A pair of animals can use overlapping ground without meeting. One may visit the shared area at night and the other by day. Their collars may record positions at different times. A large home-range overlap can describe potential opportunity for contact while saying little about whether the animals were close together at the same moment. The gold standard keeps those questions separate.
This guide builds an accessible evidence chain from estimated space use to recorded proximity. It uses fictional animals and numbers so the reasoning can be checked without exposing sensitive wildlife locations. It is not a field protocol, disease-risk assessment or animal-management recommendation. Real telemetry work needs permits, welfare safeguards, secure data handling, species knowledge and an analysis designed for the research question.
This specialist article belongs under The Gold Standard of Wildlife. It connects to The Gold Standard of Education: Learning That Transfers into Life and the verified How X Works hub.
eduKateSG · Wildlife
Find your next learning step
Separate static space-use overlap from time-aligned proximity, keep the opportunity denominator visible and test sensitivity to thresholds and error.
ROUTE 1 · CHAPTERS 1–10
Define the relationship
Define ranges, overlap, timestamps, thresholds and the central calculation.
ROUTE 2 · CHAPTERS 11–20
Control the observation
Handle missingness, sampling, estimators, behaviour and transmission routes.
ROUTE 3 · CHAPTERS 21–30
Build responsible evidence
Protect locations, test alternatives, reproduce events and correct misconceptions.
ROUTE 4 · CHAPTERS 31–40
Extend the movement model
Compare metrics, lags, devices, seasons, networks and welfare.
ROUTE 5 · CHAPTERS 41–49
Transfer and continue
Test denominators, duration, resources, simulation and verified routes.
Chapter contents
Define the relationship · Chapters 1–10
1. Ask which relationship you mean
2. Define a home range carefully
3. Work the directional-overlap example
4. Separate static overlap from dynamic interaction
5. Align the observation periods
6. Align timestamps without inventing simultaneity
7. Calculate a recorded proximity rate
8. Keep the distance threshold explicit
Control the observation · Chapters 11–20
11. Treat missing fixes as missing opportunities
12. Check whether sampling frequency can see the event
13. Avoid treating fixes as independent samples
14. Compare compatible estimators
15. Do not infer contact hotspots from habitat use alone
16. Keep behaviour and group structure visible
17. Distinguish direct and environmental pathways
18. Protect sensitive locations
Build responsible evidence · Chapters 21–30
21. Build a reproducible pair ledger
22. Correct common misconceptions
23. Questions readers often ask
24. Use simulation to test the observation design
25. Examine contact around shared resources
26. Consider species-specific sensory ranges
27. Write a proportionate conclusion
28. Define a contact opportunity denominator
Extend the movement model · Chapters 31–40
31. Consider directional following
33. Check edge effects and study boundaries
34. Transfer the reasoning to camera traps
35. Compare overlap metrics before choosing one
36. Keep core areas distinct from full ranges
37. Examine lagged association
38. Separate pair opportunity from population opportunity
Transfer and continue · Chapters 41–49
41. Treat sensor proximity as another measurement system
42. Distinguish contact probability from contact count
43. Propagate uncertainty into network edges
44. Compare seasons without changing the measurement
45. Validate selected events with independent evidence
46. Design a welfare-conscious study
47. Communicate results without exposing animals
Home-range overlap, simultaneous co-location, close approach, physical contact and transmission opportunity are different relationships. A study may need only the first. Another may require a defined distance at matching times. A disease question may also require duration, behaviour or environmental persistence.
Write the target relationship before calculating. “Do their ranges overlap?” calls for comparable space-use estimates. “Were they within 50 metres during the same ten-minute window?” calls for aligned time records and location uncertainty. “Could infection pass?” requires additional biological evidence. The strongest-looking metric is not automatically the one that answers the question.
A home range is an estimated account of an animal’s space use over a stated period and method. It is not a fence the animal recognises, and it is not simply every point joined by an arbitrary boundary. Estimates can change with sampling duration, fix frequency, movement autocorrelation, estimator settings and the behaviour represented.
