When phylogenetic assumptions are violated: base compositional heterogeneity and among‐site rate variation in beetle mitochondrial phylogenomics

The ability to generate large molecular datasets for phylogenetic studies benefits biologists, but such data expansion introduces numerous analytical problems. A typical molecular phylogenetic study implicitly assumes that sequences evolve under stationary, reversible and homogeneous conditions, but this assumption is often violated in real datasets. When an analysis of large molecular datasets results in unexpected relationships, it often reflects violation of phylogenetic assumptions, rather than a correct phylogeny. Molecular evolutionary phenomena such as base compositional heterogeneity and among‐site rate variation are known to affect phylogenetic inference, resulting in incorrect phylogenetic relationships. The ability of methods to overcome such bias has not been measured on real and complex datasets. We investigated how base compositional heterogeneity and among‐site rate variation affect phylogenetic inference in the context of a mitochondrial genome phylogeny of the insect order Coleoptera. We show statistically that our dataset is affected by base compositional heterogeneity regardless of how the data are partitioned or recoded. Among‐site rate variation is shown by comparing topologies generated using models of evolution with and without a rate variation parameter in a Bayesian framework. When compared for their effectiveness in dealing with systematic bias, standard phylogenetic methods tend to perform poorly, and parsimony without any data transformation performs worst. Two methods designed specifically to overcome systematic bias, LogDet and a Bayesian method implementing variable composition vectors, can overcome some level of base compositional heterogeneity, but are still affected by among‐site rate variation. A large degree of variation in both noise and phylogenetic signal among all three codon positions is observed. We caution and argue that more data exploration is imperative, especially when many genes are included in an analysis.     

Sexual selection and infection by ectoparasites in Wellington tree weta,Hemideina crassidens (Orthoptera: Anostostomatidae)

Abstract Sexual selection can affect the prevalence and intensity of infection of individuals by ectoparasitic mites. According to this theory, males should exhibit greater infection by parasites than females and juveniles should be less infected than adults. In the wild, I investigated whether prevalence and intensity of the chyzeriid mite, Nothotrombicula deinacridae (Dumbleton) differed between the sexes and between developmental stages in Wellington tree weta, Hemideina crassidens. Despite being under strong sexual selection, male tree weta did not exhibit greater parasitism and there was some evidence that adults and juveniles differed in prevalence. The sexual selection hypothesis was not supported in this study.     

Separating host-tree and environmental determinants of honeydew production by Ultracoelostoma scale insects in a Nothofagus forest

Abstract 1. Sugar‐rich honeydew excreted (‘produced’) by insects feeding on phloem sap is a key energy flow in a range of temperate and tropical ecosystems. The present study measured honeydew produced by Ultracoelostoma sp. (Homoptera: Coelostomidiidae) scale insects feeding on Nothofagus solandri var. solandri (Hook f.) Oerst. trees in a temperate evergreen forest in New Zealand. Simultaneous measurements of environmental variables and canopy photosynthesis were conducted to allow separation of host‐tree and environmental determinants of honeydew production. These relationships were further examined in experiments where canopy photosynthesis was manipulated by shading or plant nitrogen levels increased by foliar spray. 2. Rates of honeydew production varied nine‐fold from a maximum (± 1 SE) of 64.4 ± 15.2 mg dry mass m⁻² bark h⁻¹ in early summer (December) to a minimum of 7.4 ± 4.2 mg m⁻² h⁻¹ in winter (August). Rates of production measured 1.4 m from the base of the trees’ stems varied significantly with stem diameter, and were higher on medium‐sized (18 cm diameter) than small or large stems. 3. Rates of production were significantly related to environmental conditions over the hours preceding measurement (air temperature and air saturation deficit averaged over the preceding 24 and 12 h respectively). There was no evidence that rates of production were directly related to short‐term changes in the supply of carbohydrates from the canopy (either when compared with measurements of unmanipulated photosynthetic rate, or after sugar levels were manipulated by shading 80% of host‐trees’ leaf area), or to changes in phloem nitrogen content. 4. The results show that there is no clear effect of host‐tree carbon supply on honeydew production; if production is related to photosynthesis, the effect of this is much less important that the large and significant direct effect of environmental conditions on honeydew production.     

Impacts of plant roots on soil CO cycling and soil–atmosphere CO exchange

Carbon monoxide (CO) plays a major role in tropospheric chemical dynamics. Accordingly, global CO budgets have been reasonably well documented. Atmospheric CO consumption by soils contributes significantly to these budgets, with the magnitude of the sink generally considered to reflect a balance between microbial uptake and abiological production. However, assays of live fine roots showed that diverse intact plants produced carbon monoxide at net rates ranging from 2 to 3000 mµg gdw⁻⁻¹ d⁻⁻¹. CO production was greater for legumes than nonlegumes, and primarily associated with nodules. Excised roots from woody and herbaceous plants produced CO at comparable rates. CO production rates were similar for roots of intact plants and roots excised from those plants. The magnitude of net CO fluxes from roots was determined in part by the balance between simultaneous production and consumption. Surface sterilization of roots indicated that CO consumption was due, in part, rhizoplane CO-oxidizing bacteria, but maximum CO consumption rates were typically only a small fraction of net production rates. Assays in a Brazilian agroecosystem indicated that root CO production affects soil–atmosphere CO exchange. Estimates of global CO production rates indicated that roots contribute about 170–260 Tg CO to the soil atmosphere annually, an amount comparable to current estimates of atmospheric CO uptake by soils, and much larger than estimates of net abiological soil CO production.     

