Latitude, solar elevation angles and gap-regenerating rain forest pioneers


Correspondence author. E-mail:


1. Tropical rain forests have more species-rich tree assemblages than forests at higher latitudes, but is this because they comprise a wider array of niches or functional types? We address this by considering one tree functional type – light-demanding canopy trees with fast foliage turnover and growth – that is common in the tropics and subtropics, but virtually absent from mid-latitude rain forests. Although often referred to as ‘tall pioneers’ or ‘large pioneers’, they are by no means confined to early-successional stages, also recruiting directly to the canopy in old-growth stands by rapid growth beneath tree-fall gaps.

2. We also explored the influence of latitude on tree-fall gap light environments as a possible constraint on the geographic distribution of this functional type, using the YPLANT program to simulate light interception and potential carbon gain by seedlings of the Australian rain forest pioneer Polyscias murrayi beneath idealized gaps at tropical, subtropical and cool temperate sites (latitudes 17, 29 and 42°S, respectively). P. murrayi grows quickly to heights of 20–25 m, has high photosynthetic capacity and respiration rates, and a leaf life span of 6–9 months.

3. Simulated light interception and potential carbon gain were strongly influenced by latitude, and by the interaction of latitude with position within an idealized tree-fall gap of 100 m2. Potential net daily carbon gain of P. murrayi was strongly positive beneath the gap centre at latitude 17, and beneath the poleward (i.e. southern) gap margin at latitude 29, but negative beneath both the gap centre and margin at latitude 42. Light interception and carbon gain were also influenced by geographic variation in sunshine hours, which were highest at latitude 29 and lowest at latitude 42. A larger gap of 300 m2 permitted positive net carbon gain at all latitudes, although rates were again predicted to be highest beneath the gap centre in the tropics.

4.Synthesis. YPLANT simulations supported the hypothesis that sun angles could prevent trees with high metabolic rates from invading old-growth mid-latitude rain forests, where light environments suitable for their establishment will be scarce. Geographic variation in forest light environments is therefore likely to influence the range of viable functional types at different latitudes.


Latitude is the single most conspicuous correlate of global patterns of species richness in terrestrial ecosystems (Willig, Kaufman & Stevens 2003). Perhaps nowhere is the latitudinal diversity gradient more evident than in rain forests, which occur over a range of more than 50° of latitude in both hemispheres. A single hectare of lowland tropical rain forest often harbours >100 tree species (Richards 1996), typically declining to 25–100 in the subtropics, 10–25 at warm temperate latitudes and 2–10 in cool temperate forests (Floyd 2008). In general, patterns of species richness reflect geographic variation in rates of both origination and maintenance of species (Jablonski, Roy & Valentine 2006). Species origination rates may be high in the tropics because of the effects of temperature and growing season length on mutation rates, generation times and speed of selection (Rohde 1992). High species maintenance rates also seem possible in the tropics, because of their relative climatic stability compared with higher latitudes (Wallace 1878).

It has been conjectured that geographic variation in the range of niche availability also shapes patterns of tree species richness, by influencing species maintenance rates (Whittaker 1969; Ricklefs 1977; Loehle 2000). Potential contributing factors include latitudinal variation in the range of light environments present in forests (Ricklefs 1977) and relaxed constraints on plant form and function in the benign physical environment of the humid tropics. If there are in fact more ways for trees to ‘make a living’ in the tropics than at higher latitudes, we would expect the former to harbour a wider range of life-history traits, and of traits related to the acquisition, allocation and conservation of resources – or perhaps more varied combinations of such traits. Relevant traits include those associated with leaf economics, xylem anatomy and physiology, maximum height and seed size (Reich, Walters & Ellsworth 1992; Westoby et al. 2002; Wright et al. 2004; Poorter et al. 2009). However, we are not aware of any comparison of the distribution of such traits, or trait combinations, across humid forest assemblages at different latitudes.

Fast-growing, light-demanding canopy or emergent trees that recruit directly in tree-fall gaps represent one functional type that may be largely confined to tropical and subtropical rain forests. Familiar examples include Ceiba pentandra L., Ochroma pyramidale (Cav. ex Lam.) Urb. (Malvaceae) and many species of Cecropia in the neotropics (Poorter & Bongers 2006), Aleurites spp. (Euphorbiaceae) in Malesia and northern Australia (Grubb 1996), and Polyscias spp. (Araliaceae) in eastern Australia (Hyland et al. 2002; Floyd 2008). To our knowledge, nothing comparable to these taller pioneers has been described from temperate rain forests. Like their tropical counterparts, mid-latitude rain forest assemblages include small, fast-growing trees and shrubs that are prominent in early successional stages, e.g. Embothrium coccineum J.R. & J.G. Forst. (Proteaceae) in South America (Lusk 2002) and Pseudopanax arboreus (Murray) K. Koch (Araliaceae) in New Zealand (Wardle 1991). Temperate rain forests also include long-lived, light-demanding trees that establish mainly after coarse-scale disturbance and persist as scattered emergents in late-successional stages by virtue of their great longevity, despite regenerating poorly in old-growth stands. Examples include Nothofagus dombeyi (Mirb.) Blume (Nothofagaceae) in Chile (Veblen 1980), Podocarpus totara D. Don ex Hook. (Podocarpaceae) and Agathis australis (D. Don) Lindl. (Araucariaceae) in New Zealand (Ogden & Stewart 1995), and Pseudotsuga menziesii (Mirb.) Franco (Pinaceae) in the Pacific Northwest (Franklin & Hemstrom 1981). Although sometimes described as ‘long-lived pioneers’, their seedlings can often be found in forest understoreys, even though canopy thinning or opening is apparently required for development to larger size-classes (e.g. Ogden, Wardle & Ahmed 1987). Traits differentiating them from gap-regenerating tropical pioneers include great longevity (generally >500 years), well-defended wood, modest growth rates and relatively slow foliage turnover (Loehle 1988; Ogden & Stewart 1995).

