ArticlePDF Available

18S rRNA variability maps reveal three highly divergent, conserved motifs within Rotifera

December 2021

License
CC BY 4.0

Authors:

Carl von Ossietzky Universität Oldenburg

Background 18S rRNA is a major component of the small subunit of the eukaryotic ribosome and an important phylogenetic marker for many groups, often to the point of being the only marker available for some. A core structure across eukaryotes exists for this molecule that can help to inform about its evolution in different groups. Using an alignment of 18S rDNA for Rotifera as traditionally recognized (=Bdelloidea, Monogononta, and Seisonacea, but not Acanthocephala), I fitted sequences for three exemplar species ( Adineta vaga , Brachionus plicatilis , and Seison nebaliae , respectively) to the core structure and used these maps to reveal patterns of evolution for the remainder of this diverse group of microscopic animals. Results The obtained variability maps of the 18S rRNA molecule revealed a pattern of high diversity among the three major rotifer clades coupled with strong conservation within each of bdelloids and monogononts. A majority of individual sites (ca. 60%) were constant even across rotifers as a whole with variable sites showing only intermediate rates of evolution. Although the three structural maps each showed good agreement with the inferred core structure for eukaryotic 18S rRNA and so were highly similar to one another at the secondary and tertiary levels, the overall pattern is of three highly distinct, but conserved motifs within the group at the primary sequence level. A novel finding was that of a variably expressed deletion at the 3' end of the V3 hypervariable region among some bdelloid species that occasionally extended into and included the pseudoknot structure following this region as well as the central “square” of the 18S rRNA molecule. Compared to other groups, levels of variation and rates of evolution for 18S rRNA in Rotifera roughly matched those for Gastropoda and Acanthocephala, despite increasing evidence for the latter being a clade within Rotifera. Conclusions The lack of comparative data for comparable groups makes interpretation of the results (i.e., very low variation within each of the three major rotifer clades, but high variation between them) and their potential novelty difficult. However, these findings in combination with the high morphological diversity within rotifers potentially help to explain why no clear consensus has been reached to date with regard to the phylogenetic relationships among the major groups.

Available via license: CC BY 4.0

Content may be subject to copyright.

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

https://doi.org/10.1186/s12862‑021‑01845‑2

RESEARCH ARTICLE

18S rRNA variability maps reveal three highly

divergent, conserved motifs withinRotifera

Olaf R. P. Bininda‑Emonds*

Abstract

Background: 18S rRNA is a major component of the small subunit of the eukaryotic ribosome and an important

phylogenetic marker for many groups, often to the point of being the only marker available for some. A core structure

across eukaryotes exists for this molecule that can help to inform about its evolution in diﬀerent groups. Using an

alignment of 18S rDNA for Rotifera as traditionally recognized (=Bdelloidea, Monogononta, and Seisonacea, but not

Acanthocephala), I ﬁtted sequences for three exemplar species (Adineta vaga, Brachionus plicatilis, and Seison nebaliae,

respectively) to the core structure and used these maps to reveal patterns of evolution for the remainder of this

diverse group of microscopic animals.

Results: The obtained variability maps of the 18S rRNA molecule revealed a pattern of high diversity among the

three major rotifer clades coupled with strong conservation within each of bdelloids and monogononts. A majority of

individual sites (ca. 60%) were constant even across rotifers as a whole with variable sites showing only intermediate

rates of evolution. Although the three structural maps each showed good agreement with the inferred core structure

for eukaryotic 18S rRNA and so were highly similar to one another at the secondary and tertiary levels, the overall pat‑

tern is of three highly distinct, but conserved motifs within the group at the primary sequence level. A novel ﬁnding

was that of a variably expressed deletion at the 3’ end of the V3 hypervariable region among some bdelloid species

that occasionally extended into and included the pseudoknot structure following this region as well as the central

“square” of the 18S rRNA molecule. Compared to other groups, levels of variation and rates of evolution for 18S rRNA

in Rotifera roughly matched those for Gastropoda and Acanthocephala, despite increasing evidence for the latter

being a clade within Rotifera.

Conclusions: The lack of comparative data for comparable groups makes interpretation of the results (i.e., very low

variation within each of the three major rotifer clades, but high variation between them) and their potential novelty

diﬃcult. However, these ﬁndings in combination with the high morphological diversity within rotifers potentially help

to explain why no clear consensus has been reached to date with regard to the phylogenetic relationships among the

major groups.

Keywords: Evolution, Phylogeny, Conservation, Hypervariable, Monogononta, Bdelloidea, Seisonacea,

Acanthocephala, Deletion, Rate of evolution

permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the

original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or

other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line

to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory

regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this

licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco

mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Background

Together with numerous ribosomal proteins, 18S rRNA

forms a major component of the small subunit of the

eukaryotic ribosome. e single stranded RNA mol-

ecule itself has a characteristic and complicated sec-

ondary structure [see 1, 2], whereby it repeatedly folds

back upon itself, with the resultant base-pairing or lack

Open Access

BMC Ecology and Evolution

*Correspondence: olaf.bininda@uni‑oldenburg.de

AG Systematics and Evolutionary Biology, IBU–Faculty V, Carl von

Ossietzky Universität Oldenburg, Carl von Ossietzky Strasse 9–11,

26111 Oldenburg, Germany

Page 2 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

thereof creating stems and loops, respectively. Two or

more stem-loop regions can also combine to form one

of 14 diﬀerent types of three-dimensional pseudoknot

[3], thereby contributing to the tertiary structure of the

molecule. In addition, eukaryotes share nine homolo-

gous regions in the molecule (and which are also pre-

sent in the prokaryotic homologue 16S rRNA) that are

especially variable and are labelled as the hypervariable

regions V1—V9 [1]. (e eukaryotic V6 region, however,

is noticeably less variable compared to the other regions

[1] and to such a degree that it is not counted among

the hypervariable regions by some authors (e.g., [4])).

e molecule itself is encoded by the 18S rDNA gene,

a distinction that I will maintain throughout this paper

although the terms rRNA are rDNA are often used inter-

changeably in the literature.

