Formica picea and F. candida (Hymenoptera: Formicidae): Synonyms or Two Species?

: Ants of the « Formica picea — F. candida » complex are widespread across Eurasia. However, it is still a matter of debate if it constitutes one or two species. In this study, we collected a sample of specimens from different parts of Eurasia, sequenced the mitochondrial cox1 and cytb genes, as well as three nuclear loci— wg , Top1


Introduction
The taxonomic position of many widespread and abundant ant species remains uncertain [1,2]. Among them are the species of the «black bog ant» complex, Formica picea Nylander, 1846 and F. candida Smith, 1878. Members of this complex are found from West Europe to Kamchatka, as well as from the valleys of northern rivers to Tibet (Figure 1).

Introduction
The taxonomic position of many widespread and abundant ant species remains uncertain [1,2]. Among them are the species of the «black bog ant» complex, Formica picea Nylander, 1846 and F. candida Smith, 1878. Members of this complex are found from West Europe to Kamchatka, as well as from the valleys of northern rivers to Tibet (Figure 1).    Table 1. Locations refer to Figure 1. Sequences obtained in this study shown in bold; *, terra typica; **, type specimen. If an accession was deposited under another species name, this is acknowledged in parentheses; references to published articles are given whenever possible.
Morphological variations within the complex were studied by Dlussky [11] based on extensive collections. Dlussky [11] concluded that there is no ground to split it into several species. Later on, Bolton [12] suggested that the name F. candida has the priority over F. picea; this was picked up by some-but not all-of the scientists. Seifert [13] suggested that F. picea and F. candida are two separate species, based on the features of pubescence and some morphometric parameters. According to Seifert [13], the distribution of F. picea includes Europe, the Caucasus, and the West Siberian Lowlands, and that of F. candida is in Central Asia, the Altai Mountains, and East Siberia. The habitats of F. picea include the bogs and European mountains above the forest belt, while F. candida lives in diverse habitats, including river floodplains and dry steppes of Asia.
However, Zakharov et al. [14] refuted Seifert's [13] conclusion, stating that the variation of pubescence within the European populations of F. picea is higher than the proposed differences between the species. They also stated that there is a smooth gradient of morphological variation from Europe to the Far East. Based on this, the authors returned to treating F. picea and F. candida as synonyms.
Molecular genetic analysis is increasingly used to resolve problematic questions in ant taxonomy, due to its high resolution and its potential to detect cryptic taxa. [15] There were several molecular genetic attempts to resolve the relationships in the «F. picea-F. candida». Goropashnaya [3] performed a phylogenetic analysis based on the mitochondrial cytb gene. She demonstrated that the complex is split into two clades: one restricted to Europe and West Siberia and the other, to Central Asia. Later Goropashnaya et al. referred to these clades as a pair of species [5]. Antonov and Bukin [16] studied a sample of cytb sequences of specimens from several regions of the Palearctic identified as F. candida and F. picea. They concluded that genetic distances within F. candida are higher than those between the two species. Important datasets were also obtained by Chen et al. [7] and Schär et al. [4], based on the mitochondrial cox1 gene.
Therefore, the status of the «F. picea-F. candida» species complex remains uncertain. The problem with molecular data is that very similar sequences can be deposited under different species names depending on the authors' viewpoint on this matter. Moreover, some authors sequenced the cytb gene, and the others sequenced cox1, so the resulting samples cannot be compared directly.
We collected a sample of the «F. picea-F. candida» complex specimens from the Northern Palearctic. Moreover, we studied the diversity, geographic and habitat distribution, and ecology of its population from the north of the Russian Far East [17]. The most important specimens were identified by Dlussky and Radchenko, who are specialists in the genus Formica; both consider this complex a single species (Radchenko: [18], personal communication). Here, we attempt to resolve the issue on the «F. picea-F. candida» complex using molecular analysis, based on the mitochondrial cox1 and cytb genes, as well as three nuclear loci, with the integration of the available GenBank sequences obtained by other authors.

Materials and Methods
We collected a set of Formica specimens from Eurasia ( Figure 1, Table 1). DNA was isolated from single ethanol-fixed worker ants using the commercial silica columns (BioSilica, Novosibirsk, Russia), as described in [4]. A fragment of the mitochondrial cytochrome c oxidase subunit 1 (cox1) gene was amplified using the universal primers LCO1490m The obtained DNA fragments were visualized by agarose gel electrophoresis; unincorporated primers and nucleotide phosphates were removed using the shrimp alkaline phosphatase/E. coli exonuclease I mix (New England Biolabs, Ipswich, MA, USA). Sanger sequencing was performed on a 3130xl DNA Analyzer (Applied Biosystems, Framingham, MA, USA) in the SB RAS Genomics Core Facility (ICBFM SB RAS, Novosibirsk, Russia) using both forward and reverse primers. The obtained sequences were deposited in GenBank under accession numbers ON228270-ON228284, ON220872-ON220887, and ON862896-ON862909. Sequences obtained by other authors were also used in this study (see Table 1 for GenBank accessions). This dataset includes the cytb sequence for the F. candida neotype (AY786154).

