Supplementary Figure S1. --Phylograms of L1 sequences with open reading frames covering the degenerate PCR region from A) the megabat Pteropus vampyrus and B) the microbat Myotis lucifugus. The 54 P. vampyrus elements were obtained from the group of 69 sequences described in Table 2 after removal of identical sequences. The 54 unique M. lucifugus elements were randomly chosen from the analogous group of 919 M. lucifugus sequences described in Table 2. Sequences were aligned using ClustalW, then adjusted by hand. Paup*4.0b10 was used to infer neighbor joining trees based on nucleotide sequences using distances calculated with the F84 model of sequence evolution. Terminal branch length for many of the M. lucifugus elements is zero because they are extremely closely related and negative terminal branch lengths generated by the neighbor joining algorithm were set to zero. Ncra002b.seq is a recently inserted element from Notiosorex crawfordi. MegabatLin1 and MegabatLin2 are the consensus sequences for the two Megabat lineages. Other labels are accession numbers for trace files in the NCBI trace file database (
The two large polytomies of P. vampyruselements are consistent with L1 extinction in a group with two active L1 lineages. The larger clade is referred to as lineage 1 in the text and the smaller clade as lineage 2. In both clades there wereclusters of two to four very closely related elements at the tips of long branches. We examined each cluster more closely to see if it arose from recent transposition or whether these clusters represent multiple reads (with error)of the same sequence in a database with 2x coverage, alleles at the same locus, or sequences recently duplicated by retrotransposition or other mechanisms.
The eight clusters in lineage 1 have three to six changes at conserved amino acid positions that are shared by all members of the cluster, and four of the eight clusters have shared insertions or deletions of one to five codons that maintain the reading frame. There are three clusters in lineage 2. Two of these have three changes at conserved amino acid positions that are shared by both members of the cluster, and one has a single shared change at a conserved site; this later cluster, consisting of Pv1372818202 and Pv1381847691, is considered in more detail below. Elements within each cluster are likely to be either closely related alleles or multiple reads of the same sequence with the few differences being due to sequencing errors that exist in the trace file archive. These closely related elements might have also arisen by unequal crossing over or gene conversion. Thus, each cluster of closely related P. vampyrus sequences appears to have been derived from a single element by processes other than normal L1 retrotransposition.
Any of theP. vampyrusL1 sequences with open reading frames could have arisen recently or could be older elements that have not yet accumulated a stop codon in this region. Although there is no lineage in the phylogenetic tree with the pectinate structure typical of an active lineage, we examined each sequence for an evidence of recent activity by counting indels and changes at conserved amino acid positions. One sequence that belonged to the lineage 2 cluster described above had a single such change, while the remainder had two or more. The five sequences with fewer than three such changes were used in a BLAST search of the P. vampyrus trace filesof sequences to detect sequences with at least 92% similarity to the target sequence. An active lineage would have been evident by presence of sequences that were of intermediate divergence (98 to 95%) from an ORF-containing L1. None of the sequences detected recently active L1 lineages. Pv1384497679 detected one other identical sequence and no others within 92% similarity; it has 93% similarity to the consensus of lineage 1. Thus, this sequence does not represent a recently active L1 lineage. Pv1321490918 did not detect any other sequences that were more similar to it than they are to the consensus sequence of lineage 1 (95%). Pv1360620301 detected two sequences that were 99 to 100% similar to it and each other, but did not detect any other sequences that were more similar to it than they are to the consensus sequence of lineage 1 (94%). Thus,Pv1384497679, Pv1321490918, and Pv1360620301 are inactive members of lineage 1 and do not represent a recently active L1 lineage. Pv1372818202 and Pv1381847691 were 99 to 100% similar to each other and to two other sequences, but did not detect any other sequences that were more similar to them than they are to the consensus sequence of linage 2 (94%). Thus, these sequences are inactive members of lineage 2 and do not represent a recently active L1 lineage.