disproportional plastome-wide increase of substitution...

Article

FastT

rackDisproportional Plastome-Wide Increase of SubstitutionRates and Relaxed Purifying Selection in Genes ofCarnivorous LentibulariaceaeSusann Wicke,*,y,1 Bastian Schäferhoff,y,1 Claude W. dePamphilis,2 and Kai F. Müller11Institute for Evolution and Biodiversity, University of Muenster, Muenster, Germany2Department of Biology and Institute of Molecular Evolutionary Genetics, Pennsylvania State UniversityyThese authors contributed equally to this work.

*Corresponding author: E-mail: [email protected].

Associate editor: Charles Delwiche

Abstract

Carnivorous Lentibulariaceae exhibit the most sophisticated implementation of the carnivorous syndrome in plants.Their unusual lifestyle coincides with distinct genomic peculiarities such as the smallest angiosperm nuclear genomes andextremely high nucleotide substitution rates across all genomic compartments. Here, we report the complete plastidgenomes from each of the three genera Pinguicula, Utricularia, and Genlisea, and investigate plastome-wide changes intheir molecular evolution as the carnivorous syndrome unfolds. We observe a size reduction by up to 9% mostly due tothe independent loss of genes for the plastid NAD(P)H dehydrogenase and altered proportions of plastid repeat DNA, aswell as a significant plastome-wide increase of substitution rates and microstructural changes. Protein-coding genesacross all gene classes show a disproportional elevation of nonsynonymous substitutions, particularly in Utricularia andGenlisea. Significant relaxation of purifying selection relative to noncarnivores occurs in the plastid-encoded fraction ofthe photosynthesis ATP synthase complex, the photosystem I, and in several other photosynthesis and metabolic genes.Shifts in selective regimes also affect housekeeping genes including the plastid-encoded polymerase, for which evidencefor relaxed purifying selection was found once during the transition to carnivory, and a second time during the diver-sification of the family. Lentibulariaceae significantly exhibit enhanced rates of nucleotide substitution in most of the 130noncoding regions. Various factors may underlie the observed patterns of relaxation of purifying selection and substi-tution rate increases, such as reduced net photosynthesis rates, alternative paths of nutrient uptake (including organiccarbon), and impaired DNA repair mechanisms.

Key words: carnivorous plants, chloroplast genome, substitution rates, purifying selection, Lentibulariaceae,noncoding DNA.

IntroductionThe carnivorous syndrome in plants comprises the ability toattract, retain, trap, kill, digest prey, and finally absorb andmake use of the nutrients resulting from digestion (Juniper1986). Carnivory evolved several times among angiospermswith at least seven independent origins in five different ordersof flowering plants (Albert et al. 1992; Soltis et al. 2005;Ellison and Gotelli 2009; Pereira et al. 2012), with theorder Lamiales being the hotspot of carnivorous plantdiversity. Lamiales possess the greatest trap type diversityand a minimum of three independently evolved carnivo-rous lineages (Byblidaceae, Philcoxia from Plantaginaceae,Lentibulariaceae) and several subcarnivorous taxa (Plachnoet al. 2006; Schäferhoff et al. 2010), which are capable totrap or kill prey such as insects but cannot digest andabsorb the released nutrients.

With well more than 350 species, the cosmopolitanLentibulariaceae are by far the most diverse carnivorousfamily, exhibiting extreme physiological adaptations andwhat has been described as the most complex leaf structuresknown in the entire plant kingdom (Juniper et al. 1989).

Pinguicula, sister to the so-called bladderwort lineage thatcomprises Utricularia and Genlisea, has the simplest bodyplan, capturing prey with the help of (often) pale-greenleaves borne in a rosette that function as flypaper traps.Genlisea, the corkscrew plant, possesses Y-shaped and heli-cally twisted underground rhizophylls (eel traps) used to pas-sively capture and digest protozoa (Barthlott et al. 1998). Incontrast, the bladder traps of Utricularia work by means oflow pressure, capturing small organisms from the surround-ing water or soil, and have been characterized as having themost complex leaf modifications (Juniper et al. 1989). In con-trast to Pinguicula, bladderworts and corkscrew plants lack aroot system, and thus solely rely on the uptake of nutrientsthrough prey and by means of foliar absorption (Adamec1997, 2013).

Lentibulariaceae also exhibit remarkable traits at the mo-lecular level. For instance, the smallest genomes reportedfrom angiosperms belong to species of Genlisea andUtricularia (Greilhuber et al. 2006). DNA substitutional ratesin several loci representing all three genomic compartmentswere found to be significantly higher in members of the

� The Author 2013. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, pleasee-mail: [email protected]

Mol. Biol. Evol. 31(3):529–545 doi:10.1093/molbev/mst261 Advance Access publication December 15, 2013 529

at Pennsylvania State University on Septem

ber 2, 2014http://m

be.oxfordjournals.org/D

ownloaded from

,, , , plus -,over ; Adamechttp://mbe.oxfordjournals.org/

bladderwort lineage than in other angiosperms (Jobson andAlbert 2002; Müller et al. 2004; Ibarra-Laclette et al. 2011). Thefinding that plastid genes of Lentibulariaceae evolve at ele-vated substitution rates was unexpected as plastid genomes,and particularly their coding regions, usually evolve in a re-markably conservative manner across land plants (Palmer1985; Raubeson and Jansen 2005; Wicke et al. 2011), withthe majority of the ~110–120 unique genes being indispens-able for photosynthesis and a few other metabolic andgenetic pathways. Relatively few autotrophic lineages demon-strate unusual modes of plastid genome and plastid geneevolution (Palmer et al. 1988; Cosner et al. 2004; Guisingeret al. 2010, 2011; Fajardo et al. 2012; Sloan et al. 2012), includ-ing large-scale reconfiguration of the typically compactquadripartite structure that consists of a large and smallsingle-copy region (LSC, SSC) and two large invertedrepeats (IRs).

The most dramatic changes in plastid gene content havebeen reported for plants with a heterotrophic mode ofnutrition (Wolfe et al. 1992; Nickrent et al. 1997; Funk et al.2007; McNeal, Kuehl, et al. 2007; Delannoy et al. 2011;Braukmann and Stefanović 2012). Like bladderworts, many(obligate) parasitic plants exhibit highly accelerated rates,which result (at least in part) from reduced purifying selectionin photosynthesis and photosynthesis-related plastid genes(e.g., dePamphilis et al. 1997; Young and dePamphilis2005; McNeal, Kuehl, et al. 2007). Although no case ofC-heterotrophy in carnivorous plants has been documented,available data indicate regular carbon uptake via carnivory,particularly under challenging environmental conditions(Adamec 1997; Pavlovič et al. 2009; Adamec 2011). Previousstudies (reviewed in Adamec 2011) suggest a trend towardhigh rates of molecular evolution in lineages where thenormal uptake of inorganic nutrients via roots is accompa-nied or replaced by some other mode of nutrient acquisition.Naturally, this involves diversely modified biochemical pro-cesses and, thus, may dramatically affect the molecular evo-lution in a variety of genetic loci. It is therefore comparable tosome extent with the situation in parasites.

To investigate whether the nutritional benefit of carnivorymight impact the evolution of the photosynthetic machinery,we sequenced the plastid genomes of representatives of allthree main lineages of Lentibulariaceae. Using a comparativeapproach and a series of sophisticated analyses of rate varia-tion under maximum likelihood, we discriminate between therole of positive selection, relaxed purifying selection, and back-ground mutation rates. Here, we present evidence for changesin the degree of positive or negative selection acting on the

functionally diverse array of plastid genes and discuss it in thelight of the degree of dependency on prey in the family.A fine-scale analysis of rate shifts in nonprotein-coding re-gions and across all plastid genes assesses the degree of selec-tional changes on plastid protein-coding regions. Above that,we examine the evolution of microstructural changes acrossthe plastid chromosomes of the carnivores with special regardto the strength of purifying selection and functional losses.The results of this study promote our understanding of ge-nomic footprints of carnivory and reveal genome-level con-vergences between carnivores and parasites.

Results

Structure of Plastid Genomes in Lentibulariaceae

We completely sequenced the plastid genomes (plastomes)of Pinguicula ehlersiae, Utricularia macrorhiza, and Genliseamargaretae as representatives of the three genera of the car-nivorous Lentibulariaceae. In terms of structure, their plas-tomes resemble those of the vast majority of angiosperms.With a total length of 153,228 bp, Utricularia possesses thelargest plastome of the three sequenced species (table 1). Theplastid genomes of Pinguicula and Genlisea are 147,147 bp and141,255 bp in lengths, respectively, and colinear with that ofUtricularia (fig. 1). None of the three species show structuralreconfigurations such as inversions or gene relocations intoother plastid segments. Relative to Utricularia and otherlamiids, disruptions of gene synteny occur, however, in theLSC and SSC of Genlisea, and in the SSC of Pinguicula due tothe deletion of genes encoding subunits of the NAD(P)Hdehydrogenase complex. Considering phylogenetic relation-ships, ndh gene loss of Genlisea and Pinguicula must haveoccurred two times independently within Lentibulariaceae.In Genlisea, the genes ndhC, D, F, G, H, J, and K were lostfrom the plastome; the genes ndhA, E, and I reside as trun-cated pseudogenes. NdhB is retained with an intact readingframe in both Genlisea and Pinguicula. The latter also retainsreading frames for ndhA, D, E, G, H, I, J, and K although theseshow notably long indels and/or frameshifts, implying thatsome of these may be pseudogenes. We could not detectndhC and F in Pinguicula.

