Mammals on the EDGEConservation Priorities Based on Threat and Phylogeny

Imagine standing in front of the Tree of Life with a chainsaw you don't want to use. You know cuts are coming — global extinctions are running up to a thousand times faster than the background rate — and you don't have the resources to save everything. Which branches do you protect first? That's the uncomfortable, practical question Isaac, Turvey, Collen, Waterman, and Baillie set out to answer. Their answer starts with a simple but powerful idea: don't just count species, count the evolutionary history they carry, and then factor in how close they are to vanishing. They call the first piece Evolutionary Distinctiveness, or ED. Think of ED as how much unique history rides on a species' shoulders. You trace the path from that species back to the root of its family tree and on each branch, you split the branch's length among all the species that share it. Add those shares up. That sum is ED. In plainer terms: long, lonely branches count a lot; short, crowded ones count a little. They demonstrate this with a seven-species toy tree: one species that sits on a couple of shared branches plus a short tip ends up with an ED of about 2.23 million years. That example makes the point: ED is not just "how long is your terminal twig?" It's also the heft of the deeper limbs you share, divided among your cousins. They implement this on a near-complete, dated mammal supertree. We're talking four thousand five hundred and ten of the four thousand five hundred and forty-eight extant species in Wilson and Reeder, so about ninety-nine percent coverage. The tree is ultrametric, which means all tips end at the present and branch lengths measure time, so ED ends up in millions of years. Real trees have messiness baked in, and they tackle it head-on. Missing species? They reassign their weight to the closest relatives, so if one hedgehog species is omitted, its five congeners are each treated as carrying six-fifths of a species' share. Polytomies — those places where the tree doesn't resolve into clean bifurcations — tend to inflate ED because big unresolved nodes make subtending branches look more distinct than they are. To correct that, they simulate a five thousand species tree under a constant-rate birth and death process — speciation at zero point one, extinction at zero point zero eight — and measure how ED should scale with the number of descendants. The upshot is a simple adjustment: expected ED on a branch of length L scales like L times a factor that declines with how many lineages share it, roughly L times one point zero eight one minus zero point two six seven times the natural log of the number of descendants. For tiny clades of three, the mean ED is about eighty-one percent of the clade's age. They also lean into uncertainty in node ages by computing ED three times — using the mean, upper, and lower ninety-five percent confidence limits for branch lengths — and then taking the geometric mean across those runs. That down-weights scores hinged on very wobbly dates. So what does the mammal tree look like through this lens? ED spans over six orders of magnitude, from about zero point zero five eight million years in some murid rodents to ninety-seven point six million years in the duck-billed platypus, Ornithorhynchus anatinus. The distribution is close to log-normal, with a median around seven point eight six million years and a geometric mean near six point two eight million years. Most species' ED is dominated by the branches near the tips — in practice, much of the signal comes from those last few splits — but it's not as simple as "long terminal branch equals high ED." They point to the three-toed sloths Bradypus tridactylus and Bradypus variegatus, whose total ED is about twenty point four million years even though their most recent common ancestor is only around one million years old. Deep, shared branches can carry a lot of history, and ED faithfully splits that among the inheritors. One question conservationists worry about is comparability: can you take ED scores from different groups and line them up without biasing toward giant clades? Isaac and colleagues probe that by taking Critically Endangered species from across the mammal orders and watching how each species' cumulative ED grows as you expand the clade from the tip toward the root. The punchline is reassuring. In clades of roughly sixty species, you've already captured about eighty-eight percent of the ED you'd get from the full tree. Past about one hundred eighty species, you gain virtually nothing. That leads to a practical rule of thumb: trees with a minimum of around one hundred species yield ED scores you can safely compare across groups. The rank order of ED barely budges once you clear that threshold. Does ED line up with extinction risk? A little, but not in a way you can lean on. Least Concern species tend to have lower ED than the rest — that difference is statistically solid, with a big sample consisting of an F statistic with one and four thousand one hundred and eighty degrees of freedom and a very small p-value. But ED explains only a sliver of the variation in threat status, with an overall r squared around zero point six tenths of a percent. That's actually the point. Distinctiveness and danger aren't the same axis. So you need a way to put them on the same map. That's the second piece: EDGE, for Evolutionarily Distinct and Globally Endangered. Take ED, and multiply it by an extinction risk weight that jumps as you climb the Red List: Least Concern gets zero, Near Threatened or Conservation Dependent gets one, Vulnerable gets two, Endangered gets three, Critically Endangered gets four. Each step is treated as