Maximizing agro-ecosystems’ carbon assimilation has focused the attention of researchers for many years. Increasing net primary production is key for the economies of wood and food production systems and it plays a major role in the global carbon cycle (Ryan et al., 1997; Evans, 2013). Multiple environmental factors can constrain the photosynthetic carbon gain of the plant or system. For example, drought stress can impose both stomatal and non-stomatal limitations to photosynthesis (Scalisi et al., 2020; Marino et al., 2021), high temperature can inhibit photosynthesis (Djanaguiraman et al., 2020) and promote respiration leading to carbon starvation (Murphy and Way, 2021), unbalanced tree nutrition can reduce net photosynthesis (Loustau et al., 1999; Sperling et al., 2019), and viruses can reduce mesophyll conductance to CO₂ and Rubisco activity (Sampol et al., 2003).

In addition to these abiotic and biotic factors, the plant’s internal physiological and morphological factors can also strongly inhibit or enhance photosynthesis. Among these intrinsic factors affecting photosynthesis, leaf age, and sink-source balance are particularly important. A leaf’s potential photosynthetic performance depends significantly on its age. Photosynthesis is parabolic through the season, increasing with leaf expansion, reaching its maximum at full leaf development, and declining with senescence (Jurik, 1986; Warren and Adams, 2001; Xu et al., 2019). Reductions in the carboxylation rate, electron transport (Warren, 2006), nitrogen remobilization, decrease in the Rubisco to chlorophyll ratio (Warren and Adams, 2001) and chloroplast degradation (Camp et al., 1982; Wittenbach et al., 1982) accompany the final leaf decline.

The periods between leaf development and senescence represent a significant fraction of the leaf lifespan; hence, changes in the timing of these phenological stages can strongly modify leaf lifetime carbon assimilation (Wilson et al., 2001; Niinemets et al., 2012; Guzmán-Delgado et al., 2017). The proportion between source and sink organs within a branch or a tree can affect photosynthesis directly by enhancing or inhibiting it or indirectly by modifying leaf ontogeny. Researchers have contrasting views on whether the sources (organs that are the net producers of photoassimilates) or the sinks (organs that are the net importers of photoassimilates) have the main effect on photosynthesis and the most recent conclusion is that both, combined, should be considered (Evans, 2013).

Iglesias et al. (2002) girdled, partially thinned fruits, and defoliated citrus trees to modify their source-sink ratio. They found that leaf photosynthetic rates decreased after fruit thinning and increased after leaf removal. Andrade et al. (2019) demonstrated that reduced leaf photosynthetic rates in fruit-thinned nectarine trees was associated with lower stomatal conductance and sugar accumulation in leaves and not with impaired photochemical processes. In contrast, the increase in leaf photosynthetic rates in response to defoliation is often associated with a reduction in total non-structural carbohydrates (NSC) in leaves. However, Pinkard et al. (2011) found that, in Eucalyptus globulus, such an increase in photosynthesis in response to defoliation was linked to the degree of resource availability, with the largest increase being attained in source-limiting treatments and with no increase being observed in sink-limiting treatments. Genotype and environmental conditions can also affect the response of carbon assimilation to thinning and defoliation in maize (Abeledo et al., 2020). Source-sink ratio interacts with leaf age, accelerating or delaying leaf senescence (Paul and Foyer, 2001; Sekhon et al., 2012) and impacting orchard carbon assimilation potential at the season’s end (Marino et al., 2018b). Most studies agree on the central role of NSC hyperaccumulation and nitrogen remobilization in triggering early leaf senescence when source-sink ratios are manipulated (Christensen et al., 1981; Ceppi et al., 1987; Sekhon et al., 2012), but the exact mechanism of action is complex and still not clear.

