Phylogenetics and molecular clocks reveal the repeated evolution of ant-plants after the late Miocene in Africa and the early Miocene in Australasia and the Neotropics

Please download to get full document.

View again

of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Similar Documents
Information Report

How To, Education & Training


Views: 25 | Pages: 14

Extension: PDF | Download: 0

Ant–plant symbioses involve over 110 ant species in five subfamilies that are facultative or obligate occupants of stem, leaf or root domatia formed by hundreds of ant-plant species. The phylogenetic distribution and geological ages of these
  Phylogenetics and molecular clocks reveal the repeated evolutionof ant-plants after the late Miocene in Africa and the earlyMiocene in Australasia and the Neotropics Guillaume Chomicki and Susanne S. Renner Systematic Botany and Mycology, Department of Biology, University of Munich (LMU), Munich 80638, Germany  Author for correspondence: Guillaume ChomickiTel: +498917861285Email: Received:  1 August 2014 Accepted:  4 December 2014New Phytologist   (2015) doi : 10.1111/nph.13271 Key words:  ant  –  plant symbioses, domatia,extrafloral nectaries (EFNs), mutualism,myrmecophytes, radiations, symbioses. Summary   Ant  –  plant symbioses involve over 110 ant species in five subfamilies that are facultative or obligate occupants of stem, leaf or root domatia formed by hundreds of ant-plant species.The phylogenetic distribution and geological ages of these associations, and the frequency ofgains or losses of domatium, are largely unknown.   We compiled an up-to-date list of ant domatium-bearing plants, estimated their probabletrue number from model-based statistical inference, generated dated phylogenies that include c  . 50% of ant-plant lineages, and traced the occurrence of domatia and extrafloral nectarieson a 1181-species tree, using likelihood and Bayesian methods.   We found 681 vascular plants with domatia (159 genera in 50 families) resulting from mini-mally 158 inferred domatium srcins and 43 secondary losses over the last 19Myr. The oldestAfrican ant  –  plant symbioses are younger than those in Australasia and the Neotropics. Thebest statistical model suggests that the true number of myrmecophytes may approach 1140species.   The phylogenetic distribution of ant-plants shows that domatia evolved from a range ofpre-adapted morphological structures and have been lost frequently, suggesting that domatiahave no generalizable effect on diversification. The Miocene srcin of ant  –  plant symbioses isconsistent with inferred changes in diet and behaviour during ant evolution. Introduction The fossil record and molecular clock dating show that ants andplants have been coexisting for at least 120 Myr (Brady   et al. ,2006; Moreau  et al. , 2006; Bell  et al. , 2010; Magall  on  et al. ,2013; Moreau & Bell, 2013). Traits that support a long history of ant  –  plant interactions include elaiosomes, fatty appendages onseeds meant for ant dispersers that may have occurred as early as75 Myr ago (Ma) (Dunn  et al. , 2007). Extrafloral nectaries(EFNs), involving a defence mutualism through sugar secretionrecruiting ant mutualists, are known from Oligocene fossils(Pemberton, 1992) and evolved over 450 times in vascular plants(Weber & Keeler, 2013). A third type of ant  –  plant mutualisminvolves ants living in myrmecophytes, plants with modifiedstructures to host ants (domatia). No fossil ant domatia areknown, nor has there been a phylogenetic analysis focusing onthese structures and the geological times when they arose or werelost. For the other two ant-related plant traits, namely elaiosomesand EFNs, recent analyses suggest that they fostered diversifica-tion, implying that mutualistic interaction with insects may haveimpacted macroevolutionary patterns (Lengyel  et al. , 2009; Weber & Agrawal, 2014). In the