On the Rediscovery of Volvox Perglobator (Volvocales, Chlorophyceae) and the Evolution of Outcrossing from Self-Fertilization
Total Page：16
File Type：pdf, Size：1020Kb
Evolutionary Ecology Research — Volume 19 (2019) APPENDIX On the rediscovery of Volvox perglobator (Volvocales, Chlorophyceae) and the evolution of outcrossing from self-fertilization Erik R. Hanschen1,2,*, Dinah R. Davison2, Patrick J. Ferris2, Richard E. Michod2 1 Division of Bioscience, Los Alamos National Laboratory, Los Alamos, New Mexico 87544 2Department of Ecology and Evolutionary Biology, University of Arizona *Correspondence: [email protected] Morphological tree estimation We accounted for the substantial effect of lost Volvox section Volvox species (Volvox merrillii, Volvox amboensis, Volvox prolificus) on ancestral state reconstruction (Figure S3) by estimating a tree from morphological traits for the seven species with genetic data. If the morphological tree and genetic tree are congruent, morphology may be used to estimate species relationships. We used previously defined morphological traits (Nozaki and Itoh 1994) and traits which serve to morphologically identify species (Isaka et al. 2012), for a total of 39 traits. After removing invariant traits, seven traits remained (Table S3). Other traits (cell number, colony length, etc.) are not included as they are highly variable within species (Isaka et al. 2012; Nozaki et al. 2015). It was not possible to use PartitionFinder (Lanfear et al. 2016) to automatically identify the optimal model of evolution due to an insufficient number of traits; therefore, trees were estimated using both BINCAT and BINGAMMA models. Trees were made with RAxML version 8.0.20 (Stamatakis 2014) using 1,000 bootstraps. Both morphological trees had poor bootstrap support (below 32, Figure S4), and both trees were discordant with the chloroplast and ITS trees (normalized Robinson-Foulds distance, 50 to 75% of nodes different; Robinson & Foulds, 1981). Morphological traits cannot be used to accurately estimate the chloroplast and ITS trees (Figure S4), therefore we cannot reliably use morphological traits to infer the phylogenetic placement of lost Volvox section Volvox species. Furthermore this approach is not feasible as mating system, the trait we reconstruct, is included in this matrix. Chloroplast tree estimation We utilized a recently estimated volvocine chloroplast phylogeny (Hanschen et al. 2018a). Briefly, this phylogeny uses the coding sequences of five chloroplast genes (ATP synthase beta- subunit, atpB; P700 chlorophyll a-apoprotein A1, psaA; P700 chlorophyll a-apoprotein A2, psaB; photosystem II CP43 apoprotein, psbC; and the large subunit of Rubisco, rbcL; 6,021 base pairs total). The ingroup was defined as the smallest clade containing Chlamydomonas -1- reinhardtii and Volvox carteri. The data matrix included sequences for 97 volvocine OTUs and seven outgroup taxa (Table S1). The outgroup consisted of non-volvocine algae, including two taxon from the immediate sister group and one taxa from each of five other major groups (Herron and Michod 2008; Hanschen et al. 2018a). The best partitioning scheme and nucleotide substitution models were determined using PartitionFinder version 2.1.1 (Lanfear et al. 2016) using AICc and a greedy search algorithm with branch lengths linked (Hanschen et al. 2018a). A concatenated phylogeny was generated using Bayesian Markov chain Monte Carlo implemented in MrBayes version 3.2.2 (Ronquist et al. 2012). Four independent Bayesian runs of four chains each (three heated chains and one cold chain) were run for 2.5×107 generations with a burn-in of 5×106 generations. Trees were sampled every 100 generations. We considered the runs to have adequately sampled the solution space when the standard deviation of split frequencies was below 5×10-3. Post burn-in trees were combined and assembled to construct a majority-rule consensus phylogram and calculate posterior probabilities. An ultrametric tree, necessary for ancestral-state reconstruction, was calculated using a penalized likelihood function in the ape R package with a correlated model with no age constraints (Sanderson 2002; Paradis et al. 2004). The chloroplast tree, with 97 volvocine taxa and seven outgroup taxa, was independently constructed using maximum likelihood (ML) methods using RAxML version 8.1.12 (Stamatakis 2014) with the rapid bootstrap analysis and the partition scheme previously identified (Hanschen et al. 2018a) by PartitionFinder (Lanfear et al. 2016), generating 200 ML replicate trees to estimate bootstrap support. Taxa of the same species were previously identified (Hanschen et al. 2018b) using a single rate Poisson Tree Processes (PTP) method (Fujisawa and Barraclough 2013). A maximum likelihood approach with a default p-value of 0.001 was used (Hanschen et al. 2018b). Multiple