A genome-wide alignment between the Lentibulariaceaeplastomes and that of the closely related sesame (Sesamumindicum, Pedaliaceae) reveals several mutational hotspots(fig. 1). This includes the entire SSC as well as the regionaround the ndh genes in the LSC (fig. 1). Further hotspotsof sequence variation accumulate near tRNAs and short(

Relative to the sesame plastome, the number of insertionsand deletions is slightly lower in Pinguicula (cRIN: 206, cRDN:213) than in the bladderwort lineage (Utricularia–cRIN: 211,cRDN: 238; Genlisea–cRIN: 226; cRDN: 257; fig. 2a). Althoughthe number of simple sequence repeats (SSRs) is not evidentlydifferent between carnivores and noncarnivores, we observeda slight elevation of AT-rich di- and tetranucleotide motifsrepeated more than five or three times, respectively, in thebladderwort lineage (fig. 2b).

Similar to most flowering plant plastomes,Lentibulariaceae have few repeats of more than 12 nt inlength (fig. 2c). Lentibulariaceae exhibit a slight increase par-ticularly in reverse-complement repeats (palindromic) com-pared with sesame, and we detected a large number of directand palindromic repeats especially in Utricularia. UnlikeSesamum, all Lentibulariaceae also possess a small numberof repeats larger than 50 bp but none larger than 100 bp(except for the large IR). The distributions of larger repeatsin the plastomes are very similar between Sesamum andPinguicula (fig. 1), occurring not only in nontranscribedspacers (e.g., trnQ-rps16, psaA-ycf3, 30rps12-trnV) but also inthe ycf2 gene region. Several more repeats accumulate inthese regions of the Utricularia and Genlisea plastomes,which also harbor repeat hotspots in the SSC, in the rpoB-psbD fragment, and several more dispersed repeats in the IRs(fig. 1).

Synonymous and Nonsynonymous Substitution Rates

Across all plastid protein-coding genes, we observe a generalelevation of nucleotide substitution rates in Lentibulariaceaecompared with noncarnivorous lamiids (fig. 3a). Rates parti-cularly increase in the bladderwort lineage (PMWU< 0.001 fordS and dN; table 2). Pinguicula, in contrast, shows significantacceleration only in its synonymous rates (PMWU = 0.039).Rate differences are already detectable along the branchesleading to Lentibulariaceae and the Utricularia/Genlisea sub-clade (fig. 3b), implying shared ancient rate accelerations.

FIG. 1. Physical maps and sequence similarity between Lentibulariaceaeand closely related noncarnivorous species. The organization of theplastid genome is depicted as a physical map for each species, showing

FIG. 1. Continuedthe relative strandedness of genes, which are colored according to theirfunctional classes. In some instances, tRNA genes appear as black barsbecause of their minimal length. Gene names are provided on the left,and asterisks indicate the number of introns in a specified gene.Pseudogenes are indicated by �, whereas a red arrow marks the dele-tion of a specified gene region. Based on a plastome alignment of thefour species with progressiveMauve, similarity plots to the right of eachplastome map indicate the hotspots of divergence. Regions (representedby thin bars) colored mauve represent segments conserved among allgenomes, while differently colored ones mark those present only in asubset of taxa. The height of a bar shows the degree of conservation, andgaps in a plot indicate that this particular region could not be aligned toany of the other genomes, suggesting this fragment to be species spe-cific. The approximate localizations of larger repeats (>12 nt) are illus-trated below the similarity plot, with a two-digit number indicatingthe length and orientation of the repeat (F, forward; P, palindromic;IR, large IR).

531

Substitution Rates and Purifying Selection in Lentibulariaceae . doi:10.1093/molbev/mst261 MBE at Pennsylvania State U

niversity on September 2, 2014

http://mbe.oxfordjournals.org/

Dow

nloaded from

--5 3 to ,inverted repeat-',-to -http://mbe.oxfordjournals.org/

Relative to noncarnivores, 23 protein-coding genes in thePinguicula plastome show notable rate deviations, with 20genes being elevated in nonsynonymous rates (fig. 3c; supple-mentary table S1, Supplementary Material online). Of these,13 genes are likely to have experienced rate elevation afterthe split from Genlisea/Utricularia, rather than in theLentibulariaceae ancestor. In Utricularia and Genlisea, 42and 52 genes, respectively, evolve significantly faster with,again, more genes having elevated dN than dS (fig. 3c; sup-plementary tables S1 and S2, Supplementary Material online).

Rate acceleration apparently affects all genes of theLentibulariaceae plastid genomes similarly. Higher rates (dNand dS; fig. 3) can be observed in genes for both the photo-synthesis light and dark reactions (photosystems, electrontransport, RuBisCO) as well as photosynthesis-associatedprocesses (CO2 uptake, heme attachment), housekeepinggenes (ribosomal proteins, polymerase), and genes for light-independent processes (lipid synthesis).

Changes of Selection in Plastid Genes

Significantly relaxed purifying selection in Lentibulariaceaerelative to noncarnivorous lamiids is inferred for the plastid-encoded subunits of the ATP synthase complex (atp genes),photosystem I (psa genes), and the plastid-encoded polymer-ase (PEP, rpo genes). However, the precise modes of selec-tional change differ between these complexes (fig. 3c andtable 3). Although ! in PEP genes appears to be significantly

different in the noncarnivorous clade, the carnivorous ances-tor, and the carnivore clade (i.e., three selectional regimes,likelihood ratio test (LRT) from the best model: P< 0.001),relaxation of purifying selection apparently has affected theATP synthase complex predominantly during the transitionto the carnivorous lifestyle (LRT, best model: P = 0.003).Purifying selection in genes of the photosystem I is relaxedonly in the Lentibulariaceae crown group (LRT, best model:P< 0.001).

Not all genes with elevated rates of nonsynonymouschanges show relaxation of purifying selection. Genewisetests reveal that not all genes of a functional complex displaysimilar patterns of purifying selection. Based on Akaike infor-mation criterion (AIC) model selection, a single global!maybe assumed for four of the six plastid-encoded atp genesand three of five psa genes (supplementary table S2,Supplementary Material online). Purifying selection is relaxedin atpB (LRT, best model P = 0.003), atpI (LRT, best modelP = 0.029), psaB (LRT, best model P = 0.006), and psaC (LRT,best model P< 0.001) of Lentibulariaceae compared withnoncarnivores. RpoB as well as rpoC1 also evolve under re-laxed purifying selection in Genlisea compared with noncar-nivores (rpoB–LRT, best model P = 0.004; rpoC1–LRT, bestmodel P = 0.003). RpoA is the only subunit of the polymerasecomplex to exhibit a significant increase of ! in the branchleading to Lentibulariaceae (LRT, best model P = 0.002; sup-plementary table S2, Supplementary Material online).

0

5

10

15

20

25

30

Sesamum Pinguicula Utricularia Genlisea

No

. Rep

eats

(>12

nt)

Plastid repetitive DNA (>12 nt) forwardrepeatspalindromic repeats

0

10

20

30

40

50

60

70

80

Sesamum Pinguicula Utricularia Genlisea

No

. SS

Rs

(NNNN)≥3(NNN)≥3(NN)≥5

0

100

200

300

400

500

600

700

>30nt

16-30nt

2-15nt

100

200

300

400

500

600

700

Insertions

Deletions

SesamumPinguicula

Utricularia Genlisea

Simple sequence repeatscRIN and cRDN(a) (b)

(c)

FIG. 2. Repeat DNAs in Lentibulariaceae plastid genomes. Number and ratio of (a) 1 nt-corrected relative insertions (cRIN) and deletions (cRDN) ofthree different length categories, (b) di-, tri-, and tetra-SSRs of at least five, three, or three repetitions, respectively, and (c) forward repeats andpalindromic repeats of more than 12 bp length in Sesamum and Lentibulariaceae.

532

Wicke et al. . doi:10.1093/molbev/mst261 MBE at Pennsylvania State U



Dow

nloaded from

--Supporting information online:http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1Supporting information online:http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1-, While --http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1to -to - -- http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/

FIG

.3.

Gra

phi

cals

umm

ary

ofch

ange

sin

nuc

leot

ide

subs

titu

tion

sra

tes

and

sele

ctiv

ere

gim

esin

pla

stid

gen

esof

Len

tibu

lari

acea

e.N

onsy

non

ymou

s(d

N)

and

syn

onym

ous

rate

s(d

S)fr

om78

pla

stid

pro

tein

-cod

ing

gen

esin

Lent

ibul

aria

ceae

and

non

carn

ivor

ous

clos

ere

lati

ves

are

rep

rese

nte

das

biva

riat

ebo

xplo

ts.(

a)Bo

xplo

tba

ghu

llsfo

rdN

and

dSin

Ping

uicu

la,U

tric

ular

ia,a

nd

Gen

lisea

.Pva

lues

show

nar

efr

omW

ilcox

onte

sts

com

par

ing

the

ran

geof

dNan

ddS

for

each

ofth

eth

ree

carn

ivor

es,r

esp

ecti

vely

,wit

hal

lnon

carn

ivor

es.(

b)dN

and

dSfo

rin

tern

albr

anch

es.(

c)G

rap

hica

lsum

mar

ysh

owin

gth

ege

nes

wit

hel

evat

eddN

(abo

vebr

anch

es)

ordS

(bel

owbr

anch

es).

Bran

chle

ngt

his

pro

por

tion

alto

the

num

ber

ofge

nes

that

show

rela

xed

pur

ifyin

gse

lect

ion

atth

esp

ecifi

cn

ode.

An

aste

risk

indi

cate

sge

nes

wit

hac

cele

rate

dra

tes

atal

lchi

ldn

odes

.T

henu

mbe

ran

ddi

stri

buti

onof

sign

ifica

ntl

ydi

ffere

nt

sele

ctiv

ere

gim

esfo

ra

spec

ified

gen

e(l

eft)

orfu

ncti

onal

com

ple

x(r

ight

,bol

dfac

e)is

dep

icte

dbe

low

the

tree

,whe

rep

osit

ion

and

len

gth

ofa

box

indi

cate

the

pos

itio

nof

asi

gnifi

can

tdi

ffere

nt

omeg

aei

ther

alon

gth

ebr

anch

lead

ing

toth

eca

rniv

ores

oron

lyin

the

carn

ivor

ecl

ade

orbo

th.F

orex

amp

le,t

wo

sele

ctiv

ere

gim

esar

ein

ferr

edfo

rrp

oB,w

ith

omeg

aon

the

bran

chle

adin

gto

Lent

ibul

aria

ceae

(T)

bein

gla

rger

than

omeg

ain

the

rest

ofth

etr

ee,i

.e.,

the

Len

tibu

lari

acea

e“c

row

ngr

oup

”su

btre

e(A

)an

dth

en

onca

rniv

ores

(B).