e phylogenetic utility of 18S rDNA arguably stems in

part historically from practical considerations. Together

with 5.8S rDNA, 28S rDNA, and two internal and two

external transcribed spacers, it forms an array that is tan-

demly repeated throughout the eukaryotic genome (e.g.,

ca. 300 + copies clustered across ﬁve chromosomes in

humans [5]). e multicopy nature of the gene made it

easier to extract and amplify in the pre-PCR era and con-

certed evolution meant that the many copies are virtually

identical [6, 7], thereby sidestepping questions of paral-

ogy. In addition, 18S rDNA as a gene is found universally

among eukaryotes and possesses conserved ﬂanking

regions that facilitated primer design. However, its prac-

tical utility was augmented by the broad phylogenetic

information content yielded by its structural characteris-

tics, with the slower, more conserved stem regions (due

to the constraints of the base pairing) providing resolu-

tion deeper in the tree to compliment the more recent

information provided by the faster, less conserved loops

and especially by the hypervariable regions. Indeed,

much of the basis for deep phylogenies within Metazoa

and beyond derive from phylogenetic analyses of 18S

rDNA or other rDNA molecules more generally, both

at the sequence and, more recently, the meta-sequence

(i.e., structural) levels (e.g., [8–10]). At the other end of

the spectrum, certain hypervariable regions, or parts

thereof, have been promoted as possible species barcod-

ing regions in diatoms (e.g., [11, 12]) and other “protists”

(e.g., [13]). Although the early promise of 18S rDNA

as “the” phylogenetic marker has not been realized, it

remains one of the most widely sequenced genes across

all organismal groups, especially in a phylogenetic con-

text [14].

Indeed, 18S rDNA is often one of only a few, if not the

only, phylogenetic marker sequenced for a given group.

A case in point is Rotifera, a historically recognized phy-

lum of approximately 2000 named species of microscopic

animals [15] for which the only comprehensive molecular

phylogeny to date encompasses 53 species (plus numer-

ous outgroup species) sequenced for up to four markers

including 18S rDNA as part of a total-evidence analysis

with morphological characters [16]. Although countless

studies based on either morphological or molecular data

conﬁrm the monophyly of the three major rotifer clades–

Bdelloidea (ca. 388 species), Monogononta (ca. 1623

species), and Seisonacea (four species; more commonly

referred to as Seisonidae)–their relationships to one

another remain unclear [see 15], in part because of their

highly distinct natures. Bdelloids are obligate asexuals

(i.e., only parthenogenetic females are known) that can

also undergo anhydrobiosis and whose genome has been

shaped in part by (ancient) gene exchange. Monogon-

onts, by contrast, are facultative asexuals that undergo

cyclic parthenogenesis, possess only a single gonad and

comprise many species presenting dwarf males to various

degrees. Finally, seisonids are obligate sexuals with no

male dwarﬁsm and that live as ectoparasites or commen-

sally on diﬀerent species of the crustacean genus Nebalia,

sometimes with diﬀerent seisonid species living on diﬀer-

ent body parts of the same host [17, 18]. Even the appli-

cation of molecular phylogenetics has failed to ﬁrmly

resolve the relationships of these taxa to one another

[15].

My goal in this paper is not so much to resolve the phy-

logenetic relationships within Rotifera, nor to address

the question as to whether Acanthocephala nest within it

(see “Methods”), but rather to examine rates of molecular

evolution of the 18S rDNA gene within this traditional

grouping based on newly obtained sequences combined

with those present in GenBank. In so doing, however, my

analyses reveal a pattern of highly distinct, yet conserved

motifs among the major clades that might explain why

there has been so little consensus about their interrela-

tionships to date [see 15].

Results

Examination of the uncorrected, average pairwise dis-

tances across groups determined using PAUP* 4.0a166

[19] revealed a pattern (Table1A) whereby each of the

taxonomic groups Bdelloidea, Monogononta, Seisonacea

were all highly distinct from one another as well as from

the platyhelminth outgroup Calicophoron calicopho-

rum (> 14% divergence), but showed little internal varia-

tion (< 3%). is pattern is also reﬂected indirectly in the

increase in the percentage of gaps in the Rotifera align-

ment compared to either of the Bdelloidea or Monogon-

onta alignments (Table2). In addition, sequences from

monogonont species were, on average, slightly more

similar to those of the outgroup than they were to the

remaining rotifer clades. e use of corrected distances

Page 3 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

calculated using a GTR model of evolution revealed the

same pattern (Table1B), with average pairwise distances

being on a par with (within clade comparison) or slightly

higher than (between clade comparisons) the uncor-

rected distances.

e inferred structural maps of the three exemplar roti-

fer species—Adineta vaga (Bdelloidea), Brachionus plica-

tilis (Monogononta), and Seison nebaliae (Seisonacea)

(Fig.1; see also Additional ﬁle1)—each showed a good

ﬁt to the eukaryotic core structure for 18S rRNA pro-

posed by Van de Peer and colleagues [20–22]. Unusual,

however, is that most bdelloids (23 of 29 fully informative

species; potentially the incompletely sequenced species

Otostephanos jolantae and Zelinkiella synaptae as well)

possess a unique and variably expressed deletion that

spans 92 bps comprising the last 14 bps of region V3 and

extends into the non-hypervariable region beyond this

(Fig.2). Six motifs of diﬀerent lengths are present includ-

ing the absence of the deletion and no species expresses

the deletion in its full length (maximum deletion length

is 68bp). Although the deletion begins with a set of four

nucleotides displaying very high relative rates of evolu-

tion (Fig.1A), the patterns of the motifs suggest that the

deletion originates from its 3′ end in a non-hypervariable

region of the 18S rRNA molecule that is otherwise virtu-

ally constant in its sequence composition across Rotifera

and encompasses most, if not all, of the pseudoknot fol-

lowing the V3 region in the eukaryotic core structure. e

deletion is also variably expressed among species within

each of the genera Adineta, Embata, Mnobia, Philodina

and Rotaria (but not Abrochtha, Dissotrocha, or Habro-

trocha) as well as in the species Dissotrocha aculeata,

Dissotrocha macrostyla, Philodina citrina, and Philodina

megalotrocha, with second GenBank sequences for each

of these species (accession numbers JX494743, JX494745,

JX494740, and JX494741, respectively) possessing even

longer deletions. Altogether, this variability at both the

genus and species levels together with the apparent pres-

ence of a more restricted deletion in the monogonont

species Lindia tecusa and Lindia torulosa (Fig.2) would

suggest the convergent evolution of the deletion motifs

barring any sequencing or identiﬁcation errors for these

GenBank sequences.

