Olive growing has been shifting from traditional low-density orchards to intensive orchards, including super intensive (or super high density, SHD) orchards (up to > 1500 trees hectare^-1), which are still a minority but are increasing (Guerrero-Casado et al., 2021). As the olive system intensifies, low vigor and early and abundant bearing are essential traits (Tous et al., 1999; Rallo et al., 2007), indispensable to make SHD orchards economically viable (De Benedetto et al., 2003). However, despite much research on tree vigor and dwarf cultivars (Barranco, 1997; Sonnoli, 2001; León Moreno, 2007), few cultivars appear to combine such traits. In particular, only few cultivars are suitable for SHD orchards, where canopy volume is necessarily limited by the close spacing and the size of the harvesting machine (i.e. an over-the-row machine). Most other cultivars “escape” quickly from this small volume, and therefore require heavy pruning (Vivaldi et al., 2015). This stimulates strong vegetative regrowth, reducing fruiting (Jerie et al., 1989). In fact, cultivar response to different pruning types is an essential characteristic for olive cultivar suitability to SHD orchards (Vivaldi et al., 2015). The cultivars most used for SHD orchards are Arbequina, Arbosana and Koroneiki (Tous et al., 2014) and few more recent ones are being considered, like Chikitita (Rallo et al., 2008), Oliana (Butler, 2014) and Lecciana (Camposeo et al., 2021). These cultivars are thought to have lower vigor and to be earlier bearing and less alternate bearing than other cultivars (Moutier et al., 2004; Moutier, 2006; Tous et al., 2006; Camposeo et al., 2008; Moutier et al., 2008; Godini et al., 2011; Caruso et al., 2014; Farinelli and Tombesi, 2015; Díez et al., 2016). However, only few studies report detailed data on plant vigor, biomass partitioning, tree architecture and fruiting characteristics that distinguish such cultivars from those that are not suitable to SHD orchards. In this review, the literature available on all these aspects is summarized and interpreted.

Plant vigor depends on the cultivar (genotype) and its interaction with the environment, including agronomic management. When new orchards are established, plant density is decided based on plant vigor (Del Río et al., 2002). Dwarfing is often induced by grafting on dwarfing rootstocks (Donadio et al., 2019). Dwarf trees allow increased orchard density due to their low vigor and limited crown size (Tombesi and Farinelli, 2016). Smaller trees and higher density allow early bearing and early entrance of the orchard into full production (Koumanov and Tsareva, 2017).

In cultivated plants, the ratio of biomass invested in the harvested part over the total biomass of the plant is called harvest index (HI), and expresses the efficiency of dry matter partitioning (Donald, 1962). In tree crops, the HI is often indicated as harvest increment (HIn), which is defined as the ratio between the harvested part and the total above-ground biomass increment, over one year or a longer period (Cannell, 1985). Current fruit tree cultivars have higher HI than their wild counterparts (Patrick, 1988). HI may be 0-20% in forest trees, and up to 75-80% in cultivated species (Cannell, 1985; Fischer et al., 2012). The higher productivity of cultivated plants results from their greater HI, while their photosynthetic rates do not differ (Loomis, 1983). This implies that tree growth is driven by the competition for assimilates between vegetative organs and reproductive structures (Kramer and Kozlowski, 1979; Spurr and Barnes, 1980; Grossman and DeJong, 1995a; Grossman and DeJong, 1995b).

Reducing tree growth, by water stress (Mitchell et al., 1989), dwarfing rootstocks (Preston, 1958; Avery, 1970), containing roots with drip irrigation (Mitchell and Chalmers, 1983), pruning (Geisler and Ferree, 1984), shoot removal and/or chemical control of vegetative growth (Williams et al., 1986; Rugini and Pannelli, 1993; Mulas et al., 2011), can all increase fruit yield, suggesting competition between vegetative growth and fruit production.