Pistacia vera has a strong apically dominant growth habit with the primary sink, the fruits, borne of the previous year’s growth, and the primary source, the leaves, borne of the current year’s growth ( Figure 1 ). Pistachio trees have marked annual changes in the source-sink ratio within the branch that strongly affect productivity. A heavy fruit load will precipitate abscission of the reproductive buds and later cause early leaf senescence (Marino and Marra, 2019; Benny et al., 2020; Benny et al., 2022). The result in the following season within the individual branch is no fruit or alternate bearing, and less shoot growth due to decreased NSC reserves (Barone et al., 2014). These characteristics make pistachio trees perfect candidates to investigate the effect of manipulating source-sink relationships on leaf photosynthesis as their leaves age. Previous works in pistachio reported higher photosynthetic rates in bearing branches versus non-bearing branches only in the first part of the season (Vemmos, 1994, Baninasab et al., 2010; Sajjadinia et al., 2010). After July, a decline was observed in photosynthetic rates in bearing branches and trees that was associated with an early leaf senescence precipitated by the branch crop load. However, all these studies have been performed in rainfed or supplemental irrigated orchards, where water stress may have hindered the potential of branches to achieve higher photosynthetic rates, particularly later in the season when soil water resources are at their minimum. In addition, the branches or trees analyzed were always naturally bearing or no bearing fruits due to the alternate bearing phenomenon typical of this species. These branches or trees may have started the season with very different levels of resources, including starch (Crane and AI-Shalan, 1977) or nutrients such as N, P, Mg, K and Ca (Brown et al., 1995; Baninasab et al., 2007). For instance, in Vemmos’s study (1994) specific leaf mass was higher in fruiting than non-fruiting trees early in the season, when differences in photosynthetic rates were observed. According to this author, the earlier leaf emergence and maturation observed could have triggered these differences in photosynthesis. Finally, none of these studies have considered the influence of branch leaf area on photosynthesis.

In this work, we manipulated source-sink ratios in pistachio branches by partial thinning and defoliation and characterized leaf photosynthetic performance through the season as dictated by tree phenological and physiological changes. The information developed can be used to evaluate the potential impacts of resource competition on the leaf’s seasonal carbon gain potential.

The manipulation treatments significantly affected the branch biometric parameters measured ( Table 2 ), with the fruit thinning influencing significantly the dry weight of all the fruits in the branch and the defoliation influencing significantly the dry weight and the area of all the leaves in the branch. The branch source: sink ratio was influenced by both manipulative treatments (defruiting and defoliation) and their interaction ( Table 2 ). The fruit thinning resulted in a ~70% reduction of sink weight per branch, from about 24 g in high-sink branches to about 8 g in low-sink branches ( Table 3 ). The defoliation reduced leaf dry weight by 55% (from 7 to 3 g) and leaf surface area by 60% (490 to 200 cm²). These changes resulted in very different source:sink ratios. Removing only the leaves led to a source sink ratio of 8 (‘low-source:high-sink’ treatment); removing both leaves and fruits resulted in a source:sink ratio of 25 (‘low-source:low-sink’ treatment); eliminating the fruits led to a source:sink ratio of 104 (‘high-source:low-sink’ treatment); the non-manipulated treatment (‘high-source:high-sink’) had a source sink ratio of 21.

The light response curves of pistachio leaves varied through the season, with higher asymptotes (A_{N max}) earlier, and lower asymptotes later ( Figure 3 ). The main exception to this trend was the curve measured at 134 GDD at the end of April; it had a lower asymptote than the curves measured in the subsequent days over the different treatments and a slower, smoother increase in A_Nmax at low PPFD values. The low-sink treatments tended to have the most compact group of curves, with A _Nmax homogenously decreasing from 20 to 10 µmol m^-2 s^-1. The low-source: low-sink treatment also showed a homogeneous reduction of the curves’ asymptotes through the season but with lower values at the end of the season (6 µmol m^-2 s^-1). High-source treatments had the highest A _Nmax, 24 µmol m^-2 s^-1, for the measurement taken at 309 GDD, and the lowest A _Nmax, 5 µmol m^-2 s^-1, for measurement taken at 2500 and 2900 GDD. Interestingly, the decrease in A _Nmax was less homogeneous in these two treatments, and the curves split into two groups due to a strong decline of A _Nmax around 2000 GDD. This gap was greater when high-sink was combined with low-sources.

