Plant Ecology & Diversity, 2016 Vol. 9, Nos. 5–6, 603–614, http://dx.doi.org/10.1080/17550874.2016.1266403

Allometric projections of time-related growth trajectories of two coexisting dipterocarp canopy species in India Cécile Antin

a,b

*, Jimmy Le Beca,b,c, Narayanan Ayyappanb, Bramasamdura Rangana Rameshb and Raphaël Pélissiera,b

a IRD, UMR AMAP, Montpellier, France; bInstitut Français de Pondichéry, UMIFRE CNRS-MAE 21, Puducherry, India; cINRA, UMR SYSTEM, Montpellier, France

(Received 2 February 2016; accepted 6 November 2016) Background: The Western Ghats of India contain the westernmost dipterocarp forests of Asia. However, only a few dipterocarp tree species actually coexist in the forest canopy among which Vateria indica and Dipterocarpus indicus are the most common. The mechanisms contributing to the coexistence of these phylogenetically closely related species have not been identified. Aims: We investigated the time-related growth trajectories in diameter, height and crown size of these two species in the Uppangala Permanent Sample Plot to determine if trade-offs in their three-dimensional developmental strategies could contribute to their long-term coexistence. Methods: From annual diameter growth data of 692 trees >9.55 cm in diameter at breast height over a 21-year period, we developed time-related diameter growth models for the two species, accounting for local density-dependent competition effects and topography. Combining the diameter growth models with static stem and crown allometries, we projected timerelated tree growth trajectories in height and crown size. Results: While both species can reach similar dimensions, V. indica grows much faster, or at least as fast as D. indicus in diameter, height and crown size in all the observed situations. Both species respond similarly to topography, but V. indica appears to be more responsive to local density-dependent competition than D. indicus. Finally, V. indica shows higher mortality and recruitment rates and a greater basal area increase than D. indicus. Conclusions: These results refute our hypothesis that D. indicus coexists with the outperformer V. indica by a growth strategy allowing selected individuals in favourable conditions to reach the canopy more quickly than their competitors. The current coexistence of the two dipterocarp species at Uppangala appears not to be at a static equilibrium; V. indica probably being in a phase of canopy stand colonisation. Keywords: density-dependent competition; Dipterocarpus indicus; stem and crown dimensions; terrain slope intensity; terrain aspect; Vateria indica

Introduction Dipterocarps are an emblematic group of phylogenetically closely related tree canopy-forming species that coexist in sympatry in Asian tropical forests. Several mechanisms of dipterocarp coexistence have been evidenced from niche partitioning with respect to light gaps (e.g. Philipson et al. 2012) or soil conditions (e.g. Sukri et al. 2012), including relationships with ectomycorrhizal fungi (Peay et al. 2010) or to temporal phenological variations and masting (e.g. Brearley et al. 2007; Kettle et al. 2011). Trade-offs in resource allocation between rapid growth in light and survival in the shade have also been investigated among dipterocarp species but mainly at the seedling stage (e.g. Dent and Burslem 2009). However, these trade-offs, though possibly variable during ontogeny (Clark and Clark 1992), are likely to persist at the adult stage (Reich 2014) and thus to contribute to resource partitioning between coexisting canopy species. Trees develop in a three-dimensional (3-D) space, and the relationship between growth and its allocation among different dimensions (e.g. stem diameter, tree height and crown size) is strongly governed by allometric rules (King *Corresponding author. Email: [email protected] © 2016 Botanical Society of Scotland and Taylor & Francis

2005). This is why foresters have long-developed approaches to characterise tree growth strategies based on allometries, especially in temperate forests (e.g. Pretzsch 2006). Allometries vary among species, reflecting the way species allocate carbon to growth. For instance, a greater allocation to height growth can allow canopy species to rapidly reach a high-light environment, thus overtopping and shading their neighbours (Poorter et al. 2005), while a greater crown expansion can allow understory species to increase their interception surface in low-light environments (Kohyama 1987). Allometries can also vary with soil conditions (Heineman et al. 2011) or with local competition intensity, leading, for instance, to height growth being enhanced in crowded conditions over lateral crown expansion (Banin et al. 2012). Steep slopes also favour asymmetric crown development (e.g. Lang et al. 2010), which may result in additional biomechanical constraints on the tree stem. This may induce feedback mechanisms requiring greater resource allocation to radial growth and thus resulting in bulkier stems (Fournier et al. 2006). The comparison among the species of allometric relationships between tree above-ground dimensions is commonly used to explore the inter-specific differences in

604

C. Antin et al.

terms of growth strategies (King and Clark 2011), including plasticity of species growth responses with respect to environmental variations (Lang et al. 2010; Harja et al. 2012). However, it is often overlooked that static allometries are size-related relationships that do not properly account for the temporal differences in exploration and occupancy dynamics of the 3-D canopy space by coexisting species. Understanding these dynamics requires integrating allometric relationships with tree growth rates in both the radial (stem diameter growth) and axial (extension growth) dimensions, the latter determining both total tree height and crown width. For obvious practical reasons, studies on adult tree growth generally focus on stem diameter growth. In tropical trees, this has been demonstrated to vary between species as an adaptation to resource acquisition, according to life history strategies (fast-growing pioneers vs. slowgrowing shade tolerants; Valladares and Niinemets 2008) or adult stature (e.g. maximum height; Hérault et al. 2011). However, stem diameter growth also varies with individual responses to the amount of resources available (plasticity), either in the soil (water and nutrients, e.g. Russo et al. 2005; Coomes et al. 2009) or above ground (crown light exposure, e.g. Moravie et al. 1999). It may also vary during ontogeny (Le Bec et al. 2015), a phenomenon that is reinforced when large trees reach the canopy and access greater light levels. While both extension growth and stem diameter growth respond to soil fertility, extension growth probably responds to competition for light more directly (Poorter et al. 2005). As a consequence, projection of time-related growth trajectories in various tree dimensions could provide new insights into the mechanisms of competition and resource partitioning among coexisting species involved in the maintenance of canopy tree diversity in tropical forests (Heineman et al. 2011). In this article, we investigated the growth trajectories of two coexisting dipterocarp canopy species, Dipterocarpus indicus and Vateria indica, in order to determine whether trade-offs in their growth strategies could contribute to their long-term coexistence in a wet evergreen forest of the Western Ghats of India. Indeed, though we have no evidence for significant differences in their regeneration niches, both species are common large canopy trees. Previous studies also showed that the stem diameter distribution of V. indica was much closer to the expectation of a demographic equilibrium (i.e. to a negative exponential function suggesting relatively constant diameter growth and mortality rates with diameter; Muller-Landau et al. 2006) than D. indicus. While V. indica had a stem diameter growth rate twice as fast as D. indicus, the latter reached the canopy with more slender stems and had a narrower crown at the adult stage than V. indica (Pascal and Pélissier 1996; Moravie et al. 1997; Antin et al. 2013; Le Bec et al. 2015). We therefore expected, from population structure and static allometries, an opportunistic growth behaviour of D. indicus, with individuals able to grow faster in height in favourable conditions, becoming emergents. In contrast, we expected that V. indica invested in a large crown to optimise light capture in the crowded canopy layer, with potential