absence of a phylogeneticframework, it is unclear whether domatia also favoured diversifi-cation.Domatia occur in numerous plant species with modifiedleaves, stems or roots that provide cavities occupied by ants(Fig. 1). Some plants with domatia in addition possess specializedfood bodies or EFNs. The domatium-living ants in return pro-vide their plant hosts with protection against herbivores, withextra nutrients, or with the physical or chemical removal of com-peting plant species (Janzen, 1967, 1969; Davidson & McKey,1993; Jolivet, 1996; Renner & Ricklefs, 1998). At least 113 spe-cies of ants from five subfamilies  –   Myrmicinae, Formicinae,Dolichoderinae, Pseudomyrmecinae, and Ponerinae  –   occasion-ally or obligatorily nest in plants (McKey & Davidson, 1993).Examples of facultative (opportunistic) ant  –  plant symbioses(involving domatia) are species of   Tillandsia   that can host over30 arboreal ant species in their interlocked leaf bases (Benzing,1970; Dejean  et al. , 1995). Examples of obligate ant  –  plant sym-bioses are those between Central American species of   Vachellia  (formerly placed in  Acacia  ) and  Pseudomyrmex   ants of the  ferrugineus   group (Heil  et al. , 2005, 2009; Orona-Tamayo &Heil, 2013). Despite a large amount of data on aspects of chemi-cal ecology, food webs and feedback mechanisms between plants  2015 The Authors New Phytologist   2015 New Phytologist Trust New Phytologist   (2015)  1 Research  and ant symbionts (reviewed by Orona-Tamayo & Heil, 2013;Mayer  et al. , 2014), little is known about the evolution of thetraits that may have facilitated domatium-based symbioses andabout the frequency of their evolutionary turnover in the tropicalregions of Australasia, Africa and the New World where mostant  –  plant symbioses occur.Phylogenetic frameworks for both the plants and the ants haveso far been developed for three ant  –  plant symbioses, one from Africa, one from Southeast Asia and one from the Neotropics. Inthe African  Leonardoxa africana  , two of four subspecies have spe-cialized domatia that were colonized in parallel by older, pre-adapted ant species (Chenuil & McKey, 1996; Brouat  et al. ,2004). Species of the Southeast Asian  Crematogaster   subgenus Decacrema   independently colonized three groups of   Macaranga  species: the  Pachystemon   group  c  . 12 Ma, a smooth-stemmedgroup  c  . 5 Ma and the  M. pruinosa   group  c  . 4.5 Ma (Quek   et al. ,2004). As in  Leonardoxa  , colonization of plant hosts requiredpre-adaptations, such as the ability to adhere to slippery stems orexcavation behaviour linked to specific morphological features of their hosts (Federle  et al. , 1997, 2000; Markst € adter  et al. , 2000;Quek   et al. , 2004). Lastly, a subgroup of Mesoamerican  Vachellia  co-diversified with  Pseudomyrmex   ants, following a single coloni-zation event c. 5 Ma and subsequent host broadening within themyrmecophytic  Vachellia   (G  omez-Acevedo  et al. , 2010). Phylo-genetic studies of   Macaranga  ,  Piper   section  Macrostachys  , Neonauclea   and  Barteria   (without phylogenies of the relevantants) have shown independent evolution of domatia within thesegenera, followed by secondary losses (Blattner  et al. , 2001; Davies et al. , 2001; Tepe  et al. , 2004; Razafimandimbison  et al. , 2005;Peccoud  et al. , 2013). Because of the need to re-associate at eachgeneration, ant  –  plant symbioses likely involve little or no co-spe-ciation but rather co-diversification, where the interacting groups (a) (b)(c) (d) Fig.1  Diversity of ant domatia. (a)  Myrmephytum arfakianum  (Rubiaceae),Arfak Mountains, Papua. The domatium is aswollen hypocotyle with a system of internalgalleries. (b)  Hoya imbricata  (Apocynaceae),Indonesia. These so-called ‘external’ domatiaare formed by leaves pressed against the hosttree. (c)  Maieta guianensis (Melastomataceae), Seringalzinho, Rio