Movement ecology explains why organisms move, how they move, how they respond to places and conditions, and what their movements change. It connects a creature's condition and capabilities with its surroundings, from a short feeding trip to dispersal and seasonal migration. A line on a map describes part of a journey; explaining that journey requires evidence about the organism, the observation method and the environment.
The gold standard is a movement explanation that keeps those connections testable. This guide shows how to distinguish displacement from distance travelled, movement from measurement error, habitat use from selection, and a crossing from successful ecological connectivity. Its numerical cases are original hypothetical examples for learning, not assessments of real animals or prescriptions for fieldwork.
This subject connects The Gold Standard, Education: learning that transfers into life and the How X Works hub. Begin with Ecology for living relationships, or Landscape Ecology for how spatial patterns relate to ecological processes.
Choose your movement ecology route
Connect organisms, movement mechanisms, tracking evidence and ecological consequences.
Route 1 · Chapters 1–3: Define movement and measure paths
Route 2 · Chapters 4–6: Observe tracks, uncertainty and ranges
Route 3 · Chapters 7–9: Interpret habitat, behaviour and contact
Route 4 · Chapters 10–11: Connect landscapes, energy and outcomes
Route 5 · Chapters 12–14: Transfer, design and practise
Chapter index
Define movement and measure paths · Chapters 1–3
Observe tracks, uncertainty and ranges · Chapters 4–6
Interpret habitat, behaviour and contact · Chapters 7–9
Connect landscapes, energy and outcomes · Chapters 10–11
Transfer, design and practise · Chapters 12–14
Imagine a small mammal leaving a sheltered patch, crossing open ground and returning before dawn. The track could involve feeding, exploration, escape, social interaction or several activities in sequence. Its shape alone does not identify the purpose. The useful question is which explanation predicts observable differences that another explanation would not.
Nathan and colleagues' movement-ecology framework connects four components: internal state, motion capacity, navigation capacity and external factors. These provide a disciplined starting point. Hunger or reproductive condition may affect motivation; body form and physiology constrain locomotion; information and orientation affect direction; food, weather, predators and barriers shape the available circumstances.
Turn a story into a comparison
Suppose the proposed explanation is that animals travel farther when nearby food becomes scarce. Measure food availability using a relevant method, define the movement period, and compare animals or periods under suitable conditions. Also consider alternatives: disturbance may increase travel, breeding behaviour may change routes, or a new tracking schedule may reveal movement that older sampling missed.
If every possible track can be explained after the fact as “searching for something,” the story is too flexible to test. A stronger prediction specifies direction and context: under comparable observation conditions, trips should lengthen as accessible food declines, with other plausible drivers considered. Even then, an observational association does not establish the mechanism by itself.
The framework is broader than vertebrate tracking. Seeds and other propagules move through wind, water or animal transport. Their movement does not require conscious decisions. Release timing, physical properties and transport conditions can interact to produce different destinations. Be careful when using words such as choice or navigation across organisms with very different capacities.
Begin any movement study with an explicit subject, scale and outcome. “How does nightly travel change during this season?” and “Does dispersal reconnect these populations?” are different questions. They may use some of the same observations, but they need different evidence before they can be answered.
CHAPTER 2 OF 14
2. Separate movement types without forcing every journey into one box
Daily movements can connect feeding sites, shelters and social partners. Dispersal concerns movement away from an origin toward another place where an organism may settle or reproduce. Migration generally involves organised movement between regions or seasonal habitats, although the details vary across taxa. These terms identify biological questions rather than simply ranking distances from short to long.
A juvenile moving two kilometres and establishing a new breeding location may be dispersing. An adult travelling ten kilometres to feed and returning to its regular shelter may be making a routine trip. Distance alone would reverse the apparent importance of these movements if the research question concerned reproductive exchange.
Follow the individual and the population separately
One animal can change its movement pattern during its life. A track recorded during a settled period may not represent its later dispersal. Conversely, a long exploratory excursion may end in a return to the original range. Observing departure is not enough to infer permanent relocation.
At population scale, individuals may differ. Some may migrate while others remain resident; age, sex, condition or circumstances can influence those strategies. A population average can conceal two distinct patterns. Inspect individuals before describing a single characteristic journey for the species.
The existing Insect Migration guide develops seasonal journeys, environmental cues and cases where population-level migration spans generations. It is especially useful for recognising that a mapped annual cycle need not be completed by one individual. Preserve that distinction when interpreting routes assembled from observations made at different times.