biomechanical feedbacks to stem growth allocation. Such a hypothesis, which could contribute to explaining why the two species can coexist in the forest canopy, is only testable, however, if the growth trajectories of the two species in all dimensions can be synchronised in time. As assessing tree age of tropical tree species remains difficult (Nath et al. 2012), projections of diameter growth from repeated measurements is the most reliable way to obtain time-related growth trajectories to make synchronised species comparisons. Even though diameter growth is easily determined from field measurements, measuring extension growth is more challenging, especially in crowded forest conditions, making direct time-related assessment of height and crown extensions difficult. We thus combined species-specific dynamic growth models calibrated on a 21-year-long series of diameter measurements with static stem and crown allometries to project time-related height and crown size trajectories. Our models also accounted for the main factors of growth variation at the study site, i.e. tree size, local competition and topography. We finally discuss the above hypothesis of a stable coexistence of V. indica and D. indicus based on the time-related growth trajectories owing to their 3-D growth strategies.

Material and methods Study site and species The study site was the Uppangala Permanent Sample Plot (UPSP) in the Western Ghats of India at 12° 32ʹ 15ʹ’ N, 75° 39ʹ 46 E (Pascal and Pélissier 1996; Pélissier et al. 2011). It was located at an altitude of 300–600 m a.s.l. and presented a marked relief dissected by streams running downward from the crest of the Ghats: ridges with gentle slopes separating watercourses alternate with deep valleys, resulting in a major contrast between easterly and westerly hillside aspects. Soil analyses in a nearby plot showed that topography was the main local factor of environmental variation and that it was a very good predictor of soil water availability at Uppangala: while soils on the ridges (gentle slopes) are several meters deep, soils on the (steep) slopes are superficial and eroded (Ferry 1992). Local climate presents a strong seasonality with 90% of total rainfall (ca. 5000 mm year−1) occurring during the monsoon period (June–October) and a mean daily temperature of ca. 25–30 °C all year round. The stand belongs to an undisturbed, old-growth wet evergreen forest in which eight sample plots (five bands and three rectangular plots overlapping the bands, see Figure S1) covering a total of 5.07 ha were established between 1990 and 1993 (Pélissier et al. 2011). In these plots, all the trees with girth at breast height exceeding 30 cm (i.e. a diameter at breast height of 9.55 cm) were mapped within 10 m × 10 m subplots, identified and equipped with permanent dendrometer bands. All subplots were georeferenced, and a digital elevation model (DEM) was derived from slope measurements (angle and direction) taken from each corner of the 10 m × 10 m subplots. The stand, composed of 101

Allometric projections of time-related growth tree species, presents an average density of ca. 650 trees ha−1 and a basal area of ca. 40 m2 ha−1, respectively. Mortality and recruitment have also been regularly recorded, and their rates have been calculated from 1993 (because of the progressive addition of new plots between 1990 and 1993) to 2011. The two dipterocarp species studied, D. indicus and V. indica (species nomenclature refers to vouchers stored at the herbarium of the French Institute of Pondicherry, HIFP), are among the most abundant and dominant species in the canopy at UPSP, accounting for 20.1% of the stand density and 40.9% of the total basal area, respectively (Pascal and Pélissier 1996). Both are ectomycorrhizal canopy or emergent tree species (Riviere et al. 2006) that exhibit irregular phenological patterns and episodic mass fruiting (Elouard et al. 1997). There is no evidence for any significant difference between these two species with regard to their preferences in terms of gap-light environments for regeneration (C. Antin, unpublished data). Tree measurements In the present article, we used a dataset of 122 D. indicus and 570 V. indica individuals measured in UPSP from 1990 to 2011. Girth at breast height was recorded yearly using permanent dendrometer bands and converted into diameter at breast height (dbh, in cm), providing more than 17,000 dbh measurements for the two species (see Figure S2). Tree height (ht, in m) and crown base height (cbht, in m) were measured for random subsamples of these trees in 1990–1994 and 2007–2008. Total tree height was measured for 585 trees in 1990–1994 using a BlumeLeiss altimeter and for 348 trees in 2007–2008 using a Haglöf vertex clinometer. Crown base height was measured with the same instrumentation for 482 trees in 1990–1994 and for 338 trees in 2007–2008. Crown radii were measured in the four cardinal directions from ground projections of the crown edge to the stem centre (375 trees in 1990–1994) and in four directions given by the largest width of the crown projection and the perpendicular width (343 trees in 2007–2008). Crown width (cw, in m) was then calculated for both periods as twice the mean crown radius. Only a small fraction of these trees were measured twice for height and crown size (see Table S1 for more details about the data set). We additionally attributed a local slope angle (in %) and aspect (in degree) derived from the DEM to each tree.

