Transect sampling is a field method in which an ecologist travels along a defined line or path through a habitat and records the organisms observed along the way. Rather than trying to count every individual in a landscape, the method uses those observations, combined with information about how far each organism was from the line, to estimate the total population across a much larger area. It has become one of the most widely used survey techniques in ecology, applied to everything from red deer in the Scottish Highlands to porpoises in marine conservation zones. The elegance of the approach hides real complexity, though, and understanding why it works (and when it doesn’t) matters for anyone interested in how we actually know how many animals live in a given place.
The Basic Idea
Imagine walking a straight line through a forest and noting every bird you see, along with how far it was from your path. You will obviously see more birds close to the line than far from it, because nearby individuals are easier to spot. Transect sampling takes that predictable falloff in detection and uses it to figure out how many birds you missed. The mathematical backbone is called a “detection function,” which describes how the probability of spotting an organism drops as it gets farther from the transect line. Once you know the shape of that curve, you can scale up your actual count to an estimate of the true number of animals in the area.
There are two main flavors. In a line transect, the observer records the perpendicular distance from the transect line to each detected organism. Only a fraction of targets are actually measured to build the detection model, and the time saved is then used to walk a longer total length of transect and accumulate a larger count-only sample, which improves precision.1Methods in Ecology and Evolution. Efficient effort allocation in line‐transect distance sampling of high‐density species In a strip transect, the observer counts everything within a fixed-width band on either side of the line, assuming that every individual inside that strip is detected. Strip transects are simpler but make a much stronger assumption, and as we’ll see, they tend to perform worse when animals move around.
Why Not Just Count Everything?
A total census sounds ideal, but it is almost never practical. Habitats are too large, too dense, or too remote for anyone to walk every square meter. Even when you can cover a lot of ground, some fraction of the organisms present will be hidden, quiet, or camouflaged. Transect sampling addresses both problems simultaneously: it gives you a structured way to sample a manageable portion of the landscape and a statistical framework for correcting the inevitable undercount.
The method’s flexibility is a big part of its appeal. On land, a transect might be a researcher walking a trail through grassland. At sea, it could be the track of a ship or aircraft scanning the ocean surface. Underwater, divers swim in a straight line at a constant speed, counting organisms within a set width or estimating distances to each individual they see.2PubMed Central. Effect of underwater visual survey methodology on bias and precision of fish counts: a simulation approach In each case, the core logic is the same: travel a line, record what you see, model the detection falloff, and estimate abundance.
How Detection Modeling Works in Plain Terms
The central assumption of line transect sampling is that organisms directly on the transect line are detected with certainty. Ecologists call this “perfect detection at distance zero.” From there, detection probability declines with distance. The detection function captures the shape of that decline, and the software package Distance, which has been in use for decades, is the standard tool for fitting these models to field data.3PubMed Central. Distance software: design and analysis of distance sampling surveys for estimating population size
The practical payoff is straightforward. If your detection function tells you that, on average, you detect about half the individuals within a certain distance band, you can roughly double your raw count for that band to estimate the true number. The real calculations are more nuanced than simple doubling, but the principle holds: the detection function is the bridge between what you saw and what was actually there.
Aerial Surveys and Large Landscapes
Some of the most impressive applications of transect sampling happen from the air. Aerial surveys allow researchers to cover enormous areas quickly, which makes them especially useful for large, mobile species spread across rangelands, tundra, or open water. Helicopters or fixed-wing aircraft fly a series of parallel transect lines, and observers count animals and estimate distances from the flight path.
An aerial line transect study of red deer in northern Scotland estimated about 2,240 animals, which lined up closely with a ground-based census count of 2,270 conducted in the same area the year before.4Journal of Environmental Management. Evaluation of Aerial Line Transect Methodology for Estimating Red Deer (Cervus elaphus) Abundance in Scotland That kind of agreement between a fast, relatively inexpensive aerial survey and a labor-intensive full census is exactly why the method caught on.
Aerial transects have also been used for tule elk in California, where researchers estimated a total population of about 674 animals across several herds, with individual herd estimates ranging from roughly 109 to 400.5California Fish and Wildlife Journal. Use of aerial distance sampling to estimate abundance of tule elk across a gradient of canopy cover and comparison to a concurrent fecal DNA spatial capture-recapture survey Results like these are always presented with confidence intervals, because the estimates carry real uncertainty. But they give wildlife managers a working number to plan around, which is far better than no number at all.
