Biological Journal of the Linneun Society (ZOOl), 74: 87-111. With 15 figures
doi:l0.1006/bij1.2001.0577, available online a t http;//www.idealibrary.com on
I D E bL8
@
Simultaneous analysis of 16S, 28S, COI and
morphology in the Hymenoptera: Apocrita evolutionary transitions among parasitic
wasps
MARK DOWTON"
Institute of Conservation Biology, Department of Biology, Wollongong University, Wollongong, NSW
2522, Australia; and Centre for Evolutionary Biology and Biodiversity, Department of Applied and
Molecular Ecology, Waite Campus, Adelaide University, Glen Osmond, SA 5064, Australia
ANDREW D. AUSTIN
Centre for Evolutionary Biology and Biodiversity, Department of Applied and Molecular Ecology,
Waite Campus, Adelaide University, Glen Osmond, SA 5064, Australia
Received 21 January 2001; accepted for publication 12 June 2001
Simultaneous analysis of morphological and molecular characters from the 16s rDNA, 28s rDNA and cytochrome
oxidase 1 genes was employed to resolve phylogenetic relationships among the apocritan (Insecta: Hymenoptera:
Apocrita) wasps. Parsimony analyses, employing a broad range of models, consistently recovered the Proctotrupomorpha as a natural group, the Megalyridae and Trigonalidae as sister groups, a clade comprising the
Monomachidae, Diapriidae, and Maamingidae, the Vanhorniidae and Proctotrupidae as sister groups, the Proctotrupoidea as polyphyletic, and the Evaniomorpha as a grade (but including the Ichneumonoidea, Aculeata, and
Stephanidae). The Proctotrupomorpha, containing virtually all of the wholly endoparasitic lineages, was consistently
recovered as an apical clade, with the remaining groups forming a paraphyletic grade below them. Although the
relative placement of the groups forming this basal grade varied among analyses, the most commonly recovered
arrangement is consistent with the ancestral biology being ectoparasitism of coleopteran, wood-boring larvae.
Furthermore, the recovery of the ectoparasitic-containing proctotrupomorphs (Chalcidoidea and, in some analyses,
Ceraphronoidea) as apical lineages argues that these biologies are reversals.
0 2001 The Linnean Society of London
ADDITIONAL KEY WORDS: ectoparasitoid - endoparasitoid - phylogeny - parsimony - mixed-model analysis six-parameter parsimony - host-switching - ancestral biology - saturation analysis.
INTRODUCTION
thousand, although probably less than 25% have so
far been described (LaSalle & Gauld, 1993).
The single origin of parasitism evident in the
Hymenoptera (Rasnitsyn, 1988; Whitfield, 1992;
Vilhelmsen, 1997), together with the large number
of parasitic wasps compared with their non-parasitic
sister group (the sawflies) argues that a massive adaptive radiation occurred with the biological transition
from phytophagy, the ancestral biology of the order,
(Rasnitsyn, 1980; Vilhelmsen, 1997) to parasitism
(Wiegmann, Mitter & Farrell, 1993). This radiation is
Robust phylogenies provide a framework for understanding the evolution of complex biologies. Parasitism
is one such biology that has evolved independently
within a number of insect orders, Of these, the parasitic
Hymenoptera (wasps) comprise the vast majority of
species, numbering in excess of several hundred
~
* Corresponding author. E-mail: [email protected]
00244066/01/090087 + 25 $35.00/0
87
0 2001 The Linnean Society of London
88
1\12. DOWTON and A. D. AUSTIN
also evident in the diversity of lifestyles displayed by
the parasitic wasps: they utilize virtually every insect
order as well as arachnids as hosts; they can be ectoparasitic (feed externally), endoparasitic (feed internally), solitary (one parasitoid per host), gregarious
(many parasitoids per host), hyperparasitic (feed on
other parasitoids), idiobiontic (permanently paralyze
their host thus altering its development), koinobiontic
(allow host development to continue), among other
variants of the parasitic lifestyle.
Although much effort has been directed a t discovering the sequence of evolutionary transitions that
has led t o this diversity of parasitic lifestyles among
the Hymenoptera, there has thus far been little success
a t producing a robust phylogeny for the parasitic wasps
(Rasnitsyn, 1980, 1988; Whitfield, 1992; Dowton &
Austin, 1994; Dowton et al., 1997; Carpenter &
Wheeler, 1999; Ronquist et al., 1999). This contrasts
with the now clear picture that has emerged in recent
years for the relationships among the various sawfly
groups based exclusively on morphological data (e.g.
Vilhelmsen, 1997, 2000). For the apocritan wasps,
morphological and molecular datasets disagree in detail (see Whitfield, 1998, for an overview), with neither
yielding convincing evidence based on the conventional
measures of bootstrap support (Felsenstein, 1985) or
bremer indices (Bremer, 1988).The most likely reasons
for the instability of previous phylogenetic estimates
is either insufficient taxonomic sampling, insufficient
characters, or a combination of these two (Poe & Swofford, 1999). In the present study, we dramatically
increase both the density of taxonomic sampling for
the Apocrita. and the number of characters analysed.
Largely. the present study aims to resolve relationships
among the various parasitic groups, particularly
among the Proctotrupomorpha, Evaniomorpha (sensu
Rasnitsyn, 1988) and several small families. We have
not concentrated on relationships within the Chalcidoidea, Ichneumonidae, or Aculeata (which includes
the predatory wasps, ants and bees) as these groups
are demonstratably monophyletic (e.g. Gibson, 1986;
Sharkey & Wahl, 1992; Brothers & Carpenter, 1993;
Campbell et al., 2000) based on extensive morphological and/or molecular evidence. Rather, we have
sought to determine their likely sister groups. Our
data recover certain, but by no means all, apocritan
relationships across a range of analytical models. As
such, the current study broadly identifies the phylogenetic organization of the Apocrita, and represent our
best current estimate of the relationships of this group.
MATERIAL AND ANALYTICAL STRATEGIES
TAXONOMIC SAMPLING
The taxa sampled are listed in Table 1. The outgroup
comprised three symphytan exemplars from taxa pos-
tulated as closely related t o the Apocrita (Rasnitsyn,
1988; Vilhelmsen, 1997). In order t o represent the
Apocrita more completely compared with our previous
molecular studies (Dowton et al., 19971, taxonomic
sampling was more than doubled [84 apocritan taxa
sampled here, compared with 35 in Dowton et al.,
19971. Further, taxonomic coverage was significantly
improved, particularly within the Proctotrupoidea,
with representatives from all 13 recognized apocritan
superfamilies (and 36 families) included. More importantly, previous studies included some groups containing only apically derived representatives, whereas
care was taken to include, where known, basally derived members in the present study. Examples of this
include the Chrysidoidea from the Aculeata (Brothers
& Carpenter, 1993), and the Mymaridae from the
Chalcidoidea (Gibson, 1986). Nevertheless, future examinations of the Apocrita may benefit from a broader
sampling of aculeates, chalcidoids, and ichneumonids.
SEQUENCEGENERATION
Prior assessments of apocritan phylogeny were based
on a single gene fragment [16S rDNA (16s): Dowton &
Austin, 1994; Dowton et al., 19971. Sampling multiple
genes from distinct subcellular compartments is likely
to improve phylogenetic accuracy (e.g. Wheeler, Cartw i g h t & Hayashi, 1993). In the present study, portions
of the mitochondria1 1 6 s and cytochrome oxidase 1
(COI), and the nuclear 28s rDNA (28s) genes were
amplified as previously described (Dowton & Austin,
1995, 1998). As indicated in Table 1, amplifications
were not successful for some gene fragments in isolated
taxa (e.g. the 16s gene in Stephanidae). These were
coded as missing data, and accounted for c. 1Oo/o of the
total molecular dataset. No taxa were included that
were missing data for more than one gene portion. All
newly generated sequences have been deposited in
GenBank, under accession numbers AF379 857AF380 031.
SEQUENCE ALIGNMENT
There are a range of options for aligning nucleotide
sequences that can be broadly categorized as distancebased, parsimony-based and those utilizing information on secondary structure. Methods that successively
align sequences in an order determined by a n initial
pair-wise similarity estimate [such as Clustal; Thompson, Higgins & Gibson (1994)l were not used because
of the wide range of alignments produced depending
on the gap, gap open and gap extension penalty costs
used, with no obvious justification for choosing one
alignment over another. Parsimony-based alignments
[e.g. MALIGN: Wheeler & Gladstein (1994); POY
Wheeler (1996)] are, in our hands, too processor-intensive to be effective for a dataset of this size. It is difficult
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
89
Table 1. List of taxonomic groups surveyed, a n d the gene portions sequenced
Sub-order
Superfa m i ly
Family
Symphyta
Cephoidea
Cephidae
Siricoidea
Xiphydriidae
Orussoidea
Orussidae
Apocrita
Ceraphronoidea
Megaspilidae
Ceraphronidae
Chalcidoidea
Aphelinidae
Chalcididae
Encyrtidae
Eulophidae
Eupelmidae
Mymaridae
Pteromalidae
Torymidae
Cynipoidea
Cynipidae
Figitidae
Ibaliidae
Euaniidae
Evaniidae
Gasteruptiidae
Ichneumonoidea
Braconidae
Ichneumonidae
Taxon
Library
reference
16s
Hartigia trimaculata (Say)
M30
/
Xiphydria mellzpes (Harris)
X
J
Orussus terminalis (Newman)
M31
VI
Conostigmus sp.