For an original classroom case, compare three records: an animal returns each night, another leaves once and settles elsewhere, and a third disappears after its transmitter stops reporting. The first two support different movement descriptions. The third supports uncertainty about both movement and observation; it should not automatically be called dispersal, migration or death.
Ask what additional evidence would change each classification. Repeated seasonal journeys, establishment at a new site, successful reproduction or verified device failure can answer different parts of the question. A useful classification makes the evidence requirements clearer instead of hiding missing information behind a label.
Displacement connects a starting position to an ending position. Distance travelled follows the route between them. An animal can travel extensively and finish close to where it started. That difference matters when interpreting energy use, habitat encounters or progress toward a destination.
Consider a fictional track in a flat local coordinate system measured in metres. At 20:00 the animal is at (0, 0); at 20:10 it is at (300, 0); at 20:20 it is at (300, 400); at 20:30 it returns to (0, 0). The recorded straight-line steps are 300 m, 400 m and 500 m, because the last step is the hypotenuse of a 300–400–500 triangle.
Work both answers
The sum of recorded step lengths is 1,200 m. Over thirty minutes, that gives 40 m per minute, or about 0.667 m per second, as the average speed implied by the recorded segments. Net displacement is zero. Average velocity over the whole period is therefore zero, even though the animal was moving during the journey.
These values describe the simplified track, not necessarily the complete physical route. The animal may have curved around obstacles or paused between recorded fixes. If positions are exact, straight-line segments cannot exceed the length of the actual paths joining them. But measurement error can add artificial zigzags, so an observed sum with noisy coordinates is not automatically a lower bound on real travel.
If only the first and last positions had been recorded, the apparent movement would be zero. That would mean the observation schedule missed the excursion, not that the excursion never happened. Conversely, collecting many inaccurate positions could produce apparent movement from a stationary tag.
When comparing tracks, state the coordinate system, time interval, treatment of missing fixes and location-error assumptions. Use appropriate distance calculations for geographical coordinates rather than treating degrees of longitude and latitude as metres. For long routes, the geometry of the Earth becomes relevant as well.
The practical lesson is to name the metric before interpreting it. “Farther from the start,” “more distance travelled,” “faster progress” and “more active” are related phrases, but they are not interchangeable measurements.
Every tracking method samples movement through an observation process. GPS estimates positions at selected times. Camera traps record detections within their coverage. Acoustic receivers detect compatible signals under particular conditions. Accelerometers measure motion-related signals on the device. None supplies an unlimited view of everything an organism does.
Movebank's data model distinguishes animals, tags, deployments and sensor events. This is useful because a device can be redeployed, an animal can carry different devices, and measurements outside a deployment should not be attributed to the animal. Record the link between organism and instrument rather than assuming a tag identifier always means the same individual.
Match the schedule to the duration
Suppose an animal visits a narrow crossing for four minutes. A tracker records a position every thirty minutes. Under a simplified assumption that visit timing is uniformly distributed relative to the sampling clock and fixes succeed, the chance that at least one scheduled fix falls inside that single visit is 4 ÷ 30 ≈ 13.3%.
This calculation concerns a fix during the visit. It is not the probability of detecting any crossing evidence: fixes before and after may sometimes suggest passage, while multiple visits change the sampling opportunity. It also assumes a fixed four-minute duration and perfect timing, conditions that real data may not meet.
The example explains why absence of a recorded point inside a corridor cannot automatically show absence of corridor use. Increase temporal resolution where appropriate, combine complementary observations, or adopt a question that the available schedule can support. The best choice depends on device limits, welfare and the event of interest.
More frequent sampling can shorten battery life or increase data volume. A project intended to observe an entire seasonal journey may need a different compromise from one studying brief feeding bouts. Select that compromise before interpreting a track as complete.
For camera traps, distinguish a missed passage from a passage outside the monitored route. For acoustic systems, account for receiver coverage and conditions influencing detection. Across methods, the question remains the same: what movements could occur without leaving the evidence this system records?
A location estimate contains uncertainty. Two successive positions that differ by a few metres may represent real movement, measurement variation or both. Terrain, vegetation, device configuration and the measurement system can also affect whether a location is obtained at all.