A three-way ANOVA was performed to a reduced dataset, with the measurements starting June 1^st, and including the defoliation treatment effects applied at that time ( Table 4 ). Results of a two-way ANOVA performed using only the sink treatment and the entire dataset starting in April showed the same significance of the three-way ANOVA reported in Table 4 for the factors Sink and Leaf Age. The three-way ANOVA demonstrated that the sink and source levels did not interact to affect the photosynthetic parameters monitored. The effect of leaf age on photosynthetic parameters was consistently highly significant ( Table 4 ). Maximum photosynthetic rate (A_Nmax) decreased through the season as leaf aged, attaining 50% lower values in fall with respect to spring (~ 10 µmol m^-2 s^-1 versus 20-25 µmol m^-2 s^-1, respectively) ( Figure 4A ). The quantum yield of assimilation (ϕ) was constant through the season with average values of 0.04 µmol m^-2 s^-1 ( Table 5 ), and only very late in the season (after 2600 GDD) ϕ values were slightly lower ( Figure 4B ). The curvature convexity factor (θ) values ranged between 0.7 and 0.9 and showed a poor seasonal pattern ( Figure 4C ). Dark respiration (Rd) highest rates of 3.7 µmol m^-2 s^-1 were recorded earlier in the season, until 309 GDD, then values decreased constantly by 40%, to 2.3 µmol m^-2 s^-1 observed at 934 GDD. Rd rate increased again at 1500 GDD and then was reduced to minimum values at 2900 GDD ( Figure 4D ). Source manipulation influenced A_Nmax and R_d independently from leaf age ( Table 4 ), with 20% higher A_Nmax values in the high-source than in the low-source branches (averages of 16.9 and 13.7 µmol m^-2 s^-1, respectively) and 10% lower R_d in the low-source branches with respect to the high-source branches (averages of 2.6 and 2.8 µmol m^-2 s^-1) ( Table 5 ). The most significant impact of branch source on A _Nmax was observed at 2100 GDD ( Figure 4A ), when A _Nmax was 40% higher in high-source branches ( Figure 4A ). The most significant impact of branch source on R_d was observed later in the season, at 2800 GDD, when R_d was 25% higher in high-source branches ( Figure 4D ). The branch sink influenced A_Nmax and θ only in interaction with leaf age ( Table 4 ). Specifically, high-sink branches had 30% higher A _Nmax at ~1500 GDD, but then 30% lower A _Nmax at 2500 GDD than low sink branches ( Figure 4A ) and θ was 40% higher in high-sink branches versus low-sink branches only at 2200 GDD ( Figure 4C ).

Both branch sink and source influenced leaf NSC content ( Table 6 ), with branches with high sink and low source having 20-25% more sugars (30 mg g^-1) and starch (13 mg g^-1) than branches with low sink and high source (24 mg g^-1 of sugars and 9 mg g^-1 of starch) ( Figure 5 ). Only the sink influenced leaf nitrogen content ( Table 6 ), and high-sink branches had 25% less nitrogen than low-sink branches (17 vs 20 mg g^-1, respectively) ( Figure 6 ). No interaction between the two treatments was observed on any of the measured parameters that, nevertheless, varied with leaf age ( Table 6 ) Interestingly, the effect of the source on leaf nitrogen and starch contents showed a significant interaction with leaf age ( Table 6 ). Specifically, between 2000 and 2500 GDD, a strong decline in leaf nitrogen content paired with an increase in leaf starch content was observed in low-source branches ( Figures 5 , 6 ). For leaf nitrogen content, also the sink showed an interaction with leaf age ( Table 6 ). Differences in leaf nitrogen content were greater around 2000 GDD, when low-sink branches had 30% more nitrogen than high-sink branches, and decreased as time progressed, with low-sink branches having only 15% more nitrogen at 2500 GDD ( Figure 6 ).

Plant growth, reproduction and survival depend on their leaves’ ability to maximize carbon gain through the season. Hence, understanding the processes that impair or enhance leaf photosynthesis is essential to improve agroecosystems’ performance under variable conditions. As seen in Figure 3 and reported for other tree crop species, leaf maximum photosynthetic rates (A_Nmax) decreased through the season as the leaves aged (Jurik, 1986; Reich et al., 1991; Stirling et al., 1994; Warren and Adams, 2001; Warren, 2006). Interestingly, branch source-sink ratio affected the relation between A_Nmax and leaf age. The A_Nmax seasonal decline was gradual only in branches with a high source-sink ratio ( Figures 3B, D ). In branches with a low source-sink ratio, A_Nmax dropped abruptly by 70%, from 25 to 8 μmol m^-1 s^-1 in a two-week period, between 1700 and 2000 GDD. This data is also in line with previous studies on seasonal photosynthetic changes in pistachio reporting a reduction in A_Nmax at the beginning of July (Vemmos, 1994), especially pronounced in bearing branches and trees.

