Socotra Plants: The Trees That Grow Nowhere Else

On the Diksam Plateau in February there is a plant whose trunk can measure 2.4 metres across, wider than most doorways, and which at that moment carries no leaves at all. What it carries instead are trumpet flowers, pink shading into magenta, up to 13 cm across, erupting straight out of bare grey wood. That is the Socotra bottle tree, and for a few weeks a year it does something that reads as a mistake: it spends its stored water on a flowering display in the middle of the dry season, when nearly everything around it has shut down.

This is the part of Socotra's flora that rewards attention. The island is usually described through a single silhouette, the dragon blood tree, and the description usually stops there. The more interesting story is in the mechanics: what these plants are storing, who pollinates them, what eats their seedlings, and why several of them are quietly running out of successors.

What shaped this flora

The short version is isolation plus drought plus rock.

Socotra's plant and animal lineages have been separated long enough for the divergence to be visible at species level rather than just at population level. The clearest documented case on the island is a bird, the Socotra golden-winged grosbeak, once lumped with its Arabian and Somali relatives and now recognised as a full species purely on the strength of how far isolation pushed it apart. The same evolutionary pressure has run through the flora, which is why the plants on this island have no close analogues you can point to elsewhere.

The second force is water, or the lack of it. Socotra's summer is severe, and the plants that dominate the island's imagery are all doing some version of the same trick: storing water in tissue and spending it slowly. The bottle tree's swollen trunk base, the caudex, is a reservoir. A mature specimen can hold hundreds of litres. That is what allows it to flower in the dry months rather than waiting for rain like a sensible desert plant.

The third is substrate and elevation. Most of the marquee species are limestone specialists, growing on rocky terrain, cliff faces and canyon walls rather than in soil. Elevation then sorts them: bottle trees run from sea level to roughly 1,000 metres, frankincense forms woodland between about 500 and 800 metres on the limestone plateaus, and the mist-catching upper slopes of the Hagher Mountains hold a different, moister assemblage again. You can drive across three floras in an afternoon.

The dragon blood tree, and the bird that keeps it going

Dracaena cinnabari is the most photographed plant on the island and the one most often written about badly. The usual paragraph describes the umbrella crown and stops.

Here is the part worth knowing. The dragon blood tree does not reproduce alone. The Socotra starling (Onychognathus frater) nests exclusively in the hollow branches of dragon blood trees, feeds on their berries, and disperses the seeds. That is an unusually tight dependency in both directions: one of the very few bird species anywhere that relies on a single tree species for its nest sites, paired with a tree whose seed dispersal runs through that same bird. Wherever dragon blood tree regeneration fails, the starling eventually follows. If you want the animal side of that relationship, it is covered in the Socotra wildlife guide.

Regeneration is the problem. The decline in dragon blood tree recruitment is climate-driven and is the reason the starling's own long-term outlook is flagged despite a currently stable population. The groves you photograph on the Diksam Plateau are, functionally, an aging population. The berries ripen in the late dry season, roughly March to May, and that is when the starling flocks concentrate in the groves and the whole system is most visible at once.

Worth noting for anyone planning a shoot: the tree's common name and its epithet cinnabari both point to the red resin it produces, which is where the "dragon blood" comes from. The research behind this guide does not carry a detailed account of the resin's traditional uses, so treat any confident claim you read elsewhere about specific historical applications with the scepticism it deserves.

The bottle tree and its narrow window

Adenium obesum subsp. socotranum is the second silhouette on the island and the one with the sharpest calendar.

It is the largest member of its genus by a wide margin. Specimens reach 4-5 metres tall on a caudex up to 2.4 metres across, which produces something closer to a miniature baobab than to the desert rose sold in garden centres. The trunk is smooth, pale grey-green, and water-storing. Leaves are dark green with a lighter midrib and appear only from mid-summer onward, in dense terminal rosettes.

The window is February and March. That is when it flowers while entirely leafless, which is the whole point: flowers on bare sculptural wood, no foliage competing for the eye. There is a secondary flush in October and November, and from May to September the plants look dormant and sparse. If your trip falls in the summer you will see the form but not the event.

Densities are highest on the rocky limestone of Diksam and at Homhil Protected Area, where large specimens grow alongside dragon blood trees and cucumber trees on the trekking path. The canyon walls at Wadi Kalisan, the cliff faces of Moomi Plateau and the rocky approach to Hoq Cave all hold specimens worth walking for, often growing directly out of crevices.

Two things about the bottle tree that do not make the standard write-up. First, the conservation picture is unresolved: the subspecies has no formal IUCN assessment, sources disagree about whether it should be treated as Vulnerable or as effectively data deficient, and no census exists. Anyone quoting you a firm status for the endemic subspecies is going beyond the evidence. Second, the plant's success as an ornamental is a threat to it. Sold internationally as "desert rose", it commands high prices in the succulent trade, and wild-harvested specimens are smuggled off the island.