Record the animal, observation period, estimator, probability contour, coordinate system and data-screening rules. Comparing a 95% utilisation distribution for one animal with a minimum convex polygon for another mixes methods. The shapes may look comparable on a map while describing different statistical objects.
Imagine Animal A has an estimated 100-hectare home range and Animal B has an 80-hectare home range during the same month. Their intersection covers 40 hectares. The shared area is 40% of A’s range and 50% of B’s range.
Both percentages are correct because overlap is directional when the denominators differ. Saying “the overlap is 45%” by averaging them creates a third quantity that may not answer the study question. Report the intersection and each directional proportion. Then remember what remains unknown: how intensively each animal used those 40 hectares and whether they used it at the same times.
CHAPTER 4 OF 49 · Define the relationship
4. Separate static overlap from dynamic interaction
Static overlap compares distributions or areas across a period. Dynamic interaction asks how two movement paths relate in time. Two animals can have identical home ranges while avoiding one another temporally. They can also have modest range overlap but repeatedly meet at one resource.
Use language that preserves the distinction. “Their estimated ranges shared 40 hectares in June” is a static statement. “Their recorded locations were within 50 metres in two aligned hourly fixes” is a dynamic observation under a stated threshold. Do not rewrite the first as evidence that the animals socialised, competed or transmitted disease.
Overlap computed from January–March for one animal and July–September for another cannot support a claim about contemporaneous contact opportunity. Even if both ranges cover the same place, the animals may not have occupied it in the same season.
Restrict the comparison to the interval when both animals were monitored, or explain why another design is justified. Record gaps. A long common calendar interval can still contain few simultaneous valid fixes. Effective overlap in time matters alongside overlap on a map.
CHAPTER 6 OF 49 · Define the relationship
6. Align timestamps without inventing simultaneity
Suppose Collar A records at 10:00 and 11:00 while Collar B records at 10:20 and 11:20. Pairing nearest times creates a 20-minute mismatch. Whether that is acceptable depends on movement speed, location accuracy and the contact definition.
Choose the tolerance before looking for encounters. Preserve actual time differences in the result. Interpolation can estimate positions between fixes under a movement model, but estimated positions are not observed fixes. Label them and propagate the model’s uncertainty. A software join on nearest timestamps is a processing step, not proof of simultaneous presence.
Continue the fictional example. During 60 aligned hourly opportunities, both collars provide valid positions. Twelve paired positions place both animals somewhere inside the shared home-range area. Only two paired positions put them within the study’s 50-metre proximity threshold.
The observed proximity proportion is 2 divided by 60, or about 3.3% of valid paired opportunities. The proportion of paired opportunities with both animals inside the overlap is 12 divided by 60, or 20%. Neither number is the 40- or 50-percent static overlap. Each has a different numerator and denominator, which should be reported.
A 50-metre threshold may be useful for one question and irrelevant for another. Physical contact generally requires a much smaller relationship; acoustic or olfactory interaction can occur across a larger distance. Environmental transmission can occur without simultaneous proximity.
Run a sensitivity table where appropriate: count pairs within 25, 50 and 100 metres, using the same valid opportunities. If conclusions change sharply, report that dependence rather than selecting the threshold that creates the desired story. Thresholds should follow biological reasoning and measurement limits, not visual convenience.
GPS positions contain error. If each fix may be uncertain by tens of metres, a measured separation of 45 metres does not cleanly prove a true distance below 50 metres. Terrain, canopy, device behaviour and processing can change the error pattern.
For an elementary bound, if two positions each have a horizontal uncertainty radius of 15 metres, a recorded 45-metre separation could correspond to a true separation roughly between 15 and 75 metres under a simple worst-case geometry. This bound is deliberately conservative and not a substitute for a device-specific error model. It shows why a hard biological claim should not rest on a near-threshold pair without uncertainty.
