Effects of light availability and growth rate on leaf lifespan of four temperate rainforest Proteaceae

Although comparative studies have revealed much about the environmental correlates of leaf lifespan and its integration with other leaf traits, a comprehensive theory of leaf lifespan is still lacking. The influence of growth rate on self-shading may be a key proximate determinant of both interand intra-specific variation in leaf lifespans. If this were the case, we would expect leaf lifespans of fast-growing light-demanding species to respond more strongly to light environment than those of shade-tolerant species. We monitored growth and leaf survival of juvenile trees of four temperate rainforest Proteaceae in southern Chile, in order to explore the influences of light environment and self-shading on leaf lifespan. Leaf lifespans tended to decrease with increasing diffuse light availability, and slopes of these relationships were steeper in two light-demanding species (Embothrium coccineum, Lomatia hirsuta) than in two more shade-tolerant species (Lomatia ferruginea, Gevuina avellana). This pattern mirrored interspecific variation in relationships of height growth with light availability, height growth rates of the two light-demanding species responding more strongly to light availability than did growth of L. ferruginea and G. avellana. Path analysis suggested that light availability influenced leaf lifespans primarily through the influence of growth on self-shading: when rate of leaf production was held constant by multiple regression, light availability per se had no significant influence on leaf lifespans of any of the four species. However, 29 to 79 % of intraspecific variation in leaf lifespan remained unexplained by light environment and leaf production rate. If self-shading is fact the main proximate control on leaf lifespan, information on the elevational distribution of photosynthetic photon flux may enhance the explanatory power of studies of this nature.


INTRODUCTION
Despite a vast amount of empirical work on leaf demography in recent decades (e.g., Reich et al. 1997, Reich et al. 2004, Wright et al. 2004), a comprehensive theory of leaf lifespan is still lacking.Some environmental correlates of leaf lifespan have been clearly established (e.g., Aerts 1995, Poorter & Bongers 2006), and interspecific work shows associations of leaf lifespan with other leaf traits such as photosynthetic capacity and leaf mass per area; these cross-species correlations are very strong in global datasets spanning more than two orders of magnitude in leaf lifespan (Wright et al. 2004).Yet these trait associations break down, and in some cases are even reversed, when intraspecific comparisons are made across a range of light environments (Lusk et al. 2008a).
Prolongation of leaf lifespan appears to be associated with both adaptation and acclimation to shade.Several studies have reported that leaves live longer on shaded plants than on conspecifics growing in well-lit environments (e.g., Lusk 2002, Reich et al. 2004, Vincent 2006).Studies in diverse evergreen forests have shown that inherently slow-growing shade-tolerant species retain their leaves longer than light-demanding associates (e.g., Reich et al. 1995, Lusk 2002, Poorter & Bongers 2006).However, this parallel in evolutionary and plastic responses of leaf lifespan to shade is associated with divergent trends in leaf mass per area: whereas shade-tolerant evergreens have larger mass per area (LMA) than their lightdemanding associates, plastic responses to shade invariably involve a reduction in LMA (Lusk et al. 2008a).Despite their lower LMA, shade leaves tend to have a larger structural fraction than sun leaves, which may enable long lifespan by reducing their attractiveness to herbivores (Lusk et al. 2010).
What may bridge inter-and intraspecific patterns is the influence of self-shading on leaf lifespan (Ackerly & Bazzaz 1995).As leaves age their carbon balance deteriorates because of increased shading by newer leaves, as well as declining physiological capacity (Reich et al. 2009).Although "self-shading" is a convenient term, shading by growth of neighbouring plants may be an equally important influence on leaf lifespans (Vincent 2006).Plants might be expected to shed leaves when their net carbon gain declines to zero; although a modelling study by Reich et al. (2009) found that almost all leaves had positive daytime carbon balances at the age of their mean life-span, their calculations did not include night-time leaf respiration, or the costs of supplying leaves with water.Zero net carbon balance seems likely to be reached sooner in fast-growing plants than in slowgrowing plants, partly because of more rapid self-shading in the former (Ackerly & Bazzaz 1995), and partly because of the linkage between plant growth rates and respiration rates, which determine the minimum light levels required for positive net carbon gain (Walters & Reich 2000, Lusk 2002).As slow metabolism and slow growth are characteristic of both adaptation and short-term responses to shade, such an impact of self-shading is likely relevant to both inter-and intra-specific variation in leaf lifespan.
If rates of self-shading are the proximate determinant of leaf lifespans, we would expect leaf lifespans of light-demanding species to respond more strongly to light environment than those of shade-tolerant species.Photosynthesis and growth of the former are usually highly responsive to light availability, whereas shade-tolerant taxa often show modest photosynthetic plasticity, and growth of their seedlings often saturates at moderate light intensities (e.g., Read 1985, Poorter 1999, Valladares et al. 2000, Lusk 2002).We addressed this question by measuring juvenile height growth, leaf production rates and leaf lifespans of four Chilean rainforest Proteaceae across a wide range of light environments.The four tree species differ widely in reported light requirements.