Average relative TIGER values across all sites were

high, indicating slow rates of evolution (Table3, Fig.3),

largely because > 60% of all sites (i.e., > 1000 bp) with

suﬃcient coverage for the TIGER analyses within a

given data set (Bdelloidea, Monogononta, and all Rotif-

era) were constant. e more homogenous bdelloid and

monogonont data sets showed even greater proportions

of constant sites (85.5% and 75.5%, respectively) and

each also presented slower average relative rates across

sites than did the entire rotifer data set. However, even

the variable sites alone, which would include the hyper-

variable regions, were not unduly fast, with average

Table 1 Average pairwise distances (A, uncorrected p distances ± SE; B, GTR distances ± SE) within (along the diagonal) and between

(below the diagonal) the major rotifer clades determined using PAUP* v4.0a166 [19]

The number of pairwise comparisons is given in parentheses; it can be lower than the theoretical maximum of (n2 − n)/2 because of undened distances (e.g., when

the sequences of a species pair do not overlap). The corresponding average distances within all Rotifera were 0.0701 ± 0.0005 (uncorrected p) and 0.0787 ± 0.0006

(GTR) (each 19,499 comparisons) and between all Rotifera and Calicophoron calicophorum were 0.1539 ± 0.0021 (uncorrected p) and 0.1748 ± 0.0028 (GTR) (each 198

comparisons)

Taxon Bdelloidea Monogononta Seisonacea

Bdelloidea 0.0219 ± 0.0015 (595)

Monogononta 0.1704 ± 0.0002 (5668) 0.0270 ± 0.0001 (13,039)

Seisonacea 0.2226 ± 0.0011 (35) 0.1759 ± 0.0008 (162) n/a (0)

Calicophoron calicophorum 0.2140 ± 0.0015 (35) 0.1406 ± 0.0006 (162) 0.2191 (1)

Bdelloidea 0.0226 ± 0.0008 (595)

Monogononta 0.1969 ± 0.0002 (5668) 0.0278 ± 0.0001 (13,039)

Seisonacea 0.2692 ± 0.0016 (35) 0.2041 ± 0.0010 (162) n/a (0)

Calicophoron calicophorum 0.2556 ± 0.0021 (35) 0.1568 ± 0.0007 (162) 0.2646 (1)

Table 2 Statistics relating to the diﬀerent alignments used in

this study

The percentage of gaps is corrected to ignore terminal gaps so as to better

reect the contribution of indels over potentially incomplete sequences

Data set Aligned length Percentage of gaps

Bdelloidea 1837 3.8% (2204 of 58,247 cells)

Monogononta 1868 3.2% (9077 / 286,203)

Rotifera 2021 10.8% (40,501 / 375,458)

Rotifera plus the outgroup

Calicophoron calicopho-

rum

2175 17.3% (70,540 / 407,479)

Page 4 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

1810

1770

1780

U U C G U A G G U G

CUGCGU

AUC

1790

1800

1640

1650

G U

U G G U UG

AGCUG

ACU

1120

1130

1140

AAGA A UU G A C GG A A G G AC A

AC G

ACU

AAU

C U AC U UC G A

AAGUGGGU

CUU

UAGU

G C A C U A UUU

AUGUGCUG

GUCCUUUGUA

1150 1160

1180

1190

1200

1210

1470

1480

1490

1500 1510

1520

1530

1540

1550

1560

1570

1580

1590

1600

1610

1620

1630

A C U A

C G

A U

UGA A U G U

U A

G U G

A G

A U C U

GCU

U G G U U

CG A G C GG C U C U

GUCGUUCGGU

UCGAA

CUCA

CUG

AUUUAG

GUAA

1660 1670 1680

1690 1700

1710

1720

1730

1740

1750

1760

UUG

A G U

U G

A C U

GGU

ACU

AUCAU

U G

A U

A G G

CCU

UAAC

GUU G

UAU

UAA

CGC

G C G

C C

60 70

100

110

120

130

140

150

160

170

180

190

200

210

220

230

240

250

260

270

280

290

300

310

320

330

340

350

360

CAGAGAAU

GGUUU

GGU

U C UG

370

380 390

400

410

420

430

440

AAU

GGAC

AG G U U

U C

A U U

520

540

A A CA A

550

C A

480

490

500

510

530

560

GCC

570

580

590

600

610 U

UCUU

[

]

1170

CACUUUUAGGA

GAG

U C

U U

U C U U G A

U U

A G G U G

GUU

UC G

A A

A A C

AGA

U C

C U U

C U A A

A U

A G U CU A A A G U U C G

GGACUUU

AGACU

UUAG

AGG

C U

U G

1230

1240

1250

1260

1270

1280

1290

1300

1310

1320 1330

1340 1350 1360

1370

1380

1390

1400

1410

1420

1430

1440

1450

G C 1460

C C

1220

GUU

A A U

CAA

G A

AGA

CCAU

G C C

A C

U A

GGU

620

630

640

980

990

1000

1010

1020

1030

1040

1050

1060

1070

1080

1090

1100

1110

UUU

650

660

670

680

690

700

710

720

730

740

750

A A UGAU U

AUUUU

G A A A U U G A G AU

830

860

GCUUUA

850

U A

CGU

UGUGGUG

GAG

U U C

UGGA

C G C C A C A

G G

A C G

CAG

G C G

870

880

890

900

910

920

930

940

950

960

970

760

A A AA

U A

770

800 A

790

GAG

820

UGGA A

AGA

0 < ≤ 0.2

0.2 < ≤ 0.4

0.4 < ≤ 0.6

0.6 < ≤ 0.8

0.8 < ≤ 1.0

= 1.0 (constant)