CHAPTER 10 OF 49 · Define the relationship
10. Distinguish opportunities from events
If an animal pair remains close for three hours and collars record every ten minutes, eighteen positive pairs may represent one sustained encounter rather than eighteen independent events. Conversely, a one-hour schedule can miss several short encounters between fixes.
Define an event-separation rule appropriate to the question, such as requiring a gap outside the threshold before a new event begins. Report the raw positive fixes and derived event count. The rule changes the interpretation and should not be hidden inside software defaults.
CHAPTER 11 OF 49 · Control the observation
11. Treat missing fixes as missing opportunities
When one collar fails to record, the pair cannot be classified from that scheduled opportunity. Replacing a missing position with “no contact” biases the result downward. Dropping all poor-quality periods can also bias results if missingness is related to habitat or behaviour.
Report scheduled opportunities, valid paired opportunities, missing pairs and the reason when known. Compare missingness across animals and habitats. A complete-looking percentage with an invisible denominator is less useful than a qualified result with a transparent observation account.
CHAPTER 12 OF 49 · Control the observation
12. Check whether sampling frequency can see the event
An hourly schedule cannot reliably describe contacts lasting a few minutes. The collars can be accurate at each recorded time and still miss the phenomenon between times. Higher frequency may improve temporal resolution but affects battery life, data volume and animal-handling plans.
Match frequency to the event duration and research goal before deployment. After data collection, keep claims inside the design’s resolving power. Do not reconstruct a detailed social network from sparse fixes merely because a mapping package can draw one.
CHAPTER 13 OF 49 · Control the observation
13. Avoid treating fixes as independent samples
Animal movement is temporally autocorrelated: a position is related to earlier positions. Hundreds of frequent fixes do not necessarily provide hundreds of independent pieces of information about a range or overlap. This affects confidence intervals and comparisons.
Use methods suited to the movement process and effective sample size. The research by Winner and colleagues develops overlap inference with autocorrelated kernel density estimation and propagates uncertainty from the home-range estimates. For learners, the key principle is simpler: more rows do not automatically erase dependence.
Different home-range estimators and tuning choices can yield different boundaries and overlap values from the same tracking data. Winner and colleagues note that bias in the underlying range estimates propagates into overlap estimates. A difference between two pairs may reflect methodology as well as behaviour.
Use the same justified workflow for comparisons, or model the differences explicitly. Preserve confidence intervals where available. A point estimate of 0.55 is not automatically meaningfully larger than 0.35 when uncertainty is broad or sampling strategies differ.
CHAPTER 15 OF 49 · Control the observation
15. Do not infer contact hotspots from habitat use alone
Animals may prefer the same habitat features yet use them at different times or in different parts of the landscape. Research on wild pigs found that individual resource-selection patterns did not simply predict hotspots of contact. The contact model treated locations of actual contact separately from available, non-contact locations inside overlapping ranges.
This distinction is valuable for students: shared preference creates an opportunity structure, not a guaranteed encounter. Test the contact record itself. Avoid converting a habitat map into a social or disease network without the required temporal evidence.
CHAPTER 16 OF 49 · Control the observation
16. Keep behaviour and group structure visible
Two animals from the same social group may travel together, producing many close fixes. A solitary adult may enter the same landscape at different times. Aggregating all pairs into one contact rate can mix these mechanisms.
Record sex, age class, group membership or other relevant attributes only when lawfully and ethically available. Analyse at the pair level before forming population summaries. Weighting choices matter: a heavily sampled pair should not silently dominate a mean intended to describe typical pairs.
CHAPTER 17 OF 49 · Control the observation
17. Distinguish direct and environmental pathways
Recorded proximity supports questions about direct or near-direct encounters under the threshold. Some pathogens persist in water, soil, food sites or other environmental media. Animals can use the same place days apart and still have a plausible indirect pathway.
Define the temporal window from appropriate biological evidence. Do not call all shared space contact, and do not assume lack of simultaneous proximity rules out environmental transmission. The movement record is one layer in a larger causal account.