605

very low annual dbh increment observed in UPSP. Finally, we separated basal area fluctuations due to tree growth and due to recruitment and mortality (see Table 1). Diameter growth models We fitted diameter growth using species-specific maximum likelihood mixed models. Following the general design proposed in Le Bec et al. (2015), we introduced in the model: (i) an individual tree random effect in order to account for temporal autocorrelation of the growth data over the 21-year period of observation (Kohyama et al. 2005), (ii) tree dbh and a local competition index (as well as their log transformations) as time-varying covariates and (iii) slope and aspect (as well as their sine and cosine transformations) as time-invariant covariates. Variable selection was made with a backward step-wise procedure based on Akaike information criterion, which led to the two following growth models: V: indica ! Δdbhit ¼ a#dbhit þ b#logðdbhit Þ þ c#logðg15it Þ

þ d#logðdbhit Þ#logðg15it Þ þ e#logðg15it Þ#dbhit

þ f #sineðaspecti Þ þ γi þ εit

D: indicus ! Δdbhit ¼ a # logðdbhit Þ þ b # logðg15it Þ þ γi þ εit

(1)

(2)

Stand demography

In these equations ∆dbhit is the annual diameter increment calculated for each individual tree i, as the difference between its diameter at t and t–1 divided by ∆t in days to account for the slightly variable census intervals (from 305 to 426 days). The term dbhit is the diameter of tree i at time t. The term g15it is an index of local competition corresponding to the total stand basal area of neighbours within a 15-m radius around tree i at time t, an index that was found to better explain tree growth than others at the study site (see Le Bec et al. 2015). The term aspecti is the local aspect around tree i, with a sine transformation that emphasises east–west oppositions in slope orientation. The term γi is the individual random effect (on the intercept) for tree i, which represents how much the growth trajectory of that tree deviates consistently over time from its species growth response. Finally, εit is the residual growth for tree i at time t (assumed independent and identically distributed).

Tree mortality and recruitment rates were calculated as the total number of dead or recruited trees (between 1993 and 2011) divided by the initial abundance of the given species for V. indica, D. indicus and all the other species grouped together. Basal area was calculated from dbh. In case of missing dbh measurements in 1993 or 2011, we used dbh measured in 1994 or 2010, considering the bias introduced as negligible given the

Allometric relationships To determine if topographical variation observed in UPSP accounted for variation in allometric relationships, we first carried out an analysis of variance for assessing the effects of slope, aspect and their interaction on ht:dbh, cbht:dbh and cw:dbh ratios. The slope factor was as in Antin et al.

606

C. Antin et al.

Table 1. Demographic parameters over the period 1993–2011 for V. indica, D. indicus and for all other species grouped together in UPSP, Western Ghats of India.

Species All species V. indica D. indicus Other species West gentle (0.72 ha) V. indica D. indicus Other species West steep (1.06 ha) V. indica D. indicus Other species East gentle (1.52 ha) V. indica D. indicus Other species East steep (1.10 ha) V. indica D. indicus Other species

Initial density (trees ha−1)

Mortality rate (% year−1)

647.6 115.4 22.5 509.7 720.5 138 29.3 553.2 725.7 128.3 17.9 579.4 656.8 144.5 26.9 485.4 618 64.5 27.3 526.2

0.88 0.56 0.43 0.98 0.80 0.62 0.80 0.84 0.91 0.66 0.29 0.98 0.61 0.33 0.13 0.72 1.25 0.85 0.36 1.34

Recruitment Initial basal Basal area increase rate (% area (m2 related to growth year−1) ha−1) (%) 0.86 0.62 0.58 0.92 0.86 0.56 1.06 0.92 0.56 0.49 0.29 0.58 0.76 0.43 0.40 0.97 1.25 1.47 0.72 1.24

41.8 12 6.4 23.4 52.6 15.9 11.9 25.3 45.8 14.8 5.3 25.6 42.5 9.6 9 23.9 37.1 9.7 4.5 22.9

21.6 30.7 14.8 18.8 17.1 22.2 10.9 16.7 20.3 26.5 11.7 18.4 22.2 44.9 13.1 16.4 26.7 30.3 29.7 24.5

Basal area increase related to recruitment and mortality (%)

Basal area increase (%)

−13.8 −9.3 −6.1 −18.2 −12.8 −8.2 −15.6 −14.5 −12.6 −7.7 0 −18.1 −11 −9.6 −0.6 −15.5 −21 −17.6 −12.6 −24.1

7.8 21.4 8.7 0.5 4.3 14 −4.7 2.2 7.6 18.7 11.7 0.3 11.2 35.4 12.6 0.9 5.7 12.7 17.1 0.4

Variables were calculated according to slope (steep vs. gentle) and aspect (east vs. west) categories. Mortality and recruitment rates were assessed for the whole period of observation (17 years). As new plots were progressively added between 1990 and 1993, we calculated demography from 1993 to 2011.

(2013), i.e. simplified in two classes (steep slopes ≥ 30° and gentle slopes < 30°), while aspect was east (< 180° from north) and west (≥ 180° from north). Note that given the 15-year interval between tree height and crown size data that have been measured twice for some trees, we considered them as independent observations for allometries. Only the slope factor had a significant effect for the three ratios considered for D. indicus, whereas both slope and aspect were significant for V. indica (Table S2). We thus chose to fit separate allometric relationships for trees on gentle and steep slopes for D. indicus, and for trees on gentle or steep and east- or west-facing slopes for V. indica. As both species clearly present an asymptotic height, allometric relationships between height and dbh were determined by a Weibull-type function (Temesgen and Gadow 2004): h ¼ hmax # ð1 ! eða#dbh^bÞ Þ

(3)

where h is alternatively tree height (ht) or crown base height (cbht). The parameters hmax, a and b were estimated with a non-linear least square regression (Bates and Watts 1988). The residuals of these models met the assumption of normality, whereas the use of a power function linearised by log transformation would have led to an overestimation of the predicted variable for smallest and largest trees (Figure S3). Allometric relationship between crown width (cw) and dbh was determined by a log-linear function: logðcwÞ ¼ a þ b # logðdbhÞ

(4)