Frair and colleagues' research synthesis on GPS errors distinguishes imprecision from habitat-related differences in fix acquisition and examines their consequences for ecological inference. Its historical device-performance values should not be treated as specifications for modern equipment. The durable lesson is to investigate the error mechanism in the relevant study conditions.
Bound a simple displacement problem
Assume two recorded positions are 20 m apart, and each position is known, for this hypothetical example, to lie within 15 m of the true position. The true displacement can range from zero to 50 m. The maximum follows from 20 + 15 + 15; the lower bound is max(0, 20 − 15 − 15).
This is a geometric bound under assumed hard error limits, not a 95% confidence interval and not a claim that all displacements in the range are equally likely. If the errors instead follow probability distributions, uncertainty should be represented accordingly. Do not turn a manufacturer's typical accuracy into a guaranteed maximum error without evidence.
The result prevents an unsupported conclusion that the animal definitely moved 20 m. It does not imply that every movement below 30 m is impossible to study. Better calibration, different sensors, repeated information and suitable error models can change what is identifiable.
For an apparent long jump, check timestamps, duplicated records, deployment boundaries and location-quality information before treating it as exceptional behaviour. A point implying an implausible speed deserves review, but a rigid speed cutoff can also remove genuine rapid movements. Preserve flagged records and the reasons for decisions.
If uncertainty is large relative to the width of a habitat strip, assigning every fix to one side of the boundary can misclassify habitat use. Examine sensitivity to location uncertainty and map accuracy. A sharp coloured point on a map should not make the underlying measurement seem more certain than it is.
A home-range estimate summarises an animal's space use under a particular method and observation period. It is not automatically a defended territory, an impermeable boundary or the full set of places the animal could reach. Seasonal changes and exceptional excursions can alter what a single outline means.
Some methods draw boundaries around observed positions; others estimate a utilisation distribution describing relative space use. A contour associated with 95% of an estimated utilisation distribution is a probability-content statement within that model. It should not be described as a guarantee that the animal will remain inside it or as a 95% confidence boundary for the true range.
Noonan and colleagues' home-range study examines how autocorrelation and effective information affect estimation. Many closely spaced fixes can describe one individual's movements in great detail while contributing less independent information than their raw count suggests. Appropriate analysis depends on movement properties and the inferential question.
A rectangle is an explicit teaching model
Suppose an animal's observed positions fit within a rectangle 800 m long and 500 m wide. The rectangle's area is 400,000 square metres, or 0.4 square kilometres. That calculation is exact for the rectangle. It does not establish that every part was used, that the animal prefers all of it, or that the rectangle is a suitable home-range estimator.
If a reservoir occupies half the rectangle and the focal terrestrial animal never enters it, the outline includes a large area with a different biological interpretation. A more elaborate estimator is not automatically better, but its assumptions should match the question and the animal's movement.
Compare like periods and methods. A month of tracking and a year of tracking may reveal different extents because they include different seasons and rare excursions. A larger estimate can reflect longer observation rather than a behavioural expansion.
For population-level claims, track multiple relevant individuals and represent their differences. Ten thousand fixes from one animal do not become ten thousand independent animals. The sampling unit must follow the question being asked.
An animal's recorded locations show use under the observation process. Selection asks whether use differs from what was available under a specified comparison. A habitat can contain many fixes because it is widespread, readily accessible, frequently visited, or easier for the device to record. These explanations should not be collapsed into preference.
Calculate a simple use-to-availability ratio
Assume a defined accessible area contains 20% woodland, while 40% of suitable, comparably sampled locations fall in woodland. The simple ratio is 0.40 ÷ 0.20 = 2. Woodland is represented twice as strongly in the recorded use as in that availability definition. The number is descriptive; it is not a fitted selection coefficient or proof that woodland improves survival.
Now change the accessible area so that woodland comprises 50%. The same observed use gives 0.40 ÷ 0.50 = 0.8. The apparent interpretation changes because the comparison changed. This is why availability must be biologically justified rather than selected to obtain a preferred result.
Step-selection functions compare observed movement steps with alternative steps from the same starting location. They make local movement opportunities more explicit than a single broad availability map, but still depend on how alternatives and covariates are constructed. They are models of a defined comparison, not direct access to an animal's intentions.
For a beginner's exercise, place an animal beside a river with a single bridge. A circular buffer may include apparently nearby habitat that cannot be reached within the next short interval. Modify the available alternatives to reflect the barrier and the timescale. Explain how the comparison now answers a different, more realistic question.