= uninformative

450

460

470

UGA A A U

820

Fig. 1 The three exemplar rotifer 18S rRNA molecules: A Adineta vaga (Bdelloidea), b Brachionus plicatilis (Monogononta), and c Seison nebaliae

(Seisonacea). Individual nucleotides are coloured according to their relative rate of evolution (for Bdelloidea, Monogononta, and all Rotifera,

respectively) as determined using TIGER (see inline legends) and hypervariable regions V1–V9 are labeled and outlined in blue. In addition, the

maximal extent of the variable deletion at the 3’‑end of the V3 region in bdelloids (see Fig. 2) is highlighted using square parentheses. Each map

was created initially using VARNA v3.93 [59] and modiﬁed using Adobe® Illustrator® 2020 to match the traditional topology of the eukaryotic core

structure as closely as possible. Empty circles were added at the terminal ends of the sequences as needed to present the presumed full length of

the molecule for each species

Page 5 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

GCC

570

580

590

560

UCCU

AGU

U G G U UG

AGCUG

ACU

1130

1140

AG G A A U

1150

1640

1650

U G A C GG A A GGGC A

CCA

CAC

C C U G CU C CG

GGAGUGGG

GCU

CCAG

AGC G CG A G

CUCGCGUU

UGCCCUUUGU

1160

1170

1180

1190

1200

1210

1470

1480

1490

1500

1510

1520

1530

1540

1550

1560

1570

1580

1590

1600

1610

1620

1630

U A C U

C C

G A

UUG A A U G GU

U U

A G U

G A U C

U C

GGA

UU G G U C CGGA G G U G

C C G

CGCCUCU

GCCGA

GAUC

ACU

AUCAUUUA

AGUA

1660 1670 1680 1690 1700

1710

1720

1730

1740

1750

1760

1770

U U C

G U A G G U

CCUGC

GGA

1780

1790

1800

1810

GGGGAAU

GG U U C

GAU

U C C G

370

380

UCA

CAU

C A

A C C

GGU

AUCGU

U G

AUA

U G

U G G

CUA

AUA

UUGU

G A

UAA

AGC

G C U

C C

GCAG

CCUU

CG A G G

100

110

120

130

140

150

160

170

180

190

200

210

220

230

240

250

260

270

280

290

300

310

320

330

340

350

360

390

400

410

420

430

440

450

460

470

480

490

500

510

520

530

540

600

610

CUG G U

700

CGU

UGCGGUG

GAG

UUC

UGGA

CG C C GCA

A C G

CAA

G C G

880

890

900

910

920 930

940

950

960

970

710

720

730

740

650

660

670

680

690

A A UGAU U

870

UUU

850

ACCUCGGU

A G

A U U G A G G U

830

840

860

AUUUUG

AGA A

820

UUA

750

800 810

770

U G G U A A

790

U A C

760

G A G

AAUU

C G A

780

GUU

A A U

CAA

AGA

CCAU

G C C

A C

U U

GGU

620

630

980

990

1000

1010

1020

1030

1040

1050

1060

1070

1080

1090

1100

1110

1120

CACUG

GAUUGAGAG

UU U C U U G AU U C G G U G

GUU

C C G

A A

A A C

AGA

U C

C C U

C U A A

UUAG

GGG

C G

C U

G C

C G

1230

1240

1250

1260 1270

1280

1290

1300

1310

1320

1330

1340

1350

1390

1400

1410

1420

1430

1440

1450

C G

1220 1460

G U A C G U C G A U

GUCG

GCGUAC

1360 1370

1380

AUCUA

550

0 < ≤ 0.2

0.2 < ≤ 0.4

0.4 < ≤ 0.6

0.6 < ≤ 0.8

0.8 < ≤ 1.0

= 1.0 (constant)

