The "tree of life" represents the genealogy history of species and the "tree" with no doubt is metaphysically prior to any particular species concept [1]. As there is no consensus concerning the meaning of the term ‘species’, we choose practical one from the others, for the purpose of inferring relationships of Xenacoelomorpha with Metazoans.
There are two commonly used concept ill-named ‘biological species concepts’(BSC)and ‘phylogenetic species concepts’ (PSC). BSC defines species as “groups of interbreeding natural populations that are reproductively isolated from other such groups.” and PSC –“A species is the smallest diagnosable cluster of individual organisms within which there is a parental pattern of ancestry and descent.” [2,3]. The former needs in the filed endeavor, but the later bends us to take advantage of the dramatic progressing genome data size predicting phylogenetic position of Xenacoelomorpha. Choosing PSC to predict species genealogy by using gene genealogy is the main logic of this article.
The phylogenetic species concept includes three levels: genes, organisms and species level. Same epistemology reminds us to take coalesced homolog genes genealogy history as the representative of both organisms and species genealogy once the picked genes is “majority enough” and their "tree" topology is in concordance with each other [1]. Obviously, there is no way currently to get the single objective tree with existing species [4], because of the evolutionary speciation events [5-8] and genes fundamental but not totally sufficent "bearing alone" roles in species evolution history. Practically, the prediction should be compatible with job purpose and the methodology should be self-adapting.
The methodology compatible with PSC includes three steps: 1. homology assessment to get an data matrix; 2. inferring species tree by genes trees’ topology; 3. methodological self-adapting [9]. We can not deal all three steps simultaneously because of computational limitation. Thus, the so called "majority enough" corresponding to the first step claims for enough data source and sufficient number of homolog genes for concatenating, considering biological phenomenon such as Lateral Gene Transfer (LGT), in order to represent species genealogy. Unfortunately, current computational methods in step one using nuclear genome or transcriptome data is inadequate to fulfil the “majority criteria” [10,11,12,13] finding the position of Xenacoelomorpha in the Metazoans concordance tree, regardless of a hundred or a thousand of genes concatenated, the alignment matrix is far less than the propotion of LGT alone in human nuclear genome (8% of total nuclear genes) [7]. Let alone other non-phylogenetic signals existing and the never proven “common history” of most nuclear genes [13].
Nevertheless, the mitochondrial genome as data source meets the “majority criteria” [1] using present alignment method though not 100% of genes in this kind of plasmon being included. Furthermore, with particularly uniparental reproduction in Metazoan [14], independent adaptive evolution rates [15], conservative biosynthesis function been selected by environment, physical segregated from nuclear genome [16], mitochondrial genes genealogy shall represent the genealogy of related species at a certain extent, which can be depicted as ‘primary concordance tree’ of Metazoan, to position Xenacoelomorpha.
The position of Xenacoelomorpha in the tree of life remains a unresolved question
Morphological studies previously suggest Xenoturbella Bocki as a turbellarian flatworm in 1949 and in 1997 position Xenoturbella either as a basal bilateria or with molluscan [17-21]. we don’t discuss their result in this article as PSC we chosen.
Molecular studies based primarily on nuclear genes place Xenoturbellida as sistergroup of the Ambulacraria (echinoderms + hemichordates) [2,3,12], a basal Deuterostomes [12,23], or sistergroup to all other bilateria [10,11,24]. Using mitochondrial sequences prefer to support a basal Deuterostome placement [12,26] .
We briefly summarise their reasult and find out that the prediction based on nuclear genes can not congruent with each other. Although some problem has been resolved, such as sequence contaminated [27,28], rRNA compositional bias [10,11,29], or topolgy misleading by inadequate choice of other Metazoan clades [10,11,12,22] but their limited alignment matrix are not majority enough to represent nuclear genome as we discussed above. Conversely, most mitochondiral genome analyses leads to congruent prediction but being questioned by their unrepeatable results with low statistical support. Both nuclear and mitochondria data suggest to group Xenoturbella with Acoelomorph as a clade named Xenacoelomorpha.
Phylogenomics inference from mitochondrial sequences to position Xenacoelomorpha always been complicated by several misleading factors such as compositional heterogeneity [30], Long Branch Attraction (LBA) [9] and accelerated substitution rates, which violate the models of evolution. Each of them have complex source of cause and can not been absolutely separated from each other. Unfortunately, useful approaches has been used to alleviate system error but barely satisfactory.
In 2006, Bourlat SJ aimed to study the relationships among all Deuterostome groups with mitochondrial genome, leading to a result that Xenoturbella is a sister group of Ambulacraria but can not group Cephalochordate with Echinoderm. The global topology error probably caused by lack of phylogenetic signal within 170 nuclear proteins and dissatisfactory choice of species–the misplacing of Branchiostoma sp. among different datasets analyses [22].
In 2007 Marleen Perseke expanded Bourlat SJ’s dataset and using Phylobayes CAT model to alleviate compositional bias getting a position that Xenoturbella is an early branch of Deuterostome with unreliable PP support, which claims for far more suited model and improved alignment matrix either [25].
Later, Bourlat SJ carried out Bayesian analyse with mtZOA leading to a basal placement of Xenoturbella within Deuterostome but the prediction of other clades in the same tree being fiercely criticized–grouping the urochordates with the echinoderms remains enigmatic in his study [26].
In 2011 Kevin J. Peterson’s team conducting 37 well chosen species, mainly because of their low sequence saturation, supporting Xenacoelomorpha as a basal Deuterostomes. His arguing about grouping Acoelomorph and Xenoturbella together is generally accepted. Meanwhile, the low support of internode between Chordata and Ambulacraria, the compositional heterogenoity of mitochondrial genome and “slightly hand made” alignments make his work unrepeatable [10-12].
In 2017, Helen E. Robertson inferring phylogenetic tree from mitochondrial genes retrieves Xenacoelomorpha as an early branch of Deuterostome but the statistical support for his prediction is week, probably caused by dissatisfactory choice of species [23].
Here, we have first illustrated the philosophical foundation why it is necessary to predicting Xenacoelomorpha’s genealogy with mitochondrial genome instead of a bag of nuclear genes. Afterwards, we briefly surveyed the relevant researches related to Xenacoelomorpha’s phylogenetic position. In the next section, our analysis, inheriting the achievements of predecessors, using currently more suitable alignment methods to avoid “hand made”, building varies matrix with filtered species directly accounts for compositional heterogeneity, excluding outgroups to elaborate LBA, leads to a robust consensus topology which hold the view that Xenacoelomorpha flatworms are basal Deuterostome. In sharp contrast, inferring phylogenetic position of Xenacoelomorpha by nuclear datasets is hazardous with LBA menace and can not fill the gap between methodology and philosophy.