Also distinguish conditions at an endpoint from conditions encountered along a route. An animal can end in woodland after crossing open ground. Sparse fixes may not identify the precise path. If the mechanism concerns exposure during travel, an endpoint-only analysis may miss the relevant conditions.
Finally, inspect individuals and seasons. A population-wide association can conceal strong differences in reproductive state or local accessibility. Report those limits alongside any general conclusion about habitat selection.
A slow, winding track might suggest feeding; a fast, directed segment might suggest travel. Those interpretations are plausible in some settings, but they are not universal definitions. Slow movement can occur during searching, resting with measurement noise, handling food or negotiating obstacles. Different activities can produce similar spatial signatures.
Hidden Markov models can represent sequences as transitions among unobserved states that generate different distributions of observed movement variables. A model might distinguish short, tortuous steps from longer, straighter ones. Naming those states resting and travelling adds a biological interpretation that requires evidence beyond the statistical separation.
A GPS and video validation study shows why this matters: movement-derived states can differ from observed behaviour, particularly when coarse intervals mix activities. Its practical implications include validating state labels where possible and matching temporal resolution to the behaviours being studied. A state assigned to a step need not describe the animal's behaviour at one exact endpoint.
A mixed-activity interval
In a fictional twenty-minute interval, an animal feeds for eight minutes, travels for seven and rests for five. The time proportions are 40%, 35% and 25%. A model forced to assign one state to that interval cannot preserve all three proportions through the label alone.
If the animal ends 300 m from its starting position, that displacement does not reveal how the twenty minutes were divided. A similar endpoint could arise from a direct trip followed by rest or a circuitous search. Add appropriately validated accelerometer, video or direct-observation evidence when the research question requires the distinction.
Validation itself needs a relevant sample. Video obtained only in daylight may not establish the accuracy of labels at night. A model trained on one species, age class or attachment configuration may not transfer unchanged. Record the reference observations, disagreements and conditions under which classification remains uncertain.
The right conclusion can therefore be modest and useful: “The model identifies two movement patterns; auxiliary observations support a feeding interpretation for part of the slower pattern.” That is more informative than turning every coloured section of a track into a certain account of behaviour.
Two animals can occupy overlapping home ranges without meeting. One may use a feeding patch at dawn and the other after dusk. Their ranges establish a possible spatial opportunity, while contact requires evidence about timing, distance and the type of interaction being considered.
The specialist guide to home-range overlap and recorded contact examines this distinction in depth. It is the next route when the question concerns proximity networks, shared resources or exposure. Here, the root principle is that movement relationships must preserve time as well as space.
Work the observation denominator
Suppose two tracked animals have 80 times at which both locations can be compared under a predefined synchronisation rule. At eight of those times, their estimated separation is below a chosen threshold. Recorded proximity is 8 ÷ 80 = 10% of comparable observations.
If 100 paired observations were planned but twenty lack a usable location for one animal, using 100 as the denominator gives 8%. That describes detections per planned opportunity, not proximity frequency among comparable observations. Neither percentage establishes physical contact, contact duration or transmission of an infection.
The distance threshold must also be interpreted with location error. If uncertainty is large relative to the threshold, classifications near its boundary can be unstable. Report sensitivity to reasonable thresholds and time-alignment choices rather than selecting whichever value produces the most striking network.
Asynchronous records create another trap. Positions five minutes apart are not simultaneous merely because they are the closest available pair. A fast-moving animal may travel substantially in that interval. Interpolation can be useful under a justified movement model, but it creates estimates and uncertainty rather than additional observed positions.
Indirect interactions may require a different analysis. An animal can encounter material left by another long after the first animal departed. Then the relevant time window depends on the process, not just simultaneous proximity. Define the ecological mechanism before choosing a distance threshold or drawing a contact edge between two individuals.
Connectivity can concern passage, settlement, survival or reproductive exchange. A wildlife crossing that records many passages provides evidence about use. It does not, by itself, establish that populations are demographically connected or that genes are moving successfully between them.
Landscape Genetics develops the link between landscape structure and reproductive exchange. Movement ecology supplies a complementary part of that explanation: where individuals travel, the barriers or opportunities they encounter, and what happens after arrival. These evidence types can support one another without becoming interchangeable.
