Moa ghosts exorcised? New Zealand’s divaricate shrubs avoid photoinhibition



  • 1 An intriguing feature of the New Zealand flora is the high frequency of ‘divaricate’ shrubs and tree juveniles. These plants have numerous interlacing wide-angled branches with small leaves and no or few leaves in the outer canopy. They comprise 10% of native woody species, and have evolved in 18 different plant families.
  • 2 We tested the hypothesis that the leaves of divaricate plants are sensitive to cold-induced photoinhibition, and that self-shading by outer branches reduces light intensities enough to prevent photodamage. In a field experiment, leaves of three divaricate species (Aristotelia fruticosa A. Cunn., Corokia cotoneaster Raoul and Coprosma propinqua Hook. f.) inside the north (sunny) side of the shrubs were exposed to one of three experimental treatments over winter: (i) control leaves which were not manipulated; (ii) exposed leaves which had their outer screen of branches pruned away leaving them open to full sun; or (iii) shaded leaves which were exposed by pruning, then sheltered from direct sunlight with shade cloth.
  • 3Experimental removal of the shielding branches in winter led to rapid (< 20 days), large (23–31%) and persistent (> 3 months) reductions in the maximum photosynthetic capacity (Amax) and photochemical efficiency (dark adapted Fv/Fm) in exposed leaves, but not in shaded leaves. Full recovery of photosynthetic capacity and photochemical efficiency was displayed by Coprosma propinqua, but very little capacity for recovery was displayed by Aristotelia fruticosa or Corokia cotoneaster.
  • 4When the effects of self-shading and photoinhibition were combined, control leaves photosynthesized faster than exposed leaves at all ambient irradiances in A. fruticosa, and in bright light (> 500 µmol m−2 s−1) in C. cotoneaster.
  • 5Our data show that by shielding their leaves within an outer screen of branches, the three divaricate species studied reduce photoinhibition of photosynthesis. We contend that this architectural self-shading in divaricate plants maximizes potential carbon fixation by minimizing photoinhibition.


The isolation of the New Zealand archipelago since tertiary times (Rattenbury 1962) and multiple invasions from several different geographical areas (Cockayne 1928) have resulted in a flora that features many peculiarities. Endemism is high (Mildenhall 1980), evergreens are unusually common for a temperate region, and hybridization is extensive (Webb & Kelly 1993). Another widespread peculiarity is the occurrence in the New Zealand flora of small-leaved tangle-branched shrubs. Plants with this distinctive growth form have been collectively called ‘divaricate’ shrubs (Cockayne 1912), and together comprise approximately 10% of the woody flora of New Zealand (Greenwood & Atkinson 1977). While classifications of divaricate shrubs vary (Kelly 1994), there is general agreement that within New Zealand at least 18 distinct families all have representatives with the divaricate habit (Lloyd 1985). Furthermore, there are at least 11 New Zealand species from nine separate genera for which the divaricate habit is restricted to the juvenile stage. All this suggests some strong selective benefit in shrubs and juvenile trees from the divaricate habit. However, there has been extensive debate about what this selective benefit might be.

The term divaricate refers to branching at a wide angle (Allan 1961), and this interpretation has been favoured by some (Cheeseman 1925; Went 1971; Eagle 1982). However, the term has also been applied in a wider sense to describe the characteristic growth form of interlaced small-leaved shrubs (Greenwood & Atkinson 1977; Kelly 1994; Taylor 1975; Wardle 1963; Wilson & Galloway 1993). Another important characteristic of the divaricate growth form is that many species tend to have no or few leaves in the outer canopy (Kelly 1994); most leaves are found in the interior of the plant. While many plants do exhibit divaricate branching in the literal sense, both in New Zealand (e.g. Rubus squarrosus and Hebe vernicosa (Cheeseman 1925) and elsewhere (Lloyd 1985; McQueen 2000), the frequency of the divaricate growth habit in New Zealand is uniquely high.