Dendmcerus carpenteri (Curtis)
Aphanogmus sp.
Ceraphmn sp. 1
Ceraphmn sp. 2
M85
M254
M87
M247
M250
J
Encarsia formosa (Gahan)
Aphytis melinus (De Bach)
Brachymeria phya (Walker)
Leptomastix dactylopii (Howard)
Meli to bia austral ica Girault
Eusandalum sp.
Gonatocerus sp.
Trichilogaster sp.
Pteromalus puparum (L.)
Megastigrnus sp.
Encarsia
Aphytis
M47
M84
M159
M220
M239
M45
Pteromalus
M139
J
J
J
J
J
J
Xestophanes sp.
Anacharis zealandica Ashmead
Ibalia leucospoides (Hochenwarth)
M233
M64
M4
X
Evania sp. 1
Evania sp. 2
Euania sp. 3
Gasteruption sp. 1
Gasteruptioin sp. 2
Gasteruption sp. 3
Eufoenus sp.
M2
M253
M242
M19
M248
M113
M21
J
J
Ascogaster sp.
Diospilus sp.
Dolopsidea sp.
Jarra maculipennis Marsh & Austin
Megalohelcon ichneumonoides Tobias
Mirapotes sp.
Neoneurus mantis Shaw
Sigalphus sp.
Toxoneumn abdominalis Cresson
Ichneumon promissorius (Erichson)
Venturia canescens (Gravenhorst)
Xorides praecatorius (F.)
M161
M123
M141
M35
M63
M252
M155
Don.AC
M148
M7
Venturia
Xorides
J
J
28s
COI
J
J
dI
X
X
X
J
J
I
I
4
I
V
J
J
$'
J
J
/
J
I
/
J
J
J
J
J
con tinued
90
~~
M. DOWTON a n d A. D. AUSTIN
Table 1
-
~~
continued
Sub order
Superfn in 1 /I
Family
Taxon
Library
16s
reference
Mtgn lvrot dea
Rilegdl yridae
Plrrt?.gastmi dea
Scelionidae
Platygastridae
Pmclofrupoidea
lliapriidae
Heloridae
bbdmingidae
Monomachidae
Pelecinidae
Proctorupidae
Roproniddae
Vanhorniidea
Stc.phanoidea
Steph anidae
Apo iden
Apidae
Sphecidae
Ch Iyidoidea
Chrysodidae
Tiphiidac,
Ihpoidea
Formicidae
Vespidae
Megalyra sp. 1
Megalyra sp. 2
M69
M157
Baryconus sp.
Ceratobaeus sp.
Scelio fulgidus (Crawford)
Sparasion sp.
7'rimorus sp.
Trzssolcus basalis (Wollaston)
Allotropa sp.
Aphanoinerus sp.
Arnztus sp.
Inostemma sp.
M195
M145
M3
M143
M172
Aclista sp. 1
Diphompria sp. 1
Spilomicrus sp. 1
Spilomicrus sp. 2
Spilomicrus sp. 3
Aclista sp. 2
Basalys sp.
Genus indet. (Betylinae)
Diphoropria sp. 2
Helorus sp. 1
Helorus sp. 2
Maaniinga rangi
Monomachis antipodalis Westwood
Monomachis sp. (Chile)
Pelecinus polyturator (Drury)
Apoglypha sp.
Brachyserphus abruptus (Say)
Exallonyx obsoletus (Say)
Codrus sp.
Disogmus areolator (Haliday)
Phaenoserphus viator (Haliday)
Repmnia garnzani (Ashmeadf
Vanhornia eunemidarum (Crawford)
M227
M65
M66
M231
M232
M226
M67
M228
M229
M88
M93
M70
M241
M223
M22
M92
M94
M95
M89
M97
M96
M18
M12
M e s s c h u s bicolor (Westwood)
Megischus texanus Cresson
Schlettemrius cinctipes (Cresson)
M147
M39
M5
Taeniogonalas gundlachii (Cresson)
Orthogonalys pulchella (Cresson)
Lycogaster sp.
M13
M16
N3
Apis mellifera (L.)
Sceliphron sp.
Apis
Primeuchmeus sp.
Rhagigaster sp.
M48
M222
M y m e c i a forficata (F.)
Wspula germanica (Fabricius)
Myrmecia
M224
X
i
Trissolcus
M245
M246
M142
M244
x
X
i
X
i
X
X
X
\
M40
,
X
28s
COI
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
t o be confident of finding the shortest trees with a n
aligned dataset comprising more than 20 taxa (e.g.
Swoffordet al., 1996);it must therefore be more difficult
to find the shortest tree when the length and the
relative alignment of sequences can be varied during
tree-searching. Secondary-structure-based alignments
are possible for 16S, and we have used these in previous
studies [e.g. Dowton & Austin (1994)l. Such models
identify stem and loop regions, permitting the alignment of putatively homologous structures. However,
in our experience, the stem regions can be trivially
aligned by eye due to conserved primary sequence
homology, while the secondary-structure model cannot
help with the alignment of the length-variable loops.
No generally applicable model for the 28s gene is
available, and, in our experience (and that of others,
e.g. Belshaw & Quicke, 1997; Campbell et al., 2000),
the hymenopteran 28s secondary structure is too variable for this to be an effective strategy. For these
reasons, we have chosen a conservative approach to
sequence alignment, limiting the dataset to just those
portions of the genes for which alignment is straightforward; i.e. removing length-variable regions (as described by Swofford et al., 1996). Although this will
undoubtedly remove some information, it focuses the
analyses on confident statements of character homology. We would rather sequence additional genes to
increase the informational content of our dataset than
base our analyses on questionable areas of alignment.
The sequence alignments can be downloaded from
http j//www.waite. adelaide .edu.aufinsects.
PHYLOGENETIC ANALYSIS
Similarly there are a range of available options for
phylogenetic analysis. Distance-based methods, besides producing phenetic estimates of evolutionary
relationship, perform poorly (e.g. Hillis, Huelsenbeck
& Cunningham, 1994). Maximum parsimony and maximum likelihood are both philosophically justifiable,
and either may be more or less appropriate for a
particular dataset. A reasonable strategy is to analyse
the data employing both methodologies, considering
those relationships that are consistently resolved as
robust. However, there are practical limitations when
using maximum likelihood. Besides being computationally intensive (particularly for large datasets),
multiple data partitions must be analysed under a
single model of nucleotide evolution. We have documented the extreme composition and rate biases in the
mitochondria1 genes of certain hymenopteran groups
(Dowton & Austin, l995,1997a,b), properties not evident in hymenopteran nuclear genes (e.g. Dowton &
Austin, 1998; Belshaw et al., 2000). Thus, genes from
these two subcellular compartments may demand distinct models of analysis. Further, maximum likelihood
91
Table 2. Substitutional transformation costs for each of
the molecular data partitions
~~
Data
partition
28s
16s
COI-1
COI-2
Transitions
~~
Transversions
AG
CT
AT
AC
CG
GT
1
1
2
2
1
1
2
1
1
2
2
2
2
2
3
3
2
2
4
2
3
3
1
2
cannot yet be used to analyse combined molecular/
morphological datasets, which is the ultimate aim of
this study.
Maximum parsimony was performed using PAUP v.
4.0b4a (Swofford, 1999). Three symphytan representatives (Hartigia, Xiphydria, Orussus) were specified as outgroup taxa. The shortest trees were found
by heuristic search, adding taxa randomly (100 replicates), holding a maximum of 5 trees in memory (i.e.
addseq =random nreps = 100 nchuck = 5 chuckscore =
1).Preliminary analyses, with 200 random replicates,
indicated that this was generally a sufficiently intensive search strategy. Nevertheless, it remains possible
that some searches did not discover all shortest trees.
Due to the molecular evolutionary idiosyncracies
referred to above, we also analysed the molecular data
using six-parameter parsimony [initially described by
Williams & Fitch (1990), and adapted by Cunningham
(1997) and Stanger-Hall & Cunningham (1998)],with
different models defined for each data partition. Sixparameter parsimony applies different costs t o the six
different substitutionclasses (A+&, C-T, A-T, A-C,
CHG, G-T) according to their inferred frequencies,
which were estimated as described by Stanger-Hall &
Cunningham (1998). Briefly, the shortest tree was
found by unweighted parsimony, and used to estimate
the frequency of substitutions, based on maximum
likelihood analysis using the general time-reversible
model, as this considers both unequal base composition
and multiple substitutions. The values estimated for
each molecular partition are presented in Table 2. It
was not possible to calculate paramaters for the COI
3rd codon position (COI-3) partition due to the extremely high number of inferred AT-transversions.
In order to guard against arriving a t phylogenetic
conclusions overly reliant on such ‘parameter-rich’
models, we also performed analyses using a range of
‘simpler’ models (such as unweighted parsimony) in
order to discover those relationships robust to the
model of analysis, The range of models assessed is
described in Table 3. Thus, we assessed the robustness
of the results by identifying those relationships that
were resilient t o changes in the analytical model (i.e.
