Introduction

The macadamia (Macadamia integrifolia Maiden & Betche) is a nut tree belonging to the Proteaceae family, native to tropical and subtropical Australian forests (^{Peace et al., 2005}). It was introduced into Brazil in 1931 (^{Dierberger & Marino Netto, 1985}) and in 1940 the Agronomic Institute began the only macadamia breeding program in the country (^{Sobierajski et al., 2006}). Orchards are generally established by using grafted trees to anticipate yield, maintain cultivar characteristics, and increase the homogeneity of nut quality (^{Melo et al., 2019}). The technique usually used for vegetative propagation is grafting (^{Campo-Dall'Orto et al., 1988}). It takes 18 months for the seedling to be ready for planting in the orchard (^{Russell et al., 2016}). In previous studies, the rate of grafting success reached 80% (^{Campo-Dall'Orto et al., 1983}). However, this rate is lower than 50% in commercial nurseries. These factors increase the cost of seedlings and reduce their availability to growers making it difficult to expand the culture in Brazil (^{Melo et al., 2019}).

The macadamia tree produces a nut of great acceptance and high value in the international market (^{Penoni et al., 2011}; ^{Maro et al., 2012}). In recent years, there has been an increase in the demand for seedlings of macadamia in Brazil (^{Perdoná & Soratto 2015}; ²⁰¹⁶).

The propagation by semi-hardwood cuttings can reduce the time and costs of seedling production (^{Bell, 1993}). Therefore, establishing an effective and affordable protocol of seedling production by semi-hardwood cuttings will provide an advance for macadamia in Brazil, stimulating orchard expansion.

^{Pereira et al. (1987)} studied the rate of rooting in cuttings of five cultivars from two seasons. The authors observed significant differences for cultivars and seasons but no interactions between them. The emission of roots in cuttings of macadamia is the major difficulty for vegetative propagation. Therefore, the use of synthetic growth regulators, such as auxin can be an alternative to overcoming this problem (^{Garbelini, 2009}). Another option is the use of plant growth regulators, e.g., indole butyric acid (IBA) that, when compared to a synthetic auxin, shows lower sensitivity to biological degradation (^{Fachinello et al., 2005}). These plant growth regulators also stimulate root growth (^{Alvarenga, 1990}). However, the ideal IBA concentration for cutting immersion varies among different species and cultivars (^{Hartmann et al., 2002}), and it is still unknown for macadamia. On such a background, the aim of this study was to identify the macadamia cultivars with the best rooting performance of semi-hardwood cuttings and test different IBA concentrations and seasons for seedling production by semi-hardwood cuttings.

Materials and methods

Six macadamia cultivars were propagated by semi-hardwood cuttings. The percentage of survival, callus formation, and rooting were evaluated. The cultivars were selected for their characteristics of yield and nut quality. The cultivars HAES 344, HAES 816, and HAES 660 were developed by the Hawaii Agricultural Experiment Station (HAES) and the Agronomic Institute (IAC) developed the cultivars IAC 4-12B, IAC 9-20, and IAC 4-20.

The semi-hardwood cuttings were collected from 12-year-old trees in Dois Córregos, São Paulo, Brazil, for four months (October and November 2018, and February and March 2019). The materials were cut from the medium treetop with 15-20 cm of length and 3-5 cm of diameter; three leaves were kept for all cuttings, as recommended by ^{Russell et al. (2016)}. The semi-hardwood cuttings were soaked in IBA (0, 5, or 10 g L^-1) (NEON^®, Suzano, Brazil) for 15 min, followed by planting in a propagation tube (120 ml) filled with substrate (peat-vermiculite-limestone, Carolina Soil^®, Pardinho, Brazil). In addition, the tips of cuttings collected in November, February and March were also soaked in Clonex^® (IBA 3 g L^-1, for 15 min) (Growth Technology, Taunton, UK). To prevent the cuttings from drying out, they were planted on the same day of collection in a mist-house.

The humidity in the mist-house was maintained using a controller (Fascitec NTI 12 - AC, São Bernardo do Campo, Brazil) for intermittent watering and to avoid accumulation of water on the leaves. The experimental design was a two-way completely randomized design with four replicates and five cuttings by plot/month, of which 20 cuttings were the control treatment (distilled water). Thus, 1,800 cuttings were evaluated.

After 120 d, the semi-hardwood cuttings were removed from the tubes, washed, and evaluated for survival, callus, and root development. The data were transformed by the ^{BoxCox method (Box & Cox, 1964)} to meet the normality assumption. The data were analyzed using the R software (^{R Development Core Team, 2019}), and the cluster analyses and threshold were calculated using the R-package 'facto-extra' (^{Kassambara & Mundt, 2017}).