This phenomenon has long attracted speculation as to its cause (Cockayne 1912; Kelly 1994). One hypothesis attributes it to lingering effects of recently extinct moa (Atkinson & Greenwood 1989; Carlquist 1974; Greenwood & Atkinson 1977; Taylor 1975). Within their browse zone, the divaricate form supposedly conferred resistance to these giant, flightless herbivorous birds. Diamond (1990) went so far as to claim that the ghosts of moa could still be seen in their residual influence on divaricate New Zealand shrubs. The moa hypothesis is an appealing explanation for why only New Zealand has so many divaricates, because moa were also unique to New Zealand. However, subfossil moa gizzards contain large numbers of twigs of divaricate plants, showing that this form did not completely prevent browsing (Burrows 1980; Burrows 1989).

The main alternative hypotheses propose that climatic factors favour the divaricate form. Supposed climate benefits include sheltering the leaves from frost, wind or low humidity (McGlone & Clarkson 1993; McGlone & Webb 1981). However, experimental work has failed to find any significant amelioration of the temperatures or humidities experienced by the leaves of divaricates (Kelly & Ogle 1990). Moreover, none of these climatic factors is uniquely strong in New Zealand (Greenwood & Atkinson 1977). One environmental factor which has not been considered for its effect on divaricates is strong irradiance, which inhibits photosynthesis (Ball 1994; Powles 1984). Photoinhibition occurs when light is absorbed by the photosynthetic apparatus in excess of the photosynthetic requirement, leading to a light-dependent decline in photosynthesis (Osmond 1981). There are two major mechanisms by which this can occur: (i) direct damage to the photosystem II (PSII) reaction centres (photodamage; Krause 1988) and/or (ii) increased energy dissipation by protective quenching mechanisms (photoprotection) (Björkman & Demmig-Adams 1994; Demmig-Adams & Adams 1994; Demmig-Adams et al. 1995; Gilmore 1997). Photoinhibition can be exacerbated by any environmental stresses (e.g. cold temperatures, drought) that limit the capacity of the photosynthetic apparatus to use photon fluxes received (Osmond 1994). Photoinhibition is increasingly recognized as an important factor in plant ecology (Ball et al. 1994). Divaricate shrubs are most abundant in exposed, frosty inland habitats in the central South Island of New Zealand (McGlone & Clarkson 1993; Wilson & Galloway 1993), where they may often be exposed to bright sunshine following a cold night. These are conditions which would promote cold-induced photoinhibition. Indeed, photoinhibition affects non-divaricate tree seedlings in this area (Ball 1994).

In this study, experimental manipulations of plant canopies under field conditions were used to examine the responses of three divaricate species to the combined effects of high light and low temperature. We propose that the primary selective advantage of the divaricate syndrome is that it mitigates the effects of photoinhibition by reducing the irradiance incident on photosynthetically active tissues. We tested the hypothesis that the leaves of divaricate plants are sensitive to cold-induced photoinhibition, and that self-shading by outer branches reduces irradiance enough to prevent photodamage. Finally, we consider the significance of the observed responses to carbon gain in these plants.

Materials and methods


Field measurements and experiments were conducted at the University of Canterbury field station at Cass (171°45·5′ E, 43°02′ S, 650 m altitude), in the Arthur’s Pass region of the South Island of New Zealand. The Cass study site has been the location for previous work on leaf temperatures in divaricate shrubs (Kelly & Ogle 1990) and photoinhibition in non-divaricate trees (Ball 1994), and is in the area that contains the highest concentration of divaricate species (Wilson & Galloway 1993). Field experiments were conducted between 22 June and 2 September, 1998. An automatic weather station at the site recorded light and temperature during winter 1998. Ambient intensity of photosynthetically active radiation, expressed as photon flux density (PFD), was measured using a Li-Cor 190SB quantum sensor (Li-Cor Inc, Lincoln, Nebraska). Air temperature was measured inside a Stevenson screen at 1·5 m above ground level. Fifteen-minute averages of PFD and air temperature were recorded using a Campbell CR10X data logger (Campbell Scientific Inc, Logan, Utah). This enabled us to investigate the relationship between daily minimum temperature and daily PFD. As the minimum temperature was almost always achieved just before dawn, maximum irradiance in the photoperiod following the minimum temperature is presented. During the study, daytime maximum air temperatures ranged from 4 to 16 °C. There were 42 frost events with a minimum temperature of –8·3 °C, and frosts were usually followed by bright sunshine (the 13 highest daily mean PFDs followed frosts).