92
M. DOWTON and A. D. AUSTIN
Table 3. Description of analytical models used for the molecular data partitions
Model pseudonym
Model description
molL
molU ex. (201-3
molW
All characters included, and treated as unordered
COI-3 excluded, all remaining treated as unordered
COI-3 excluded, individual six-parameter step matrices applied to each remaining partition
As for GP, except that character weights varied to reduce bipartition incongruence (character weights
used were 28s: 16s:COI-1: COI-2=2: 1: 1: 1)
All morphological characters treated as unordered
Certain morphological characters treated as ordered, as described in Ronquist et al. (1999)
mol6Pwts
sensitivity analysis, sensu Wheeler, 1995). We did not
assess support using the bootstrap procedure (Felsenstein, 1985),as support declines with increased taxonomic sampling (Sanderson & Wojciechowski, 2000),
particularly for large datasets such as the one analysed
here. We similarly did not assess support using the
bremer index (Bremer, 1988), as these are not comparable between datasets analysed under different
assumptions, becoming inflated in analyses employing,
for example, six-parameter parsimony and/or ordered
morphological characters.
ASSESSMENT OF PAWTITION HETEROGENEITY
There is controversy surrounding the combination of
data partitions, particularly when judged incongruent.
In our experience, nuclear and mitochondrial hymenopteran gene phylogenies are generally judged
incongruent (e.g. see Belshaw et al., 2000), but this
may be due to the different mutational biases operating
in these two subcellular compartments (see above,
under ‘Phylogenetic analysis’). Furthermore, such
heterogenous datasets are precisely the ones that,
when combined, are most likely to recover phylogeny
accurately-the phylogenetic patterns to emerge from
such analyses will reflect their common history, rather
than the selective pressures (e.g. towards a high ATcontent in mitochondrial genes) that have shaped the
evolution of characters from a single subcellular compartment (Wheeler et al., 1993). Our preference is that
data partitions should always be combined, unless
there is compelling evidence that the datasets have
distinct histories. Instead, we used the results of partition heterogeneity tests to refine our six-parameter
parsimony models. Generally, partition heterogeneity
was assessed using the Incongruence Length Difference test [ILD; Farris et al. (1994)l as invoked using
PAUP v. 4.Ob4a (Swofford, 1999). For these tests, the
search settings were; addseq =random nreps = 5
nchuck = 2 chuckscore = 1.
MORPHOLOGICAL DATAMATRIX
We used the recently published morphological datamatrix generated by Ronquist et al. (1999), available
electronically a t httpj//www.zoologi.uu.se/systzoo/staff/
ronquist/Hymenoptera.txt.Characters are coded at the
family level, with polymorphic families coded as the
most likely ground-plan state where possible (see Konquist et al., 1999, for a discussion). We did not include
a hypothetical ancestor, contra Ronquist et al. (1999),
because our study focuses on apocritan relationships,
whereas these authors employed a hypothetical ancestor to root the tree for the whole order. Instead, we
used the morphological coding for the three outgroup
symphytan families (Cephidae, Xiphydriidae and
Orussidae) for which we had molecular data to root
the apocritan analyses. To facilitate combination of
these two datasets, we removed families from the
morphological matrix for which we did not have molecular data, and duplicated families where we had
multiple representatives (i.e. terminal taxa from the
same family were coded identically). Lifestyle transitions were mapped onto apocritan cladograms using
MacClade v. 3.06 (Maddison & Maddison, 1996).
ON THE EXCLUSION OF COI-3 DATA
Among taxa separated by long periods of evolutionary
history, molecular positions that are relatively free
from selective constraints are likely to have suffered
multiple, hidden substitutions. There is controversy
concerning whether such positions should be included
in phylogenetic analyses. Some argue that such positions provide information at apical nodes, and random noise a t deeper levels (e.g. Kallersjo, Albert &
Farris, 1999), whereas others claim that they provide
misleading signal due to non-random noise, grouping
together taxa with, for example, convergently shared
compositional bias (e.g. Huelsenbeck & Nielsen, 1999).
In our opinion, such ‘noisy’characters tend to be excluded a priori from morphological datasets, because
only characters perceived t o be of low variability among
the groups t o be studied are chosen. Similar exclusion
of highly variable molecular characters is necessarily
more overt, because they are collected along with the
less variable characters when gene fragments are
sequenced.
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
0.8 -
93
16s
0.6 -
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
1 ,
0.8
1
/I
COI-3
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
Figure 1. Saturation analysis of molecular character partitions. Pairwise uncorrected and corrected distances were
estimated using PAUP v. 4.0b4a (Swofford, 1999), using the HKY85 (Hasegawa, Kishino & Yano, 1985)model to correct
for multiple substitutions. Uncorrected distances were then plotted on the X-axis, corrected distances on the Y-axis.
The extent of hidden substitutions is indicated by the distance of the points to the right of the diagonal line.
There are accepted methods for identifying molecular character sets that have suffered multiple,
hidden substitutions. Corrected and uncorrected pairwise sequence differences are estimated; character sets
that differ significantly in these two measures have
presumably undergone sufficient hidden substitutions
for their phylogenetic utility to become questionable.
Such a ‘saturation’ analysis is presented in Figure 1.
In line with expectation, the character set most free
to vary (COI-3) is the most saturated, with corrected
pair-wise differences departing markedly from uncorrected estimates. For this reason, we investigated
the effect of excluding this data partition in some
analyses, as part of our investigation into the robustness of the inferred phylogeny.
RESULTS AND DISCUSSION
ANALYSIS OF MOLECULAR PAWITION HETEROGENEITY
We recognized five molecular partitions: 16S, 28S, and
COI 1st and 2nd codon positions (COI-1 and COI-2)
and COI-3. Table 4 lists the bipartition heterogeneity
(ILD) scores for all pairwise comparisons, excluding
those involving COI-3,both before and after generation
of substitutional cost-matrices as outlined by StangerHall & Cunningham (1998), wherein separate costs
are given to each of the six possible substitutions based
on the log of their inferred frequency. Low ILD values
(suggesting incongruence) were evident between many
of the partitions when analysed under equally
weighted parsimony (Table 4, column 1). The effect
94
M.DOWTON and A. D. AUSTIN
Table 1. Results of bipartition heterogeneity (ILD) tests
h e t w w n a1I molccul ar partitions, under different models
of evolution Model abbreviations are a s described in
‘rabl c 3
_______
_____
L ex COI 3
6P
6P wts
0 01
0 01
0 05
0 02
0 85
10
006
001
001
001
005
097
098
005
085
NA
NA
NA
~
28s 165
28s COI 1
28s COI 2
16s COI 1
1 % C(JI 2
(’01-1 C‘OI-2
was greatest between nuclear and mitochondria1 data
partitions, but was also evident among certain mitochondrial data partitions (e.g. 16sand COI-1). Analysis
of congruence after generation of substitutional costmatrices produced mixed results (Table 4, column 2)
- although I t improved congruence between 28s and
16s.congruence was decreased between 28s and COI2 and was unchanged between 285 and COI-1. We
then investigated the impact of weighting the various
data partit ions fractionally (retaining the substitutional cost-matrices), and will report on these
tests more completely elsewhere (Dowton & Austin,
submit1t.d). Briefly, reducing the weight of the mitochondria] partitions markedly increased congruence
(Table 4 $column 3).
PHYI,C)GEIVIYI‘IC ANALYSIS OF THE MOLECULAR DATA
The molecular data were analysed under a range of
models. which are described in Table 3. There are a
range of evolutionary models that can be applied t o
these data, and the choice of those presented here was
made to reasonably represent the breadth of possibilities - from the simplest (all character changes
weighted equally. or ‘unordered’) t o the most parameter-rich (with separate six-parameterstep-matrices
applied to each molecular partition). Strict consenses
of the most parsimonious trees found from each of the
analyses described in Table 3 are presented in Figures
2-5. Below, we briefly discuss the robustness of the
molecular analyses t o changes in the model. A more
complete discussion is presented after inclusion of the
morphological data.
The most robust higher level relationship resolved
by analysis of the molecular data was Rasnitsyn’s
(1988)I’roctotrupomorpha, including the Chalcidoidea,
Cynipoidea, Platygastroidea and Proctotrupoidea.
Nevertheless, the Heloridae (Proctotrupoidea) were
only recovered inside the Proctotrupomorpha in analyses employing six-parameter parsimony (6P) (Figs 4
and 5). In unordered analyses, they generally fell out
as the sister group to the Evaniidae (Figs 2 and 3).
Within the Proctotrupomorpha, the Chalcidoidea and
Platygastroidea were generally recovered as sister
groups (Figs 3-5); the Vanhorniidae were always recovered as the sister group to the Proctotrupiclae;
and the Monomachidae, Diapriidae, and Maamingidae
were always recovered as a clade. The placement of
the Cynipoidea was the most variable, although they
were always recovered as part of the Proctotrupomorpha. This contrasts with our previous molecular
analysis, which recovered the Cynopoidea outside the
Proctotrupomorpha (Dowton et al., 1997), but is curiously consistent with an earlier analysis employing
fewer taxa (Dowton & Austin, 1994).
