Adventitious root formation is a complex process regulated by a number of endogenous and exogenous factors. Enzymes regulating auxin metabolism re­present a key factor that affects adventitious root formation in almost all phases of rhizogenesis (Hartmann et al., 2001). The enzymes that appear most often to be involved in the formation of adventitious roots are (Kavrayan & Aydemir, 2001; Qaddoury & Amssa, 2003; Porfirio et al., 2016a): peroxidase (POD; donor: H ₂O ₂ oxidoreductase, EC 1.11.1.7), polyphenoloxidase (PPO; monophenol, dihydroxy-L-phenylalanine oxy­gen oxidoreductase, EC 1.14.18.1) and indole-3-acetic acid oxidase (IAAox).

Peroxidase activity has been suggested to serve as a good indicator of the rooting potential (Rout et al., 2000; Husen, 2012). The initial reports of Van Hoof & Gaspar (1976) related rhizogenesis with a reduction in POD activity. From then on, several researchers described that adventitious rooting in various species occurs after POD activity peaked and followed by a subsequent decrease (Metaxas et al., 2004; Vatulescu et al., 2004; Husen & Pal, 2007). However, the opposite trend has also been observed in some species such as Castanea sativa × C. crenata (Gonçalves et al., 1998), disputing in part POD's role as a biochemical marker (De Klerk, 1996).

Polyphenoloxidase is another enzyme involved in the initiation and development of adventitious roots (Kose et al., 2011). According to Yilmaz et al. (2003), PPO may play a role in the organisation and development of root primordia, as it takes part in cell division and differentiation and it can directly regulate the synthesis of phenolics. Furthermore, Cheniany et al. (2010) reported that the changes in PPO activity appeared to be greater in easy-to-root cultivars and suggested the use of PPO as a marker of the onset and duration of the different phases of rooting.

A close relationship between IAAox activity and adventitious rooting has also been reported (Güneş, 2000), as this enzyme is involved in the regulation of IAA levels. According to Vatulescu et al. (2004), the reduction of IAAox activity could inhibit rooting when the endogenous level of IAA is supra-optimal, while a sub-optimal concentration usually promotes rooting. There is also a close relationship between POD, PPO and IAAox (Shinshi & Noguchi, 1975). It is generally accepted that some POD isoenzymes possess IAAox activity, as well as PPO activity (Srivastava & van Huystee, 1977). It is believed that PPO does not influence root formation directly, but its effects rather occur through a disturbance in POD activity, revealing a possible inverse relationship between these two enzymes (Cheniany et al., 2010; Porfirio et al., 2016a).

There is a wide variation in rooting ability of semi-hardwood cuttings among olive cultivars (Fontanazza, 1993). The exact reason behind the difficulty of rooting of an olive cultivar is unknown and it is believed that in some cases the differences in the rooting capacity may be due to the anatomical structure of the cuttings. The presence of a continuous sclerenchyma ring as well as the thick cortex in stem cuttings' base may act as a physiological barrier to adventitious root initiation or as a mechanical barrier to the emergence of the newly formed roots and has been negatively correlated with the rooting ability (Ayoub & Qrunfleh, 2008; Amissah et al., 2008). Furthermore, the success of the rooting process has been partially associated with the location of the potential root initiation sites as well as the ability of target cells to be de-differentiated (Avidan & Lavee, 1978; Amissah et al., 2008).

The objectives of this study were: a) to obtain information over the role of the enzymes POD, PPO and IAAox during the different phases of adventitious rooting process in an easy and a difficult-to-root olive cultivar, and b) to examine how their changes are related with differences on adventitious rooting ability. Additional objectives included the identification of the sites of root primordia initiation and the anatomical differences between 'Arbequina' and 'Kalamata' stems. Attention was also given to whether these anatomical differences could be interpreted as a barrier to the initiation and emergence of adventitious roots.

The presence of high concentrations of auxin during the de-differentiation of the cells is of utmost importance for cuttings' rooting (Sagee et al., 1992; Gaspar et al., 1997). Nonetheless, in the later stages of growth and development of root primordia, high concentrations of auxin may be inhibitory thus its reduction is necessary (Gaspar et al., 1997). Therefore certain enzymes (especially POD, PPO and IAAox) can contribute directly or indirectly to the regulation of auxin concentration during Ar (Hartmann et al., 2001). However, there is very little information so far regarding the role of oxidative enzymes in adventitious rooting in olive. Macedo et al. (2013) and Porfirio et al. (2016b) have investigated and related the role of PPO and POD activities mainly in olive microcuttings, while Aslmoshtaghi & Shahsavar (2016) are among the few researchers who studied the enzyme activities in olive semi-harwood cuttings.

