120 years of untangling the divaricate habit: a review - UC Research Repository

Page created by Jeanne Harvey
 
CONTINUE READING
Available online at www.science.canterbury.ac.nz/nzns

120 years of untangling the divaricate habit: a review
Kévin J. L. Maurina* & Christopher H. Luskb
a
 School of Science, The University of Waikato, Hamilton, New Zealand;
b
 Environmental Research Institute, The University of Waikato, Hamilton, New
Zealand
*The University of Waikato – School of Science, Private Bag 3105, Hamilton
3240, New Zealand, kjlm1@students.waikato.ac.nz
(Received August 2020, revised and accepted January 2021)

    The evolution of divaricate plants in New Zealand has been the subject of long-
    running debate among botanists and ecologists. Hypotheses about this remarkable
    case of convergent evolution have focused mainly on two different types of selective
    pressures: the Plio-Pleistocene advent of cool, dry climates, or browsing by now-
    extinct moa. Here, we review the scientific literature relating to New Zealand
    divaricates, and present a list of 81 taxa whose architectures fall on the divaricate
    habit spectrum. We recommend a series of standardised terms to facilitate clear
    communication about these species. We identify potentially informative areas of
    research yet to be explored, such as the genetics underlying the establishment and
    control of this habit. We also review work about similar plants overseas, proposing
    a list of 53 such species as a first step towards more comprehensive inventories;
    these may motivate further studies of the ecology, morphology and evolutionary
    history of these overseas plants which could help shed light on the evolution of
    their New Zealand counterparts. Finally, we compile published divergence dates
    between divaricate species and their non-divaricate relatives, which suggest that
    the divaricate habit is fairly recent (< 10 My) in most cases.

Keywords: convergent evolution; divaricating shrubs; heteroblasty; moa; New
Zealand; structural plant defences

Introduction                                                Central Europe, where plants do not show
                                                            similar architectures. These plants are
The earliest mention we have found of                       nowadays recognised as a collection of shrubs
what we call today “divaricating plants” or                 and early growth stages of heteroblastic trees
“divaricates” was made in 1896 by German                    bearing small leaves on tangled branches
botanist Ludwig Diels. He described them                    diverging at wide angles.
as “systematically distant descendants of                      Such a case of convergent evolution
the New Zealand forest flora that converged                 naturally attracted much attention from
towards a xerophytic structure” (Diels 1896,                local and overseas botanists and ecologists.
pp. 246-247, translated from German).                       The centre of this attention was to identify
He expressed surprise at seeing apparently                  putative selective forces that may have
drought-adapted species in climates that                    driven this evolution. Diels (1896) initially
are generally more humid than in his native                 proposed drought as the main selective factor,

                New Zealand Natural Sciences (2021) 46: © New Zealand Natural Sciences
New Zealand Natural Sciences 46 (2021)              2

and McGlone & Webb (1981) considered             Characterising the diversity of
that frost and wind might also have been         divaricating habits: variations on a
important. Diels’ climatic hypothesis remained   New Zealand theme
largely unchallenged until Greenwood &
Atkinson (1977) developed the moa-browsing
hypothesis that several authors had previously   Past attempts at defining the divaricate habit in
hinted at (e.g. Denny 1964; Carlquist 1974;      New Zealand
Taylor 1975), igniting a passionate debate
                                                 “Divaricate” comes from a Latin root
that is still ongoing today. Concurrently, a
                                                 meaning “stretched apart”, which in botany
non-selective evolution process was proposed
                                                 refers to the usually wide angle at which
by Went (1971): the horizontal transfer of
                                                 branches of these species grow from the
“divaricate” genes; it however was strongly
                                                 stem on which they originate. Indeed, the
criticised on theoretical grounds (Tucker
                                                 branching angle of divaricating species is on
1974; Greenwood & Atkinson 1977) and
                                                 average more than 70°, sometimes over 90°
has not been empirically investigated so far.
                                                 (Bulmer 1958; Greenwood & Atkinson 1977),
Rationale for and content of this review         whereas their broadleaved relatives branch
                                                 on average at < 55° (Kelly 1994). However,
Although about 120 years have passed since       simplifying the definition of a divaricate
the first publications on the topic, the real    species by its branching angle is misleading:
debate around the evolution of divaricates       Pott & McLoughlin (2014) and Pott et al.
only started in the late 1970s. Yet, no recent   (2015) discussed the evolutionary adaptations
literature review (e.g. Wilson & Lee 2012)       of shrub or low-growing tree species of the
offers an exhaustive account of all the          extinct gymnosperm family Williamsoniaceae
scientific material published about these        by making a parallel between them and New
plants. The aim of this review is to provide     Zealand divaricates, claiming that they share
a comprehensive resource for anyone with an      similar architectures. Although the species
interest in divaricate plants.                   they described undeniably branched at wide
    First, we review past attempts at defining   angles, they did not look anything like what
the divaricate habit and describing its          New Zealand researchers call “divaricates”:
variability in New Zealand. We propose           they bore much larger leaves (4-25 cm long, cf.
a series of terms to try to standardise the      < 2 cm in most New Zealand divaricates), and
vocabulary to be used when discussing these      their branches were not interlaced. Likewise,
species (in bold in the text). We also report    many examples of extant species can be cited
and discuss observations of divaricate-like      as having wide branching angles while not
species overseas, compiling a list of such       satisfying the definition of a divaricate, e.g.
occurrences.                                     Araucaria heterophylla (Salisb.) Franco or Piper
    We then review the published hypotheses      excelsum (G.Forst.). Indeed, the divaricate
that have been formulated to explain how         habit in New Zealand is also defined by a
such a diversity of architectures was selected   collection of other traits, including: small
in the New Zealand flora, and comment on         leaves (leptophyll and nanophyll classes of
the weight of evidence for or against each       Raunkiær 1934); interlaced and abundant
hypothesis. Finally, we examine the handful      branching; relatively long internodes
of studies that, rather than focusing on the     compared to the size of their leaves (Kelly
evolution of these species, have looked at       1994 and references therein; Maurin & Lusk
developmental aspects of these peculiar          2020)—although some species show “short-
architectures. We conclude our review by         shoot development” (Tomlinson 1978), i.e.
pointing out new areas of research that might    stubby shoots with densely crowded nodes
enhance our understanding of divaricate          and leaves. The exact set of features used
plants.                                          to define the habit however varies between
                                                 authors (Kelly 1994; Grierson 2014; see Table
Maurin & Lusk: Untangling the divaricate habit: review                    3