Three common species regarded as divaricate (Kelly 1994) were used in this study. Coprosma propinqua (Rubiaceae) is a member of a genus distributed throughout the South Pacific, although 44 of 45 New Zealand species are endemic. Coprosma propinqua ranges over almost the entire country, and is only absent from the extreme north and at high altitudes (Salmon 1996). It is variable in stature, being prostrate and mat-like on exposed coastal sites, yet it can attain heights of up to 5 m in sheltered areas. At Cass, C. propinqua assumes a shrub habit to about 3 m. Shrubs appear green in summer but leaves often develop a brownish or bronzed coloration in winter. Aristotelia fruticosa (Elaeocarpaceae) exhibits remarkable leaf and branching dimorphism between juvenile and later phases. Juvenile plants have deeply toothed leaves and monopodial branching. This form is retained when plants grow in sheltered conditions. In older plants in exposed sites, regular dieback of the terminal shoots produces a sympodial architecture (Tomlinson 1978). There is an obvious increase in leaf size with increasing depth into the canopy. At Cass A. fruticosa appears to favour south-facing sites, where it assumes a tight divaricate form. Corokia cotoneaster (Escalloniaceae) has very distinctive zig-zag interlacing branchlets (Greenwood & Atkinson 1977). Corokia cotoneaster is found in open shrublands throughout New Zealand; it occurs in great abundance at Cass and the surrounding areas, and tolerates a wide variety of conditions including a range of soil fertility, as it is found on alluvial fans of almost any age.

Of the three species, A. fruticosa and C. cotoneaster display the strongest architectural dimorphism, with branches bearing a few small leaves in the outer canopy and many more larger leaves in the inner canopy. Coprosma propinqua, whilst displaying the highly branched shoot structure and small leaves typical of divaricate plants, possesses full-sized leaves throughout the canopy.


Four Li-Cor 190SB quantum sensors were used on each of three Campbell CR21X dataloggers, to measure PFD attenuation into the canopies of the three species of divaricate shrub. The outermost sensor was located at the periphery of the shrub on the north side; the remaining three were located at 10 cm increments into the canopy. Instantaneous PFD was simultaneously measured on each sensor every 10 s and recorded as 5-min averages. This was repeated for each species on 6 days; on each day new plants were randomly chosen for measurement. PFDs from the three depths within the canopy were expressed as a percentage of the PFD at the periphery of the canopy, calculated for each 5-min period during daylight hours over the 6 days and then averaged for each species.

On eight randomly selected replicate plants per species, leafy shoots 20 cm inside the north (sunny) side of the shrubs were exposed to one of three experimental treatments over winter: (i) control leaves which were not manipulated; (ii) exposed leaves which had their outer screen of branches pruned away leaving them open to low-angle winter sunlight; or (iii) shaded leaves which were pruned, then sheltered from direct sunlight with a piece of vertically oriented shade cloth but remained exposed to the sky. Because of the interlaced nature of the canopies of these species, the pruning of outer branches did not result in direct wounding of the shoots to be used for subsequent measurements. Physiological measurements were made at treatment initiation (23 June–6 July), on 14–15 July and finally on 1–2 September.