Telemetry data can expose dens, nests, migration bottlenecks and rare-species locations. The most precise data are not automatically suitable for a public article, classroom map or open download. Separate analytical access from public display.
Use invented coordinates for teaching. Follow project rules for masking, aggregation and access. Check image backgrounds and metadata as well as coordinate columns. An apparently harmless screenshot can reveal a sensitive site. This continues the root guide’s principle that accurate information still needs responsible distribution.
Animal C and Animal D have only 10 hectares of estimated range overlap. During 30 valid paired fixes, both animals are inside that small intersection six times and within 25 metres four times. Their static overlap area is smaller than A and B’s 40 hectares, yet their recorded close-proximity proportion is 4/30, or 13.3%, greater than A and B’s 3.3% under its 50-metre example.
The comparison does not prove stronger social bonds because thresholds, schedules and species context may differ. It demonstrates the central logic: a larger shared area does not mechanically determine a larger recorded contact rate.
CHAPTER 20 OF 49 · Control the observation
20. Test the conclusion against alternative explanations
A cluster of close fixes might reflect attraction, shared food, constrained terrain, a den, collar error or interpolation. Low contact might reflect avoidance, different schedules, missing data or inadequate sampling frequency. List plausible alternatives before naming a mechanism.
Then ask what new evidence separates them: field observations, finer temporal sampling, habitat covariates, device checks or an independent season. A limitation should guide the next test. It should not be a ceremonial sentence added after a confident causal claim.
For each pair, record the common monitoring interval, range estimator, overlap quantities, scheduled paired times, valid paired times, threshold, positive fixes, event rule, missingness and uncertainty method. Include software and data versions where relevant.
This ledger lets another analyst reconstruct the denominator and distinguish a map overlap from a contact event. It also helps identify pairs that should not be compared. A clean graphic without this account may be attractive but difficult to audit.
Overlapping ranges do not prove animals met. No observed close fixes do not prove they never met. More frequent fixes do not automatically create independent evidence. A 95% home-range contour is not a 95% chance that an animal is inside at every moment. A contact threshold is an analytical rule, not a universal biological boundary.
Likewise, uncertainty does not make the data useless. It changes the claim. A study may identify potential contact zones, bound plausible encounters or design better sampling even when it cannot establish physical contact.
CHAPTER 24 OF 49 · Build responsible evidence
24. Use simulation to test the observation design
Before deployment, simulate movement paths with known encounters and sample them at the proposed fix schedule. Compare true simulated events with detected events. Change event duration, speed, missingness and location error. The exercise shows which contacts the design can and cannot resolve.
Simulation does not predict the animals’ real behaviour. It tests the measurement system under stated conditions. If five-minute encounters disappear under an hourly schedule, later analysis cannot recover them reliably without stronger modelling assumptions. The design can then adjust frequency, duration or the research question before animals are handled.
Preserve the simulation code, parameters and random seed. Report detection performance as conditional on the simulated scenarios, not a universal accuracy claim. This turns sampling limitations into an explicit planning decision.
CHAPTER 25 OF 49 · Build responsible evidence
25. Examine contact around shared resources
Water points, feeding sites, crossings and shelters can concentrate animals. A high event count near one feature may reflect resource distribution rather than social attraction. Compare resource availability and use outside contact events.
Avoid baiting or disturbing animals to create observations. Use authorised passive data and appropriate null models. Management interpretation should distinguish a resource hotspot from an interaction preference.
CHAPTER 26 OF 49 · Build responsible evidence
26. Consider species-specific sensory ranges
Animals can interact through sound, scent or sight beyond a GPS distance chosen for physical proximity. The relevant range varies with species, environment and signal. One universal threshold cannot capture every interaction.
Define the behavioural mechanism first. If evidence supports only spatial closeness, report that. Do not upgrade it to communication or contact merely because such interaction is possible.