1 in Supplementary Material for a list of traits    thus do not prevent the outgrowth of lateral
used by past authors). Finally, New Zealand         branches. This first definition of the term
divaricates are notably lacking in spines,          “filiramulate” emphasised the wiry branches
except for Discaria toumatou Raoul which            that may be flexuose to truly divaricating, and
has spinescent congeners in Australia and           divaricate plants were therefore considered
South America. Some divaricating species,           a type of filiramulate species. However, this
such as Melicytus alpinus (Kirk) Garn.-Jones        definition of “filiramulate” has not been
and Aristotelia fruticosa Hook.f., have been        widely adopted by the scientific community.
considered spinescent by some authors (e.g.             The lack of a consensus word-based
Greenwood & Atkinson 1977; Burns 2016),             definition of the divaricate habit led to two
but we argue that their pointed branchlets          attempts to find a mathematical quantification
are not sharp enough to pierce the skin and         of divaricateness. Atkinson (1992) focused on
therefore probably did not have the same            branch density (number of lateral branches
adaptive value as actual wounding spines or         subtended per cm of main branch) and
thorns.                                             branching angle; Kelly (1994) also focused
    Because the divaricate habit has evolved        on branching angle, and included leaf size
independently multiple times in the New             and density (the relative width of the leaves
Zealand flora, it appears under different           to the size of the internodes that bear them).
structural forms that were tentatively grouped      Although these two indices emphasise
by various authors to form classifications.         different features of the divaricate habit,
Bell (2008) recognised four branching               they correlate well for New Zealand species
pattern types in divaricate species: branching      (Kelly 1994; Grierson 2014). In spite of these
at wide angles (e.g. Aristotelia fruticosa), zig-   indices, which are rarely used in the literature,
zagging by sympodial (e.g. juvenile form of         consensus definitions of the divaricate habit
Elaeocarpus hookerianus Raoul) or monopodial        and its variations are still lacking.
(e.g. Muehlenbeckia astonii Petrie) growth, and
                                                    Heteroblastic divaricate species
“fastigiate”. The use of “fastigiate” (meaning
narrow branching angles) to categorise              Although most of the species showing the
divaricate plants may seem paradoxical,             divaricate habit keep it their whole life, some
but Bell’s (2008) example, Melicytus alpinus,       heteroblastic species produce a divaricating
sometimes does show a fastigiate habit in           form early in life, then later switch to a non-
shaded habitats. In our experience, however,        divaricating form (Cockayne 1958). Very few
in sunny environments M. alpinus has wider          quantitative data exist regarding the age before
branching angles and is compactly interlaced.       the non-divaricating form appears (Table 1),
Tomlinson (1978) tried to assign divaricate         which may depend on the degree of exposure
species to Hallé et al.’s (1978) architectural      to sunlight in many cases (Cockayne 1958), or
models, without success. Halloy (1990)              even on latitude at least in Sophora microphylla
defined five groups based on branching              Aiton (Godley 1979). We propose referring to
patterns and assigned one species per group         them as heteroblastic divaricate species; the
as examples, but his proposal has been largely      term “habit-heteroblastic” used by Philipson
ignored.                                            (1963) for such species is inadequate as it does
    These variations around the features            not mention “divaricate”, and the juvenile and
which characterise the divaricate habit led         adults forms of some heteroblastic divaricate
Wardle & McGlone (1988) to propose                  species do not only differ in architecture but
the word “filiramulate” to describe lianes          also in leaf shape (e.g. Pennantia corymbosa
and shrubs with reduced apical buds that            J.R.Forst. & G.Forst.). Both forms often
have some (but not all) of the traits usually       coexist on the same individual at least for
regarded as integral to the divaricate habit.       some time, and the transition can be abrupt
These reduced buds exert a weakened apical          (“metamorphic” species (Ray 1990), such
dominance (Wardle & McGlone 1988), and              as in Pennantia corymbosa; see Figure 1 in
New Zealand Natural Sciences 46 (2021)          4

Table 1. Published quantitative measurements and estimations of the age reached by heteroblastic
divaricate species before their adult form appears.

 Species                          Duration of the juvenile form    Reference
 Elaeocarpus hookerianus Raoul    At least 60 years, depends on Cockayne (1958)
                                  light conditions (source not
                                  specified by the author)
 Prumnopitys taxifolia (D.Don)    (1) Up to 60 years (source not (1) Dawson & Lucas (2012)
 de Laub.                         specified by the author)        (2) Lusk (1989)
                                  (2) At least 47 years (based on
                                  ring counts)
 Sophora microphylla Aiton        (1) ca. 15 years (source not (1) Cockayne (1958)
                                  specified by the author)            (2) Godley (1979)
                                  (2) variable according to location:
                                  from absence of juvenile form in
                                  some parts of the North Island,
                                  to ca. 3.5 years in the Auckland
                                  region and at least 23 years in the
                                  south-east of the South Island
                                  (based on field observations and
                                  a common garden experiment)

Supplementary Material), or gradual with           early form of some heteroblastic divaricate
transitional forms between the divaricating        species are capable of flowering, such as
bottom and the non-divaricating top of the         those of Pennantia corymbosa (Beddie 1958;
plant (“allomorphic” species (Ray 1990),           Cockayne 1958) or Plagianthus regius (Poit.)
such as in Hoheria sexstylosa Colenso). Day        Hochr. subsp. regius (Cockayne 1958): they
et al. (1997), studying the transition of the      should therefore not be termed “juvenile”.
heteroblastic divaricate Elaeocarpus hookerianus   However, alternative terms such as “young”
from its juvenile form to its adult form,          and “old” carry ambiguities of their own,
described a distinctive transitional form          so it is not obvious to us how to improve
characterised by a less plastic growth pattern     upon “juvenile” and “adult”, which have
than the juvenile form, while not showing          become deeply anchored in the literature. We
the morphological attributes that identify         however recommend the use of juvenile/
the adult form.                                    adult form instead of the more commonly
   The ubiquitous use in the literature            used juvenile/adult “stage” or “phase” to
of the adjectives “juvenile” and “adult”           avoid the confusion between growth habit
(sometimes “mature”) to name, respectively,        and reproductive state that Jones (1999)
the early divaricating form and the ultimate       pointed out.
non-divaricating form of heteroblastic                Two hypotheses have been proposed to
divaricate species, is potentially misleading.     try to explain the origin of heteroblastic
Jones (1999) criticised the use of “juvenile”      divaricate species:
to describe early forms of heteroblastic
species because it better characterises a phase    1.   Hybridisation between a divaricate
of plant development that is incapable of               species and a non-divaricate relative
sexual reproduction. She therefore suggested
that “juvenile” should be restricted to non-       It is well known that some divaricate species
flowering stages of heteroblastic species.         hybridise with broadleaved congeners (e.g.
Yet, it was observed in New Zealand that the       Dansereau 1964; see lists of known (and
Maurin & Lusk: Untangling the divaricate habit: review                     5