The Proctotrupoidea (included families listed in
Table I), were not recovered as a monophyletic group
in any molecular analysis. Ignoring the misplacement
of the Heloridae, they were generally recovered as
a grade, but then disrupted by the inclusion of the
Cynipoidea (Figs 3,4). The only analysis in which they
were recovered as a grade excluding the Cynipoidea,
and including the Heloridae, was in the 6Pwts analysis
(Fig. 5).
The Ichneumonidae and Braconidae were not always
recovered as sister groups, and only strictly so in
the 6Pwts analysis (Fig. 5). Analysis of single gene
fragments indicated that 28s data generally supported
this affiliation, whereas 16s did not (data not shown).
We attribute this to the very different AT-content of
ichneumonid and braconid mitochondrial genes. Braconid mitochrondrial genes have the highest AT-content
of all Apocrita, and ichneumonids the lowest. In addition, the Aculeata were not recovered close to either
the braconids o r the ichneumonids, contra Rasnitsyn
(1988) and our previous molecular analyses (Dowton
& Austin, 1994; Dowton et al., 1997). Generally, the
Aculeata were recovered among a clade with the Stephanidae, Megalyridae and Trigonalidae [Figs 3-0; i.e.
a subset of Rasnitsyn’s ( I 988) Evaniomorpha].
Rasnitsyn’s (1988) Evaniomorpha were not recovered as a clade, contra our prior molecular analysis
(Dowton et al., 1997). At best, they were recovered as
a grade in the 6P:wts analysis (Fig. 5), but generally
were disrupted by the presence of the Braconidae and/
or the Ichneumonidae among them. Nevertheless, the
Ceraphronoidea were always recovered outside of the
Proctotrupomorpha, in agreement with our previous
analyses, and contra Ronquist et al. (1999). Within the
‘evaniomorph’ families, the Megalyridae were always
recovered as the sister group to the Trigonalidae. Further, the Stephanidae were generally closely affiliated
with the Megalyridae and Trigonalidae, but never as
the sister to all remaining Apocrita as postulated by
studies post-Rasnitsyn (1988) (see Whitfield 1998, for
review) (Figs 3-5).
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
Hartigia
Xiphydria
Conostigmus
Dendrocerus
Aphanogmus
I-
Schlettererius
Mepischus b.
IMegischus t.
J Helorus sp. 1
Helorus sp. 2
Evania sp. 2
Evania sp. 1
Evania sp. 3
Orussus
Proctotrupomorpha
ex. Heloridae
1
Symphyta
1
Ceraphronoidea
]
Stephanidae
3
Heloridae
]
Evaniidae
95
*
Venturia
Ichneumon
Xorides
Megalohelcon
Dolopsidea
Jarra
Neoneurus
Diospilus
Ascogaster
Toxoneuron
Mirapotes
Sigalphus
Megulyra sp. 1
Megalyra sp. 2
Orthogonalys
Taeniogonalas
Lycogaster
7Eufoenus
]
‘tGasterLption 6 . 2
Rhagigaster
Primeuchroeus
Apis
Sceliphron
Myrmecia
Vespula
I
Vanhornia
Disogmus
Brachyserphus
Apoglypha
Exallonyx
Codrus
Phaenoserphus
Ropronia
Pelecinus
Anacharis
Ibalia
Xestophanes
Sparasion
Aphanomerus
Allotropa
Inostemma
Amitus
Scelio
Trimorus
Baryconus
Ceratobaeus
Trissolcus
Megastigmus
Trichilogaster
Eusandalum
Melittobia
Pteromalus
Leptomastix
Encarsia
Brachymeria
Aphytis
Gonatocerus
Monomachis
Monomachis a.
Diphoropria sp. 1
Diphoropria sp. 2
Betylinae M228
Aclista sp. 1
Aclista sp. 2
Maaminga
Basalys
Spilomicrus sp. 2
Spilomicrus sp. 1
Spilomicrus sp. 3
J
Ichneumonidae
Braconidae
J
) Megalyridae
]
Trigonalidae
Aculeata
- Vanhorniidae
Proctotrupidae
I
Cynipoidea
Platygastroidea
1
Chalcidoidea
J
3
Monomachidae
I
Diapriidae (pt)
-
Maamingidae
Diapriidae (pt)
Figure 2. Maximum parsimony analysis of molecular data, all character transformations weighted equally (molU
model). Strict consensus of 10 shortest trees, length 7820. Asterisk indicates misplaced outgroup taxon. Although the
three symphytans were specified as outgroup taxa, PAUP* does not enforce this when shorter trees are found. In this
case, the shortest tree did not recover the ingroup (the Apocrita) as monophyletic, instead Orussus fell within the
Ichneumonoidea.
96
M. DOWTON and A. D. AUSTIN
~
Ichneumon
Xorides
Rhagigaster
Sceliphron
Apis
Myrmecia
Vespula
Primeuchroeus
Megischus h.
Sch lettereriu
Megischus t.
J
Symphyta
]
Ichneumonidae
11
Aculeata
i
Stephanidae
3 Megalyridae
]
Trigonalidae
3
Heloridae
]
Evaniidae
3
Ceraphronoidea
1
Gasteruptiidae
J
Jo&o
Neoneurris
Diospilus
Mirapotes
Toxoiisuron
Ascogaster
Sigalnhus
Proctotrupomorpha
ex. Heloridae
Anacharis
Ibalin
Xestophanes
Vanhornia
Disognius
Apoglypha
Brachyserphus
Phaenoserph us
Codriis
Exallonyx
Manminga
Monomach is
Monomachis a.
Diphoropria 5p 1
Diphoropraa sp 2
Betvhnae M228
A c l h sp. 1
Aclistn sp. 2
Basnlys
Spilomicrus sp. 2
Spilomicrus sp. 1
Spilomicrus sp. 3
Encarsia
Braehymeria
Trichilogosfer
Ap hy t is
Gonatocerus
Leptomastix
Eusandalurn
Megastigmus
Melittobia
Pteronzalus
A p ha name rii s
Allotropa
Inostentma
Sparasion
Aniitus
Dimorus
Scelio
Baryconus
Ceratohneus
D1ssolt.us
Braconidae
3
Cynipoidea
-
Vanhorniidae
I
Proctotrupidae
- Maamingidae
3 Monomachidae
1
1
Diapriidae
J
I
1
Chalcidoidea
J
Figure 3. M a x i m u m p a r s i m o n y a n a l y s i s of m o l e c u l a r d a t a , COI-3 c h a r a c t e r s excluded (molU ex. COI-3 model). Strict
c o n s e n s u s of 15 s h o r t e s t t r e e s , length 5629.
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
Orussus
Hartigio
Xiphydria
Venturia
Ichneumon
Xorides
Sceliphron
Primeuchroeus
Myrmecia
Apis
Schlettererius
Megischus b.
Megischus t.
__
3
Symphyta
]
Ichneumonidae
1
Aculeata
97
J
]
Stephanidae
3
Megalyridae
]
Trigonalidae
1I
Braconidae
J
Gasteruptiidae
]
Evaniidae
1
Ceraphronoidea
J
Maaminga
Monomachis
Monomachis a.
Basalys
Spilomicrus sp. 2
Spilomicrus sp. 1
Spilomicrus sp. 3
Diphoroprsa sp. 1
Diphoropria sp. 2
Betylinae M228
Aclista sp. 1
Aclista sp. 2
r Anacharis
Xestophanes
Ibalia
Helorus sp. 1
Helorus sp. 2
Ropronia
Pelecinus
Vanhornia
Disogmus
Brachyserphus
Aposhha
Codrus
Exallonyx
Phaenoserphus
[ Melittobia
Pteromalus
~
LJF-6
Proctotrupomorpha
- Maamingidae
3 Monomachidae
1
1
J
3
Diapriidae
Cynipoidea
Heloridae
- Vanhorniidae
Proctotrupidae
J
1
Chalcidoidea
J
Sparasion
Allotropa
Inostemma
Aphanomerus
Amitus
Scelro
Trimorus
Baryconus
Ceratobaeus
Trissolcus
1
Platygastroidea
Figure 4. Maximum parsimony analysis of molecular data using six-parameter parsimony (6P model). Strict consensus
of 4 shortest trees, length 7804.
98 M. DOWTON a n d A. D. AUSTIN
z
I
I
/
Ichneumonoidea
1
Hartigia
Xzphydria
Euania sp. 2
Euania sp. 1
Euaraia sp. 3
Eufoenus
Gasteruption sp. 3
Gasteruption sp. 2
Gasteruption sp. 1
Megalyra sp. 1
Megalyra sp. 2
Orthogonalys
r Taeriioponalns
7 LycogGter
Rhagigaster
Vespula
r Myrniecia
7r Aois
LL S'celiphron
Megischus 1.
Meeischus h.
-L SchY2ettereriu.s
Primeuchroeus
Conostzgmus
Dendrocerus
Cernphron sp. 1
Aphanogmus
Ceraphron sp. 2
Xorides
Venturia
1
Evaniidae
1
Gasteruptiidae
3
Megalyridae
i
I
]
1
]
Trigonalidae
Aculeata
Stephanidae
Ceraphronoidea
Ichneumonidae
Ja rra
L
Proctotrupomorpha
Neoneurus
Sigalphus
Diospilus
Ascogaster
Mirapotes
Toxoneuron
Maaminga
Monomachis
Monomachis a.