An original sequence calculation
Imagine 100 animals begin a dispersal attempt. Under a simplified scenario, 80% reach the far side of a barrier. Of those that cross, 50% settle in suitable habitat. Of those that settle, 25% reproduce successfully during the defined observation period. The expected numbers are 80 crossing, 40 settling and 10 reproducing.
The final fraction is 0.80 × 0.50 × 0.25 = 0.10, or 10%. These are conditional probabilities along a specified sequence, so multiplying them does not require pretending the stages are independent. The values are hypothetical and do not predict any real species' success.
If passage improves to 90% while the later conditional probabilities remain unchanged, the expected reproductive fraction becomes 11.25%. That is a gain of 1.25 percentage points, not ten percentage points. If settlement becomes harder because habitat beyond the crossing deteriorates, improved passage may produce even less benefit. State what is assumed to remain constant.
Corridor planning therefore needs to include destinations and, for some journeys, stopovers. USGS work on migration and corridors connects routes with stopover and seasonal habitats. Protecting the narrow line while losing necessary resources along it can leave the biological journey unsupported.
A useful monitoring design names outcomes at each relevant stage. It also considers which individuals were observed and whether their journeys represent the wider population. Many detections from a small number of repeat users should not be described as many successful dispersers.
The shortest route is not necessarily the least costly or most successful. Terrain, currents, wind, cover, food and danger can change the consequences of travelling a particular distance. Organisms operate with limited time and energy, but a track alone rarely identifies the exact trade-off being made.
Compare two fictional routes
Route A is 600 m long across exposed ground. Assume, purely for this exercise, a locomotion cost of 2 joules per metre. Its travel cost is 1,200 J. Route B is 900 m through a more favourable surface, with an assumed cost of 1 joule per metre. Its cost is 900 J. B is 50% longer but costs 25% less under these assumptions.
These values are invented to demonstrate a calculation. Real energetic costs vary with species, speed, slope, gait, body mass and conditions. The example does not establish that covered routes are usually cheaper. It shows why distance cannot stand in for energy without a justified conversion.
Now suppose B requires an additional 500 J of climbing. Its total becomes 1,400 J, making A less costly in this limited accounting. If A has greater predation risk, however, energy alone still does not determine the preferred route. Avoid adding an arbitrary risk score to joules as if the units were naturally interchangeable.
Moving air and water introduce a second distinction: velocity relative to the surrounding medium differs from velocity relative to the ground. A drifting organism can move far geographically with little self-propelled motion. A swimmer can expend effort while making little upstream progress. Interpret the reference frame before explaining performance.
The Insect Migration guide provides a more detailed route into wind-assisted journeys. For the broader movement question, record environmental conditions at the relevant place and time. A seasonal average may miss the particular current or gust encountered during a movement event.
Where energy or danger is central, seek independent observations or validated models. A winding track is not sufficient evidence that an animal wasted energy, and a straight track is not proof that it followed an optimal plan.
A study records positions every five minutes for twenty animals during one season. Twelve animals have at least one confirmed corridor passage, and forty passages are recorded in total. A second study uses hourly positions for ten animals in another season and records three animals with confirmed passage. A report claims corridor use fell from 60% to 30% because the corridor became less suitable.
Before reading the answer, identify what each percentage describes, why the forty passages do not change the first denominator, and which changes prevent a straightforward causal interpretation. Then propose a comparison that would better test the claim.
Answer and reasoning
The first proportion is 12 ÷ 20 = 60% of monitored animals with recorded passage. Forty passages can include repeated use by the same individuals. It does not mean forty different animals crossed. The second proportion is 3 ÷ 10 = 30% of monitored animals with recorded passage under its own observation conditions.
The sampling interval, season and animal sample changed. Brief passages may be easier to confirm with five-minute records than hourly records. Seasonal motivation and accessibility may differ, and the two samples may differ in age, sex, condition or capture location. The observed percentages therefore do not isolate corridor suitability.
One diagnostic is to resample the fine-resolution tracks to the coarser schedule using several plausible clock alignments, then compare how many known passages remain detectable. This evaluates part of the observation problem within the original dataset. It does not recreate missing movements in the second dataset or remove seasonal and sampling differences.
A stronger study would use comparable observation designs across relevant seasons and animals, include independent passage monitoring where appropriate, and evaluate habitat change with a suitable comparison. An intervention analysis may require observations before and after change plus comparison locations or another defensible design. The correct design depends on the mechanism and practical constraints.