potential) hybrids compiled by Cockayne            arose from heteroblastic divaricate ancestors
1923, Cockayne & Allan 1934 and Greenwood          which later lost their forest-adapted adult
& Atkinson 1977). These hybridisation events       form in response to new selective pressures
were hence proposed as a source for the            in more open environments. It was first
origin of heteroblastic divaricate species         suggested by Cockayne (1911, p. 25–26;
(Godley 1979; 1985). Carrodus (2009)               1958, p. 141) and further developed by Day
addressed the question of whether Pittosporum      (1998a). It is difficult to see how to test such a
turneri Petrie, a heteroblastic divaricate         hypothesis, which may explain why it has not
small tree, is a hybrid between Pittosporum        been the subject of published research so far.
divaricatum Cockayne, a divaricating shrub,
                                                   The divaricate habit in New Zealand and overseas
and Pittosporum colensoi Hook.f., a broadleaved
tree. The study used plastid and nuclear DNA       Variations of the divaricate habit are found
markers as well as a morphological analysis        in ca. 81 taxa in New Zealand (Appendix
and found evidence supportive of such              1), including heteroblastic divaricate taxa.
an event, e.g. that P. tuneri shows an ISSR        80 are Eudicots, one is a Gymnosperm, and
band and morphological traits (for example         they represent 20 families. According to
in leaves, flowers and fruits) that combine        statistics about the New Zealand vascular
those of the putative parents. They however        flora produced by De Lange et al. (2006),
suggested more investigation: their cross-         this number represents almost 13% of
pollination experiments between P. divaricatum     indigenous woody spermatophytes. We
and P. colensoi did not produce progeny, and       refer to all these species as divaricates, a
given the limitations of the ISSR technique        term that encompasses architectures that
they recommend using more nuclear markers          fall on a spectrum with two extremes. On
in more individuals. Shepherd et al. (2017) and    one end, there are the true divaricates (or
Heenan et al. (2018) used chloroplast DNA          truly divaricating species), i.e. species with
and microsatellite markers respectively to         the most characteristic traits of the habit
study hybridisation and introgression events       (such as tightly interlaced tough branches
in New Zealand Sophora L.: even though their       with relatively long internodes compared
findings showed that these species hybridise       to leaf size, and leaves < 2 cm in length);
readily, they reported little support for the      typically shrubs that are common in open
hypothesis that the heteroblastic divaricate       environments such as forest margins. To
species Sophora microphylla arose through          characterise the other end of the spectrum,
hybridisation between divaricate species           we propose to use the term semi-divaricate
Sophora prostrata Buchanan and the non-            as used by Greenwood & Atkinson (1977);
divaricate species Sophora tetraptera J.F.Mill.    these are species with traits that are not as
    However, as Godley (1985) makes explicit,      typical as the traits of the true divaricates,
his hypothesis allows for multiple generations     such as slender branches in a more open
after an initial hybridisation and for selection   architecture, and larger leaves—sometimes
of the heteroblastic divaricate form from          species that appear clearly divaricate in open
a variable population of hybrid derivatives        areas tend towards a semi-divaricate habit
(such as a hybrid swarm). Therefore, genetic       when growing in the shade (Philipson 1963;
signal of a hybrid origin might be weak            Christian et al. 2006; pers. obs.).. Furthermore,
and difficult to detect in studies employing       we use the term divaricate habit to refer to
only modest numbers of genetic markers.            the habit as a phenomenon, which manifests
                                                   itself through a variety of architectures that
2.   Neotenous loss of a putative adult            we refer to as divaricating habits.
     non-divaricate form                               Although divaricates are present in a wide
                                                   range of environments throughout New
A mirror image of the previous hypothesis,         Zealand, several environmental patterns
this hypothesis states that divaricate species     in their abundance have been noted. They
New Zealand Natural Sciences 46 (2021)            6

can be found in most forest types and              reported by Tucker (1974), and the South
successional shrublands (Wardle 1991),             African shrub species with dense, cage-like
from the coast to alpine environments              architectures studied by Charles-Dominique
(Greenwood & Atkinson 1977). Divaricates           et al. (2017). Most overseas divaricate-like
have been reported as especially common in         plants also differ notably from all but one
open environments such as forest margins           New Zealand divaricates by the presence of
(McGlone & Webb 1981), though relevant             wounding spines. A striking example is the
quantitative data are lacking. The percentage      African boxthorn (Lycium ferocissimum Miers;
of divaricate species in woody assemblages         see Figure 2 in Supplementary Material), a
increases from north to south (McGlone             South African species naturalised in New
et al. 2010). Quantitative analyses have           Zealand, which has tough interlaced branches
shown strong associations with frosty (and         similar to those of some New Zealand
to some extent, droughty) climates such as         divaricates but bears sharp spines. However,
are typical of the eastern South Island (Lusk      this spinescence can sometimes be rather
et al. 2016; Garrity & Lusk 2017) where            weak, for example in Australian species of
notably divaricate species often comprise the      Melicytus J.R.Forst. & G.Forst. (Stajsic et al.
majority of arborescent assemblages (Lusk et       2015). There are however some overseas
al. 2016). It has been stated that divaricates     divaricate look-alikes that show the same
are commonest on fertile young soils, such         traits as New Zealand divaricates, for example
as those derived from recent alluviums or          Tetracoccus hallii Brandegee (Picrodendraceae),
volcanic ashes (Greenwood & Atkinson 1977;         a non-spiny shrub with seemingly tough,
McGlone et al. 2004). Consistent with this         interlaced branches, branching at wide angles
proposal, the largest known concentrations         and bearing small leaves (descriptions and
of divaricate species occur on alluvial terraces   pictures from SEINet Portal Network 2020
derived from mudstone in the Rangitikei and        and Calflora 2020) from south-west USA
Gisborne areas (Clarkson & Clarkson 1994).         (distribution data from GBIF 2020 and
However, an analysis of > 1,000 plots by           Calscape 2020).
Lusk et al. (2016) did not detect a significant        We propose a list of the species that the
association with terraces, or with any other       studies cited above claim as “divaricate” and
topographic position.                              that we agree do resemble the architectural
    Even though broadly similar plants occur       models we see in New Zealand divaricates
in many other regions of the world, few of         (Appendix 2). We sug gest the name
them show the full range of traits that are        divaricate-like to describe these species in
typical of New Zealand divaricates. Species        order to emphasise their resemblance with
showing aspects of the divaricate habit have       New Zealand divaricates, yet stressing the fact
been reported from Madagascar (Grubb               that they often present distinguishing features
2003; Bond & Silander 2007), Patagonia             (discussed above) and that they evolved
(Wardle & McGlone 1988; McQueen 2000)              in environmental conditions that were
or South America in general (Böcher 1977),         somewhat different from those experienced
mainland Australia and Tasmania (Bulmer            by the ancestors of New Zealand divaricates
1958; Mitchell et al. 2009; Thompson 2010;         (reviewed below).
Stajsic et al. 2015), Arizona and California in
the USA (Carlquist 1974; Tucker 1974) and
New Guinea (Lloyd 1985). The reported
species and their close relatives indeed
show branching patterns similar to what
is seen in New Zealand divaricates, but
they often present rather large leaves. This
is for example the case with the North
American Quercus dunnii Kellogg ex Curran,
Maurin & Lusk: Untangling the divaricate habit: review                7