Results and discussion

From the 1,800 semi-hardwood cuttings, 9.8% showed callus, 5.7% developed roots, and 84.5% did not survive. This high mortality caused the data to be unbalanced, which prevented the use of an analysis of variance. The statistical analysis was performed by descriptive and multivariate analysis.

The best season for removing semi-hardwood cuttings for the presence of callus was October 2018 (22.8%), considering all cultivars and doses of IBA and Clonex^® (Fig. 1A). However, the best results for root development were observed for those plants collected in November 2018 (10.0%; Fig. IB). The lowest semi-hardwood cutting mortality was observed when the removals occurred in October 2018 (71.9%), while the highest rates occurred in February 2019 (90.2%) and March 2019 (92.3%; Fig. 1C).

FIGURE 1 Similarity matrix (Euclidean distance) among collection seasons of semi-hardwood cuttings of Macadamia integrifolia, considering all cultivars and doses of IBA and Clonex^®. The similarity matrix considered the following parameters: A) presence of callus, B) root development, and C) mortality. The grouping method (cluster analyses) was based on the K-means algorithm (^{Kassambara & Mundt, 2017}). 

The control treatment (without IBA application) showed the highest rate for the presence of callus (42.4%), considering all cultivars and collecting seasons (Fig. 2A). However, this treatment showed the lowest rooting rate (11.8%). The other treatments showed no statistical differences for root development with 31.4% for 10 g L^-1 IBA, 29.4% for Clonex^®, and 27.4% for 5 g L^-1 IBA (Fig. 2B). The mortality was statistically lower in semi-hardwood cuttings treated with Clonex^® (19.5%) than with the other doses of IBA (Fig. 2C).

FIGURE 2 Similarity matrix (Euclidean distance) among doses of indole butyric acid (IBA) and Clonex^® for semi-hardwood cuttings of Macadamia integrifolia treatment, considering all cultivars and collecting seasons. The similarity matrix considered the following parameters: A) presence of callus, B) root development, and C) mortality. The grouping method (cluster analyses) was based on the K-means algorithm (^{Kassambara & Mundt, 2017}). 

The cultivar HAES 660 showed the highest presence of callus (35.6%) considering all seasons of collection and doses of IBA and Clonex^® (Fig. 3A). However, only 12.0% of the semi-hardwood cuttings of this cultivar effectively developed roots (Fig. 3B). Regarding root development, the cultivar IAC 4-12B showed the highest rate (37.0%), followed by the second group (formed by TAC 9-20' and TAC 4-20') that showed the second-highest performance (Fig. 3B). The mortality rates by cultivar were similar to each other, but the cultivars HAES 660 (14.8%) and IAC 4-12B (15.6%) showed the lowest values (Fig. 3C).

FIGURE 3 Similarity matrix (Euclidean distance) among cultivars of Ma-cadamia integrifolia 'IAC 4-12B', 'IAC 9-20', 'IAC 4-20', 'HAES 344', 'HAES 660' and 'HAES 816', considering all collecting seasons and doses of IBA and Clonex^®. The similarity matrix considered the following parameters: A) presence of callus, B) root development, and C) mortality. The grouping method (cluster analyses) was based on the K-means algorithm (^{Kassambara & Mundt, 2017}). 

The descriptive analysis regarding the cultivars in all IBA doses and cutting seasons showed 'HAES 660' as the best cultivar regarding the presence of callus (60%), also exhibiting the lowest mortality (40%). These rates were observed for those plants collected in October 2018 and subjected to the control treatment (Tab. 1). The callus is formed by new meristematic cells near the phloem (^{Hartmann et al, 2002}). The presence of calluses in species of hard rooting can precede root emission (^{Bitencourt et al., 2010}). However, the development of callus and roots can be regarded as independent events. For instance, ^{Singh and Ansari (2014)} observe high rates for callus development in Dalbergia latifolia (91.5%) and Gmelina arborea (75.0%) in the control treatment, without these calluses evolving into roots (3.1 and 0%, respectively). The authors suggest that the presence of callus in treatments without hormones is a reaction of the plant to restore the injured cutting and not a prerequisite for root induction. These two events may simultaneously occur, because both processes involve cell division and depend on favorable conditions (^{Preti et al, 2012}).

TABLE 1 Percentage of semi-hardwood cuttings with presence of callus (Callus), root development (Root) and mortality (Mort.) in macadamia [Macadamia integrifolia) cultivars collected in four seasons and treated with indole butyric acid (IBA; g L^-1) and Clonex^®. 

Clonex^® - commercial product; NA - not evaluated.

