Decline in photosynthetic nitrogen use efficiency with leaf age and nitrogen resorption as determinants of leaf life span


A. Escudero, Department of Ecology, School of Biology, University of Salamanca, Salamanca 37071, Spain (fax +34 923 294515; e-mail


  • 1Cost-benefit models predict that leaf life span depends on its initial photosynthetic rate and construction cost and on the rate of decline in photosynthesis with age. Leaf gas exchange rates and N contents were measured in nine woody evergreen Mediterranean species with different leaf life spans to determine the effects of leaf ageing on photosynthetic N use efficiency (PNUE). N costs of leaf construction were assumed to be in part dependent on N resorption from senescing leaves.
  • 2Leaf ageing had significant negative effects on photosynthetic rates per unit leaf area. As N content per unit leaf area did not decline until the end of leaf life, PNUE also decreased with age. PNUE generally declined faster in species with a shorter leaf life span. There were no significant interspecific differences in maximum CO2 assimilation rates per unit leaf area and in N resorption that could be related to differences in leaf life span.
  • 3As PNUE decreases with leaf age, shedding of the older leaves and retranslocation of N to the current year's leaf biomass would result in an increase in the mean instantaneous efficiency of use of the N retranslocated. However, total CO2 assimilation can be improved by such shedding only when the increase in the efficiency of use of the remobilized N compensates for the remaining N lost in the shed leaves.
  • 4The photosynthesis of the old leaf cohorts exceeded the increase in photosynthesis that would be obtained from the N retranslocated to the younger leaves, given the observed efficiencies of N resorption. The retention of old leaves thus resulted in a higher whole-canopy CO2 assimilation, despite their low PNUE.


Different terrestrial plant species show large differences in leaf life span, which are correlated with nutrient limitation. Evergreenness was interpreted as an adaptation contributing to improved nutrient use efficiency, thanks to a more prolonged use of nutrients in leaf biomass (Small 1972). Detailed ecophysiological studies, however, have demonstrated that maximum assimilation rates and instantaneous nutrient use efficiencies are usually negatively correlated to leaf life span (Reich et al. 1992, 1999), mainly because longer leaf life span requires a structural reinforcement of the leaf that negatively affects photosynthetic rates (Westoby et al. 2002; Wright et al. 2002). Interspecific differences in leaf longevity have been analysed using cost-benefit models (Williams et al. 1989; Kikuzawa 1991). According to these, a plant should replace its leaves when the net C gain by a leaf per unit time over its entire life span has reached a maximum (Kikuzawa 1991; Ackerly 1999). The basic assumption for this prediction is that a plant can retain only a finite number of leaves at any one time, so that production of a new leaf entails concurrent loss of an older one (Kikuzawa & Ackerly 1999). The alternative scenario is that leaves should be maintained until their instantaneous net photosynthetic rate drops to zero (Kikuzawa 1991; Givnish 2002).

The assumption of a limited number of leaves per plant is reasonable when water, light or nutrients are limited. For example, the commonly observed nutrient resorption during leaf senescence permits the re-use of nutrients. The demand for nutrients from growing tissues may drive nutrient resorption and senescence of older leaves (Mooney 1983; Aerts 1996; Killingbeck & Whitford 2001). Several studies have shown that photosynthetic nitrogen use efficiency (PNUE) tends to decline as the leaves age (Hom & Oechel 1983; Sobrado 1994; Kitajima et al. 1997). Nutrients resorbed from old leaves may be used to increase the biomass of young leaves, thus enhancing the average nutrient use efficiency of the leaf biomass. Nutrient limitation may be the reason why leaves are discarded before they reach zero CO2 assimilation, provided that nutrients may be used more efficiently in younger leaves (Westoby et al. 2002).

It has been predicted that leaf life span will be short when the initial photosynthetic rate of the leaf is high, and long when construction costs are high or the rate of decline of photosynthetic rate with leaf age is slow (Kikuzawa 1991). For a given nutrient availability, nutrient costs of leaf construction should depend mainly on nutrient losses in leaf litter, such that efficient nutrient resorption from senescing leaves may reduce these costs (Kikuzawa 1995). Efficient resorption should therefore favour a short leaf life span. Nevertheless, resorption percentages are not significantly greater in short-lived foliage (Aerts 1996), although other authors have observed a slight tendency to lower resorption efficiencies in species of longer leaf life span (Del Arco et al. 1991; Aerts et al. 1999). The lack of a consistent relationship between leaf life span and nutrient resorption may be due to the other factors that determine the final C balance of the leaves, such as the initial assimilation rate and the rate of decline of assimilation with advancing leaf age.