A review of theories about                          evidence show that New Zealand was not
the evolution of New Zealand                        completely submerged during the Late
divaricates                                         Oligocene (reviewed by Mildenhall et al.
                                                    2014), particularly the 23 Myo Foulden
                                                    Maar deposit (near Middlemarch, Otago),
The climatic hypothesis                             which notably contains fossils of diverse
                                                    land plants (e.g. Lee et al. 2016). Moreover,
Since its Upper Cretaceous separation from
                                                    recent molecular dating of the age of New
Gondwana (Wallis & Trewick 2009), New
                                                    Zealand lineages strongly suggest that some
Zealand has undergone wide-ranging climatic
                                                    extant terrestrial plant and animal groups
changes. There is some debate as to the
                                                    most probably originated from a Gondwanan
climate of the Upper Cretaceous: some argue
                                                    vicariance (Wallis & Jorge 2018; Heenan &
this period was probably warmer than today
                                                    McGlone 2019).
(e.g. Fleming 1975), others that it was similar
                                                        Diels (1896) was the first to hypothesise
to present-day climates (e.g. Mildenhall 1980;
                                                    an important role of Pleistocene climate
Kennedy 2003). Hornibrook’s (1992) review
                                                    in shaping the modern New Zealand flora,
of marine fossil evidence indicates mostly
                                                    and as far as we are aware his work is the
subtropical climates during the Paleogene,
                                                    first attempt to explain the evolution of
although a sudden cooling event may have
                                                    the divaricate habit. He proposed that, by
occurred around the Eocene-Oligocene
                                                    reducing transpiration, the divaricate habit
boundary; temperatures then warmed to
                                                    helped plants cope with droughty climates
a local peak around 16 Mya, during the
                                                    created in the eastern South Island by the
Miocene; the climate remained subtropical
                                                    uplift of the Southern Alps. Cockayne
until a Late Miocene cooling, with further
                                                    (1911) proposed that the divaricate habit
cooling from the Pliocene. The combined
                                                    was a response to past windy and droughty
effects of this global cooling and of the
                                                    Pleistocene steppe climates, especially in the
rapid uplift of the Southern Alps during the
                                                    South Island. Similarly, Rattenbury (1962)
Kaikoura Orogeny (Batt et al. 2000) created
                                                    hypothesised that the divaricate habit was
local frosty and droughty environments,
                                                    an adaptation to dry or cool Pleistocene
especially in the eastern South Island. These
                                                    climates, and suggested an effect of the cage-
new climates are likely to have reduced plant
                                                    like architecture as a windbreak, reducing
growth on many sites (Lusk et al. 2016), as
                                                    transpiration. Wardle (1963) suggested that
shown by comparisons of juvenile annual
                                                    the divaricate habit continues to be adaptive
height growth rates of the small broadleaved
                                                    in the present-day drier forest and shrub
tree Aristotelia serrata J.R.Forst. & G.Forst. on
                                                    environments of eastern New Zealand.
modern sites that differ in growing season
                                                        McGlone & Webb (1981) fur ther
length (Bussell 1968; Anton et al. 2015).
                                                    developed the climatic hypothesis, joining
    Besides these climatic variations, a
                                                    the debate started by Greenwood and
progressive submergence greatly reduced
                                                    Atkinson with the moa-browsing hypothesis
the extent of the New Zealand landmass
                                                    (Greenwood & Atkinson 1977; see next
from the Upper Cretaceous to the Early
                                                    section). They suggested that the divaricate
Miocene (85–22 Mya; Landis et al. 2008). It
                                                    habit represents the response of the “largely
reaching a peak around 25–23 Mya known as
                                                    subtropical” Tertiary flora of the isolated
the Oligocene marine transgression (Cooper
                                                    New Zealand archipelago to the near-treeless
1989), at which point the surface of the New
                                                    glacial periods of the Pleistocene; this habit
Zealand mainland was about 18% of its
                                                    may have protected growing points and leaves
present-day surface area (Cooper & Cooper
                                                    from wind abrasion, desiccation and frost
1995). Landis et al. (2008) argued that, at that
                                                    damage, which occurred unpredictably in
time, New Zealand was probably completely
                                                    the weakly seasonal New Zealand climates
submerged, but this idea is now clearly
                                                    of the Quaternary. McGlone & Webb (1981)
refuted. Geological and paleobiological
New Zealand Natural Sciences 46 (2021)                   8

also argued that the cage-like architecture        The photoprotection variant of the climatic hypothesis
of the divaricate habit also provides a
                                                       Howell et al. (2002, see also Howell 1999),
milder microclimate within the plant which
                                                   proposed that the shading of inner leaves by
promotes higher rates of photosynthesis. The
                                                   the cage-like divaricate architecture protects
transition from the juvenile form to the adult
                                                   them from high irradiance on cold mornings
form in heteroblastic divaricate species occurs
                                                   after frosts, thus minimising photoinhibition
above the height of the most damaging frosts
                                                   and photodamage. It is a derivative of the
during temperature inversions on clear nights,
                                                   climatic hypothesis that includes the effect of
and the absence of the habit on offshore and
                                                   solar radiation as a selective pressure under
outlying islands can be explained by their
                                                   stressfully cold climatic conditions. Howell
more oceanic, hence milder and less frosty,
                                                   et al. (2002) tested this hypothesis with an
climates. Burns & Dawson (2009) however
                                                   experiment involving the pruning of the
noted that the heteroblastic divaricate species
                                                   outer branches of three divaricate species,
Plagianthus regius from the mainland has a
                                                   which resulted in a reduced photosynthetic
heteroblastic divaricate subspecies (P. regius
                                                   capacity of the inner leaves of these shrubs
subsp. chathamicus (Cockayne) de Lange) on
                                                   for at least 3 months. This experiment was
the historically avian-browser-free Chatham
                                                   criticised by Lusk (2002), who pointed out
Islands: they propose that, because P. regius is
                                                   that the failure to include non-divaricate
a recent immigrant on the Chatham Islands,
                                                   species as a control undermined the authors’
its juvenile form has not been counter-
                                                   conclusions: without further research, we
selected yet.
                                                   cannot know if non-divaricate plants would
    The climatic factors suggested as selective
                                                   respond in a similar way to pruning of their
forces are certainly not peculiar to New
                                                   outer branches.
Zealand, whereas divaricate-like forms are
much less common in other regions with             Empirical appraisal of the climatic hypothesis
similar climates (Dawson 1963). McGlone
& Webb (1981) argued that what made                Experimental tests have produced little
New Zealand unique in the evolution of its         support for the climatic hypothesis. Although
subtropical flora in response to the cold, dry     past climatic conditions cannot be reliably
and windswept environments that appeared           reproduced in a controlled experiment, it is
during the Quaternary was its isolation from       possible to estimate the differential response
sources of steppe-adapted floras, apparently       of divaricate and non-divaricate species when
believing that such floras might have provided     they are subjected to present-day climatic
plants with more conventional physio-              conditions similar to those hypothesised to
morphological responses to cold, dry climates.     have selected the divaricate habit during the
This argument appears to overlook the fact         Pleistocene.
that divaricate shrubs are also common in the         Kelly & Ogle (1990) were the first to
Patagonian steppe, although those species          publish a test of the response of divaricating
are invariably spinescent (McQueen 2000).          habits to climatic conditions. They studied
Furthermore, if wind was one of the drivers        the effect of air temperature, humidity, frost
of the evolution of the divaricate habit, it is    and wind on internal and external leaves of
strange that few divaricate species are found      a divaricate species and both juvenile and
in some very windy parts of New Zealand            adult forms of a heteroblastic divaricate
(Greenwood & Atkinson 1977): although              species. While they did not show a significant
they are often prominent in the vegetation of      difference in leaf temperature and air
windswept areas such as Cook Strait (Wardle        humidity between the inside and the outside
1985), they present a low species richness         of divaricating habits, they did show that the
there (Gillham 1960).                              habit provides some protection against frost.
                                                      Keey & Lind (1997) used four species
                                                   showing various divaricating habits to test the
                                                   effect of different branching architectures on
Maurin & Lusk: Untangling the divaricate habit: review                  9

