Nuclear genes and plastomes gave two confident, incompatible answers for a stubborn node in Sapindales; three nested chromosomal fusions give a third, and make it stick.
Deep nodes in the plant tree are where sequence data goes to argue. Sapindales holds a textbook case: Biebersteiniaceae, a single genus of Central Asian herbs with an unusually low chromosome count (2n = 10), has been placed in several positions depending on which genes you read. Nuclear single-copy genes say one thing, plastomes say another, and both say it with high support. That is gene tree conflict from rapid ancient divergence and reticulation, not noise to average away. The usual response is to add loci; this paper adds a different kind of character.
An and colleagues assembled a chromosome-level genome for Biebersteinia heterostemon, built a species tree from collinear genes (the subset whose order is conserved across genomes), then reconstructed ancestral karyotypes across the order. The collinear-gene tree recovers a third topology, distinct from both earlier answers: Biebersteiniaceae sister to the Anacardiaceae, Meliaceae, and Rutaceae clade, with Sapindaceae outside that group and Nitrariaceae diverging first. The chromosomal evidence lines up node by node. One fusion is shared by Anacardiaceae, Meliaceae, and Rutaceae; a second is shared by those three plus Biebersteiniaceae; a third unites all of them with Sapindaceae and is absent in Nitrariaceae. Rearrangements are rare enough, and hard enough to reverse, to work as synapomorphies where nucleotides have saturated.