Basalys
Diphoropria sp. 1
Diphoropria sp. 2
Spilomicrus sp. 3
Spilomicrus sp. 1
Spilomicrus sp. 2
Betylinae M228
Aclista sp. 1
Aclista sp. 2
Helorus sp. 1
Helorus sp. 2
Ropronia
PeEecinus
Vanhornia
Disogmus
Brachyserphus
Apoglypha
Codrus
Exallonyx
Phaenoserphus
Xestophanes
Anacharis
Ibalia
Leptomastix
Trichilogaster
Pteromolus
Encarsin
Eusandalum
Megastignius
Melittohia
Brachymeria
Aphytis
Gonatocerus
sparn.sion
Allotrupa
Inustem nia
Aphanomerus
Aniitus
Scelio
Trimorus
Baryonus
Ceratobaeus
TkIssolcus
Braconidae
-
Maamingidae
3 Monomachidae
1
Diapriidae
J
3
Heloridae
-
Vanhorniidae
I
]
Procto trupidae
Cynipoidea
1
i
Chalcidoidea
J
Platygastroidea
Figure 5. Maximum parsimony analysis of molecular d a t a using six-parameter parsimony, with mitochondria1
partitions downweighted (6Pwts model). Strict consensus of 2 shortest trees, length 5822. a s t e r i s k indicates misplaced
taxon.
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
SIMULTANEOUS ANALYSIS OF MOLECULAR AND
MORPHOLOGICAL DATA
Ronquist et al. (1999) presented analyses of a large
suite of morphological characters [i.e. a corrected version of Rasnitsyn’s (1988 data)] applied to resolve
hymenopteran relationships. In their study, some characters are treated as ordered. Although they argued
that they investigated “common assumptions of the
way that morphological characters evolve”, they did
not present the range of analyses initially suggested,
including those with losses treated as irreversible. In
the present study we investigated just two alternative
models: treating all characters as unordered, and treating those characters as ordered that were so indicated
in their character list. Figures 6-9 are analyses in
which all morphological characters are treated as unordered, with molecular data treated under the range
of models outlined in Table 3, while Figures 10-13 are
analyses in which some morphological characters are
treated as ordered, with molecular data again treated
under the range of models outlined in Table 3. Below,
we summarize the relationships recovered under this
range of analyses.
The Proctotrupomorpha sensu Rasnitsyn (1988)
were generally recovered as a clade (Figs 7-8, 10-12),
except in those analyses in which the mitochondrial
partitions were downweighted (Figs 9, 13). In all analyses, the Heloridae were recovered inside the Proctotrupomorpha, in contrast to the molecular analyses.
The Ceraphronoidea were only recovered inside the
F’roctotrupomorpha (i.e. as resolved by Ronquist et al.
1999) in those analyses in which the mitochondrial
partitions were downweighted. Thus, the position of
the Ceraphronoidea was sensitive t o the model of the
analysis. In contrast to this, the clade comprising
the Diapriidae, Monomachidae and Maamingidae were
always recovered regardless of the model employed
(Figs 6-13), as were the Vanhorniidae +Roctotrupidae. We previously reported a close relationship
between the Vanhorniidae and Proctotrupidae (Dowton
et al., 1997), with the Vanhorniidae inside the Proctotrupidae. The more extensive character sampling presented here suggests that these two families should
remain distinct. The resolution of the sister group
relationship between the Chalcidoidea and Platygastroidea was robust to the addition of morphological
characters in some analyses (Figs 7-8, 11-12), generally when the COI-3 partition was excluded (Figs 7,
11) or when six-parameter parsimony was used (Figs
8, 12). In the remaining analyses, this sister group
relationship was disrupted by either the Cynipoidea
(Figs 6 , 10, 13) or the Ceraphronoidea (Fig. 9). As in
the molecular analysis, the Proctotrupoidea were not
recovered as a natural group in any of the combined
analyses.
99
In contrast to the molecular analyses, the Ichneumonoidea were always recovered as a natural group
(Figs 6-13) in combined analyses, although their relative placement to other Apocrita varied. The Aculeata
were never recovered as the sister group t o the Ichneumonoidea, contra Rasnitsyn (1988) and our previous molecular analyses (Dowton & Austin, 1994;
Dowton et al., 1997). Instead, the aculeates were generally recovered as a relatively basal apocritan lineage
(e.g. Figs 7-9, 11-13). Intriguingly, the recent analysis
of extant morphological data similarly recovered the
aculeates basally [fig. 7 in Ronquist et al. (1999)]. We
do not consider that the placement of the aculetes
in the present study is due to the sole influence of
morphological characters, a s molecular characters also
placed them relatively basally (Figs 2 4 ) .
With respect to the families comprising the Evaniomorpha sensu Rasnitsyn (1988), the Megalyridae
were generally recovered as the sister group t o the
Trigonalidae (Figs 7-10, 12). The relative placement
of the Stephanidae was sensitive to the model of analysis but were generally recovered close to the Megalyridae and Trigonalidae. The Evaniidae + Gasteruptiidae (i.e. Evanioidea) were rarely recovered as a
natural group (Figs 7, 9, 13).
BIOLOGICAL TRANSITIONS AMONG THE APOCRITA
The strategy of assessing phylogeny across a range of
analytical models makes mapping biological transitions difficult, particularly because many relationships are sensitive to the model of analysis. In
particular, the lack of robustness of the basal lineages
of the Apocrita make inferring the groundplan biology
problematic. Nevertheless, certain higher level consistencies allow us to make some broad conclusions. A
consistent aspect of all analyses was the resolution of
the Proctotrupomorpha as a natural, derived group,
leaving aside for the moment the questionable placement of the Ceraphronoidea. Further, the remaining
(i.e. non-proctotrupomorph) superfamilies were generally recovered as a grade below them (see Fig. 3 for
the only exception), rather than as a sister clade. This
suggests that certain of Rasnitsyn’s Evaniomorpha,
the Ichneumonoidea and the Aculeata form a basal
grade in the Apocrita. Of the combined analyses, the
most consistent basal lineages are the Stephanidae
(often affiliated with the Megalyridae + Trigonalidae),
Ichneumonoidea and the Aculeata (e.g. Figs 7, 10, 12).
Although intuitively little could be concluded from
such a broad assemblage, it has been pointed out
that the basal lineages of each of these groups are
ectoparasitic (Fig. 14) on wood-boring coleopteran
larvae (Whitfield, 1992; Fig. 15). Specifically, the
Stephanidae, Ichneumonidae and Scolebythidae (Aculeata) contain members that exhibit this biology. The
--
100 M. DOWTON and A. D. AUSTIN
,___.
i
_ _ ~ ~
~ _ _
-
Orussus
Hartigia
Xiphydrio
[ Megalyra sp. 1
Megalyra SP. 2
Megischus b.
Megischus t.
Schlettererius
Orthogonalys
Taeniogonalos
Lycogaster
Eufoenus
'
Ichneumonoidea
Gasteruptzon sp. 2
Myrmeein
Vespula
Rhagigaster
Primeuchroeus
Apis
Sceliphron
Ichneumon
Xorides
Venturia
Megalohelcon
Dolopsidea
Jorra
Neoneurus
Diospzlus
Ascogaster
Toxoneuron
Mirapotes
Sigalphus
Ropronia
Pelecinus
Vanhornia
Disogm us
Brachyserph us
Apoglyp ha
Exallonyx
Codrus
Phamoserph u s
Sparasion
Aphanom.erus
Amitus
Allotropa
Inostemma
Scelio
Trimorus
Baryconus
Trissolcus
Cerotobaeus
Helorus sp. 1
Helorus sp. 2
Euania sp. 2
Euania sp. 1
Evania sp. 3
Conostigni us
Dendrocerus
Ceraphron sp. 1
A p h anogmus
Ceraphron sp. 2
Monomachis
Monomachis a.
Diphoroprio sp. 1
Dzphoropria sp. 2
Betylinae M228
Aclzsta sp. 1
Aclista sp. 2
Maominga
Basa1,ys
Spzlomicrus sp. 2
Spilornicrus sp. 1
Spilomicms sp. 3
Anacharis
Ibalsa
Xestophanes
Aphytis
Gonotorerus
Brarhymeria
Megastzgmus
Melittobio
Pteronialus
Leptomastix
Encorsia
Eusondaluni
Trichilogaster
Symphyta
1
1
]
Megalyridae
Stephanidae
Trigonalidae
1
~
J
]
Aculeata
Ichneumonidae
1
1
Braconidae
'
I
I
s
G
-
Vanhorniidae
Proctotrupidae
J
7
J
3
1
]
Heloridae
Evaniidae
1
J
Ceraphronoidea
3
Monomachidae
1
Diapriidae ( p t )
-
1]
~
Maamingidae
Diapriidae ( p t )
Cynipoidea
Chalcidoidea
J
Figure 6. Simultaneous analysis of molecular and unordered morphological data (mol U, morphU model). Strict
consensus of 12 shortest trees, length 8313.
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
Migikhus b.
Schlettererius
Megischus t.