The parameters a and b were estimated with a linear least square regression. The residuals of these models met the assumption of normality (not shown). Projection of time-related growth trajectories We then used the diameter growth model and allometric relationships to integrate growth trajectories over time for both species and accounting for the various topographic situations. We started integration from dbht=0 = 10 cm and calculated dbh predictions at a yearly time step t with: dbhtþ1 ¼ dbht þΔdbhðdbht Þ

(5)

where ∆dbh(dbht) is estimated from parameters fitted in Equations 1 and 2, for V. indica and D. indicus, respectively. We projected time-related trajectories of both species with a competition index (g15) fixed to 40 m2 ha−1 (mean observed value of g15i at UPSP) and considered two situations for V. indica with aspect alternatively fixed to 90° (easterly aspect) or 270° (westerly aspect). We stopped trajectory projections at 120 cm dbh corresponding to the maximum observed diameter in UPSP for both species. We finally estimated ht, cbht and cw at the same yearly time steps, by using the parameters estimated while fitting allometric relationships (Equations 3 and 4). b

ht ¼ hmax # ð1 ! eða#dbht^ ÞÞ

(6)

cwt ¼ a þ b # logðdbht Þ

(7)

Allometric projections of time-related growth Finally, a 3-D representation of the development of a hypothetical tree combining the predicted dbh, stem height and crown dimensions was made by using the Allostand model (Barbier et al. 2012) for both dipterocarp species in various topographic situations. All the above analyses were carried in the R environment, v. 3.0.2 (R Development Core Team 2011). Results Stand and population demography While the stand density and basal area varied with respect to topography, V. indica and D. indicus populations were quite homogeneously distributed except in steep east-facing slopes, where V. indica was almost half as dense than in other topographies (Table 1). Overall the observed population dynamics were faster for V. indica (0.56% and 0.62% year−1 in terms of mortality and recruitment) than for D. indicus (0.43% and 0.58% year−1) except on gentle westfacing slopes where it was the reverse (Table 1). Finally, total stand basal area increased by more than 7.8% between 1993 and 2011 and was proportionally greater for V. indica (21.4%) than for D. indicus (8.7%) or for the other species (0.5%) (Table 1). The greatest increment in basal area between 1993 and 2011 was observed on gentle east-facing slopes for V. indica (35.4%) and on steep east-facing slopes for D. indicus (17.1%), i.e. where V. indica is less dense and shows the smallest increment in basal area (12.7%). The only negative increment in basal area was recorded for D. indicus in gentle west-facing slopes (−4.7%), i.e. where its turnover was the fastest (Table 1). These figures suggest that the growth and demography of the two dipterocarp species vary in a non-consistent manner according to slope and aspect at UPSP. Furthermore, both species showed contrasted dbh distributions: V. indica showed a density function decreasing exponentially with tree size (and thus relatively close to the expectation of a demographic equilibrium), whereas D. indicus showed an atypical density function with a greater proportion of large trees (Figure S4). Moreover, the relative frequency of small tree size classes decreased over time for V. indica, whereas the relative frequency of large tree size classes decreased for D. indicus. Variation in growth and allometries The diameter growth models fitted to the 21-year series of dbh increments showed that V. indica grew faster than D. indicus. Mean tree growth increased with tree dbh, with larger variation for V. indica, and decreased with competition, more notably for small trees of V. indica; D. indicus was less responsive to increased competition (Figure 1 and Table S3). In terms of allometries, D. indicus and V. indica showed contrasting responses to environmental factors: ht–dbh allometry showed an asymptotic height significantly higher for D. indicus than for V. indica and an asymptotic height significantly higher on gentle slopes than on steep slopes for both species (Figure S5 and

607

Table S4). As a consequence, large trees of D. indicus were more slender at a given diameter than those of V. indica. The cbht–dbh allometry showed, for a given dbh, a higher cbht for D. indicus than for V. indica and a higher cbht on gentle slopes than on steep slopes, except for V. indica on westerly aspects for which no difference was observed between gentle and steep slopes (Figure S5 and Table S4). The cw–dbh allometry showed that large trees had wider crowns on steep slopes and that, for a given dbh, crowns of V. indica were wider than crowns of D. indicus, except on steep east-facing slopes where V. indica had the narrowest crowns (Figure S5 and Table S4). Time-related growth trajectories Projection of time-related trajectories showed a general trend towards more rapid growth for V. indica than D. indicus whatever the topographic situation. At t = 200 years (>10 cm dbh), D. indicus reached a dbh of 37.9 cm, whereas V. indica reached a dbh of 50.5 cm on westerly aspects and 87.7 cm on easterly aspects (Figure 2). V. indica reached the maximum observed dbh (120 cm) at t = 249 years on easterly aspects, at t = 328 years on westerly aspects; whereas D. indicus reached the same maximum dbh at t = 505 years. Both species showed a continuous increase in growth rate over time (see Figure 2). V. indica also showed a more rapid and early height growth on easterly aspects than on westerly aspects, and than D. indicus, regardless of the slope (Figure 3). However, D. indicus always reached a higher maximum height than V. indica (see also Figure S5). Overall, both species showed a shift of growth allocation in the stem from height growth to stem diameter growth during ontogeny (Figure 4), with a predicted decrease in the slenderness ratio, slower for D. indicus than for V. indica. The predicted decrease in the slenderness ratio was also slower on gentle slopes than on steep slopes for both species and slower on westerly aspects than on easterly aspects for V. indica. It is noticeable that V. indica reached a low slenderness ratio much earlier on gentle east-facing slopes than in other topographic situations or than D. indicus in all topographic situations. Growth in crown width (cw) was also more rapid for V. indica than for D. indicus (Figure 4). At t = 200 years and on gentle slopes, D. indicus reached a cw of ca. 7 m, and V. indica reached ca. 10 m on westerly aspects and ca. 16 m on easterly aspects, whereas on steep slopes, D. indicus reached a cw of ca. 8 m and V. indica reached ca. 9 m on westerly aspects and ca. 14 m on easterly aspects. Both species showed a decreasing crown depth– crown width ratio with time, indicating a shift from crowns deeper than wide to crowns wider than deep. But while D. indicus showed a lower initial ratio than V. indica (at t = 0, i.e. dbh = 10 cm), the shift occurred later than for V. indica (Figure 4). We note that D. indicus maintained an equilibrium between crown dimensions from t = 200 to t = 400 years on gentle

608

C. Antin et al.

Figure 1. Predicted dbh growth according to initial dbh with g15 (local basal area within a 15 m radius around tree) fixed at 40 m2 ha−1 (left insert) and predicted dbh growth according to the competition index g15 with dbh alternatively fixed to 10, 50 and 80 cm (right insert) for D. indicus and V. indica in UPSP, Western Ghats, India. All other covariates are invariant and have been fixed to their average values.