the surrounding airflow patterns. Although          especially stressfully cold temperatures. In
they did not compare these species to non-          parallel, Schneiderheinze (2006) studied
divaricate species, they showed that dense          photoinhibition in divaricate and non-
branching patterns produce calmer zones,            divaricate species under high light loads and
which may imply that they create a more             other stressful conditions, such as drought.
favourable growing environment for leaves           She found plants of both habits showed
and other fragile organs by reducing wind           similar levels of photoinhibition under high
damages.                                            irradiance, whether the plants were water-
    Darrow and colleagues experimentally            stressed or not. Here again, the hypothesis
compared the frost resistance (2001) and            as formulated by Howell et al. (2002; i.e.
water use efficiency (2002) of juvenile and         protection from photoinhibition under cold
adult forms of heteroblastic species, most          conditions) was not tested, but the study still
of them divaricate at a juvenile stage. Darrow      provided a valuable insight into the absence
et al. (2002) found that most (though not all)      of significant photoprotection in divaricate
divaricate juvenile forms had lower water           species compared to their non-divaricate
use efficiency than the corresponding adult         relatives.
forms, concluding their results were not                Recently, an observational approach was
consistent with the climatic interpretation         taken by Lusk et al. (2016), who examined the
of the divaricate form. Darrow et al. (2001)        environmental correlates of the proportion of
compared the frost tolerance of the leaf            divaricate species in arborescent assemblages
tissues of juvenile and adult forms of five         throughout the main islands of New Zealand.
heteroblastic divaricate species by chilling        They concluded that divaricate species are
leafy twigs overnight in thermostatically           generally more diverse and prominent at
controlled freezers. However, their findings        frosty and droughty sites. Garrity & Lusk
are of limited relevance to the climate             (2017) also used an observational approach by
hypothesis, as this approach does not address       correlating climatic data with the distribution
the effect of leaf size on night-time chilling      of 12 congeneric pairs of divaricate and
under a clear sky (cf. Lusk et al. 2018),           larger-leaved species of the main islands of
nor any potential effect of stem vascular           New Zealand. They found that divaricate
anatomy on freeze-thaw embolism. In a               species were significantly favoured by colder
similar vein, Bannister et al. (1995) studied the   mean annual temperatures, and especially by
development of frost tolerance of detached          colder minimum July temperature, but there
leaves of some divaricate and non-divaricate        was little evidence of an association with
species of Pittosporum Banks & Sol. ex Gaertn.      droughtier environments. Their results also
over the course of autumn and winter. As was        showed little support for the photoprotection
the case for Darrow et al. (2001) this study        hypothesis, as divaricate species tended
of tissue-level responses to frost did not test     to predominate in cold environments
the potential roles of any of the characteristic    irrespective of winter solar radiation levels.
leaf or stem traits of divaricates in conferring    These two different observational approaches
frost resistance .                                  concur in showing that short frost-free
    A test of the photoprotection hypothesis        periods and cold climates in general favour
was provided by Christian et al. (2006), who        the abundance and diversity of divaricate
compared carbon gain versus structural              species, but do not quite agree on the effect
costs of three congeneric pairs of divaricate       of drought. Given the limited number
and non-divaricate species under different          of species encompassed by Garrity &
intensities of light exposure. They showed          Lusk (2017), as well as evidence that the
that the costs of divaricating habits may           largest concentrations of divaricate species
be too high to be compensated by the                occur on middle North Island sites subject
photoprotection it provides, although               to significant water deficits (Clarkson &
they did not subject their samples to               Clarkson 1994), the balance of the evidence
New Zealand Natural Sciences 46 (2021)           10

indicates that both frost and drought favour       about how the ancestors of moa reached
divaricate species.                                New Zealand (Allentoft & Rawlence 2012):
    Finally, a key component of the divaricate     they may have inhabited the New Zealand
habit is small leaf size, which is known to        landmass from the time it started to separate
be advantageous under harsh climates. A            from Gondwana about 80 Mya (the “Moa’s
study by Lusk et al. (2018) compared leaf          Ark” of Brewster 1987); alternatively their
temperature during clear winter nights in          ancestors might have reached New Zealand
relation to leaf size for 15 native New Zealand    either by walking before 60 Mya, when the
species, including four congeneric pairs of        New Zealand landmass was still connected to
divaricate and non-divaricate species. They        a disintegrating Gondwana, or by flying after
observed that small leaves chilled significantly   the complete separation. This last possibility
less than large leaves. Their conclusions          is consistent with recent molecular evidence
provide experimental support to leaf energy        that the closest living relatives of moa appear
balance theory, which predicts that large          to be tinamous (Phillips et al. 2010; Mitchell
leaves should be more vulnerable to frost          et al. 2014), a group of volant birds. If the
because they cool below air temperatures on        earliest ancestors of moa to inhabit Zealandia
frosty nights whereas the smallest leaves stay     were volant, fossil evidence suggest that their
close to air temperature (Parkhurst & Loucks       descendants have been large flightless birds
1972; Wright et al. 2017). Although this           since at least 16-19 My ago (Tennyson et al.
effect does not explain the three-dimensional      2010). All moa species were extinct by about
structure of the divaricate habit, it suggests     the mid-15th century CE (Perry et al. 2014),
that the characteristically small leaves of        apparently because of hunting (Allentoft et
divaricates may have provided an adaptive          al. 2014).
value in open habitats with short annual                Moa subfossil remains are more common
frost-free periods (see also Lusk & Clearwater     on the South Island than on the North
2015, a similar but less conclusive study on       Island (Anderson 1989); moreover, they are
a smaller scale). Additionally, a study of the     more concentrated in the east of the South
relationship between leaf dimensions and           Island (Anderson 1989). However, this does
environmental variables in South African           not necessarily mean that moa were more
species of Proteaceae concluded that small         abundant in the eastern South Island than
leaves promotes convective heat dissipation        elsewhere in the country, since the subfossil
under dry conditions and limited wind,             record is probably influenced by preservation
enabling them to avoid overheating when            biases: natural moa bone deposits are mainly
water shortage forces stomatal closure             in alkaline swamps and limestone caves, which
(Yates et al. 2010). This effect was confirmed     are near-ideal preservation environments
on Australian Proteaceae by Leigh et al.           (Atkinson & Greenwood 1989) that happen
(2017). The small size of the leaves of most       to be more common in the eastern South
divaricates may therefore enable them to           Island than in most other parts of the country
cope with drought better than large-leaved         (Anderson 1989). Furthermore, an estimation
competitors.                                       of population size and distribution of the
                                                   different moa species based on mitochondrial
The moa-browsing hypothesis
                                                   DNA and fossil record of Dinornis spp.
 “Moa” is the Māori name for a group of            suggests, in contrast, that moa populations
now-extinct large (1-3 m and 10-250 kg;            were more numerous on the North Island
Atkinson & Greenwood 1989; Worthy &                than on the South Island (Gemmell et al.
Holdaway 2002) flightless birds (“ratites”) of     2004). Therefore, it seems difficult at present
the endemic order Dinornithiformes. Nine           to draw clear conclusions about geographic
species are currently recognised, belonging        variation in moa densities.
to six genera and three families (Worthy &             Although the potential influence of moa
Scofield 2012). There are several hypotheses       browsing on the evolution of the divaricate
Maurin & Lusk: Untangling the divaricate habit: review                 11