In fact, fruit growth requires abundant photosynthates and trunk, branch, and, especially, root growth decreases at increasing fruit load (Lakso and Flore, 2003). In fruit trees, fruit and vegetative growths overlap for large periods, inducing strong competition for resources (Forshey and Elfving, 1989; DeJong, 1999; Wünsche and Ferguson, 2005). This competition is well documented for adult trees of many species (Stevenson and Shackel, 1998; Costes et al., 2000; Berman and DeJong, 2003), including for olive (Monselise and Goldschmidt, 1982; Rallo and Suárez, 1989; Obeso, 2002; Connor and Fereres, 2005; Lavee, 2007; Dag et al., 2010; Castillo-Llanque and Rapoport, 2011). The effects of this competition are even more dramatic in young fruit trees, where preventing fruit set greatly increases tree growth (Chandler and Heinicke, 1926; Verheij, 1972; Forshey and Elfving, 1989; Embree et al., 2007). This is the case also for young olive trees (Rosati et al., 2018a; Rosati et al., 2018b; Paoletti et al., 2021). In fact, deflowering has been proposed as an effective tool to accelerate the growth of young olive trees, thus shortening the unproductive phase in new orchards (Famiani et al., 2022).

Achieving a balance between vegetative growth and fruiting is a research priority in horticulture, given that further gains in productivity are thought to be obtainable by reducing the vegetative growth necessary to allow fruit growth (Elfving, 1988). Yet, a sufficient development in vegetative organs is necessary to intercept radiation and to absorb water and nutrients. In fruit trees, this competition has been studied mostly on plant parts, because it is easier than working at the whole-tree level. Most frequently, individual shoots have been studied (e.g. Rallo and Suárez, 1989; Acebedo et al., 2000), comparing shoot growth of fruiting and non-fruiting trees (Cimato and Fiorino, 1986; Proietti and Tombesi, 1996). Less frequently authors have considered populations of modules (Hasegawa and Takeda, 2001), or whole branches (Castillo-Llanque and Rapoport, 2011). Only a whole-tree approach, however, can provide quantitative descriptions of the competition between vegetative growth and fruit growth. Such approaches are infrequent, mostly found in older works (Verheij, 1972; Forshey and Elfving, 1989), reporting defruiting or rootstocks effects. Furthermore, even in these works, vegetative growth was compared between “off” and “on” years/trees/shoots, not allowing to provide quantitative relationships between vegetative growth and fruiting at the whole-tree level, as was done more recently in young olive trees (see below).

There is scant data on the HI for adult olive trees. In adult Arbequina trees, Villalobos et al. (2006) reported an HI of 50%, considering only the aerial parts, which corresponds to an HI of 35% if the aerial parts are assumed to be 70% of the total biomass increment.

Quantitative data on biomass partitioning in young olive trees is also scarce (Scariano et al., 2008; Di Vaio et al., 2012; Di Vaio et al., 2013). Recently, however, it has been reported that deflowering (i.e. preventing fruit set) young olive trees, strongly increased vegetative growth, and eliminated cultivar differences in vigor between Frantoio and Arbequina (high and low vigor, respectively), suggesting that differences in cultivar vigor arise from differences in biomass partitioning into fruit (Rosati et al., 2018a; Rosati et al., 2018b; Paoletti et al., 2021). This work was done at the whole-tree level and thus provided quantitative relationships between fruit production and reduction in vegetative biomass accumulation. Tree growth was inversely related to fruiting efficiency also across 12 cultivars (Rosati et al., 2017). All of this demonstrates that resource competition is a strong determinant of tree growth in young olive trees, and suggests a source limitation to tree growth (Rosati et al., 2018b).

Evidence for a source limitation to plant growth appears to contrast with previous findings that growth is, instead, sink limited (Fatichi et al., 2014; Palacio et al., 2014) and that photosynthesis is controlled by sink availability (Boussingault, 1868; Gucci et al., 1991; Gucci et al., 1995; Iglesias et al., 2002). In fact, at times photosynthesis is higher when fruit is present, suggesting sink limitation in the absence of fruit (sinks). This usually occurs when girdling or other extreme treatments are applied in the absence of fruit (Quentin et al., 2014), or with external applications of sugar (Iglesias et al., 2002; Ribeiro et al., 2017), at elevated CO₂ concentrations (Ainsworth et al., 2004), in rooted leaves without other growing organs (Sawada et al., 1986), or with continuous light (Sawada et al., 1986; Layne and Flore, 1995).

