Scientists estimate how many humpback whales live across the North Pacific by combining photographs of identifiable whales with information about which individuals were seen again. No research team can watch the entire ocean at once. A catalogue provides a record of the animals documented, and a statistical model helps estimate how many were missed. Both depend on where and when observers looked, how well they could identify whales and which animals had a chance to appear in more than one sample. These details explain why a photograph from British Columbia can be valuable to a large study even though it shows only one whale.
The photograph is a record of one animal
A humpback often exposes the underside of its tail flukes when it begins a deeper dive. The markings and shape of the flukes help distinguish it from other humpbacks. A photograph with clear features, a reliable date and a useful location can be matched to images from earlier encounters. When researchers agree on the match, the two observations contribute to the history of one whale. A photograph showing a humpback with no identifiable features may still document the species in an area, but it cannot play the same role in an individual encounter history. North Pacific humpback photo-identification dataset study
Two images of the same flukes on the same day are often different views of one encounter. The North Pacific dataset study defines an encounter in terms of an identified individual recorded on a specific day and at a specific location, supported by an image. A different day or location can produce another encounter with that animal. The rules matter because a photographer can take many frames while the number of whales seen stays the same.
A study may also keep images that support a probable match separate from ones that support a firm match. If a researcher mistakenly treats the same animal as two individuals, the dataset has too many apparent whales and too few repeat encounters. If two different whales are treated as one, the reverse problem occurs. Consistent decisions about image quality and identity help keep those errors from shaping the estimate.
Why repeated matches help estimate abundance
Imagine two sets of photographs taken during periods defined by a study. The first contains 30 identified whales. The second contains 30, including 15 seen in the first set. Those repeated matches tell you there is considerable overlap between the sets. If the second set contained only three previously seen whales, the overlap would be smaller. Under suitable assumptions, a smaller share of repeat observations can indicate that the two sets came from a larger group of animals. It can also point to differences in where and how the researchers collected the records.
This approach is called photographic capture–recapture, or mark–recapture. In this setting, “capture” means documenting an individual in a photograph. The animal’s natural markings allow the researchers to recognise it again. The proportion of returning IDs helps a model estimate the whales that were available to be seen but did not appear in the sample. The overlap is informative only if the model accounts for the way the samples were collected and which whales could show up in each one. Barlow and colleagues’ North Pacific capture–recapture study
The arithmetic in a simple example is easier than the research problem. Whales can enter or leave an area between observation periods. Some are more likely to be photographed than others. A model may account for migration, differences in detection or changes over time, depending on the question. The result should be read with the stated assumptions and an interval or other measure of uncertainty where the authors provide one.
Why coverage matters across the North Pacific
Humpbacks use seasonal feeding and breeding areas across the basin. A whale seen in British Columbia may later be photographed in Hawaiʻi or Mexico. A feeding area can contain animals from more than one wintering group, and a wintering area can contain whales from more than one feeding region. If each organisation kept its catalogue isolated, many cross-season matches would be hard to find. Collaboration makes it possible to connect records that otherwise look like separate observations of different whales. Canadian North Pacific humpback whale status assessment
The SPLASH project brought together teams across the North Pacific to study abundance, population structure and other questions. Researchers worked in both feeding and wintering areas and used comparable methods. This made it easier to build a wider set of identification histories. It also gave scientists a way to see which parts of the ocean and annual cycle had good coverage and where observations were sparse. NOAA’s North Pacific SPLASH project archive
Coverage is especially important when comparing years. More groups of photographers can produce more distinct IDs even if abundance is stable. A region with poor weather may generate fewer usable photos despite having whales present. A new study that includes previously underrepresented sites might find additional animals simply because people looked in more places. Research that models abundance tries to handle differences in the probability of documenting a whale.
How a large catalogue becomes a scientific dataset
A collection may contain images from research organisations, professional photographers, tour operators and other contributors. The images have to be matched and linked to encounters before the collection can support a particular analysis. Good records include enough location and time information to place an encounter in the right region and period. Reviewers also need to identify duplicates, flag unclear photographs and decide which data meet the requirements for the study.
A photograph can be worth retaining even when it is excluded from one model. An unclear fluke image might support an account of behaviour or a note that a whale was present. A clear photograph taken outside the years covered by an analysis might add to a later project. A report that describes how many records it began with and which records it used gives readers a way to understand how those choices affect the result.