habit had been suggested by previous authors        (McGlone & Clarkson 1993), the largest
(e.g. Denny 1964; Carlquist 1974; Taylor            known concentrations have been reported
1975), Greenwood & Atkinson (1977) were             from fertile terraces derived from mudstone
the first to fully develop and argue this idea.     (Clarkson & Clarkson 1994). On the other
First postulating that moa fed by clamping          hand, Greenwood & Atkinson (1977) noted
and pulling vegetation in the same manner           that divaricates are largely absent from areas
as present-day ratites, they hypothesised           where moa did not live, such as offshore
that the tough and highly tensile branches          islands, or where moa could not reach them,
of many divaricate species are difficult to         such as growing on cliffs or as epiphytes.
tear off, while the interlaced structure kept       Although Myrsine divaricata A.Cunn. is
leaves and growing tips out of easy reach.          abundant on some of the subantarctic islands
Hence, browsing on these plants would be            of New Zealand (McGlone & Clarkson 1993;
less energetically rewarding than browsing on       Meurk et al. 1994), which are unlikely to have
broadleaved species. Greenwood & Atkinson           harboured moa, Greenwood & Atkinson
(1977) did not completely exclude a cutting         (1977) attributed such occurrences to recent
ability of moa beaks, later acknowledging that      colonisation from the mainland. Kavanagh
the feeding behaviour of moa could not be           (2015) lent support to this interpretation by
confidently inferred because fossil skulls do       comparing some traits used to describe the
not retain all the relevant tissues (Atkinson &     divaricate habit between related species of
Greenwood 1989). A recent study simulating          New Zealand mainland and Chatham Island
the force of moa jaw muscles however                (historically moa-free, with a flora largely
concluded that different moa species fed            derived from the mainland): he concluded
in various different ways, including cutting        that the absence of moa may have relaxed
(Attard et al. 2016). This appears to confirm       the selection for traits that deterred moa
the findings of studies of moa gizzard              browsing on the main islands of New
contents, which concluded that that divaricate      Zealand.
twigs consumed by moa had been sheared                  Greenwood & Atkinson (1977) also
rather than broken off (Burrows 1980; 1989;         examined the bearing of the height of
Burrows et al. 1981). These findings were later     transition between the juvenile in adult forms
corroborated by a study of coprolites (Wood         in heteroblastic divaricate species on their
et al. 2008), yielding the same conclusion that     hypothesis. They claimed that, in such species,
divaricate species were by no means exempt          the shift from the juvenile divaricate form
from moa browsing (reviewed by Wood et              to the adult non-divaricate form happens
al. 2020).                                          around 3-4 m high; this height corresponds
    Moreover, Greenwood & Atkinson (1977)           to the approximate height of the tallest moa,
used evidence from the distribution of              implying that the adult form in these species
divaricate plants to support their hypothesis.      only appears at heights where it is safe from
One the one hand, they pointed out that             browsing. Burns & Dawson (2006) brought
divaricate plants often grow on lowland             support to this claim from New Caledonia:
river terraces and swamps, which offer              they mentioned that heteroblastic species
high nutrient levels and hence high plant           there (which do not have a divaricating
productivity and nutrient content. They             juvenile form) seem to shift form at about
explained that divaricate species should be         the estimated height of the flightless birds
more subjected to moa browsing in such              which once lived there, although they called
places, a sensible claim given that at least some   for quantitative support for this observation.
studies show a positive correlation between         There are however multiple counter-examples
herbivore abundance and soil fertility (e.g.        to Greenwood & Atkinson’s (1977) claim.
Kanowski et al. 2001). Even if divaricate           Field observations sometimes reveal that
species have been reported from low fertility       the shift can happen significantly lower; for
soils, such as the acidic soils of Stewart Island   example, Cockayne (1911) reported that
New Zealand Natural Sciences 46 (2021)            12