Figure 2. Time-related growth trajectories in dbh (left) and dbh growth (right) as predicted by species-specific growth models for D. indicus and V. indica according to slope (steep vs. gentle) and aspect (east vs. west) categories in UPSP, Western Ghats, India.

slopes, whereas the shift from vertical extension to horizontal extension occurred faster on steep slopes. Projections of the time-related growth trajectories of V. indica and D. indicus in stem diameter, tree height and crown dimensions in different topographic situations are summarised in Figure 5. Discussion Whereas the comparison of static allometric relationships between these species has already been investigated (e.g. Pascal and Pélissier 1996; Antin et al. 2013), here we advance by introducing the temporal dimension within the projected time-related growth trajectories. Our results bring new insights into the way the growth trajectories vary between the two species and according to terrain aspect and slope intensity. However, these differences do not yet indicate trade-offs in the growth strategies of the two species that could explain their long-term stable coexistence in the forest canopy as we expected. We further discuss an alternative speculative hypothesis of a transient

coexistence after an undocumented large-scale disturbance event. V. indica is more responsive to light environment than D. indicus We showed that, unlike D. indicus, V. indica has a greater stem diameter growth on easterly exposed hillsides (receiving sunlight in the morning) than westerly exposed hillsides. It may be due to some combination of less cloud cover in the morning and/or inhibition of photosynthesis by greater water stress in the afternoon. Moravie et al. (1999) showed that diameter growth of V. indica was mainly driven by the light environment at UPSP, as has been shown for other species in tropical forests (King et al. 2005). On the other hand, V. indica exhibits a greater crown volume on east-facing slopes than on westfacing ones, which may also explain its greater growth rates by increasing the total amount of light interception surface (Kohyama 1987). We also found that the growth response to competition was more than twice as strong for V. indica as compared to

Allometric projections of time-related growth

609

Figure 3. Time-related growth trajectories in tree height (top) and tree height growth (bottom) as predicted from species-specific diameter growth models and height–diameter allometries for D. indicus and V. indica according to slope (steep vs. gentle) and aspect (east vs. west) categories in UPSP, Western Ghats, India.

that for D. indicus. The competition index selected in the diameter growth model considers all the neighbouring trees in a radius of 15 m around each tree and so partly includes the effects of above-ground and below-ground competition. These results suggest that V. indica is more sensitive to the light environment and other growing conditions than D. indicus. In other words, V. indica appears more plastic than D. indicus in terms of stem diameter growth. Nevertheless, D. indicus exhibits a basal area increase similar to that in V. indica (29.7% and 30.3%, respectively) on steep east-facing slopes, where initial density and basal area were the lowest (618 trees ha−1 and 37.1 m2 ha−1, respectively). This basal area increase is more than twice than the increase for D. indicus in other topographic situations where initial density and basal area were greater (657–726 trees ha−1 and 42.5– 52.6 m2 ha−1, respectively). This suggests that D. indicus may be sensitive to competition in these conditions, but we were not able to detect this with the growth model we fitted. Both species respond similarly to topography For both species, we showed that projected height growth is faster on gentle slopes. In contrast, no effect of slope was detected on stem diameter growth. As a result, we observed that total heights are taller for a given stem diameter on gentle slopes, so that trees are slenderer there than elsewhere. Robert and Moravie (2003) proposed that this reduced slenderness on

steep slopes in UPSP could be the result of increased stratification of tree crowns on the slope thus reducing competition for light and consequently reducing investment in height growth as compared to gentle slopes. We observed that the slenderness ratio decreased more rapidly on steep slopes for both species. This shows that the allocation pattern between height growth and stem diameter growth is affected by local environmental conditions for both species (Sumida et al. 1997; Rozendaal et al. 2015). The shift from prevailing allocation to height growth towards prevailing allocation to stem diameter growth not only depends on age or size (Silveira et al. 2012) but is also influenced by environmental conditions. Crown width growth and crown base increase are also faster on gentle slopes than on steep slopes for both species. While D. indicus shows a lower initial crown depth:crown width ratio than V. indica (at t = 0, i.e. dbh = 10 cm) in all topographic situations, we noticed that V. indica reached a lower ratio much earlier on gentle east-facing slopes than in other topographic situations and lower than D. indicus in all topographic situations. This suggests that V. indica established earlier and at a lower height in the canopy than D. indicus. V. indica grows faster than D. indicus V. indica exhibits a faster growth rate than D. indicus for most of the tree dimensions and in most of the situations observed at UPSP. The faster stem diameter growth of V.

