Click below for larger images and slideshow more...

A “tree cabbage” cultivated in the Canary Islands. This strain of cabbage probably represents a derivative of the ordinary Brassica oleracea, related to frost-free conditions.

A strain of tree cabbage has been cultivated in the British Channel Islands and can be grown in other places where frost is minimal.

Sonchus oleraceus is a common annual weed that is familiar throughout temperate zones. Like many weeds, it grows when the soil is wet, and forms seeds by the time conditions for growth become too dry. Sonchus oleraceus probably resembles the ancestors of the Sonchus that colonized the Canary Islands and Madeira.

Sonchus acaulis on the Canary Islands is a perennial, but lacks aboveground woody stems.

Sonchus tectifolius is a small shrub with several stems, adapted to growing in steep lava cliffs on Tenerife.

Sonchus pinnatus is a shrub branched at the base; its narrow leaf pinnae are suited to the sunny habitat.

Sonchus leptocephalus is shrub found in sunny places along the northern coast of Tenerife.

The leaves of Sonchus leptocephalus have very narrow segments, a morphology correlated with its sunny habitat.

During the drier season of the year, Sonchus leptocephalus loses its leaves; the stems are succulent and survive the dry season.

In the wet laurel forests of Tenerife, one can find the sparsely-branched rosette tree Soncus abbreviatus.

Sonchus abbreviatus compared in size to a Volkswagen. Having a terminal rosette of leaves is a way for a species derived from a rosette herb to reach the best illumination possible in a shady situation.

The leaves if Sonchus abbreviatus are thin and broad, features adapted to wet forest.

The summit of Madeira, viewed through an opening in the wet, dense, laurel forest.

Sonchus fruticosus is a rosette tree that grows in openings in the laurel forest of Madeira.

Scaevola mollis, a shrub on the upper slopes of Konahuanui, Oahu, is a shrub about one meter high, and is typical of the Pacific purple-fruited scaevolas.

Scaevola glabra is a tree that grows in very wet forest on the island of Kauai.

Scaevola glabra has leathery leaves like those of other wet-forest trees, an adaptation

Silene struthioloides is a small shrub on the floor of Haleakala, Maui; other Silene species are herblike, branched near the ground surface.

A transection of wood of Silene struthioloides; the vessels are very narrow (an adaptation to the dry conditions), not much wider than the fibers.

Cyanea leptostegia is one of the most spectacular of Hawaiian lobelioids. It forms unbranched palmiform rosette trees towering to 10 m or so in the cloud forest of Kauai.

A tangential section of the mature wood of Cyanea leptostegia shows a juvenile feature: all of the ray cells are vertically elongate.

Three plants of Viola maviensis from the Pepe’opae bog of Molokai. The plants start by growing horizontally like mainland violets but then bend upwards and become woodier.

The wood of Viola chamissoniana is entirely rayless. Raylessness is found in secondarily woody plants.

Paedomorphosis in wood is associated with slow subdivision of fusiform cambial initials and ray cells into vertically shorter cells. Before secondarily woody plants were studied, rays with an abundance of ray cells were not appreciated by wood anatomists. This chart places such rays (and raylessness) into evolutionary contexts.

The “tree lettuces” endemic to the Juan Fernandez Islands west of Chile form striking growth forms. This is a lowland Dendroseris, D. litoralis (photograph by Karl Skottsberg).

The most of amazing of the Juan Fernandez tree lettuces is this rosette tree, Dendroseries regia (photograph by Karl Skottsberg).

Much taller than a man is Begonia parviflora, which grows in cloud forest areas in the Andes of Peru. Each of the stems develops woodiness at the base.

The leaves of Begonia parviflora are large and thin, adapted to the misty cloud forest environment.

The flowers of Begonia parviflora are, as the species name indicates, small. Both male flowers (one open, above center) and female flowers (left, right, and bottom margins) are present in the inflorescence.

Plants of Coreopsis gigantea along of Malibu coast of southern California, flowering in late March toward the end of the rainy season. If one saw these plants at the end of summer, in September, one would see only the succulent stems bearing dead leaves. Succulence and drought deciduousness are mechanisms that can be readily evolved in order to adapt to seasonal dryness. Succulent stems like this are killed by frost (as can be shown by cultivation of C. gigantea inland), so this species occurs mostly on the island of southern California, but also in a few frost-free pockets on the nearby mainland coast..