Here we highlight the latitudinal distribution of tall, gap-regenerating pioneers, by reviewing data on typical potential heights and leaf life spans in rain forests growing at temperate versus tropical/subtropical latitudes. We then model the influence of latitude on light availability and potential carbon gain beneath tree-fall gaps (Ricklefs 1977) as one possible explanation for the latitudinal distribution of these taxa. Average solar altitude angles decrease with increasing latitude, meaning that less direct beam radiation penetrates into tree-fall gaps in temperate old-growth forests than in their tropical counterparts (Canham et al. 1990; Prentice & Leemans 1990; Weishampel & Urban 1996). This is in addition to the general geometric effect of latitude on the incidence of solar radiation outside the atmosphere, and on the mean path length of the solar beam through the atmosphere (Monteith 1972). If the potential for seedling carbon gain is lower in tree-fall gaps in the temperate zone than in the tropics, light-demanding species there would have less chance of growing fast enough to overhaul the advance growth of shade-tolerant species, and might even be unable to fix enough carbon to survive. We used the YPLANT program (Pearcy & Yang 1996) to simulate the effect of latitude on light interception and potential carbon gain of seedlings of a common Australian rain forest pioneer (Polyscias murrayi, Araliaceae) in tree-fall gaps.

Materials and methods

Functional diversity in tropical/subtropical versus temperate rain forests

We used leaf life span as an indicator of species light requirements, and examined the relationship of this trait with potential height in tropical/subtropical and temperate rain forests. Although most existing data on species shade tolerance or successional status are not expressed in comparable units across sites or regions, interspecific variation in light requirements and growth rates within both tropical and temperate evergreen-dominated rain forest assemblages is related to leaf life span and other leaf economic traits (e.g. Lusk 2002; Poorter & Bongers 2006). We chose leaf life span as a focal trait of interest because its interpretation is more straightforward than that of other leaf economic traits such as leaf mass per area, nitrogen content or photosynthetic capacity. As life span is influenced by the light environment in which a leaf grows (e.g. Lusk 2002; Reich et al. 2004), we only included life spans obtained from what were described as ‘sun leaves’, or from microsites with ≥10% canopy openness. Our review included all evergreen rain forest tree and shrub species for which both height and leaf life span data could be found using ISI Web of Science.

We also sought to compare gas exchange, and wood and seed traits of different tree functional types in tropical/subtropical and temperate rain forests. To this end, we obtained data on photosynthetic capacity, dark respiration rates, wood density and mean conduit diameters of as many as possible of the species occurring in our height/leaf life span dataset. We classified species with leaf life spans ≤12 months as ‘pioneers’ and those with longer leaf life spans as ‘non-pioneers’. Although this threshold is arbitrary, it corresponds well with data published by Coley (1988), who reported that of 20 Panamanian rain forest species with saplings found only in treefall gaps, 19 (95%) had leaf life spans of ≤12 months. Conversely, 15 out of 21 species described as ‘shade-tolerant’ had leaf life spans >12 months. A potential height threshold of 20 m was used to split the pioneer species further into ‘short’ and ‘tall’ categories. Although separating tree species into three groups suited our comparative purposes, we note that the literature indicates a continuum of variation in height and leaf life span in rain forest assemblages (e.g. Coley 1988; Poorter & Bongers 2006).

The approach to modelling light interception and carbon gain

Version 3.1 of YPLANT (Pearcy & Yang 1996) was used to model the effect of latitude on average daily light interception and carbon gain by juvenile trees beneath idealized tree-fall gaps. Simulations using YPLANT have been shown to correspond well with actual measurements of light interception and carbon gain (Valladares & Pearcy 1998). YPLANT inputs are the three-dimensional geometry of leaf and stem arrangement, a description of leaf shape, leaf photosynthetic capacity and respiration rate, plus a digital ‘canopy file’ describing canopy structure above the plant, that is used to model the light regime. The latter is usually derived by analysing a hemispherical photograph with software such as CANOPY (Rich 1990), or HEMIVIEW (Delta-T Devices Ltd, Cambridge, England), and enables estimation of the photosynthetic photon flux density (PPFD: μmol photons m−2 s−1) incident on each leaf surface at different times of day. Potential carbon assimilation rate can be calculated, using PPFD response curves generated from species-specific measurements of gross photosynthetic capacity (Amax) and dark respiration rate (Rd), using the Thornley method (Thornley 1976). YPLANT enables the user to model the effect of latitude and time of year on light interception and potential carbon gain.