A strong final sentence might say: “During 60 valid paired hourly fixes, two recorded separations were below 50 metres; the estimated monthly ranges overlapped by 40 hectares, but device error and between-fix encounters remain unresolved.”
This sentence keeps observation, static overlap, threshold and limits together. It is more useful than “the animals frequently interacted,” which introduces a behavioural conclusion the data did not establish.
CHAPTER 28 OF 49 · Build responsible evidence
28. Define a contact opportunity denominator
The denominator can be scheduled paired fixes, valid paired fixes, minutes of simultaneous monitoring or time spent inside shared space. Each produces a different rate. Choose the denominator that matches the claim.
If two collars overlap for 30 days but one records only at night, calendar days exaggerate effective opportunity for a daytime-contact question. Build an availability timeline and sum intervals in which both devices and animals could contribute under the design. Publish the excluded time and reason. A rate without its opportunity denominator cannot be compared safely.
CHAPTER 29 OF 49 · Build responsible evidence
29. Account for unequal monitoring among pairs
Pair A–B may have 1,000 aligned fixes while pair C–D has 50. A pooled fraction of positive fixes will be dominated by A–B. An equal-pair mean gives both pairs equal influence but makes the sparse pair influential.
Decide whether the target is a typical observation opportunity or a typical animal pair. Present both when useful and include uncertainty. Do not silently pool. This is the same weighting question that appears across ecology: the unit of inference should determine the unit of analysis.
Two proximity events with the same minimum distance can differ in duration. One may be a brief crossing; another may last hours. Duration can matter for behaviour and transmission, but sparse fixes only bound it.
If a pair is outside the threshold at 10:00, inside at 11:00 and outside at 12:00, the true duration is somewhere within the surrounding interval under simple assumptions, not exactly one hour. Report bounds or model-based estimates with assumptions. Avoid assigning the whole interval automatically.
Movement interaction may be asymmetric. B may tend to enter locations shortly after A, while A does not follow B. Symmetric range overlap cannot express direction.
Use time-lagged, pair-directed analyses where justified and compare against an appropriate null movement model. A sequence alone does not prove intentional following; shared trails or resources can generate the pattern. Directional language should remain proportional to evidence.
Some proximity will occur by chance because animals share space and time. A raw event count is more informative when compared with an expectation under independent movement that preserves relevant features such as home ranges, activity cycles and sampling schedule.
Simple random shuffling can destroy these features and create an unrealistic baseline. Explain the null model and validate it. A result exceeding the null suggests association under the model; it does not by itself identify the biological cause.
CHAPTER 33 OF 49 · Extend the movement model
33. Check edge effects and study boundaries
If tracking data are clipped to a study area, apparent overlap and contact opportunities near the boundary can be distorted. Animals may move outside the observed region, and missing external positions can change range estimates.
Record the monitoring extent and whether devices continued beyond it. Avoid interpreting the map frame as an ecological boundary. Sensitivity analysis with alternative extents can reveal whether the result depends on clipping.
CHAPTER 34 OF 49 · Extend the movement model
34. Transfer the reasoning to camera traps
Two species photographed at the same camera station during a month share a site but may never be present together. To examine temporal co-occurrence, compare timestamps within a stated window and account for camera downtime, trigger performance and independence rules.
The method transfers: define spatial overlap, align observation periods, define proximity in time, keep opportunities and uncertainty visible, and avoid inferring contact from shared site use alone. The sensors change, but the evidence chain remains.
CHAPTER 35 OF 49 · Extend the movement model
35. Compare overlap metrics before choosing one
Home-range overlap can be expressed through intersection area, directional proportions, utilisation-distribution similarity or other indices. These metrics emphasise different relationships. Two pairs can change rank when the metric changes because one pair shares a small intensively used core while another shares a large lightly used boundary.