Megalyra sp. 1
Megalyra sp. 2
Orthogonalys
Taeniogonalas
Lycogaster
Rhagigaster
Primeuchroeus
Vespula
Myrmecia
Ichneumon
_i Xorides
Venturia
Megalohelcon
Dolopsidea
Jarra
Neoneurus
Diospilus
Mirapotes
Toxoneuron
Ascogaster
Sigalphus
Conostigmus
Dendrocerus
Ceraphron sp. 1
Aphanogmus
Ceraphron sp. 2
Evania sp. 2
Euania sp. 1
Euania sp. 3
]
Stephanidae
3
Megalyridae
]
Trigonalidae
1
]
101
Aculeata
Ichneumonidae
'I-
Ichneumonoidea
I
Ropronia
Helorus sp. 1
Helorus sp. 2
Anacharis
Ibalia
Xestophanes
I-
Proctotrupomorpha
~
-
Pelecirzus
Vanhornia
Disogmus
Apogbpha
Exallonyx
Brachyserphus
Codrus
Phaenoserphus
Maaminga
Monomachis
Monomachis a.
Diphoropria sp. 1
Dzphoropria sp. 2
Betylinae M228
Aclista sp. 1
Aclista sp. 2
Basalys
Spilonzicrus sp. 2
Spilomicrus sp. 1
Spilonzicrus sp. 3
4
4
11
Braconidae
3
Evaniidae
3
Gasteruptiidae
3
Heloridae
]
Cynipoidea
-
Vanhorniidae
1
Proctotrupidae
-
Maamingidae
I
Diapriidae
3 Monomachidae
i
1
I
Sparasion
Scelio
Trimorus
Baryconus
nissolcus
Ceratobaeus
Ceraphronoidea
1
I
Chalcidoidea
Platygastroidea
Figure 7. Simultaneous analysis of molecular and unordered morphological data (mol U ex. COI-3, morphU model).
Strict consensus of 27 shortest trees, length 6113.
102
M. DOWTON and A. D. AUSTIN
L
]
LI
.
L
I
I
Orussus
Hartigia
Xtphydria
PriinPirchrows
Rhngigaster
Vespula
Myrm w i n
Apis
Sceliphron
Megnlvra SQ. 1
Megalyrn sp. 2
Lyrogcisfer
Orth ogoiia l y s
Tnrniogonnlna
Schletterwius
M~grsr1iu.sh.
Meggischus t.
Euania SQ. 2
Euania sp. 1
Eoariia SQ. 3
Eufornus
Gnstvruption s p 3
Gasteruptiori sp 1
Gastcruptron sp. 2
Conostigrnus
Denn'rowrus
Cer-ophron sp. 1
Aplinnogni u s
Ceraphron S Q . 2
Vuirturici
r
Ichneumonoidea
iI
I
I
1
Trigonalidae
1
Stephanidae
1
Evaniidae
1
i
I
1
U
~
1
~
i
Proctotrupomorpha
Gasteruptiidae
:
Ceraphronoided
]
Ichneumonidacl
Jarrn
DoZopsidi~a
1
I
Aculeata
) Megalyridae
Neonertriia
~
Symphctn
c
Diospr (us
Mirnpofes
To.miiruron
Ascogaster
Signlphus
Man iniriga
Mononinehis
il.fonomnchis a.
Basn1,vs
Spilolnicrus sp. 2
Spilomicrus sp. 1
Spi1omicru.s sp. 3
Diphoroprrci sp. 1
Diphoroprio sp. 2
Betylinae M22h
Aclis/u sp. 1
Ac'lista sp. 2
Anacharis
Ihalin
Xestoph n I 1 1's
Hclorus sp. 1
Helorus S Q . 2
Ropron ia
Pelecrn it s
Vanhoriirn
Dt.srigniu,v
Brac'hysc~rph11s
Apoglyplr a
Braconidae
i
-
1
Maamingidae
Monomachidar
1
1
-
1
Diapriidae
Vanhorniidae
Proctotrupidae
i
Meliftohia
P fcrorn a111.5
Eusnndnlur~i
Lupton1astr.x
Eri c'a rsin
Megnstrgrnli,\
Brcrrhyiner~a
Trichilr~grr,ster
Aplphytrs
Gori otowrus
spnra.sro1i
Allotroprr
Iiiostein nin
Aphnn~in7rr-us
Aniitus
s~elio
TrlI?loru.s
Bar-~co~i
11s
Cerntobotws
Trr s.solr 1I.S
1
Ii
Chalcidoidea
I
J
7
I
1
i
Platygastroidea
Figure 8. Simultaneous analysis of m o l e c u l a r and unordered morphological data (molGP, morphU model). Strict
conscnsus of 2 shortest trees, length 8293.
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
Orussus
Hartzgia
Xiphydria
Rhugigaster
Vespula
Myrmecia
Primeuchroeus
Jarria
Neoneurus
Sigalphus
Diospilus
Ascogaster
Mirapotes
Toxoneuron
Euania sp. 2
Euania sp. 1
Euania sp. 3
Eufoenus
Gasteruption sp. 3
Gasteruption sp. 1
Gasteruption sp. 2
Schlettererius
Megischus 6.
Megischus t.
Mega1 ra sp. 1
Megabra sp. 2
Orthogonalys
Taeniogonalas
Lycogaster
Vanhornia
Disogmus
Brachyserphus
APoglYPha
Codrus
Exallonyx
Phaenoseruhus
Helorus sp. 1
Helorus sp. 2
Ropronia
Proctotrupomorpha
inc. Ceraphronoidea
1J
Aculeata
]
Ichneumonidae
1
Braconidae
Evaniidae
Gasteruptiidae
Stephanidae
Megalyridae
Trigonalidae
Vanhorniidae
Proctotrupidae
Heloridae
-
~
-
Monomachis a.
Betylinae M228
Aclista sp. 1
Aclista sp. 2
Basalys
Diphoropria sp. 1
Diphorupria sp. 2
Spilomicrus sp. 3
Spilomicrus sp. 1
Spilomicrus sp. 2
Xesto hanes
Anaclaris
Ibalia
103
3
Maamingidae
Monomachidae
Diapriidae
J Cynipoidea
7
Chalcidoidea
J
Ceraphronoidea
Platygastroidea
Trissolcus
J
Figure 9. Simultaneous analysis of molecular a n d unordered morphological data (molGPwts, morphU model). Strict
consensus of 2 shortest trees, length 6301.5.
104
M. DOWTON and
____
A. D. AUSTIN
~
Ichneumonoidea
Ichneumon
‘Xorides
Megalohelcon
] Ichneumonidae
Jcirra
Neoneurus
Diospilus
Ascogaster
Toxoneuron
Mirapotes
Sigalphus
Megischus b.
Meptschus t
ScKkttererius
Megalyro sp. 1
Megalyra sp. 2
Orthogonalys
Taeniogonalas
Lycogaster
Eufoenus
Gasteruption sp. 1
Gusteruption sp. 3
Gusteruption sp. 2
Primeuchrovus
Rhagigoster
Apis
Sceliphron
Myrmecio
Vespula
Evania sp. 2
Evania sp. 1
Euania sp. 3
Conostigmus
Dendrocerus
Ceraphron sp. 1
Aphanogmus
CeraDhron SD. 2
Braconidae
Stephanidae
Megalyridae
] Trigonalidae
1
1
Gasteruptiidae
Aculeata
] Evaniidae
I
Ceraphronoidea
) Heloridae
-
Vanhorniidae
Proctotrupidae
J
Proctotrupomorpha
ALIvtrvpa
Inostemma
J
-
Maamingidae
) Monomachidae
1
1
Diapriidae
J
lbalia
Xestophones
[Aphytis
Gonatocerus
- _ _ Brachymerio
__ Megastigmus
-_i Melittobia
Pteromalus
] Cynipoidea
-
1
Chalcidoidea
Figure 10. Simultaneous analysis of molecular a n d morphological d a t a , i n which some morphological transformations
a r e ordered (molU, morphU/O model). Strict consensus of 8 shortest trees, length 8376.
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
Orussus
Hartigia
Megalyra sp. 1
Megalyra sp. 2
Megischus b.
Schlettererius
Megischus t.
Ort h ogonal,ys
Lycogaster
Taeniogona las
3
Symphyta
3
Megalyridae
]
]
Trigonalidae
Euania sp. 2
]
Evaniidae
J
Ceraphronoidea
[ Xiphydria
2
Euania sp. 31
Evania
Eufoenus
Conostigmus .
Dendrocerus
Ceraphron sp. 1
Aphanogmus
Ceraphron sp. 2
Prinietrchroeus
Rhagigostc'r
Vespula
Mvrmecm
4Sceliphron
A ~ L S
7
Venturia
t
Ichneumonoidea
c Ichneumon
Xorides
Megalohelcon
Dolopsidea
Jarra
Neoneurus
Toroneuron
Ascogaster
Stephanidae
Aculeata
J
]
Ichneumonidae
1
I
J
Proctotrupomorpha
105
Ropronia
Helorus sp. 1
]
Helorus SD. 2
Pelecmus'
Vanhornaa
Disogmus
APodYPha
Brarliysrrphus
Phaenoserphus
Codrus
Exallonvx
Maam inga
Mononiachis
)
Monomachis a.