Dark-adapted maximal photochemical efficiency of PSII (Fv/Fm, variable fluorescence/maximal fluorescence yield; Bolhar-Nordenkampf & Oquist 1993) was measured using a portable chlorophyll fluorometer (MINI-PAM, Photosynthesis Yield Analyzer, Walz, Effeltrich, Germany). Measurements were made throughout the day at ambient temperature on leaves selected at random among species and treatments. For each stem, three leaves were randomly chosen at each visit. Leaf clips (DLC-8) were used to dark-adapt each leaf for exactly 15 min before each recording was taken. The order of sampling was retained for each sampling day, thus leaves were sampled at the same time each day. This sampling strategy was employed in the field to reduce variation in Fv/Fm caused by differences in relaxation kinetics and short-term regulatory reductions in the quantum efficiency of PSII in the light.

Five of the eight plants from each of the three species were randomly selected, and the response of photosynthesis to photon flux density (PFD) for each of the three treatments within those plants was determined about 50 days after the start of the experiment (6–16 August). Shoot gas exchange was measured using a Li-Cor 6400 infra-red gas analyser equipped with CO2 and light-control modules. Light-response curves were generated following measurements of steady-state responses of photosynthesis (A) to PFD. Photon flux densities were supplied by a blue-red light source mounted above the leaf cuvette in 11 steps from 0 to 1500 µmol m−2 s−1. Measurements were made at each PFD set point when photosynthetic gas exchange had equilibrated (taken to be when the coefficient of variation for the CO2 partial pressure differential between the sample and reference analysers was below 1%). This condition was typically achieved in 4–5 min after a stable PFD set point had been reached. Measurements were made on leaves assigned at random throughout the course of the day at leaf temperatures of 15 °C. Water vapour pressure deficit was generally held between 1·0 and 1·5 kPa. Combined leaf and stem areas, required to calculate net photosynthesis per cm, were measured for all selected stems using calipers. Stems were regarded as two-dimensional, so projected stem area was measured. Leaves were regarded as ellipsoid, so length and width were recorded for each leaf. This method of calculating area was accurate (±5%) when compared to values attained using an area meter (Delta-T Devices, Burwell, Cambridge, UK). For statistical analysis Amax was calculated as the mean of all points from PFD ≥ 500 µmol m−2 s−1 (beyond which photosynthesis was saturated), and the apparent quantum yield of photosynthesis was calculated from the initial slope of the light-response curve (below 100 µmol m−2 s−1).


Two-way anova was used to test for the main effects and interactions of species and treatment on photosynthetic characteristics using s-plus version 3·3 (MathSoft Inc., Seattle, USA). The three treatments (control, exposed, shaded) were blocked within eight randomly selected plants per species, and plant was used as a block effect in all anovas.


In all three species, the outer branches had an attenuating effect on PFD at short distances into the canopies (Fig. 1). The level of attenuation was similar between species, with PFD at 30 cm into the shrubs reduced to 20–40% of that outside the canopy.

Figure 1.

Percentage reduction in photon flux density with depth into the canopies of three species of divaricate shrub. Percentage attenuation was calculated for each 5-min measurement average, and values averaged over all times of the day from six replicate days.

During the 71 days following the initiation of the experimental treatments in mid-winter (22 June), plants experienced 42 frosts. Following these sub-zero temperatures, daily maximum PFDs were between 800 and 2000 µmol m−2 s−1 (Fig. 2a), while average PFDs for the hour following sunrise were in the range 100–1100 µmol m−2 s−1 (Fig. 2b). While high PFDs did sometimes occur following relatively warm nights, maximum PFDs following daily minimum temperatures above zero were quite variable, and generally much lower than those following frost events.

Figure 2.

The relationships between (a) minimum air temperature at sunrise and maximum photon flux density (PFD) during the following photoperiod, and (b) minimum air temperature at sunrise and average PFD for the hour following sunrise at the Cass field site during this study.