Jau,Amazonas, Brazil. The domatia consist of leafpouches at the base of the lamina. (d)  Macaranga indistincta  (Euphorbiaceae) with Crematogaster   (Myrmicinae) ants, Sabah,Borneo. Note the Beccarian bodies and theentrance holes. The inset shows alongitudinal section of an  M.pearsonii  stemdomatium, showing the cultivation of scaleinsects by  Crematogaster   ants. Photos: (a, b)Andreas Wistuba; (c) Nigel Smith; (d) EduardLinsenmair; inset, Brigitte Fiala. New Phytologist   (2015)   2015 The Authors New Phytologist   2015 New Phytologist Trust Research New Phytologist 2  diversify by host broadening or switching (Ehrlich & Raven,1964; Cruaud  et al. , 2012; de Vienne  et al. , 2013).Domatia might be selectively favoured in plants living in nutri-ent-poor habitats, such as epiphytes (Janzen, 1974), plants thatalready have EFNs patrolled by nectar-foraging ants, or plantspatrolled by ants tending scale insects (Ward, 1991; Davidson &McKey, 1993). Wilson & H € olldobler’s (2005) dynastic-succes-sion hypothesis moreover posits that the transition from a dietinvolving predation on ground-dwelling insects to one involving secretions from tended hemipterans or from nectar glands,associated with aboveground living, occurred relatively late in thehistory of ants, coinciding with the evolution of angiosperm-dominated tropical forests that provided complex habitats. If such transitions in diet and habitat indeed evolved recently, thatis, no earlier than the Eocene, then myrmecophytes inhabited by arboreal ants might be relatively young, something that can betested with clock-dated phylogenies for relevant plant clades.By assembling a new list of domatium-bearing vascular plantspecies worldwide, a large phylogenetic framework for ant-plants,and dated phylogenies for half of all myrmecophyte lineages, weaddress the following questions about the evolution of ant  –  plantsymbioses.: (1) How often have domatia been gained or lost (a question answerable with minimal estimates from trait recon-structions on phylogenies)? Where in the land plants do we findthe highest concentrations of myrmecophyte srcins and the larg-est clades with myrmecophytic species and how clustered arethey? (2) Where are these clades located geographically? (3) How old are they? And (4) are there significant differences in the agesof myrmecophyte lineages in the Neotropics, Australasia and Africa? Such age differences might be expected because the Afri-can forests were more affected by Miocene and Pliocene climateoscillations than were Australasian and Amazonian forests (vanZinderen Bakker & Mercer, 1986; Jacobs, 2004). Materials and Methods Known ant-plants, types of domatia and inference of thelikely total ant-plant number  In order to assemble a species-level list of ant-plants we con-ducted a literature search in Google Scholar ( using the terms ‘myrmecophytes’, ‘domatia’, ‘antplants’ and ‘ant/plant symbiosis’; we also searched monographsof relevant genera, such as  Cecropia  ,  Myrmecodia  ,  Neonauclea  , Triplaris   and  Ruprechtia  . We incorporated the genus-level myrm-ecophyte lists of Davidson & McKey (1993), McKey & David-son (1993) and Jolivet (1996), and an unpublished list providedby Camilla Huxley-Lambrick in November 2013. The taxo-nomic assignment of species to genera and families was updatedfollowing recent literature and during GenBank ( ) searches for DNA sequences of myrm-ecophytes. We define a myrmecophyte as a plant species that hasa structure to host ants (a myrmecodomatium); this includesexternal domatia (Fig. 1b), but excludes plant structures usedby ants to make a nest (e.g. the root system of   Coryanthes  ,Orchidaceae). We classified domatia into eight types: (1) stem domatia, any hollow stem or twig, independent of the order or number of shoot axes transformed into domatia; (2) leaf pouches, all pouchdomatia formed on the petiole and/or