Regarding root development, TAC 4-20' showed the highest rate of rooting (30%) when collected in November 2018 and subjected to treatment with IBA at a dose of 5 g L^-1 (Tab. 1). For this variable, all cultivars developed by the Agronomic Institute showed higher results than those found for the Hawaiian cultivars. Cultivar IAC 4-12B was the most stable with 20,20,15 and 25% of rooting in November 2018 (IBA doses 0, 5,10 g L^-1, and Clonex^®, respectively; Tab. 1). This cultivar also showed high values of rooting in the February collection season (20% - Clonex^®) and March 2019 (25% for IBA dose of 10 g L^-1 and 20% for Clonex®; Tab. 1). Cultivar IAC 9-20 also showed a high rooting rate (20%) in the October 2018 collection season in semi-hardwood cuttings treated with 10 g L^-1 of IBA (Tab. 1). ^{Russell et al. (2016)} observe a rooting rate of 44% in M. integrifolia cultivars. ^{Entelmann et al. (2014)} obtain rates of 22.9% of rooting when studying the cultivar Aloha 10-14 with an IBA concentration of 3 g L^-1.

According to our data, the capability of the cultivars for rootstocks for the presence of callus, rooting, and survival were classified as nine distinct groups by the cluster analyses (Fig. 4). Cluster 3 was formed by TAC 4-20' (IBA at a dose of 5 g L^-1), TAC 4-12B' (IBA at a dose of 5 g L^-1) and TAC 4-20' (control treatment). These cultivars showed the highest rooting rates and moderate presence of callus. Cluster 7 was formed by the cultivars with lowest mortality rates (TAC 9-20' - control treatment and 'HAES 660' - IBA at doses of 5 and 10 g L^-1). The cultivars grouped into cluster 8 ('HAES 344' - control treatment and IBA at a dose of 5 g L-¹, and 'HAES 660' - control treatment) showed a high rate of presence of callus; however, these cultivars did not develop roots. These clusters are in agreement with the dendrogram obtained by the similarity matrix (Euclidean distance) (Fig. 5).

FIGURE 4 Grouping by multivariate features (cultivars, indole butyric acid, and collecting seasons) of semi-hardwood cuttings of Macadamia integrifolia. Cultivars (dose of IBA - g L^-1): 1 - HAES 344 (0), 2 - HAES 344 (5), 3 - HAES 344 (10), 4 - HAES 344 (Clonex^®), 5 - HAES 660 (0), 6 - HAES 660 (5), 7 - HAES 660 (10), 8 - HAES 660 (Clonex^®), 9 - HAES 816 (0), 10 - HAES 816 (5), 11 - HAES 816 (10), 12 - HAES 816 (Clonex^®), 13 - IAC 4-12 B (0), 14 - IAC 4-12 B (5), 15 - IAC 4-12 B (10), 16 - IAC 4-12 B (Clonex^®), 17 - IAC 4-20 (0), 18 - IAC 4-20 (5), 19 - IAC 4-20 (10), 20 - IAC 4-20 (Clonex^®), 21 - IAC 9-20 (0), 22 - IAC 9-20 (5), 23 - IAC 9-20 (10), and 24 - IAC 9-20 (Clonex^®). The threshold was obtained by the K-means algorithm (^{Kassambara & Mundt, 2017}). 

FIGURE 5 Similarity matrix (Euclidean distance) among Macadamia integrifolia cultivars HAES 816, HAES 660, HAES 344, IAC 9-20, IAC 4-20, and IAC 4-12 B subjected to treatment with indole butyric acid (0 g L^-1, 5 g L^-1 and 10 g L^-1) and Clonex^® and collected in different seasons. The threshold was obtained by the K-means algorithm (^{Kassambara & Mundt, 2017}). 

The high mortality rates observed in this experiment can be related to diseases caused by high environmental moisture, probably due to the identified presence of Cladosporium sp. on the leaves of the semi-hardwood cuttings. High mortality rates (46.2%) were also observed by ^{Russell et al. (2016)} evaluating the rooting ability of 32 macadamia cultivars in Australia. These authors found fungi from Nectria spp., seriously affecting M. integrifolia.

Conclusions

Considering all the cultivars evaluated in this study, the macadamia cultivar IAC 4-12B showed the highest rooting. This cultivar, along with HAES 660, showed the lowest mortality rate. Therefore, it can be concluded that IAC 4-12B is an interesting option for macadamia producers.

Although there were no statistical differences between indole butyric acid and Clonex^® for root development, this last treatment showed the lowest mortality.

Under the conditions of Brazil, November showed the highest rooting values for semi-hardwood cuttings, and October showed the highest values for the presence of callus and low mortality for semi-hardwood cuttings.