Choose the metric from the biological question and estimator, not after seeing which result looks strongest. Define its range and symmetry. If A-in-B differs from B-in-A, publish both directional values. When comparing studies, verify that probability contours, smoothing and overlap metrics match. A shared word such as “overlap” does not guarantee comparable calculations.
CHAPTER 36 OF 49 · Extend the movement model
36. Keep core areas distinct from full ranges
A 50% core contour describes concentrated use differently from a 95% range. Two animals may have broad outer-range overlap but separate cores, or overlapping cores inside mostly separate ranges. These patterns can suggest different contact opportunities, but they still require temporal evidence.
Report the contour with every value. Do not subtract a core overlap from a full-range overlap as if the two were simple nested areas under all estimators. Examine the utilisation distributions and uncertainty. A map should not present an arbitrary contour as a hard territorial boundary.
Simultaneous proximity is only one dynamic relationship. One animal may follow another after a delay or visit a site shortly after it. Calculate lagged distances only when the research question justifies them, and preserve the lag direction.
For example, compare A at time t with B at t+30 minutes, then reverse the direction. Similar results do not prove following; both may respond to a shared resource schedule. Lag analysis creates hypotheses that require behavioural and environmental evidence. Testing many lags also raises a multiple-comparison problem, so pre-specify a meaningful range where possible.
CHAPTER 38 OF 49 · Extend the movement model
38. Separate pair opportunity from population opportunity
Some pairs share no space during the common tracking interval and therefore have no opportunity for direct proximity under the model. Others overlap substantially. Averaging contact rates across all pairs can mix structural zeros with observed non-contact among available pairs.
Report how many pairs had overlapping ranges, aligned data and valid proximity opportunities. A population network built only from collared animals also represents the sampled subset, not every individual. Uncollared animals can mediate interactions. Keep the sampling frame attached to any network conclusion.
CHAPTER 39 OF 49 · Extend the movement model
39. Check collar synchronisation and clock drift
Collars scheduled for the same interval may not record at exactly the same second. Device clocks can drift, time zones can be mishandled and daylight-saving conversions can shift records. A silent one-hour offset can manufacture or erase apparent proximity.
Store timestamps in an unambiguous standard, retain device time metadata and test known deployment events. Plot time differences between paired fixes. If a correction is applied, preserve the original field and transformation. Temporal accuracy is part of spatial contact evidence, not clerical housekeeping.
CHAPTER 40 OF 49 · Extend the movement model
40. Handle three-dimensional separation
Horizontal distance alone can misclassify animals separated vertically in trees, cliffs, water columns or buildings. Altitude and depth data have their own uncertainty and may be unavailable. A two-dimensional threshold should be described as such.
Where vertical separation matters and reliable measurements exist, use a three-dimensional metric or stratify by layer. Do not combine noisy altitude with precise horizontal positions without checking error. For many studies, the honest conclusion is recorded horizontal proximity rather than physical closeness.
CHAPTER 41 OF 49 · Transfer and continue
41. Treat sensor proximity as another measurement system
Bluetooth, radio or acoustic proximity loggers can record device encounters more frequently than GPS, but signal strength depends on orientation, vegetation, body position and hardware. A logged encounter is not automatically a fixed physical distance.
Calibrate devices under relevant conditions and preserve detection thresholds, missed-event estimates and false triggers. Combining logger and GPS data can strengthen interpretation if their clocks and identities align. Do not use one sensor as infallible truth for the other.
CHAPTER 42 OF 49 · Transfer and continue
42. Distinguish contact probability from contact count
The probability of at least one recorded contact in a day and the number of recorded contact events answer different questions. A pair with one long encounter may have many positive fixes but one event. A pair with three short encounters may have fewer positive fixes but more events.
Choose the response variable before analysis. Report exposure time and valid observation coverage. Count models, binary models and duration models make different distributional assumptions. A network edge weighted by raw fixes should not be interpreted as event frequency unless the event construction supports that meaning.