= uninformative

Fig. 1 continued

Page 6 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

AGG

C C

530

U A U G U C U U G A G U AACUA

540 550

490

500

510

520

GAA

CAUGU

C A C U

GAU

AGUG

AUUGUA A A

U G

AUA

UG G U

CCA

GC G A C

GUCGU

UAAAGUCGAAC

UGUU C G G C U UU G

ACU

AUAU

UAA

AGC

G C U

C C

100

110

120

130

140

150

160

170

180

190

200

210

220

230

240

250

260

270

280

290

300

310

320

330

340

350

360

370

CGGGGAAU

GG U U C

GAU

U C CG

380 390

400

410

420

430

440

450

460

470

610

V4 C

1830

1780

1790

U U U

C C

G U A G G

CCUGC

1800

1810

1820

1680

U A U A

U U

G A

U G A A CGAU

U U

A G U

G A A U

GAA

UU U A U

U U G U G

CAUGA

GUAAA

AUUC

AUU

AUCGUUUA

UAUA

1690 1700 1710 1720 1730

1740

1750

1760

1770

GAGU

U G G U UG

AGCUG

ACU

1130

1140

1150

AGGA A UU G A C GG A A GGGC A

CCA

AC G

CAA

AUAU

GCCC

GGU

GGGU

GUU

CUAG

AGCA C G AG

CUCGUGCU

UGCCCUUUGU

1160 1170

1180

1190

1200

1210

1220

CACUG

GAUCGAGAG

UU U C U U G AU U C A G U G

C G

C U

C G

1230

1240

1250

1260

1270

1280

1290

GUU

CCG

A A

AAC

AGA

U C

U C UGC U A A

A G U G U G

CACACU

UUAGAGA

1300

1310

1320

1330

1340

1350

1360

1370

1380

1390

1400

1410

1420

1430

1440

1450

1460

1470

G C 1480

1490

1500

1510

1520

1530

1540

1550

1560

1570

1580

1590

1600

1610

1620

1630

1640

1650

1660

1670

650

660

670

680

690

700

710

720

730

740

750

A A UGAU U

ACUC

AGUUG

C A A U U

UU A A

G A G U

830

840

860

870

850

UUG

760

770 U

810

780

790

800

GAG

A U

640

620

990

A U

G C

890

CGU

CAUGACG

GAG

UUC

UUGA

CG U C A U G

A C G

CUA

G C G

900

910

920

930

940

950

960

970

980

UAG

880

480

600

GCC

570

580

590

AGA

CCGUA

G C C

A C

GGU

630

1000

1010

1020

1030

1040

1050

1060

1070

1080

1090

1100

1110

1120

AUCA

UGA

820

0 < ≤ 0.2

0.2 < ≤ 0.4

0.4 < ≤ 0.6

0.6 < ≤ 0.8

0.8 < ≤ 1.0

= 1.0 (constant)

= uninformative

Fig. 1 continued

Page 7 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

TIGER values for Bdelloidea and Monogononta being

slightly less than 0.5 and only decreasing to around 0.4

for all Rotifera (Table3). Indeed, the rates for variable

sites tended to cluster around these values such that

variable sites that evolved extremely slowly (TIGER

rate > 0.6) were all but absent as were those that evolved

very rapidly (TIGER rate < 0.4) with the possible excep-

tion of across Rotifera as a whole.

ere was no diﬀerence in the inferred relative rates

of evolution between paired sites in each of the three

data sets (column 2 in Table4) and together these sites

(“stems”) evolved slightly slower than unpaired sites

(“loops”) for Bdelloidea only (column 3 in Table 4).

Highly signiﬁcant diﬀerences in relative rates were

present between regions inferred as being within

hypervariable versus non-hypervariable regions,

regardless whether the former regions were pooled

for the analysis or analysed individually (columns 4 to

6, respectively, in Table 4). e average relative rates

for most hypervariable regions were faster than those

for the pooled non-hypervariable regions (Fig.4), with

V6 being the notable, expected exception in addition

to several hypervariable regions for bdelloids. In addi-

tion, the rates in the Rotifera data set were generally

faster than those in either the bdelloid or monogonont

data sets, reﬂecting in part the sequence diﬀerences

between the motifs in the latter two. For the latter data

sets, bdelloids generally presented slower rates than

Fig. 2 Partial alignment of the 18S rDNA gene showing the variably expressed deletion present at and extending beyond the 3’ end of the V3

hypervariable region in bdelloids and two species in the monogonont genus Lindia. The locations of the entire V3 region as well as the pseudoknot

following it are indicated at the top of the alignment. The taxonomic groups are numbered as (1) Calicophoron calicophorum (Platyhelminthes,

outgroup), (2) Bdelloidea, (3) Seisonacea, and (4) Monogononta (selected species)

Table 3 Relative TIGER rates of evolution across the three data sets for all sites and variable sites only, extended to include

Acanthocephala

Sites where less than 15% of the species possess sequence data are excluded. Rates are presented as the mean ± SE with no correction for the non‑independence

between paired sites

Data set TIGER rate of evolution (all sites) TIGER rate of

evolution (variable

sites)

Bdelloidea 0.926 ± 0.022 (n = 1703) 0.485 ± 0.031 (n = 244)

Monogononta 0.873 ± 0.020 (n = 1817) 0.467 ± 0.022 (n = 434)

Rotifera 0.760 ± 0.017 (n = 1969) 0.391 ± 0.014 (n = 777)

Acanthocephala 0.744 ± 0.015 (n = 2313) 0.371 ± 0.012 (n = 942)

Syndermata (= Acanthocephala + Rotifera) 0.661 ± 0.013 (n = 2480) 0.326 ± 0.009 (n = 1249)

Page 8 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

did monogononts with the exception of hypervariable

region V3.

Substitution hotspots were present along the entire

molecule (Fig.5), but more common and with higher rel-

ative rates in the hypervariable regions (with the excep-

tion of V6). However, most of these hotspots did not

express unduly high relative rates of evolution. Relative

rates within the hypervariable regions could also diﬀer as

exempliﬁed by region V4, where the 5′ half of the region

displays noticeably higher rates than the 3′ end where

two pseudoknots are located. Again, rates of evolution at

any given position were generally the highest for Rotifera

and lowest for Bdelloidea among the three data sets.

Indels were numerous in each of the three exemplar

species when aligned against one another in the Rotifera

data set, and disproportionately so in the hypervariable

regions, but were very short on average (usually < 2 bp

long; Table 5). Only hypervariable region V7 showed

a tendency toward having longer indels despite it being

noticeably shorter in rotifers than in the outgroup species

C. calicophorum (as well as in Daphnia pulex, and Loric-

era foveata; see Methods). Region V6 is noteworthy inso-

far as no indels were inferred between the three major

rotifer clades (Table5).

Discussion

Altogether, my examination of the evolution of 18S rDNA

in rotifers revealed a clear pattern whereby the high mor-

phological disparity among the three major clades is

matched by their molecular disparity for this molecule.

e high number of very short indels together with the

extremely restricted sequence variation within each of

the two largest clades (Bdelloidea and Monogononta)

also indicates that the disparity derives mostly from sub-

stitutions, and then predominantly in the hypervariable

regions. However, despite their name, even the hypervar-

iable regions are largely conserved within the three major

rotifer clades. By way of comparison, as well as to under-

score the degree of sequence conservation within the

major clades, the analogous average uncorrected pair-

wise distance for an alignment of 69 18S rDNA GenBank

sequences for Acanthocephala (see Additional ﬁle2) that

includes all its four major subgroups within the clade is

13.1%. Although this value approaches that of the average

pairwise divergence among the three rotifer 18S rDNA

motifs (from 17.0 to 22.2% from Table1 or 18.7% ± 1.5%

among the three exemplar species only), it is an order of

magnitude higher than that for within each of the major

rotifer clades (< 3%; compare Table1), a taxonomic sta-

tus that arguably also applies for Acanthocephala (see

Methods). Similarly, TIGER rates of evolution for Acan-

thocephala (Table3) are comparable to those for Rotifera

as a whole, rather than to those for either Bdelloidea or

Monogononta.

is extreme pattern in Rotifera is also easily visual-

ized through a maximum-likelihood phylogeny of the

sequences examined in this study together with those

of Acanthocephala for context (Fig.6; Additional ﬁle3).