610

C. Antin et al.

Figure 4. Time-related trajectories of slenderness ratio (h:dbh), crown width and crown shape (crown depth:crown width) as predicted from species-specific diameter growth models and height-stem diameter allometries for D. indicus and V. indica according to slope (steep vs. gentle) and aspect (east vs. west) categories in UPSP, Western Ghats, India.

indica could be partly attributed to its less dense wood than that of D. indicus (0.48 and 0.59 g cm−3, respectively; see Bhat et al. 1990; Devagiri et al. 2013), as it has been showed that stem growth is negatively correlated with wood density (King et al. 2005). As both species show very close scaling relationships between crown width and stem diameter – although V. indica tends to have wider crowns than D. indicus – V. indica exhibits a faster growth in crown width. More surprisingly, while D. indicus has a slenderer stem than V. indica, the latter also exhibits a faster height growth than D. indicus on easterly aspects, and both species show a similar height growth on westerly aspects. While we observed a general good concordance between extension growth patterns inferred from time-related growth trajectories on the one hand and from direct estimations on the other hand, this is not the case for crown width trajectories of V. indica and height growth trajectories for small trees in the range 15–80 cm dbh for D. indicus (see

Figure S6). This emphasises a potential limit to our method of time-related reconstruction, which cannot account for possible changes in allocation patterns that could have occurred before 1990 because of different conditions and forest dynamics. In particular, D. indicus individuals as old as 300–500 years according to our estimations may not represent the actual conditions of competition that partly determine the resource allocation patterns between stem, height and crown extension. While stand density remained almost constant over the study period (−0.02% trees ha−1 year−1), stand basal area increased by 7.8% (up to 11.2% on gentle east-facing slopes) indicating a progressive canopy closure and therefore a more competitive environment. This kind of change might both slow down the individual crown width extension of V. indica (by lack of space) and promote a faster height growth for some individuals of D. indicus (by selection). V. indica seems to do better than D. indicus at establishing into the canopy, and this result might contribute to

Allometric projections of time-related growth

611

hypothesised originally. It seems instead that V. indica simply grows faster than D. indicus in all dimensions and in most topographic situations. This is the main reason for the more than doubling in basal area of V. indica over the 21 years as compared to that of D. indicus. Such an increase in basal area shows that the current dynamics have not reached equilibrium, at least not a static equilibrium. Le Bec (2014) showed, from simulations, that the current dynamics observed in UPSP could be the result of an undocumented large past disturbance (ca. 50% mortality, over 100 years ago). Our present study tends to confirm this hypothesis and that the stand could currently being colonised by V. indica. We also observed that D. indicus had a greater increase in basal area than the mean increase for all other species, suggesting that it is not likely to be excluded from the stand by V. indica in the near future. Limitations of the study

Figure 5. Representation of the 3-D development of a hypothetical tree combining the predicted dbh, stem height and crown dimensions at 10, 50, 100, 150 and 250 years (above 10 cm dbh) for V. indica and, additionally, at 350 and 500 years for D. indicus in various topographic situations in UPSP, Western Ghats, India. dbh is represented by a doubling size in order to emphasise the visual differences between species and topographic situations (slope and aspect).

explaining its high density (17.8% of the stand density) and basal area (28.7 % of the stand basal area) in the stand. Moravie et al. (1997) developed a canopy regeneration model for UPSP, which showed that, in the absence of major disturbances, a dominant tree can be replaced when it dies by suppressed trees that survived in its crown periphery. With such a mechanism, V. indica would outperform D. indicus by rapidly overshadowing trees of the same size, as its growth in height and crown width is faster. We also observed that V. indica showed the greatest increase in density and basal area on gentle east-facing slopes, where the mortality rate is the lowest, indicating low disturbance. A hypothesis of a temporal shift in species dominance of V. indica and D. indicus in the forest canopy Comparing two canopy and emergent dipterocarp species, we found no evidence of existing trade-offs between height growth and stem diameter growth as we

We are aware that our time-related projections of growth trajectories based on repeated stem diameter growth data may not exactly represent the growth trajectory of an average individual (Zuidema et al. 2009). The high temporal autocorrelation of tree growth (Kohyama et al. 2005) and the increased frailty of slow growing individuals (Rüger et al. 2011) mean that large trees are likely to be represented mostly by fast growing individuals, while juvenile trees exhibit a more heterogeneous composition – a ‘selection’ effect, well known in population demography (Cam et al. 2002). However, we believe that our approach offers a relevant tool to compare how growth is distributed among diameter classes and to compare the growth patterns of the two dipterocarp species. As a matter of fact, when we compared the growth patterns in height and crown size inferred from combining diameter growth models and static allometries with direct estimations for the few trees that have been measured twice in height and crown size (once in 1990–1994 and again in 2007–2008), we found a good concordance between the two approaches except for a slight overestimation of the predicted crown width increment for V. indica and a slight underestimation of the predicted height growth in the range 15–80 cm dbh for D. indicus (see Figure S6). These slight discordances do not invalidate our general findings and conclusions. An alternative to improve the prediction of growth trajectories would be to rely on long-term growth series through direct dendrochronological studies. However, comparison between direct and indirect methods for deriving growth trajectories for tropical species in the Western Ghats of India demonstrated that both methods are only consistent when annual growth rings can be assessed properly (Nath et al. 2012). As many trees of evergreen tropical forests do not form distinct annual growth rings, indirect estimations remain for the time being the most reliable approach.

612

C. Antin et al.

Conclusion In this study, we proposed an original approach for investigating the contribution of trade-offs in growth strategies to the long-term coexistence of D. indicus and V. indica in a wet evergreen forest of the Western Ghats of India. As we assumed that comparing static allometries was not sufficient for investigating the role of growth strategies in species coexistence, we projected time-related growth trajectories in stem diameter, stem height and crown size of these two species by integrating allometric relationships with tree growth models. Our results highlighted the importance of synchronising in time the growth trajectories of the two species in all dimensions in order to compare their growth strategies. Whereas the comparison of static allometric relationships between these species leads to the hypothesis that D. indicus coexists with the faster growing V. indica by a growth strategy allowing selected individuals in favourable conditions to reach the canopy more quickly than their competitors (Pascal and Pélissier 1996; Moravie et al. 1997; Antin et al. 2013), we conclude from our results that there is no trade-off in the growth strategies of the two species that could explain their long-term stable coexistence. While D. indicus and V. indica show contrasting responses to environmental factors in terms of allometries, V. indica grows much faster, or at least as fast as D. indicus in diameter, height and crown size in all the observed situations. Combined with the observation that the growth and demography of the two dipterocarp species vary in a non-consistent manner according to slope and aspect at UPSP, we offer an alternative hypothesis of a transient coexistence after a possible undocumented largescale disturbance event, V. indica probably being in a phase of canopy stand colonisation. A stand dynamic simulation model could allow deeper investigation of the dynamic coexistence of both dipterocarp populations in Uppangala. Acknowledgements UPSP is a joint research station of the Karnataka Forest Department, Bangalore, and the French Institute of Pondicherry. We are very grateful to the many field workers, technicians, engineers and researchers who contributed to its long-term monitoring. We thank Francis Brearley and three anonymous reviewers for their thoughtful and constructive comments and Philippe Verley for assistance in realisation of Figure 5.

