Transposons, Genome Size, and Evolutionary Insights in Animals

Despite many studies, the C-value enigma sensu Gregory [2005], that is the expansion of the genome size mainly due to the accumulation of non-coding DNA not related to the genetic complexity and to the evolutionary position, remains to date one of the most fascinating problems which is still not fully understood in the organization and evolution of the genome.

One of the peculiarities of these changes in genome size is that they seem to influence, regardless of their sequences, various parameters of the cell which have an impact on morphological and functional characteristics of the organism. These characteristics are subject to natural selection, therefore influencing major evolutionary processes.

To explain the variability in genome size, several theories have been discussed. According to some of these theories, there would be no causal relationship between the amount of DNA and cellular parameters, and the accumulation of DNA would depend simply on different levels of tolerance of each species towards insertion, amplification, and storage of repetitive sequences [Pagel and Johnstone, 1992; Charlesworth et al., 1994]. Other hypotheses imply that variations in genome size would always have a causal role in changes in cellular parameters and in those traits of the organism that are subject to natural selection and therefore to stabilizing selection for the optimum genome size [Bennett, 1971, 1973; Cavalier-Smith, 1982, 1985]. Finally, other hypotheses speculate that the genome size of a species evolves until the loss of non-coding self-replicating DNA through small deletions has an equal rate of DNA gain through long insertions [Petrov, 2002a].

The most recent studies on genome sequencing have allowed a significant progress in the knowledge of the organization and composition of the genome. Moreover, studies on mobile DNA and their propensity to independently amplify and to specifically insert in the host genome have changed the perspective of the C-value enigma.

Many works have shown a general positive correlation among different parameters of the genome, especially between repeated sequences including transposons and the genome size within numerous eukaryotes [Elliot and Gregory, 2015a] and mainly in vertebrates [Chalopin et al., 2015]. However, this correlation appears to especially affect the quantitative aspects of transposons, since a limited correlation between the diversity of transposons and genome size was noted only for genomes <500 Mb [Elliot and Gregory, 2015b].

In this regard, changes in the percentage of transposons and genome size in 161 species of animals, whose entire genome has been sequenced until now, including 88 deuterostomes, 68 protostomes, and 2 cnidarians, 2 ctenophores, and 1 placozoan species, were examined in the light of major evolutionary transitions in order to verify the influence that these sequences had on evolutionary processes such as speciation and adaptation to the environment. Data on genome sizes were obtained from the Animal Genome Size Database (www.genomesize.com), and minimum and maximum values for each taxon are shown in table 1.

Most of the studies on genome size involved vertebrates, while those focusing on invertebrates are very limited (no more than 1% of all living invertebrate species have been studied). The situation regarding genome sequencing and transposon analysis is more balanced. In fact, 54.7% of these studies focused on chordates, especially vertebrates, and 42.2% on protostomes and 3.1% on primitive metazoans (table 2).

The information on genome size and composition of primitive metazoans is very limited. The genome size is generally low, ranging from 0.04 to 1.84 pg/N (table 1). The percentage of transposons was studied only in 5 species and seems not to be correlated with the genome size. Indeed, in Trichoplax the percentage of transposons is only 0.13%, the lowest among animals, while in ctenophores it is <10%. In cnidarians, despite the small genome size, the percentage of transposons shows high values comparable with those identified in many protostomes and deuterostomes. Although these data are very scarce, they might suggest that at the origin of metazoans genome sizes and perhaps even the percentage of transposons were generally low [Gregory, 2005] and that transposons would have amplified independently in various phyla in the early phases of evolution.

Among protostomes, arthropods and especially insects have been extensively investigated. Within insects genome sizes are limited, ranging from 0.09 to ∼4 pg/N, with most of the species abutting values <1.5 pg/N. Exceptions are represented by orthopterans with values reaching up to 16.93 pg/N. An interesting aspect concerns the differences between the holometabolous (with average values <2 pg/N) and hemimetabolous (possessing values significantly higher) insects [Gregory, 2005; Hanrahan and Johnston, 2011]. A comparison of the percentage of transposons indicates that these differences depend predominantly on their expansion (fig. 1), as shown by hemipterans and orthopterans having on average a high percentage of transposons compared to other orders (table 2).

In arachnids, genome sizes vary 8×, from 0.74 to 5.7 pg/N with an average of 2 pg/N. These arthropods do not metamorphose but show various molts during growth, and, similarly to hemimetabolous insects, they do not have a genome size limit of 2 pg/N [Gregory and Shorthouse, 2003].

Higher and more variable values can be observed among crustaceans, ranging from 0.16 to over 50 pg/N, with an average of about 3 pg/N and with most of the species not exceeding 6 pg/N. Genome size values >20 pg/N are extremely rare among invertebrates and crustaceans, and they are limited to species living in extreme environments, such as polar regions, especially the Antarctic, and the deep seas (mainly in hydrothermal vents) [Gregory, 2005; Rees et al., 2007, 2008; Bonnivard et al., 2009; Dufresne and Jeffery, 2011].

The other arthropod groups harbor a fairly limited genome size that only rarely exceeds 5 pg/N [Gregory and Shorthouse, 2003; Gregory, 2005; Hanrahan and Johnston, 2011].

Data collected so far in arthropods suggest that a small genome size was a widespread common ancestral condition within this phylum and that during evolution increases in genome size independently occurred in different lineages [Hanrahan and Johnston, 2011].

In mollusks the occurrence of a whole genome duplication has been speculated on [Yoshida et al., 2011]. The genome size has been investigated mainly in bivalves, and in this group, values range from a minimum of 0.43 pg/N to a maximum of 7.85 pg/N with an average of 1.8 pg/N. One of the highest values has been found in the Antarctic bivalve Neobuccinum eatoni [Libertini et al., 1993]. Among gastropods, it was noted that terrestrial species display a genome ∼2× larger than their freshwater relatives [Hinegardner, 1974; Vinogradov, 2000; Gregory, 2005]. Hinegardner [1976] hypothesized an increase in genome size during evolution from more generalized to more specialized species. However, the latest results contrast with this hypothesis and suggest instead that more generalized mollusk species possess larger genomes [Rodriguez-Juiz et al., 1996].