genetically-unique individuals of the same species, as well as species with unknown mating system (heterothallic outcrossing or homothallic selfing) were trimmed from the final tree (which contained 69 species) using the R package ape (Paradis et al. 2004). -2- Ancestral-state reconstructions Maximum likelihood ancestral states were reconstructed using the R package diversitree version 0.9-9 (FitzJohn 2012). Two models of character evolution were evaluated for the evolution of selfing and outcrossing (Table S2): (1) an equal rates of change for both transitions between states (ER model) and (2) all rates different for both transitions between states (ARD model). Model fit was compared using the AIC (Akaike 1974), corrected for small sample size (Burnham and Anderson 2002), which should reveal the best-fitting model without including unnecessary parameters. The best-fitting model was ARD (ÄAICc = 8.59). Alternative root state models were evaluated by comparing their likelihoods while holding the best-fitting transition model constant. These models included an equal probability for each state, probabilities based on the frequency of each state among species on the tree, or fixed in either selfing/outcrossing state). We found that the ÄAICc values for alternative root state models were indistinguishable (ÄAICc < 8.2×10- 4 for all reconstructions). Therefore, the root prior was weighted based on the observed frequency of each state among taxa across the tree (FitzJohn 2012) following (Hanschen et al. 2018b). The state with the highest probability was considered the most likely for a given node and considered to be significantly supported at a given node only if it was at least 7.39 times (if the natural logarithm of the ratio of two likelihoods is greater than 2) more likely than the alternative state (Pagel 1999). Tree topology in phylogenetic simulations A small proportion (4.46%) of trees reconstructed an alternative evolutionary history with mostly outcrossing ancestors and numerous (5-7) independent origins of selfing in Volvox section Volvox (Figure 3C, upper left black box). In order to better understand these reconstructions, we analyzed 1,000 trees for each method of assigning lost Volvox species to a branch ((1) Equal, (2) Proportional, (3) Inverse Proportional). We extracted all trees with an -3- outcrossing evolutionary history (Equal, 17; Proportional, 13; Inverse Proportional, 86), and an equal number of trees with a selfing evolutionary history (Figure 3C, lower right black box). There were significantly more trees with this outcrossing evolutionary history in the Inverse Proportional sample than the Equal sample (G test of independence, G = 48.247, p = 3.7×10-12) or the Proportional sample (G test of independence, G = 57.706, p = 3.0×10-14). We tested two hypotheses for how different tree topology has resulted in radically different reconstructions: (1) when outcrossing V. prolificus is sister to outcrossing V. rousseletii or outcrossing V. perglobator, a selfing evolutionary history is inferred (and when V. prolificus is not sister to V. rousseletii or V. perglobator, an outcrossing history with numerous origins of selfing is preferred over three independent reversions), and (2) a tree topology with more recent speciation events results in an outcrossing evolutionary history. The frequency with which V. prolificus is sister to another outcrossing species is not significantly different in outcrossing and selfing evolutionary histories (G test of independence, G = 2.755, p = 0.097). Therefore, the topology of V. prolificus relative to other outcrossing species does not differentiate these evolutionary histories. Alternatively, the average branch length of the three lost Volvox species strongly correlates with outcrossing or selfing evolutionary history (one sided Mann-Whitney U test, W = 16300, p = 2.2×10-16, Mann and Whitney 1947). More recent speciation events strongly correlate with an outcrossing ancestor with numerous origins of selfing. This is consistent with the observation of more outcrossing evolutionary histories in the Inverse Proportional sample, which preferentially added lost Volvox species to short branches (which seem to be more recent, Figure 2). If speciation events are extremely recent, then the inferred reverse transition rate must be extremely high, which is discordant with the rest of the volvocine tree (Figure S2), resulting in an alternative reconstruction with numerous (5-7) independent origins of selfing. A small number of simulations (4.46%) predicted a largely outcrossing ancestry in Volvox section Volvox (Figure 3C, upper left black box). However, these reconstructions are the consequence of extremely recent simulated speciation events within Volvox section Volvox. The -4- disparate geographical distribution of outcrossing Volvox section Volvox species (V.