The Engineer That Britain Killed Twice

ABy the time the last British beaver was killed, some time in the sixteenth century, the animal had been hunted for two things: a dense pelt, and castoreum, a secretion of the scent glands that apothecaries sold against headache. Its return has been slower than its removal. A licensed trial began at Knapdale in Argyll in 2009; an unlicensed population on the River Otter in Devon was documented from 2013 and eventually allowed to remain; and in 2022 the species acquired legal protection in England. Beavers do not merely occupy a stream, they rebuild it. A single family may raise between six and fourteen dams along eight hundred metres of channel, impounding as much as a thousand cubic metres of water where none stood before. Much of the modern case for reintroduction rests on what that held water does to a flood. Whether the effect recorded in the first trial catchments belongs to the animals or to the streams they happened to be released into has divided the field ever since.

BMarguerite Lowther, a fluvial hydrologist with the Upper Tay Catchment Programme in Perth, holds that the mechanism is genuine and that policy has squandered it. Between 2011 and 2023 her group gauged seventy-eight headwater reaches, thirty-nine of them dammed and thirty-nine paired with those for gradient, geology and land use. Across three hundred and forty storms the dammed reaches passed their peak flow a median of twenty-seven per cent lower, and roughly ninety minutes later, than their partners. Over the same years the length of river in the region covered by release licences roughly tripled, an expansion that reads in agency reports as ambition. What did not change was the share of licensed river capable of holding a dam at all. Two-thirds of the licensed length ran through channels whose banks stood too high and whose gradient was too slight for any dam to hold, because a stretch of river that nobody farms is a stretch a minister can licence without argument.

CThe account makes a prediction, and Lowther has tested it. If impounded water sets the limit, attenuation ought to scale with the number of dams. At nineteen reaches carrying more than eight dams the peak reduction reached a median of forty-one per cent, against twelve per cent where only one or two stood. The gain was confined, however, to storms delivering less than thirty-five millimetres of rain in a day. Above that threshold the ponds were already brimming when the rain arrived, and the dammed reaches behaved almost exactly like the undammed ones. The asymmetry is inconvenient, since the events that empty a town are precisely the ones the ponds cannot absorb. Lowther therefore presents what her instruments recorded as a property of ordinary winters rather than of the storms that reach the news, and she ranks the flood argument as the weakest of the reasons for putting the animal back.

DPiotr Halvorsen, who studies channel morphology at the Institute for Lowland Rivers in Toruń, replies that a dam on the wrong channel is only a pond in a trench. Two centuries of drainage and dredging have left many lowland rivers running well below the fields they once flooded, and water held back inside such a trench merely climbs the walls of it. At forty-one lowland sites his group mapped the ground actually put under water. Where the banks stood more than 1.6 metres above the summer channel, a dam inundated a median of 0.3 hectares of floodplain, against 5.1 hectares where the banks stood lower. Bank height of that kind now enters several licensing models, although it was treated as a detail when the mapping began. He concedes two things. All forty-one sites lie within a single basin on glacial till, and the bank heights predating drainage were reconstructed from nineteenth-century estate maps rather than measured.

EThe argument left the seminar room some time ago. Since 2023 the licensing rules in England have obliged an applicant to show that a proposed release site can spill safely onto land, instead of leaving objectors to prove that it cannot, and applications unable to supply that evidence have been refused. Lowther points out, awkwardly for her own position, that an incised channel takes no notice of a licence boundary, so the planning she argues for cannot reach the condition that decides the outcome. Halvorsen answers that this appetite for problems which can be drawn on a map is exactly what keeps incision out of the rules, since a trench offers no line an inspector can walk. Both regret one habit above all others: that the field goes on reporting kilometres of river recolonised rather than hectares of floodplain reconnected, a figure that flatters everyone who quotes it.

FA settlement of sorts is emerging in which the two accounts are divided between channel types rather than ranked against one another. On unincised headwater streams, where the water can leave the channel as soon as it is held, dam density governs what a catchment gains and the licence lines are worth arguing over; on the deep lowland channels, legal protection secures the animals' presence without buying the reconnection that would alter anything downstream. The unresolved question is whether a floodplain wetted again after two centuries comes to carry the same wet-woodland plants and invertebrates as one never drained, and no monitoring series yet runs long enough to answer it. What both researchers reject is the assumption behind the earliest release targets, namely that an animal restored to a map has been restored to the system that once contained it.