The concept: working against the establishment.  Secondary woodiness is the process of increase in woodiness during the evolution of a group of plants.  The starting-point may be an herb, it may be an herb with some woodiness at its base, or it may be a small shrub.  The end-point may be a tree, or a rosette tree, or a large shrub.  If this doesn’t sound very startling to you, you need to know that in the 1950s and 1960s, botanists didn’t really believe in secondary woodiness.  They had the idea that woodier was, in any family of dicotyledons, more primitive, and that herbaceous members of any given family were specialized.  This curious but pervasive idea was in part a byproduct of the fact that most families are represented in the tropics, and most of them are woodier there.  The temperate zone is relatively rich in herbaceous species.  The herbaceous habit is how a plant gets through a cool winter—or a very long dry summer.  Botanists were located mostly in cities of the north temperate zone—they weren’t located in places where secondary woodiness is common.  I had the advantage of spending most of my life in southern California, where frost is not severe in most regions and therefore there are secondarily woody plants, such as woody sages and woody composites.  I also had the advantage of visiting the Hawaiian Islands and then other islands at an early age, and seeing instances of what could only be secondary woodiness. 
   You also need to know that before the 1950s, wood anatomists almost never studied wood of secondarily woody plants.  Wood anatomy was a field fostered by forestry, and often pursued in forestry institutes.  Such institutes saw little merit in studying wood that could not be used for construction or at least furniture.  And indeed, the timber industry subsidized forestry institutes, so the bias in favor of studying woodier plants and not studying less woody plants persisted.  Not only was I familiar with regions where secondary woodiness occurs, I have never been associated with a forestry institute, nor have I ever accepted (or been offered) money from the timber industry.  I viewed studying wood of plants that were secondarily woody—or even herbaceous with a small amount of wood at their bases—as a wonderful and largely unexplored opportunity.
   To be sure, I began my studies on wood anatomy by studying a family—Asteraceae—in which woodiness is relatively uncommon in most geographical areas, but in which some conspicuous instances of secondary woodiness do occur.  In fact, many temperate botanists are unfamiliar with woody composites, and think of the family as essentially herbaceous.  An ideal family in which to look at secondary woodiness.
    Conceivably, I could include this account of secondary woodiness in the Wood Evolution section of this website rather than the Island Biology section.  In fact, the phenomenon of paedomorphosis in wood overlaps to an appreciable extent with secondary woodiness, and paedomorphosis is covered in the Wood Evolution section of this website.  But secondary woodiness is remarkably characteristic of some oceanic island plants, and is certainly integral to the syndrome of evolution one finds on islands.  In a sense, it’s related to dispersal: nonwoody plants travel to islands by long-distance dispersal more readily than woody plants.  Many mainland forest trees mostly have large fruits and seeds that don’t disperse long distances.
   Above and beyond the connection with island features, secondary woodiness on islands was disbelieved by a number of workers.  The Europeans were notable in that regard.  I early saw that a number of the groups on the Hawaiian Islands exemplified secondary woodiness—the lobelioids and the composites most conspicuously.  The Hawaiian story was clear to me.  But long after I thought I had demonstrated the phenomenon of secondary woodiness on islands, some European workers firmly asserted that islands such as the Canary Islands were refuges of ancient plants, and if some of the plants, such as the laurels (which clearly do have a woody ancestry and are thus have primary woodiness), were once on the European mainland, then all of the woody species on the Canary Islands must be relictual, and their woodiness an ancient feature!  In fact, that contention by Europeans collapsed only recently, as evidence from DNA cladograms provided overwhelming evidence for secondary woodiness in such Canarian genera as Echium and Sonchus.  Those workers dismissed paedomorphosis in wood as of no significance.  And they conveniently never dealt with the Hawaiian flora—if they had, they would have been driven to truly amazing (mis)interpretations of the rather numerous instances of secondary woodiness on those islands..  However, times have changed, and secondary woodiness and paedomorphosis are now widely accepted.