YPLANT 3.1 also enables the user to simulate light interception under a range of atmospheric conditions. This is achieved by varying the atmospheric transmission coefficient Rs/Ro, where Ro is the global solar irradiance at the top of the atmosphere, and Rs is the global solar irradiance at the top of the canopy (Roderick et al. 2001). When Rs/Ro is high (i.e. clear sky) most irradiance arrives as direct beams, whereas diffuse light predominates under overcast conditions, when this coefficient is low. We simulated light interception and carbon gain under both clear and overcast skies, setting Rs/Ro at 0.3 to mimic overcast conditions, and at 0.79 to reproduce the effect of a clear sky (Roderick et al. 2001). Under sunny conditions, plants are periodically illuminated by direct light (‘sunflecks’) when the solar disk coincides with canopy openings, but receive only small amounts of diffuse light between sunflecks. Data on mean daily sunshine hours at tropical, subtropical and cool temperate rain forest sites in eastern Australia (Table 1) were used to weight the output from simulations under clear and overcast skies, thus obtaining overall estimates of average daily carbon gain. For example, a figure of 6.8 sunshine hours per day (out of a possible 12 h of daylight) was obtained by averaging data from the three meteorological stations located within the tropical rain forest region of northern Queensland (Table 1). The weight given to the output from simulations under a clear sky at latitude 17 was therefore 6.8/12.0 = 0.567.

Table 1.   Geographic and climatic data for Australian tropical, subtropical and cool temperate rain forest regions in which juvenile tree light interception and carbon gain was simulated, including all stations that record sunshine hours. Data courtesy of the Australian Bureau of Meteorology (
LocalityLatitude (°S)Longitude (°E)Altitude (m)Mean maximum warmest month (°C)Mean minimum coldest month (°C)Mean annual precipitation (mm)Mean daily sunshine hours
  1. *Although there are other meteorological stations within the subtropical rain forest region of northern New South Wales and southern Queensland, none records sunshine hours.

 Kairi Research Station17.22145.5769728.911.112956.8
 Pin Gin Hill17.53145.974031.215.335906.7
 South Johnstone17.61146.001831.215.032946.3
       Mean 6.8
Cool temperate       
 Savage River41.49145.2135220.13.319394.8
 Lake St. Clair42.10146.2273518.80.315114.9
       Mean 4.6

Plant material for simulations

We modelled light interception and carbon gain by juvenile Polyscias murrayi (F.Muell.) Harms (Araliaceae) (Fig. 1), one of the most common and widespread pioneer trees in the tropical and subtropical rain forests of eastern Australia (Hyland et al. 2002; Floyd 2008). As well as regenerating profusely after major disturbances that expose mineral soil, P. murrayi establishes in tree-fall gaps in old-growth stands. It has large pinnate compound leaves up to 120 cm long, often grows as much as 12 m in height before branching, and attains heights of 20–25 m. Monitoring of leaf survival on saplings indicated leaf life spans of 6–9 months (Lusk & Kooyman, unpublished data).

Figure 1.

 Overhead and lateral views of a three-dimensional crown reconstruction of a ∼650 mm tall juvenile Polyscias murrayi used in YPLANT simulations of light interception and potential carbon gain in tree-fall gaps.

We used a magnetic digitizer to capture the architecture of three juvenile P. murrayi, ranging from 620 to 660 mm tall. This was carried out in subtropical rain forest in Nightcap National Park (NSW, Australia) located at 28º38′S, 153º20′E and at elevation of 380 m a.s.l. Mean annual temperature at the nearest meteorological station (Whian Whian) is estimated at 17.6 °C and mean annual rainfall at 2314 mm, with a minimum in late winter to early spring (Bureau of Meteorology, The selected seedlings were growing at microsites with between 10% and 20% canopy openness, as quantified by hemispherical photography. The sampled plants were thus growing in light environments similar to those experienced beneath single tree-fall gaps in a 25-m-tall canopy. Although taller plants will experience higher light levels in tree-fall gaps than seedlings of the size that we worked with, pioneers such as P. murrayi are rarely present as advanced growth prior to gap formation, and addressing our question required us to assess light interception and carbon gain potential in the environments that their seedlings typically establish in.