the shift in Sophora microphylla can happen           Lowry’s view (Atkinson & Greenwood 1980).
as low as 1.4 m, and we observed a shift in           Consequently, this view raised the question
Pennantia corymbosa happening at about 2 m            of why the divaricate habit, if it is not a
high (Figure 1 in Supplementary Material).            specialised moa-deterring adaptation, is much
Conversely, some homoblastic divaricate               scarcer in other regions where non-ratite
species can reach heights significantly above         browsers existed (McGlone & Webb 1981).
the size of the tallest moa without showing               Indirect support for the moa-browsing
any relaxation of their divaricating habit;           hypothesis came from a fossil of a small-
McGlone & Clarkson (1993) report such                 leaved woody species with wide-angle
instances with individuals of Coprosma                opposite branching that was discovered by
crassifolia Colenso, Melicope simplex A.Cunn.         Campbell et al. (2000). It was estimated to
and Myrsine divaricata more than 5 m high;            date from 20-16 Mya, which corresponds
individuals of the latter species exceeding           to the Early Miocene, whereas the climatic
this height were also recorded by Veblen &            conditions usually put forward as the drivers
Stewart (1980).                                       of the evolution of the divaricate habit did
    Finally, a crucial point of Greenwood &           not occur before the Pliocene (i.e. not before
Atkinson’s (1977) argument is the fact that           5.333 Mya, Cohen et al. 2013, updated).
the New Zealand flora is unique in having             Despite the absence of information about
co-evolved with ratites but without browsing          the three-dimensional structure of the
mammals. This phenomenon did not occur                plant when alive, 12 out of 15 experts they
in areas where divaricate-like species co-            consulted agreed it was most likely a divaricate
evolved with ratites: in Madagascar, now-             species (potentially extinct), and had rather
extinct elephant birds shared the island with         varied ideas about what genus it could belong
giant tortoises and giant lemurs (Bond &              to. They noted the presence of “small acute
Silander 2007); in Patagonia, Darwin’s rhea           broken processes protrud[ing] from the
grazed side-by-side with diverse mammals,             branchlets at irregular intervals”, which look
such as equiids, camelids and giant ground            like spines even though they are not opposite.
sloths (McQueen 2000); in Australia, emus             Even though the processes might have been
coexisted with many different herbivorous             defensive spines that would be of little use
mammals, mostly marsupials (Roberts et                against moa beaks, this discovery appears
al. 2001). Although these regions have all            consistent with the moa-browsing hypothesis.
undergone megafaunal extinctions, they                    According to the moa-browsing hypothesis,
still host browsing mammals, and with the             the divaricate habit could be nowadays seen
exception of Madagascar they have retained            as an anachronism (Greenwood & Atkinson
their ratites as well. No ratites or ratite fossils   1977). As such, it was hypothesised that
are known from North America; they are                divaricate species may not be adapted to the
known only from former Gondwanan lands                current browsing pressure of introduced
(Briggs 2003).                                        mammals because their costly ratite-resistant
    Greenwood & Atkinson (1977) originally            architecture was thought to be useless against
hypothesised that the divaricate habit evolved        mammals (Bond et al. 2004). Diamond
as a deterrent to moa browsing. Lowry (1980)          (1990) imported the concept of “ghost”
instead suggested that the main effect of the         from overseas cases of anachronisms (later
divaricate habit is to help the plant survive         reviewed by Barlow 2000) when defending
browsing by spacing and multiplying palatable         the hypothesis that divaricates are adapted
growing tips, with a side-effect of making the        to a now-extinct fauna. However, the
browsing less energetically rewarding. This           conclusions of Pollock et al. (2007) about the
idea that the divaricate habit enables plants         preferences of ungulates for New Zealand
to survive rather than to prevent browsing            woody plants, as well as a study by Lusk
led Atkinson and Greenwood to reconsider              (2014) on the regeneration of divaricate and
their 1977 hypothesis by acknowledging                non-divaricate species in a forest remnant
Maurin & Lusk: Untangling the divaricate habit: review                 13

that had been subject to ungulate browsing         New Zealand landmass as early as 80-60 Mya
for decades, indicate that the divaricate habit    (reviewed by Allentoft & Rawlence 2012),
may also be effective in deterring mammal          the divaricate habit may not have become
browsing. Ungulates indeed tend to avoid           advantageous as an anti-browsing defence
some (though not all) divaricate species until     until Plio-Pleistocene climatic constraints
more attractive foods are depleted (Forsyth        on plant growth resulted in juvenile trees
et al. 2002; Lusk 2014).                           being exposed for longer to ground-dwelling
                                                   browsers. During this period the combination
Experimental appraisal of the moa-browsing
                                                   of global cooling (Hornibrook 1992) and
hypothesis
                                                   rapid uplift of the Southern Alps (Batt et al.
The moa-browsing hypothesis was first tested       2000) created widespread frosty, droughty
experimentally by Bond et al. (2004), who          environments in the eastern South Island.
fed juvenile and adult form foliage of two         The relatively fertile alluvial soils of these
heteroblastic divaricate species to present-       environments may have attracted high levels
day ratites (emus and ostriches). They found       of browsing, but frost and drought would
that the high tensile strength of divaricate       have reduced the ability of juvenile trees
branches reduces breakage, that the high           to grow rapidly out of the browsing zone,
branching angles make the twigs difficult          even in well-lit microenvironments such as
to swallow because birds cannot use their          treefall gaps. Evidence for a much earlier
tongue to properly orient the twigs, and that      origin of divaricate plants, for example in
small and widely spaced leaves increase the        the more benign climates of the Miocene or
time and the energy required to consume leaf       Oligocene, would refute both this hypothesis
biomass. These results brought support to the      and the original climate hypothesis, and would
hypothesis that the divaricate habit represents    point to moa browsing as the sole driver of
an adaptation to deter moa browsing.               divaricate evolution if no other factor can
However, whether the feeding behaviour             be identified.
of the present-day ratites reliably reflect the
                                                   The light trap hypothesis and its appraisal
feeding behaviour of extinct moa is a matter
of debate (reviewed above).                        The light trap hypothesis, formulated by Kelly
    A more elaborate cafeteria experiment was      (1994), relies on the conclusions of Horn
conducted a few years later by Pollock et al.      (1971) that a multi-layered leaf distribution
(2007), comparing the offtake of deer, goats       (i.e. leaves distantly scattered among multiple
and ostriches from five divaricate species         layers in the canopy) is more efficient at
compared to five congeneric non-divaricate         capturing a higher proportion of sunlight
species. Their general finding is that features    than mono-layered architectures (i.e. leaves
of the divaricate habit, such as small leaves      distributed in a dense layer, the umbra of
and stem toughness, deter ungulates as well        the outermost leaves completely obscuring
as ratites.                                        the innermost leaves). Photosynthesis of
                                                   most plants is indeed saturated well below
The moa-climate synthetic hypothesis
                                                   full sunlight, the saturation point varying
The idea that selection for the divaricate habit   with, for example, species’ successional status
may have been driven by both past climatic         (e.g. Bazzaz & Pickett 1980). The scattered
conditions and the effect of moa browsing          distribution of the leaves of divaricates
has been suggested several times since the         over multiple branch layers therefore allows
debate started (Wardle 1985; 1991; Cooper          inner leaves to be in the penumbra of the
et al. 1993; Bond & Silander 2007). Lusk et        outer leaves, thus better distributing light
al. (2016) proposed a synthetic hypothesis         harvest throughout the canopy. The light
with a specific mechanism integrating              trap hypothesis appears consistent with a
browsing and climatic factors. Although            modelling study of the impact of penumbral
the ancestors of moa may have reached the          effects on shoot-level net carbon gain of
New Zealand Natural Sciences 46 (2021)           14