In almost all other protostomes, except in very rare cases like the flatworm Otomesostoma, whose genome reaches 20 pg/N, a quite similar situation has been described with a range that goes from a minimum of about 0.1 pg/N to a maximum of 5 pg/N [Rodriguez-Juiz et al., 1996; Gregory, 2005]. Despite many species of annelids, nematodes, and flatworms being parasites and generally having small genome sizes [Sundberg and Pulkkinen, 2015], no difference in the percentage of transposons between free-living and parasitic species was noticed [Zheng et al., 2013].

Examining the presence and the proportion of repetitive DNA in various protostome phyla it is evident that in all cases the expansion of the genome depends on the expansion of various classes of transposons (table 2; online suppl. fig. 1; for all online suppl. material, see www.karger.com/doi/10.1159/000444429). In insect species harboring genomes <2 pg/N there is a linear and direct relationship between the percentage of transposons and the genome size (fig. 1A), which indicates that the expansion of the genome is caused mostly, if not completely, by the amplification of transposons. However, taking into account the orthopterans displaying genomes >6 pg/N, the above relationship shows a logarithmic development (fig. 1B). This trend is also confirmed when looking at all arthropods and protostomes (online suppl. figs. 1, 2). These data suggest that exceeding certain values, the increase in genome size is not fully justified by the amount of transposons, but it could depend on other sequences or on the preservation of a certain percentage of transposons without any selective pressure causing the accumulation of mutations over time and masking its repetitiveness. Concerning the classes of transposons in various protostomes, and especially within the group of insects, certain homogeneity was noticed at the level of genus or family, while there is a considerable diversity, both among the various phyla and within each phylum or subphylum (table 2).

While studying genome sizes in deuterostomes, the most evident aspect is the clear difference found between vertebrates and other deuterostomes including primitive chordates. Indeed, in the latter, genome sizes are small, while in vertebrates they are on average higher, more variable, and can reach very high values of >100 pg/N [Gregory, 2005]. This could be explained by the hypothesis that origin and some important steps in the evolution of vertebrates would have been characterized by a duplication of the entire genome. Such duplications would have occurred at the origin of vertebrates, at the separation of gnathostomes and agnates, and at the origin of teleosts after the separation of actinopterygians and sarcopterygians. These duplications coincide with a burst of new character appearances and with the acquisition and the increase in phenotypic complexity [Meyer and Schartl, 1999; Panopoulou et al., 2003; Donoghue and Purnell, 2005; Panopoulou and Poustka, 2005; Volff, 2005]. However, the significant differences identifiable within certain classes cannot be explained only by genome duplication events, suggesting on one hand that they were not the only causes of changes in genome size and that genomic dimensions have been affected by many factors on the other [Chalopin et al., 2015]. A very significant increase in genome size not due to duplication of the entire genome is known in lungfishes and salamanders, whose origin and early stages of evolution have occurred in conjunction with the transition from aquatic vertebrates to terrestrial ones.

Although some hypotheses suggest that in vertebrates the most primitive species possess small genomes [Gregory, 2005], during evolution the genome size does not appear to have followed just one trend. For each class of the subphylum different trends are observed, leading to a genome expansion in some classes and to a contraction in others.

Beside lungfishes and amphibians, the cartilaginous fishes display the largest genome size ranging from 1.5 pg/N in the chimaeras to 17 pg/N in sharks with an average of 5.7 pg/N [Stingo et al., 1980; Gregory, 2005]. Although the lowest values are found in the chimaeras, which are considered to be the most primitive group among chondrichthyes, it is not possible to identify a clear trend in genome size evolution within this class. Larger cells and nuclei were observed in cartilaginous fish living in cold temperatures. Although a direct relationship between genome size and cell and nucleus size is well known, it has been noticed that cells of selachian species living in cold water are larger than those of species living in warm water even if they have the same genome size [Hardie and Hebert, 2003]. Higher average genome sizes were also observed in some deep-sea selachians, although this correlation is not significant [Sion et al., 2004].

The percentage of transposons has been studied so far only in the chimaera Callorhinchus milii [Chalopin et al., 2015]. However, C₀t analyses, although based on a limited number of species, would suggest that the increase in genome size in this fish is correlated to an increase in moderately repetitive DNA [Morescalchi and Olmo, 1982; Olmo et al., 1982; Stingo et al., 1989] (online suppl. fig. 3). Since it is known that transposons belong to the above-mentioned fraction [Krebs et al., 2013], it is presumed that even in this class the expansion of the genome could depend on the amplification of transposons. Reassociation kinetics have also shown that in some species the increase of DNA would also be accompanied by a doubling of the so-called single-copy fraction that is largely made up of structural gene sequences [Krebs et al., 2013]. This could be a remnant of a primitive whole genome duplication [Olmo et al., 1982].

From an evolutionary point of view the ray-finned fishes and especially teleosts are the most successful group among the vertebrates. They comprise 99% of the 30,000 species of extant fishes and, along with mammals, exhibit the highest rate of diversification in the course of evolution [Benton, 2000; Olmo, 2006].

The origin of teleosts would have been characterized by a specific whole genome duplication that would coincide with a burst of character acquisitions and with an increased phenotypic complexity [Vanderpoele et al., 2004; Donoghue and Purnell, 2005; Volff, 2005]. Loss or sub-function partitioning of duplicated genes would have been involved in the generation of phenotypic variability of these fishes [Volff, 2005], which actually experienced more frequent gene linkage disruptions than other vertebrates [Ravi and Venkatesh, 2008]. Despite their great evolutionary success, the teleost fishes show a remarkable uniformity in genome size, with values ranging from a minimum of 0.4 to a maximum of 4.4 pg/N with an average of 1.2 pg/N, one of the lowest among all the deuterostomes. A correlation between genome size and extreme environments has been noted also in bony fish species: mesopelagic and bathypelagic species display larger genomes than surface water fishes [Ebeling et al., 1971], and the cold-water species are larger in genome size than warm-water species [Hardie and Hebert, 2003].

