How Canals, Walls and Fences Split Animal Populations Without Killing One

A tiger can survive a bullet wound. What it cannot always survive is a canal it will not cross. Linear infrastructure, canals, highway medians, border fences, railway embankments, irrigation channels, does not kill animals in the way a snare or a speeding truck does. It works more slowly, and the damage shows up in a spreadsheet of genetics long before it shows up in a body count.

Individual animals often survive the construction of a canal or a fence. The population is what fractures. When two groups of the same species can no longer interbreed, each group begins to lose genetic diversity through a process called genetic drift, random changes in allele frequency that accumulate faster in small, isolated groups than in large connected ones. A 2007 study in the journal Molecular Ecology tracking tigers across central India found measurably reduced gene flow between populations separated by roads and canals, even where the animals themselves were not being killed by traffic. The mechanism is not mortality. It is the slow arithmetic of a closed gene pool.

The barrier does not need to be physically impassable. A canal three metres wide is swimmable for most large mammals. A chain-link fence can be dug under. What stops movement is the combination of the structure with everything around it: the service road running alongside the canal, the human activity that concentrates along any linear structure, the light and noise that make the corridor feel exposed. Elephants in the Nilgiris have been documented avoiding culverts under railway lines not because they cannot fit through them but because the culvert exits onto an open embankment with no tree cover. The barrier is architectural, not physical. For smaller animals, rodents, reptiles, amphibians, even a tarmac road surface raised a few centimetres above grade can function as a wall. A study tracking the Indian spiny-tailed lizard found that individuals almost never crossed metalled roads, despite being capable of covering that distance in seconds on open ground.

Once a population is isolated, the consequences compound. The first generation loses access to mates from outside the fragment. The second generation begins to express recessive genetic disorders that a larger, more diverse population would have diluted. Immune response narrows. Reproductive success drops. Litter sizes fall. Cub survival rates decline. The population looks stable on a headcount for years while the genetic architecture underneath it is collapsing. This is what conservation biologists call an extinction debt: the population is functionally doomed before the numbers reflect it. The lion population of the Gir Forest in Gujarat is the most studied example in India. Isolated from all other lion populations for over a century, Gir lions carry extremely low genetic diversity. A single disease event, canine distemper reached the population in 2018 and killed over two dozen animals, can move through a genetically uniform group far faster than it would through a diverse one.

The solutions are engineering problems, not conservation philosophy. Wildlife crossings, overpasses wide enough to carry vegetation, underpasses with natural substrate, culverts designed to the shoulder height of the target species, restore connectivity without removing the infrastructure. The effectiveness depends on placement and design. A crossing positioned where animals already move, identified through camera trap data and track surveys, works. One placed where it is convenient for the contractor does not. India's National Highway Authority has built underpasses along NH-44 through the Pench-Kanha corridor, and camera trap data has confirmed tiger and leopard use of several of them. The Bangalore-Mysore highway expansion included mammal underpasses after pushback from wildlife biologists who mapped movement corridors before construction began. These are not expensive relative to the total project cost. A wildlife underpass on a national highway typically adds less than one percent to the per-kilometre construction budget. The barrier to building them is not money. It is the absence of a legal requirement to conduct a connectivity assessment before a linear project is approved.

Fences can be gapped. Highways can have underpasses retrofitted. Canals are harder to fix after the fact because they carry water under pressure or gravity and cannot simply be bridged at arbitrary points. The canal networks of Punjab, Haryana, and the Krishna delta have fragmented blackbuck and nilgai populations across agricultural plains where these animals once moved freely. Retrofitting crossing points requires coordination between irrigation departments and forest departments that rarely work together. The Cauvery basin's canal grid has effectively partitioned what was once continuous elephant range in parts of Karnataka. There is no clean engineering fix for a canal that is also a working irrigation system. The only real solution is to assess connectivity before the canal is dug, to identify where a population corridor crosses the planned route and either reroute that section or build a crossing into the original design. That requires wildlife biologists in the room when irrigation engineers are drawing the alignment, which is not currently standard practice in India.

A fence that never kills anything and a canal that every animal in the vicinity could swim are not neutral infrastructure. They are slow partitions, and the species on either side of them are already diverging.