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 constant n/a

TIGER relative rate

200

400

600

800

1000

1200

1400

1600

Number of positions

Bdelloidea

Monogononta

Rotifera

Fig. 3 Histogram of relative TIGER rates of evolution across the three rotifer data sets (green, Bdelloidea; red, Monogononta; and blue, Rotifera). n/a

stands for sites represented by less than 15% of the species in a given data set and so with insuﬃcient coverage for the TIGER analyses

Page 9 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

Table 4 Statistical summary for testing of potential diﬀerences in the relative TIGER rate of evolution between diﬀerent categories of positions

Paired and unpaired positions were taken as proxies for stems and loops, respectively. Because paired rates were not signicantly dierent from one another (column 2), each paired position was represented by only its

single, average rate in column 3 to avoid problems with pseudoreplication. Only those positions that could denitely be assigned as being paired versus unpaired based on the structural map of the exemplar species (see

Fig.1) were included in the respective analyses

Data set Between paired positions Stems vs. loops Non-hypervariable vs.

hypervariable regions – counts

of constant vs. variable

Non-hypervariable vs.

hypervariable regions – rates Among hypervariable as well

as pooled non-hypervariable

regions

Bdelloidea Wilcoxon W = 2211.5; Z = 1.055;

p = 0.292; nnon‑zero,total = 88, 437 Mann–Whitney U = 1.604 × 105;

Z = 2.969; p = 0.003;

nstems,loops = 443, 776

χ2 = 105.70, df = 1; p = 8.60 × 10–25 Mann–Whitney U = 2.722 × 105;

Z = 10.608; p = 2.74 × 10–26;

nnon,hyper = 1022, 663

Kruskal–Wallis H = 75.3;

p = 2.12 × 10–38

Monogononta Wilcoxon W = 5142.0; Z = 1.881;

p = 0.060; nnon‑zero,total = 131, 458 Mann–Whitney U = 1.827 × 105;

Z = 1.890; p = 0.059;

nstems,loops = 461, 834

χ2 = 162.95, df = 1; p = 2.57 × 10–37 Mann–Whitney U = 2.613 × 105;

Z = 13.217; p = 7.04 × 10–40;

nnon,hyper = 1076, 683

Kruskal–Wallis H = 169.8;

p = 6.92 × 10–59

Rotifera Wilcoxon W = 11,856.0; Z = 1.0087;

p = 0.313; nnon‑zero,total = 209, 454 Mann–Whitney U = 1.866 × 105;

Z = 1.913; p = 0.056;

nstems,loops = 461, 859

χ2 = 130.18, df = 1; p = 3.74 × 10–30 Mann–Whitney U = 2.718 × 105;

Z = 11.190; p = 4.54 × 10–29;

nnon,hyper = 1079, 701

Kruskal–Wallis H = 174.3;

p = 2.03 × 10–43

Page 10 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

1100 bp

250 bp

Fig. 4 Sizes and relative rates of evolution (as determined using TIGER) of each of the hypervariable and pooled non‑hypervariable regions for the

three rotifer data sets (green, Bdelloidea; red, Monogononta; and blue, Rotifera). Error bars represent standard errors and are subsumed by the data

point when not visible

Fig. 5 Relative TIGER rates of evolution presented as a rolling average of the 35 positions centred on the focal position for each of the three rotifer

data sets (green, Bdelloidea; red, Monogononta; and blue, Rotifera). The locations of the hypervariable regions are indicated by bars at the top of the

graph

Page 11 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

Table 5 Summary statistics regarding indels in each of the three exemplar rotifer 18S rRNA molecules from the complete Rotifera alignment (i.e., excluding Calicophoron

calicophorum)

For the pooled, non‑hypervariable regions (“non”), terminal gaps, which likely represent incomplete sequences, were excluded from the calculations

Region Start position in the

aligned rotifer data set Length in the

aligned rotifer data

set

Adineta vaga (Bdelloidea) Brachionus plicatilis (Monogononta) Seison nebaliae (Seisonacea)

Number of

nucleotides Number

of indels Average

length of

indel

Number of

nucleotides Number

of indels Average

length of

indel

Number of

nucleotides Number

of indels Average

length of

indel

V1 64 39 33 4 1.5 33 3 2.0 35 3 1.3

V2 187 125 104 10 2.1 102 10 2.3 104 6 3.5

V3 524 83 74 7 1.3 75 6 1.3 77 4 1.5

V4 700 286 228 27 2.1 233 23 2.3 236 30 1.7

V5 1174 50 44 6 1.0 44 5 1.2 41 7 1.7

V6 1374 42 42 0 n/a 42 0 n/a 42 0 n/a

V7 1491 79 51 3 9.3 48 4 7.8 64 4 3.8

V8 1660 73 60 7 1.9 60 6 2.2 68 4 1.3

V9 1880 54 48 4 1.5 46 4 2.0 34 8 2.5

Non n/a 1190 1135 50 1.1 1130 52 1.2 1129 51 1.2

Page 12 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

Here it is the pattern that is of chief interest rather than

the relationships per se, with Bdelloidea and Seisonacea

forming sister taxa at the ends of extended branches to

the exclusion of Monogononta, which shows much less

divergence from the rotifer common ancestor. Monogon-

onta also displays extremely reduced molecular diver-

gence between its members, despite being by far the best

represented of the three major clades. Finally, Acantho-

cephala again display as much internal molecular diver-

gence in the phylogeny as for all true rotifers.