Also in fish a clear and direct correlation between the percentage of transposons and the increase in genome size was identified (table 2). Moreover, it has also been hypothesized that speciation events could be associated with retrotranspositional bursts [Volff, 2005].

The typical tendency of teleosts to preserve small genomes is evident in pufferfishes, which possess the smallest and most compact genomes of all vertebrates. They have a lower percentage of repeated sequences (especially DNA transposons) and shorter intronic sequences. This situation would depend both on a high rate of intron and transposon loss and on a higher level of indels (insertions/deletions) [Imai et al., 2007; Loh et al., 2008; Noleto et al., 2009; Guo et al., 2012].

One of the most controversial steps in the evolution of vertebrate genome sizes is the transition from aquatic to terrestrial environments involving lobe-finned fishes and amphibians.

The extant lobe-finned fishes include the coelacanths, very popular in the Devonian, and are represented today only by the Latimeria genus with 2 species dwelling in the deep waters of Africa and Indonesia and the lungfishes, which are shown by molecular studies to be the direct ancestors of the tetrapods [Amemiya et al., 2013; Biscotti et al., 2016].

The 2 species of Latimeria have a moderate genome size of ∼3 pg/N of which 20% consist of non-LTR transposons [Amemiya et al., 2013; Chalopin et al., 2015].

The lungfishes are freshwater fish that date back to the Devonian. Widely spread in the early Carboniferous, they began to decline in the Mesozoic. Currently, there are 6 living species of lungfish belonging to 3 genera: Neoceratodus, Protopterus, and Lepidosiren. Lungfishes have huge genomes, the largest among animals, exceeding 100 pg/N in Lepidosiren.

By studying the size of the bone lacunae (an indirect measure of genome size) in fossil and living lungfishes, Thomson [1972] noted that cell sizes (and presumably genome sizes) were uniformly small in the Devonian and that a progressive and significant increase took place independently in Neoceratodus and Lepidosiren lineages since the Carboniferous, reaching its climax at the beginning of the Mesozoic. This increase accompanied a progressive evolutionary decline. The only lungfish that has been studied for the composition of the genome is N. forsteri, in which ∼40% of the DNA consists of non-LTR transposons, mainly CR1 and L2 [Sirijovski et al., 2005; Metcalfe et al., 2012]. A similar percentage of transposons (between 35 and 40%) was also inferred in Lepidosiren [Metcalfe et al., 2012].

A similar study on the size of the bone lacunae in fossil and living amphibians [Thomson and Muraszko, 1978] suggested that cell and genome sizes were relatively small at the origin of this class and that any increase occurred secondarily and separately in different class lineages. Furthermore, it is speculated that a genome size of 2.5-5 pg/N would have been also the baseline for coelacanths, lungfishes, and for all living tetrapods. Similar values of genome and cell size are indeed common in several species of frogs, lepospondyl amphibians, living and extinct non-avian reptiles, and mammals [Organ et al., 2007, 2011]. In this regard it has been speculated that a genome size included in the above range would represent the ancestral and characteristic value of the entire sarcopterygian lineage from which the large genomes of Dipnoi and salamanders and the small genomes of birds would be secondarily derived [Organ et al., 2011].

Amphibians include 3 extant orders which differ in genome size: the frog and Apoda genomes are moderate, ranging from 0.95 to about 13 pg/N, urodeles possess instead very high DNA values between 13.5 and 60 pg/N [Gregory, 2005]. Increases in amphibian genome sizes mainly depend on the increase in repetitive DNA, especially in the moderately repetitive C₀t analysis fraction [Morescalchi and Olmo, 1982] and on a lengthening of the introns, which in Ambystoma mexicanum and in other salamanders are longer than in humans, chickens, and frogs [Smith et al., 2009; Eo et al., 2012; Sun et al., 2012a; Voss et al., 2013]. The presence of transposons has been studied only in 2 frog species and in 7 species of salamanders: the primitive Cryptobranchus and 6 species of plethodontids, one of the most advanced family of the suborder. In the 2 frog species the percentage of transposons is ∼40%, with a prevalence of DNA transposons in Xenopus and LTR retrotransposons in Nanorana (table 2) [Sun et al., 2015]. In urodeles, transposons range from 25 to 50%, and almost all belong to Gypsy LTR, DIRS, and ERV 1, which indicates that salamander transposons all have the same origin (table 2) [Sun et al., 2012a, b; Sun and Mueller, 2014].

Some authors speculate that Lissamphibia have originated in the Permian, others in the early Triassic [Marjanovic and Laurin, 2007]. The oldest urodelian fossils date back to the middle Jurassic [Gao and Shubin, 2003]. Therefore, it is not easy to imagine the scenario in which the very large genomes found in extant salamanders would have originated. By comparing the genome sizes to the time of speciation in 28 species of salamanders, it is possible to infer that the largest genomes are found in the oldest species and that a gradual contraction happened up to the lowest values found in more recent species (fig. 2; online suppl. table 1). Considering the uniformity of transposon percentages, it can be assumed that the increase in the size of the salamander genomes is derived from a burst of a single group of LTR retrotransposons during the Paleocene era [Sun and Mueller, 2014] when the primitive cryptobranchids appeared. Subsequently, due to the loss of transposon sequences, there would have been a progressive reduction in genome size until the Miocene.

The non-avian reptiles own moderate genome sizes ranging from 1.1 to 5.4 pg/N, values similar to those assumed for the basal-most lobe-finned fishes and tetrapods. Transposons account for on average 30% of the entire genome, and they all are non-LTR retrotransposons. Birds are characterized by some of the smallest genomes among vertebrates, ranging from 1 to 2.2 pg/N with an average value of 1.4 pg/N. The percentage in transposons is also very low with an average of 6.8% (table 2). It has been suggested that such small genomes have evolved to acquire high metabolism levels, essential for the ability to fly [Hughes and Hughes, 1995]. However, the study of fossil cell sizes in dinosaurs has shown that the reduction of cell and genome sizes occurred between 230 and 250 Mya in saurischians (the lineage from which birds have originated), long before the appearance of the first birds, and this would have depended on a drastic reduction of non-LTR transposons in this lineage [Organ et al., 2007].