Aha

shke

yin

dica

tes

thre

ese

lect

ive

regi

mes

,an

da

blac

krh

ombu

sm

arks

the

gene

wit

ha

high

erom

ega

only

inth

ebl

adde

rwor

tlin

eage

(in

clud

ing

the

bran

chle

adin

gto

the

blad

derw

ort

linea

ges)

com

par

edto

Ping

uicu

lap

lus

non

carn

ivor

es(s

eeta

ble

3,su

pp

lem

enta

ryta

bles

S2an

dS3

).PS

I,p

hoto

syst

emI

gen

es,P

EP,p

last

id-e

nco

ded

pol

ymer

ase

gen

es.

533




Dow

nloaded from

http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/

Remaining intact genes of the NDH complex in Utriculariaand Pinguicula differ significantly in their respective ! whencompared with noncarnivores. Although ndhE (LRT, bestmodel: P = 0.031) and ndhG (LRT, best model: P = 0.046)show relaxed purifying selection, ndhH (LRT, best model:P = 0.069) and ndhJ (LRT, best model: P = 0.065) exhibitonly marginally different !. NdhB, which is retained also inGenlisea, shows no significant change of !.

Significant relaxation of purifying selection occurs inanother 19 plastid genes across all functional complexes(graphically summarized in fig. 3). These changes apply tothe entire Lentibulariacae, except for the psbH gene, inwhich the change of ! is confined solely to the bladderwortlineage (supplementary tables S2 and S3, SupplementaryMaterial online).

There is a tendency of genes in close proximity to oneanother to exhibit similar changes of selection (atpI/rps2,ndhE/G/psaC, clpP/psbB, psbH/petB/D, rps7/12, atpB/rbcL,rpl33/rps18, psaB/rps14, rpl36/rps8/14, rpl22/rps19/rpl2;figs. 1 and 3). Most of those genes are organized as neighbor-ing genes within the same plastid operon-like transcriptionunit (hereafter: operon). In several cases (petB, rpl2, rpl14,rpl22, rps7, rps12), relaxation of purifying selection in neigh-boring genes has set in independently, that is, at a differenttime during the evolution of Lentibulariaceae (fig. 3c; supple-mentary tables S2 and S3). PsbK and cemA represent the onlycases where we find relaxation of purifying selection (psbK–LRT, best model P = 0.020, cemA–LRT, best model P = 0.046),but not in their neighboring genes. However, the cemA flank-ing genes petA and ycf4 show significant elevation ofnonsynonymous changes in both species of the bladderwortlineage or only in Genlisea (fig. 3c; supplementary table S1,Supplementary Material online), respectively. Genlisea showsacceleration of dN also in psbI, which is localized adjacent topsbK (fig. 3c; supplementary table S1, Supplementary Materialonline).

Evolution of Plastid-Noncoding DNA

Relative to noncarnivorous lamiids, all Lentibulariaceae ana-lyzed here exhibit notably higher rates of nucleotide substi-tutions across the vast majority of the 130 plastid noncodingregions (ncDNA; fig. 4 and table 2). Plastid ncDNA of single-copy regions shows evident sequence divergences in both

introns and intergenic spacers (PMWU< 0.001, all species),whereas ncDNA in the IRs evolves more conservatively andsimilar to noncarnivores (fig. 4 and table 2). Notable diver-gences in spacers and introns are seen most clearly in the IR ofGenlisea (PMWU = 0.049), even though an analogous trendexists also in Utricularia albeit only marginally significant(PMWU = 0.081). Overall, Genlisea accumulates more nucleo-tide changes in ncDNAs than its sister lineage Utricularia(PMWU< 0.001), which evolves much faster than Pinguicula(PMWU< 0.001). No significant differences exist in noncodingregions of the IR between pairs of carnivorous lineages(Pinguicula vs. Utricularia: PMWU = 0.691; Pinguicula vs.Genlisea: PMWU = 0.377, Utricularia vs. Genlisea:PMWU = 0.729; table 2). Notably, rates are elevated in theSSC of Pinguicula and Genlisea, coinciding with the loss ofndh genes. Patterns of ncDNA evolution in the SSC regionof the Utricularia plastome are, however, convergent withthose of noncarnivores. The divergence of ncDNA is mostextreme in the intergenic spacers trnH-psbA, trnK-rps16,rps16-trnQ, and trnG-R, whereas short or internally tran-scribed spacers (e.g., rpoC1-B) show the greatest sequenceconservation (figs. 1 and 4).

We detected a strong correlation between high sequencedivergence and low GC contents (P�< 0.001, �=�0.685;fig. 4; supplementary fig. S1, Supplementary Materialonline). The percentage of GC in nontranscribed spacerstends to be considerably lower than elsewhere in the plas-tome, even though a correlation is insignificant.

Rate acceleration in plastid spacers and introns matchesthe distribution of genes with accelerated dN and/or dS(PSign = 0.016) or relaxed purifying selection in Utriculariaand Genlisea ((PSign = 0.019), respectively. This trend is alsoobvious in Pinguicula, albeit here with marginal significanceonly (PSign = 0.052), indicating that Pinguicula exhibits fewerrate changes in protein-coding regions than in its noncodingregions.

Evolution of Plastome Microstructure

Lentibulariaceae plastomes accumulate much more inser-tions and deletions (indels) than Sesamum with 108 indelsin Pinguicula, 180 in Utricularia, and 242 in Genlisea. Totalingup to 1,030 indels in Genlisea versus 789 in Sesamum (fig. 2),the ratio of insertions to deletions is similar in all taxa.

Table 2. Results Evaluating the Differences of Relative Nucleotide Substitution Rates across All Coding (dN, dS, and dN + dS) and NoncodingRegions from Lentibulariaceae versus Noncarnivores.

Noncarnivores (Sesamum + Lindenbergia + Mimulus) Noncarnivores (Sesamum + Lindenbergia)

78 Protein-Coding Genesa 130b Noncoding Regions

dN + dS dN(min-max)

PWilcoxon-Value (dN)

dS(min-max)

PWilcoxon-Value (dS)

SC + IR l (min-max, SC) SC l (min-max, IR) IR

Pinguicula 0.072 0–0.123 0.746 0–0.201 0.039

Tab

le3.

Om

ega

Val

ues

and

Res

ults

ofM

axim

umLi

kelih

ood

Tes

tsEv

alua

tin

gD

iffer

ent

Mod

els

ofSe

lect

ion

alC

han

ges

inEi

ght

Fun

ctio

nal

Cla

sses

ofPl

asti

dG

enes

betw

een

Len

tibu

lari

acea

ean

dLa

miid

Rel

ativ

es.

Sin

gleu

Mod

elT

wo

Sele

ctiv

ePa

ths

Th

ree

Sele

ctiv

ePa

ths

Bra

nch

(T)

vers

us

Cla

de

A/B

Cla

de

A/B

ran

chve

rsu

sC

lad

eB

Cla

de

Ave

rsu

sC

lad

eB

/Bra

nch

Cla

de

A,

Cla

de

B,

Bra

nch

Gen

eC

lass

xgl

ob

alx

A+

Bx

TP

Val

uea

xA

+T

xB

PV

alu

eax

Ax

B+

TP

Val

uea

xA

xB

xT

PV

alu

ea

Lent

ibul

aria

ceae

node

:Pi

ngui

cula

/Utr

icul

aria

/Gen

lisea

(A)

vers

us

nonc

arni

vore

s(B

)an

dth

ebr

anch

lead

ing

toLe

ntib

ular

iace

ae(T

)

atp

0.08

2<0.

094<

0.10

7

0.08

0<0.

092<

0.10

5

0.07

6<0.

121<

0.18

0

0.34

10.

097<

0.11

5<0.

135

0.05

9<0.

073<

0.08

9

0.00

340.

096<

0.11

5<0.

137

0.06

5<0.

079<

0.09

4

0.01

40.

095<

0.11

5<0.

136

0.05

9<0.

073<

0.08

9

0.07

5<0.

120<

0.17

9

0.01

4

pet

0.05

7<0.

070<

0.08

5

0.05

5<0.

069<

0.08

4

0.03

4<0.

084<

0.16

5

0.68

10.

060<

0.07

9<0.

102

0.04

3<0.

060<

0.08

1

0.23

40.

059<

0.07

9<0.

104

0.04

6<0.

062<

0.08

2

0.31

50.

058<

0.07

9<0.

103

0.04

3<0.

060<

0.08

1

0.03

4<0.

083<

0.16

3

0.49

9

psa

0.03

2<0.

039<

0.04

8

0.03

2<0.

040<

0.05

0

0.00

8<0.

025<

0.05

7

0.34

60.

041<

0.05

3<0.

067

0.01

7<0.

026<

0.03

6

0.00

20.

044<

0.05

6<0.

074

0.01

8<0.

026<

0.03

5

<0.

001

0.04

4<0.

058<

0.07

4

0.01

7<0.

026<

0.03

6

0.00

8<0.

025<

0.05

7

0.00

2

psb

0.03

4<0.

040<

0.04

8

0.03

2<0.

039<

0.04

7

0.02

9<0.

060<

0.10

4

0.28

860.

036<

0.04

6<0.

057

0.02

6<0.

034<

0.04

4

0.12

330.

035<

0.04

5<0.

057

0.02

8<0.

037<

0.04

6

0.29

80.

034<

0.0 4

5<0.

056

0.02

6<0.