Under conditions of N limitation, use of N stored in old leaves may allow increases in the growth of new leaf biomass. Nutrients that are resorbed are directly available for plant growth, while nutrients recycled through litterfall may not become available for plant uptake for a long time (Aerts 1996). As leaf ageing usually reduces PNUE, retranslocation of N to the new leaf biomass should increase the efficiency of use of the N recovered. However, plants are not capable of removing all N from senescing leaves. The remainder of the leaf N is lost to the soil when the leaves are shed. Accordingly, the effects of leaf senescence and removal of N on total CO2 assimilation must depend on the efficiency of N resorption and on the rate of decline of photosynthetic rate with leaf age.

Our purpose was to study the relationship between leaf longevity and photosynthetic performance at the time of the year most favourable for gas exchange in a Mediterranean environment, i.e. when it is most probable that photosynthesis will be limited by low N availability. We measured CO2 assimilation rates and PNUE (calculated as the ratio between the instantaneous CO2 assimilation rate and leaf N content) in leaves of different age classes along with resorption efficiencies of N from senescing leaves. Although the literature contains many references to the maximum photosynthetic rates of different species, data on photosynthetic rates over time are comparatively scarce (Kikuzawa & Ackerly 1999). Consequently, despite the importance of the loss of photosynthetic capacity as a determinant of the long-term carbon budget of leaves, there are not many quantitative data on the rate of decline of CO2 assimilation with leaf age in species of different leaf life spans. Moreover, although resorption efficiency strongly affects the integrated nutrient use efficiency of a leaf, there are few studies in which CO2 assimilation rates, leaf life span and nutrient resorption efficiencies have been measured simultaneously (Aerts & Chapin 2000).

Total CO2 assimilation of the leaf biomass can be improved by the shedding of older leaves only when the increase in photosynthesis from resorbed N exceeds the photosynthesis of the leaves lost (Franklin & Agren 2002), assuming that the costs of retranslocation are zero or small (Givnish 2002). This condition is satisfied if the ratio (100 × PNUE in the old leaves/PNUE in the young leaves) is less than the percentage of N recovered from senescing leaves before abscission. Otherwise, retention of the older leaves would result in a higher total CO2 assimilation for the whole leaf biomass. Accordingly, under N limitation, maximum leaf life span must be constrained by both the rate of decline in PNUE with leaf age and the efficiency of N resorption, and the balance between these factors will determine one minimum relative PNUE for a leaf to be retained. We hypothesize that the percentage of N resorption may be taken as a predictor of this minimum value.

Materials and methods

study species and area

The species selected for experimental work (Table 1) were: Taxus baccata L., Ilex aquifolium L., Pinus halepensis Miller, P. pinaster Aiton, P. pinea L., P. sylvestris L., Quercus coccifera L., Q. rotundifolia Lam. and Q. suber L. All these species are evergreen with either relatively prolonged (Pinus spp. and T. baccata) or with relatively short leaf life spans (Quercus spp. and I. aquifolium). In all species, mortality of the old leaf cohorts occurred mainly during the spring and summer (Escudero & Del Arco 1987), coinciding with the period of expansion of the new leaf cohort (Mediavilla & Escudero 2003a).

Table 1.  Mean ± SE leaf longevity, maximum current and total leaf biomass and leaf N pools of the different species. Data are means of 3 years of sampling
SpeciesMean leaf longevity (months)Current-year leaf biomass (g m−2)Total peak leaf biomass (g m−2)Current-year leaf N pool (g m−2)Total peak leaf N pool (g m−2)LAI (m2 m−2)
Quercus suber15.0 ± 2.7373 ± 67.5614 ± 1116.50 ± 0.919.65 ± 1.313.42 ± 0.62
Quercus coccifera15.6 ± 4.4241 ± 67.9367 ± 1043.32 ± 0.835.09 ± 1.211.63 ± 0.46
Quercus rotundifolia23.7 ± 2.8271 ± 32.0608 ± 71.93.46 ± 0.617.21 ± 1.282.46 ± 0.29
Ilex aquifolium25.0 ± 7.3352 ± 42.0898 ± 1115.22 ± 0.7411.2 ± 1.404.94 ± 0.62
Pinus pinea35.5 ± 7.2216 ± 43.8565 ± 1152.07 ± 0.554.93 ± 1.251.84 ± 0.37
Pinus halepensis36.1 ± 2.5133 ± 93.0365 ± 25.61.43 ± 0.163.80 ± 0.251.19 ± 0.08
Pinus sylvestris48.8 ± 5.4174 ± 19.1476 ± 52.42.31 ± 0.225.42 ± 0.521.71 ± 0.19
Pinus pinaster51.2 ± 3.5124 ± 84.0413 ± 28.11.18 ± 0.223.32 ± 0.710.85 ± 0.06
Taxus baccata62.1 ± 4.6132 ± 97.0684 ± 50.31.40 ± 0.104.31 ± 0.302.91 ± 0.21