The genome sizes of mammals have a range similar to that of non-avian reptiles, from a minimum of 1.7 to a maximum of 8.4 pg/N and an average of 3.5 pg/N, close to the value for man. Regarding transposable elements, mammals present an average of 40%, the highest of all amniotes, if compared to 29% in non-avian reptiles and 6% in birds (table 2). Also in this class, transposons have caused an elongation of the introns [Wang et al., 2012].

A particular situation was described in the vespertilionids: since 36 Mya this group has experienced a rapid adaptive radiation, leading to the onset of the mammal family most rich in species (>400). This radiation coincided with an initial burst of DNA transposons that would have facilitated the rapid diversification of these bats [Platt et al., 2014].

Analyzing the trends in genome sizes with the percentage of transposons in chordates, there is a situation similar to that seen in protostomes. In fact, even in this phylum, for relatively low values (<5 pg/N) there is a significant linear and direct correlation between the increase in genome size and the increase in transposon percentage which suggests that the expansion of the genome size is caused primarily, if not only, by the expansion of transposable sequences (fig. 3A). Vice versa, if we consider species with genomes >5 pg/N, the correlation shows a logarithmic pattern which suggests that the repetitive DNA, and in particular transposons, are not the only cause of the increase in the genome size (fig. 3B). Alternatively, it is assumed that a certain percentage of transposons are stored without being subjected to any selective pressure and are thereby free to accumulate mutations that in time would mask the original repetitiveness.

Similarly to what is known in protostomes, even in deuterostomes there is certain variability with regard to the different classes of transposons. In general in the ray-finned fish DNA transposons are more frequent, while in tetrapods, except for the salamanders, the non-LTR retrotransposon subclass is more frequent (table 2).

In animals, as in all eukaryotes, the contribution of transposons to the changes in genome size would only be quantitative, since a limited correlation between the diversity of transposons and genome size was found restricted to species harboring genomes <500 Mb and since trends in this correlation are not the same in animals and in plants [Elliott and Gregory, 2015b].

Although the range of variation in genome size is different in protostomes and deuterostomes, there are common characteristics. There is a correlation between genome size and intronic length [Moriyama et al., 1998; Vinogradov, 1999a; Wang et al., 2012; Zhang and Edwards, 2012]. Moreover, the rate of change in genome sizes is mainly dependent on variations in the proportion of repetitive DNA, especially transposons, which is in turn proportional to the initial genome size [Oliver et al., 2007]. Finally, up to certain values, the ratio of the increase in genome size to the increase in the percentage of transposons is linear, while beyond certain limits the trend shows a logarithmic pattern, indicating that further expansions of the genome depend on DNA sequences whose degree of repetitiveness is not detectable (online suppl. fig. 4).

There are many factors influencing changes in genome size: whole genome duplication, segmental duplication, DNA repeat proliferation, polyploidy, etc. However, most of the results obtained so far and mentioned in the previous section point out that the most important cause of genome expansion is represented by the amplification of transposable elements [Kidwell, 2002; Chalopin et al., 2015]. Given their ability to rapidly replicate, transposable elements represent one of the best mechanisms, if not the very best, able to relatively quickly determine changes in genome size [Dufresne and Jeffery, 2011].

In this regard there are some important issues that require in-depth analysis:

• the causes of amplification of transposable elements;

• the mechanisms interacting with transposon activity impacting the expansion and the contraction of the genome;

• the influence of genome size and transposon percentage on structural and functional characteristics of the genome and the cells;

• the effect of changes in genome size and transposon percentage on evolutionary processes and in particular on the adaptation to changing environmental conditions.

Barbara McClintock [1984] formerly suggested that the activity of mobile elements in the genome represented a reaction to environmental stressors. This hypothesis was supported by several other authors [Capy et al., 2000; Kidwell, 2002; Chénais et al., 2012; Piacentini et al., 2014], and numerous cases of transposon activation due to environmental stress conditions have been observed in plants, where one of the most influential environmental factors is represented by temperature [Vitte and Panaud, 2005; Kelly and Leitch, 2011; Chénais et al., 2012; Ito, 2013; Wheeler, 2013; Ishiguru et al., 2014; Kim et al., 2014]. Examples of correlations between environmental stressors and transposon activity are rarer in the animal kingdom. However, similar connections were noted in Drosophila melanogaster, where differences in the rate of transposition were related to the development of temperature [Capy et al., 2000; Kim et al., 2014; Piacentini et al., 2014] and to the development of a resistance to pesticides [Chénais et al., 2012]. In human cells a reorganization of the transcriptome after thermal shock was described, putatively involving SINE transposon sequences [Wheeler, 2013]. Changes in the proportion of repetitive DNA in relation to different latitudes and altitudes were observed in some invertebrates [Fielman and Marsh, 2005; Dufresne and Jeffery, 2011] and vertebrates [Litvinchuck et al., 2007]. Moreover, an indirect indication of the relation between transposon activity and environmental stresses would be the large genome size observed in some fish and crustaceans living in the deep waters [Rees et al., 2007, 2008; Bonnivard et al., 2009].

Transposon activation induced by environmental stressors causes deleterious effects in certain cases but favors an increase in genetic variability in others. Therefore, it represents an effective adaptive response to drastic environmental changes [Piacentini et al., 2014].

One of the most extreme changes in genome size and in the percentage of transposons regards vertebrates involved in the water-to-land transition characterized by the appearance of huge genomes such as those of lungfishes and salamanders. The activation of transposons as a result of environmental stress may be the best explanation for the rapid emergence of these huge genomes, although there is currently no direct evidence.

The transition from lungfishes to tetrapods and the early stages of the amphibian evolution began in the late Devonian and continued until the Carboniferous, a very long period with extreme climate changes.

