Islands make up about 6% of Earth’s land, yet a new checklist finds 4,128 native bee species on just 305 of them, about one in five described bee species, and 1,742 that have never been recorded on a mainland.

Peer reviewed · Observational, compiled-data study (checklists and museum, database and photo records) · 305 islands

Islands hold a far bigger share of the world’s bees than their size suggests, according to a global analysis published in Nature on 18 November 2026. Leon Marshall of Naturalis Biodiversity Center in Leiden, Nicolas Vereecken of the Université libre de Bruxelles and seven colleagues combined species checklists with more than 600,000 bee records. About 42% of the island species live only on islands, many of them in Madagascar, New Guinea and Hawaii. The catch: much of the tropics is thinly sampled, so the true counts there are probably higher.

What you need to know

  • Finding: 4,128 native bee species were recorded on 305 islands, roughly 20% of the about 21,000 described bee species; 1,742 of them are known only from islands.
  • Method: The team matched expert-vetted checklists (mainly the Discover Life bee database) to a global island map, added 663,189 georeferenced records, and used statistical models to relate species counts to island area, isolation, biome and climate.
  • Source: Marshall et al., “Global patterns and drivers of bee richness and endemism on islands”, Nature, 18 November 2026 (open access).
  • Why it matters: The islands with the most bee species (Mediterranean ones such as Sicily and Cyprus) are not the ones with the most species found nowhere else (large tropical islands), which changes where protection would count most.
  • Limitation: This is a compilation of existing records, not a new field survey; sampling is uneven, and the drivers it identifies are statistical associations, not tested causes.

Evidence at a glance

Evidence summary for the global island bee study
ItemAssessment
Evidence statusPeer reviewed. Observational study of compiled data: checklists, published island faunas, museum and database records and Research Grade iNaturalist photos. A preprint version was posted on bioRxiv in March 2026.
Primary sourceLeon Marshall, Nicolas J. Vereecken and seven co-authors, “Global patterns and drivers of bee richness and endemism on islands”, Nature, published 18 November 2026, doi:10.1038/s41586-026-11325-6. Received 28 February 2026, accepted 16 October 2026. CC BY-NC-ND 4.0.
Evidence strengthModerate, because the counts rest on expert-checked lists and the main patterns held after adjusting for sampling effort, but the data are observational and some islands could only be analysed as whole archipelagos.
Principal limitationUneven sampling. Parts of Oceania, Indomalaya and Australasia are under-sampled, and better-studied islands showed higher counts, so tropical totals are likely underestimates.
Coverage at a glance0 independent credible newsroom reports found as of 20 November 2026, 07:00 UTC (SciQuest web searches in English). One Naturalis Biodiversity Center press release (18 November) appears on EurekAlert!, NatureToday and Phys.org; copies of the same release were not counted. Coverage is not evidence that a finding is correct.

Sources: the Nature paper (full text, figure legends, methods and author information); Naturalis release on EurekAlert!; bioRxiv record 10.64898/2026.02.28.708622; SciQuest coverage search at the time stated.

Is this a big deal?

It is a solid step rather than a surprise. Island biology has long leaned on birds, mammals and plants, and the authors note that global summaries of insects on islands are “virtually absent”. This paper gives bees a worldwide baseline that others can test. Its most useful result is a split: the places where bee species pile up and the places where unique bee species pile up are largely different.

What scientists found

The richest island for bees is Sicily, with 636 native species on about 25,000 km2. Cyprus, at about 9,300 km2, has 384, the most of any oceanic island (one never joined to a continent by a land bridge). Both sit in the Mediterranean-type biome, where limited but predictable rain brings a short, intense flowering season. Yet few of their bees are unique: Cyprus has 23 island endemics, and Crete 14.

The tropics flip that pattern. New Guinea has 284 native bee species, and 210 of them are found nowhere else. Madagascar has 226, of which 201 are endemic. On the Hawaiian Islands, all 65 native wild bees belong to one lineage of yellow-faced bees, Hylaeus (Nesoprosopis), and every one is endemic to the archipelago.