All these species occurred at six sites close to the city of Salamanca (central-western Spain) between latitudes 41°50′ N and 40°20′ N and longitudes between 5°20′ W and 6°25′ W. Altitudes ranged between 700 and 1500 m above sea level. Owing to this altitudinal range, some climatic differences between these sites were observed. Annual rainfall ranges from 300 mm to more than 1000 mm at the sites situated at the greatest altitude, with most falling during winter and spring. As a consequence, water limitation is usually absent during spring and early summer. All sites, however, underwent a summer drought. The soils, dystric cambisol in all cases, are poor in organic matter and in nitrogen, with a low pH and medium/low water retention capacity. It may be expected that CO2 assimilation in spring and early summer will mainly be limited by the low availability of nutrients, especially nitrogen.

The sites consisted of sparse populations (between 50 and 100 specimens ha−1) of mature (more than 100 years old) individuals. Mean heights were about 4–10 m. All specimens selected for the study were fully sun-exposed.

sampling methods

At each site, three, four or five mature specimens of each species were selected randomly on each sampling date, i.e. at monthly intervals over 3 years (from 1997 to 1999). A composite sampling of branches with leaves from different crown positions of each canopy was undertaken for each selected individual. The samples were immediately taken to the laboratory and the branches were separated into annual segments (shoots) of different age classes. Only one flush of leaf growth was observed in all species. Accordingly, all the leaves born in one particular year were considered to belong to the same age class. All the shoots bearing leaves of a given age were identified as belonging to the same age class. Branches produced in spring on a 1-year-old-shoot were ascribed to a new shoot cohort. Subsamples of 40–50 shoots per sampling date and per age class were used for demographic analyses and for measurements of leaf biomass. The number of leaves or needles per shoot was counted each month for each age class. The samples were oven-dried at 70 °C to constant mass and the mean dry mass per leaf was determined. Projected leaf area was measured by image analysis (Delta-T Devices LTD, Cambridge, UK).

Four 0.25 m2 litter traps were randomly placed below the canopies of each species in each plot. The specimens selected were sufficiently isolated from specimens of other species to ensure that only litterfall from the species sampled would be collected. The litterfall was collected monthly over 3 years and mean monthly mass for each species was estimated after drying the litter at 70 °C to constant mass. Total annual leaf litter production per unit ground area was calculated for each species by adding up the monthly values.

Total N concentrations in leaf and litter samples were determined by combustion (NA-2100 autoanalyser, CE-Instruments, ThermoQuest, Milan, Italy).

gas-exchange measurements

In Mediterranean climates, leaf ageing within a growing season parallels a dramatic increase in the intensity of drought, whose effects mask those of ageing. The effect of ageing can only be isolated from other factors by comparing the assimilation rates achieved at the beginning of the growth season by each leaf age-class (photosynthetic capacity in field conditions).

Net photosynthesis and stomatal conductance of different leaf age classes were measured simultaneously with a portable photosynthesis system (Li-6200, LiCor Inc, Lincoln, NE, USA). Leaves were measured under ambient CO2 concentrations, air temperatures, relative humidities and saturating irradiances (above c. 1000 µmol m−2 s−1). As the purpose of these observations was to measure gas exchange rates under near optimal ambient conditions, all measurements were performed from 07.00 to 09.00 h local solar time on sunny days during late spring and early summer from 1996 to 2001. Photosynthetic rates were measured for between 20 and 30 fully expanded leaves of each age class from four to six individuals per species. Given the low leaf area index typical of these Mediterranean woodlands (Table 1), even the oldest leaves received full sunlight during at least part of the day. Immediately after the gas-exchange measurements, each leaf was harvested, taken to the laboratory, and its area, dry mass and N concentration determined.