Pretreatment with auxin improved rooting in 'Ar­bequina' (76% in summer, 40% in autumn and 37% in spring), while the difficult-to-root 'Kalamata' presented only 3% rooting (Fig. 1). The very low rooting ability of 'Kalamata' is in agreement with the results of Avidan & Lavee (1978) and Wiesman & Lavee (1995), and this was the reason why this cultivar was characterised by the researchers as a difficult-to-root. The present results showed that the auxin treatment was unable to stimulate root initiation even a massive callus was formed in the majority of 'Kalamata' cuttings. The failure of auxin to stimulate rooting in 'Kalamata' suggests that other physiological and/or biochemical factors and/or the different anatomical structure might also be involved in this process.

Many biochemical studies showed a close relationship between PPO, POD, IAAox and adventitious rooting (Ríos et al., 1997; Tchinda et al., 2013). In the present study, the difficult-to-root 'Kalamata' cuttings exhibited generally higher enzymatic activities at day 0 and during the rhizogenesis phases compared to 'Arbequina' (Figs. 2, 3 and 4). It is noteworthy the fact that PPO and POD activities were, in average, two times higher in 'Kalamata' compared to 'Arbequina', while for the same cultivar IAAox activity presented even an eightfold increase during the initiation phase (3-5 DAP). Ludwig-Müller (2003) also found that the activity of the above enzymes was higher in difficult-to-root species compared to easy-to-root ones. Therefore, the particular high activity of the above enzymes may inhibit the rooting of 'Kalamata' cuttings and could partly justify its recalcitrancy to root. Bansal & Nanda (1981) came to a similar conclusion about the role of these enzymes in the rooting capacity of various forest species.

It is noteworthy that in both cultivars a similar pattern of enzymatic activities was observed between control and IBA treated cuttings (Figs. 2-4). These observations are in agreement with Tchinda et al. (2013), as well as with Qaddoury & Amssa (2004), who found that in the control cuttings POD and IAAox activities followed the same pattern as in IBA treated. Data also showed that control cuttings exhibited higher enzymatic activities, but in some cases the detected differences were not statistically significant. Τhe present results appear to contradict the findings of other researchers, where IBA treated leafy cuttings of Ricinodendron heudelotii (Tchinda et al., 2013) as well as date palm offshoots (Qaddoury & Amssa, 2004) exhibited higher enzymatic activities. Generally, PPO, POD and IAAox presented increased activities in both cultivars during spring and summer, while in autumn the enzymatic activities decreased (Figs. 2-4). However, it is difficult to establish any relation between the above observed seasonal enzymatic activities and rooting in both cultivars.

PPO, POD, IAAox activity underwent significant changes during the rooting period (RP) (Figs. 2-4). Similar changes in the enzymatic activities have been reported for various plant species such as almond (Caboni et al., 1997), poplar (Güneş, 2000) and vine cuttings (Kose et al., 2011). The determination of POD activity in the basal region of 'Kalamata' cuttings during RP showed that the enzyme activity peaked within 3-5 DAP (Fig. 3A-C). The increase in this enzyme activity was relatively high compared to 'Arbequina' and since peroxidases are known to be involved in IAA catabolism (Caboni et al., 1997; Gaspar et al., 1997), it seems that POD negatively affected the auxin balance at meristematic regions, that corroborate with previous reports made in different species ( Arbutus unedo, Taxus baccata and peach rootstock GF-677) (Metaxas et al., 2004; Molassiotis et al., 2004).

It is well known that POD activity has a fundamental role in root initiation and changes in the activity of this enzyme have been used as a predictive marker of the rooting process (Hartmann et al., 2001; Metaxas et al., 2004). In 'Arbequina', POD activity increased significantly 15 DAP in all experimental seasons, as was also observed in 'Kalamata' cuttings (Fig. 3A-C). An increase in POD activity 15 DAP has been also described in other species like date palm ( Phoenix dactylifera) (Qaddoury & Amssa, 2003) and apple rootstock MM106 (Naija et al., 2008). According to Porfirio et al. (2016b) this increase coincides with the expression phase in olive, confirming the well-known function of POD in cell wall formation and modification (Molassiotis et al., 2004; Konieczny et al., 2014). This increase in POD activity in the present study corresponded to the observed histological changes leading to callus or root formation in 'Arbequina' cuttings (Fig. 6D) as was also observed in olive microcuttings of the 'Galega vulgar' (Macedo et al., 2013). Nonetheless, this POD function seems to occur only in 'Arbequina' as in 'Kalamata' cuttings no root formation was observed.

IAAox activity in 'Kalamata' exhibited the same pattern of increase as POD with maximum enzyme activity 3 DAP in summer and 5 DAP in spring and autumn, followed by a subsequent decrease (Fig. 4A and C). On the other hand, IAAox in the easy-to-root 'Arbequina' decreased during the induction phase (from day 0 to day 1) (Fig. 4). These results are in agreement with those of Husen (2012) and Wiesman et al. (1988), who pointed out that the low IAAox activity during the induction phase seems to be responsible for the better adventitious rooting by preserving the source of free auxin. According to Mato & Vieitez (1986) and Basak et al. (2000), it appears that low IAAox activity favors rooting in Castanea sativa and other species, as was also observed in 'Arbequina' cuttings. Generally, the low enzyme activity in 'Arbequina' compared to 'Kalamata', is likely to regulate endogenous auxin concentration to appropriate levels, in order to trigger the rooting initiation phase and subsequently the rooting expression phase.