Our results show the development of severe photoinhibition following exposure of inner canopy leaves to the combined effects of high light and low temperature. After 21 days exposure to elevated light and cold temperature, the fluorescence parameter Fv/Fm was significantly lower in exposed treatments in all three species (treatment effect: A. fruticosa, F2,45 = 18·1, P < 0·001; C. cotoneaster, F2,45 = 24·8, P < 0·001; C. propinqua, F2,45 = 14·0, P < 0·001; Fig. 3). This indicates severe light-dependent reduction in the photochemical efficiency of PSII for leaves exposed to normal daylight. No such reduction occurred in either shade or control treatments. The reductions in photochemical efficiency that occurred in the 21 days following treatment initiation were maintained until 71 days in two of the three species. Substantial recovery was demonstrated in exposed leaves of C. propinqua between 21 and 71 days (Fig. 3), although the treatment effect was still significant (F2,45 = 5·34, P = 0·009). There was no such recovery for A. fruticosa or C. cotoneaster (Fig. 3; F2,45 = 26·2, P < 0·001 and F2,45 = 43·7, P < 0·001, respectively).

Figure 3.

Response of leaf fluorescence in three divaricate shrub species to cold-induced photoinhibition. Exposed treatments had the outer shading screen of branches removed; shaded treatments had outer branches replaced with shade cloth. The figure shows the time course of changes in the dark-adapted maximal photochemical efficiency of PSII Fv/Fm (± SEM) for A. fruticosa, C. cotoneaster and C. propinqua. The dotted line indicates when light response curves (Fig. 4) were measured.

At day 50, exposed leaves showed markedly lower rates of photosynthesis than control and shaded leaves (Fig. 4a–c). The reduction in photosynthetic performance is particularly obvious in exposed leaves of A. fruticosa and C. cotoneaster between 300 and 1500 µmol m−2 s−1. In all cases, shade treatment foliage was indistinguishable from the control. The light-saturated photosynthetic rate (Amax) of exposed leaves was reduced by 53% in A. fruticosa (5·10 versus 2·39 µmol m−2 s−1; F2,8 = 6·82, P < 0·05) and 49% in C. cotoneaster (3·96 versus 1·93 µmol m−2 s−1; F2,8 = 8·28, P < 0·05), but by only 11% in C. propinqua (F2,8 = 0·48, NS). The quantum yield of photosynthesis responded in a similar manner to Fv/Fm and Amax, and was reduced by 29% in A. fruticosa (0·035 versus 0·025; P < 0·05) and 57% in C. cotoneaster (0·30 versus 0·013; P < 0·05), but not significantly affected in C. propinqua (0·029 versus 0·028).

Figure 4.

Response of leaf gas exchange in three divaricate shrub species to cold-induced photoinhibition. Exposed treatments had the outer shading screen of branches removed; shaded treatments had outer branches replaced with shade cloth. (a–c) Photosynthetic light-response curves for all treatments on day 50 plotted against photon flux density at the leaf surface. (d–f) Photosynthetic light-response curves in exposed and control treatments plotted as a function of photon flux density outside the shrub. Actual rates of photosynthesis in control leaves under field conditions will be modified by the reduced irradiance they experience due to the screen of outer branches – for the control treatment the photon flux density used to calculate photosynthetic rate has been rescaled to account for the measured amount of self-shading.

Actual rates of photosynthesis in control and shaded leaves under field conditions will be modified by the reduced irradiance they experience due to the screen of outer branches. The branches reduced irradiance 20 cm into the plants to 45% of ambient in A. fruticosa, 30% in C. cotoneaster and 20% in C. propinqua (Fig. 1). When the light-response curves for control and exposed leaves are rescaled to show the combined effects of shading and depression of photosynthesis on day 50 (Fig. 4d–f), A. fruticosa control leaves photosynthesized more quickly than exposed leaves at all light intensities, and C. cotoneaster control leaves had higher rates in bright light (> 500 µmol m−2 s−1). In contrast, because exposed leaves of C. propinqua recovered more rapidly from the initial depression in photosynthesis (Fig. 2c), and because this species showed the deepest shading at 20 cm into the canopy, by day 50 control leaves in C. propinqua had lower photosynthetic rates than exposed leaves at all PFDs.