This photosynthetic down regulation is probably linked to the accumulation of photosynthesis end-products, such as starch (Paul and Foyer, 2001) or soluble sugars (Franck et al., 2006; McCormick et al., 2006), which cannot be quickly exported due to the lack of sinks (Ainsworth et al., 2004). In some cases, the photosynthetic down-regulation may be observed only when some nutrient deficiency occurs (Pieters et al., 2001; Pan et al., 2017) or only in the short term (Gucci et al., 1991; Pan et al., 2017), while in well-nourished plants, or in the long term, the plants resume sink activity and the down regulations vanishes. All of this might suggest that sugar accumulation in the leaves, due to insufficient export to limited external sinks might cause the negative feedback to photosynthesis. However, it has been recently found that this downregulation begins before sugar accumulation become substantial, suggesting that, rather than the sugar content, the signal for photosynthesis downregulation might be the change in sugar turnover (Nebauer et al., 2011). Paul and Foyer (2001) reviewed the subject and concluded that “photosynthesis responds to and is controlled by whole plant source-sink balance, controlled by whole plant nutrient balance, principally by the carbon and nitrogen status”.

In other studies, photosynthesis was not found to be affected by source-sink manipulations (Egli and Bruening, 2003; Matsuda et al., 2011). In other studies yet, photosynthesis was found to decrease, rather than increase, when fruit was present, and this was attributed to the competition for nutrient, in situation of nutrient deficiency (Zhang et al., 2013; Saa and Brown, 2014; Bote and Jan, 2016). Under water, temperature, or other stress, evidence is increasing that the stress acts directly on the sinks, reducing their sink activity (i.e. growth) before photosynthesis is affected, explaining why sugars concentrate in many stress situations (Muller et al., 2011; Fatichi et al., 2014; Palacio et al., 2014).

In addition to sink-regulation, photosynthesis can be downregulated via stomatal closure (DeJong, 1986), which, in turn, increases leaf temperature and reduces transpiration (Li et al., 2007). This occurred in defruited trees, which had many more new and more vigorous shoots, implying higher transpiration, thus justifying stomatal closure and limitation to leaf photosynthetic rates (though not of whole-canopy photosynthesis). Li et al. (2007) found similar results, additionally reporting that the stomatal limitation increased leaf temperature, further down-regulating photosynthesis. Similar conclusions were reached by Rosati et al. (2018b).

Given the evidence for both sink and source limitation to photosynthesis, it is reasonable to assume that plant growth can be both sink and/or source limited, depending on the situation (Paul and Foyer, 2001; Ainsworth et al., 2004). However, under natural conditions and in the absence of stress, photosynthesis downregulation is unlikely (Stitt, 1991).

In olive, under natural conditions leaf photosynthesis was not downregulated in the absence of fruit (Proietti, 2000; Proietti et al., 2006), although downregulation occurred with girdling, but when no fruit was present (Proietti and Tombesi, 1990). Haouari et al. (2011) reported that photosynthesis was reduced when no fruit was present, however shoot tips had been removed, artificially enhancing sink limitation. It may be concluded that, in olive, in the presence of active sinks, photosynthesis downregulation via accumulation of photosynthates is unlikely. Removing fruit may lead to some stomatal limitation to leaf photosynthetic rates when leaf area is increased due to increased source availability for vegetative growth.

Leaves are the main contributors of autotrophic activity in most plants and crops. However, to explore the environment and find sufficient light, leaves need supporting structures. In trees, these are woody structures, including shoot stems, branches, trunk, and roots. These structures are necessary, yet unproductive, thus representing a “cost”. In natural environment, with unselected trees, this cost is high because the trees need to invest more in woody structures to compete for light with other trees. In cultivated situations, the competition for light is reduced by orchard management and selected fruit tree cultivars are smaller (i.e. have lower growth rates) and invest proportionally less biomass in supporting woody structures, and proportionally more in leaves and fruit (Cannell, 1985; Patrick, 1988; Obeso, 2002). The higher productivity of modern cultivars is achieved mostly in this way (Forshey and McKee, 1970; Archbold et al., 1987), while photosynthetic abilities have not changed (Evans, 1976; Loomis, 1983). Hence, understanding the biomass partitioning into woody structures, fruit and leaves is fundamental for yield improvements.