Moreover, the variability among the rotifer clades

is restricted to at most 40% of the 18S rRNA molecule,

with the remaining sites being constant across all Rotif-

era. is value coincidentally matches the proportion of

the molecule that comprises hypervariable regions. How-

ever, even though the hypervariable regions are indeed

signiﬁcantly more variable than the non-hypervariable

ones (both in the number of variable sites and their

rates; Table 6), variable sites are found in both regions.

Although the species sampling in this study was neces-

sarily restricted, the taxonomic diversity of Rotifera was

well represented insofar as exemplars from all major tax-

onomic subdivisions within each of Rotifera, Bdelloidea,

and Monogononta (see Methods) were present in the

data set. It could be argued that the number of constant

sites is slightly overestimated insofar as gaps between

the major clades were often preferred to substitutions.

However, the increase in length compared to length of

the entire molecular is negligible (about 10%; see Table2)

and this problem would apply chieﬂy to the entire rotifer

data set. In addition, the high proportion of constant sites

across Rotifera matches that inferred across Gastropoda

by Weigand etal. [23] and for Acanthocephala (Table3).

us, the general patterns observed here, especially the

highly distinct motifs for the major clades, are likely to be

accurate. In addition, preliminary results from 28S rDNA

and MT-CO1 conﬁrm this general pattern of highly dis-

tinct motifs (results not shown), although not to the same

extent as seen for 18S rDNA.

e lack of comparative data for other, comparable

clades prohibits a good assessment of the novelty of

the observed patterns across eukaryotes and the loss of

0.2

73.5

40.6

100.0

81.9 100.0

Monogononta

Acanthocephala

Bdelloidea

Seisonacea

Fig. 6 Maximum‑likelihood phylogeny of the 18S rDNA sequences examined in this study using RAxML v8.2.12 [60]. The analysis consisted of a fast

bootstrap search followed by a thorough search for the maximum‑likelihood topology [61] under a GTR + Γ model, with the gamma distribution

being approximated initially through a CAT model [62]. The tree was rooted on Calicophoron calicophorum. Values above selected nodes represent

bootstrap support [63] and the scale bar represents the average number of substitutions per site per unit time. Species names have been removed

for clarity and the major clades are labeled as well as colour‑coded (green, Bdelloidea; red, Monogononta; blue, Seisonacea; black, Acanthocephala)

Page 13 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

readily available structural data through the demise of

the European Ribosomal Database [24] is strongly felt in

this regard. As mentioned, the high proportion of con-

stant sites is comparable to that seen across Gastropoda,

although the latter show a more even distribution of rates

across the entire spectrum from slow to fast than was the

case for Rotifera, albeit with a strong spike at intermedi-

ate rates [23]. Again, there is also a strong similarity in

many parameters between Rotifera as a whole and Acan-

thocephala. Although the variability map of Van de Peer

et al. [25] shows comparatively few constant sites and

many very fast ones, it is at the level of all Eukaryota and

so hardly comparable in terms of taxonomic breadth. e

latter, however, does indicate that the pseudoknot follow-

ing the V3 region is relatively conserved across eukary-

otes, making its variable deletion within Bdelloidea all

that more unusual. Even more extraordinary is that the

deletion in both Rotaria neptunoida and Rotaria tar-

digrada includes the central “square” of the 18S rRNA

molecule from which its three main arms originate and

thereby could aﬀect the tertiary structure of the entire

molecule in some unknown and potentially severe

fashion.

Regardless of the potential patterns in other taxonomic

groups, one explanation for the pattern observed here

is that the three rotifer crown groups are of relatively

recent origin and, with the exception of Seisonacea, have

undergone rapid adaptive radiations. If true, this scenario

would imply a high degree of morphological plastic-

ity in Monogononta in particular given both the higher

number of species as well as morphological diversity

across the group compared to Bdelloidea. What remains

unclear, however, is why monogonont sequences, on

average, remain more similar to that of a relatively dis-

tant platyhelminth outgroup (Table 1) instead of to the

remaining rotifers or why this clade as a whole also does

not subtend an extended branch like Bdelloidea and Sei-

sonacea (and even Acanthocephala) as would be expected

for a recent radiation within an otherwise ancient group.

Of particular interest in this general context would be

the sequencing of the remaining seisonid species. Given

the lack of any obvious widespread dispersal abilities in

seisonids, perhaps in concert with the hypothesis that

they are among the oldest of the rotifer clades [17], the

apparently exclusive association between them and

their Nebalia hosts could be ancient, which agrees with

the extended branch leading to S. nebaliae. Less clear,

however, is whether the individual associations are also

ancient or of more recent origin, especially given that

diﬀerent seisonid species can be found on the same host

species if not the same host individual [17, 18]. Addition-

ally, or alternatively, it could be that the three clades have

independently reduced their rates of molecular evolu-

tion. However, it is not clear what the mechanism behind

this would be and why the same process has not occurred

in Acanthocephala, especially given that this taxon does

indeed appear to nest within Rotifera [also 15, 16, 26–

28]. Possible explanations for the latter discrepancy could

lie with the endoparasitic lifestyle of all acanthoceph-

alans as compared to the free-living true rotifers (with

the exception of Seisonacea) or that the crown group is

simply older and so shows more within-group molecular

diversity.