From the data obtained, we calculated leaf mass per area (LMA), leaf N content per unit area (N/area), net CO2 assimilation per unit leaf area (A/area) and instantaneous PNUE (A/N). The A/g ratio (CO2 assimilation per unit area (A)/stomatal conductance to water vapour (g)) was used as an estimate of water use efficiency for a given vapour pressure deficit (intrinsic water use efficiency).


The mean number of leaves per shoot of a given age on each census date was used to construct static life tables (Begon & Mortimer 1986), which made it possible to calculate the mean leaf life span for each species according to standard methods. Due to the gradual leaf fall, maximum leaf life span was much longer than mean leaf life span and in all species we were able to measure photosynthesis in leaves older than the estimated mean leaf life span.

Leaf mass per shoot was calculated each month over the 3-year sampling period by multiplying the mean number of leaves per shoot (Li) by the average dry mass per leaf (Mi), where i = age of the shoot in months (i was set to 1 in the month in which the maximum leaf number per shoot was attained).

The leaf biomass produced per shoot was estimated as the sum of the leaf mass per shoot at the age when the leaves attained their maximum individual mass (and stopped growing) plus the previous losses of leaf biomass per shoot owing to the shedding of the leaves that died before reaching their maximum mass:

image(eqn 1)

where x is the age at which the leaves reached their maximum individual mass.

With this procedure we estimated the total leaf biomass produced per shoot throughout its entire lifetime (i.e. the length of time elapsed between shoot initiation and the loss of the last leaf on the shoot). In evergreen species, with overlapping annual leaf cohorts, our calculation is roughly equivalent to estimating the annual leaf biomass production per shoot of each of the different cohorts and then summing over all cohorts.

As the specimens studied were assumed to be at steady-state with respect to leaf biomass, the total leaf litter produced annually by each plant was assumed to be equivalent to the leaf biomass produced annually by the same plant, because the loss of leaf mass caused by the resorption of leaf components during the senescence period was relatively low (less than 5% for all species studied). Thus, the leaf biomass per cohort for each month for each species was estimated on a ground area basis as:

biomass of leaves of age i (g m−2) = leaf mass per shoot of age i (g) × total annual leaf litter production (g m−2)/leaf biomass produced per shoot (g)(eqn 2)

This allowed the data to be expressed as leaf biomass in units comparable to the leaf litter production measured below the canopy of each species. The total leaf area index was estimated by dividing leaf biomass per unit ground area by mean LMA.

The leaf N pool size in each month was estimated by multiplying the leaf biomass by the N concentration. The annual N loss in leaf litterfall was calculated by adding up the quantities of N in the monthly leaf litterfall.

The fraction of the leaf N pool annually resorbed prior to leaf fall was then calculated as:

(maximum N pool size − N in leaf litterfall)/maximum N pool size(eqn 3)

where N quantities are measured in g m−2 ground area.


Linear regression analyses were used to examine relationships between leaf longevity and leaf biomass. Given the strong relatedness among many of the species, instead of using the species as independent data points, we analysed our data set by phylogenetically independent contrasts (Felsenstein 1985). Classification of species in families and higher groupings followed the phylogenies published by Soltis et al. (2000), while generic delineation followed Liston et al. (1999) and Manos et al. (1999). Independent contrasts for leaf biomass and LAI were obtained using CAIC vs. 2.0.0 (Purvis & Rambaut 1995), and were regressed against those produced for leaf life span. Linear regression analyses were also used to calculate the slope of decline in PNUE with advancing leaf age for each species. Differences among species in the slope of decline in PNUE with advancing leaf age were then analysed by phylogenetically independent contrasts. A two-tailed t-test was used to check whether the mean PNUE of the different leaf cohorts was significantly different to the hypothesized minimum PNUE for a leaf to be retained in the canopy. All the statistical analyses were performed using the SPSS Statistical Package (SPSS Inc., Chicago, IL, USA).