Picture two kinds of island. A Mediterranean island is like a busy train station near a big city: plenty of bees from the nearby continent arrive, but they keep mixing with mainland relatives, so few become new species. A remote tropical island is more like a rarely visited outpost: few bees get there, but those that do have room to split into many local species. The comparison has limits, because the paper describes these routes as likely explanations, not processes it measured.

Island bees: the richest islands are not where the unique species are Left: native bee species on selected islands, and how many live nowhere else. Right: what a tenfold larger island adds, by biome. Native species vs endemic species all native species endemic (found only there) CyprusMediterranean-type 384 23 Canary IslandsMediterranean-type 128 57 New Guineatropical 284 210 Madagascartropical 226 201 Maluku Islandstropical 102 62 Hawaiian Islandsall one lineage, Hylaeus 65 65 Cyprus count = single-island endemics; archipelagos count species endemic to the whole island group. Sicily has the most species (636). Make an island 10 times bigger and expect about this many times more species Mediterraneanslope 0.50 ×3.2 Tropicalslope 0.28 ×1.9 Temperateslope 0.18 ×1.5 Desertslope 0.02 ×1.05 Borealslope 0.02 ×1.05 ×1 would mean no gain. Multiplier = 10 to the power of the slope, from the paper’s model (other factors held fixed); approximate, because the model works on log(species + 1). An association, not a measured experiment. Data: Marshall et al., “Global patterns and drivers of bee richness and endemism on islands”, Nature (2026), doi:10.1038/s41586-026-11325-6.Multipliers: SciQuest calculation from the paper’s published slopes. SciQuest original graphic.
Bee-rich Mediterranean islands such as Cyprus have few species of their own, while New Guinea, Madagascar and the Hawaiian Islands are dominated by endemics. Island size also matters far more for bee numbers in Mediterranean-type biomes than in deserts. Data: Marshall et al., Nature (2026). The size multipliers are SciQuest calculations from the paper’s published model slopes and are approximate. SciQuest original graphic.

Size matters too, and how much depends on climate. In the team’s model, a Mediterranean-type island ten times larger than another is predicted to hold about 3.2 times as many bee species. For tropical islands the figure is about 1.9, for temperate ones about 1.5, and for desert and boreal islands area makes almost no difference.

Show the maths: where the 3.2 comes from

The model links the logarithm of species number to the logarithm of island area. A slope (β) of 0.50 means that multiplying area by 10 multiplies species by 100.50 ≈ 3.16. The paper’s slopes are 0.50 (Mediterranean-type), 0.28 (tropical, 100.28 ≈ 1.91), 0.18 (temperate, 100.18 ≈ 1.51) and 0.02 (desert and boreal, 100.02 ≈ 1.05). Because the model uses species + 1 and holds other factors fixed, these are approximate predictions for a typical island, not exact counts.

Try it: how much does island size matter for bees in each climate?

Predicted: island B holds about 3.2 times as many native bee species as island A.

How this is calculated: species multiplier = (area ratio)slope, using the biome slopes published by Marshall et al. (2026), with isolation, island type and sampling held fixed. It is an approximate model prediction from a statistical association, not a promise for any real pair of islands, and ratios far beyond those seen in the data are extrapolation. Try 100 times larger in a desert.

Remoteness lowers the count as well. Richness fell as distance from the mainland grew, in the same way across all biomes, though in most biomes the effect was weaker than that of area. Oceanic islands held fewer species than continental islands or continental fragments such as Madagascar.

How they found it

Nobody has visited all these islands with a net. The team, which includes John Ascher of the National University of Singapore, co-author of the Discover Life world bee checklist, started from that checklist and cross-checked it against published faunas for places such as the Azores, Fiji and Madagascar. “Some islands are rarely if ever visited by specialist bee collectors,” Ascher said in the Naturalis press release. Validated iNaturalist photos helped fill gaps, though the team labelled those records as uncertain.

They matched each list to a global island map, kept islands of at least 5 km2 and at least 1 km offshore, and took area, isolation and plant data from the GIFT flora database. They also gathered 663,189 georeferenced bee records from GBIF, the Atlas of Living Australia and iDigBio to score how thoroughly each island had been sampled. That score went into the models, so the main patterns could be checked against a simple map of where collectors had been.

The richness model explained 66% of the variation between islands using area, isolation, island type, biome and sampling, and 75% once it allowed for differences between archipelagos.

