P. stipulacea fruits were manually collected from five selected trees in September 2013. The harvesting took place in a Caatinga vegetation area in the Natural Heritage Private Reserve “Não me deixes” Farm (4˚49'34''S, 38˚58'9''W and 210 m above current sea level), in Quixadá County, Ceará State, Brazil.

The fruits were dried in an oven (45˚C/3days) and processed for seed extraction. After processing, seeds were separated from impurities (part of fruits, leaves, petioles) and from other seeds which were damaged by insects or that were broken. They were then placed in plastic containers and stored (for two months) in a cold chamber at an average temperature of 12˚C and relative humidity of 55% until the beginning of the experiments.

The experiments were conducted in the Seed Analysis Laboratory (LAS―Laboratório de Análise de Semen- tes) from the Plant Science Department, at Federal University of Ceará (UFC). Botanical material containing P. stipulacea leaves and fruits was collected for specimen preparation and was deposited at Prisco Bezerra Herbarium―EAC, Federal University of Ceará, under protocol number 054121 (EAC).

After they were mixed, 100 seeds were randomly selected for individual measurement. Biometric featuring was performed by means of digital caliper (0.01 mm) through which the following variables were measured: length, width, thickness and weight (in precision scale). The length was measured from the base to the apex and width and thickness were measured on seeds midline. Data were subjected to descriptive statistics in Excel application in order to calculate arithmetic mean, standard deviation, standard error, coefficient of variation and confidence interval for each biometric feature.

P. stipulacea thousand seed weight was calculated in accordance with the Rules for Seed Analysis recommendations [13] . The number of seeds per kilogram was determined from the thousand seed weight results.

The dormant seeds [1] were manually scarified with n. 80 sandpaper, opposite to the micropyle, in order to wear the integument. After scarification, they were treated with sodium hypochlorite to avoid fungi attack during germination. The seeds were placed in Becker and carefully and completely immersed in 5% sodium hypochlorite for 5 minutes. They were then transferred to a fine mesh steel sieve, rinsed in running water and dried with paper towel.

The treated seeds were sown on two sheets of germitest filter paper arranged in 9.50 cm diameter Petri dishes. The Petri dishes with the sheets of germitest filter paper were previously autoclaved at 120˚C for 20 minutes. The substrate was moistened with distilled water in the ratio of two and a half times the weight of the paper [13] .

Temperature and light effects were checked by using the completely randomized experimental design. Treatments were distributed in 4 × 4 factorial arrangement, and subjected to constant temperatures of 20˚C, 25˚C and 30˚C and alternating temperatures from 20˚C to 30˚C, under a 12-hour photoperiod and the following light conditions: white, darkness, red, and far red, with four replicates of 25 seeds. During alternating temperatures, the light period corresponded to the higher temperature. As for the white light condition, the Petri dishes were placed in transparent plastic bags to prevent water loss. In the absence of light (darkness), seeds were individually wrapped in aluminum foil and placed inside black plastic bags. For the simulation of red light (660 nm) and far red light (730 nm), the method described by Almeida and Mundstock [14] was adopted. The red light was obtained by using two sheets of red cellophane whereas the far red light, was obtained by using two sheets of red cellophane and two sheets of navy blue cellophane as described by Bickford and Dunn [15] .

The Petri dishes were placed in a BOD (Biological Oxygen Demand) germination chambers regulated in their constant and alternating temperature regimes. The germinated seeds were daily recorded for 10 days, and the evaluation criterium was the presence of a radicle of at least 2 mm. Security green light was used to count seeds subjected to darkness, red and far red light. During this period, the Petri dishes were carefully remoistened with distilled water when necessary. Finally, the following variables were evaluated:

a) Germination percentage, % G = (N/A) × 100, where: N = number of germinated seeds and A = total number of sown seeds.

b) Germination speed index, according to Maguire [16] : GSI = G₁/N₁ + G₂/N₂ + ... + G_n/N_n, where: G₁, G₂ and G_n = number of germinated seeds in each day and N₁, N₂ ... N_n = number of days elapsed since the day of sowing.

c) Mean germination time, according to Ranal and Santana [17] : MGT = (ΣG_i * T_i)/ΣG, with the results expressed in days, where: G_i = number of germinated seeds within a given time interval T_i, G = number of germinated seeds, T_i = days of germination.

d) Length: the length of seedlings root and shoot was measured separately, discarding the cotyledons, with the help of a millimeter ruler.

e) Dry weight: obtained from material placed in paper bags duly identified and dried in the greenhouse at 80˚C, with forced air circulation for 24 h. After this period, the seedlings were weighted in precision scale and the results were expressed as g/seedling.