Two important events framed this quick decline in A_Nmax. The first was at 1500 GDD, when the A_Nmax of the high-sink branches increased by 28% with respect to the previous measurement ( Figure 4 ). This increase was not observed in the low-sink branches, which were assimilating significantly less carbon than the high-sink branches (18 μmol m^-1 s^-1 vs. 25 μmol m^-1 s^-1). Higher photosynthetic rates in branches with higher fruit loads are commonly reported in the literature (Iglesias et al., 2002; Andrade et al., 2019). Mid-July (~1.500 GDD) in pistachio is a period of intense carbon demand since the kernels are rapidly growing (Marino et al., 2022). Hence, the high sink strength for carbon exerted by the fruits in the highly loaded branches leads to leaf NSC depletion and eventually the upregulation of photosynthesis. The lack of differences in A_Nmax among treatments before 1500 GDD suggests that the high-sink branches, with three clusters, were not source-limited until the onset of kernel growth, but also that the low-sink branches, with a single fruit cluster, were not sink-limited. However, based on our recently published carbon budget model for pistachio (Marino et al., 2022), the branch photosynthetic apparatus at that time assimilates four times what is needed to support one single cluster with 16 nuts. Therefore, we hypothesize that A_Nmax was not inhibited before 1500 GDD in low-sink branches as NSC are not hyperaccumulating in leaves ( Figure 5 ) but are being translocated to stronger sinks, such as heavily loaded branches and storage, that get depleted in the first part of the season (Tixier et al., 2020). This observation suggests that species with strong fruit sinks, such as pistachio, have seasonal changes in sink strength exerted by the growing fruits, which greatly affect the tree’s physiological responses to treatments and, potentially, management strategies. This characteristic may explain the unique behavior of this species observed in some irrigation studies where assimilation rates differ markedly for similar values of stomatal conductance rates depending on the fruit phenological stage (Gijon et al., 2011) and load (Marino et al., 2018a). In research studies, sampling and measurements should be replicated in bearing and non-bearing branches, or the bearing status of the branch or tree should be reported and considered when interpreting the results. In terms of orchard management, the data from this study can be implemented in carbon budget models and the results used to develop improved practices that consider tree crop load and its effect on leaf photosynthetic performance. The second significant impact of branch source-sink ratio on leaf photosynthesis was observed late in the growing season, at 2200 GDD ( Figure 3 ). As described above, A_Nmax started to decline after 1500 GDD, more rapidly in branches with a low source-sink ratio ( Figure 3 ). At 2200 GDD, the low-source branches had significantly lower A_Nmax than the high-source branches (9-13 μmol m^-1 s^-1 vs. 16 μmol m^-1 s^-1, respectively), and at 2600 GDD, the high-sink branches had lower A_Nmax than the low-sink ones. In both cases, the lower A_Nmax in branches with low source-sink ratios was associated with higher leaf NSC concentration ( Figure 5 ). These results differ from studies showing that source-limited conditions produced by defoliation are generally associated with low levels of NSC and an increase in A_Nmax (Iglesias et al., 2002; Pinkard et al., 2011), and sink-limited conditions created by fruit thinning are generally associated with high leaf NSC concentration and lower A_Nmax (Iglesias et al., 2002; Andrade et al., 2019). Our data suggest that the substantial decline in A_Nmax observed at the end of the season could be the indirect result of branch source:sink ratio on leaf senescence. Specifically, the lowest branch source:sink ratio precipitated leaf senescence. The reduction of the curvature factor (θ) in highly loaded branches observed exactly at this date ( Figure 4 ) support this theory, since lower θ has been associated with lower chlorophyll content (Leverenz, 1987; Leverenz, 1988), which in turn is a main symptom of leaf senescence, and correspond to a period when large differences in leaf nitrogen content were observed between high-sink and low-sink treatments ( Figure 6 ).