034<

0.04

4

0.02

9<0.

059<

0.10

4

0.23

7

rpl

0.28

3<0.

316<

0.35

2

0.28

2<0.

317<

0.35

4

0.19

1<0.

308<

0.46

2

0.93

40.

294<

0.34

0<0.

391

0.24

1<0.

287<

0.33

9

0.30

30.

296<

0.34

6<0.

402

0.29

6<0.

3462<

0.40

2

0.28

00.

296<

0.34

6<0.

401

0.24

1<0.

287<

0.33

9

0.18

8<0.

304<

0.45

7

0.55

1

rpo

0.19

0<0.

202<

0.21

4

0.18

2<0.

194<

0.20

7

0.25

7<0.

316<

0.38

3

0.00

20.

220<

0.23

8<0.

257

0.14

9<0.

164<

0.18

0

<0.

001

0.21

1<0.

229<

0.24

9

0.16

5<0.

180<

0.19

6

0.00

30.

209<

0.22

8<0.

247

0.14

9<0.

164<

0.18

0

0.25

4<0.

313<

0.37

9

<0.

001

rps

0.74

2<0.

803<

0.86

6

0.73

0<0.

792<

0.85

7

0.70

7<0.

989<

1.33

4

0.48

10.

796<

0.88

4<0.

977

0.63

8<0.

718<

0.80

5

0.14

50.

782<

0.87

420<

0.97

4

0.66

5<0.

743<

0.82

6

0.25

50.

780<

0.87

2<0.

971

0.63

7<0.

718<

0.80

4

0.70

3<0.

985<

1.32

9

0.32

3

ndha

0.82

8<0.

873<

0.92

0

0.85

5<0.

905<

0.95

6

0.41

0<0.

58<

0.77

8

0.09

10.

783<

0.85

2<0.

925

0.82

8<0.

888<

0.95

0

0.69

0.80

7<0.

892<

0.98

3

0.80

7<0.

864<

0.92

3

0.77

90.

856<

0.94

5<1.

038

0.82

9<0.

888<

0.95

1

0.39

4<0.

560<

0.75

3

0.21

0

Bla

dder

wor

tlin

eage

:U

tric

ular

ia/G

enlis

ea(A

)ve

rsu

sno

ncar

nivo

res

plu

sPi

ngui

cula

(B)

and

the

bran

chle

adin

gto

the

blad

derw

ort

linea

ge(T

)

atp

0.08

2<0.

094<

0.10

6

0.08

1<0.

0933<

0.10

6

0.06

1<0.

112<

0.18

5

0.60

90.

086<

0.10

7<0.

130

0.07

4<0.

087<

0.10

2

0.20

20.

085<

0.10

7<0.

133

0.07

6<0.

089<

0.10

3

0.26

70.

084<

0.10

6<0.

132

0.07

4<0.

087<

0.10

2

0.06

0<0.

110<

0.18

2

0.44

0

pet

0.05

7<0.

070<

0.08

5

0.05

4<0.

067<

0.08

2

0.06

3<0.

139<

0.26

0

0.14

50.

053<

0.07

4<0.

101

0.05

1<0.

067<

0.08

6

0.67

20.

046<

0.06

7<0.

094

0.05

5<0.

071<

0.09

0

0.79

30.

045<

0.06

6<0.

093

0.05

1<0.

067<

0.08

6

0.06

3<0.

139<

0.26

0

0.34

6

psa

0.03

2<0.

040<

0.04

8

0.03

0<0.

037<

0.04

6

0.03

6<0.

075<

0.13

5

0.10

60.

039<

0.05

3<0.

069

0.02

3<0.

031<

0.04

1

0.02

10.

035<

0.05

0<0.

067

0.02

6<0.

034<

0.04

4

0.11

80.

035<

0.04

9<0.

067

0.02

3<0.

031<

0.04

1

0.03

5<0.

074<

0.13

3

0.04

7

psb

0.03

4<0.

040<

0.04

8

0.03

2<0.

039<

0.04

7

0.03

2<0.

061<

0.10

5

0.22

80.

035<

0.04

6<0.

060

0.02

9<0.

037<

0.04

6

0.24

70.

032<

0.04

4<0.

058

0.03

1<0.

039<

0.04

8

0.56

60.

032<

0.04

3<0.

058

0.02

9<0.

037<

0.04

6

0.03

2<0.

061<

0.10

4

0.35

4

(con

tin

ued)

535




Dow

nloaded from


Of these, 256 indels are found in plastid protein-coding re-gions; an exception are the plastid encoded photosystem Igenes (psa), which lack any length variation. The highestnumbers of indels accumulate in ycf2 (38 indels), rpoc2(30), accD (21), rpl22 (18), matK (17), and ndh genes, withndhF (19) in particular. Across all genes, the longest insertionsor deletions (>9 nt) occur in accD, ycf2, ndh, and PEP genes.

Accumulation of microstructural changes in the differentprotein-coding genes does not generally coincide with ele-vated nucleotide substitution rates in Lentibulariaceae(P�= 0.257; table 4). A correlation can be found, however,for PEP (P�= 0.031) and, with marginal significance, in 50Sribosomal protein genes (P�= 0.054; fig. 5).

In plastid ncDNA, indel evolution is evidently related tonucleotide substitutional patterns (figs. 4 and 5b). Spacersand introns with high sequence divergence significantlyaccumulate more indels than moderately evolving ncDNAregions (P�< 0.001; supplementary fig. S1, SupplementaryMaterial online). Especially, GC-poor regions accumulate in-sertions and deletions (P�< 0.001), perhaps reflecting theobservation of proliferating SSRs in the bladderwort-lineagerelative to Pinguicula and noncarnivorous lamiids.

Autapomorphic indels account for the vast majority ofmicrostructural changes in Lentibulariaceae. Visual inspectionsuggests that only very few are family-specific insertions ordeletions. Absence/presence counts revealed that only fiveindels are apparently unique to Lentibulariaceae in protein-coding regions as opposed to 146 in noncoding regions. Thesenumbers should, however, be regarded with some caution,because an in-depth analyses of indel history requires animproved taxon sampling.

Notably low rates of microstructural changes occur appar-ently every 2–3 kb (figs. 1 and 4). In contrast to the LSC and IR,this pattern is less prominent in the SSC, likely because of theobserved gene losses in the large ndh operon. Distance-relatedalternation of GC and substitution rates is much less pro-nounced (fig. 5).

Discussion

Architecture of Plastid Genomes in Lentibulariaceae

Complete plastome sequencing and comparative genomeanalyses revealed distinct paths of plastid genome evolutionin the three carnivorous genera of Lentibulariaceae. The struc-ture of Lentibulariaceae plastomes is mostly similar to that ofclosely related noncarnivorous lamiids (fig. 1), with theUtricularia lineage evolving most conservatively in terms ofgene synteny. Similar results were also reported for the plas-tome of U. gibba based on a draft assembly of the organellargenomes (Ibarra-Laclette et al. 2013), which was publishedduring the final stages of writing this manuscript and couldtherefore not be considered in our analyses. Local disruptionsof gene order occur in P. ehlersiae and G. margaretae becauseof the loss of some ndh genes. Phylogenetic relationships andthe kind and number of missing genes or pseudogenesstrongly suggest that the loss of the ndh genes has occurredindependently in the two lineages. As 8 out of 11 genes weredeleted in G. margaretae and four gene losses are observed inTa

ble

3.C

onti

nue

d Sin

gleu

Mod

elT

wo

Sele

ctiv

ePa

ths

Th

ree

Sele

ctiv

ePa

ths

Bra

nch

(T)

vers

us

Cla

de

A/B

Cla

de

A/B

ran

chve

rsu

sC

lad

eB

Cla

de

Ave

rsu

sC

lad

eB

/Bra

nch

Cla

de

A,

Cla

de

B,

Bra

nch

Gen

eC

lass

xgl

ob

alx

A+

Bx

TP

Val

uea

xA

+T

xB

PV

alu

eax

Ax

B+

TP

Val

uea

xA

xB

xT

PV

alu

ea

rpl

0.28

3<0.

316<

0.35

2

0.28

2<0.

317<

0.35

4

0.17

8<0.

307<

0.48

0

0.93

70.

314<

0.37

3<0.

438

0.24

4<0.

283<

0.32

6

0.10

20.

322<

0.38

9<0.

463

0.24

8<0.

285<

0.32

6

0.08

00.

322<

0.38

8<0.

463

0.24

5<0.

284<

0.32

7

0.17

2<0.

299<

0.46

9

0.21

3

rpo

0.19

0<0.

202<

0.21

4

0.19

0<0.

202<

0.21

5

0.15

7<0.

206<

0.26

2

0.92

20.

217<

0.23

8<0.

260

0.16

7<0.

181<

0.19

6

0.00

10.

223<

0.24

7<0.

272

0.16

9<0.

183<

0.19

7

<0.

001

0.22

2<0.

246<

0.27

1

0.16

8<0.

182<

0.19

6

0.15

4<0.

198<

0.25

3

0.00

3

rps

0.74

2<0.

803<

0.86

6

0.72

7<0.

789<

0.85

4

0.73

5<1.

039<

1.40

9

0.42

10.

742<

0.83

8<0.

943

0.70

4<0.

779<

0.86

0

0.61

90.

711<

0.81

5<0.

929

0.72

2<0.

796<

0.87

5

0.87

60.

705<

0.80

8<0.

920

0.70

3<0.

779<

0.86

0

0.73

04<

1.03

28<

1.40

07

0.70

4

ndha

0.82

8<0.

873<

0.92

0

0.82

5<0.

876<

0.92

9

0.67

8<0.

852<

1.03

8

0.90

20.

752<

0.84

4<0.

942

0.82

9<0.

884<

0.94

1

0.71

40.

596<

0.76

9<0.

958

0.83

7<0.

888<

0.94

2

0.60

00.

488<

0.63

8<0.

801

0.82

8<0.

883<

0.94

0

0.91

4<1.