During the Carboniferous, in lungfishes there was a progressive increase in genome size starting from values of 2.5-5 pg/N and leading to values >100 pg/N found in extant species. This increase, independently taking place in different lineages, was likely accompanied by a dramatic reduction of morphological and taxonomic evolution [Thomson, 1972; Stanley, 1975]. This evolutionary decline is certainly due to drastic environmental changes, since in the second half of the Carboniferous (Pennsylvanian) severe climate changes with temperature drops and aridification phenomena took place, also provoking mass extinction [Sahney et al., 2010]. Lungfishes, living in shallow freshwater, are particularly sensitive to temperature changes and drying. Therefore, increases in genome size were very likely the consequence of a burst of transposon amplification (probably non-LTR) induced by stressful situations.

A similar increase in genome size was observed in amphibians whose fossil forms appeared in the late Devonian and achieved the highest evolutionary success in the Carboniferous. In some amphibian fossils, especially temnospondyls, a genome expansion would have taken place, leading to high values similar to those of some living salamanders [Thomson and Muraszko, 1978]. Again, this expansion would have occurred during the latter stages of the amphibian radiation in the late Permian and early Triassic period, in which ∼70% of the terrestrial amphibian families probably died [Erwin, 1994; Roelants et al., 2007]. Also in this class it is therefore likely that the increases in genome size depended on a burst of transposons stimulated by environmental stress.

The presence of the enormous genomes possessed by living salamanders is less clear to define, also because genome sizes are much smaller in frogs and Apoda [Gregory, 2005]. The oldest urodelian fossils date from the middle Jurassic [Gao and Shubin, 2003], long after the disappearance of temnospondyls [Marjanovic and Laurin, 2007, 2013]. It is therefore not possible to determine whether the genome expansion of the living salamanders is a direct result of the expansion observed in temnospondyls or whether it is a new and independent one.

As mentioned in the previous section, the large genomes of extant urodeles derived from an amplification of LTR transposons dated back to the end of the late Cretaceous, another period characterized by mass extinctions. Therefore, it is possible that even this LTR transposon burst would have been stimulated by severe conditions of environmental stress.

In some cases transposons can provide evolutionary advantages to the host as for the molecular domestication that leads to the appearance of new functional genes or regulatory sequences or the ability to increase the genetic variability through their mutagenic action [Kidwell, 2002; Schmidt and Anderson, 2006; Volff, 2006; Boehne et al., 2008; Chalopin et al., 2012; Lee and Kim, 2014; Piacentini et al., 2014; Platt et al., 2014; Chalopin et al., 2015]. In most cases, however, their activity has a deleterious effect: they can alter gene expression by fitting into regulatory sequences, their insertion can result in deleterious mutations or even in extensive chromosomal rearrangements [Petrov et al., 2003], and they can impose an exaggerated functional load to the host due to an increased replication of foreign DNA [Kidwell, 2002; Johnson, 2007; Dufresne and Jeffery, 2011; Chenais et al., 2012; Sun et al., 2012a; Lee and Kim, 2014]. In response to this situation the organisms have triggered various mechanisms to curb the activity of transposons and keep the genome size within certain limits.

Comparing genome size and the percentage of transposons in different species of plants and animals, it seems clear that the dimensions of the genomes are the result of a complex balance between gain and loss of repeated sequences [Petrov et al., 2003]. The propensity of transposon insertion would not be random, but it would depend on specific compositional and functional characteristics, including the initial genome size in some organisms [Dufresne and Jeffery, 2011]. Lynch and Conery [2003] have hypothesized that below a minimum genome size the insertion of a transposon would not be possible; at intermediate sizes only some species allow insertions, while above a specific value all species would be contaminated.

The control of the genome size is implemented through 2 phases: the inactivation of transposons with a consequent limitation of their insertion and the deletion of transposons and other repetitive DNAs.

Transposon inactivation depends largely on 2 mechanisms: methylation [Johnson, 2007; Kelly and Leitch, 2011; Wheeler, 2013] and interference of small RNAs such as plant small interfering RNAs (siRNAs) and animal PIWI-interacting RNAs (piRNAs, probably evolved as a defense mechanism against viruses) [Arensburger et al., 2010; Ito, 2013]. Initially, methylation was thought to be the only responsible mechanism for transposon control in plants, and only later was it noted that transposon silencing depends on the interference of siRNAs [Ito, 2013]. The prevailing mechanism in animals is piRNAs [Wheeler, 2013; Klenov et al., 2014], which have a very ancient origin given their presence in the sponge Amphimedonand in the cnidarian Nematostella [Grimson et al., 2008].

Transposon silencing through piRNAs entails the recognition of aberrant double-helix RNAs derived from transposons and their subsequent processing to small RNAs (sRNAs) [Slotkin and Martienssen, 2007]. Small single-stranded RNAs derived from an sRNA filament may drive the degradation on transposon sequences or they may affect DNA methylation and histone modification to prevent the transcriptional activity of transposons [Kelly and Leitch, 2011].

The role of piRNAs in transposon inactivation and the resulting control over expansion of the genome has been demonstrated by a genome comparison between D. melanogaster and Aedes aegypti. The latter harbors a genome 8× larger than the former and with a much higher percentage of transposons (table 2). The diversity of transposon sequences is comparable in both species; however, only 19% of Aedes piRNAs map on mobile elements compared to 51% in Drosophila. As a result the action of piRNAs against transposons is much smaller in the first species, easily allowing their insertion and a subsequent increase in genome size [Arensburger et al., 2010]. A similar situation could be observed in a comparison between teleosts and mammals. In the former, where genomes are on average smaller and exhibit the highest diversity of transposons among all vertebrates [Chalopin et al., 2015], the piRNA genes have a faster evolution and this may represent an adaptive mechanism to counteract the greater variability in transposons. Moreover, a greater variability in teleost piRNAs has been suggested to be linked to the external fertilization where gametes are subjected to a greater risk of transposon invasion [Yi et al., 2014].