CHAPTER 43 OF 49 · Transfer and continue
43. Propagate uncertainty into network edges
Contact networks often draw an edge whenever a pair meets a threshold. Near-threshold location error, missing data and event rules make some edges uncertain. Treating every edge as certain can change apparent hubs and connectivity.
Use sensitivity networks under several justified thresholds or probabilistic edge weights where the model supports them. Report which conclusions persist. A visually central animal may be central only under one classification rule. This matters when network results guide surveillance or intervention.
CHAPTER 44 OF 49 · Transfer and continue
44. Compare seasons without changing the measurement
Seasonal movement can alter range size, overlap and fix success. Comparing a dry-season 30-minute schedule with a wet-season two-hour schedule mixes behaviour and observation design. Standardise or model the difference.
Use common temporal windows, compatible estimators and matched thresholds. Preserve reproductive stage, migration and management events where relevant. A higher wet-season overlap may reflect larger estimated ranges from different sampling rather than stronger association. Repeat the full evidence chain rather than comparing map colours.
CHAPTER 45 OF 49 · Transfer and continue
45. Validate selected events with independent evidence
Field observations, camera traps or environmental sensors can check some candidate encounters. Validation should account for each method’s detection field and identification limits. A camera 200 metres away cannot disprove a collar-based event outside its view.
Select events and non-events under a pre-defined scheme rather than validating only dramatic contacts. Preserve reviewer blinding where feasible. Independent evidence can estimate classification performance and reveal systematic errors, but it does not turn unvalidated records into certainty.
The analytical question does not override animal welfare. Device mass, attachment, capture, recapture and battery strategy require ethical and regulatory review. Higher fix frequency can improve temporal resolution while shortening deployment or increasing handling.
Choose the minimum data burden capable of answering the question. Use existing data when lawful and suitable. Stop rules and adverse-event reporting belong in the design. This article’s arithmetic examples deliberately omit capture operations; they should not be read as instructions to collar wildlife.
CHAPTER 47 OF 49 · Transfer and continue
47. Communicate results without exposing animals
A public map can show general patterns using aggregation, displacement or synthetic examples while authorised researchers retain precise data. The transformation should be documented so readers do not measure the display as if it were the analytical layer.
Explain what the map supports: possible shared-use areas, recorded proximity under a threshold or an uncertain network. Avoid headlines that turn two collar fixes into a story of friendship, conflict or infection. Responsible communication protects wildlife and preserves scientific meaning.
Does 50% overlap mean the animals meet half the time?
No. It describes a spatial overlap quantity with a stated denominator, not a temporal meeting proportion.
Can camera records verify a collar-based contact?
They can add evidence when time, location, detection field and individual identification are suitable. A camera’s non-detection does not automatically disprove presence.
Is a closer threshold always better?
No. It must fit the biological question and location accuracy. A threshold smaller than measurement error can create unstable classifications.
Can this method determine disease transmission?
No. It can characterise recorded proximity. Transmission also depends on pathogen, route, duration, susceptibility and other evidence.
CHAPTER 48 OF 49 · Transfer and continue
48. Continue through the verified wildlife routes
Return to The Gold Standard of Wildlife for responsible observation and conservation evidence. Use Bounding Habitat Residence Time Between Tracking Fixes for uncertain time within habitats, Bounding Travel Distance Without Counting Location Error as Movement for path and position uncertainty, Matching Habitat Use with Each Animal’s Available Environment for use–availability design, and Why Two Survey Visits Can Leave Occupancy and Identification Error Unresolved for repeated survey ambiguity.
Winner et al., Statistical inference for home range overlap develops overlap estimates and confidence intervals while accounting for autocorrelation and sampling limitations. Gilbertson et al., Trade-offs with telemetry-derived contact networks for infectious disease research distinguishes spatial overlap from spatiotemporal contact definitions and sampling demands. Yang et al., Individual-level patterns of resource selection do not predict hotspots of contact models contact locations separately from available locations inside overlapping ranges. The animals, thresholds and calculations in this guide are original teaching examples.