Mean leaf life span varied between 15 months in Q. suber and 62.1 months in T. baccata (Table 1). Contrasts for current-year leaf biomass were negatively related to contrasts for leaf life span (y = 0 – 3.803x, P = 0.0399, r2 = 0.475, n = 8). However, contrasts for total leaf biomass were independent of leaf life span (P = 0.75), because the overlapping of many leaf cohorts in the canopy of species with longer leaf life spans compensated for their lower leaf biomass per cohort (Table 1). Obviously, in the species with a long leaf life span a large proportion of the total biomass consisted of leaves more than 1 year old. Likewise, the proportion of the total N pool stored in old leaf biomass was large in species with greater leaf life span, which indicates that the PNUE achieved by old leaves must make an important contribution in determining the average PNUE in these species. For example, in T. baccata, at the time of year when the total leaf mass is greatest, immediately after the expansion of the new flush, only 32% of the total leaf N pool was in the current-year leaves. For Pinus species, I. aquifolium and Q. rotundifolia, current-year foliage stored between 33 and 47% of the total leaf N pool. Finally, in the evergreens with the shortest leaf life span (Q. suber and Q. coccifera) the new leaves represented over 66% of the total leaf N pool (Table 1). Because of their high LMA, species with long leaf life span had a small LAI relative to their total leaf biomass (Table 1), and contrasts for LAI were not significantly related to leaf life span (P = 0.76).

In all species, the N concentration per unit leaf mass decreased as leaves aged (data not shown). However, this reduction was simply due to the dilution of N in a larger leaf mass per area, as the N/area did not change significantly with leaf age (Fig. 1). N resorption from older leaves did not occur until the leaves became senescent and prone to shedding. The rates of CO2 assimilation per unit area tended to decline with leaf age in all species, but intrinsic water use efficiency (A/g) only declined in the oldest leaf cohorts of T. baccata and P. sylvestris (Fig. 2). There were no clear interspecific differences in maximum A/area that could be related to differences in leaf life span. However, because of the tendency of long-lived foliage to maintain greater LMA (Fig. 1), maximum CO2 assimilation per unit leaf mass tended to decline with increasing leaf longevity (data not shown).

Figure 1.

Leaf mass (LMA) and nitrogen concentrations per unit area (N/area) as a function of age class in leaves of evergreen species with different leaf life spans. Data are means of 20–30 measurements. Bars represent 1 SE.

Figure 2.

Net CO2 assimilation rate per unit leaf area (A/area) and intrinsic water use efficiency (A/g) as a function of age class in leaves of evergreen species with different leaf life spans. Data are means of 20–30 measurements. Bars represent 1 SE.

As a consequence of the decline in A/area without changes in nitrogen per unit leaf area, PNUE decreased approximately linearly as the leaves aged (Fig. 3). When the PNUE of the different age classes was expressed as a percentage of the maximum PNUE found in young tissue, the slopes of the linear regressions fitted to the data for the decline in PNUE with leaf age of the different species tended to be steeper for short-lived foliage (Table 2). Independent contrasts for the slopes were positively related to contrasts for leaf life span (y = 0 + 0.229x, P = 0.0121, r2 = 0.677, n = 8). On average, the maximum photosynthetic N use efficiencies of the different leaf cohorts decreased at a rate of between 19 and 28% per year in the oaks and I. aquifolium, and at a rate of 9–17% per year in the other species (Table 2).

Figure 3.

Mean PNUE (+1 SE, n = 20–30) of the different leaf cohorts expressed as the percentage of the maximum PNUE. Horizontal lines represent mean N resorption percentages of each species. Above each bar, asterisks indicate that observed PNUE is larger than the mean resorption percentage (t-test). ***P < 0.001, **P < 0.01, *P < 0.05, NS = not significant.

Table 2.  Summary of linear regressions relating photosynthetic nitrogen use efficiency (percentage of maximum) to leaf age (years)
Quercus suber−23.31230.300.0001
Quercus coccifera−19.31190.140.1030
Quercus rotundifolia−28.31260.600.0001
Ilex aquifolium−19.01220.380.0001
Pinus pinea−16.91080.260.0001
Pinus halepensis−15.11170.140.0010
Pinus sylvestris−15.41190.160.0003
Pinus pinaster −8.751100.120.0020
Taxus baccata−13.51210.170.0001

There were no clear interspecific differences in N resorption efficiency related to differences in leaf life span. For example, although the lowest resorption efficiency (25%) was exhibited by the species with longest leaf life span (T. baccata), the highest (49%) corresponded to P. sylvestris, despite its prolonged leaf life span (Fig. 3). As a consequence of these low resorption efficiencies, the PNUE of old leaves expressed as a percentage of the PNUE of current-year leaves was in most cases greater than, and never significantly less than, the resorption percentage of each species (Fig. 3), despite the decrease in PNUE with age.