Methylation is another effective mechanism for the restriction of transposon activity. This is supported by a direct correlation observed in the mouse between the demethylation of the intracisternal A particles (IAPs, a family of LTR retrotransposons) and an increase in their expression, often causing diseases through their insertion into genes [Barbot et al., 2002]. A similar correlation was also observed in mammals between the hypomethylation of retrotransposons in germ cells in the early stages of development, in which these transposons are active, and their hypermethylation in somatic cells, in which they cannot be mobilized [Kazazian, 2004]. Also interesting is the increase in DNA methylation of repeats observed in several metazoans with the increase in genome size [Jabbari et al., 1997; Lechner et al., 2013] and a significant level of methylation found in another deuterostome, the echinoderm Strongylocentrotus purpuratus [Regev et al., 1998].

This mechanism, which in some cases may be influenced by sRNAs [Wheeler, 2013; Yi et al., 2014], appears to be less generalized, especially in animals. Indeed, repeated elements are generally highly methylated in plants, yeast, and vertebrates but are less methylated in protostomes, especially in insects [Albalat et al., 2012]. In this class the landscape is very variable: a clear correlation between genome size and methylation levels was not found, and species with a large genome and higher levels of genomic parasites, such as Culex, display a low level of methylation [Regev et al., 1998]. In Crassostrea it has been observed that only certain classes of repetitive elements represent methylation targets [Gonzalez and Petrov, 2009] and, unlike in vertebrates, methylation would not be the prevalent mechanism in silencing of genes located within the transposons, but it would act mainly at an intergenic level [Riviere, 2014].

Although interference of sRNAs and methylation are the main mechanisms that regulate transposon activity, other processes have been identified, such as the inhibition of transposition (especially retrotransposition) by cytosine deaminase [Dutko et al., 2005; Stenglein and Harris, 2006] or by factors involved in DNA repair [Curcio and Garfinkel, 1999; Bryk et al., 2001], even if their influence seems to be less relevant.

However, in some instances transposons could be able to activate mechanisms to neutralize the action of sRNAs and methylation, causing an increase in and expansion of the genome. This would be done mainly through the action of environmental stressors [Vitte and Panaud, 2005; Kelly and Leitch, 2011; Nosaka et al., 2012; Wheeler, 2013; Piacentini et al., 2014]. An example was observed in Antirrhinum majus where a cold-induced hypomethylation favored the transposition of Tm3 transposons [Ito, 2013].

As previously mentioned, the genome size is the result of a balance between insertion and loss of DNA sequences, especially transposons. As well as the mechanisms of transposon amplification, deletion of sequences is not accidental. A key mechanism in the control of genome size would be the so-called indel bias [Petrov, 2001; Gregory, 2004], that is the ratio of the levels of insertions to the removal of the ectopic sequences. The indels are divided into small (1-30 bp) and large (involving thousands of bases) [Sun et al., 2012b]. The mechanisms of insertion and DNA loss are different. In the genome, gene sequences are always spaced by stretches of more or less long non-genic sequences (introns and intergenic sequences). Therefore, very long deletions would bear the risk of eliminating gene sequences. This favors small deletions, which would then be the main cause of DNA loss and genome size decrease [Petrov, 2002a]. A clear negative correlation between the rate of DNA loss through small deletions and genome size is observed when comparing insects with differently sized genomes: D. melanogaster (0.16 pg/N), the Luapala cricket (1.93 pg/N), and the locust Podisma (16.99 pg/N) have a rate of DNA deletion inversely related to their respective genome sizes: in Podisma the loss of DNA is much slower than even that of man (3.5 pg/N) [Petrov et al., 2000; Bensasson et al., 2001; Petrov, 2002b]. Moreover, in some mosquito species the genome shows an inverse correlation with the DNA loss rate [Chen et al., 2015]. Similar situations have been recorded also in vertebrates. In archosaurs the indel rate is higher in birds than in turtles and crocodiles, whose transposon percentage is higher [Green et al., 2014], and in mammals the deletion rate of pseudogenes is lower in humans than in rodents [Graur et al., 1989; Ophir and Graur, 1997]. The indel bias might also explain the huge genomes of salamanders and perhaps also of lungfishes. Indeed, a study of the DNA loss in 5 species of salamanders has shown that the rate in these species is much lower than that found in species belonging to 5 classes of non-urodelian vertebrates [Sun et al. , 2012b; Frahry et al., 2015]. In all these cases the different levels of deletion would depend more on differences in the size of the elided sequences than on their frequency [Bensasson et al., 2001; Sun et al., 2012b].

From a comparison between 2 species of pufferfish and between mouse and man it was noted that in the 2 tetraodontids the loss of DNA is higher than in the other 2 species harboring larger genomes, and that this loss mainly concerns sequences at the level of introns [Loh et al., 2008]. Even in chicken, indels primarily affect introns, while they appear highly reduced (if not absent) in intergenic regions [Rao et al., 2010].

The gain of sequences may take place either through small insertions or through large insertions, and the latter occurrence is in line with the expansion of the genome due to rapid insertions and amplifications, typical for transposons [Petrov, 2002a; Dufresne and Jeffery, 2011].