Why it matters to you

This won’t change anything in your garden this year. Its value is in where conservation attention goes. Only 542 bee species have been assessed for the global IUCN Red List, including just 68 island endemics, so most island bees have no official risk status. The authors argue that protection should prioritise islands with unique evolutionary histories, from Madagascar to Hawaii, over those with the longest species lists. In the United States, most federally endangered bee species are native Hawaiian endemics.

The study also counted at least 54 introduced bee species on 143 islands. The western honey bee was by far the most widespread, introduced to 127 of the islands, and the authors cite evidence that it can harm native island wildlife.

What the evidence does not show

The study does not show that area, isolation or climate cause the patterns; it shows they go together in the data. The plant result needs the same care. Across 175 islands, plant diversity alone accounted for 49% of the variation in bee diversity, and Mediterranean-type islands held about 5.6 times more bee species than tropical islands with similar plant numbers, area and isolation. The authors say shared drivers and real dependence on flowers cannot be separated here.

It also does not measure decline. These are presence lists, not population counts over time, so they say nothing about whether island bees are increasing or falling. And because some data could only be assigned to whole archipelagos, island-level numbers for places such as the Solomon Islands are coarse.

The headline numbers vary slightly by version. The paper’s summary records 4,147 species, while its analysis counts 4,128 as native after classifying introduced species. The March preprint gave 4,140 species on 306 islands.

How this fits what we already knew

On continents, bees are unusual: their species peak not in the tropics, as most animal and plant groups do, but in dry, seasonal mid-latitude regions. The island data mirror that, with richness peaking at mid-latitudes in the Northern Hemisphere. One explanation is that short, intense flowering seasons let many specialised bees share out resources; the authors say trait studies are needed to test it.

The partial split between richness and endemism is where bees differ from island plants, whose richest large tropical islands also tend to host the most endemics. Endemic island species can be fragile: SciQuest has reported how most island species of barn owls in the Caribbean died out.

Independent perspective

We did not obtain comment from an independent bee researcher before publication. Nature lists Bo Dalsgaard and anonymous referees as the paper’s peer reviewers, and their reports are published alongside the paper. We will add outside comment if we receive it.

How other headlines framed it

We found no independent newsroom coverage at the time of writing. The Naturalis press release on EurekAlert!, also carried by Phys.org and NatureToday, leads with the more than 1,700 island-only species and calls the work the first global assessment of island bees. Nature‘s own summary line puts the 20% share first, and the paper’s abstract describes “a global checklist” without claiming to be first.

What happens next

The authors call for digitised museum collections, targeted trips to hard-to-reach islands and more iNaturalist records to close the sampling gaps. Combined with new family trees of bees, the data could show when lineages reached islands, split into new species or died out. The checklist and code are public on Zenodo, so others can test the patterns.

In one sentence

Islands hold about a fifth of the world’s described bee species, but the bee-packed Mediterranean islands and the endemic-rich tropical ones are different places, and both are still unevenly counted.

Quick check: did it click?
  1. Which island has the most native bee species in the study? Sicily, with 636.
  2. Why does Cyprus have many bee species but few of its own? The authors suggest frequent arrivals from nearby continents keep island and mainland populations mixing, which limits the formation of new local species.
  3. Does the study prove that island size causes higher bee numbers? No. It shows a strong statistical association in compiled records; it is not an experiment.

Read next

Sources and disclosure

  1. Marshall, L. et al. Global patterns and drivers of bee richness and endemism on islands. Nature (2026), published 18 November 2026. Data and code: Zenodo.
  2. Marshall, L. et al. Global Patterns and Drivers of Bee Diversity and Endemism on Islands. bioRxiv preprint, March 2026.
  3. Naturalis Biodiversity Center. First global assessment of island bees finds more than 1,700 bee species are unique to islands. EurekAlert!, 18 November 2026.

Disclosure: This article was drafted with AI assistance and checked against the sources above. The size multipliers in the graphic, the “Show the maths” box and the calculator are SciQuest calculations from slopes published in the paper. The study authors declare no competing interests. Spotted an error? See our corrections policy.


Discover more from SciQuest

Subscribe to get the latest posts sent to your email.