Capturing plant architecture

Digital capture of plant architecture (Hanan & Room 1997; Falster & Westoby 2003a) is much less time-consuming than the manual methods often used in conjunction with YPLANT. The three-dimensional leaf arrangement of each seedling was recorded using a 3SPACE® 3D-digitizer (Polhemus, Colchester, Vermont, USA), in conjunction with the software package FLORADIG (CSIRO Entomology, Brisbane, Australia). The digitizer includes a magnetic signal receiver and pointer, allowing the user to record the 3D spatial co-ordinates of the pointer within a hemisphere of 3 m diameter from the receiver. Individual plants are reconstructed virtually by recording a series of point co-ordinates, and the relevant connectivity between points. Stem segments and petioles are characterized by their elevation angle, azimuth, length and diameter. Individual leaves are characterized by their length together with the azimuth and elevation angle of two vectors on the lamina surface. The compound leaves of P. murrayi were treated as branches, each leaflet being digitized as an individual leaf. The 3D description of leaf arrangement recorded for each seedling in FLORADIG was converted to the appropriate YPLANT format using a program written in the C programming language (Falster & Westoby 2003a).

Gas exchange parameters

Simulation of carbon gain in YPLANT requires measurements of photosynthetic capacity (Amax) and dark respiration rates (Rd). A LI-6400 photosynthesis system (LI-COR Biosciences, Lincoln, Nebraska, USA) was used to measure Amax and Rd on each of the three juvenile Polyscias murrayi. Measurements were carried out on two fully-expanded leaflets in the upper crown of each plant, on summer mornings in late January, at temperatures of 25–26 °C. Amax and Rd of P. murrayi were estimated at 13.33 ± SE 1.05, and −1.30 ± SE 0.02 μmoles m−2 s−1, respectively. As mitochondrial respiration is strongly inhibited by light (Villar, Held & Merino 1995; Atkin, Evans & Siebke 1998), the daytime respiration rate was estimated as being 40% of the measured Rd (Pearcy et al. 2004). Gas exchange data from the three plants are given in Appendix S1 (Supporting Information).

Modelling gap light environments

We developed canopy files representing light environments beneath the centre and poleward margins of idealized canopy gaps of two sizes, at three different latitudes. Poleward gap margins were included in our analysis because, depending on gap size, at low sun angles they can receive more direct beam radiation than the forest floor beneath the centre of gaps (Canham et al. 1990). A gap size of 100 m2 was chosen because it corresponds roughly to average tree-fall gap sizes reported in several studies of rain forests (Denslow 1987; Ogden et al. 1991). A larger gap size of 300 m2 was also used to approximate the more open light environments arising as a result of multiple tree-falls. Latitudes of 17°, 29° and 42° were chosen to represent tropical, subtropical and temperate environments. Although rain forests are patchily distributed along the length of eastern Australia, these three latitudes are representative of the three major rain forest masses: tropical northern Queensland, subtropical northern New South Wales, and cool-temperate western Tasmania. Simulations were run at eight dates spread evenly throughout the year, including both solstices and both equinoxes.

A digital canopy file used with YPLANT divides the hemisphere into eight azimuth classes and 20 altitude classes (Pearcy & Yang 1996). The canopy file describes the distribution of canopy openness over these 160 sectors, and the time series of direct PPFD interception (sunflecks) throughout any chosen day of the year. Our starting point was a composite digital understorey light environment developed by averaging data from six hemispherical photos taken beneath an intact rain forest canopy (cf. Falster & Westoby 2003a), yielding an average canopy openness of ≈1.4%. This light environment was then modified to incorporate the effect of idealized canopy gaps. Trigonometric functions were used to determine which of the 160 sectors would be included in circular canopy gaps of 100 m2 and 300 m2 in a 25-m-tall forest canopy. For example, from the forest floor directly below the centre of a circular gap of 300 m2 (=radius 9.77 m) in a 25-m-tall canopy, the angular elevation of the gap horizon in all eight azimuth classes can be obtained as:


Beneath the southern margin of a 300-m2 gap in the same 25-m-tall canopy, in contrast, the gap horizon due north lies at 52.0º elevation, the north-western and north-eastern horizons occur at 61.1º, and no significant fraction of the gap occurs in any of the other five azimuth classes.

Modelling the asymmetric light environments that occur beneath gap margins requires knowledge of variation in solar azimuth throughout the day. Solar azimuths were calculated using VBA functions developed by Greg Pelletier, Department of Ecology, State of Washington (, and an azimuth function incorporated into the spreadsheet used to generate canopy files. Examples of canopy files are shown in Appendix S4 (Supporting Information).

Statistical analysis

Nested anova was used to test for trait differences between biomes, and across functional types within each biome. We used repeated measures anova to determine the effects of latitude and gap position on simulated light interception and carbon gain. The structure of our study, using the same virtual plants for simulations at three different latitudes, is analogous to an experiment involving repeated measurements on replicate individuals successively exposed to several different treatments, or sampled repeatedly at different dates. Analyses were carried out using spss Statistics 17.0 (SPSS Inc., Chicago, IL, USA).


