Eulaema meriana, Complexes

Sandoval-Arango, Stephania, Branstetter, Michael G., Cardoso, Carolina F. & López-Uribe, Margarita M., 2023, Phylogenomics reveals within species diversification but incongruence with color phenotypes in widespread orchid bees (Hymenoptera: Apidae: Euglossini), Insect Systematics and Diversity 7 (2), pp. 1-13 : 5

bombiformis and E. meriana Complexes

Our mtDNA dataset included a total of 1,272 bp of which 159 bp were variable. In the UCE dataset, we recovered an average of 41,950,575 bp per individual after the SPAdes assembly. Following the assembly and extraction of UCE contigs, we aligned the data using MAFFT and trimmed the sequences using Gblocks, resulting in 2415 UCE loci with a mean alignment length of 845.93 bp per loci, a total of 49,602 informative sites, and an average of 20.54 informative sites per loci.The UCE data matrices included 2205, 2139, and 2022 loci for the 75, 90, and 100% taxon completeness, respectively. Detailed sequencing and assembly statistics can be found in Supplementary Table S4. We found no differences between the IQ-TREE results of the three completeness matrices (Supplementary Figs. S2 and S 3 ), therefore subsequent analyses were performed on the 100% matrix and datasets.

For both molecular datasets, we recovered several monophyletic lineages that are mostly congruent with geography ( Fig. 2 , Supplementary Figs. S4 and S 5 ), and with previous results based on mitochondrial data only ( López-Uribe et al. 2014). These geographic clades have high support values and only some internal clades show low support (indicated by asterisks). However, we found incongruences between the two datasets, particularly for E. meriana . The mitochondrial Bayesian phylogeny recovered six main clades for the E. meriana complex, while the UCE ML phylogeny from IQ-TREE recovered three main clades for the E. meriana complex ( Fig. 2 ). The main difference in the topology is explained by a split of the individuals corresponding to the Choco and Central American regions in the mitochondrial dataset ( Fig. 2A ). In both cases, E. meriana bees from the Amazon and E. atleticana bees from the Atlantic Forest formed one monophyletic group sister to either a clade that contains all the Central American individuals ( Fig. 2B ) or to both Central American and Choco individuals ( Fig. 2A ). For the E. bombiformis complex results were very similar between the two datasets. Individuals of E. niveofasciata formed a monophyletic group that was either sister to all the E. bombiformis individuals ( Fig. 2A ) or that was clustered in a clade that contains both E. bombiformis and E. niveofasciata bees ( Fig. 2B ).

Our species tree analysis included 50 UCE loci and recovered a topology similar to that of our IQ-TREE analyses with some incongruences in the internal nodes of most clades ( Fig. 3 ). Within the E. meriana complex, there is support for an independent lineage that contains all the individuals from the Choco Region. Although without support, most individuals in the other clades grouped according to their geographic regions, except for one individual of E. meriana from Central America (ML172). The E. atleticana individuals were grouped with individuals from the Amazon Forest similar to our IQ-TREE analysis. In the E. bombiformis complex, the only clade that was supported by most trees was a lineage containing all the E. bombiformis and E. niveofasciata individuals. Clades within that lineage had low support, signaling incongruence between the species trees from independent runs and the gene trees of individual UCE loci.

The results of our divergence dating ( Fig. 4 ) indicate that most diversification within the E. meriana complex occurred during the late Pliocene and early Pleistocene (Mean age: 3.88 Mya, 5–2.5 Mya 95% HDP). Additionally, the MCRA of the chocoan lineage of E. sororia and E. chocoana separated from the E. meriana complex approximately in the late Miocene [Mean age: 7.09 Mya, 9.5–4.5 Mya 95% highest posterior density (HDP)]. Finally, the E. bombiformis complex is likely to be an older lineage (7.5–3 Mya 95% HDP) compared to E. meriana , which radiated at the end of the Miocene and early Pliocene.

Figures

Fig. 2. Phylogenetic relationships of the E. meriana and E. bombiformis species complexes based on (A) mitochondrial data (mtDNA; CO1 and Cytb) and (B) ultraconserved elements (UCE; 2022 loci). The mtDNA phylogeny was estimated using Bayesian inference in BEAST2, posterior probabilities on nodes were all above 0.9 except for nodes with asterisks (*).The UCE phylogeny was estimated with maximum likelihood using IQ-TREE and a concatenated 100% completeness matrix. Support values on nodes indicate ultrafast bootstrap (UFB) and SH-like (SH) approximate likelihood ratio test scores (SH-aLRT). All support values were above 95/95 except for nodes indicated with asterisks (* or **). One of the E. cingulata individuals (TA12) was pruned to improve the cophylogenetic visualization.The colors of the clades correspond to geographic regions outlined in Fig. 1. Green: Central America, Orange: Choco region, Blue: Amazon Forest, and Pink:Atlantic Forest.

Fig.3. Maximum clade credibility (MCC) species tree of the phylogenetic relationships in the E. meriana and E. bombiformis species complexes estimated under the multi-species coalescent model (MSC) using *BEAST in BEAST2.The species tree was estimated using the 50 most informative UCE loci from the 100% completeness dataset.Nodes without labels had posterior probabilities <0.5.The colors of the clades correspond to geographic regions outlined in Fig.1. Green: Central America,Orange: Choco region, Blue: Amazon Forest, and Pink:Atlantic Forest.

Fig. 4. Chronogram of the phylogenetic relationships in the E. meriana and E. bombiformis species complexes estimated using BEAST2 and 500 UCE loci. All nodes had a posterior probability of 1.The arrow indicates the node used for calibration of the tree and acronyms correspond to geographic regions outlined in Fig. 1. CA, Central America; CR,Choco Region;AM,Amazon Forest;AF,Atlantic Forest.In the calibration point, M = mean age, and S = confidence interval.

Fig. 5. Color phenotypes and phyloheatmap of the morphological characters of E. meriana and E. atleticana, visualized using the (A) mtDNA and (B) UCE phylogenies. Circles at the end of individual names indicate the color phenotype of that individual. Names above the phyloheatmaps indicate the character that was measured.BL, Body length; HW, Head width; ID, Intertegular distance;WTII,Width of colored bands on tergum II;WTIII,Width of colored bands on tergum III. In this phyloheatmap, each column of the measured characters was standardized to have the same variance prior to analysis.The scale below indicates how much each value deviates from the mean.The colors of the clades correspond to geographic regions outlined in Fig. 1. Green: Central America, Orange: Choco region, Blue:Amazon Forest, and Pink:Atlantic Forest.