What defines an adaptive radiation? Macroevolutionary diversification dynamics of an exceptionally species-rich continental lizard radiation – BMC Ecology and Evolution

Phylogenetic tree

Our analyses are based on a multi-gene molecular, time-calibrated phylogenetic tree ( Fig. 1 ), including 109 of the ~240 known Liolaemus species ( the total number of species is unmanageable to determine given taxonomic controversies and the miss of dependable diagnoses for respective species ), extracted from Pyron et al. ’ s [ 37 ] comprehensive tree of squamates. The tree was time-calibrated using holocene estimates obtained from molecular phylogenies of the major clades within Liolaemus [ 26 ], and based on the genus ’ fossil record [ 38 – 40 ]. We set the lineage of the Liolaemus crown group radiation ( beginning with the latest coarse ancestry between the subgenus Eulaemus and Liolaemus sensu stricto ) at 19.25 million years ago ( Mya ). This time represents the average between paleontological and molecular estimates, which place the origin of the crown group radiation, respectively, at 18.5 and 20 Mya .Fig. 1figure 1 phylogenetic relationships within the Liolaemus radiation sickness showing version in body size ( snout-vent length obtained by averaging male and female SVLs ) across species ( blacken bars, in millimeter ). Clade color indicate the eight independent groups ( or subgenus ) within the genus Full size picture

Analyses of lineage diversification

Analyses based on the time-calibrated phylogenetic tree were performed to quantify the evolutionary tempo and modality of diversification in Liolaemus, with focus on both linage and body size diverseness. To quantify historic rates of species accumulations ( i.e., tests of the prediction that diversification has slowed down over fourth dimension following an early burst ) we created a linage through-time ( LTT ) plot implemented in the R software ‘ ape ’ [ 41 ]. For the LTT curve, we first gear implemented Pybus & Harvey ’ s [ 42 ] Monte Carlo Constant Rate ( MCCR ) quiz. This psychoanalysis calculates the γ statistic for incompletely sampled phylogenies, by comparing the distribution of inter-node distances between the tree root and its temporal role center to the distribution of distances between the temporal center and the tree tips [ 43 ]. negative γ values indicate that inter-node distances between the root and the center are shorter than the distances between the center and the tips, and therefore, that most branching events occurred earlier in the evolutionary history of the clade, a design reproducible with a decline in the rate of species accretion over fourth dimension ( i.e., an ‘ early break ’ model of diversification ). When descent diversification follows a changeless rate process, the ramify events are evenly distributed throughout the tree, with γ being normally distributed and with a mean of 0. Given that incomplete taxonomic group sampling in a evolution increases type I error rates in diversification analyses, the MCCR test computes corrected γ distributions through simulations of phylogenies to the known clade size ( ~240 species in Liolaemus ) under the nothing hypothesis of a ceaseless rate pure-birth diversification process. Species are then randomly pruned from the simulate trees to replicate incomplete sampling ( 109 species are included in our corner ; see above ). Our analysis is based on 10,000 Monte Carlo simulations. The MCCR test was conducted using the ‘ laser ’ package in R [ 44 ].