123<

1.34

7

0.72

5

NO

TE.—!

valu

esar

egi

ven

wit

hlo

wer

and

upp

erco

nfide

nce

leve

ls;

AIC

scor

esw

ere

empl

oyed

todi

scrim

inat

ebe

twee

nth

edi

ffere

ntm

odel

s,an

dth

ebe

stfit

isin

dica

ted

init

alic

s.a P

valu

esar

ede

rived

from

LRT

sve

rsus

the

sing

leo

mod

el(s

eeM

ater

ials

and

Met

hods

for

deta

ils).

536




Dow

nloaded from

significantly http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1-,---triculariainguiculaenliseaeight http://mbe.oxfordjournals.org/

0246810121416

Indel Frequency

0510152025303540

% Divergence

trnH-psbApsbA-trnKtrnK-matkmatk-trnK

trnK-rps16rps16-intronrps16-trnQtrnQ-psbKpsbK-psbIpsbI-trnS

trnS-GtrnG-intron

trnG-RtrnR-atpA

atpA-FatpF-intron

atpF-HatpH-I

atpI-rps2rps2-rpoC2

rpoC2-C1rpoC1-intron

rpoC1-BrpoB-trnCtrnC-petN

petN-psbMpsbM-trnD

trnD-YtrnY-EtrnE-T

trnT-psbDpsbC-trnStrnS-psbZpsbZ-trnG

trnG-fMtrnfM-rps14rps14-psaBpsaB-psaApsaA-ycf3

ycf3-int1ycf3-int2ycf3-trnStrnS-rps4rps4-trnT

trnT-LtrnL-intron

trnL-FtrnF-ndhJ#ndhJ-ndK#ndhK-trnV#

trnVintrontrnV-trnMtrnM-atpEatpB-rbcLrbcL-accDaccD-psaI psaI-ycf4

ycf4-cemA cemA-petApetA-psbJ

psbJ-LpsbL-F psbF-E

psbE-petLpetL-G

petG-trnWtrnW-P

trnP-psaJpsaJ-rpl33

rpl33-rps18rps18-rpl20rpl20-rps12rps12-clpP

clpP-intron1clpP-intron2

clpP-psbBpsbB-TpsbT-NpsbN-H

psbH-petBpetB-intron

petB-DpetD-intronpetD-rpoArpoA-rps11rps11-rpl36

rpl36-infAinfA-rps8

rps8-rpl14rpl14-16

rpl16-intronrpl16-rps3rps3-rpl22

rpl22-rps19rpl19-rpl2rpl2-intronrpl2-rpl23rpl23-trnItrnI-ycf2

ycf2-ycf15ycf15-trnLtrnL-ndhB

ndhB-intronndhB-rps7rps7-rps12

rps12-intronrps12-trnVtrnV-rrn16rrn16-trnI

psbM-trnDtrnI-A

trnA-introntrnA-rrn23

rrn23-rrn45rrn45-rrn5rrn5-trnR

trnR-NtrnN-ycf1

rpl32-trnLtrnL-cssA

cssA-ndhD#ndhD-psaC#

psaC-ndhEndhE-G#ndhG-I#ndhI-A#

ndhA-intron#ndhA-H#

ndhH-rps15#rps15-ycf1

010203040506070

%GC

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

µ

Ses

amum

Lind

enbe

rgia

Pin

guic

ula

Utr

icul

aria

Gen

lisea

IRS

SC

LS

C

(# w

ithou

t Gen

lisea

)

(a)

(b)

(c)

(d)

FIG

.4.

Nuc

leot

ide

subs

titu

tion

rate

san

din

delf

requ

enci

esin

pla

stid

non

codi

ngD

NA

.(a)

Subs

titu

tion

rate

sfo

ral

lpla

stid

non

codi

ng

regi

ons

are

dep

icte

dfo

ral

lLen

tibu

lari

acea

ean

dtw

on

onca

rniv

ores

inth

eup

per

diag

ram

,whe

rea

row

rep

rese

nts

ap

last

id-n

onco

din

gre

gion

s(a

sin

dica

ted

inth

ebo

ttom

ofth

efig

ure)

and

the

taxo

nid

enti

tyis

mar

ked

byco

lore

ddo

tsor

squa

res.

The

degr

eeof

(b)

dive

rgen

ce,(

c)th

efr

eque

ncy

ofin

sert

ion

and

dele

tion

sev

ent,

and

(d)

the

GC

con

tent

isill

ustr

ated

for

allp

last

idin

tron

san

dsp

acer

sfr

omLe

nti

bula

riac

eae

and

the

non

carn

ivor

esba

sed

onth

eal

ign

edda

tase

ts.D

ashe

dlin

esm

ark

the

jun

ctio

ns

ofsi

ngle

copy

regi

ons

and

the

IR.“

#”m

arks

regi

ons

wit

hla

ckin

gda

tadu

eto

gen

elo

ssin

Gen

lisea

.

537




Dow

nloaded from


P. ehlersiae, we believe that these events coincide with thefunctional loss of the plastid NAD(P)H dehydrogenase com-plex. In the light of the extreme rare occurrence of functionaltransfer of genes from the plastid to the nuclear genome inangiosperms (e.g., Bock and Timmis 2008), we appraise

the replacement of these genes by nuclear copies as consid-erably low.

During photosynthesis, the NAD(P)H dehydrogenase com-plex located in the chloroplast thylakoid membrane mediateselectron cycling around the photosystem I and thereby

R² = 0.3097(pρ = 0.03)

R² = 0.2406(pρ = 0.05)

R² = 0.3459

R² = 0.2551

R² = 0.0457

R² = 0.1416R² = 0.1332

100

200

300

400

500

0 1 2 3 4 5 6 7 8 9 10 11 12 13

Nu

cleo

tid

e S

ub

stit

uti

on

s

rpo

rpl

rps

ndh

pet

psb

atpTrend across 78 genes

R² = 0.9399(pρ < 0.001)

500

1000

1500

2000

2500

3000

3500

4000

0 100 200 300 400 500 600 700 800 900

Indels

130 nc regions

Spacer

Introns

R² = 0.9293(pρ < 0.001)

R² = 0.9026(pρ < 0.001)

(a)

(b)

FIG. 5. Correlation of microstructural changes and substitution rates in different functional classes of plastid genes. The correlation of indels andsubstitution rates (total) is illustrated (a) for all plastid gene classes and across all plastid protein genes, and (b) for spacers, introns, and across all plastid-noncoding regions. The goodness of fit for each of the tested region is provided by R2 (color coded). Where present, P values from Spearman testsindicate a significant or marginally significant correlation between the number of indels and substitutions.

Table 4. Results of Spearman Tests Evaluating the Correlation of Nucleotide Substitution Rates and Indel Events in Plastid Coding and NoncodingRegions.

78 Genesa atp ndh pet psa psb rpo rpl rps 130b nc Regions Introns Spacer

n 11 11 11 11 n.a. 11 11 11 11 9 9 9

rs 0.373 0.391 0.414 0.194 n.a. 0.593 0.650 0.297 0.444 0.970 0.950 0.964

t 1.21 1.27 1.36 0.59 n.a. 2.21 2.56 0.93 1.49 10.47 8.05 9.59

df 9 9 9 9 n.a. 9 9 9 9 7 7 7

p1-tailed 0.129 0.118 0.103 0.285 n.a. 0.027 0.015 0.1881 0.085

fine-adjusts the ratio of ATP and the electron carrierNAD(P)H. The latter is used to synthesize monosaccharidesby fixating CO2 in the Calvin cycle (Peltier and Cournac 2002;Rumeau et al. 2007). The NAD(P)H dehydrogenase appears toplay an especially important role under light-stress conditionsor low CO2 concentration, under both of which ndh mutantsshowed significant impairments of photosynthesis (Horvathet al. 2000); no such effects were seen under nonstress con-ditions and high CO2 concentration. Deletion of ndh genesfrom the plastome was reported earlier for some autotrophicplants (Geraniaceae: Chumley et al. 2006, Blazier et al. 2011;Guisinger et al. 2011; conifers and Gnetales: Brauckmann et al.2009; Wu et al. 2009). In their natural habitats, carnivorousLentibulariaceae are exposed to very much the same stressfactors as most other plants, which includes the occasionaloccurrences of light stress. However, the loss of the NAD(P)Hcomplex might not affect the fitness of Pinguicula andGenlisea or other lineages lacking plastid ndh genes. It is pos-sible that these plants regulate gas exchange such that CO2concentrations remain sufficiently high in the cells or thatthey make use of organic carbon obtained through prey toovercome periods of environmental stress (e.g., Juniper et al.1989)—resembling the situation in some parasitic plants tosome extent (Krause 2011).

Elevated amounts of plastid repetitive DNA (includingSSRs) and low numbers of repeats longer than 50 bp charac-terize the plastomes of the bladderwort lineage (fig. 2). Someof these repeats colocalize with mutational hotspots in theplastomes of Lentibulariaceae (fig. 1). However, the overallabundance of longer repeats in Lentibulariaceae plastomesis still too low to formally explore a correlation betweenmutational hotspots and longer repeats, although these areknown to drive molecular evolution (e.g., McDonald et al.2011). Larger repeats are generally rare among angiospermplastomes and considered to promote illegitimate recombi-nation (Raubeson et al. 2007). Proliferation of long repeats,otherwise, has only been reported for a number of plants withhighly reconfigured plastomes (Cosner et al. 2004; Cai et al.2008; Guisinger et al. 2011; Sloan et al. 2012) or for parasiticplants undergoing relaxed selection of the photosynthesisapparatus (Wicke 2013). Impaired function of DNA repairenzymes or relaxed selection (or both in concert) havebeen suggested to contribute to the proliferation of recom-binogenic factors in plastomes (Guisinger et al. 2008, 2011;Wicke et al. 2013) and might likely result in significant accel-eration of both nonsynonymous and synonymous changesand the accumulation of indels.