Latitudinal variation in relationship of leaf life span with tree height

We found only a weak relationship between tree height and leaf life span in tropical and subtropical rain forests (R2 = 0.025), but these two variables were significantly positively correlated in temperate rain forests (Fig. 2). At a given height, tropical and subtropical species had shorter average leaf life spans than temperate species, but also a wider range of leaf life spans. Evergreen trees reaching >20 m height with leaf life spans ≤12 months were abundant in tropical and subtropical forests, but the dataset included no such taxa from sites poleward of latitude 35°. The same pattern was found when mean annual temperature was used to separate sites into two classes, above and below 17 °C (not shown).

Figure 2.

 Relationships of leaf life span with typical potential heights of evergreen trees and shrubs in tropical/subtropical (n = 182) versus temperate (n = 71) rain forests. All leaf life span data are from sun leaves, i.e. in forest canopies, or from plants growing on forest margins or beneath canopy gaps. Note the wider range of leaf life spans and the non-significant relationship between leaf life span and height in tropical/subtropical forests, and the absence of species >20 m tall with leaf life spans ≤12 months in the temperate zone. Leaf life span data were obtained from Wright et al. (2004), Poorter & Bongers (2006), Richardson et al. (in press), C.H. Lusk (unpublished data). Sources of height data appear in Appendix S2.

Leaf, stem and seed traits across biomes and functional types

Gas exchange parameters and wood density differed appreciably across functional types within each biome, but relatively little between biomes (Table 2; Fig. 3). Conversely, conduit diameters varied much more between biomes than across functional types within each biome. Seed size was not significantly affected by either biome or functional type (Table 2). In the topics and subtropics, short and tall pioneers had similar seed, gas exchange and wood traits, and both these groups were clearly differentiated from their more shade-tolerant associates by having higher rates of gas exchange and lower wood densities (Table 2; Fig. 3).

Table 2.   Summary of nested anova testing for seed, leaf and stem trait differences between biomes (temperate versus tropical/subtropical rain forests) and across functional types within biomes (short pioneer, tall pioneer, non-pioneer). All data were log-transformed before analysis, to meet the assumption of additivity of effects (Quinn & Keough 2002)
Seed mass
 Functional type (Biome)2.0800.106
 Functional type (Biome)24.343<0.0001
 Functional type (Biome)5.8820.001
Wood density
 Functional type (Biome)5.4710.001
Conduit diameter
 Functional type (Biome)0.5940.621
Figure 3.

 Seed, leaf and wood traits of tree functional types in temperate and tropical/subtropical rain forests. Graphs include all species which could be objectively classified into functional types using leaf life span and height data (Fig. 2), showing 10th, 25th, 50th, 75th and 90th percentiles, with black dots indicating outliers. Gas exchange measurement temperatures were mostly 25 °C, but ranged from 22 to 29° C. Principal data sources were Wright et al. (2004), Poorter & Bongers (2006), Royal Botanic Gardens Kew (2008), Chave et al. (2009), Zanne et al. (2009), Poorter et al. (2010). Other sources are listed in Appendix S3. Seed wet masses were converted to dry mass using the relationship log10 dry mass = 0.92 * log10 fresh mass0.94 (Moles et al. 2005).

Effects of latitude and gap position on light interception and potential carbon gain

The three replicate plants of Polyscias murrayi differed only minimally in simulated light interception and potential carbon gain (Figs 4 and 5). As a result, even quite small effects were statistically significant (Table 3).

Figure 4.

 Effect of latitude on daily light interception and potential carbon gain by foliage of juvenile Polyscias murrayi beneath an idealized 100 m2 tree-fall gap in a 25-m-tall forest canopy, modelled using YPLANT. Mean daily fluxes of three replicate plants (±SE) were estimated by averaging results obtained from eight different dates evenly-spaced throughout the year, including solstices and equinoxes. White and grey bars respectively show results obtained beneath the centre and poleward (sunny) margins of gaps. Left and centre panels show results of simulations under clear and overcast skies, respectively; right panels show weighted averages based on sunshine hours (Table 1).

Figure 5.

 Effect of latitude on daily light interception and potential carbon gain by foliage of juvenile Polyscias murrayi beneath an idealized 300 m2 tree-fall gap in a 25-m-tall canopy, modelled using YPLANT. Mean daily fluxes of three replicate plants (±SE) were estimated by averaging results obtained from eight different dates evenly-spaced throughout the year, including solstices and equinoxes. White and grey bars respectively show results obtained beneath the centre and poleward (sunny) margins of gaps. Left and centre panels show results of simulations under clear and overcast skies, respectively; right panels show weighted averages based on sunshine hours (Table 1).