conifers (Stenberg 1995) which, like New                Moreover, he suggested that this sequential
Zealand divaricates, have small effective               branching is characterised by a lack of
leaf diameters that result in short shadows;            organisational control that translates into
this modelling however does not explain                 a dimorphism between orthotropic and
the potential advantage of the architectural            plagiotropic branches. He recommended
structure of divaricating habits. Moreover,             the study of the changes in the branching
even though penumbral effects are likely to             sequence of many divaricate species over
result in higher carbon gain per unit area of           their lifetime, as he believed this could be
foliage in small-leaved species growing in              the only way to understand how the diversity
high light, Christian et al.’s (2006) data suggest      of divaricating habits was produced under a
that this advantage will be outweighed by the           possibly single selective pressure, and to draw
much higher (ca. threefold) leaf area ratio of          general conclusions about their development.
congeneric broadleaved species, resulting in                Subsequently, the development patterns
higher net carbon gain per unit of biomass              of a few divaricates were studied in the
in the latter. In divaricate species, this effect       1990s. The species were: Muehlenbeckia astonii
might be at least partially compensated by              (Lovell et al. 1991); the juvenile form of
photosynthesis in stems, brought to light in            Elaeocarpus hookerianus (Day & Gould 1997;
one instance so far: the juvenile form of the           Day et al. 1998; Day 1998a), Carpodetus
heteroblastic divaricate Prumnopitys taxifolia          serratus J.R.Forst. & G.Forst. (Day 1998a; b)
(Banks & Sol. ex D. Don) de Laub. (Mitchell             and Pennantia corymbosa (Day 1998c); Sophora
et al. 2019). More divaricate species will              prostrata and the juvenile form of Sophora
need to be investigated to determine how                microphylla (Carswell & Gould 1998). Overall,
widespread stem photosynthesis is among                 these studies concluded that such a growth
divaricates. However, why would divaricating            pattern, with many growing points scattered
habits be scarce or absent in most other                across the plant’s crown, offers a plastic
regions of the world if sunlight were the               structure that can more easily accommodate
main driver of the evolution of these peculiar          changes in environmental conditions (e.g.
architectures in New Zealand, where solar               forest canopy gap versus closed canopy
irradiance levels are similar to those of other         or seasonal changes in environmental
regions at comparable latitudes (Solargis               conditions). These case studies also agreed
2020)? The light trap hypothesis does not               that the lack of apical dominance plays a key
appear to offer a satisfying explanation of             role in the establishment of the divaricating
the evolution of the New Zealand divaricates.           habits they observed.
                                                            In parallel to the study of developmental
Insights into the development of divaricate branching
                                                        patterns, a handful of studies looked into
patterns
                                                        the hormonal control of the divaricate
If the debate surrounding divaricate plants             habit. Horrell et al. (1990) showed that a
has mainly focused on how the divaricate                gibberellic acid treatment on cuttings of
habit has evolved, a handful of studies looked          the adult form of Pennantia corymbosa and
into describing the range of growth patterns            Carpodetus serratus tends to revert them to
that give rise to the spectrum of divaricating          their juvenile form. This phenomenon did
habits, and how such patterns translate into            not occur in Elaeocarpus hookerianus, a result
adaptations to local environments.                      later confirmed by Day et al. (1998) with
   Tomlinson (1978) examined bifurcation                treatments of adult cuttings with gibberellic
ratios of 18 New Zealand divaricates,                   acid and other growth factors, including
including two heteroblastic divaricate                  a cytokinin. Day et al. (1998) also showed
species. He concluded that the interlaced               that the adult form is not precociously
structure of most divaricates is a consequence          triggered in E. hookerianus seedlings by these
of a sequential branching which may be                  treatments. In Sophora, a treatment with
supplemented by reiterative branching.                  6-benzylaminopurine (a cytokinin) reinforces
Maurin & Lusk: Untangling the divaricate habit: review                  15

the divaricateness of the juvenile form of         nevertheless do not resolve the blurry
Sophora microphylla (Carswell et al. 1996).        boundary between true divaricates and semi-
Qualitative and quantitative measurements in       divaricates, like any categorisation involving
E. hookerianus showed that the leaves of the       a degree of subjectivity.
divaricating juvenile form contain more active         In spite of rather extensive experimental
cytokinins than the non-divaricating adult         and observational evidence, no hypothesis
form or transitional form leaves (Day et al.       about the evolution of divaricates in New
1995, reviewed by Jameson & Clemens 2015).         Zealand has been decisively favoured over
A similar yet more questionable conclusion         another. Among the most plausible hypotheses
was drawn from a comparison of the ratio           however, the moa-browsing hypothesis seems
of active to storage forms of cytokinin            more supported than the climatic hypothesis,
between divaricate and non-divaricate forms        although neither are fully satisfying on their
in Sophora species (Carswell et al. 1996). In      own. The synthetic moa-climate hypothesis
contrast with the heteroblastic divaricate         has not been much discussed or tested so
species studied, the levels of cytokinins are      far, but given the evidence of both the
relatively low in the divaricate species Sophora   moa-browsing hypothesis and the climate
prostrata, suggesting that they might not play     hypothesis individually, it appears to be a
a role in the establishment of the divaricating    good candidate for a definitive answer to the
habit itself (Carswell et al. 1996). There are     divaricate question.
however too few studies about these growth             However, neo-ecological studies alone
regulators to formulate general conclusions        are unlikely to entirely resolve the origin of
about their potential effects in controlling the   divaricate plants. One way still left to explore
expression of the divaricate habit.                was suggested by Cooper et al. (1993):
                                                   using molecular phylogenetics to date the
Conclusions                                        divergences between divaricates and their
                                                   closest non-divaricate relatives. Past studies
The terms divaricate or divaricating
                                                   estimating the age of New Zealand plant
have been variously applied to around 80
                                                   lineages (e.g. reviewed by Wallis & Jorge
New Zealand species that we regard as
                                                   2018; Heenan & McGlone 2019) have not
occupying a spectrum from truly divaricate
                                                   focused on dating such divergences. Such
(small and widely-spaced leaves; wide-angle
                                                   studies, and studies on overseas groups that
branching; tough, wiry, tightly interlaced
                                                   include New Zealand representative, can still
stems) to semi-divaricate (plants that
                                                   offer isolated dates even though they might
present some but not all of these traits). This
                                                   not have sampled the closest non-divaricate
spectrum of architectural forms, which we
                                                   relative to the divaricate species they included
call divaricating habits, is the expression of
                                                   (Appendix 3). The divergence dates between
a phenomenon called the divaricate habit.
                                                   congeneric divaricate and non-divaricate
Heteroblastic divaricate species have a
                                                   species give us a first hint that the divaricate
divaricate (or semi-divaricate) juvenile form
                                                   habit may have appeared less than 10 Mya in
and a non-divaricate adult form, in contrast
                                                   most cases. Table 2 provides the theoretical
to the generally smaller (< 8 m) homoblastic
                                                   divergence dates one might expect from
divaricates that retain the divaricate form
                                                   a study specifically dating splits between
throughout their entire lives. Finally, we
                                                   divaricate and non-divaricate species under
coin the term divaricate-like to describe
                                                   the different hypotheses in play: the dates of
overseas instances of the divaricate habit
                                                   the divergences in Appendix 3 hardly favour
phenomenon, which acknowledges their
                                                   one hypothesis over the other, suggesting the
resemblances with New Zealand divaricates
                                                   need for a dating effort specifically targeting
while stressing their peculiarities. We hope
                                                   divaricate species and their closest non-
that adoption of these terms will help reduce
                                                   divaricate relatives, as suggested by Cooper
ambiguities in future research and facilitate
                                                   et al. (1993).
clear communication. Our recommendations
You can also read