Problematic in testing these hypotheses is that diver-

gence time estimates for and within Rotifera are all but

absent. Apart from a few reports of subfossilized rotifers

from Holocene peat deposits (e.g., [29, 30]), the only

other known rotifer fossils comprise contracted bdelloid

specimens or their theca encased in amber, the oldest

pieces of which have been dated to 35–40Ma ago [31,

Table 6 Counts (with percentages in parentheses) of constant versus variable sites partitioned according to the inferred hypervariable

versus non‑hypervariable regions (pooled) of the three 18S rRNA data sets

The results testing the hypotheses that the proportion of constant versus variable sites do not dier between regions of the 18S rRNA molecule is presented in Table4

(middle column)

Data set Region Counts

Total Constant Variable

Bdelloidea Not hypervariable 1030 (60.5%) 940 (55.2%) 90 (5.3%)

Hypervariable 673 (39.5%) 488 (28.7%) 185 (10.9%)

All 1703 (100.0%) 1428 (83.9%) 275 (16.1%)

Monogononta Not hypervariable 1111 (61.1%) 944 (52.0%) 167 (9.2%)

Hypervariable 706 (38.9%) 411 (22.6%) 295 (16.2%)

All 1817 (100.0%) 1355 (74.6%) 462 (25.4%)

Rotifera Not hypervariable 1138 (57.8%) 784 (40.0%) 354 (18.0%)

Hypervariable 831 (42.2%) 359 (18.2%) 472 (24.0%)

All 1969 (100.0%) 1143 (58.0%) 826 (42.0%)

Page 14 of 18

Bininda‑Emonds BMC Ecol Evo (2021) 21:118

32]. Although the theca in particular have been assigned

to the extant genus Habrotrocha [32], the contracted

nature of the specimens and the simplistic features of the

theca make it diﬃcult to determine their species iden-

tity precisely and thus caution is perhaps advised as to

whether or not either set of specimens belong to crown

group Bdelloidea. Molecular based studies place the

divergence between Rotifera and Platyhelminthes from

anywhere between 492 to 1160 Ma ago (best estimate,

824Ma ago; www. timet ree. org [33]), with the origin of

Rotifera being necessarily more recent than this, espe-

cially given that Platyhelminthes are likely not the imme-

diate sister group of Rotifera, even among extant taxa

[34, 35]. As such, it is unknown how old Rotifera are as

a group as well as what the ages of its three major crown

groups are, information that is needed to better under-

stand the pattern of evolution of 18S rDNA witnessed in

this paper, and possibly of other markers as well.

Nevertheless, the observed pattern potentially explains

the severe problems in reconstructing the phylogenetic

history of Rotifera at higher taxonomic levels. As sum-

marized by Fontaneto and de Smet [15], about the only

points of consensus in this context are the monophyly of

each of the three major rotifer clades. Indeed, morpho-

logical and molecular analyses paint diﬀerent pictures of

higher-level rotifer phylogeny. Whereas the former tend

to support Bdelloidea and Monogononta as sister taxa

(= Eurotatoria) within a monophyletic Rotifera, the lat-

ter usually cluster Bdelloidea and Seisonacea together

(as herein) within a paraphyletic Rotifera because of the

inclusion of Acanthocephala, sometimes as the immedi-

ate sister group to Seisonacea. However, the pattern of

evolution in 18S rDNA presented in this paper suggests

that traditional sequence-based phylogenetic analyses of

Rotifera could potentially be compromised by artefacts

known to arise from long-branch attraction [see 36],

which could be especially severe in this case given the

lengths of the branches involved. In fact, the major roti-

fer clades are so distinct from one another molecularly

that even analyses of MT-CO1 support the monophyly

of each of Bdelloidea and Monogononta with high boot-

strap support (results not shown), although MT-CO1 (or

at least that part obtained using the Folmer [37] primers)

normally loses phylogenetic signal above the genus level

so rapidly that its use is typically restricted to species bar-

coding [38]. Instead, inferences based on rare genomic

changes [see 39] or other types of markers, including

meta-sequence features [see 40] might be more reliable

for unravelling higher-level relationships within Rotifera.

Interestingly, results from an analysis of mitochondrial

gene order [28] do match those from traditional sequence

analyses, despite the limited taxon sampling in that study

as well as the methodological problems that are known

for analyses of gene order and that have prevented these

data from playing a more prominent role in phylogenetic

analyses to date [see 41, 42].

Fortunately, any potential artefacts caused by long-

branch attraction appear to be limited to the inferences

of the relationships among the three major clades, rather

than the relationships within each of them where the

branches are shorter (see Fig.6). At these less inclusive

levels, the inferred relationships in the 18S rDNA tree

(Additional ﬁle 3) tend to reﬂect accepted taxonomic

groups down to the genus level within Rotifera, especially

within Monogononta (see e Rotifera World Catalog

(www. rotif era. hausd ernat ur. at), despite the high propor-

tion of constant sites. As such, analogous to the case with

missing data (see [43]), it would appear that the limited

number of variable sites are both suﬃcient in number

and do not evolve unduly rapidly to provide good reso-

lution. Nevertheless, the rate variation among sites that

is present requires that some correction for rate hetero-

geneity is used [see also 20, 23] and even this might be

insuﬃcient at higher taxonomic levels within Rotifera.

Methods

Data set

Largely complete 18S rDNA sequences (one per spe-

cies; see TableS1, Additional ﬁle2) were either compiled

from GenBank or newly generated within the working

group (87 sequences, all from Monogononta), in part

for other studies (e.g., [44, 45]). All previously unpub-

lished sequences have been deposited in GenBank under

the accession numbers MT522624–MT522695 and

MT542324. e newly derived sequences were obtained

following the protocols outlined in Kimpel et al. [46]

and Wilke etal. [45]. All taxonomic names were veriﬁed

against e Rotifera World Catalog (accessed on April 10,

2020). In total, 198 rotifer sequences were used, including

162 monogonont sequences, 35 bdelloid sequences, and

one seisonid sequence. Although this amounts to only

roughly 10% of the described species diversity, all three

major rotifer clades (Bdelloidea, Monogononta, and Sei-

sonacea) were represented as were both major clades

within Monogononta (Gnesiotrocha and Pseudotrocha;

[15]) and all three within Bdelloidea (Adinetida, Philo-

dinavida, and Philodinida; [47]); Seisonacea comprises

only a single clade of four described species in two genera

[18]. Although there is mounting molecular evidence that

Acanthocephala (ca. 1330 species) nests within Rotif-

era, possibly as the sister taxon to Seisonacea (e.g., [16,

26–28]), I have restricted my primary analyses to “true”

rotifers as traditionally recognized only. e complete

18S rRNA sequence for the ﬂatworm Calicophoron cali-

cophorum (Platyhelminthes: Trematoda: Digenea; Gen-

Bank accession L06566) was added to the data set as