Table 3.   Summary of repeated measures anova testing effects of latitude and position within gap on light interception and carbon gain by juvenile Polyscias murrayi. PAR data were log-transformed before analysis
 PAR interceptionNet carbon gain
Small gap, clear sky
 Gap position7.20.0551.50.289
 Latitude × position2604.9<0.00012491.3<0.0001
Small gap, overcast
 Gap position44.20.00356.00.002
 Latitude × position3779.8<0.0001601.2<0.0001
Large gap, clear sky
 Gap position29.10.00621.50.010
 Latitude × position971.5<0.00011254.2<0.0001
Large gap, overcast
 Gap position233.3<0.000139.60.003
 Latitude × position41.90.00390.70.001

Under overcast conditions, simulated light interception and net C gain by P. murrayi beneath gaps of both sizes were dominated by latitudinal effects (Table 3). Interception of photosynthetically active radiation (PAR) and C gain under overcast conditions were invariably higher (or less negative) beneath gap centre than gap margins (Figs 4 and 5). Potential carbon gain of P. murrayi foliage under overcast conditions was negative at all three latitudes beneath small gaps (Fig. 4), but only at latitude 42 beneath large gaps (Fig. 5).

Under clear skies, although latitude was also generally the dominant influence on simulated PPFD interception and C gain (Table 3), the precise effect of latitude was strongly dependent on position within gaps. Potential C beneath the centre of gaps of both sizes declined monotonically from latitude 17 to latitude 42. Beneath gap margins, in contrast, plants were predicted to do best at latitude 29 (Figs 4 and 5).

Potential C gain on clear days was less closely related to total PAR interception (Figs 4 and 5) than to the mean daily duration of direct light (Fig. 6). This reflects the saturating response of photosynthesis to light, with high-intensity sunflecks that contribute massively to daily PPFD not making a commensurate contribution to daily C gain. For example, a doubling of PPFD from 600 to 1200 μmoles m−2 s−1 produces only a modest increase in net assimilation rate, from 11.5 to 12.7 μmoles m−2 s−1.

Figure 6.

 Effect of latitude on mean daily duration of direct sunlight beneath the centre and poleward margin of idealized 100 m2 and 300 m2 tree-fall gaps. Bars show averages from eight dates evenly-spaced throughout the year.

Using sunshine hours to weight results of simulations on clear and overcast days provided an overall picture of the influences of latitude and gap size on mean daily PAR interception and potential carbon gain (Figs 4 and 5). Substantial carbon gain was predicted beneath the centre of small gaps at latitude 17, and beneath their margins at latitude 29; however, net carbon deficit was predicted beneath both the centre and margin of small gaps at latitude 42 (Fig. 4). Large gaps, in contrast, were predicted to permit abundant carbon gain everywhere except beneath gap centres at latitude 42 (Fig. 5).


Our review of tree height and leaf life span data shows notable trait differences between low- and mid-latitude rain forests, and confirms the existence of a trait combination confined to the former (Fig. 2). Temperate rain forests appear to lack evergreen trees >20 m tall with fast foliage turnover and rapid growth that recruit directly to the canopy after tree-falls in old-growth stands.

Our simulation results are consistent with the hypothesis that the success of pioneers in canopy gaps depends on the high solar elevations characteristic of low latitudes (Figs 4 and 5). Abundant direct beam penetration into tree-fall gaps at low latitudes makes it feasible for species with high metabolic rates to survive and compete with the advanced growth of shade-tolerant species. On the other hand, limited light availability in tree-fall gaps at temperate latitudes means that gap-phase dynamics there are dominated by release of advanced growth of shade-tolerant species (e.g. Ogden et al. 1991; Lusk & Smith 1998). If our hypothesis is correct, we would expect to see a latitudinal trend in the minimum gap size required for establishment of Polyscias murrayi and other pioneers. P. murrayi would appear to be an excellent species for testing this prediction, as its distribution spans about 20° of latitude in Australia, and it has also been reported from New Guinea (Floyd 2008). However, at present we are unaware of quantitative data suitable for carrying out such a test.

The heights attained by many tropical and subtropical pioneers (Fig. 2) suggests that light-demanding species gain important benefits from reaching the ‘lumicline’ (Parker 1997) associated with the main canopy of old-growth stands. This can be understood in terms of trade-offs between current reproductive effort and future reproductive value (Pianka 1976). Even individuals recruiting beneath tree-fall gaps will encounter higher light levels upon reaching the main canopy (Parker 1997), so sustained allocation to vertical stem growth and deferral of reproduction until canopy space has been secured is likely to result in much larger seed crops than those achievable by shorter pioneers that fail to reach the canopy (Falster & Westoby 2003b). Short pioneers in tree-fall gaps also risk being rapidly overtopped and killed by taller pioneers, or by advanced growth of shade-tolerant canopy species. We are not aware of data suitable for testing the expectation of a positive correlation between age at first reproduction and potential heights of rain forest pioneers, but data in Moles et al. (2004) do confirm this pattern for tree and shrub species in general. Pioneer trees are unlikely to be exposed to this type of selection on height growth in the temperate zone, as our simulations suggest that species with high metabolic rates will simply be unable to survive beneath tree-fall gaps in mid-latitude rain forests (Figs 4 and 5). This might explain the latitudinal distribution of tall pioneers (Fig. 2). On the other hand, a low potential height and shorter life expectancy may be associated with cheaper stem construction, and hence faster height growth (Falster & Westoby 2005b), which could help pioneer trees and shrubs compete with herbaceous plants in the open, early-successional environments that they are largely confined to at temperate latitudes. Although our data review found that tall and short (sub)tropical pioneers had similar wood densities (Fig. 3), a study of an Australian rain forest assemblage found a positive correlation between wood density and potential heights of 19 light-demanding tree species (Falster & Westoby 2005a), consistent with the expectation of lower costs of height growth in short pioneers. As a result, short and tall (sub)tropical pioneers may differ in maximum height growth rates, despite having similar gas exchange traits (Fig. 3).