When the Key Assumption Breaks Down
That assumption of perfect detection on the transect line is the method’s Achilles’ heel. In many real surveys, especially aerial ones, observers miss some animals directly below them. Vegetation hides individuals, animals flush and move before being spotted, or glare and terrain make observation difficult. Mule deer detection, for example, drops with increasing brush cover, sunlight, and terrain ruggedness, even right along the transect.6The Journal of Wildlife Management. A detection probability model for aerial surveys of mule deer
When detection at distance zero is less than perfect, standard line transect estimates will undercount the true population. The solution is a hybrid technique called mark-recapture distance sampling. This approach uses two independent observers (or two passes over the same transect) so that the overlap in their detections can be used to estimate how many animals both observers missed. It corrects for both the usual visibility bias at a distance and the imperfect detection on the line itself.7Wildlife Society Bulletin. Mark‐Recapture Distance Sampling for Aerial Surveys of Ungulates on Rangelands The method has been widely adopted for ungulate surveys in open rangelands where a single observer simply cannot see everything.8Methods in Ecology and Evolution. Using mark–recapture distance sampling methods on line transect surveys
Animal Movement and Measurement Error
Animals do not sit still while you survey them, and this creates a separate category of bias. If an animal moves toward the transect line during the survey, it is more likely to be detected than one that stays put, which inflates density estimates. Simulations have shown that line transects handle this better than strip transects: in strip transects, bias reaches nearly 10% when animal speed is just 30% of observer speed, while in line transects, bias stays below 10% until animal speed hits about 80% of observer speed.9PLoS ONE. The Effect of Animal Movement on Line Transect Estimates of Abundance At very high animal speeds, though, line transects can perform just as badly or worse. The practical takeaway is that the method works best when the observer is moving substantially faster than the organisms being counted.
Measurement error adds another wrinkle. Even small errors in estimating distance can bias results. Under certain error models, even unbiased distance estimates can lead to overestimates of density, and some patterns of error strip away the “shoulder” of the detection function, making it impossible to fit the model reliably.10PubMed. Predicting and correcting bias caused by measurement error in line transect sampling using multiplicative error models This is why field training matters enormously. A poorly trained observer who consistently misjudges distances can compromise an entire dataset in ways that are not easy to detect after the fact.
The Problem with Roads and Rivers
Placing transect lines is not as simple as drawing straight lines on a map. Many studies have used landscape features like roads or rivers as transect lines because they are convenient to travel along. This creates two problems at once. First, animals may be attracted to or repelled by the feature itself, whether because of human disturbance along a road or habitat characteristics near a river. Second, the habitat along these features may not represent the broader landscape.11Journal of Applied Ecology. Accounting for non-random samples with distance sampling to estimate population density If elk congregate near water sources or avoid roadsides, a transect running along either feature will give a skewed picture of the population. The gold standard is to place transects randomly with respect to animal distribution, which usually means cutting through the landscape rather than following its built-in pathways.
Researchers working in tropical and subtropical forests have also found that organisms sampled along line transects are not always independent of each other. In densely vegetated regions, where species cluster around particular microhabitats, successive observations along a transect tend to be more similar than random draws from the area would be.12PubMed Central. Biodiversity survey and estimation for line-transect sampling This non-independence can throw off biodiversity estimates unless it is explicitly accounted for in the analysis.
Cetaceans and the Open Ocean
Marine mammal surveys push transect sampling to its limits. Whales and dolphins surface only briefly, making them easy to miss even when directly on the transect line. Sea state, glare, and swell all affect how far an observer can see, and conditions can change dramatically over the course of a single survey. Ignoring these effects leads to meaningful bias in abundance estimates for cetaceans.13Marine Mammal Science. Inferring trackline detection probabilities, g(0), for cetaceans from apparent densities in different survey conditions
To deal with this, marine surveys typically use multiple observer platforms and sophisticated correction factors. Ship-based surveys might have observers on both the flying bridge and a lower platform, creating the dual-observer setup needed for mark-recapture distance sampling. Aerial surveys over the ocean use similar principles but face additional challenges from aircraft speed and the narrow window of time in which a surfacing animal can be spotted. Despite the difficulties, transect-based surveys remain the primary method for estimating abundance of whale and dolphin populations worldwide, and the resulting numbers feed directly into international management and conservation decisions.
Drones, Cameras, and eDNA Along Transects
Technology is reshaping what a “transect” looks like. Unmanned aerial vehicles (drones) can fly pre-programmed strip transects at low altitude, collecting continuous video over areas that would be dangerous or impossible for crewed aircraft. A recent study used biweekly drone flights along strip transects to monitor harbour porpoises over 17 weeks, covering a survey area of about 3.4 square kilometers and providing data across 89% of a marine conservation zone.14Remote Sensing in Ecology and Conservation. Programmed unmanned aerial vehicles show great potential for monitoring marine megafauna in specific areas of interest Drones are cheaper to operate per flight hour than helicopters, they eliminate risk to human observers in remote or rough terrain, and they produce a permanent video record that can be re-analyzed later.
Camera traps, traditionally thought of as stationary devices, have also been incorporated into the transect framework. Researchers have developed density surface models that use camera trap distance sampling data within a hierarchical modelling framework, allowing them to map how density varies across a landscape rather than producing a single overall estimate.15Methods in Ecology and Evolution. Modelling density surfaces of intraspecific classes using camera trap distance sampling This is a significant advance for species that are difficult to observe directly, like nocturnal mammals or animals in thick undergrowth.