Evolution of Substitution Rates and Selective Regimesin Plastid Genes

Across the vast majority of plastid coding and single-copynoncoding regions, Pinguicula, Utricularia, and Genliseaexhibit elevated substitution rates compared with closely re-lated noncarnivorous lamiids (table 2 and fig. 3). Acceleratedrates occur in all gene classes of the Lentibulariaceae plas-tomes, affecting the normally highly conserved plastid-encoded fractions of photosystems, photosynthetic electron

transport chains, the thylakoid ATP synthase complex, andgenes for transcription and translation.

Plastid genes of Lentibulariaceae show disproportionally(i.e., more frequently) an elevation of nonsynonymous ratesthan of synonymous rates. This trend is especially pro-nounced in the bladderwort lineage, where more than halfof the protein-coding genes are sped up significantly in bothdN and dS. Although the number and type of faster evolvinggenes differ notably among the Lentibulariaceae species, rateacceleration is not totally random. Accelerated genes sharedby all lineages suggest changes in the molecular evolution ofsome housekeeping genes (e.g., matK, rpo genes) and electrontransport genes that have taken place already in the ancestorof Lentibulariaceae.

In contrast to the general evolutionary patterns seen acrossangiosperms (Jansen et al. 2007), Lentibulariaceae do notshow a correlation of indels and substitution rates across allplastid-coding regions, and most gene classes (except for PEP,and rpl genes to a smaller extent) have accumulated morenucleotide substitution than indels. Even though we cannotrule out sampling artifacts contributing to this result, thiscould imply that the accumulation of nucleotide substitu-tions in plastid protein-coding regions of Lentibulariaceaeproceeds faster than the accumulation of microstructuralchanges.

Changes of selective constraints either affect the entireclade of Lentibulariaceae (photosystem I and other photosyn-thesis genes, clpP, rps14; fig. 3c) or are confined to the blad-derwort lineage (ATP synthase, several photosynthesis andhousekeeping genes; fig. 3c). We also detected relaxed purify-ing selection in plastid-encoded genes for transcription (PEP),transcript maturation (matK), translation (ribosomal proteingenes), and protein turnover (clpP). The polymerase complexis inferred to have experienced a two-step relaxation of puri-fying selection in Lentibulariaceae: once during the transitionto carnivory (i.e., on the branch leading to Lentibulariaceae)and later during the formation of the bladderwort lineage, i.e.,along the branch uniting Utricularia and Genlisea (fig. 3c;supplementary tables S2 and S3, Supplementary Materialonline). Unlike in the other protein–gene classes, the evolu-tion of indels and nucleotide substitution also correlatessignificantly in the rpo genes. This deviating trend of micro-structural changes together with the detected relaxation ofpurifying selection probably indicates that PEP genes havebecome pseudogenes or are about to become pseudogenesin Lentibulariaceae. This hypothesis would invoke a shift awayfrom PEP to the nuclear-encoded polymerase in plastid tran-scription, for which there is no data available as of writing thisarticle. On the other hand, the evolution of the rpo genesthemselves may be the result of mutations in the replicationand or repair machinery.

Interestingly, some subunits of the different plastid func-tional complexes evolve under more relaxed constraints thanothers. Future studies, which will also need to consider pro-tein-domain interactions, are required to shed light on howmuch variation is permitted in protein complexes to retainprotein function.

539




Dow

nloaded from

-,-,,s -- -,; Guisinger etal.,--to-, ,S-,http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1-http://mbe.oxfordjournals.org/

In a considerable number of cases, rate elevation and re-laxation of purifying selection affect genes that are encoded inclose proximity to one another. Higher substitution rates in agene appear to influence its neighboring genes, even thoughthey might be of a different function. Interestingly, genes inthe highly reconfigured plastome of Pelargonium (Guisingeret al. 2008) show a comparable, yet less prominent, patternof higher dN/dS values in neighboring genes. “Localizedhypermutation” has also been described in Fabaceae, particu-larly around the accD-ycf4 region. A plastome-wide analysis ofsubstitution rate elevation in Lathyrus also revealed such aneighboring effect for a few other regions, although to a muchsmaller extent (e.g., rpl16-rpoA, ndhH-ycf1; Magee et al. 2010).Potential reciprocal effects of a similar kind have been ob-served with regard to the series of functional and physicalgene losses in nonphotosynthetic parasitic plants (Wicke2013; Wicke et al. 2013), where DNA deletions seem to begoverned by protection from deletion by essential genes.More data are currently being compiled to test this patternin the framework of spatial autocorrelation and potentiallyunderlying molecular mechanisms.

Evolution of Noncoding DNA

Nearly all plastid ncDNAs in Lentibulariaceae evolve with ac-celerated substitution rates compared with noncarnivores.Many of the extremely accelerated spacers and introns arelocated directly adjacent to or are enclosed by protein-codinggenes with elevated dN, dS, or relaxed purifying selection. Thispattern is particularly prominent in the bladderwort lineageand suggests the existence of various mutational hotspots inthe plastomes of Lentibulariaceae.

A strong correlation of indels and substitution rate isevident in plastid noncoding regions of all Lentibulariaceae(fig. 4). The extent of divergence in spacer regions is obviouslystrongly related to its GC content. AT richness contributes topotentially deleterious recombination. Long homopolymerstretches and microsatellite regions favor replication errors(Levinson and Gutman 1987; Wagner et al. 1990; Wolfsonet al. 1991; Müller et al. 1999), which may explain theproliferation of microstructural changes in the plastomes ofLentibulariaceae (fig. 2; supplementary fig. S1, SupplementaryMaterial online). Due to the current lack of comparable data,it remains elucidated whether the evolutionary patterns ofplastid ncDNAs are characteristic for Lentibulariceae or ifthese can also be observed in other angiosperm plastomes(Müller and Wicke 2013).

Implications of Relaxed Purifying Selection in PlastidGenes of Lentibulariaceae

Several hypothesized biological phenomena underlying ratevariation in angiosperms have already been excluded as po-tential explanations for Lentibulariaceae (e.g., generation timeeffect: Jobson and Albert 2002; Müller et al. 2004; speciationrate hypothesis: Müller et al. 2006), while differences in met-abolic rates (Martin and Palumbi 1993) and unequal effi-ciency of DNA repair (Britten 1986) are difficult to assessand still await experimental testing in this family. Taking

into account the extraordinary physiological differentiationsof Lentibulariaceae, Albert et al. (2010) formulated an alter-native mechanistic hypothesis, according to which the geno-mic peculiarities seen in Utricularia may be a consequence ofhigh levels of mutagenic reactive oxygen species (ROS) as theresult of a short circuit in the mitochondrial respiratory chain.Due to a mutation in cytochrome c oxidase (Jobson et al.2004; Laakkonen et al. 2006), this shunt possibly providesenergy from sequestering protons more rapidly for theactive prey capture mechanism (“firing”) of Utricularia traps(Albert et al. 2010). Although several lines of evidence are insupport of this hypothesis (e.g., Ibarra-Laclette et al. 2011), afew details in the observed rate patterns in plastid genomesare difficult to explain when interpreted exclusively in thelight of the ROS hypothesis (as it is currently phrased). Rateacceleration as well as relaxation of purifying selection alsooccur in Pinguicula, which does not share the COX mutationwith the bladderwort lineage. Substitutional and microstruc-tural changes are even higher in Genlisea, which captures itsprey exclusively passively, other than in Utricularia. Likewise,ultrastructural data from Utricularia traps provide evidencefor the presence of passive traps in basal lineages of the genus(Müller et al. 2004; Reifenrath et al. 2006), resulting in a morelikely evolutionary scenario of a most recent common ances-tor of the bladderwort lineage that had not captured preyactively and thus may have benefited less from the shortcircuit in the mitochondrial respiratory chain. Even thoughfull genome sequencing showed that the organellar genomesevolve more conservatively and apparently uncoupled fromthe evolutionary dynamics of the nuclear genome in U. gibba(Ibarra-Laclette et al. 2013), further comparative genomicstudies based on a dense taxonomic sampling will be veryhelpful in evaluating the role of ROS in shaping genomearchitectural traits in Lentibulariaceae. Most DNA damageis likely to occur at the immediate source of ROS, i.e., in mi-tochondria (Friedberg et al. 2006), so particularly comparativemitochondrial genome analyses are required to specificallytest the outreach of the ROS hypothesis.

As mentioned above, the patterns of rate elevation andrelaxation of selection as well as the loss of ndh genes inplastid genomes of Lentibulariaceae are also interesting be-cause of its similarities to obligate photosynthetic parasiticplants like Cuscuta (McNeal, Arumugunathan, et al. 2007;McNeal, Kuehl, et al. 2007) and Orobanchaceae (Wicke2013; Wicke et al. 2013). Given these convergences with pho-tosynthetic parasites regarding molecular evolutionary pat-terns, it stands to reason that alternative paths of acquiringnutrients may likely promote the rapid evolution of plastidgenes in Lentibulariaceae. Even though all Lentibulariaceae arephotosynthetic and capable of carbon fixation, carnivorousplants have been demonstrated to counterbalance negativephotosynthesis benefits due to unfavorable environmentalconditions by the uptake of nitrogen, phosphorous, and or-ganic carbon from their prey (Juniper et al. 1989; Adamec1997). Besides metabolized prey, a close interaction with mi-crobial communities in or near traps of Utricularia has beenspeculated to contribute to alternative paths of nutrient up-takes (Sirová et al. 2010), which therefore might change

540




Dow

nloaded from

also also -to ---http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1to be ,a,-While the t-,http://mbe.oxfordjournals.org/

selection pressures in normally highly conserved genomicfractions.