These mechanisms that control the activity of transposons and the balance between insertion and loss of ectopic sequences clearly explain the changes in genome size within low to intermediate values, but they do not completely explain the appearance and preservation of huge genomes such as those of lungfishes and salamanders. As previously mentioned, some events of transposon amplification and the resulting huge genomes may represent a mechanism stimulated by environmental stress that would provide evolutionary advantages [Piacentini et al., 2014]. However, unlike what we see in species harboring smaller genomes, in the genomes that exceed certain high values a clear and direct relationship between the increase in transposon percentage and the increase in DNA content does not seem to exist. Moreover, the increase in DNA content also depends on an amount of non-coding sequences whose repetition is no longer detectable [Kidwell, 2002; Metcalfe and Casane, 2013]. A possible reason for this occurrence could depend on the different insertion sites of the transposons into the host genome [Zhang et al., 2011]. Transposons inserting within or near genes have a mutagenic effect or may affect gene expression, being detrimental to the host genome, and therefore are subject to purifying selection and more frequent deletions [Metcalfe and Casane, 2013; Shen et al., 2013; Lee and Kim, 2014]. This occurrence is supported by the fact that in many organisms the level of indels is higher in introns than in intergenic sequences [Loh et al., 2008; Rao et al., 2010; Lee and Kim, 2014]. Conversely, transposons inserted in regions far from the genes and with low recombination, like intergenic regions and telomeric or centromeric heterochromatin, are less subject to deletions [Kidwell, 2002; Rao et al., 2010] and therefore tend to be conserved and can progressively accumulate mutations losing their repetitivity. Indeed, for example, in Neoceratodus, where most of the transposons belong to non-LTR L2 and CR1 families (present in all classes of vertebrates) [Chalopin et al., 2015], traces of CR1 and other ancient transposons that have largely changed over time were recognizable in the single-copy fraction of DNA [Sirijovski et al., 2005]. A similar situation is found also in some species of salamanders [Metcalfe et al., 2012; Metcalfe and Casane, 2013; Sun and Mueller, 2014] and in the orthopteran Podisma, a species displaying a genome much bigger than Drosophila and where a large proportion of the DNA excess derives from the accumulation of mutations in older pseudogenes [Bensasson et al., 2001].

Although transposon activity and changes in genome size cannot be considered the main drivers of evolution, they may have various relevant effects on certain evolutionary processes. Some of those effects depend on their direct action, while others are mediated by alterations carried out on structural and functional parameters of cells and organisms. One of the direct effects is the influence that transposons may have on genetic variability. Some results suggest that transposons can be considered important factors for the reorganization of the genome through chromosomal rearrangements such as duplications, inversions, and translocations (which have consequences on adaptive phenomena), and also through molecular domestication, a phenomenon giving rise to new coding genes and regulatory elements such as enhancers [Bejerano et al., 2006; Matveev and Okada, 2009; Nakanishi et al., 2012; Piacentini et al., 2014]. The amplification of transposons, especially that resulting from environmental stresses, enables a rapid accumulation of mutations that cause an increase in variability and create the basis for speciation [Piacentini et al., 2014]. An example for this situation is the adaptive radiation of vespertilionids, which corresponds to a burst of activity of transposons [Platt et al., 2014].

However, as it has been previously noted, although transposons can provide evolutionary advantages to the host, in most cases their activity has detrimental consequences [Caceres et al., 2001; Kidwell, 2002; Johnson, 2007; Dufresne and Jeffery, 2011; Chénais et al., 2012; Sun et al., 2012b; Lee and Kim, 2014], and various experimental data suggest that their effect on the rise of genetic variation and speciation might be limited to species with smaller genomes. These correlations between genome size and genetic variability appear to have generally negative effects on the evolutionary processes such as taxonomic diversity, speciation, and extinction. In many animal taxa an inverse correlation was noted between the level of speciation and genome size [Vinogradov, 2004; Kraaijeveld, 2010] and among taxonomic levels of variability, genome size, and proportion of repetitive DNA [Olmo, 2006; Kraaijeveld, 2010]. In this regard, an inverse correlation between genome size and the level of heterozygosity was observed in teleosts [Yi and Streelman, 2005] and urodeles [Pierce and Mitton, 1980]. A similar correlation was also observed between genome size and chiasma frequency in several vertebrates [Olmo et al., 1989; Peterson et al., 1994] and between chromosome changing rate and genome size in reptiles [Olmo, 2005]. Adaptive radiations observed in various vertebrate taxa, such as saurischians [Organ et al., 2007], hummingbirds [Gregory et al., 2009], pufferfishes [Volff et al., 2003], and plethodontids [Kozak et al., 2006], coincided with significant reductions in genome size followed by bursts of morphological diversification. Moreover, freshwater teleost species harboring larger genomes have a lower level of variability compared to marine species with smaller genomes [Hardie and Hebert, 2004]. Furthermore, in insects it was also noted that only those with a genome <2 pg/N have a large increase in taxonomic diversity, while in clades with larger genomes the variability is much lower [Kraaijeveld, 2010]. A final negative consequence of the increase in genome size seems to be also an increase in endangered species and the related risk of extinction [Vinogradov, 2004].

In the light of these observations the main problem to solve is whether the appearance of large genomes only depends on the tolerance of some species to the accumulation of non-coding DNA, especially transposons, or whether large genomes can provide some evolutionary advantages.

A possible answer to this question may come from an analysis of the consequences of genome amplification on some structural and functional parameters of the cell, which in turn affect the morphological and functional characteristics of the organism exposed to natural selection.

Two important cellular parameters affected by genome size are the duration of the cell cycle and the size of the nucleus and the cell.

Although studies on the relationship between the amount of DNA and the cell cycle length are limited both in plants and in animals, a strong direct correlation has been found between the genome size and the S phase and to a lesser extent with the length of mitosis, while there seems to be no correlation with the duration of the G1 and G2 phases [Nagl, 1974a, b; Grosset and Odartchenko, 1975a, b; Horner and MacGregor, 1983; Vinogradov, 1999b; Simova and Harben, 2012].

Correlations between genome size, nucleus, and cell sizes were noted in early cellular studies that led to the formulation of the concept of the nucleoplasmic ratio [Gregory, 2005]. Overall, although the size of the nucleus and the cell are determined by genetic factors [Cavalier-Smith, 2005] and are also influenced by certain phases of the cell cycle, in different species a direct and positive correlation between genome size and various nucleus and cell morphometric parameters (volume and surface) [Szarski, 1968, 1970, 1976; Olmo and Morescalchi, 1975, 1978; Kuramoto, 1981; Olmo and Odierna, 1982; Olmo, 1983; Gregory, 2005; Mueller et al., 2008; Simova and Harben, 2012] and an inverse correlation between genome size and the surface/volume ratio of the cell were noted [Olmo and Morescalchi, 1975, 1978; Olmo and Odierna, 1982; Olmo, 1983].