lamina; (3) hollow rachis,the leaf rachis axis is swollen and hollow, as in  Tachigali  ; (4) leaf base domatia, a cavity formed into the spaces of interlocked leaf bases, as in  Tillandsia  ; (5) stipular domatia, which include stipu-lar thorns, stipular pouches, either closed or open; (6) root tuberdomatia, for a transformed root tuber; (7) external domatia, fordomatia formed by epiphytes with a structure pressed against thehost tree which can be a leaf (Fig. 1b) or a modified stem; (8)hypocotyle with galleries, for the unique domatia of the Hy-dnophytinae ( Myrmecodia  ,  Hydnophytum   and related genera).Our list of ant-plant species is almost certainly incomplete dueto overlooked literature and as yet unrecorded ant  –  plant symbio-ses. To estimate the true number of myrmecophytes, we used themodel comparison framework implemented in CatchAll (Bunge,2011). By using the same search terms (‘myrmecophytes’, ‘doma-tia’, ‘ant plants’ and ‘ant/plant symbiosis’) and each genus or spe-cies name from our list (Supporting Information Table S1) inGoogle Scholar (as of 1 September 2014), we obtained the fre-quency of publications per myrmecophyte species and used thisas input in CatchAll. We compared five nonparametric models(Good-Turing, Chao1, ACE, ACE1 and Chao  –  Bunge gamma-Poisson) and five parametric models (Poisson, single exponentialmixed Poisson, and mixtures of two, three and four exponentialsmixed Poisson) to find the best-fitting estimate (Bunge, 2011). Alignments and phylogenetic analyses In order to infer the minimal numbers of gains and losses of domatia in angiosperms, we searched GenBank for the 681myrmecophytes in our species-level list. For the 323 species pres-ent, we searched for their closest relatives, using previously pub-lished phylogenies, by including other congeneric species whengenera were small, or by similarity based on the 100 highest-scor-ing BLAST hits of the myrmecophyte target sequence. We alsoincluded a representative sample of domatium-lacking families of angiosperms, gymnosperms and ferns, typically with one speciesper family except for the largest angiosperm families where onespecies per subfamily was included. The resulting matrix con-sisted of 1181 species and 3958 sequences downloaded fromGenBank ( ), comprising thenuclear  18S   rDNA and ITS regions, the plastid genes  rbcL  , matK   ,  ndhF   and  atpB  , and the plastid spacers  trnL-trnF   and atpB-rbcL  . The final matrix comprised 1181 species and 38 080aligned nucleotides, with 57% missing data (cells in the matrix filled with ‘nnn’ or ‘  —  ’), including   rbcL   (799 sequences; 32%missing data),  matK    (752; 36%),  ndhF   sequences (532; 55%), atpB   sequences (358; 69%),  18S   rDNA sequences (304; 74%),ITS sequences (600; 49%),  trnL-trnF   sequences (488; 60%) and atpB-rbcL   (135; 88%). Accession numbers are in Table S2.Tips naming was automated with Phyutility (Smith & Dunn,2008), and sequences were aligned with MAFFT v7 (Katoh &Standley, 2013). The five genes ( rbcL, matK, atpB, ndhF  ,  18S  rDNA) were aligned using standard settings. For the more  2015 The Authors New Phytologist   2015 New Phytologist Trust New Phytologist   (2015) New Phytologist  Research  3  rapidly evolving spacer regions (ITS,  trnL-trnF, atpB-rbcL  ), weselected the option ‘leave gappy regions unaligned’, with a simi-larity threshold of 0.8. This approach allowed us to align com-plete ITS sequences across land plants. Minor alignment errorswere manually corrected in Mesquite v2.75 (Maddison & Madd-ison, 2011) and the matrices were concatenated in Geneious v5.4(Drummond  et al. , 2011).Maximum-likelihood (ML) inference relied on RAxML v7.0(Stamatakis  et al. , 2008) with 100 ML bootstrap replicates andthe analysis partitioned by gene region, all under the GTR  + Γ substitution model, as selected under the AIC criterion by jmod-eltest2 (Darriba   