Although causation remains to be elucidated, relaxed pu-rifying selection in photosynthesis and photosynthesis-relatedplastome regions appears to be mirrored to some extent inaverage net photosynthesis rates of carnivorous plants. Ellison(2006) states that both the photosynthetic nutrient use effi-ciency as well as the photosynthesis rate per leaf mass isconsiderably lower in aquatic and terrestrial carnivorousplants than in noncarnivores. Estimates range between 1and 5% lowered photosynthesis rates in carnivores (Mendezand Karlsson 1999; Adamec 2011). Subarctic Pinguicula spe-cies were even reported to show a reduction of both leaf andmass-based photosynthesis rates by 30% compared with geo-graphically similarly distributed noncarnivorous plants,including hemiparasitic plants (Mendez and Karlsson 1999).These results are, however, confounded by reports of veryhigh photosynthesis rates in aquatic Utricularia species rela-tive to other aquatic plants (Adamec 2009; Sirová et al. 2010).Interestingly, these high photosynthesis rates appear to di-rectly benefit the prey capturing and cultivation of microbialcommunities in the traps of the investigated bladderworts(Sirová et al. 2010).A denser sampling of genome-scale datawill be needed to thoroughly evaluate causes and conse-quences of genomic evolutionary patterns in relation tonutritional and physiological peculiarities in carnivorousplants in general, and in Lentibulariaceae in particular.

Materials and Methods

Plant Material and Shotgun Sequencing

Pinguicula ehlersiae (cultivated at the Bonn BotanicalGardens, Germany, voucher: BONN, 3252-1998, B.Schäferhoff 60, 2009-01-05), U. macrorhiza (collected inthe Black Moshannon Lake, PA, voucher: PAC, leg./det.: B.Schäferhoff and S. Wicke, 2009-07-22), and G. margaretae(cultivated and kindly provided by Dr A. Fleischmann,University of Munich, Germany; voucher: M, A.Fleischmann Z29, 2006-09-27), representing the threegenera of Lentibulariaceae, were selected for wholegenome shotgun-sequencing. Total genomic DNA was ex-tracted from fresh plant material using a modified cetyltrimethylammonium bromide (CTAB) method, which in-cluded adjustment of CTAB extraction buffer to pH 9.8 tomaximize DNA yield. All samples were treated with RNAse(Qiagen) before purification by PEG-8000 precipitation(McNeal et al. 2006). High-molecular weight DNA (between3.5 and 5mg) was subjected to whole genome shotgun se-quencing with 454 FLX Titanium at the Genomics CoreFacility of the Pennsylvania State University, UniversityPark, PA.

Plastid Genome Reconstruction andSequence Finishing

Pyrosequencing data were quality trimmed and clipped off ofthe linker sequences plus the adjacent 15 nucleotides. Theprocessed data sets contained 555,679 reads (159 Mbp) forP. ehlersiae, 521,827 reads (121 Mbp) for U. macrorhiza, and

356,598 reads (96 Mbp) for G. margaretae. Reads were assem-bled using MIRA v3 (Chevreux et al. 1999; Chevreux 2011)under the accurate, de novo, genomic assembly mode.

Contigs were sorted according to their similarity to knownplastid protein-coding genes using the local NCBI Blast pack-age and a custom plastid gene database. Plastid-like contigswere extracted from the contig pool (BlastX evalue: 5e�30,BlastN evalue: 5e�50) and automatically postassembled inCAP3 (Huang and Madan 1999) using a minimum identity of97%, a minimal overlap of 100 nt, and a gap length of max.5 nt. This approach produced two or three supercontigs inGenlisea and Pinguicula, respectively, and six supercontigs inUtricularia, all that were joined manually. All manual overlapsand potential conflicts in contig ends as well as junctionsbetween plastome single-copy and inverted-repeat regionswere verified by PCR and Sanger resequencing. Assemblybreakpoints included the IR-SSC junction (around the ycf1region) and trnH-psbA in all species, the trnS-trnG region inUtricularia and Pinguicula, and rps16-trnQ, psaA-ycf3, andndhG-ndhI in Utricularia. Except for the ycf1 region, whichcontained very long homopolymer stretches, all contig break-points appeared to have been caused by alignment issues inshort repeat- or microsatellite-rich regions during theautomatic assembly rather than because of the presence ofdivergent copies from the mitochondrial or the nucleargenome.

We mapped all 454 reads to the reconstructed draft plas-tome sequence in Geneious Pro v5.6 to further verify thecorrect assembly of the plastid chromosome; read mappingresults are shown in supplementary figure S2 (SupplementaryMaterial online). The coverage of the final complete se-quences were 37.7� in Pinguicula (plastid read abundance:2.87%), 58.8� in Utricularia (plastid read abundance: 7.59%),and 84.8� in Genlisea (plastid read abundance: 11.94%).

Annotation and Structural Analysis

Plastome sequences were annotated with DOGMA (DualOrganellar GenoMe Annotator; identity cutoff for protein-coding genes: 35%, tRNA identity cutoff: 90%, e value: 5;Wyman et al. 2004) with manual refinements regardinggene start, gene stop, and exon–intron boundaries. The com-pletely annotated sequences have been submitted to theEuropean, American and Japanese (EMBL/GenBank/DDBJ)nucleotide database collaboration.

Similarity between Lentibulariaceae and close relatives wasassessed using progressiveMauve 3.1 (Darling et al. 2010),assuming colinear genomes with an initial seed weight of 17and a gap open penalty of �200 to account for small gapscaused by localized deletions.

Because of the good results (assessed by eye) and fewercomputational resources and time required for the multiplesequence alignment of complete plastomes, we used MAFFTv6 (Katoh and Toh 2008) under the FFT-NS-I mode and the1PAM/�= 2 scoring matrix. Based upon those alignments, weemployed the Indel Analysis tool v1.0.0 of the Galaxy suite(Blankenberg et al. 2010) to obtain genome-wide counts ofinsertions and deletions relative to a reference genome

541




Dow

nloaded from

--to -tricularia, USA& enlisea.prior toBetween 3.5 and 5 µg of , USAinguiculatriculariaenliseab---which -have enhttp://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1XXX--http://mbe.oxfordjournals.org/

(Sesamum). The results were grouped into three size classes(2–15 nt, 16–30 nt, and >30 nt) with the number of 1 ntindels being counted separately, because this type is oftenan artifact from sequencing and/or read assembly (e.g.,Moore et al. 2006). A corrected relative insertion numberand deletion number (cRIN, cRDN), respectively, was there-fore calculated as the difference between the total number ofindels and the number of 1 nt indels. For the analyses of SSRsand other repeated elements, we removed one IR of each ofthe analyzed species of Lentibulariaceae, because the IR isalready defined as a large repeat and subrepeats may there-fore be counted twice. Di-, tri-, and tetrarepeat units occurringat least five, three, and three times, respectively, were analyzedwith the help of the SSR extractor tool, excluding cases of SSRscomprising repeated subunits (e.g., TATA, TGTG). REPuter(Kurtz et al. 2001) was employed to search for all forwardand palindromic repeats of min. 12 nt length (Hamming dis-tance: 3, e value: 5e�5).

We created separate alignments for coding and noncodingregions for analyses of microstructural changes, for which wepreferred the use of automatic alignments without subse-quent manual modifications to ascertain both reproducibilityand accurate data processing by batch scripting. The conca-tenated and annotated data set comprised 78 unique pro-tein-coding genes of all Lentibulariaceae plastomes and fourrelated lamiid taxa: Mimulus guttatus (Mimulus GenomeProject, DoE Joint Genome Institute), Lindenbergia philippen-sis (HG530133), S. indicum (NC_016433), and Nicotiana taba-cum (NC_001879). The gene ycf1 was excluded because ofunresolvable sequencing/assembly errors in homopolymerstretches. Data sets were aligned with PRANK using the“align translated codons” mode (Löytynoja and Goldman2005), because this approach outperformed MAFFT v6 forsingle gene alignments in terms of quality (assessed by eye).

Mimulus was excluded from the analyses of plastid-noncoding DNA (ncDNA), as sequence data of its intergenicspacers was unavailable as of this writing. The ncDNA data setcomprised 130 unique regions from Nicotiana, Lindenbergia,Sesamum, and Lentibulariaceae. Because of gene loss, homol-ogy assessment was critical for some spacers in the Genliseaplastome, and we therefore excluded its trnF-ndhK, ndhJ-K,ndhC-trnV, as well as its SSC ncDNA. Despite pseudogeniza-tion of infA in Nicotiana (Shinozaki et al. 1986; Millen et al.2001), we included both the rpl36-�infA and �infA-rps8spacers due to the overall high conservation of the region.Alignment of spacers and introns was carried out in PRANKusing the structural model “genomic” for the alignment; a ts/tv ratio of 2, a gap rate of 0.05, and a mean gap length distri-bution of 5 (for all regions).

Rather than counting indels relative to a reference as doneto obtain a general overview across the entire plastome (seeabove), we inferred length mutational variation in codingregions and ncDNA in more detail with SeqState 1.4.1(Müller 2005) under the simple-indel coding option.Correlated evolution of microstructural changes and substi-tution rates was tested by parsimony-based optimization ofindels and total substitution rates on a specified tree asdescribed in Jansen et al. (2007).

Substitution Rate Analyses

Genewise and gene-class-specific nonsynonymous (dN) andsynonymous (dS) rate changes were analyzed by LRTs inHyphy 2.11 (Kosakovsky-Pond et al. 2004) based on align-ments of Pinguicula, Utricularia, and Genlisea with the fournoncarnivorous species Sesamum, Lindenbergia, Mimulus,and Nicotiana. Besides powerful hypothesis testing using max-imum likelihood approaches and several other usefulsequence data analysis tools, the Hyphy software offersmore flexibility than other currently available programs be-cause of its own scripting (batch) language (HBL), whichallows users to design customized tests. A specified lamiidtopology that reflects generally established phylogenetic rela-tionships of these taxa (cf. fig. 3) was the basis for all tests. Inbrief, an unconstrained likelihood function (LF) uses a fullmodel (with local parameters for both dN and dS), and thedata were optimized. The resulting maximum likelihoodestimates for the parameters dN and dS were extracted forsubsequent graphical representation. To test for the signifi-cance of differences across species, a constrained LF thatassumes dN or dS to be equal in a selected pair of taxa wasoptimized, and the difference of likelihood scores from con-strained and unconstrained functions (LRT statistic) was eval-uated by w2 distribution.