The conclusion supported now is narrower: recorded passage was found in a smaller fraction of the second monitored sample. The cause remains unresolved. That statement retains useful evidence without pretending the two observation systems measured the same thing equally well.
Transfer the principle to fish receivers, camera traps or seed traps. Whenever methods change, ask what movements each method could miss before interpreting the difference as a change in nature.
Good movement research considers the effects of observation on the organism. A device can change drag, loading, behaviour or condition; capture and attachment can also have consequences. The appropriate assessment is species-, device- and study-specific. A simple universal percentage of body mass is not a complete welfare or data-quality evaluation.
The National Park Service's account of migratory-bird tracking technology emphasises matching methods to questions and evaluating tag effects. USGS research on transmitter effects in sea ducks illustrates why both short- and longer-term behavioural changes deserve attention. Its findings should not be converted into a universal post-tagging exclusion period for other species.
More fixes and more animals answer different needs
Suppose a fixed hypothetical budget permits either five animals with 10,000 fixes each or twenty animals with 2,500 fixes each. Both designs produce 50,000 fixes. The first may describe individual paths more finely; the second samples more individuals. Neither is universally superior. Brief-event detection, individual variability, duration and the intended population inference determine what information is needed.
Before collecting data, simulate plausible movement and observation processes to examine whether the proposed design can distinguish the explanations of interest. Include missing fixes and location error. A simulation that assumes perfect observation may make an inadequate design look reliable.
Preserve timestamps, deployment records, quality flags, environmental information and analysis versions. The Ecological Informatics root explains how to maintain these meanings through data integration and modelling. When evaluating prediction, hold out the unit relevant to intended use, such as an animal, season or region, instead of randomly splitting closely related fixes without justification.
Sharing also requires judgment. Current locations of vulnerable wildlife can be sensitive, and data access does not imply unrestricted reuse. Movebank's data policy explains ownership and access considerations. Preserve sufficient documentation for responsible analysis while respecting the conditions attached to the data.
For learning, use fictional tracks or appropriately released datasets. Handling or tagging wildlife is unnecessary for understanding the calculations in this guide and requires the relevant expertise, permissions and welfare oversight in real research.
Does a straight route prove that an animal knows its destination?
No. Directional movement can arise from several mechanisms, including environmental transport, orientation responses and movement through constrained terrain. Evidence about navigation requires a design that distinguishes the proposed information or orientation mechanism from alternatives. Track shape is a clue, not a complete explanation of cognition.
Does no recorded movement mean the animal was resting?
No. A long sampling interval can miss an excursion; movement can occur within location uncertainty; the device can detach or stop reporting. Check instrument status, interval and supporting observations. A resting interpretation becomes stronger when independent evidence supports it and plausible observation failures have been examined.
Does a large home range mean poor habitat?
Not necessarily. Range size can vary with body size, season, reproductive behaviour, resource distribution, observation duration and estimation method. A habitat-quality claim needs an explicit mechanism and appropriate comparisons. Range area alone does not measure survival, reproduction or wellbeing.
What should I inspect when a movement result looks wrong?
For impossible jumps, inspect time order, coordinate systems and deployment boundaries. For a sudden activity change, inspect sampling schedules and device changes. For a strong habitat association, inspect availability, fix success and mapping uncertainty. For a contact network, inspect simultaneous observation opportunities. These are diagnostic routes, not reasons to discard unexpected biology automatically.
A practical learning sequence
Start with the three-position excursion in Chapter 3 and calculate both distance and displacement. Remove the middle observations and explain what information disappears. Then add hypothetical location uncertainty and distinguish an exact track calculation from an uncertain biological interpretation.
Next, draw two habitats separated by a barrier with one crossing. Define what is reachable in a short interval, calculate a simple use-to-availability ratio, and change the availability boundary. Finally, work the corridor transfer challenge and write a conclusion that preserves the difference between recorded passage and actual movement.
For further study, use Landscape Ecology for spatial processes, Landscape Genetics for reproductive connections, and the linked specialist guides for migration and contact. The primary research and official documentation linked here were checked on 27 October 2026.
The gold standard of movement ecology is an explanation that follows both the organism and the evidence: what moved, what could have been missed, which mechanisms remain plausible, and what consequences the observations can actually support.