In addition to the responses above, exposed leaves of A. fruticosa were also much more likely to abscise over winter (69% lost) than control (36%) or shaded (7%) leaves (F2,14 = 6·35, P = 0·011), consistent with shedding after severe photoinhibition damage, whereas neither C. cotoneaster nor C. propinqua showed a significant treatment effect on net leaf loss (F2,14 = 1·21 and 0·80, P = 0·33 and 0·47, respectively). Net photosynthesis in exposed A. fruticosa branches would therefore be reduced by smaller photosynthetic capacity in surviving leaves, and by having fewer leaves.


This study shows a clear and present benefit in New Zealand divaricate shrubs from the sheltering outer layer of sparsely leafed branches which is typical of their growth form. When these branches were removed, increased irradiances rapidly reduced the photosynthetic capacity of all three species studied. The large and sustained reductions of Fv/Fm and Amax in C. cotoneaster and A. fruticosa are characteristic of photoinhibition of photosynthesis, i.e. severe light-dependent reductions in the photochemical efficiency and the light-saturated rate of photosynthesis. Such reductions can persist for several months (Ball et al. 1991; Oberhuber & Bauer 1991). Both A. fruticosa and C. cotoneaster showed almost no capacity for recovery from high light stress. These species also show the strongest morphological differences between the outer and inner canopy, and seem to rely heavily on an architectural means of photoprotection. The other species (C. propinqua) has similar-sized leaves throughout the canopy, suffered smaller reductions in Fv/Fm and recovered relatively swiftly following exposure of inner leaves to strong irradiance. Although the precise mechanisms have yet to be elucidated, the recovery of C. propinqua suggests that it can either limit photodamage by invoking photochemical mechanisms for dissipation of excess excitation energy (Demmig-Adams et al. 1995), or rapidly repair photodamage.

Exposure of interior foliage to light conditions usually restricted to the canopy edge during cold temperatures, reduced Amax for both A. fruticosa and C. cotoneaster by ≈50% (Fig. 4). However, there was no such long-term reduction for C. propinqua. There is a potential problem when interpreting these results that should act as a caution. The observed response by leaves that developed under moderate light and were subsequently exposed to stronger light may differ from the responses of leaves that developed under high light. However, the irradiances experienced by interior leaves in our experimental plants (20–40% of incoming irradiance) are not low when compared to the deep shade experienced by understorey plants (often 0·5–3% of ambient; Chazdon et al. 1988; Turnbull & Yates 1993). Thus these experimental leaves cannot be considered heavily shade-adapted. We would also argue that photosynthetic acclimation can account for larger increases in PFD in more shade-adapted leaves (Turnbull et al. 1993). In addition, our finding that C. propinqua, which possessed the most heavily self-shaded interior leaves (20% of incident irradiance; Fig. 1), was able to recover fully from the initial transfer to a higher light regime indicates that the increase in irradiance alone was not drastic.

The concept that the divaricate habit manipulates light at the canopy level to minimize photoinhibition is a relatively novel one. However, mechanisms for regulating the light energy absorbed by leaves are well established. Such mechanisms include reflective hairs (Bisba et al. 1997), anthocyanin pigments (Adams et al. 1992), and vertical orientation of leaves (Björkman & Demmig-Adams 1994). The assumed carbon cost associated with internal leaves in the canopies of divaricate shrubs (Greenwood & Atkinson 1977; McGlone & Webb 1981) is based on the observation that such leaves suffer significant self-shading. Within the canopies of A. fruticosa, C. propinqua and C. cotoneaster, incident PFD at 30 cm ranges between 20 and 40% of that received at the edge of the canopy (Fig. 1). Photosynthesis in most higher plants is typically saturated at about 20–30% of full sunlight (Larcher 1995). This irradiance is still attained at 20 cm below the edge of the canopy for C. propinqua, and is exceeded at a depth of 30 cm for both A. fruticosa and C. cotoneaster. When incident radiation is high there will be sufficient light, even at considerable depths into divaricate shrubs, to saturate photosynthesis. Reducing an excess of light beyond saturation will incur little cost and may have major benefits.