   How and why secondary woodiness happens.  To survive a cold winter or a long dry summer, annuals die entirely and survive as seeds.  To survive winter cold, perennials die back to roots or stems.  Escape from frost is more probable for plant structures below the ground than above the ground.  Escape from frost for aboveground structures is greatest at ground level than higher up.  Dying back is not an economical strategy for a plant, so plants that die back to belowground structures or survive as seeds exhibit selection for survival under extreme conditions—and compete well in those conditions, therefore.  If cold or drought do not force particular plants to die back and sacrifice stems or foliage, those plants have a selective advantage.  Plants that die back must produce new leaves and then accumulate photosynthates each year before flowering.  Those that are not forced to die back can flower much sooner and therefore reproduce more abundantly.  Moreover, plants that do not die back reach sunnier levels and thus have a potential advantage compared with shaded plants beneath them. 
   Herbaceous plants which typically die back from cold on mainland areas, when they become island colonizers, are released from seasonality.  They can grow indefinitely.  This is shown rather amusingly by “tree cabbages.”  On the Canary Islands, one can see cabbages that form stems about two meters or so in height.  Yes, they’re the same species, Brassica oleracea, as the ordinary cabbages that don’t form such stems.  These are grown because they can furnish leaves throughout the year—a more convenient way of harvesting cabbage leaves than destroying a whole cabbage plant by cutting it off at ground level.  A similar strain of “tree cabbage” has been developed on the British Channel Islands, where frost is also minimal.  These tree cabbages eventually do flower—the flowering perhaps delayed because more energy is put into stem formation, or perhaps because there has been selection for delay in flowering. 
    Somewhat weedy herbaceous plants have reached islands in prehuman times and followed evolutionary pathways similar to those of the cabbages.  Weedy herbaceous plants are more likely to reach islands than forest trees, although islands may have moist zones ideal for forest tree growth.  Weedy herbaceous plants have abundant seeds that are small because habitats suitable for weedy plants are widely scattered, and tend to be sunny.  Small seeds are adapted to reaching widely scattered habitats.  Forest trees generally have larger seeds that are suited for germination and seedling growth in shady growth, and the larger seeds that forest trees have tend to be deposited in suitable habitats, most of which tend to be stable and near the parent tree.  If one looks at the kinds of seeds that reach islands, one sees a bias in favor of relatively small seeds (often formed in small dry fruits).  Not an overwhelming bias, but certainly a shift compared with the seeds and fruits one finds in mainland areas of similar climate.  To the extent that the spectrum of colonizers is richer in herbaceous plants and poorer in woody species than what the ecology of the recipient island potentially can support, there will be a change to increased woodiness by the herbaceous island colonizers, which will follow the economy and release from seasonality criteria described in the preceding paragraph. 
   One rather wonderful series of examples can be seen in the species of Sonchus on the Canary Islands and Madeira.  Sonchus oleraceus is one of the most familiar of garden weeds.  A short-lived annual with ragged outlines of leaves and yellow flower heads—if you think you haven’t seen it, you’re probably wrong.  Something similar to Sonchus oleraceus was probably the ancestor of the endemic species of Sonchus on the Canary Islands and Madeira.  The Canarian S. radicatus is common on cliffs.  It has a woody base, but isn’t shrubby.  The colonizer of the Canarian Sonchus species may have looked something like S. radicatus.  Several shrubby species of Sonchus have developed: S. arboreus in brighter localities, with narrower leaf lobes; S. tectifolius with broader leaf lobes in shady canyon situations.  Sonchus leptocephalus is a shrub that has succulent stems; its leaves are drought deciduous.  In the laurel forests of the Anaga peninsula is a species that can become a rosette tree (not a tall one, however), S. abbreviatus.  In the steep shady wet valleys of Madeira, near waterfalls, one can find S. fruticosus, which can become a genuine tree to 4 m tall.  What Sonchus shows us is that the difference between woodiness and herbaceousness is not a real one.  Even annuals such as S. oleraceus can develop a woody cylinder—all of the Sonchus species have secondary growth, they differ merely in having different amounts of secondary xylem. 
   On the Hawaiian Islands, there are many types of plants that have secondary woodiness.  Shrubby species of Scaevola (Goodeniaceae) have spread across the Pacific; the purple-fruited species are endemic to particular islands or island groups, whereas the white-fruited beach Scaevola, S. sericea, occurs on the coral sand of many islands.  The shrubby Scaevola species of the Pacific are lower in stature than S. glabra, which grows in shady wet forest situations.  The leaves of S. glabra are leathery, like those of typical wet forest trees; the flowers are adapted for bird pollination; and the fruits are large, an example of loss of dispersibility. 
