Rutilus tungurukensis).

In the Pliocene, functional diversity practically remains on the same level as in the end of the Late Miocene (Figure 4) with a similar composition of functional groups. Roaches (Rutilus frisii, R. robustus), pikeperches (Sander), seabreams (Pagrus), and sturgeons (Acipenser) make up the group of medium- to large-sized carni-omnivore fishes of riverine–estuarine environments, whereas the similar functional group of lacustrine–riverine environments comprises pikes (Esox), catfishes (Silurus), percids (Leobergia, Perca), and cyprinids (Abramis, Aspius, Leuciscus, Tinca). Crucian carps (Carassius) and rudds (Scardinius) are the small planti-omnnivores of slow-flowing, vegetated, lacustrine–riverine habitats with mud/silt substrates. During the Kuyalnikian, however, the groups of large carnivores (Esox, Silurus) and medium-sized carni-omnivores (percids and cyprinids) tend to be more separated, but there is practically no change in the assemblages compared to the Kimmerian (Figure 5d,e).

In the Middle–Late Pleistocene, functional diversity gradually increases (Figure 4) along with there being substantial changes in the taxonomic composition of cyprinids and percids (Table S1). The common carp (Cyprinus carpio) and the Nogaisk zingel (“Zingel nogaicus”) appear in the Calabrian as a large omnivore of slow-flowing lacustrine–riverine environments and as a medium-sized carnivore of fast-flowing riverine environments, respectively (Figure 5f). Some cyprinids re-appear after their absence in the Pliocene, such as Pelecus in the Calabrian, and Alburnus and Blicca in the Chibanian. Each belong to different functional groups: the zingel as a medium-sized omnivore of moderate-flowing riverine habitats (Figure 5f), and the bleak and the bream as carni-omnivores of slow-flowing lacustrine–riverine habitats (Figure 5g). Salmonids (Salmo sp.) also re-appear in the Tarantian in fast-flowing riverine environments (Figure 5h), joining the barbel (Barbus) and the Black Sea roach (Rutilus frisii). On the other hand, the diversity of silurids and esocids (i.e., large carnivores of lacustrine and riverine environments) gradually decreases by the Tarantian (Table S1).

4. Discussion

The revealed trends in changes in diversity reflect the impact of palaeoenvironmental and palaeogeographic processes on the evolution of actinopterygian assemblages in the south of Eastern Europe and their transition to a lacustrine–riverine fish fauna. The isolation of the Paratethys from the Mediterranean basin during the Middle Miocene led to the substantial desalination of the former [3], and, as a consequence, facilitated the dispersal of freshwater fishes from the east to the west, which resulted in notable similarities along a latitudinal gradient. During the Late Miocene, a thermophilous, limnophilous, lacustrine–riverine fish fauna existed in the freshwaters of the south of Eastern Europe represented by ancient groups (Acipenseridae, Salmonidae, and Moronidae) as well as taxa that entered the region from Asia (Cyprinidae, Siluridae, Esocidae, and Percidae) and Africa (Clariidae, Latidae). Considering its taxonomic composition and time of existence, this fauna was part of the Euro-Siberian ichthyofauna sensu Sychevskaya [55].

From the end of the Miocene, there had been a gradual decline in the taxonomic richness of the fish assemblages of the region. The increased tectonic activity in the beginning of the Pliocene resulted in irreversible changes in the hydrological regime of the Paratethys and of the river systems of its drainage basin [56,57]. Orogenetic activity led to the emergence of new mountain ridges changing the distribution patterns of air masses, which resulted in progressive cooling [58]. During the first half of the Pliocene, several thermophilic taxa disappeared from the local freshwater fish fauna, while the share of cold-tolerant elements increased, the latter being represented mainly by ecological analogues of the aboriginal taxa, but less demanding and able to survive under more contrasting climatic conditions. At the same time, the Paratethys split up into several basins divided by newly emerged terrain, which complicated faunal exchange and played a crucial role in shaping the evolutionary trajectory of the fish assemblages that existed in these basins. Later, in the second half of the Pliocene and in the Pleistocene, the freshwater fish assemblages in the south of Eastern Europe remained relatively poor in terms of taxonomic richness and diversity, being represented mainly by unpretentious elements of widely distributed genera, though at least some water bodies of the region played the role of local refugia of thermophilous species. Functional diversity slightly increased during the Pleistocene, indicating a gradual stabilisation of the ecological relationships and functional structure of the local assemblages.