Study site
The study was carried out at Parque Katalapi (latitude 41º31' S, longitude 72º45' W, elevation 60 masl) near Cordillera de Quillaipe in southern Chile.Mean annual precipitation is estimated at 2322 mm, with a marked summer minimum, and nearly all of this falls as rain.Mean annual air temperature is 11 ºC.The coldest months are June to August, with mean daily minimum and maximum temperatures of about 2 and 10 ºC, respectively.February is the warmest month, with mean daily minimum and maximum temperatures of 8 and 23 ºC, respectively.The site is a 30 ha mosaic of old fields, remnants of the previous forest cover, and patches of young regenerating forest.Cattle have been excluded from the study area for 15 years.Common tree species on the site included Nothofagus nitida, Laureliopsis philippiana, Luma apiculata, Amomyrtus luma, A. meli, Weinmannia trichosperma, Caldcluvia paniculata, Eucryphia cordifolia and Drimys winteri, as well as the four Proteaceae that were the focus of this study.

Study species
The four Proteaceae represent a wide range of reported light requirements.Embothrium coccineum and Lomatia hirsuta are frequent colonisers of disturbed habitats such as old fields and road cuttings.E. coccineum was described as "very intolerant" of shade by Donoso (1989).Its seedlings were found to have the highest understorey mortality rates of eight rainforest species studied by Lusk ( 2002), and Figueroa & Lusk (2001) reported that juveniles of E. coccineum were associated with brighter light environments than most other Chilean rainforest evergreens.Although Lomatia hirsuta was also rated as "very intolerant" by Donoso (1989), its presence in old-growth forests, especially those with an overstorey of deciduous Nothofagus obliqua (e.g., Veblen et al. 1979), suggests that it tolerates slightly more shade than E. coccineum.This is consistent with data in Figueroa & Lusk (2001), showing that L. hirsuta seedlings on Isla Grande de Chiloé occupied shadier microsites than those of E. coccineum.Gevuina avellana and L. ferruginea have been described as being of intermediate shade tolerance (Donoso 1989), although juveniles of the former can often be found growing in as little as 2 % light, in the shaded understories of old-growth stands (Lusk et al. 2008b).Nomenclature follows Marticorena & Quezada (1985)

Sampling methods
Sampling was carried out on a series of transects run across the park, including a wide range of light environments, including shaded understories and open pasture.At intervals randomly spaced between 20 and 30 m apart, the nearest juvenile (500-2200 mm tall) of each species was chosen for sampling.This system worked satisfactorily for three of the study species, but the fourth (L.hirsuta) was not common enough to be sampled in this manner.Accordingly, we selected all juvenile L. hirsuta within the requisite size range that could be located.
The height of the principal axis of each juvenile was measured to the apex, and all fully-expanded leaves on this axis were counted.12 months later, each plant was revisited, its height remeasured, and survival of leaves recorded.Abscission scars were counted to determine mortality of new leaves initiated after the start of the study period; this was particularly important for E. coccineum, as most leaves of this species appeared to live < 1 yr.Leaf longevity (years) was estimated as: where n i = initial number of leaves, n f = final number surviving from n i , and m n = mortality of new leaves initiated since the first census.
The light environment of each juvenile tree was quantified using a pair of LAI-2000 canopy analysers (Li-Cor, Lincoln, NE).One instrument was used to take measurements at each sampling point, while the other, placed in an open field, was programmed to take readings at 30-s intervals.Integration of data from the two instruments enabled estimation of percentage diffuse irradiance at each sampling point within the forest, equivalent to percentage of canopy openness over the quasi-hemispherical (148º) field of view perceived by the LAI-2000 sensors.Measurements were made on overcast days, using the full 148º field of view.Measurements with the LAI-2000 are a good surrogate of spatial variation in mean daily photosynthetic photon flux density within a stand (Machado & Reich 1999).