a doubling of risk. Mathematically, the final EDGE score behaves like a natural logarithm of the expected loss of evolutionary history for that species. If you imagine rolling the dice on the next few decades, EDGE is asking: how much history do we expect to lose if this one goes? For clarity, they leave out edge cases — species listed as Extinct in the Wild, domesticated populations of threatened taxa, and a few dozen taxonomically dubious forms — to keep the scale clear. Now the map comes into focus. Across mammals, EDGE scores range from about zero point zero five seven up to six point four eight, with a mean around two point six three. The top one hundred species aren't just a parade of charismatic megafauna. They span eighteen orders and fifty-two families, with deep representation from the under-loved corners of the mammal world: rodents, shrews and moles, bats, lagomorphs. That breadth is the point. Evolutionary history is concentrated in some very non-charismatic places. And here's the gut check. Many of these high-EDGE species aren't on anyone's conservation to-do list. In their tally, forty-two of the top one hundred had no species-specific actions proposed in plans or the literature. Protected areas? A logistic regression makes the mismatch painfully clear: species that don't occur in protected areas tend to have higher EDGE than those that do, with a strong chi squared signal. In plain English, the irreplaceable and the imperiled are often slipping through the cracks. Lipotes vexillifer, the Yangtze River dolphin, tops the EDGE list at six point four eight — and at the time may have had fewer than thirteen surviving individuals. That's what EDGE exposes: places where massive evolutionary stakes are hanging by a thread, and the thread isn't even labeled. The authors don't just rank; they stress-test the method. One big worry is taxonomic churn. If you split a widespread species into several narrowly defined ones, do the scores go haywire? Looking at primates, ED under a biological species concept versus a phylogenetic species concept correlates strongly, with an r squared of about zero point six five on a log and log scale. There are, however, meaningful changes for particular lineages. Phaner furcifer, a fork-marked lemur, sees its ED drop from thirty-eight point three three to ten point four five when it's split into four species. EDGE is honest about what that means: scores reflect how we carve up the tree. The good news is that high-ranking species generally remain high-ranking when taxonomy shifts, which is exactly the kind of robustness you want in a prioritization tool. They also show how to keep ED from being fooled by tree imperfections. Polytomies get that scaling correction derived from simulated trees — the one that shrinks ED as the number of descendant branches grows. Missing taxa get smeared onto their presumed nearest neighbors; in the hedgehog example, treating each of five species as six-fifths of a species compensates for one absent cousin. And wobbly dates get tamed by taking geometric means across three time calibrations. None of this is hand-wavy. It's a set of practical patches grounded in how ED behaves under a simple diversification model, and in empirical checks across thousands of clades. Two more realities shape how you use EDGE. First, data deserts. Roughly three hundred mammals are listed as Data Deficient, which means you can't score them at all — they might be high-EDGE and invisible. About fifty extant species were missing from the supertree, too, though the reallocation step softens that blow. Second, the method is designed to slot into, not replace, the way conservation decisions actually get made. Isaac and colleagues are careful on this point. EDGE flags irreplaceability plus urgency. It does not tell you what it costs to act, whether action is feasible, or how social and political constraints bend the field. Those layers sit on top. And yet, when you look at the list, the signal is unmistakable. EDGE pulls hidden lineages into the light. It shows that the biggest chunks of mammalian history at risk aren't always the species you've seen on billboards. It also reveals cases where updating taxonomy changes who rises to the top. Pontoporia blainvillei, the La Plata River dolphin, for example, emerges as a potential high EDGE species after new congeners are described; its ED clocks in at about thirty-six point three million years. That kind of shift matters because it updates the stakes: who's truly alone on a long branch, and how close are they to the edge? If you zoom out, the through-line is this. Evolutionary Distinctiveness captures the depth of the story a species carries. It's stable across big phylogenies, resilient to reasonable taxonomic changes, and calculated in a way that doesn't get tricked by missing data or unresolved splits. Pairing ED with extinction risk turns that story into a priority list that respects both irreplaceability and urgency. The hard numbers — the range from tenths of a million years to almost a hundred million, the eighty-eight percent of ED recovered in moderately sized clades, the forty-two out of a hundred with no actions on the books — give the method teeth. So where does that leave us? With a clearer compass. Use EDGE to find the species whose loss would fell whole, ancient branches, especially the unglamorous ones we're not watching. Then bring in the rest of the playbook — threat abatement, feasibility, and the realities on the ground — to turn flags into action. Isaac, Turvey, Collen, Waterman, and Baillie didn't write a silver bullet. They wrote a better map. And in a world where every cut takes centuries of history with it, a better map is the start of better choices.