In agreement with our expectations, instantaneous PNUE of the leaf cohorts found in the canopy was usually above the predicted minimum PNUE for a leaf to be retained, taking into account the resorption efficiency typical of each species (Fig. 3). The retention of old leaves, despite their low PNUE, can be thus explained as a means of increasing total production because of a more prolonged use of N. A reduced resorption efficiency implies that more N will be lost when the leaves are shed. Under these circumstances, delaying leaf abscission may be advantageous in terms of total production as it allows more N to be retained in the canopy, although the average instantaneous PNUE is small.

The measurements of gas exchange were carried out in late spring and early summer. In the species with leaf abscission in spring (oaks), leaves from the oldest cohort, which had survived the spring abscission period, still had a relatively high PNUE (Fig. 3), because at the time of the measurements they were still in an intermediate phase of their life cycle. It is probable that the leaves shed earlier were those that underwent a faster deterioration or occupied less illuminated positions, and that, for these reasons, reached the minimum PNUE before the surviving leaves. A significant variability in leaf life span inside the crown has been found in rain forest species (Osada et al. 2001), although, for most leaf traits, between-species variation tends to be far greater than that within species (Wright et al. 2002). In the species with leaf abscission throughout the summer (conifers and I. aquifolium), the PNUE of the oldest leaf cohort was close to the hypothesized minimum (Fig. 3), the only exception being P. sylvestris, which had a lower PNUE in its oldest leaf cohort. As the leaves of the oldest cohort in these species were close to the end of their lives at the time of measurement, the results support our hypothesis that leaves would be discarded only when the recovery of N from senescing foliage results in a higher total production. Values of PNUE below the predicted minimum may be explained by the existence of significant energetic costs of N resorption (Field 1983), which should reduce the relative PNUE at which the leaves must be shed. Possible among-species differences in such costs should also contribute to the differences in the minimum relative PNUE and in the leaf longevity of the different species.

Previous optimization models of N reallocation (Field 1983; Hirose & Werger 1987) have postulated that the distribution of leaf N contents that maximizes carbon gain over the whole array of leaves is such that the rate of change of carbon gain with changing leaf N is equal for every leaf. Accordingly, N should be allocated to different leaves so that their marginal return on further N is equal (dA/dN = constant). These models predict that the N concentration per unit leaf area should decrease with increasing depth in the canopy, as a result of the reallocation of N from shaded to more illuminated parts of the canopy. We did not, however, observe any N reallocation within the canopy during most of the leaf lifetime, because the decrease in N/mass with advancing leaf age was compensated for by an increase in LMA (Fig. 1), probably because of accumulation of ash (Kitajima et al. 2002) and/or lignin (Horner et al. 1987), which resulted in a similar absolute N content per leaf in the different age classes. N resorption did not occur until the end of leaf life, and was associated with leaf senescence, such that the resorption of N from a leaf implied the loss of that leaf. In all species, the abscission of older leaves was more or less simultaneous with the growth of the new leaf biomass (Escudero & Del Arco 1987). In oaks, most leaf abscission occurred during the spring, shortly before the expansion of the new leaf biomass. Leaf abscission occurred later in the remaining species and coincided with their later leaf emergence and slower leaf growth (Mediavilla & Escudero 2003a). It is thus likely that the resources resorbed from senescing leaves would be used to promote the growth of young leaves. The negative correlation observed between leaf longevity and current-year leaf biomass (Table 1) and between leaf longevity and N concentration per unit leaf mass (Mediavilla & Escudero 2003b) suggest that the large amounts of resources stored in old leaf biomass in species of long leaf longevity contribute to reducing the amounts of resources available for the construction of young leaf biomass, and that the amount of young leaf biomass should be greater if the old leaves are discarded. Under these circumstances, the effects of N resorption on C gain depend on the balance between the photosynthesis of the leaves lost and that of the new leaf biomass produced using resorbed N, and are determined by the decrease in PNUE with age and the efficiency of N resorption. Models based on the reallocation of N among live leaves within the canopy (Field 1983) are clearly not applicable to our species, which show no such reallocation.