We then analysed the diversification dynamics that are more likely to have shaped the LTT tendency of Liolaemus species accretion by fitting multiple evolutionary models that rely on different evolutionary processes. We used Etienne et al. ’ sulfur [ 45 ] maximum-likelihood fitting-model approach to test four alternative hypotheses of diversification. This proficiency employs a shroud Markov model ( HMM ) approach to calculate the likelihood of a phylogenetic history under multiple diversity-dependent birth-death models of diversification. These models account for the influence that species early than those included in the evolution ( i, both extinct species and species missing from the evolution ) may have on historic rates of diversification ( given that speciation rates are a function of existing species at each point in time ). consequently, this set about is comparable to the results produced by the MCCR test above as both techniques consider the potential effects of species missing from the tree [ 43 ]. Two of the four fitted models assume changeless diversification rates. These are the pure-Birth ( or Yule ) model, which assumes no extinctions, and the constant rate birth-death model ( crBD ), which allows extinctions but assumes that the rates of speciation and extinction remain constant through prison term and across lineages. The other two models, density-dependent logistic ( DDL + E ) and density-dependent exponential ( DDE + E ), assume diversity-dependence and thus quantify diversification rates as functions of changes in accumulating diversity over clock ( while accounting for extinctions, E ). While the DDL + E models linear pace changes in diversification, the DDE + E models exponentially declining speciation rates as a function of extant lineage diverseness at any point in meter. We fitted all four models under two alternative assumptions about the symmetry of missing species in the evolution. First, we assumed that the Liolaemus clade consists of its presently known 240 species. We then assumed that the genus consists of many more species than those presently reported, and that our evolution only accounts for 30 % of the ‘ real ’ diverseness of the linage. For both scenarios, we fitted all four models using the R box ‘ DDD ’ [ 45 ]. To evaluate the best-fit model, we employed the Akaike Information Criterion ( AIC ) approach [ 46 ]. We report the bias-corrected version of AIC, referred to as AICc [ 47, 48 ]. The good of meet of campaigner evolutionary models is determined by identifying the lowest AICc scores, and hence, when shown as ΔAICc scores ( the remainder between the best or lowest AICc, and the AICc of each option model ), then the best model has ΔAICc = 0 [ 47, 48 ].

Body size data

To evaluate the potential relationship between clade diversification and phenotypical development during the radiation of Liolaemus, we investigated the rates and trajectories of body size diversification. We focus on body size as it is the single most authoritative morphologic trait that influences the majority of ecological and evolutionary processes via its correlation coefficient with most components of organismal phase and affair [ 49, 50 ]. In addition, body size is much considered to be a key geomorphologic indicator of recess in lifelike populations [ 49, 51 ]. besides, in Liolaemus in particular, body size is ideally suited for diversification analyses as existing evidence suggests that its mutant is not predictably influenced by geographic/climatic clines [ 28, 30, 34 ], it varies with numbers of coexisting species ( Pincheira-Donoso, unpublished observation ), and other phenotypical traits observed to respond to ecological pressures in early lineages ( for example, consistency proportions, [ 1 ] ) vary in rather irregular ways when linked to, for example, habitat characteristics [ 30, 52, 53 ]. We used snout-vent distance ( SVL ), the traditional proxy for soundbox size in lizards [ 54 – 56 ]. For the analyses, we collated an extensive soundbox size dataset ( Additional file 1 ) consist of 6,500+ adult individuals ( adulthood was estimated based on body sizes reported in previous studies, [ 30 – 32, 34 ] ), representing > 85 % of the presently known species diversity within the genus. To obtain SVL for each species, we averaged male and female SVL values, calculated independently using the upper two-thirds of the size range available for each sex in each species [ 30, 57 ]. Although utmost SVL has been extensively used as a proxy for size in lizards, it has been shown that the use of extreme values may result in soundbox size overestimations [ 58 ]. In contrast, the use of intercede percentiles between the utmost recorded value and the mean from the entire adult sample distribution provides accurate estimates of asymptotic size [ 58 ]. The integral dataset was collected by the same person ( DPD ) to control for error arising from inter-individual measurements ( for example, [ 57 ] ). The species included in our dataset embrace the entire phylogenetic, phenotypical, ecological, and geographic diverseness known in Liolaemus [ 30, 52, 53 ], and consequently, they provide an adequate sample of the consistency size diversity in this genus ( Fig. 1 ).