   Dry situations in the Hawaiian Islands are host to plants with secondary woodiness, such as the rather amazing small shrub Silene struthioloides on the floor of Haleakala, Maui.  Its wood suggests that the conditions are dry throughout the year.  Secondary woodiness is possible where the climate is never really wet, as indicated by the woody Caryophyllaceae [ PDF ] and Lamiaceae of the Canary Islands. 
    Among the most amazing examples of secondary woodiness in the Hawaiian Islands are the lobelioids.  Shown here is a palmiform type, Cyanea leptostegia.  However, others are rosette shrubs, some much branched.  Not to mention the fabled Brighamia, a succulent in a class of its own.  All of the Hawaiian lobelioids have wood that shows paedomorphosis for the entire length of life of the plant (wood-anatomy-of-lobelioideae_1970) [ PDF ]. Some of the criteria of secondary woodiness have been mentioned elsewhere on this site (Wood Evolution—the subhead Paedomorphosis).  Species that represent phylesis from small shrubs to larger shrubs may not show juvenilistic features in wood anatomy very prominently. However, those groups that represent shift from herbaceousness to greater woodiness typically do show those features.  The flat or descending curve [ PDF ] in vessel element length and length of associated imperforate tracheary elements may also be related to lack of selection for maximal strength; one finds an ascending curve in “true” trees, where longer tracheary elements and alternate pitting of vessel elements seem indicative of increased strength.  Ray cell histology is a sensitive indicator of paedomorphosis.  Upright (vertically elongate) ray cells form a layer on the sides of multiseriate rays in most dicotyledonous woods, but if a ray consists predominantly or exclusively of upright cells for most or all of the lifespan of the plant, paedomorphosis indicative of secondary woodiness is very likely operative and secondary woodiness is indicated.  
   Taken one step further, the cells that would have matured into ray cells may be so very vertically elongate, so that the cells, when they mature, are fibers rather than ray cells.  In this case, the wood is rayless.  Thus, raylessness is an indicator of paedomorphosis—or a type of it, anyway.  Rayless woods are present in the Hawaiian violets, most of which are woody, some to 3 m.  The bases of these violets tell us a rather amazing story, because their developmental history as well as their evolutionary history can be seen there.  These violets begin with horizontal rhizomes, like those of continental herbaceous violets.  Then the stems turn upward.  This feature is unique to the Hawaiian species of Viola.  Plants with rayless woods are relatively common on islands, notably Plantago (wood-anatomy-of-plantago_1970 ???) [ PDF ] and Caryophyllaceae [ PDF ].  Rays conduct photosynthates horizontally in a woody cylinder, but if the woody cylinder has a finite diameter, lack of rays is not disadvantageous.  Rays eventually develop in some species that begin rayless—this can be found in the Hawaiian species of Cyrtandra. 
   The woody lettuces, Dendroseris, of the Juan Fernandez Islands, represent the diversification of a cichorioid colonization, and show that the growth forms involved in secondary woodiness are related to climate.  The Juan Fernandez Islands are cooler and wetter than the Canaries, and that shows in the fact that all of the Dendroseris species could be called rosette trees.  Most of them are comparable in woodiness and degree of branching to Sonchus abbreviatus in the Canarian laurel forest and S. fruticosus in similarly moist localities on Madeira. 
      Secondary woodiness may be found in the central Andes from Peru to Colombia.  Tree Papaveraceae (such as Bocconia), Asteraceae, Gentianaceae (Macrocarpaea), and Piperaceae represent this tendency.  Perhaps most startling is the tree begonia, Begonia parviflora.  Cloud forests in Andean mid-elevations are moist throughout the year and, like the east African volcanoes, are seasonless.

Secondary woodiness can be found in areas with little frost, even though seasonally very dry. Plants can more readily develop resistance to seasonal drought than to seasonal freezing.  One interesting plant is this regard is Coreopsis gigantea, which is a much larger plant than other species of Coreopsis.  It occurs mostly on the Channel Islands of Southern California, but is also native to a few restricted sites on the mainland along the Malibu coast.  Coreopsis gigantea can be killed by frost, so it is essentially limited to frost-free zones, which are more common on the offshore islands than on the mainland.  As a stem succulent, it takes up water during the winter months in the Mediterranean-type climate of southern California, and therefore would be likely to freeze in inland sites.  It is adapted to summer drought, much like Sonchus leptocephalus, by drought deciduousness.