Polyphenoloxidase is another enzyme whose activity is in close relationship with root formation, as it may oxidize auxin during growth and development of root primordia (Qaddoury & Amssa, 2003; Tchinda et al., 2013). The results of this study showed that PPO activity of 'Kalamata' cuttings reached its maximum level between 3 and 5 DAP (Fig. 2). The increase in this enzyme activity coincided with the maximum POD and IAAox activities observed for 'Kalamata' cuttings (Figs. 3 and 4). Therefore, PPO may act synergistically to IAAox and POD during the initiation of rhizogenesis (3 to 5 DAP). In 'Arbequina', PPO activity generally increased 15 DAP during all experimental seasons (Fig. 2A-C). In the following days, callus or root formation was observed in the stem of the 'Arbequina' cuttings (Fig. 6D). Similar changes have been reported in vine and Ricinodendron heudelotii (Yilmaz et al., 2003; Kose et al., 2011; Tchinda et al., 2013). Consequently, PPO activity may play a significant role in formation of root primordia in 'Arbequina' cuttings as has been also found in walnut by Cheniany et al. (2010). Nevertheless Aslmoshtaghi & Shahsavar (2016) found no clear relationship between PPO activity and root initiation in olive cuttings.

The anatomical observations of this study showed that cells divisions occurred mainly close to the basal end of the cutting surface, as was previously observed by other researchers (Jasik & De Klerk, 1997; Hartmann et al., 2001). In 'Kalamata' cuttings the cell differentiation occurred in the cortex outside the sclerenchyma ring from cells of the photosynthetic parenchyma, resulting in a swelling of the basal part of the cutting and contributing to callus formation, even excessive, which did not lead to root primordia (Fig. 7A-B). These results confirm that callus formation is a characteristic of the difficult-to-root olive cultivars, as has been also reported by Ayoub & Qrunflesh (2008) for the olive cv 'Nabali'. On the other hand, most of the 'Arbequina' cuttings rooted without callus formation while few formed callus (Fig. 6D). These results confirm firstly that in the easy-to-root cultivar, root formation may occur without callus formation and secondly that callusing and adventitious rooting are independent, even though both involve cell division (Hartmann et al., 2001).

The origin of adventitious roots on stem cuttings has been reported to be located in various tissues and it varies from one species to another (Naija et al., 2008; Agullό-Antόn et al., 2014). In the present study the origin of root initials in both cultivars was the cambial zone (Figs. 6A-C and 8A-B), as was previously observed in other olive cuttings of cvs. 'Wetaken' (Bakr et al., 1977), 'Manzanillo', 'Mission' and 'Hamed' (Salama et al., 1987).

Several anatomical studies have suggested a cor­relation between difficulty in rooting and the presence of a thick cortex and a pericyclic sclerenchyma layer (Avidan & Lavee, 1978; Amissah et al., 2008; Porfirio et al., 2016a). In the present study, in the cuttings of the difficult-to-root 'Kalamata', a noticeable thicker cortex and a continuous sclerenchyma ring were observed in comparison with the easy-to-root 'Arbequina' (Fig. 5A-D). The above two anatomical barriers may reduce the absorption and movement of exogenous applied auxin in the inner layers of 'Kalamata' cuttings, resulting in a weak auxin stimulus, not enough to induce de-differentiation of the cambial cells into root initials. This speculation was further supported by the fact that in 'Kalamata' cuttings 15 DAP, a cell differentiation occurred in the cortex outside the sclerenchyma ring resulting in callus formation (Fig. 7A-B), while the cambial cells remained inactivated, as if the auxin stimulus never reached the cambial cells and only triggered the outer layers.

However, further anatomical observations during the rooting period at 30 DAP revealed that, once the cambial cells de-differentiated and the root initials formed in 'Kalamata', the continuous sclerenchyma ring did not constitute an impenetrable barrier as it was easily penetrated by the pressure of the growing primordium (Fig. 8). The present findings support Bakr et al. (1977), Avidan & Lavee (1978) and Fabbri (1980) observations, according to whom the difficulty in adventitious rooting of olive cuttings was not related to the anatomical structure of the cutting. Thus, it can be suggested that the inability of 'Kalamata' cuttings to form root primordia is not due to the presence of sclerenchyma ring acting as a physical barrier to root emergence but a barrier that difficults exogenous applied auxins to reach the sites where root primordia are supposed to be induced and further differentiated.

Based on the above, it can be concluded that the adventitious rooting ability of 'Arbequina' and 'Kalamata' is affected by the combined interaction of auxin, enzymatic activities and the anatomical structure, rather than by a single factor alone. In fact, in 'Kalamata' the high enzymatic activities determined during the rhizogenesis period combined with this cultivar's tendency to form a massive callus, without redifferentiation to root primordia, is a more likely reason for the poor rooting percentages.