Frankincense: eleven species, most of them in trouble

This is where Socotra's botanical significance stops being a curiosity and starts being a number. Socotra holds eleven of the world's twenty-four known frankincense tree species. Nearly half of all Boswellia diversity on Earth is concentrated on one island you could drive across in a few hours, and every one of those eleven species is endemic.

The one you are most likely to meet is Boswellia elongata, which forms a true woodland at Homhil between roughly 500 and 800 metres, on the standard trekking route. It is one of the few intact Boswellia forest stands remaining anywhere. The tree runs 4-8 metres with a spreading crown and pale buff to orange-brown bark that peels away in thin papery sheets, birch-like, which makes it one of the most texturally distinctive trees on the island.

The trait that matters most is invisible until you look at a trunk closely. Socotra's Boswellia exudes its amber-yellow resin naturally, without the bark being cut. Elsewhere in the world, frankincense harvesting means deliberately wounding the tree. Here the resin simply appears as hardened nodules on trunk and branches, and Socotri communities have collected it on that basis for millennia. Frankincense has been traded off this island for at least five thousand years, and the resin is regarded as among the finest anywhere: sweeter and more complex than East African or Omani material.

If you want the more aromatic species, Boswellia ameero grows in the Wadi Ayhaft valley and the Hagher valleys, and it is harder to reach. It produces a darker resin. Note that the research is not fully settled on which Socotra Boswellia is the right flagship, and taxonomy here is still moving: a 2025 study added another species to the count.

The conservation story is the important one, and it has its own detailed treatment in the frankincense conservation piece.

The cucumber tree, and a naming problem

Dendrosicyos socotranus is the island's third distinctive trunk, and it is commonly cited as the only member of the cucumber family that grows as a true tree. The research behind this guide does not carry a formal citation for that claim, so it is offered here as the received account rather than as a verified fact.

What is documented is more practical. Locally, "bottle tree" is applied to both Dendrosicyos and Adenium, which produces real confusion in trip reports and field notes. The two are unrelated lineages and easy to separate once you know the tell: the cucumber tree carries small yellow flowers, the bottle tree carries large pink to magenta ones. If someone tells you they photographed a bottle tree and the flowers were yellow, they photographed a cucumber tree. Both grow at Homhil, alongside dragon blood trees, which is why that one trekking path is the most efficient place on the island to see the three marquee endemics in a single walk.

The plants you will walk past without noticing

The photogenic four dominate the coverage, but they are not the vegetation. Most of Socotra's mid-elevation ground, roughly 200 to 800 metres and something like seventy percent of the island's area, is arid scrubland dominated by Euphorbia, Acacia and juniper. That is the matrix. Its fruits and seeds feed the golden-winged grosbeak, which is why the island's commonest endemic bird is a scrubland bird rather than a dragon blood tree bird.

The wadis hold a different community: dense riparian shrubland including Croton, the layer the Socotra chameleon lives in at one to three metres above the ground. On the coastal plains, the dominant vegetation is halophytic dwarf-shrubland, salt-tolerant low growth on silt and sand including Tamarix and Suaeda. Less than a metre tall, unremarkable to look at, and the entire global habitat of the Socotra cisticola, estimated at under 100 square kilometres.

The point is that the island's endemism is not concentrated in the four species on the postcards. It runs through the scrub as well.

What to see, and when

October to April is the dry season and the working window for plants. February and March are the peak: bottle trees leafless and in full bloom, fresh resin exuding from Boswellia trunks, sunbirds concentrated around the flowers. June to September is the monsoon, when interior wadi access is difficult and the flowering plants are dormant.

For the practical side of shooting all this, the photo guide covers gear and light. In short: wide for the caudex and the woodland, longer glass for resin nodules and peeling bark, and backlight for the bottle tree, which turns the pale trunk luminous and the pink flowers translucent. Morning light on the plateau is better than afternoon, with softer light and less wind-driven dust.

Species-level detail, IUCN status and per-plant photography notes are in the species database, and best places sequences the sites into a route.

The pressure

Three things are working against this flora at once, and only one of them is dramatic.

The dramatic one is cyclones. Chapala and Megh, in November 2015, destroyed 38 percent of the Homhil Boswellia elongata population outright. A further 29 percent died over the next two years. A two-thirds loss from a single storm season.

The slow one is goats. Livestock browse seedlings before they reach reproductive maturity, across Boswellia, Adenium and the wider shrubland. The result is populations composed almost entirely of aging adults with no replacement generation forming. A frankincense woodland that looks healthy in a photograph can be a stand with no future in it.

The third is that almost nothing here has species-specific protection. Socotra's UNESCO World Heritage listing provides umbrella cover, and that is largely it. CITES listing for the Boswellia was considered in 2025 and opposed, on the grounds that it would eliminate the legal trade that gives local communities an economic reason to protect the trees. That is a real argument, not a cynical one, and it is worth understanding before you form a view.

The trees will still be standing when you visit. The question the photographs do not answer is whether anything is growing up underneath them.