et al. , 2012), with six rate categories. The treewas rooted on  Selaginella moellendorfii  . The tree with all tipnames is presented in Fig. S1. Molecular clock dating of myrmecophyte groups In order to infer absolute divergence times for myrmecophyte lin-eages, we generated local phylogenies that were more densely sampled than our higher-level vascular plant tree (previous sec-tion). For this, we used published datasets representing nearly half of all myrmecophyte-containing lineages: namely   Barteria  (Peccoud  et al. , 2013),  Clerodendrum  ,  Leonardoxa   (Brouat  et al. ,2001) in Africa;  Cecropia  ,  Cordia   (Weeks  et al. , 2010), Mic-oniaeae (Melastomataceae; Michelangeli  et al. , 2004),  Piper  (Tepe  et al. , 2004),  Platymiscium   (Saslis-Lagoudakis  et al. , 2008), Ruprechtia  ,  Triplaris   (Sanchez & Kron, 2008) and  Vachellia  (G  omez-Acevedo  et al. , 2010) from the Neotropics; and Dischidia  ,  Hoya   (Wanntorp  et al. , 2006), the Hydnophytinae( Myrmecodia  ,  Hydnophytum  ,  Myrmephytum  ,  Squamellaria  ,  Anthorrhiza  ),  Macaranga   (Blattner  et al. , 2001; Davies  et al. ,2001) and  Neonauclea   (Razafimandimbison  et al. , 2005) from Australasia. Accession numbers are either in Table S2 or appearnext to the respective species name in Figs S2  –  S15. Alignmentand phylogenetic analyses were performed as described above forthe 1181-species tree, except that the Q-INS-i approach wasselected in MAFFT to take into account RNA secondary structure when aligning the ITS region, as recommended forthis marker when aligning fewer than 200 sequences (Katoh &Standley, 2013).Dating for all data matrices relied on BEAST v1.8(Drummond  et al. , 2012) and the GTR  + Γ  substitution modelwith six rate categories. The tree prior was a pure-birth (Yule)tree, with MCMC chain lengths between 20 and 60 milliongenerations, sampling every 10 000th generation, with the chainlength depending on convergence as determined by examining the log files in Tracer v1.5 (Rambaut & Drummond, 2009) afterremoval of a burn-in proportion of 10% of the trees. Unlessotherwise stated below, we used uncorrelated log-normal(UCLN) clock models. For calibration, we used either secondary constraints from other dated phylogenies or nucleotide substitu-tion rates. Secondary constraints were assigned normal distribu-tion priors with a standard deviation (SD) matching the 95%confidence interval from the srcinal study when presented orotherwise a 20% SD. Specifically, the secondary calibrations were:for  Piper,  the split between  Piper   and  Peperomia   was assigned anage of 91.2    10 Myr (Smith  et al. , 2008). For  Macaranga  , thesplit between  Blumeodendron   and the  Hancea   ( Mallotus   ( Maca- ranga  )) clade was assigned an age of 86.4    5Myr, the  Mallotus  plus  Macaranga   clade an age of 59   10 Myr, and the  Macaranga  crown an age of 33.5   12Myr (van Welzen  et al. , 2014). For Triplaris  / Ruprechtia  , the split between Brunnichieae and its sisterclade was assigned an age of 69.1   25 Myr (Schuster  et al. ,2013). For  Platymiscium  , we set the split between  Riedellia   and itssister clade to 47.2    5Myr (node 47 in Lavin  et al. , 2005). For Vachellia  , we assigned the split between the ( Vachellia constricta  ( V. schottii   ( V. neovernicosa  )) clade and its sister group, whichincludes a myrmecophyte clade, an age of 12.3    3Myr (G  omez- Acevedo  et al. , 2010). In the Boraginales, the relationshipsbetween the main clades were constrained to match the topology found by Weigend  et al.  (2013) with denser sampling of taxa andgenes. We assigned the split between the ( Nama   ( Eriodictyon   ( Wi-  gandia  )) clade and the rest of the Boraginales, including Cordia-ceae, an age of 60.4   10Myr (Weeks  et al. , 2010), whichresulted in an age of 52 Myr for the  Ehretia   stem group, consistentwith Eocene  Ehretia   fossil fruits (Gottschling   et al. , 2002). Forthe Hydnophytinae ( Squamellaria, Hydnophytum  ,  Myrmephytum  ,  Anthorrhiza  ,  Myrmecodia  ), we assigned 14.5   6 Myr to thecrown group node of the sister group of Hydnophytinae (Barrab  e et al. , 2014). For  Neonauclea  , we assigned an age of 40   10Myrto the root, corresponding to the crown group of the Cinchonoi-deae (Bremer & Eriksson, 2009). For  Barteria  , we assigned thesplit of   Barteria   and  Passiflora   to 39    10 Myr using the  Passiflora  stem group age (Hearn, 2006).For clades that lack fossils and have not been clock-dated inother studies, we used published substitution rates for calibrationand strict or relaxed clock models following analyses of the extentof rate heterogeneity in Tracer. Because rates can vary greatly andmay correlate with generation time (Kay   et al. , 2006; Smith &Donoghue, 2008), we used three rates for each phylogeny,spanning the range of plausible rates. For  Leonardoxa   andthe Miconieae, we used rates of 1 9 10  9 , 2 9 10  9 , or3 9 10  9 substitutions per site per year, representative of ITS inwoody species (Kay   et al. , 2006), with a strict clock model for Leonardoxa   and UCLN relaxed clocks for  Neonauclea   and theMiconieae. For  Clerodendrum  , we used a strict clock model andrates of 1 9 10  9 , 2 9 10  9 , or 3 9 10  9 substitutions per siteper year for both ITS and the  trnL-F   region (Chase  et al. , 1993;Richardson  et al. , 2001; Kay   et al. , 2006). To calibrate the Apo-cynaceae matrix of Wanntorp  et al.  (2006), which consists of twoplastid spacer regions and nuclear ITS, we used a strict clock andrates of 2.5 9 10  9 , 3.5 9 10  9 , or 4.5 9 10  9 substitutions persite per year, consistent with noncoding plastid and ITS substitu-tion rates in other herbaceous perennials (Manen & Natali,1995; Richardson  et al. , 2001; Kay   et al. , 2006) For  Cecropia  , webuilt a combined  trnL-F, rbcL   and  matK    matrix and used a strictclock with substitution rates of 0.8 9 10  9,  1.2 9 10  9 or2 9 10  9 substitutions per site per year, based on rates for theseloci in other woody groups (Chase  et al. , 1993; Richardson  et al. ,2001; Lavin  et al. , 2005). We cross-validated age estimates against those from publishedstudies with overlapping taxon sampling. The trees obtained New Phytologist   (2015)   2015 The Authors New Phytologist   2015 New Phytologist Trust Research New Phytologist 4  from each clock run were summarized with TreeAnnotatorv1.8.0, with a 10% burn-in and showing only nodes  ≥ 0.98 pos-terior probability. Time-calibrated trees are shown in Figs S2  –  S15. Sister-group geographic mapping  We selected 20 sister clade pairs from our 1181-species tree orpublished phylogenies and then downloaded the geographicranges of these closest relatives from the Global Biodiversity Information Facility (GBIF) ( Theclosest relatives were  Acacia cochliacantha  ,  Adenia cynachifolia  ,  Androsiphon adenostegia  ,  Conceveiba pleistemona  ,  Cordia collococa  , Cordia ecalyculata  ,  Cuviera subuliflora  ,  Euphronia guianensis  , Henriettea succosa  ,  Korthalsia jala  ,  Leucosyke australis  ,  Ludekia borneensis  ,  Mallotus brachythyrsus  ,  M. nudiflorus  ,  M. ficifolius  , Macbridenia peruviana  ,  Microsorum linguiforme  ,  Piper aequale  , Psychotria hawaiensis   and  Ruprechtia triflora  . The distributionswere plotted on a world map using DIVA-GIS (Hijmans  et al. ,2005). The mean annual temperatures were downloaded from WorldClim ( Ancestral state reconstructions In order to reconstruct gains and losses of domatia, we scoreddomatium absence (0) and presence (1) for all 1181 species inour tree based on our World myrmecophyte list (Table S1). Ancestral reconstruction relied on maximum likelihood (ML)implemented in Mesquite using the highest scoring likelihoodtree and the Markov two-parameter model (Lewis, 