The absence of N reallocation among live leaves may be explained by the low LAI of our species (Table 1). Hirose & Werger (1987) found that reallocation of N within the canopy had negligible effects on total A for an LAI of around 2.1. Only Q. suber and I. aquifolium had LAI above 3, and most of our species were below 2, thus suggesting that reallocation of N among leaf cohorts would have little effect on total CO2 assimilation. Furthermore, N concentrations per unit leaf mass were low in all of our species (data not shown). A further reduction in the N concentration in old leaves as a result of N withdrawal could lead to a severe reduction in both photosynthesis and PNUE (Field & Mooney 1986), which could drive their senescence.

The rate of decline of photosynthetic capacity with increasing leaf age seems to be lower in species with longer leaf life spans (Field & Mooney 1983; Reich et al. 1992; Hiremath 2000; Mediavilla & Escudero 2003b). Such a trend (Table 2) may contribute to explaining the observed differences in leaf life span. However, the reduction in PNUE with increasing leaf age was significant in all species except Q. coccifera. Consequently, a careful estimation of the proportions of the total photosynthetic N pool present in each of the different leaf cohorts is essential for a proper calculation of total CO2 assimilation and of the average PNUE of the whole canopy.

The decline in the CO2 assimilation rate with leaf age could be a consequence of self-shading, as new leaves are produced in upper-canopy positions, rather than of physiological deterioration (Field & Mooney 1983; Kitajima et al. 1997; Ackerly 1999). In fact, some studies have found that the decrease in photosynthetic capacity with leaf ageing was paralleled by decreases in leaf N contents, such that PNUE did not vary with leaf age (Mooney et al. 1981; Field & Mooney 1983). However, in the present study N content did not decline with leaf age, and the decline in CO2 assimilation rate caused a strong decline in PNUE, as previously observed (Hom & Oechel 1983; Sobrado 1994; Kitajima et al. 1997).

Given the low LAI of the study species, it is, however, unlikely that photosynthesis by the leaves of most age-classes would be significantly limited by shading. In fact, our measurements of gas exchange were made at saturating light levels for all leaf age classes because it was easy to find sunlit leaves even for the oldest age classes. The reduction in light levels within the canopy may be more intense in fast-growing species but, in slower-growing species, age-related senescence processes, i.e. physiological deterioration, may be more important than shading in determining the reduction of photosynthesis, and thus PNUE, with leaf age (Ackerly 1999). The decline in A/area was paralleled by a decrease in stomatal conductance (data not shown), although the latter is unlikely to decrease PNUE. A/g remained constant throughout most of the leaf life (Fig. 2), suggesting that the reduction in stomatal conductance with leaf age would rather be a response to the lower assimilatory capacity of the mesophyll (Farquhar & Sharkey 1982). This indicates that certain characteristics of older leaves, such as a greater LMA, may have some influence over the decrease in PNUE. A negative correlation between PNUE and LMA in interspecific comparisons has been reported by several authors (Reich et al. 1991; Lloyd et al. 1992; Mediavilla et al. 2001), and may be due to a greater relative allocation of N to non-photosynthetic functions in leaves with a high LMA, as well as a different partitioning of photosynthetic N between light harvesting complexes, electron transport and CO2 fixation (Poorter & Evans 1998). It is possible that the same trends occur as leaves age and that they are the cause of declining PNUE.

Contrary to our expectations, maximum initial A/area was not correlated with leaf life span. Differences in CO2 assimilation rates related to differences in leaf longevity have been reported frequently when assimilation rates are expressed per unit leaf mass (Reich et al. 1992, 1999). However, the differences tend to disappear when gas-exchange rates are expressed per unit leaf area (Körner 1995).

Under water-limited conditions, another requisite for old leaves to be retained is that water use efficiency should not decrease with leaf ageing. Unlike N, water is expended during gas exchange and it cannot be re-utilized in the future. Consequently, leaves should be discarded as soon as their water use efficiency declines below the mean efficiency of the canopy. Although CO2 assimilation rates decreased with leaf age, A/g remained more or less constant throughout most of the leaf lifetime (Fig. 2), the only exception being the oldest leaves of P. sylvestris, which were close to the end of their lives at the time of measurements. In T. baccata, however, the three oldest leaf cohorts had a low A/g, but this species tends to occupy moist sites in the mountains and it is probably less water-limited than the other species. The retention of older leaves thus allows the whole-canopy photosynthetic capacity to be increased by a more prolonged use of N and, in most species, does not reduce the mean water use efficiency of the canopy.


This paper has received financial support from the Spanish Ministry of Education (Project No. AMB95-0800) and from the Regional Government of Castilla-León (Projects No. SA 47/95 and SA 72/00B).