Results
We sequenced fragments of the mitochondrial cox1 and cytb loci for our sample (Table 1) and constructed phylogenetic trees. The final alignment for cox1 contained 593 bp; for cytb, 693 bp. On the cytb tree ( Figure 2) the «F. picea-F. candida» complex was split into two big clades. Nucleotide distances between them were as large as those among other Formica species. One of the clades included the specimens from Kyrgyzstan, Europe, the Urals, the Caucasus, and Tibet. The sequences from Kyrgyzstan that included the F. candida neotype fell into a well-supported subclade. The rest of the specimens (shown on Figure 2 as F. picea) formed a group with no statistical support. This group included F. picea (including a specimen from southern Finland, the terra typica of F. picea), as well as those submitted to GenBank as F. candida. The second clade (shown on Figure 2 as Formica sp.) contained the specimens from West and East Siberia, the Russian Far East, China, and Korea. For cytb, the average p-distances between F. candida and F. picea were 1.34%; between F. picea and Formica sp., 3.25%; between F. candida and Formica sp., 3.35%. The mean p-distance within F. picea was 0.81%; within F. candida, 0.06%; within Formica sp., 1.05%.
The tree constructed using the cox1 gene ( Figure 3) was built on more specimens than the cytb tree. However, there were no data for Kyrgyzstan, where the neotype of F. candida was described, or other regions of Central Asia, and there were few F. picea specimens. On the cox1 tree, the «F. picea-F. candida» complex was also found to contain two clades: one corresponding to F. picea (including terra typica, i.e., the region where the original type specimens were collected) and the other to Formica sp. The average p-distances between F. picea and Formica sp. were 2.52%; within the species, they were 0.57 and 0.38%, respectively. We should also note that the cox1 tree contained as many as three clades, with specimens identified as F. gagatoides.
The final alignment for the wg locus included 312 bp; for Tpo1, 717 bp; for ITS2, 678 bp. Sequences of wg and TPO of F. picea, Formica sp., and F. gagatoides were identical, except for some degenerate positions. In the ITS sequence, F. picea and Formica sp. could be distinguished by one A<>G substitution in position 216 of the alignment, while these two taxa differed from F. gagatoides by two substitutions and one short indel.

Discussion
In this study, we analyzed the dataset of the «F. picea-F. candida» complex in order to find out if F. picea and F. candida are synonyms or two distinct species. The results were unexpected. The sample, indeed, contained two clades: one with F. candida and F. picea specimens and the other representing a previously unknown taxon (Figure 2). For mitochondrial sequences, genetic distances between these clades were as high as between the well-established species of the genus Formica. This is in line with earlier findings that recovered multiple cryptic species within the genus Formica [24,25], as well as other wellstudied genera, such as Pheidole [26,27], Myrmica [28,29], Lasius and Cardiocondyla [24], etc. For the nuclear sequences, however, the differences were not as high: wg and Tpo1 gene fragments were identical, and only one substitution distinguished F. picea from Formica sp. On the other hand, F. gagatoides was also very close to these taxa, differing from F. picea by two substitutions and one indel in ITS2, with two other genes being identical. It is well known that nuclear DNA has a lower substitution rate compared to the mitochondrial genome, and closely related ant species may have few distinctive positions or none at all. From the data obtained in [4], one can see that there are very few differences in the nuclear sequences of the whole F. picea/F. gagatoides/F. lemani/F. fusca/F. neorufarbis group. This is the usual situation that is found in ants, e.g., in the well-studied Formica rufa group, several species acknowledged by myrmecologists cannot be distinguished by nuclear markers and even show significant interspecific hybridization [30,31]. A similar case can also be observed in the genus Proformica [32]. Thus, we can conclude that the Asian Formica sp. is apparently distinct genetically from F. picea and F. candida, but it is unclear whether it should be regarded as a separate species or as a subspecies.
F. candida was initially described by Smith [33] based on a single specimen. This specimen was collected "On the road across the Pamir, from Sarikol to Panja", which is probably somewhere near the easternmost part of China, on the border with Tajikistan. According to [13], this specimen was lost, so he fixed a neotype from Kyrgyzstan. This location is quite remote (about 315 km straight line distance in a mountainous area) from the original one. Our sample contained a cytb sequence of the neotype (AY786154). We can state that, so far, the F. candida clade is limited to the available specimens from Kyrgyzstan and was not found elsewhere.The type specimen of F. picea is old and highly damaged [13]. It was collected in the vicinity of Helsingfors (currently Helsinki, Finland) with no precise information on its whereabouts. Both locations available in our sample are several dozen km from Helsinki. Given the fact that no other group of the complex was found in Finland, we can suggest that these accessions represent the «real» F. picea.
Whether F. picea and F. candida are reciprocally monophyletic remains an open question. On our cytb tree F. picea was not supported as a clade (Figure 2), but more data are needed to verify that.
Based on the data on the type specimen origin, we cannot assign the Formica sp. clade to either F. picea or F. candida and have to conclude that it represents a new taxon that is still to be described. From the maps, we can see that all three ant taxa have distinct distributions: Formica sp. is found in Asian Russia, Siberia, and East Asia; F. candida, in Central Asia; F. picea, in Europe, with a single finding in Tibet. As suggested by multiple authors, morphological differences between these taxa are elusive [11,13,18]. There might be some differences in ecological preferences of these species: the European F. picea is found in swamps or in the alpine belt of mountains. The Asian Formica sp. is considered to be eurytopic in Siberia, while in the Russian Far East, it is found mostly in floodplains ( [11] and our observations). However, it remains to be proved that these ecological differences coincide with the distribution boundaries of these species.

Conclusions
By trying to resolve the relationships within the «F. candida-F. picea» complex, in order to separate these two species, we found a new third member of this complex. It is important to note that genetic distances between F. candida and F. picea are significantly smaller than between them and the newly found taxon. The latter seems to not be the only cryptic taxon in this group: we obtained preliminary evidence that F. gagatoides might also represent as many as three cryptic taxa.