An important evolutionary interaction is the influence that the genome size has on the duration of embryonic and larval development through the dimensions of the cell and the duration of the cell cycle, especially of the S phase, the extent of which is 50× shorter in embryonic cells compared to somatic cells [Callan, 1972]. Moreover, it should be remembered that during early development the G1 and G2 phases are extremely short and should not affect the total cycle duration [Tang, 2010].

An inverse correlation between genome size and duration of the development has been noted in some invertebrates and anamniote vertebrates, especially amphibians [Goin et al., 1968; Oeldorf et al., 1978; Horner and MacGregor, 1983; Jockusch, 1997; Gregory, 2002a], in which also an inverse correlation between genome size, growth rate, and differentiation rate in the process of regeneration has been noted [Sessions and Larson, 1987]. Conversely, in amniote vertebrates any correlation between genome size and parameters of development was not noticed [Olmo, 2003; Gregory, 2002a, b, 2005].

The evolutionary importance of the developmental and larval period is evident in frogs and salamanders, where species, reproducing in temporary pools and having a very rapid embryonic and larval development that allows them to adapt to arid environments, have very small genome and cell sizes and a very short cell cycle. Conversely, large genomes and cell sizes are present in species living in water-rich but colder environments and having longer embryonic and larval periods, hindering metamorphosis during unfavorable periods of the year [Goin et al., 1968; Oeldorf et al., 1978].

Increasing the genome size as well as the duration of the development could have an impact also on the developmental complexity [Gregory, 2002a, 2005]. In insects the increase in DNA content determined the gradual transition from species with complete metamorphosis (holometabolous, limited to species with <2 pg/N) to species with incomplete (hemimatabolous) or absent metamorphosis (ametabolous). In urodeles the increase in the genome allowed the transition from a biphasic life cycle to increasingly frequent cases of paedomorphosis or even to obliged neoteny [Gregory, 2002a, 2005]. In plethodontids it was observed that the ancestral forms probably had a shorter larval period than current forms [Bonett et al., 2014], implying a co-evolution between genome increase and paedomorphosis/neoteny. The progressive development of paedomorphosis and neoteny characterizes also the evolution of the current lungfishes harboring very large genomes [Joss, 2006].

Another important evolutionary parameter related to genome and cell size is the metabolic rate. In many animals an inverse relationship between genome/cell size and metabolic rate was noted [Licht and Lowcock, 1991; Vinogradov, 1995, 1997; Gregory and Hebert, 1999; Gregory, 2003, 2005], even if a precise relationship was determined between cell size and metabolic rate [Monnickendam and Balls, 1973; Szarski, 1976; Starostova et al., 2009], and in some cases no precise correlation was indicated with the genome size [Starostovà et al., 2009]. A particularly influential parameter of the metabolic rate, especially in larger cells, is the cell surface/volume ratio, which limits the exchange of nutrients and gases between the cell and the surrounding environment [Szarski, 1976; Olmo, 1983, 2003]. Also important are the parameters that govern the exchanges between the nucleus and cytoplasm such as the nucleoplasmic ratio [Szarski, 1976], the nuclear surface/volume ratio, and the frequency of the nuclear pores [Olmo, 1983].

A particular consequence of the metabolic level has been hypothesized to be the ability to fly [Hughes and Hughes, 1995]. In birds stronger flyers have genomes significantly smaller than weak flyers and flightless birds. It has been also hypothesized that the reduction in genome size was necessary to acquire the high metabolic levels essential for the flight [Hughes and Hughes, 1995; Hughes, 1999], a reduction depending on the shortening of intron lengths and of other sequences such as transposons [Hughes and Hughes, 1995]. A similar situation was also described in bats, which possess the smallest genomes among mammals and also the shortest introns [Van den Bussche et al., 1995; Gregory, 2005; Zhang and Edwards, 2012]. However, the correlations between small genomes, small cells, and high metabolic rates are not related only to the ability to fly, because it was noted that a reduction in cell size and length of introns had occurred within the archosaurian lineage from which birds originated long before the first flying birds [Waltari and Edwards, 2002; Organ et al., 2007].

Flying insects achieve the highest known mass-specific rates of O₂ consumption in the animal kingdom [Suarez, 2000]. Except for the orthopterans, they have the lowest genome sizes among animals and similarly show a reduction in the length of introns within smaller genomes [Wang et al., 2014]. However, there is no difference in the correlation between the genome size and the metabolic level among orthopterans and groups harboring smaller genomes.

Other correlations were assumed between genome size and other evolutionarily important functional parameters such as the control of the extra/intra-cellular solute composition [Vinogradov, 1998], which, however, does not seem sufficiently supported by the experimental point of view. It is important to note that all the above-mentioned correlations indicate that variations in the dimensions and in the percentage of genomic non-coding DNA are also indirectly subject to selective pressures through their effects on the cell and on the morphometric and functional parameters in the organism.

Although causes and consequences of the interaction between transposon activity and expansion of the genome are demonstrated, their direct or indirect influence on the evolutionary processes, as well as the causes of the so-called C-value enigma are not fully elucidated, but some points seem to be quite defined.

Given their ability to autonomously expand, transposons certainly represent the main cause of the increase in the genome. Their influence would, however, be only quantitative, since it is not possible to notice an equal correlation between genome size and variety of transposons in all eukaryotes.

The current levels of genome size found in many eukaryotes are the results of a balance between the activity of transposons and selective pressures acting on various levels:

• a first pressure on a genomic level due to the defense mechanisms put into practice by the cells to limit any harmful action of transposons;

• a selection pressure aimed at maintaining optimal levels of structural and functional parameters within cells and the organism;

• a selective pressure acting on the effects that changes at the genomic and cytological level have had on morphological and functional characteristics which are important for the adaptation to different environmental conditions.

Relevant genome increasing occurred at particular times of evolution, such as the conquest of the land or adaptation to extreme environments, e.g. the deep sea or very cold areas, due to burst amplification of transposons stimulated by environmental stresses.