2001), whichallows for different forward and backward change frequencies.Domatium presence in a common ancestor was assumed if theML probability was  ≥ 70%. We added a single gain for genera with domatium-bearing species (Table S1) that were notincluded in our 1181-species matrix. We also inferred the evolution of extrafloral nectaries on our1181-species tree, using the same approach. We scored EFN-bearing species as 1, and EFN-lacking species as 0, according tothe World List of plants with extrafloral nectaries (Keeler, 2008). We also mapped EFNs onto the  Macaranga   and the  Vachellia  chronograms. In addition to the ML approach implemented inMesquite, we inferred ancestral states (both for EFNs anddomatia) in  Vachellia   and  Macaranga   using the Bayesian revers-ible-jump MCMC approach for discrete characters implementedin BayesTraits (Pagel & Meade, 2007) on a sample of 2000 treesfrom BEAST (burn-in excluded), thereby taking into accounttopological uncertainty. The chain was run for 50 9 10 6 genera-tions, and rate coefficients and ancestral states were sampled every 1000 th generation. We ensured that the acceptance rate wasbetween 20% and 40% as recommended in the manual. Results Frequency and geography of ant  –  plant symbioses, and thedistribution of domatium types and growth forms Our world list of myrmecophytes includes 681 species in 159genera and 50 families (Table S1, which also provides informa-tion on geographic ranges). Our modelling approach to estimatethe true number of myrmecophytes (including ones not yet docu-mented or missed in our literature search) yielded 1139 speciesunder the best-fit model (1-exponential mixed Poisson, Table 1). Ant  –  plant symbioses are almost exclusively tropical. Excep-tions are species of   Vachellia   ranging into South Texas and Afri-can  Vachellia drepanolobium   south of the Tropic of Capricorn.There are strong diversity asymmetries in absolute species num-bers, with overall  c.  7 times more ant-plant species than plant-antspecies. This asymmetry is present in all three biogeographicregions (Fig.2) and may be strongest in Australasia, although thatmight be an artefact of the lack of taxonomic knowledge of Aus-tralasian ants and cryptic species complexes (personal communi-cations from M. Janda, Czech Academy of Sciences, October2013, and V. Witte, University of Munich, June 2014). Closestrelatives of ant-plant clades for which we could evaluate geo-graphic ranges were all distributed in the tropics and absent fromtemperate regions (Fig. 2).Domatium-bearing plants are present in one family of ferns,absent in gymnosperms, and generally widespread in angio-sperms, although they are absent in basal eudicots. The highereudicots, however, contain the majority of myrmecophytes, withRubiaceae having the highest number (162 species, Table S1),followed by Melastomataceae (144 species, Table S1). The ances-tral reconstruction implies 158 independent srcins and 43 lossesof domatia (Fig.3). In some genera, such as  Cecropia  ,  Dischidia  , Table1  Predicted total number of ant domatium-bearing species from model estimatesModel TauEstimated totalspecies SE Lower CB Upper CBBest model 1 exponential mixed Poisson 8 1139 40 1067 1224Model 2a Poisson 5 805 16 778 840Model 2b 2 exponential mixed Poisson 10 1160 72.8 1037 1325Model 2c 2 exponential mixed Poisson 40 1159 49.5 1072 1267Non-P1 Chao1 2 842 28.4 795 908Non-P2 ACE1 10 1070 59.6 969 1205Tau is the upper frequency cut-off; SE, the standard error of the estimate; Lower and Upper CB, the 95% confidence bound. The best model (first line) isfollowed by the three next best-fit parametric models (Models 2a  –  c) and the two best-fitting nonparametric models (non-P1, P2).  2015 The Authors New Phytologist   2015 New Phytologist Trust New Phytologist   (2015) New Phytologist  Research  5
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!