Using custom R scripts, unpaired Wilcoxon tests with se-quential alpha error (Bonferroni) correction were employedto evaluate cross-species differences of dN, dS, and ! amongLentibulariaceae and noncarnivores. Results of pairwise anal-yses were visualized as bivariate boxplots (Rousseeuw et al.1999).

Changes in ! across clades and branches were evaluatedusing the HBL scripts “SelectionLRT” and “TestBranchDNDS”(Frost et al. 2005; Kosakovsky-Pond et al. 2007). SelectionLRTtests whether differences in selection exist between twoclades, A and B, and the branch T leading to A, using LRTsagainst a global-! model. AIC is then used to select frommodels, assuming either two different ! (one for A + Band one for T; one for A + T and one for B; one for A andone for B + T) or three different! (each one for A, B, and T).The test requires two distinct clades consisting of two distinctbranches with at least two terminal taxa each. SelectionLRTwas run genewise and for concatenated data sets, the latterrepresenting distinct functional complexes with roles in eitherphotosynthesis or housekeeping. As we aimed at unravelingselective changes coinciding with the transition to carnivoryand with further differentiation of the bladderwort lineagewithin Lentibulariaceae, two sets of analyses were conducted:1) treating the Lentibulariaceae crown group as clade A, thebranch leading to Lentibulariaceae as T, and the noncarnivo-rous rest of the tree as B, and 2) treating the terminalbranches Utricularia + Genlisea as A, the branch leading tothis clade as T, and Pinguicula plus the noncarnivores as B.Taxa lacking a certain gene were excluded from the respectivetests (e.g., infA without Nicotiana). As both Pinguicula andGenlisea have lost ndhC, ndhD, ndhF, and ndhK, theLentibulariaceae clade collapsed (with respect to the require-ments for the SelectionLRT script), so tests for selectional

542




Dow

nloaded from

--simple sequence repeats ()--533 s--esamum-likelihood ratio tests ()--``''if ,``''i.-ii.-http://mbe.oxfordjournals.org/

changes in these genes between Utricularia and noncarni-vores were conducted with TestBranchDNDS, which allowstesting differences in ! between a single branch and the restof a given tree. All tests were run under the MG94xREVmodel, allowing for equivalent amino acid changes and as-suming two independent gamma/beta distributions foracross-site variation in both dN and dS and three differentrate classes.

A custom HBL script was employed to perform pairwisenucleotide substitution rate tests under the GTR + G + Imodel in ncDNA regions between each of theLentibulariaceae taxa, Sesamum, and Lindenbergia; Nicotianawas used as outgroup. LRTs were used to test for signifi-cance of rate changes in single spacers/introns as describedabove; across-species differences were evaluated bysequential Bonferroni-corrected unpaired Wilcoxon tests.Nonparametric Spearman tests were used to test for correla-tions between different genomic traits (GC content, indelfrequency, nucleotide divergence). A Sign test was used toevaluate whether the distribution of rate elevation or relaxa-tion of purifying selection in protein-coding genes match rateelevation in the adjacent ncDNA regions; ndh genes andadjacent ncDNA regions were excluded from the test.

Supplementary MaterialSupplementary figures S1 and S2 and tables S1–S4 are avail-able at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).

Acknowledgments

We thank Andreas Fleischmann (LMU Munich) and the BonnBotanical Gardens for fresh plant material, and Lena Landherrand Paula Ralph (all Penn State University) for excellent tech-nical advice. Thanks are also due to the PIs of the MimulusGenome Sequencing Project for giving us early access to plas-tome data, and Yan Zhang (Penn State University) for theMimulus plastome annotation. We also thank three anony-mous reviewers and the Associate Editor for helpfulcomments and suggestions on an earlier version of this man-uscript. This study was funded by the Germany ScienceFoundation, Deutsche Forschungsgemeinschaft (DFG) grantMU2875/2 to K.F.M.

ReferencesAdamec L. 1997. Mineral nutrition of carnivorous plants: a review. Bot

Rev. 63:273–299.Adamec L. 2009. Photosynthetic CO2 affinity of the aquatic carnivorous

plant Utricularia australis (Lentibulariaceae) and its investment incarnivory. Ecol Res. 24:327–333.

Adamec L. 2011. Ecophysiological look at plant carnivory: why are plantscarnivorous? In: Seckbach J, Dubinski Z, editors. All flesh is grass..Plant–animal interrelationships. Cellular origin, life in extreme hab-itats and astrobiology, Vol. 16. New York: Springer Science +Business Media B. V. p. 455–489.

Adamec L. 2013. Foliar mineral nutrient uptake in carnivorous plants:what do we know and what should we know? Front Plant Sci. 4:10.

Albert VA, Jobson RW, Michael TP, Taylor DJ. 2010. The carnivorousbladderwort (Utricularia, Lentibulariaceae): a system inflates. J ExpBot. 61:5–9.

Albert VA, Williams SE, Chase MW. 1992. Carnivorous plants: phylogenyand structural evolution. Science 257:1491–1495.

Barthlott W, Porembski S, Fischer E, Gemmel B. 1998. First protozoa-trapping plant found. Nature 392:447.

Blankenberg D, Von Kuster G, Coraor N, Ananda G, Lazarus R, ManganM, Nekrutenko A, Taylor J. 2010. Galaxy: a web-based genome anal-ysis tool for experimentalists. Curr Protoc Mol Biol. 89:19.10.1–19.10.21.

Blazier JC, Guisinger-Bellian MM, Jansen RK. 2011. Recent loss of plastid-encoded ndh genes within Erodium (Geraniaceae). Plant Mol Biol. 76:1–10.

Bock R, Timmis JN. 2008. Reconstructing evolution: Gene transfer fromplastids to the nucleus. Bioessays 30:556–566.

Brauckmann T, Kuzmina M, Stefanović S. 2009. Loss of all plas-tid ndh genes in Gnetales and conifers: extent and evolution-ary significance for the seed plant phylogeny. Curr Genet. 55:323–337.

Braukmann T, Stefanović S. 2012. Plastid genome evolution in mycohe-terotrophic Ericaceae. Plant Mol Biol. 79:5–20.

Britten RJ. 1986. Rates of DNA sequence evolution differ betweentaxonomic groups. Science 231:1393–1398.

Cai Z, Guisinger MM, Kim H-G, Ruck E, Blazier J, McMurtry V, Kuehl J,Boore J, Jansen RK. 2008. Extensive reorganization of the plastidgenome of Trifolium subterraneum (Fabaceae) is associated withnumerous repeated sequences and novel DNA insertions. J MolEvol. 67:696–704.

Chevreux B. 2011. Sequence assembly with MIRA3—the definitiveguide. [cited 2012 Nov]. Available from: http://mira-assembler.sour-ceforge.net/docs/DefinitiveGuideToMIRA.pdf.

Chevreux B, Wetter T, Suhai S. 1999. Genome sequence assembly usingtrace signals and additional sequence information. Comput Sci Biol(Proc Germ Conf Bioinforma). 99:45–56.

Chumley TW, Ferraris JD, Mower JP, Fourcade HM, Calie PJ, Boore JL,Jansen RK. 2006. The complete chloroplast genome sequence ofPelargonium x hortorum: organization and evolution of the largestand most highly rearranged chloroplast genome of land plants. MolBiol Evol. 23:2175–2190.

Cosner ME, Raubeson LA, Jansen RK. 2004. Chloroplast DNA rearrange-ments in Campanulaceae: phylogenetic utility of highly rearrangedgenomes. BMC Evol Biol. 4:27.

Darling AE, Mau B, Perna NT. 2010. progressiveMauve: multiple genomealignment with gene gain, loss, and rearrangment. PLoS One 5:e11147.

Delannoy E, Fujii S, des Francs CC, Brundrett M, Small I. 2011. Rampantgene loss in the underground orchid Rhizanthella gardneri highlightsevolutionary constraints on plastid genomes. Mol Biol Evol. 28:2077–2086.

dePamphilis CW, Young ND, Wolfe AD. 1997. Evolution of plastid generps2 in a lineage of hemiparasitic and holoparasitic plants: manylosses of photosynthesis and complex patterns of rate variation. ProcNatl Acad Sci U S A. 94:7367–7372.

Ellison AM. 2006. Nutrient limitation and stoichiometry of carnivorousplants. Plant Biol. 8:740–747.

Ellison AM, Gotelli NJ. 2009. Energetics and the evolution of carnivorousplants—Darwin’s “most wonderful plants in the world”. J Exp Bot.60:19–42.

Fajardo D, Senalik D, Ames M, Zhu H, Steffan SA, Harbut R, Polashock J,Vorsa N, Gillespie E, Kron K, et al. 2012. Complete plastid genomesequence of Vaccinium macrocarpon: structure, gene content, andrearrangements revealed by next generation sequencing. Tree GenetGenomes 9:489–498.

Friedberg EC, Walker GC, Siede W, Wood RD, Schultz RA, Ellenberger T.2006. DNA repair and mutagenesis. 2nd ed. Washington (DC): ASMPress.

Frost SDW, Liu Y, Pond SLK, Chappey C, Wrin T, Petropoulos CJ, Little SJ,Richman DD. 2005. Characterization of human immunodeficiencyvirus type 1 (HIV-1) envelope variation and neutralizing antibodyresponses during transmission of HIV-1 subtype B. J Virol. 79:6523–6527.

543




Dow

nloaded from

-``''Sif informationhttp://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/mst261/-/DC1http://www.mbe.oxfordjournals.org/http://www.mbe.oxfordjournals.org/(multi-sheet excel file) are available online

disproportional plastome-wide increase of substitution...

Documents