Statistical analysis

With the data, ANOVA was performed, regression, and the measurements were compared with the SNKα test (p≤0.05), using SISVAR 5.3 (^{Ferreira, 2010}), because it uses the distribution of the Tukey test. The Keuls method is relatively easy to apply and considers the total number of treatments used in the experiment, and is a multiple amplitude test that evaluates the significance of a set of differences, making a balance between Errors of Type I and II.

Rooted cuttings, presence of calluses and mortality

The percentages of rooting (from 8 % to 58 %) were relatively low; however, the types of cuttings (Pr>F 0.002), IBA concentrations (Pr>F 0.002), immersion times (Pr>F 0.026) and the interactions between these factors (Pr>F 0.000) showed significant differences (p<0.05) (Tables 1 and 2). The positive interaction with the same factors was verified by ^{Dias et al. (2012a)} with vegetative propagation for mini cuttings of Anadenanther amacrocarpa and different IBA concentrations.

Table 1 Effect of the concentration and immersion time in indolebutyric acid (IBA) and the position of the cutting with the percentage of cuttings rooted (EE), callus formation (C), mortality (M) and the mean values of the number of roots (NR) and root length (LR) in Bertholletia excelsa. 

Values with different letters in a column are statistically different (p≤0.05).

Table 2 Effect of the concentration and immersion time in indolebutyric acid (IBA) and the position of the cutting, on the mean values of the root number (NR) and root length (LR) in Bertholletia excelsa juvenile cuttings. 

The mean values with a different letter in a column are different (SNK test; p≤0.05); † Data transformed to x =0 5; NR (dp±2.93); CR (dp±2.74).

^{Wendling and Xavier (2005)} obtained 80-100 % of rooting when using concentrations of 500 to 1000 mg L^-1 of IBA in cuttings of Eucalyptus clones. According to ^{Santos et al. (2011)}, in addition to the effect of the growth regulator, the specific characteristics of each species also affect the rooting of cuttings, as was observed in their study with 20 forest species. These results highlight that the rooting of cuttings presented differentiated behavior and that for the start of rooting, intrinsic factors related to the plant itself influenced as well as extrinsic ones linked to environmental conditions (^{Hartmann et al., 2002}; ^{Pio et al., 2003}).

The low percentage of root formation is explained because the cuttings responded differently to the action of the hormone; some need to remain in contact with the growth regulating solution for longer and others root sporadically or they do not root. This difference is due to the physiological variations between cuttings from the same plant, although they may be juvenile; in addition, the process can be inhibited based on the concentration of the growth regulator. These results suggest that juvenile plant material is not always favorable for the rooting process of woody species from seeds, as found with B. excelsa cuttings. ^{Fachinello et al. (2005)} point out that despite the capacity for rooting of woody species, the potential of a cutting to form roots varies with the species and type of crop due to the interaction of factors such as manipulation of the plant, good nutritional status, period of recollection and age of the plant, length and diameter of cuttings, presence of leaves and buds, hormonal treatment, illumination, temperature and relative moisture.

The use of IBA increased the number of roots per cutting, and promotes a greater percentage of rooting according to ^{Endres et al. (2007)}. The same was observed by ^{Gratieri-Sossella et al. (2008)}, in studies with Brazilwood (Caesalpinia echinata) and ceibo (Erythrina crista-galli), respectively.

In absence of IBA the formation of calluses varied from 66 to 100 %, especially in the apical position (Table 1; Figure 1). The presence of calluses in the sections of cuttings occurred independently from the presence of the growth regulator or position of the cut. ^{Hernández et al. (2013}), with regard to the vegetative propagation with cuttings of Cariniana estrellensis, report that the presence of callus is produced with or without the presence of IBA, both in apical and in intermediate cuttings.

Figure 1 Callus formation on the base of the Bertholletia excelsa cutting. 

^{Fachinello et al. (2005)} point out that the callus is formed from a mass of parenchyma and disorganized cells, as a result of the lesions of the tissues of the xylem and phloem during the preparation of the cutting. The natural auxin produced in the leaves and buds moves towards the inferior part of the cuttings stimulating the rooting activity and accumulating in the base of the cutting, together with sugars and other nutrients (^{Janick, 1996}), acting as inhibiting substances in the initiation of the preformed adventitious roots and potentiating the formation of calluses at the base of the cuttings (^{Hartmann and Kester, 2011}).