Modelling body size evolution

We investigated the evolutionary dynamics of body size throughout the phylogenetic history of Liolaemus using two quantitative approaches based on our time-calibrated evolution. First, we quantified the tempo and modality of body size diversification by fitting four option models that identify different evolutionary dynamics : the Brownian-motion model ( BM, which describes a random walk of trait evolution along branches in the evolution, with increase in trait variance centered around the initial rate at the root of the tree, and increasing with the distance from the tree root ; [ 59 ] ), the Ornstein-Uhlenbeck model ( OU, which assumes that once traits have adaptively evolved, stabilizing selection pulls the trait values around an adaptive optimum for the trait ; [ 60 ] ), the Early-Burst or “ niche-filling ” model ( EB, which describes exponentially increasing or decreasing rates of development over time based on the assumption that niches are saturated by accumulating species within a linage ; [ 8 ] ), and the Delta model ( a time-dependent model of trait evolution, which describes the effects that early versus late evolution in the tree have on the rates of trait evolution ; it returns a δ value which indicates whether recent evolution has been fast when δ > 1, or slow when δ < 1 ; [ 61 ] ). Comparisons of good of fit for these models were performed through the Akaike Information Criterion ( AIC ) [ 46 ]. survival of the best evolutionary model is based on the same AICc approach described above for model-selection of linage accumulation. Model implementation and match was conducted with the R box ‘ geiger ’ [ 62 ]. We then investigated whether the distribution of body size in Liolaemus has evolved around a given number of SVL optima ( i.e., whether stabilizing survival has promoted macroevolutionary convergences of the trait against one or more such peaks ), using the ‘ surface ’ package in R [ 63, 64 ]. This surface method acting fits an adaptive radiation model in which lineages on a evolution may experience convergent shifts towards adaptive optimum on a macroevolutionary Simpsonian landscape, importantly, without assumptions of whether some lineages correspond to particular optimum [ 63, 64 ]. Based on an OU exemplar [ 60 ] in which all species are pulled against a one adaptive optimum in morphospace, SURFACE employs a bit-by-bit model survival approach path based on AICc, which allows for identification of the best model and the numbers and positions of adaptive peaks ( i.e., trait ‘ regimes ’ ), and hence, for convergence towards these optima over evolutionary fourth dimension [ 63, 64 ]. We then modelled torso size disparity through time ( DTT ). Based on size data from extant species ( see above ), this set about calculates the mean disparity for the trait over time, and compares the observe body size disparity with that expected under a nothing model of Brownian-motion by simulating soundbox size evolution 10,000 times across the tree [ 12 ]. then, the modal consistency size disparity obtained from the actual and the simulated data are plotted against node long time to calculate the morphologic disparity index ( MDI ). This index quantifies the overall remainder in relative disparity for the trait among and within subclades ( i.e., differences in the crop of variation ) compared with the expectation under the nothing Brownian gesture mannequin [ 13, 62, 65 ]. damaging MDI values indicate lower than expected trait proportional disparity under Brownian motion ( i.e., broken average subclade relative disparity ), which in turn indicates that most disparity occurs among subclades, and therefore, that they occupy smaller and more isolate areas of the morphospace [ 12 ]. In line, incontrovertible MDI values indicate that relative disparity among subclades shows stronger overlap in morphospace occupation [ 12 ]. Trait disparity analyses were conducted using the R box ‘ geiger ’ [ 62 ]. The plot projecting the Liolaemus evolution onto the body size morphospace ( against time since the root ), based on ancestral node estimations using maximum likelihood [ 66 ] is shown in Fig. 3 ( see legend for details ), and was built using the R software ‘ phytools ’ [ 67 ]. We ultimately investigated the influence of body size on macroevolutionary linage diversification in Liolaemus. We employed the phylogenetic likelihood-based approach Quantitative State Speciation and Extinction ( QuaSSE ) implemented in the R package ‘ diversitree ’ [ 68 ]. This method fits evolutionary models based on the distribution of extant characters ( body size ) on a evolution, under the presumption that diversification follows a birth-death action and that a species can be characterized by its average value of the quantify trait, which affects diversification through its effect on the speciation-extinction rates ( where rate of speciation is λ, and the pace of extinction is μ, see [ 69 ] ). evolutionary models are fitted by adding a ‘ drift ’ or ‘ directing ’ parameter ( φ ), which describes the deterministic ( or directing ) component of character development. That is, the expect rate of character change over clock as a routine of survival or other serve which determines a directional tendency [ 68, 70 ]. therefore, this term does not refer to genetic float specifically. After adding the drift condition, the likelihood functions created by QuaSSE trace diversification by a ceaseless, linear, sigmoid, or hump-shaped routine of log body size [ 68 ]. recognition of the best evolutionary model is performed via the AIC approach ( see above ) .

generator : https://thefartiste.com
Category : Tech

About admin

I am the owner of the website thefartiste.com, my purpose is to bring all the most useful information to users.

Check Also


Manage participants in a zoom meeting webinar

Call the people who attend the meet as follows Alternate host host Who scheduled the …

Leave a Reply

Your email address will not be published.