Plant architecture is an important topic in horticulture (Hallé and Oldeman, 1970; Oldeman, 1974; Hallé et al., 1978), particularly in tree crops. Plant architecture results from the temporal and spatial arrangements of the plant parts, and results from morphological traits of shoots and branches (Lauri, 2007). The plant develops its shape through a specific growth pattern or “architectural model”, which represents its basic growth strategy. Analyzing a plant’s architecture is important in order to understand its growth, branching pattern and productivity, and to develop crop models. The architectural parameters more often studied are growth, branching, the lateral vs. apical position of reproductive structures, and the morphological differentiation of axes, (Hallé and Oldeman, 1970; Hallé et al., 1978; Barthélémy et al., 1997; Caraglio and Barthélémy, 1997; Barthélémy and Caraglio, 2007).

Both endogenous (i.e. genetic) and exogenous (i.e. environmental) factors affect the plant’s architecture (Hallé et al., 1978). Genetic factors change across cultivars within a species. In apple, for instance, where studies on tree architecture abound (Costes et al., 2006), traits such as branching density (Lespinasse and Delort, 1986; Forshey et al., 1992), branching frequency (Lauri, 2007) and flowering abundance on lateral shoots (Lauri and Lespinasse, 2001), all vary largely across cultivars, although they are also modulated by environmental variables like temperature, for instance (Cook and Jacobs, 1999; Labuschagné et al., 2003; Naor et al., 2003). In apple and other species, the proportion of short and long shoots also varies across genotypes (Lespinasse and Delort, 1986).

Rootstocks can affect the plant size and branching pattern, thus they are used to control tree size, allowing higher tree densities in modern orchards (Costes et al., 2006). Dwarfing rootstocks increase biomass partitioning into fruit and decrease it into wood, thus increasing the HI, as found in pear, apple, and cherry (Atkinson and Else, 2003). However, this strategy does not work in olive where grafting with dwarfing rootstocks has failed to induce earlier and higher yields in vigorous cultivars (Baldoni and Fontanazza, 1990; Troncoso et al., 1990; Pannelli et al., 1992; Barranco, 1997; Pannelli et al., 2002). This is probably related to the fact that, in olive, the formation of sufficiently long 1-year-old shoots is necessary for flowering and fruiting, while other fruit species, like apple (Costes et al., 2006), can produce flowers and fruit on short shoots (brachyblasts) such as bourses and spurs, allowing abundant fruiting with little vegetative growth. The olive does not have brachyblasts (Castillo-Llanque and Rapoport, 2011), and fruit set only occurs on sufficiently long 1-year-old shoots (macroblasts). Therefore, an abundant bloom (and yield) is possible only with sufficient previous-year vegetative growth. Hence, vigor reduction reduces fruiting sites and potential yield (Rosati et al., 2013). Thus, agronomical practices reducing vigor are unlikely to increase yield efficiency. In fact, using root constriction techniques, while reducing tree growth and size in olive, does not improve fruiting efficiency, leaving cultivar differences in fruiting efficiency unaltered (Paoletti et al., 2023). Therefore, in olive, while removing fruit increases vigor, reducing vigor through agronomical practices does not increase fruiting, suggesting that abundant fruiting is the cause and not the consequence of low vigor.

Since dwarfing rootstocks are unlikely to increase yield and yield efficiency in olive, the preferred way to increase orchard density has been the selection suitable cultivars. Arbequina, Arbosana and Koroneiki (Tous et al., 2014), are the most used cultivars worldwide for SHD orchards, and are considered low-vigour, early bearing, and with low alternate bearing (Moutier et al., 2004; Moutier, 2006; Tous et al., 2006; Camposeo et al., 2008; Godini et al., 2011). However, until recently, no studies reported detailed data on plant vigor and on which tree architectural characteristics might contribute to distinguish these cultivars from others, less suitable for SHD orchards. In other fruit species, such as pear and apple, plant growth and reproduction (i.e. yield) are closely related to the morphology of the axes and the axes’ position within the canopy or, in other words, to the architecture of the plant (Normand et al., 2009). In olive, the fruit distribution in fruiting shoots has been studied across different cultivars (Moutier et al., 2004). Other architectural features have been studied in non-fruiting young seedlings (Hammami et al., 2011; Hammami et al., 2021). Detailed studies on the genetic differences in olive tree architecture and their effect on plant growth and fruiting was reported in Rosati et al. (2013), where the two cultivars most used in SHD orchards, i.e. Arbequina and Arbosana, were compared to 19 other cultivars, and in subsequent work comparing Arbequina to Frantoio (Rosati et al., 2018c). The results will be discussed in the following sections.