Diphoropria sp. 1
Diphoropria sp. 2
Betylinae M228
Aclista sp. 1
Aclista sp. 2
Basalys
Spilomicrus sp. 2
Spilomicrus sp. 1
Spilomicrus sp. 3
Leptonmstix
Pteronialus
Encarsia
Eusandalum
Megastigmus
Melittobia
Brachymeria
Trichilogastw
Aphytis
Gonatocerus
Sparasion
Aphanomerus
Amitus
Allotropa
Inostemnia
Scelio
Trimorus
Baryconus
Trissolcus
Ceratobaeus
1
Braconidae
Cynipoidea
Heloridae
Vanhorniidae
Proctotrupidae
Maamingidae
Monomachidae
Diapriidae
i
1
Chalcidoidea
J
1
Platygastroidea
J
Figure 11. Simultaneous analysis of molecular a n d morphological d a t a , i n which some morphological transformations
a r e ordered (molU ex. COI-3, morphU/O model). Strict consensus of 77 shortest trees, length 6173.
LO6
XI. DOWTON and A. D. AUSTIN
-. Ill.,.l"
Hartigia
Xzphydria
Primeuchroeus
Rhagigaster
Vespuln
r Myrniecia
___.~_____
r--
1 Symphyta
1
3
Aculeata
Megalyridae
1 Trigonalidae
1 Stephanidae
Euania sp. 2
Euania sp. 1
Evanin sp. 3
Eufoenus
Gastrruption sp. 3
Gasteruption sp. 1
Gusteruption sp. 2
Conostigmus
Dendrocerus
Ceraphron sp. 1
Aphanogniu s
Ceraphron SP. 2
Venturza
Ichneumon
Xorides
Megalohelcon
Jarra
Dolopsidea
7
Neoneurus
Diospilus
Mirapotrs
Toxoneuron
Ascogaster
Sigalphus
Maaminga
Monorriach is
Monomachis o
Basalys
Spilomicrus sp. 2
Spilomicrus sp. 1
Spilomicrus sp. 3
Diphoropria Sp. 1
Diphoropria sp. 2
Betylinae M228
iAclista SD. 1
Evaniidae
r-
-1
Proctotrupomorpha
Gasteruptiidae
Ceraphronoidea
3
Ichneumonidae
Braconidae
Maaminedae
Monomachidae
Diapriidae
]
Cynipoidea
3
Heloridae
-
Vanhorniidae
I
Proctotrupidae
Chalcidoidea
-
Allotropn
lnostemmn
Aphanomerus
Anzitus
J
1
Platygastroidea
J
Figure 12. Simultaneous analysis of molecular and morphological d a t a , i n which some morphological transformations
a r e ordered (molGP, morphU/O model). Strict consensus of 4 s h o rt e s t trees, length 8356.
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
Orussus
Hart igia
Xiphydria
Primeuchroeus
Rhagi aster
vespufi
Myrmecia
Apis
Sceliphron
Megalyra sp. 1
C fiiegalyra sp. 2
Megischus b.
LSchlettererius
Megischus t.
Orthogonalys
r
4~a~iw;"las
Evania sp. 2
Euania sp. 1
Euania sp. 3
Eufaenus
Casteruption sp. 1
Gasteruption sp. 3
Gasteruption sp. 2
I
Aculeata
3
Megalyridae
]
Stephanidae
]
Trigonalidae
1
Evaniidae
I3
107
Gasteruptiidae
Ichneumonidae
3
Neoneurus
Sigalphus
Diospilus
Ascogaster
Mirapotes
Toxoneuron
Conostigmus
Dendrocerus
Aphanogmus
Ceraphron sp. 2
Ceraphron sp. 1
Sparasion
Allotropa
Inostemma
Aphanomerus
Amitus
Scelio
Dimorus
Baryconus
Ceratobaeus
Dissolcus
Ropronia
Helorus sp. 1
Helorus sp. 2
Pelecinus
Vanhornin
Disogmus
Brachyserphus
Apogbpha
Codrus
Exallonyx
Phaenoserph us
Maaminga
Monomachis
Monomachis a.
Basalys
Diphoropria sp. 1
Diphoropria sp. 2
Spilomicrus sp. 3
Spilomicrus sp. 1
Spilomicrus sp. 2
Betylinae M228
Aclista sp. 1
Aclista sp. 2
Xestophanes
Anncharis
Ibalia
Encarsia
Megastigm us
Eusandalum
r Melittohia
Pteromalus
Leptomastix
Brachymeria
Trichilogaster
Aphytis
Gonatocerus
1
Braconidae
i
:
1
"
;
Proctotrupomorpha
inc. Ceraphronoidea
~
Ceraphronoidea
3
Heloridae
-
Vanhorniidae
1
-
3
Proctotrupidae
Maamingidae
Monomachidae
Diapriidae
J
J
1
Cynipoidea
Chalcidoidea
Figure 13. Simultaneous analysis of molecular a n d morphological d a t a , i n which some morphological transformations
are ordered (molGPwts, morphU/O model). Strict consensus of 8 shortest trees, length 6361.5.
108 91. DOWl'ON and A. D. AUSTIN
b
4-
Orussidae
Aculeata
Megalyridae
Trigonalidae
Stephanidae
Evaniidae
(fasteruptiidae
Ceraphronoidea
Ichneumonidae
Braconidae
Maamingidae
Monomachidae
Diapriidae
Cynipoidea
Heloridae
Ectoparasitic
Endoparasitic
Biology unknown
Vanhorniidae
Proctotrupidae
Chalcidoidea
I
Pla tygastroidea
Figure 14. Transitions between ectoparasitism and endoparasitism during the evolution of the Apocrita. Groundplan
biologies for. the included families were mapped onto the cladogram presented in Figure 8 (reduced to a family-level
cladogram). We do not consider this a preferred phylogeny, but one that retains many of the relationships identified
as robust to the method of analysis. Note that the Chalcidoidea are coded as endoparasitic. Although they contain
many ectoparasitoids, the putative basal lineages are egg endoparasitoids. The Megalyridae were coded as ectoparasitic
(ShLiw. 1990)>while the Ceraphronoidea were coded as ectoparasitic based on the biology of Megaspilidae (Gauld &
Bolton, 1996).
Trigonalidne is the only group which departs from this
pattern in that its members have an unusual biology
where eggs are deposited onto foliage and are then
ingested by larval caterpillars o r sawflies (Weinstein
& Austin. 1991). Furthermore, the remaining basal
lineages (the Evanioidea and Megalyridae) have biologies closely aligned with the putative groundplan.
The hulacidae (Evanioidea) are endopararasitic on
wood-boring beetle larvae, while the Megalyridae attack xylophagous, coleopteran larvae. Nevertheless,
we wish t o stress that the transitions depicted in
Figures 1-1 and 15 do not represent our 'preferred
hypothesis', but exhibit a range of relationships t h a t
were relatively robust t o the choice of analytical model.
LVhere resolved, the basal lineages of the F'roctotrupomorpha are generally endoparasitic (Figs 4-8,
10-13; mapped in Fig. 14). This is primarily because
the only superfamily within the Proctotrupomorpha to
contain ectoparasitoids are the Chalcidoidea, which
was always recovered apically. The Ceraphronoidea
also contain ectoparasitoids, and similarly fell out apically when recovered within the Proctotrupomorpha
(Figs 10-1 3 ) , although our molecular data and recent
general assessment of the Ceraphronoidea (see Whitfield, 1998)argue strongly for this group being placed
outside the Proctotrupomorpha. These findings suggest
t h a t endoparasitism is the groundplan state for the
Proctotrupomorpha, with a reversion to ectoparasitism
within the Chalcidoidea. Indeed, the extant basal lineages of the Chalcidoidea s. 1. (Mymarommatoidea and
Mymaridae: Gibson, 1986; Campbell et ul., 2000) are
endoparasitoids. Although such a reversion is generally considered controversial (see Whitfield, 1998),
we have documented similar phylogenetic evidence
for a return to ectoparasitism among the Rraconidae
(Dowton & Austin, 1998). We would argue that parsimony should be allowed to arbitrate the groundplan
biology for the Proctotrupomorpha, rather than intuition.
Further, within the Proctotrupomorpha one of the
most intriguing biological scenarios comes from the
sister group relationship between the Platygastroidea
and Chalcidoidea predicted by our molecular data,
which are arguably more robust that the available
morphological data (but see Gibson, 1999). In this
scenario, the likely ground-plan state for this clade is
110
______
XI. DOWTON and A. D. AUSTIN
Belshaw, Ferdinand0 Bin, Paul Dangerfield, John
Early, F Felipe, Scott Field, G. Fitt, Ian Gauld, Gary
Gibson, E. Grissell, Paul Hanson, €3. Hatami, John
Heraty. Marianne Hellcrs, P. Horne, G. Jackson, John
Jennings. Norm Johnson, Mike Keller, D. Kent, J.
King. J Kitt, M. Kulbars, Nina Laurence, Lewis, Patrick rvlardulyn, Lubomir Masner, Gwen Mayo, D.
Murray, Ian Naumann, J. O'Hara, Donald Quicke,
Nathan Schiff, Scott Shaw, Aguiar Sharkov, D. Smith,
R. Storey, Gary Taylor, N. Tonkin, G. Tribe, G. Walter,
Q. Wang, A. Wells, Bob Wharton, J i m Whitfield, N.
Zareh Thanks are also due to Rob Whelan for kindly
providing laboratory and office space to the senior
author
REFERENCES
Austin AD, Field SA. 1997. The ovipositor system on
sixlionid and platygastrid wasps (H-ymenoptera: Platygastroidea): comparative morphology and phylogenetic
imp1icat.ions. Int.ertebrate Taxonomy 11: 1-87.