We simulated light interception and carbon gain in only one pioneer species, potentially raising questions about the representativeness of our chosen species. Gap-dependence is primarily a function of rates of metabolism and tissue turnover (Lusk 2002; Poorter & Bongers 2006; Baltzer & Thomas 2007), and we found that the dark respiration rate of P. murrayi was only slightly higher than the median value of measurements carried out on 38 species identified as tall pioneers (Fig. 3), suggesting that the metabolic rate of our chosen species – and hence, its degree of dependence on direct beam radiation – is representative of this functional type. Although differences in foliage display efficiency also influence light interception and carbon gain (Pearcy et al. 2004), a study of 24 species in a tropical rain forest understorey showed that differences in light absorbed by individual plants were determined more by the light available at each particular site than by architectural differences between individuals or species (Valladares, Skillman & Pearcy 2002). Similarly, Poorter & Werger (1999) reported that differences in crown architecture between juvenile trees growing in different light environments were more significant than those observed when species were compared in a common light environment.

Does our assumption of invariant respiration rates at different latitudes lead to underestimation of net carbon gain in cool temperate environments? Although, in the short-term, respiration is a function of temperature, acclimation and/or natural selection tend to minimize observed differences between respiration rates of plants growing at different latitudes (Lambers, Chapin & Pons 1998). What little data is available suggests that tropical and temperate rain forest species have similar respiration rates when measured at their respective ambient temperatures during the growing season (e.g. Lusk 2002; Pearcy et al. 2004). Latitudinal differences in temperature are most evident during winter, and Miyazawa & Kikuzawa (2005) reported that winter respiration rates of six Japanese temperate evergreens were significantly lower than rates measured during the growing season. However, wintertime depression of Amax also appears to be common in mid-latitude regions (Awada et al. 2003; Miyazawa & Kikuzawa 2005), suggesting that both photosynthetic income and respiratory costs of evergreens are likely lower in winter. It is therefore unclear to what extent latitudinal variation in temperature could alter our conclusions about potential carbon gain.

Although our simulation results reflect the high metabolic rate of P. murrayi, the real life requirement of pioneers for high-light environments is also a function of their short leaf life spans. Differential adaptation of evergreen trees to light environments is associated with a well-known trade-off between performance and persistence of leaves (Walters & Reich 1999; Poorter & Bongers 2006; Lusk et al. 2008). Leaves with little structural reinforcement and high nitrogen content drive rapid carbon gain and growth in high light, but their short life spans (as well as their high respiration rates) are a liability in the shade, where high rates of carbon gain are not possible. A more complete model might thus incorporate the effect of growth and tissue turnover on long-term plant carbon balance (e.g. Walters & Reich 1999, Table 5), as well as respiration of stems and roots and night-time respiration by foliage.

As we modelled only net daytime carbon gain by crowns of juvenile Polyscias murrayi, without considering other components of plant carbon balance, the value of our simulations lies primarily in revealing latitudinal trends, rather than absolute rates of potential carbon gain beneath tree-fall gaps. However, predicted carbon deficits of P. murrayi crowns in certain environments (Figs 4 and 5) do strongly suggest the non-viability of the species there. Potential carbon gain of juvenile P. murrayi declined with increasing latitude beneath the centre of canopy gaps, and peaked at latitude 29 beneath poleward gap margins (Figs 3 and 4). At latitude 42, simulations showed foliage of P. murrayi to be in carbon deficit beneath openings similar in size to typical single tree-fall gaps (100 m2), consistent with the absence of gap-regenerating tall pioneers from the cool temperate rain forests of south-eastern Australia. Although positive carbon gain was possible beneath the poleward margin of an idealized 300 m2 gap, the rarity of gaps of this size (Ogden et al. 1991; Hubbell et al. 1999) might make it difficult for pioneers to maintain a presence in old-growth forests. Furthermore, gap margins are not necessarily adequate environments for regeneration of pioneer species as, without further tree-falls, light levels there are likely to decline rapidly because of encroachment by overstorey trees bordering the gap (Kneeshaw & Bergeron 1998). Weishampel & Urban (1996) developed a spatially-explicit forest gap model incorporating solar angles, which predicted that basal area of light-demanding species should decrease with increasing latitude. We note, however, that temperate rain forests include some relatively light-demanding emergent species whose great longevity enables them to persist into late-successional stages despite regenerating poorly in old-growth stands. These are often conifers, some such as Pseudotsuga menziesii in the Pacific Northwest and Podocarpus totara and Agathis australis in New Zealand attaining ages of >1000 years (Franklin & Hemstrom 1981; Ogden & Stewart 1995).