Perhaps the most surprising recent development is the pairing of transect sampling with environmental DNA. Instead of recording organisms visually, a researcher collects water samples along a transect and sequences the DNA fragments shed by organisms in the environment. A study along a 2.5-kilometer transect through a kelp forest ecosystem found that eDNA could distinguish vertebrate community assemblages from habitats separated by as little as about 60 meters, and it revealed cryptic species that visual dive surveys had missed entirely.16PubMed Central. Assessing vertebrate biodiversity in a kelp forest ecosystem using environmental DNA The transect structure gives the eDNA results spatial resolution, so you know not just what species are present but roughly where in the habitat they concentrate.
Density Surface Models
Traditional transect analysis produces a single density estimate for the entire surveyed area. Density surface models go further by mapping how abundance varies across space. The approach works in two stages: first, distance sampling accounts for detectability; then a spatial model relates the corrected counts to environmental variables like depth, vegetation type, or distance from shore.17Methods in Ecology and Evolution. Spatial models for distance sampling data: recent developments and future directions The result is a heat map of predicted density rather than a single number, which is far more useful for conservation planning.
These models have been extended to accommodate data from multiple surveys collected using different methods, including conventional distance sampling, double-observer distance sampling, and strip transects.18PubMed Central. Extending density surface models to include multiple and double-observer survey data Being able to combine data across survey types and years makes long-term monitoring programs much more powerful, because you are not locked into a single protocol forever.
Large-Scale Monitoring Programs
Transect sampling underpins some of the longest-running and most influential wildlife monitoring efforts in the world. The UK Breeding Bird Survey, for example, uses a stratified random sampling scheme across the United Kingdom, the Isle of Man, and the Channel Islands. In each randomly selected 1-by-1-kilometer grid square, data are collected along two 1-kilometer-long transects subdivided into 200-meter sections.19Global Ecology and Biogeography. The Breeding Bird Survey of the United Kingdom Thousands of volunteers walk these transects every year, generating population trend data for hundreds of bird species. The design is deliberately simple enough for trained non-professionals to carry out consistently, which is essential when you need continent-scale coverage over decades.
This kind of program illustrates a broader point about transect sampling: the method scales. The same basic logic that works for one biologist walking a kilometer through woodland works for a national network of volunteers, a fleet of aircraft covering a sea basin, or a drone autonomously flying a grid over a coral reef. What changes is the technology and the correction factors, not the fundamental approach of traveling a line, recording observations, and modeling what was missed.
When Transect Sampling Is the Wrong Tool
No single method works for every situation, and transect sampling has clear limitations. For species that are extremely rare or patchily distributed, transects may produce too few detections to fit a reliable detection function. A study comparing mobile acoustic transects with stationary surveys for bats found that the mobile transects underrepresented bat species richness across all vegetation communities and in both dry and wet seasons. Most critically, they failed to detect three rare species, one of which was federally endangered.20PubMed Central. Mobile acoustic transects miss rare bat species: implications of survey method and spatio-temporal sampling for monitoring bats For rare species, stationary long-duration monitoring or targeted methods like camera trapping and mark-recapture may be more appropriate.
Transect sampling also struggles in very dense or structurally complex habitats where visibility is severely limited. If you can barely see five meters into the undergrowth, the detection function collapses to a very narrow effective strip width, and your estimates become imprecise. In these settings, plot-based methods or capture-recapture approaches may yield better results. The method’s sweet spot is open or semi-open environments with moderate animal densities, mobile organisms that can be seen or heard from a distance, and a survey platform (on foot, in a vehicle, in an aircraft) that moves faster than the target species.
Acoustic and Passive Transects
Visual detection is not the only way to conduct a transect survey. Acoustic transects are common for cetaceans, bats, and birds, where the observer (or a recording device) detects organisms by sound rather than sight. The same distance sampling framework applies: the probability of detecting a call decreases with distance from the transect, and the detection function can be modeled from the data. Ship-towed hydrophone arrays, for instance, record whale vocalizations along a transect and use the arrival time differences between hydrophones to estimate the distance to each calling animal.
Passive acoustic monitoring can also be paired with the transect concept by placing recording devices at intervals along a line, creating a spatial sequence of listening stations. This hybrid approach borrows the spatial structure of a transect while avoiding the need for the observer to physically travel the line during recording. It works well for species that vocalize reliably but are otherwise difficult to see, and it allows surveys to run continuously for weeks or months rather than being limited to one-off passes.
What Makes a Good Transect Design
Placement matters more than most people realize. Transects should be positioned so that they are representative of the habitat you want to estimate density for, and so that organisms do not cluster toward or away from the line itself. Random or systematic placement with a random starting point is the standard recommendation. Parallel lines spaced evenly across the study area tend to work well because they spread effort across the landscape and reduce the chance that local patchiness in animal distribution will dominate the result.
Length matters too. Longer total transect length generally improves precision more than adding additional short transects, because it increases the count sample. The tradeoff is logistical: a single very long transect through rugged terrain may be impossible to walk in a day, and conditions may change over the course of the survey. In aerial surveys, fuel and flight time constrain how much ground can be covered. Practical design is always a negotiation between statistical ideals and field realities, and experienced survey designers know which compromises cost the least in terms of bias and precision.