Belshaw K, Quicke DW. 1997. A molecular phylogeny
of thtz Aphidiinae (Hymenoptera: Braconidae). Molecular
Ph,y/ogenr,lics & Ecolution 7: 281-293.
Belshaw K, Dowton M, Quicke DLJ, Austin AD.
2OOO. Estimating ancestral geographical distributions: a
C;andwan:m origin for aphid parasitoids? Pmceedings of
tht> R i q a ' Society of London B 267: 491496.
Bremer K. 1988. The limits of amino acid sequence data
i i i angiosperm phylogenetic reconstruction. Evolution 42:
;9;,-xo:3.
Brothers 1)-J, Carpenter JM. 1993. Phylogeny of Aculeata:
Chrysidoidea and Vespoidea (Hymenoptera). Journal of
Hynimapteran Research 2: 227-304.
Camphell B, Heraty J, Rasplus J-Y, Chan K, SteffenCampbell J. Babcock C. 2000. Molecular systematics
of the Chalcidoidea using 28S-D2 rDNA. In: Austin AD,
Ihwton h2, eds. Hymenoptera: Evolution, Biodiversity and
Rio/ogfco/ Control. Collingwood, vie: CSIRO, 59-73.
Carpenter J M , Wheeler WC. 1999. Towards simultaneous
analysis of morphological and molecular data in Hymen~lptera.Zoologica Scnpta 28: 251-260.
Cunningham CW. 1997. Is congruence between data
partitions a reliable predictor of phylogenetic accuracy?
Empirically testing an iterative procedure for choosing
iiniong- phylogenetic methods. Systematic Biology 46:
.H54--478.
Dowton M, Austin AD. 1994. Molecular phylogeny of
!he insect order Hymenoptera: Apocritan relationships.
Procwdings of the National Acadeiny of Sciences USA
91: 9911-9915.
Do nt o n M, Austin AD. 1995. Increased genetic diversity
,II niitochondrial genes is correlated with the evolution
I J parasitism
~
in the Hymenoptera. Journal of Molecular
Er'OjUfi(J7?41: 958-963.
Dowton M, Austin AD. 1997a. Evidence for AT-transversion hias in wasp (Hymenoptera: Symphyta) mito,:hondrial genes and its implications for the origin of
parasitism. Journal of Molecular E i d u t i u n 44: 398-405.
Dowton M, Austin AD. 1997b. The evolution of strandspecific compositional bias. A case study in the hymenopteran mitochondria1 16s rRNA gene. hloliw lar Biology
& Euolution 14: 109-112.
Dowton M, Austin AD. 1998. Phylogenetic relationships
among the microgastroid wasps (Hymenoptera: Hraconidae): combined analysis of' 16s and 28s rIlNA genes,
and morphological data. Molecular Phylogenetics & E w h -
tion 10: 354-366.
Dowton M, Austin AD, Dillon N, Bartowsky E. 1997.
Molecular phylogeny of the apocritan wasps: The Proctotrupomorpha and Evaniomorpha. Systematic Ento~r;.o/ogy
22: 245-255.
Farris JS, Kallersjo M, Kluge AG, Bult C. 1994. Testing
significance of incongruence. Cladistics 1 0 31 5-31 9.
Felsenstein J . 1985. Confidence intervals on phylogenies:
a n approach using the bootstrap. Ecdutiorr 39: X - i 9 1 .
Gauld ID, Bolton B. 1996. The HyinPnoptem. Oxford:
Oxford University Press.
Gibson GAP. 1986. Evidence for monophyly and relationships of Chalcidoidea, Mymaridae, and Mymarommatidae (Hymenoptera: Chalcidoidea: Terebrantes).
Canadian Entomologist 118: 205-240.
Gibson GAP. 1999. Sister-group relationships of the
Platygastroidea and Chalcidoidea (Hymenoptera) - a n
alternative hypothesis to Rasnitsyn (1988).Zoologica
Scripta 28: 125-138.
Hanson PE, Gauld ID. 1995. The Hyinenoptera o f Costa
Rica. Oxford: Oxford University Press.
Hasegawa M, Kishino H, Yano T. 1985. Dating the
human-ape split by a molecular clock of mitochondrial
DNA. Journal of Molecular Evolution 2 2 160-174.
Hillis DM, Huelsenbeck JP, Cunningham CU'. 1994.
Application and accuracy of molecular phylogenies. Science 264: 671-677.
Huelsenbeck JP, Nielsen R. 1999. Variation in the
pattern of nucleotide substitution across sites. -/rjurna/
of Molecular Evolution 48. 86-93.
Kallersjo M, Albert VA, Farris JS. 1999. Homoplasy
increases phylogenetic structure. Cladistita 15: 91-93.
LaSalle J, Gauld ID. 1993. Hymenoptera and Biodicersity
Wallingford, UK: CAB International.
Maddison WP, Maddison DR. 1996. Mal:Clndr L'. 3.06.
Sunderland, MA: Sinauer Associates, Inc.
Poe S, Swofford DL. 1999. Taxon sampling revisited.
Nature London 398: 299-300.
Rasnitsyn AP. 1980. The origin and evolution of the
Hymenoptera. Trudy Paleontologicheskofir, Institutci. Akaderniya Nauk SSSR 174: 1-192 (in Russian).
Rasnitsyn AP. 1988. An outline of evolution (of the
hymenopterous insects (Order Vespida). Orientoi Irz,secto
22: 115-145.
Ronquist F, Rasnitsyn A€', Roy A, Eriksson K, Lindgren M. 1999. Phylogeny of the Hymenoptera: .4cladistic.
reanalysis of Rasnitsyn's ( 1988) data. ZodCJglCf.2 Scriptn
28: 13-50.
Sanderson MJ, Wojciechowski MF. 2OOO. Improved
SIMULTANEOUS ANALYSIS OF APOCRITAN RELATIONSHIPS
bootstrap confidence limits in large-scale phylogenies,
with an example from Neo-Astragalus (Leguminosae).
Systematic Biology 4 9 671-685.
Sharkey MJ, Wahl DB. 1992. Cladistics of the Ichneumonoidea (Hymenoptera). Journal of Hymenoptera
Research 1: 15-24.
Shaw SR. 1990. Phylogeny and biogeography of the
parasitoid wasp family Megalyridae (Hymenoptera).
Journal of Biogeography 17: 569-581.
Stanger-Hall K, Cunningham CW. 1998. Support for
a monophyletic Lemuriformes: overcoming incongruence
between data partitions. Molecular Biology & Evolution
1 5 1572-1577.
Swofford DL. 1999. PAUP*, Phylogenetic analysis using
parsimony. Sunderland, MA: Sinauer Associates, Inc.
Swofford DL, Olsen GJ, Waddell PJ, Hillis DM. 1996.
Phylogenetic Inference. In: Hillis DM, Moritz C, Mable
BK, eds. Molecular Systematics. Sunderland, MA Sinauer,
407-514.
Thompson JD, Higgins DG, Gibson TJ. 1994. CLUSTAL
W improving the sensitivity of progressive multiple
sequence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice.
Nucleic Acids Research 2 2 4673-4680.
Vilhelmsen L. 1997. The phylogeny of lower Hymenoptera
(Insecta), with a summary of the early evolutionary
history of the order. Journal of Zoologxcal and Systematic
Evolutionary Research 35: 49-70.
Vilhelmensen L. 2000. The ovipositor apparatus of basal
111
Hymenoptera (Insecta): phylogenetic implications and
functional morphology. Zoologica Scripta 29: 319-345.
Weinstein P, Austin AD. 1991. The host relationships of
trigonalyid wasps (Hymenoptera: Trigonalyidae), with a
review of their biology and catalogue to world species.
Journal of Natural History 25: 399433.
Wheeler WC. 1995. Sequence alignment, parameter sensitivity, and the phylogenetic analysis of molecular data.
Systematic Biology 44: 321-331.
Wheeler W. 1996. Optimization alignment The end of
multiple sequence alignment in phylogenetics? Cladistics
12: 1-9.
Wheeler WC, Gladstein DS. 1994. MALIGN: A multiple
sequence alignment program. Journal of Heredity 85:
417418.
Wheeler WC, Cartwright P, Hayashi CY. 1993. Arthropod phylogeny: a combined approach. Cladistics 9:
1-39.
Whitfield JB. 1992. Phylogeny of the non-aculeate Apocrita
and the evolution of parasitism in the Hymenoptera.
Journal of Hymenopteran Research 1: 3-14.
Whitfield JB. 1998. Phylogeny and evolution of hostparasitoid interactions in Hymenoptera. Annual Review
of Entomology 4 3 129-151.
Wiegmann BM, Mitter C, Farrell B. 1993. Diversification
of carnivorous parasitic insects: extraordinary radiation or
specialized dead end? American Naturalist 142 737-754.
Williams PL, Fitch WM. 1990. Phylogeny determination
using the dynamically weighted parsimony method.
Methods in Enzymology 183 615-626.