Statistical analysis
One of our interests was in the slope of relationships of leaf lifespan and growth with light availability.Major axis tests are appropriate when the primary interest is in slopes and intercepts of relationships, rather than their strength (Warton et al. 2006).We therefore used SMATR (Falster et al. 2006) to calculate the slope and intercept of the major axis of these relationships, instead of ordinary least squares regression.All variables were log 10 -transformed prior to analysis, in order to meet the assumption of additivity of effects.
As leaf lifespans are likely to be influenced by production of new leaves, rather than by height growth per se, we also examined relationships of new leaf production with light availability.This approach is supported by the findings of Ackerly & Bazzaz (1995), who showed that the light environments of individual leaves deteriorate steadily as they are progressively shaded by production of new leaves.As the relationship of leaf production rate with light availability appeared to differ in shape across species, we explored various least squares fits to determine which best described the relationship for each species.Finally, as light availability could potentially influence leaf lifespan through mechanisms other than selfshading, we used path analysis to explore dependence of leaf lifespan on both light availability and new leaf production.This was achieved by multiple regression of leaf lifespan on light availability and leaf production rate, and simple regression of leaf production rate on light availability (Quinn & Keough 2002).

Leaf lifespan
Mean leaf lifespans in high light (> 10 %) ranged from 0.7 yrs in Embothrium coccineum to 4.3 years in Gevuina avellana (Table 1).All three of the species present at microsites receiving < 10 % diffuse light had somewhat longer leaf lifespans there than in high light.
Overall, leaf lifespan showed a weak negative relationship with diffuse light availability (Fig. 1).Scaling exponents of this relationship differed significantly among species (P = 0.047), being steeper in lightdemanding Lomatia hirsuta and E. coccineum than in the more tolerant L. ferruginea and G. avellana.However, this relationship was a long way from being statistically significant in Gevuina avellana, and significant at P < 0.05 only in Embothrium coccineum.

Growth
Height growth was positively correlated with diffuse light availability, although this relationship was only marginally significant in L. ferruginea (Fig. 2).Scaling exponents differed significantly among species (P = 0.003), and were steeper in light-demanding Lomatia hirsuta and E. coccineum than in the more tolerant L. ferruginea and G. avellana.
The fastest height growth was recorded in light-demanding E. coccineum and L. hirsuta (Table 1), some individuals of these species growing > 1 m in high light (Fig. 2).
Leaf production of all species responded positively to light availability (Fig. 3), although this relationship was only marginally significant (P = 0.08) for L. hirsuta.The shape of the relationship between rate of leaf production and light availability differed among the four species.Whereas an exponential fit best described the relationship for light-demanding E. coccineum and L. hirsuta, a power function gave the highest r 2 values for the more shade-tolerant L. ferruginea and G. avellana (Fig. 3).