Among the characteristics that distinguish the most used cultivars for SHD orchards is early bearing and a low alternate bearing tendency (Moutier et al., 2004; Moutier, 2006; Tous et al., 2006; Camposeo et al., 2008; Godini et al., 2011; Camposeo et al., 2022), as well as higher numbers of flowers and fruits per node (Rosati et al., 2013). However, fruit number is not the best parameter to assess the yield potential, as fruit size varies largely among cultivars (Barranco, 1999) and there is compensation between fruit number and size (Rosati et al., 2010). This compensation is determined by the different sink strength associated with different fruit size and cell number (Rosati et al., 2012; Rosati et al., 2020). Varietal differences in fruits size are already pre-determined at bloom in the form of ovary size differences (Rosati et al., 2009; Rosati et al., 2012), therefore the compensation between size and number must occur early, and affects both pistil abortion (Rosati et al., 2011) and fruit set (Rosati et al., 2010). Cultivars suitable for SHD orchards are also characterized by higher branching and lower wood to leaf biomass ratio, both at the shoot and the whole-tree level, as discussed in section 5. In fact, when comparing Arbequina and Arbosana to many other cultivars not suitable for SHD orchards, Rosati et al. (2013) found that some cultivars (e.g. Rosciola, Maurino) had flowering and fruiting parameters similar to them at the single shoot level, but not accompanied by high branching and smaller diameters of woody structures. On the other hand, other cultivars had similar branching qualities (e.g. Piantone di Mogliano, Piantone di Falerone), but insufficient fruiting. Only when both architectural and fruiting features were combined together, Arbequina and Arbosana were segregated from all other cultivars, suggesting that both features are required to achieve high yields in small canopy volumes, and thus suitability to SHD systems.

Olive tree growth appears to be mostly source limited and vegetative growth competes with fruit production. Therefore, yield and growth depend on partitioning of resource toward these two sinks. Defruiting makes trees grow more, while fruit presence slows down growth and produces shorter shoots. Short shoots have lower wood to leaf biomass ratio and thus greater and earlier ability to export carbon, which, in turn, supports fruit growth. However, in most cultivars, the short shoots formed during a year of abundant production will not flower and set fruit profusely the following year, setting off the alternate bearing cycle. Some cultivars, i.e. those most suitable for SHD systems, bear more fruit per node, thus allowing fruiting also on the short shoots formed in an “on” year. This reduces alternate bearing and increases fruiting efficiency by constantly producing shorter shoots with lower wood to leaf biomass ratio. However, these cultivars also have higher branching and thinner woody structures, both features increasing the leaf area and the number of fruiting sites per unit of wood biomass and per unit of canopy volume. Increased leaf area increases radiation capture and thus resource availability. Increasing the number of fruiting sites per unit of wood biomass saves resource investments in non-productive sinks (roots, trunk and branches), thus liberating resources which can be exported for fruit set and growth, increasing fruiting efficiency (resource partitioning towards fruit). Increased resource availability for fruit (from increased leaf area and reduced partitioning into woody structures) might in fact explain, at least in part, the greater flowering and fruit set ability of such cultivars.

All of this suggests that possessing the right architectural characteristics is essential for cultivar breeding and selection for higher yield and yield efficiency, in intensive and, particularly, in super intensive systems where canopy volume is necessarily limited. Under these conditions, in fact, low branching not only reduces directly resource availability for fruit, but it also induces rapid growth outside the allowed volume, requiring intense pruning, which, in turn, reduces fruiting sites and thus fruiting. With lower fruiting, vigor increases further, requiring even more pruning, triggering a vicious circle of producing more vegetation and lower yields (i.e. low fruiting efficiency). High branching, instead, directly reduces canopy growth in terms of volume, but also liberates resources thus increasing fruiting, which, in turn, controls vigor further. With reduced vigor, less pruning is required, allowing greater fruiting, triggering a virtuous circle. These concepts are visualized and summarized in Figure 2 .

AR: Conceptualization, Funding acquisition, Investigation, Supervision, Visualization, Writing – original draft, Writing – review & editing. AP: Investigation, Writing – review & editing. EL: Investigation, Writing – review & editing. FF: Conceptualization, Investigation, Supervision, Writing – original draft, Writing – review & editing.