Funding This work was supported by the French Academy of Agriculture through a Dufrenoy grant to CA; French Academy of Agriculture [Grant Dufrenoy 2007].

Disclosure statement No potential conflict of interest was reported by the authors.

Supplemental data Supplemental data for this article can be accessed here.

ORCID Cécile Antin

http://orcid.org/0000-0002-0385-5886

Notes on contributors Cécile Antin is a researcher in forest and agroforest ecology. Her interests include forest dynamics and rules governing biomass allocation in trees in undisturbed forests as well as in agroforests. Jimmy Le Bec is a researcher in forest and agroforest ecology. His interests include forest and agroforest dynamics, tree demography and individual tree variability. Narayanan Ayyappan is a researcher in tropical forest ecology in charge of the management of the Uppangala Permanent Sample Plot. His interests include plant systematics, diversity, structure and dynamics of disturbed and undisturbed forests. Bramasamdura Rangana Ramesh is a researcher in tropical forest ecology. His interests include phytogeography, tropical forest tree dynamics and forest landscape ecology. Raphaël Pélissier is a senior researcher in tropical forest ecology, with a special interest in the spatial organisation of tree species diversity, and in 3-D forest stand structure and dynamics.

References Antin C, Pélissier R, Vincent G, Couteron P. 2013. Crown allometries are less responsive than stem allometry to tree size and habitat variations in an Indian monsoon forest. Trees. 27:1485–1495. Banin L, Feldpausch TR, Phillips OL, Baker TR, Lloyd J, Affum-Baffoe K, Arets EJMM, Berry NJ, Bradford M, Brienen RJW, et al. 2012. What controls tropical forest architecture? Testing environmental, structural and floristic drivers. Global Ecology and Biogeography. 21:1179–1190. Barbier N, Couteron P, Gastellu-Etchegorry J-P, Proisy C. 2012. Linking canopy images to forest structural parameters: potential of a modeling framework. Annals of Forest Science. 69:305–311. Bates DM, Watts DG. 1988. Nonlinear regression: iterative estimation and linear approximations. In: Nonlinear regression analysis and its applications. Hoboken (NJ): John Wiley & Sons, Inc. p. 32–66. Bhat KM, Bhat KV, Dhamodaran TK. 1990. Wood specific gravity in stem and branches of eleven timbers from Kerala. Indian Forester. 116:541–546. Brearley FQ, Proctor J, Suriantata NL, Dalrymple G, Voysey BC. 2007. Reproductive phenology over a 10-year period in a lowland evergreen rain forest of central Borneo. Journal of Ecology. 95:828–839. Cam E, Link WA, Cooch EG, Monnat J-Y, Danchin E. 2002. Individual covariation in life-history traits: seeing the trees despite the forest. The American Naturalist. 159:96–105. Clark DA, Clark DB. 1992. Life history diversity of canopy and emergent trees in a neotropical rain forest. Ecological Monographs. 62:315–344. Coomes DA, Kunstler G, Canham CD, Wright E. 2009. A greater range of shade-tolerance niches in nutrient-rich forests: an explanation for positive richness-productivity relationships? Journal of Ecology. 97:705–717.

Allometric projections of time-related growth Dent DH, Burslem DFRP. 2009. Performance trade-offs driven by morphological plasticity contribute to habitat specialization of Bornean tree species. Biotropica. 41:424–434. Devagiri GM, Money S, Singh S, Dadbhawal VK, Patil P, Khaple A, Devakumar AS, Hubbali S. 2013. Assessment of above ground biomass and carbon pool in different vegetation types of south western part of Karnataka, India using spectral modeling. Tropical Ecology. 54:149–165. Elouard C, Houllier F, Pascal J-P, Pélissier R, Ramesh BR. 1997. Dynamics of the dense moist evergreen forests – long term monitoring of an experimental station in Kodagu District (Karnataka, India). India: Institut Français de Pondichéry. Ferry B 1992. Etude des humus forestiers de la région des Ghâts occidentaux (Inde du Sud) facteurs climatiques, édaphiques et biologiques intervenant dans le stockage de la matière organique du sol [PhD dissertation]. [France]: Université de Nancy. Fournier M, Stokes A, Coutand C, Fourcaud T, Moulia B. 2006. Tree biomechanics and growth strategies in the context of forest functional ecology. In: Ecology and biomechanics: a mechanical approach to the ecology of animals and plants. Boca Raton (FL): Taylor and Francis, CRC Press. p. 1–33. Harja D, Vincent G, Mulia R, Van Noordwijk M. 2012. Tree shape plasticity in relation to crown exposure. Trees. 26:1275–1285. Heineman KD, Jensen E, Shapland A, Bogenrief B, Tan S, Rebarber R, Russo SE. 2011. The effects of belowground resources on aboveground allometric growth in Bornean tree species. Forest Ecology and Management. 261:1820–1832. Hérault B, Bachelot B, Poorter L, Rossi V, Bongers F, Chave J, Paine CET, Wagner F, Baraloto C. 2011. Functional traits shape ontogenetic growth trajectories of rain forest tree species. Journal of Ecology. 99:1431–1440. Kettle CJ, Maycock CR, Ghazoul J, Hollingsworth PM, Khoo E, Sukri RSH, Burslem DFRP. 2011. Ecological implications of a flower size/number trade-off in tropical forest trees. Plos ONE. 6:e16111. King DA. 2005. Linking tree form, allocation and growth with an allometrically explicit model. Ecological Modelling. 185:77– 91. King DA, Clark DA. 2011. Allometry of emergent tree species from saplings to above-canopy adults in a Costa Rican rain forest. Journal of Tropical Ecology. 27:573–579. King DA, Davies SJ, Supardi MNN, Tan S. 2005. Tree growth is related to light interception and wood density in two mixed dipterocarp forests of Malaysia. Functional Ecology. 19:445– 453. Kohyama T. 1987. Significance of architecture and allometry in saplings. Functional Ecology. 1:399–404. Kohyama T, Kubo T, Macklin E. 2005. Effect of temporal autocorrelation on apparent growth rate variation in forest tree census data and an alternative distribution function of tree growth rate. Ecological Research. 20:11–15. Lang AC, Härdtle W, Bruelheide H, Geißler C, Nadrowski K, Schuldt A, Yu M, Von Oheimb G. 2010. Tree morphology responds to neighbourhood competition and slope in speciesrich forests of subtropical China. Forest Ecology and Management. 260:1708–1715. Le Bec J 2014. Rôle de la variabilité intraspécifque des processus démographiques en forêts hétérogènes pour caractériser les stratégies écologiques de coexistence des espèces [PhD dissertation]. [Montpellier]: AgroParisTech. Available online at: https://hal.archives-ouvertes.fr/tel-01116564/document. Le Bec J, Courbaud B, Le Moguédec G, Pélissier R. 2015. Characterizing tropical tree species growth strategies: learning from inter-individual variability and scale invariance. Plos ONE. 10:e0117028. Moravie M-A, Durand M, Houllier F. 1999. Ecological meaning and predictive ability of social status, vigour and competition