Lv et al. (2020) reported similar results in wheat with increased reactive oxygen species production and degradation of chlorophyll-protein complexes and carbohydrates. Abeledo et al. (2020) found defoliation and non-thinning anticipated senescence in corn, but the response was affected by genotype and environmental conditions. This point could explain the different trend observed by other authors (e.g., Kumar et al., 2019 and Sekhon et al., 2012), who found anticipated leaf senescence induced by the lack of sinks. However, the consensus in the literature is that NSC hyperaccumulation in the source (leaf) tissue was associated with premature senescence (Christensen et  al, 1981; Ceppi et al., 1987; Sekhon et al., 2012), consistent with what we observed in our study for the lowest source:sink ratio treatment ( Figure 5 ).

It is beyond the scope of this study to elucidate the mechanisms associated with leaf senescence in response to source: sink balance. However, an interesting and unexpected result was that the interaction of the two treatments (fruit thinning and defoliation) on photosynthesis was never significant, potentially suggesting a different response mechanism of pistachio to sink-limited and source-limited conditions. This different mechanism was indicated by N content changes. Declining A_Nmax in high-sink branches was accompanied by a clear reduction in leaf N content during senescence in early August ( Figure 6 ). However, defoliation did not significantly affect leaf N levels. This result suggests N remobilization to the growing kernels could be a major factor in reducing photosynthesis in heavily loaded branches at the end of the season (Mei and Thimann, 1984; Morris and Pauisen, 1985). In addition, defoliation but not fruit thinning affected leaf dark respiration (R_d) rates, with leaves from low-source branches having significantly lower R_d, especially toward the end of the season ( Table 3 and Figure 4B ). Earlier studies on the impact of defoliation on respiration rates show contrasting results, with both increases and decreases being reported (Ovaska et al., 1993; Reich et al., 1993; Layne and Flore, 1995; Pinkard et al., 2007). In our case, the decline in R_d under defoliation corresponded perfectly to the decline in A_Nmax and to the NSC accumulation in leaves. A parallel decrease between A_Nmax and R_d is commonly observed since the two parameters may be affected by similar factors (Amthor, 2012) and may suggest a decrease in biosynthesis and prioritization of assimilate storage. Storage implies lower respiratory costs than biosynthesis (Amthor, 2012). On the contrary, non-defoliated branches showed a respiration peak prior to R_d decline, a typical symptom of senescence onset and generally associated with an increase in metabolic rate to support the energetic needs of remobilization.

This study demonstrates that the source-sink balance at the branch and, potentially, tree level influences the photosynthetic capacity of individual pistachio leaves. An unequal source-sink ratio can promote higher photosynthesis during period of high sink activity, but also accelerate leaf senescence, precipitating a decline in photosynthetic activity late in the growing season. Fall assimilation is important for deciduous species, and increased leaf lifespan can enhance reserve accumulation to support winter dormancy and the following year’s bloom. Based on these results, heavily fruiting branches may have impaired photosynthetic capacities starting in middle-late August. Management could be optimized based on tree crop load. Specifically, in years with high crop load, thinning strategies could be used to improve multi-year tree carbon balance and productivity, and irrigation reductions or cut-offs during post-harvest could be implemented without significantly impacting tree carbon resources. Also, the timing for zinc fall applications could be selected considering also tree crop load and associated leaf photosynthetic and translocation potential. Finally, breeding programs could benefit from the results of this study. If stay-green traits in high-yielding branches or trees are associated with lower alternate bearing, a highly desired trait for growers, they could be used for cultivar selection and in crosses.

Our data suggest that studies focusing on photosynthetic capacity or primary gross production of tree crop species with strong alternate bearing behaviors, such as Pistacia spp., should consider the physiological changes the trees undergo during high bearing and low bearing years. Data from this study can be potentially implemented in available carbon budget models to facilitate the adaptation of management practices to tree sink-source balance.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

GM, PG-D and BB-U conceived the experiment; GM, PG-D, ES, JA, BB-U set up the experiment and collected the data. GM, PG-D and EF-S analysed the data. All authors contributed to data interpretation. GM and PG-D wrote the manuscript. All authors contributed to the article and approved the submitted version.