^{Iritani et al. (1986)} and ^{Caldwell et al. (1988)} affirm that the roots can appear directly from caulinar tissue or solely through the callus. The formation of calluses must be due to the presence of equilibrium of the auxin applied (IBA) in the presence of an endogenous cytokinin and, according to ^{Hartmann and Kester (2011)}, this equilibrium can also occur when cytokinin is applied with the endogenous auxin. Thus, the presence of calluses in B. excelsa cuttings would indicate that if the cuttings remain for a longer period in the rooting environment, there would be an increase in the percentage of rooted cuttings. However, the roots that arise from differentiation of callus cells rarely have vascular connections with the cuttings, so propagation from these is not recommended (^{George et al., 2008}).

With regard to mortality, the cuttings in the basal position, immersed 1 s with 3000 mg L^-1 of IBA presented higher percentage (83.3 %) in comparison to other types of position and immersion time, suggesting that there was phytotoxicity at that concentration. Among treatments, only the apical cuttings in absence of IBA did not present mortality (Table 1). ^{Bastos et al. (2004)} studied the rooting of cuttings from starfruit (Averrhoa carambola) and suggest that there was phytotoxicity under different IBA concentrations and immersed 10 s, and that the survival capacity of the apical cuttings in absence of the growth regulator was higher than those exposed to IBA.

Number and length of roots

The cuttings presented, in average, the formation of only one root (Figure 2). The treatment that caused a higher number of roots was the one with concentration of 3000 mg L^-1 of IBA (Table 1).

Figure 2 Aspects of roots on the Bertholletia excelsa cuttings at 180 days. 

The middle cuttings submerged 1 s at 1000 mg L^-1 of IBA showed a higher percentage of rooting (58.3 %), followed by the basal cuttings at the same time of immersion and the same concentration (41.7 %) (Table 1). However, the latter did not differ from the middle and basal cuttings, immersed 1 min in 3000 mg L^-1 of IBA. In absence of IBA, for the three types of cuttings used, there was no root formation, as well as in the apical cuts submerged for 1 s in a concentration of 3000 mg L^-1 of IBA and immersed for 1 min at 1000 mg L^-1 of IBA (Table 1).

In the regression equation, the total number of roots obtained through direct morphogenesis tended towards a linear increase in the number of roots produced by cuttings, as the concentration of the growth regulator increases (Figure 3).

Figure 3 Effect of eight concentrations of IBA (mg L^-1) on the number of roots produced in Bertholletia excelsa cuttings. r=0.864. 

Within each IBA concentration, the cuttings in the apical position had the lower percentage of rooting (Table 1). In this sense, ^{Dias et al. (2012b)} point out that the lower rates of rooting occur in apical cuttings, since they have the meristem, which is one of the principal places of natural auxin synthesis and, therefore, the exogenous application of synthetic growth regulators on the base is not needed; if it takes place, it causes a hormonal imbalance and inhibits the appearance of roots (^{Hartmann et al., 2002}; ^{Lopes et al., 2011}). The greater mean root lengths were obtained in treatments with cuttings in the middle position, submerged for 1 s at 3000 mg L^-1 of IBA (7.0 cm) and in 1000 mg L^-1 of IBA (6.4 cm) without significant differences between them. The smallest mean root length (1.5 cm) was observed with the cuttings in the apical position, immersed for 1 s in the 1000 mg L^-1 concentration of IBA (Table 2).

The linear regression showed a growing trend with an increase in the concentration of IBA, indicating that it would eventually have better results in higher concentrations. However, per cutting there was broad variation in the root length. This fact could be due to genetic factors, since each cutting presented unique characteristics and, therefore, the requirements for propagation also tended to be unique (Figure 4).

Figure 4 Effect of eight concentrations of IBA (mg L^-1) on the root length in Bertholletia excelsa cuttings. R=0.947. 

These results are similar to those found by ^{Xavier et al. (2003)}, who found significant differences in the rooting of Cedrela fissilis cuttings. ^{Hernández et al. (2012)} in a study with Piptadenia gonoacantha obtained higher average values than in the middle and basal cuttings with the application of IBA (6000 mg L^-1), so the three types of cuttings responded to the application of IBA.

According to the observations performed, after six months, the surviving cuttings were green and strengthened, suggesting a later initiation and elongation of adventitious roots. However, the success of this technique depends on the type of plant material, the species and its nutritional conditions, the type of growth regulator, the concentrations used and the immersion time. Due to the low percentage of rooting and time for rooting, it is necessary to make adjustments for the technique to be viable and to become an alternative for the production of seedlings throughout the year, especially in situations where the seed is a limiting input.