613

indices in a tropical rain forest (India). Forest Ecology and Management. 117:221–240. Moravie M-A, Pascal J-P, Auger P. 1997. Investigating canopy regeneration processes through individual-based spatial models: application to a tropical rain forest. Ecological Modelling. 104:241–260. Muller-Landau HC, Condit RS, Chave J, Thomas SC, Bohlman SA, Bunyavejchewin S, Davies S, Foster R, Gunatilleke S, Gunatilleke N, et al. 2006. Testing metabolic ecology theory for allometric scaling of tree size, growth and mortality in tropical forests. Ecology Letters. 9:575–588. Nath CD, Boura A, De Franceschi D, Pélissier R. 2012. Assessing the utility of direct and indirect methods for estimating tropical tree age in the Western Ghats, India. Trees. 26:1017–1029. Pascal J-P, Pélissier R. 1996. Structure and floristic composition of a tropical evergreen forest in south-west India. Journal of Tropical Ecology. 12:191–214. Peay KG, Kennedy PG, Davies SJ, Tan S, Bruns TD. 2010. Potential link between plant and fungal distributions in a dipterocarp rainforest: community and phylogenetic structure of tropical ectomycorrhizal fungi across a plant and soil ecotone. New Phytologist. 185:529–542. Pélissier R, Pascal J-P, Ayyappan N, Ramesh BR, Aravajy S, Ramalingam SR. 2011. Tree demography in an undisturbed Dipterocarp permanent sample plot at Uppangala, Western Ghats of India. Ecology. 92:1376. Philipson CD, Saner P, Marthews TR, Nilus R, Reynolds G, Turnbull LA, Hector A. 2012. Light-based regeneration niches: evidence from 21 dipterocarp species using sizespecific RGRs. Biotropica. 44:627–636. Poorter L, Bongers F, Sterck FJ, Wöll H. 2005. Beyond the regeneration phase: differentiation of height–light trajectories among tropical tree species. Journal of Ecology. 93:256–267. Pretzsch H. 2006. Species-specific allometric scaling under selfthinning: evidence from long-term plots in forest stands. Oecologia. 146:572–583. R Development Core Team. 2011. A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Reich PB. 2014. The world-wide “fast–slow” plant economics spectrum: a traits manifesto. Journal of Ecology. 102:275–301. Riviere T, Natarajan K, Dreyfus B. 2006. Spatial distribution of ectomycorrhizal Basidiomycete Russula subsect. Foetentinae populations in a primary dipterocarp rainforest. Mycorrhiza. 16:143–148. Robert A, Moravie M-A. 2003. Topographic variation and stand heterogeneity in a wet evergreen forest of India. Journal of Tropical Ecology. 19:697–707. Rozendaal DMA, During HJ, Sterck FJ, Asscheman D, Wiegeraad J, Zuidema PA. 2015. Long-term growth patterns of juvenile trees from a Bolivian tropical moist forest: shifting investments in diameter growth and height growth. Journal of Tropical Ecology. 31:519–529. Rüger N, Huth A, Hubbell SP, Condit R. 2011. Determinants of mortality across a tropical lowland rainforest community. Oikos. 120:1047–1056. Russo SE, Davies SJ, King DA, Tan S. 2005. Soil-related performance variation and distributions of tree species in a Bornean rain forest. Journal of Ecology. 93:879–889. Silveira AP, Martins FR, Araújo FS. 2012. Are tree ontogenetic structure and allometric relationship independent of vegetation formation type? A case study with Cordia oncocalyx in the Brazilian caatinga. Acta Oecologica. 43:126–133. Sukri RS, Wahab RA, Salim KA, Burslem DFRP. 2012. Habitat associations and community structure of dipterocarps in response to environment and soil conditions in Brunei Darussalam, Northwest Borneo. Biotropica. 44:595–605.

614

C. Antin et al.

Sumida A, Ito H, Isagi Y. 1997. Trade-off between height growth and stem diameter growth for an evergreen Oak, Quercus glauca, in a mixed hardwood forest. Functional Ecology. 11:300–309. Temesgen H, Gadow K. 2004. Generalized height–diameter models—an application for major tree species in complex stands of interior British Columbia. European Journal of Forest Research. 123:45–51.

Valladares F, Niinemets Ü. 2008. Shade tolerance, a key plant feature of complex nature and consequences. Annual Review of Ecology, Evolution, and Systematics. 39:237–257. Zuidema PA, Brienen RJW, During HJ, Güneralp B. 2009. Do persistently fast‐growing juveniles contribute disproportionately to population growth? A new analysis tool for matrix models and its application to rainforest trees. The American Naturalist. 174:709–719.