Individual histories also allow researchers to ask questions besides abundance. They can look at how often whales return to a feeding area or which wintering grounds are linked to a set of feeding locations. These analyses can use some of the same records while grouping observations differently. A migration match is evidence about two places used by one animal. An abundance estimate involves patterns of detection and matching across the study population.
How to read a published number
Start by identifying the area and the period. The estimate might cover the whole North Pacific, animals associated with a particular wintering ground or whales using a smaller feeding area. These results can overlap because a whale can use more than one place. Adding estimates from feeding and breeding areas can count the same animals twice. Read the description of what the study treated as its population or study unit before comparing numbers.
Check whether the number is a total of documented individuals, an estimate of abundance or an estimate of change. A collaborative study modelled annual abundance for North Pacific humpbacks from 2002 through 2021. Its results are modelled estimates based on a large identification dataset. They answer a different question from the number of photographs the project received or the number of distinct whales represented in all years combined. North Pacific humpback population modelling study
Then look for uncertainty. An interval gives you information about how precisely a study estimated the result under its approach. Two estimates may differ without demonstrating a clear change, particularly if they use different areas or methods. Conversely, a long series of comparable observations can support a trend even if the exact number in each individual year is uncertain. When you quote a figure, include enough context for someone to tell which result it represents.
How other surveys add context
A ship or aerial survey can record animals along planned routes. Researchers can use the locations and effort to estimate density or distribution, depending on the survey design. That provides a different view from a collection of repeated photographs. A survey may establish that whales are concentrated in particular waters without identifying every one. An identification project can follow an individual across seasons while leaving some areas with little coverage.
An acoustic study can detect whales that are vocalising within range of a hydrophone. The result can provide information about presence and seasonality, even when no vessel is available to photograph them. Repeated songs may come from the same whale, so a count of detections is a different measure from a count of individuals. A combination of methods helps researchers understand where their strongest evidence lies and what they may have missed.
What one Campbell River encounter can contribute
If a humpback lifts its tail during a trip, a clear image from your existing viewing position may provide evidence of identity. Keep the time, general location and uncropped original if you plan to share it. If the animal shows only the back, you can still record what you observed. Follow the crew’s viewing instructions and let the encounter determine which view is possible. A tour provides one moment in a much longer record of whales using the coast. Eagle Eye’s Campbell River whale watching tour
Ask the guide whether they recognise the whale and how they can tell. A known animal might have a long history of returning to British Columbia. A first good photo can establish a new encounter without indicating where the whale spent its previous years. That is one way the local view connects to the larger study. One observer adds an accurate record, and many observers and researchers make it possible to compare the records across an ocean. Marine Education and Research Society’s data contribution guidance
Why the timing of each sample can change the answer
Suppose observers collect one group of photographs early in a feeding season and another late in the season. The proportion of matches could reflect whales coming and going as well as the number present. If the research question is how many use the area at some point in the whole season, those movements matter in a different way than they would for an estimate on a particular date. The authors need to define which of those quantities they are estimating.
The same issue arises across years. A series can contain a strong early sample from one feeding area and a stronger later sample from another. Some whales will have different chances of being documented in the two periods. A study that accounts for those differences can make the change easier to interpret. Check the methods before assuming the raw number of matches and the estimated population changed in the same direction.
Frequently asked questions
Are whales caught to count them?
In photographic capture–recapture, a capture is a photo-supported identification. A recapture occurs when researchers recognise the same whale in a later sample. Its natural markings serve as the reference.
Is the number in a catalogue the population size?
The catalogue records animals that were documented. Population estimation uses additional information about repeat encounters and the chances of detecting whales. Check the geographic and seasonal scope before comparing catalogue totals with estimates.
Why can local sightings go up while a wider estimate falls?
Whales can move among feeding areas and observers can change where or how often they look. A local count tells you about encounters in one setting. The basin-wide estimate uses evidence from several areas and periods.
Can a single photograph contribute to a larger study?
A dated and located photograph with an identifiable whale can add an encounter. Its value for a particular study depends on the image, the supporting information and whether it falls within the areas and periods researchers are analysing.