Our results also highlight the importance of geographic variation in cloud cover as a potential control on the distributions of plant species. Low sunshine hours at latitude 42 added critically to the effects of solar elevation angles on light interception and carbon gain (Figs 4 and 5). Potential net carbon gain under clear skies was positive beneath the centre and (especially) margins of large (300 m2) gaps at latitude 42, but negative under the overcast conditions that prevail at this latitude (Table 1). Low sunshine hours in western Tasmania are far from being an idiosyncrasy of Australian geography, as mid-latitude humid forests in New Zealand, South America and the Pacific Northwest also grow in cloudy climates (London, Warren & Hahn 1989), with few sites receiving >6 sunshine hours a day. Cloud cover is also frequent in equatorial regions, but lowland tropical sites generally receive more sunshine hours than the middle latitudes of both hemispheres (London, Warren & Hahn 1989).

Although we can find no equivalent of tall pioneers in evergreen-dominated temperate rain forests, some fast-growing light-demanding trees appear to maintain an overstorey presence in old-growth deciduous temperate forests via gap-phase regeneration. Liriodendron tulipifera L. (Magnoliaceae) is the best-known example, persisting in forests older than its maximum life span of c. 300 years (Busing 1995). However, this species is believed to require somewhat larger gaps (>400 m2) than those examined here (Runkle 1985), in keeping with the expectation that the minimum gap size required for establishment of light-demanding trees will increase with latitude. What little age data is available suggests that most tropical and subtropical rain forest pioneers are shorter-lived than L. tulipifera (Lieberman et al. 1985; Alvarez-Buylla & Martinez-Ramos 1992).

Other environmental factors besides sun angles and cloud cover could also potentially restrict the distribution of fast-growing pioneers. For example, the fast growth of tropical pioneers might depend on the possession of highly-conductive xylem (Poorter et al. 2010) that is not viable in frosty climates. On moist sites, xylem anatomy and physiology are believed to be shaped by a trade-off between performance and cold tolerance: large conduits conduct water very efficiently (conductivity ∝r4) but are highly susceptible to cavitation after freeze-thaw cycles (Sperry & Sullivan 1992; Feild & Brodribb 2001). Our data review showed that tropical and subtropical rain forest trees in general had much larger conduits than their temperate counterparts, but there was only weak evidence that conduit diameters of (sub) tropical pioneers differed from those of their slower-growing shade-tolerant tropical associates (Table 2; Fig. 3). At this stage it is therefore unclear to what extent climatic constraints on viable conduit diameters limit the distribution of fast-growing pioneers, although observations on trees planted poleward of their natural limits might be informative.

As pioneers can constitute a sizeable fraction of tree species richness in old-growth tropical rain forest, their geographic distribution contributes significantly to global patterns of tree species richness. Of the 174 tree species recorded in tree-fall gaps on Barro Colorado Island, 47 (27%) were classified as ‘pioneers’ by Hubbell et al. (1999), and other research shows that most light-demanding trees found in the old-growth forest at that site are tall enough to reach the canopy (Bohlman & O’Brien 2006). In a moist tropical forest in Bolivia, 25 out of 54 tree species studied by Poorter, Bongers & Bongers (2006) were identified as ‘pioneers’, most of which attained heights > 20 m, although it is not clear how well this selection of tree species represents the makeup of the local old-growth assemblage. Our own data (Kooyman & Lusk, unpublished data) suggest that pioneers would constitute a smaller proportion of tropical and subtropical rain forest assemblages in Australia than they appear to at the neotropical sites studied by Hubbell et al. (1999) and Poorter, Bongers & Bongers (2006).

In summary, we have demonstrated that an important tree functional type in low-latitude rain forests is virtually absent from their mid-latitude counterparts. Furthermore, we have shown that the effect of latitude on light availability beneath tree-fall gaps could largely explain this pattern. If this mechanism is correct, rising global temperatures are unlikely to have much effect on the latitudinal distribution of tall pioneers in rain forests. Although latitudinal effects on light availability in gaps have been recognized for some time (Canham et al. 1990; Prentice & Leemans 1990; Weishampel & Urban 1996), this is the first study to provide quantitative support for the hypothesis that latitudinal variation in the range of light environments in forests has a hand in shaping gradients of species richness, by influencing niche availability (Ricklefs 1977).


We thank Tanja Lenz for carrying out simulations and data analysis and for preparing Fig. 1, Bob Pearcy for providing YPLANT 3.1 and very generous advice on its workings, Daniel Falster for developing the original canopy file template that we modified for our procedures, Greg Pelletier for kindly making available his VBA functions for calculating solar azimuth, Colin Prentice for helpful comment and the ARC Discovery scheme for funding this research. Comments by two anonymous reviewers helped improve the manuscript.