Path analysis
Path analysis suggested that light availability influenced leaf lifespan of all four species primarily through the influence of new leaf production on self-shading (Fig. 4).This relationship was strongest in E. coccineum, with leaf production rate explaining nearly 70 % of the variance in leaf lifespan of this species, and weakest in L. hirsuta, which was found in only a limited range of light environments.The lack of statistical significance in the case of G. avellana at least partly reflects the small sample size of this species (n = 14).When leaf production rate was held constant by multiple regression, light environment per se was not found to significantly affect leaf lifespan of any of the four species.
The explanatory power of our model of variation in leaf lifespan differed widely between species.Whereas the combined effects of leaf production rate and light availability explained nearly 71 % of the variance in leaf lifespan of E. coccineum, this figure was only 21 % for L. hirsuta (Fig. 4).
Gevuina avellana typically retained its leaves for four to six years on average, whereas most individuals of the most light-demanding species (Embothrium coccineum) underwent a complete change of foliage during the 12 months (Fig. 1).In both tropical and temperate rainforests, strongly lightdemanding trees and shrubs often have leaf lifespans of < 1 year, whereas the most shadetolerant taxa often hold their leaves for > 4 years (Coley 1988, Lusk 2002, Poorter & Bongers 2006).
As predicted, responsiveness of leaf lifespan to light environment appeared to be related to species shade tolerance and growth rate.Leaf lifespan of light-demanding E. coccineum and Lomatia hirsuta scaled more steeply with light availability than that of relatively shade-tolerant G. avellana and L. ferruginea (Figs. 1 and 2).Similarly, a comparative study of the tropical rainforest genus Piper found that light availability was strongly negative correlated with leaf lifespan of two light-demanding species, whereas weaker relationships prevailed among more shade-tolerant species (Williams et al. 1989).On the other hand, Lusk (2002) reported rather "individualistic" species responses of foliage turnover to light availability in a temperate rainforest, without finding any clear relationship with shade tolerance.Reich et al. (2004) reported that prolongation of leaf lifespans in low light was a generalized phenomenon in an Amazonian rainforest, without documenting any relationship with species' light requirements.
Path analysis supported the postulated role of self-shading as an important proximate control on leaf lifespan (Ackerly & Bazzaz 1995).Light availability per se had only a very weak negative influence on leaf lifespans of all species when leaf production rate was held constant (Fig. 4), suggesting that mechanisms other than self-shading played little part in the observed correlations of leaf lifespan with light availability (Figs. 1 and 4).The relatively high leaf mortality rates of open-grown individuals of all species therefore likely reflect rapid diffuse light availability or leaf production rate (Fig. 4).Leaves growing on the main stem of a plant receiving most of its light from high elevational angles (e.g., growing directly beneath an opening in an otherwise intact canopy) will undergo a rapid decline in light availability (e.g., Ackerly & Bazzaz 1995).However, such declines are likely to be more gradual in plants receiving abundant side-light (e.g., growing on forest margins), despite the potential for rapid growth in such environments.This hypothesis is consistent with experimental manipulations of seedling light environments in a glasshouse experiment (Ackerly & Bazzaz 1995): leaf lifespans of tropical pioneer seedlings were increased when self-shading was attenuated by using reflective material to increase light availability to old leaves.Similarly, Hikosaka et al. (1994) found that training Ipomoea vines to grow declines in light interception and carbon gain of individual leaves as they age, because of shading by vigorous new growth (Ackerly & Bazzaz 1995).In contrast, light environments of leaves would be expected to change more gradually in slow-growing plants with low leaf production rates.The evidence for control of leaf lifespan by self-shading was strongest in E. coccineum, the fastest-growing and most light-demanding of the four species, and weakest in L. hirsuta.This latter result probably reflects the restricted range of light environments in which we were able to find juvenile trees of L. hirsuta.
The elevational distribution of canopy openness was not examined in this study, but may account for some of the unexplained variation in leaf lifespans.Depending on species, between 29 and 79 % of the variance in leaf lifespan could not be explained by global horizontally led to prolongation of leaf lifespan, presumably as a result of reduced self-shading.Although canopy structure was quite variable at our study site, we did not obtain information on the elevational distribution of photosynthetic photon flux.This hypothesis could readily be tested in the field by using hemispherical photos to provide detailed information on the elevational and azimuthal distribution of light (Rich 1990).
In conclusion, two lines of evidence presented here support the postulated role of self-shading as a major proximate determinant of leaf lifespan.Firstly, the negative relationship of leaf lifespan with light availability was steepest in light-demanding species (Fig. 1), whose growth rates responded mostly dramatically to variation in light availability (Fig. 2).Second, when leaf production rate was held constant by multiple regression, light availability per se had little effect on leaf lifespan of any species (Fig. 4).Nevertheless, factors that remained unquantified in this study, such as the elevational distribution of canopy openness, must be invoked to explain 29 to 79 % of the intraspecific variance in leaf lifespan of the four Proteaceae that we studied.More sophisticated quantification of plant light environments could help address some of this unexplained variation.

Fig. 3 :
Fig.3: Influence of light availability on production of new leaves of four Proteaceae, over a 12-month period.Fitted lines/curves show best fits for each species.Best fits were achieved with a power function for Embothrium coccineum (r 2 = 0.489) and Lomatia hirsuta (r 2 = 0.200), and with an exponential function for Gevuina avellana (r 2 = 0.582) and L. ferruginea (r 2 = 0.423).Solid lines and curves show significant fits at P < 0.05; dashed lines show non-significant relationships.

Fig. 4 :
Fig. 4: Path diagrams showing dependence of leaf lifespans of four Proteaceae on light availability and growth.Values above black arrows show coefficients of determination.Residuals show proportion of variance not explained by the model.Diagramas de vía mostrando la dependencia de la longevidad foliar de cuatro Proteáceas sobre la disponibilidad lumínica y el crecimiento.Las cifras sobre las flechas negras muestran los coeficientes de determinación.Los residuales indican la proporción de la varianza no explicada por el modelo.

TABLE 1
Muñoz (1980) height, growth and mean leaf lifespan data of four Proteaceae on which measurements of height growth and leaf lifespan were made.Potential height data obtained fromMuñoz (1980).