Phylogeny and classification of Cucujoidea and the recognition of a new superfamily Coccinelloidea (Coleoptera: Cucujiformia)

A large‐scale phylogenetic study is presented for Cucujoidea (Coleoptera), a diverse superfamily of beetles that historically has been taxonomically difficult. This study is the most comprehensive analysis of cucujoid taxa to date, with DNA sequence data sampled from eight genes (four nuclear, four mitochondrial) for 384 coleopteran taxa, including exemplars of 35 (of 37) families and 289 genera of Cucujoidea. Maximum‐likelihood analyses of these data present many significant relationships, some proposed previously and some novel. Tenebrionoidea and Lymexyloidea are recovered together and Cleroidea forms the sister group to this clade. Chrysomeloidea and Curculionoidea are recovered as sister taxa and this clade (Phytophaga) forms the sister group to the core Cucujoidea (Cucujoidea s.n .). The nitidulid series is recovered as the earliest‐diverging core cucujoid lineage, although the earliest divergences among core Cucujoidea are only weakly supported. The cerylonid series (CS) is recovered as monophyletic and is supported as a major Cucujiform clade, sister group to the remaining superfamilies of Cucujiformia. Currently recognized taxa that were not recovered as monophyletic include Cucujoidea, Endomychidae, Cerylonidae and Bothrideridae. Biphyllidae and Byturidae were recovered in Cleroidea. The remaining Cucujoidea were recovered in two disparate major clades: one comprising the nitidulid series + erotylid series + Boganiidae and Hobartiidae + cucujid series, and the other comprising the cerylonid series. Propalticidae are recovered within Laemophloeidae. The cerylonid series includes two major clades, the bothriderid group and the coccinellid group. Akalyptoischiidae are recovered as a separate clade from Latridiidae. Eupsilobiinae are recovered as the sister taxon to Coccinellidae. In light of these findings, many formal changes to cucujiform beetle classification are proposed. Biphyllidae and Byturidae are transferred to Cleroidea. The cerylonid series is formally recognized as a new superfamily, Coccinelloidea stat.n. Current subfamilies elevated (or re‐elevated) to family status include: Murmidiidae stat.n. , Teredidae stat.n. , Euxestidae stat.n. , Anamorphidae stat.rev. , Eupsilobiidae stat.n. , and Mycetaeidae stat.n. The following taxa are redefined and characterized: Cleroidea s.n. , Cucujoidea s.n. , Cerylonidae s.n. , Bothrideridae s.n. , Endomychidae s.n. A new subfamily, Cyclotominae stat.n. , is described. Stenotarsinae syn.n. is formally subsumed within a new concept of Endomychinae s.n.     

Molecules,morphology and minute hooded beetles: a phylogenetic study with implications for the evolution and classification of Corylophidae (Coleoptera: Cucujoidea)

Eight genes (nuclear: 18S, 28S, H3, CAD; mitochondrial: 12S, 16S, COI, COII) and morphology were used to infer the evolutionary history of Corylophidae, some of the smallest free‐living insects. The study included 36 corylophid exemplars, representing approximately 60% of the known generic diversity of the family and 16 cucujoid outgroup taxa. Multiple partitioning strategies, molecular datasets, combined datasets and different taxon sampling regimes using maximum likelihood and mixed‐model Bayesian inference were utilized to analyse these data. Most results were highly concordant across analyses. There was strong agreement across (i) partitioning strategies, (ii) maximum likelihood and Bayesian inference analyses of the molecular data, and (iii) Bayesian inference of the molecular data alone and Bayesian inference of the combined morphological and molecular data when all terminal taxa were included. When a strict taxon sampling protocol was employed so that only single generic exemplars were included, deep relationships were affected in the resulting phylogenetic hypotheses. Under such narrow sampling strategies, deep phylogenetic relationships were also sensitive to the choice of generic exemplars. Although it is often challenging to obtain single representatives for many taxa in higher‐level phylogenetic analyses, these results indicate the importance of using denser taxon sampling approaches even at the specific level for genera included in such studies. Molecular data alone support Anamorphinae (Endomychidae) strongly as the sister group of Corylophidae. In combined data analyses, Coccinellidae is recovered as the sister group to Corylophidae. In all analyses, Corylophidae and the subfamily Corylophinae are recovered as monophyletic. The monophyly of Periptyctinae was untested, as only a single species was included. All included corylophine tribes were recovered as monophyletic with the exception of Aenigmaticini; Aenigmaticum Matthews forms the sister group to Orthoperus Stephens and Stanus?lipiński et al. is recovered as the sister group of Sericoderus Stephens. Stanus tasmanicus?lipiński et al. is transferred to a new genus, Pseudostanus Robertson, ?lipiński & McHugh gen.n. incertae sedis. We propose a new tribe, Stanini Robertson, ?lipiński & McHugh trib.n. for Stanus bowesteadi?lipiński et al. and a new concept of Aenigmaticini sensu.n. to include only the nominate genus. Anatomical transitions associated with corylophid miniaturization are highlighted. Key phenotypic modifications and elevated rates of substitution in nuclear rRNA genes are evident in a subgroup of Corylophinae that includes the most diminutive species. Other taxonomic and evolutionary implications are discussed in light of the results.