The water restriction simulation on P. stipulacea germination under laboratory conditions was tested by means of polyethylene glycol solutions (PEG 6000), which satisfactorily simulates low water potentials. PEG is chemically inert and does not show toxicity to the seeds [18] - [21] .

The seeds were placed in Petri dishes (9.50 cm diameter) and the germitest paper substrate was moistened with distilled water (witness) and PEG 6000 solution in different concentrations. The used polyethylene glycol concentrations (PEG 6000) were −0.2; −0.4; −0.6; −0.8; −1.0 and −1.2 MPa water potential, obtained for the constant temperature of 30˚C, according to Villela et al. [18] . The experiment was set up in completely randomized design with four replicates of 25 seeds each.

The Petri dishes were kept in germination chamber (BOD) under constant temperature of 30˚C for a 12-hour photoperiod. The germitest paper substrate and PEG solutions were replaced every 48 hours aiming at maintaining the experiment initial conditions. The counts of germinated seed were carried out daily for 14 days, and the germination percentage (% G) was determined at the end of the experiment.

All the dependent variables were analyzed with respect to normality by using Kolgomorov-Smirnov test, and the homogeneity of variances was analyzed by using the Levene test. After the two criteria were met, the data were subjected to ANOVA, and averages were compared by Tukey’s test at 5% significance level. When they were not met, the data were subjected to non-parametric statistics by the Kruskal-Wallis test and evaluated by non- parametric multiple comparisons at 5% significance level [22] .

P. stipulacea seeds are uneven regarding size, and show variation in length (6.71 to 8.67 mm), width (4.78 to 6.32 mm) and thickness (1.88 to 3.07 mm). The unit seed weight ranged from 42.9 to 80.28 mg. Seeds descriptive statistics is presented in table 1. The thousand seed weight was of 44.134 g, which allows inferring that a kilogram of P. stipulacea seeds can contain 22,658 seeds.

Temperature and light quality (white, darkness, red and far red) evaluations for P. stipulacea seeds germination percentage (% G) showed similar means (H_(3.15) = 32.96, p = 0.0048). The data presented in table 2 show that the species germinated regardless the presence and absence of light, both at the constant temperatures of 20˚C, 25˚C and 30˚C and at the alternating temperature from 20˚C to 30˚C.

The ANOVA result for P. stipulacea seed germination showed that abiotic factors such as temperature, light and their interactions exerted significant effects for all variables under analysis, with the exception of the light for seedling dry weight (Table 3).

The interaction between light and temperature influenced the germination speed index (GSI). It is possible to see that the highest germination speed index (10.52) occurred at 30˚C under far red light condition, and the lowest one (4.3) was observed under darkness condition at 20˚C (Table 4).

There was a significant statistical effect of temperature and light treatments on P. stipulacea seeds Mean Germination Time (MGT). The lower MGT was obtained at 30˚C under red and far red light conditions. The

Means followed by the same letter do not significantly differ from each other by nonparametric multiple comparisons at 5% probability.

^**Significant at 1% probability level by F test; ^*Significant at 5% probability level by F test and ns: non-significant.

Means followed by the same lowercase (lines) and uppercase (columns) letters do not significantly differ from each other by the Tukey’s test at 5% probability.

largest MGT was obtained at 20˚C in all evaluated light conditions (Table 5).

The white light was more efficient in stimulating P. stipulacea seedlings’ root growth. The highest root length mean occurred at 20˚C (2.55 cm) under white light condition, although the other temperatures evaluated under the same light condition were statistically similar (Table 6).

The lowest mean shoot length occurred at 20˚C in the four evaluated light conditions. There was the highest mean shoot length in all treatments (Table 7).

The dry mass was lower at 20˚C regarding the four evaluated light conditions. There were small variations in the means obtained from the other treatments (Table 8).

P. stipulacea germination percentage was significantly influenced by different treatments with polyethylene glycol (PEG 6000), (F = 331.29, p < 0.001). Seed germination was not affected by PEG 6000 concentrations up to −0.4 MPa. However, it was reduced at the concentrations of −0.6 and −0.8 MPa. There was no seed germination at −1.0 and −1.2 MPa (Figure 1).