Another of the hypothesized limitations associated with the divaricate habit is that height growth is slow and plants could be overtopped by species with stronger apical dominance (Greenwood & Atkinson 1977). However, the sites which divaricate shrubs typically inhabit do not usually support taller vegetation. In addition, in the recent past New Zealand’s climate has often restricted the development of forests, for example during the Pliocene–Pleistocene glaciation (Mildenhall 1980). Current natural tree lines in New Zealand are typically determined by cold temperatures, waterlogged soils, salinity or strong winds. These potential stresses reduce the light requirement to saturate photosynthesis and therefore the irradiance at which light is received in excess. It may be that divaricate plants are found in all these areas because they can reduce the irradiance to leaves and ameliorate photoinhibitory effects when in combination with environmental stresses.

It was initially proposed that current conditions are not harsh enough to account for the divaricate form, and that cooler, drier and windier conditions in the Pleistocene had driven the convergent evolution of the divaricate form (Cockayne 1912). Following a long period of relatively stable subtropical to warm temperate climate throughout the Tertiary, the onset of cooler and drier conditions in the Pleistocene (2 million years ago) was relatively rapid (Mildenhall 1980). This would have had a significant effect on New Zealand’s largely subtropical flora (McGlone & Webb 1981). More recently, it has been suggested that while the conditions throughout the early Quaternary undoubtedly played a major role, the divaricate habit is an adaptation to still-existing conditions (Wardle 1963). Data presented in this study indicate that cold-induced photoinhibition of photosynthesis can occur under existing climatic conditions. The strong irradiance experienced during winter at our field site, which is at the centre of the distribution of divaricate plants in New Zealand, is an important environmental factor. But are the stresses which exacerbate photoinhibition more likely to be a problem in New Zealand than in other areas? Several lines of evidence suggest that they are. First, the New Zealand climate is relatively oceanic and aseasonal, so cold conditions may come at any time of year (McGlone & Webb 1981; Sturman & Tapper 1996). Second, clear skies and moderate latitude combine to give strong irradiance, even in winter (Fig. 2). While the maximum PFDs recorded on clear days at Cass are similar to those observed in other regions during winter, e.g. Austria (Oberhuber & Bauer 1991) and Colorado, USA (Adams et al. 1994; Logan et al. 1998), these events, especially those associated with very cold mornings and moderately warm days, are frequent. For these reasons, deciduousness (which is rare in New Zealand; Batcheler 1989) or winter dormancy are not favoured because too many opportunities for photosynthesis are lost (McGlone & Clarkson 1993), and the divaricate growth form may provide a viable alternative.

In conclusion, the severity and long duration of the reductions in photosynthesis we have measured indicate that avoiding cold-induced photoinhibition of photosynthesis could be of selective advantage to divaricate shrubs in the New Zealand climate. It has been said in the past that the divaricate growth form imposes a cost of self-shading (McGlone & Webb 1981), which then must be compensated for by other advantages such as frost protection or reduced moa browsing. Our data show that self-shading may confer a direct benefit: transiently exposed leaves suffer a potentially damaging surfeit of light over winter, and in the most strongly divaricate species net photosynthesis is greater where the leaves are partially shaded. An intriguing outcome of our observations is that divaricate plants may differ quite markedly in their susceptibility to photoinhibition and their ability to recover from it. It is through these functional differences that we further hypothesize that the extent of the divaricate form (e.g. the strong architectural dimorphism between outer and inner canopy) will be related to the susceptibility of leaves to stress-induced photoinhibition. While limiting photoinhibition may not be the sole explanation for divarication in every species of divaricate shrub in New Zealand, our findings provide a clear enough benefit to question the need to see moa ghosts (Diamond 1990) still shaping New Zealand’s vegetation.


This research program is supported by the Marsden Fund of the Royal Society of New Zealand (Grant UOC001-2001). We gratefully acknowledge the advice and assistance of Dr Matt McGlone in developing this research. We thank Dr Osbert Sun for the use of the Mini-PAM leaf fluorescence analyser.