The most recent stage in the evolution of the freshwater fish fauna of the south of Eastern Europe began at the end of the Pleistocene and the beginning of the Holocene, when under the conditions of progressive warming and weak tectonic activity there was a repeated limnisation of river systems of the region [59]. The changes in climate, increased heat capacity, and general mineralisation of fresh waters also led to the decline of cryophilic and cold-tolerant elements in the fish fauna.

The most notable changes in the taxonomic composition of actinopterygian assemblages of the region took place in the Late Miocene, when representatives of the families Clariidae, Moronidae, Sciaenidae, and Gobiidae disappeared. Salmonids disappeared as well and re-appeared only in the Tarantian (genus Salmo); few of their remains were recovered from the sandstones of Novgorod-Siverskyi and the Dnipro alluvium, areas that likely served as spawning sites. Another group of anadromous fish, the sturgeons (Acipenseridae) appeared in the regional fauna at the end of the Late Miocene, during the Pontian, represented by Acipenser gueldenstaedtii. In taxonomic terms, the largest groups during the entire timespan considered were the cyprinids (Cyprinidae) and percids (Percidae), occupying various positions in the functional space of the fish assemblages.

In terms of ecological and functional roles, the most important parameters determining the distribution of fish taxa appeared to be substrate, presence or absence of aquatic vegetation, and belonging to a certain trophic and reproductive guild. Current velocity and trophic group were also important factors. Most species and genera were phytophilous, which were more common in slow-flowing or standing water bodies with well-developed underwater vegetation and sandy bottoms. For these fish, vegetation acted as both a food base and spawning substrate, and it also could have been used as shelter and a schooling place for juveniles. Slow currents and loose bottom sediments, in turn, allowed plants to take root and thereby potentially contributed to the further overgrowth of the water body.

The distribution of fish by the type of environment they prefer was less specific, although most species were confined to lacustrine–riverine and riverine habitats, while only a few taxa preferred brackish conditions and thus were usually restricted to the lower reaches of rivers and estuaries. Body size apparently had no substantial role in determining the distribution of the identified taxa by functional groups and fishes of all size classes were represented in each environment and habitat. Forms with a terminal mouth position predominated in all the samples studied, while those with inferior or superior mouth positions were of minor importance, and fishes having the ventral mouth (e.g., sturgeons) occurred very rarely in the region.

The analysis of the trophic and reproductive specialisation of individual fish taxa allows for an estimation of the potential roles these fish played in palaeocommunities. Most species of large functional groups were omnivorous and carnivorous (predatory), although plantivorous forms (e.g., Scardinius haueri, S. erythrophthalmus, S. ponticus, Scardinius sp.) were also present. In contrast, there were few specialised malacophagous species (e.g., Rutilus frisii, Rutilus cf. R. frisii, and R. robustus).

The number of plantivorous fish taxa in the south of Eastern Europe remained consistently low throughout the entire late Cenozoic, reaching its maximum in the Pontian stage (n = 4, or 15.1% of all identified taxa), while only one extant species—the common rudd, Scardinius erythrophthalmus—existed in the water bodies of the region during the late Middle and Late Pleistocene (

Table 2. The number (n) of fish taxa representing different trophic guilds and their proportion (%) in the studied palaeoassemblages.

Period	Stage	Trophic Guild								Σ
		Plantivorous		Malacophagous		Carnivorous		Omnivorous
		n	%	n	%	n	%	n	%
Pleistocene	Tarantian	1	4.5	1	4.5	8	36.4	12	54.6	22
	Chibanian	1	4.3	2	8.7	8	34.8	12	52.2	23
	Calabrian	2	7.7	3	11.5	11	42.3	10	38.5	26
Pliocene	Kuyalnikian	2	8.7	3	13.0	11	47.9	7	30.4	23
	Kimmerian	3	10.3	4	13.8	15	51.8	7	24.1	29
Miocene	Pontian	4	15.4	3	11.5	12	46.2	7	26.9	26
	Maeotian	2	7.1	2	7.1	10	35.8	14	50.0	28
	Sarmatian	3	7.9	2	5.3	19	50.0	14	36.8	38

). Malacophagous taxa were equally represented in the fossil record of the area (four species in the Kimmerian and one—the Pontic roach, Rutilus frisii—in the Tarantian). The number of carnivorous fish taxa was relatively high (n = 19, 50.0%) during the late Sarmatian, although it gradually decreased over time, with a reduction in their share in the palaeoassemblages to up to eight taxa (36.4%) in the Tarantian. Omnivorous fish were also numerous, and their representation, in contrast, steadily increased from the second half of the Pliocene (30.4%) to the Late Pleistocene (54.6%). Thus, there was a trend towards an increase in the number of generalists and the loss of specialised forms in the region’s fish fauna.

In terms of the representatives of different reproductive guilds (

Table 3. The number (n) of fish taxa representing different reproductive guilds and their proportion (%) in the studied palaeoassemblages.

Period	Stage	Reproductive Guild								Σ
		Lithophilous		Phytophilous		Pelagophilous		Various Grounds
		n	%	n	%	n	%	n	%
Pleistocene	Tarantian	7	31.8	12	54.5	2	9.1	1	4.6	22
	Chibanian	5	21.7	14	60.9	3	13.0	1	4.4	23
	Calabrian	7	26.9	15	57.7	3	11.5	1	3.9	26
Pliocene	Kuyalnikian	7	30.4	10	43.5	5	21.7	1	4.4	23
	Kimmerian	7	24.1	12	41.4	6	20.7	4	13.8	29
Miocene	Pontian	5	19.2	12	46.2	6	23.1	3	11.5	26
	Maeotian	8	28.6	14	50.0	3	10.7	3	10.7	28
	Sarmatian	10	26.3	16	42.1	6	15.8	6	15.8	38

), phytophilous fishes comprised the largest and the most stable group (n = 10–16), followed by lithophilous forms, the proportion of which ranged from 19.2% in the Pontian to 31.8% during the Tarantian stage. The third most numerous group (mostly during the Late Miocene and Pliocene) were pelagophilous fishes, although their number considerably decreased in the Pleistocene. The number and proportion of fishes that used various spawning grounds was always the smallest (n = 3–6; 10.7–15.8%). Moreover, it sharply decreased to 4.4% in the Pliocene, from which point only one species (Sander lucioperca) represented this group in the material from the samples studied.

To conclude, the studied late Cenozoic actinopterygian assemblages of the south of Eastern Europe represented a typical limnophilous, lacustrine–riverine fish fauna, which, in the Late Miocene, comprised a relatively large number of thermophilic elements. In the Pliocene, due to tectonic activity and gradual cooling, this fauna’s composition transformed with the disappearance of warm-loving species and the expansion of cold-tolerant taxa. Subsequent warming and mineralisation of freshwaters during the Pleistocene reduced the number of representatives of the latter group and reshaped the composition and functional structure of the fish assemblages.

All these changes are reflected in the values of the analysed diversity indices, including taxonomic and functional diversity, as well as in the changes that occurred in the composition of various functional groups within the fish assemblages that existed during subsequent geological stages.

We must emphasise, however, that the results of our analyses should only be considered as general trends, because the diversity indices of palaeoassemblages are a priori lower, especially when compared to recent communities, due to taphonomic selectivity and the incompleteness of the fossil record. The analysed data and the obtained results of this study characterise the identified assemblage, not the entire palaeocommunity. Therefore, the disappearance and appearance/re-appearance of various taxa in the fossil record may as well be related to specific conditions of fossil accumulation.

Nevertheless, the analysed sample of nearly 12,000 identified specimens collected from 53 localities for nearly 70 years reveals changes and patterns that could have been characteristic for the entire palaeocommunity. Further research and findings of fish fossils from the region will certainly shed more light on how actinopterygian diversity evolved in the water bodies of the south of Eastern Europe.