Germination analysis in different substrates of Tabebuia cassinoides (Lam.) DC. of high restinga forest, Ubatuba, São Paulo, Brazil

Tabebuia cassinoides (Lam.) DC., popularly known as caixeta, is a well-known species for its timber harvesting. The species is dominant in flooded high restinga forest where we collected fruits and seeds from Núcleo Picinguaba, Serra do Mar State Park, Ubatuba, Brazil. The restinga, or sandy coastal forest, is a type of vegetation formed by sandy sediments from marine deposition. Thus, we asked whether the sand substrate would produce better germination indices than germitest paper or vermiculite due to its similarity to specie natural condition. We placed a total of 60 seeds in Gerboxes with duplicates of the three substrates. We placed each Gerbox with ten seeds randomly in the germination chamber at 30 °C, photoperiod of 12 h, and water at proper field capacity. We periodically evaluated the seeds and analyzed the germination percentage (%), germination speed index (GSI), and mean germination time (MGT). Our results showed that sand was the most favorable substrate for germination of caixeta seeds for at least 35 days, corroborating the initial hypothesis. We hope that our results can improve the sustainable management and conservation of the species.


Introduction
The Atlantic Forest is the second-largest forest of the American continent, which extends along the Brazilian coast (Tabarelli et al., 2005;Ribeiro et al., 2009). Historically, the Atlantic Forest stretched continuously covering 15% of Brazilian territory and spread over seventeen states, but nowadays, the fragmentation of the forests is observed as a consequence of deforestation and anthropic actions, remaining only 11% to 16% of of tropical rainforest species since these species vary greatly concerning dormancy types and time required for germination onset (Cardoso, 2004).
Since the Tabebuia cassinoides occurs in a flooded high restinga forest, we asked which is the best substrate for germination of its seeds. Many tropical trees have higher germination indices in paper or vermiculite substrates (Mello and Barbedo, 2007;Martins et al., 2012), however, we hypothesize that sand substrate would be better due to its similarity to the restingas natural condition.

Species of study
The Tabebuia cassinoides (Lam.) DC., Family Bignoniaceae, popularly known as Caixeta or Pau-Caixeta, is a Brazilian endemic species that occur in the ombrophilous dense forest of Atlantic Forest at São Paulo, Rio de Janeiro, Minas Gerais e Espírito Santo States (Lohmann, 2019). T. cassinoides is a hygrophilic plant that occurs in soils permanently humid or flooded in Brazilian coastal range (Lorenzi, 2008), presenting formation of adventitious roots and hypertrophic lenticels which are important characteristics to tolerance flooding (Kuniyoshi, 1993).
T. cassinoides has 6-12 m height, trunk 30-40 cm in diameter, simple leaves 12-22 cm long without stipple, flowering season between July and January and occurrence of its ripe fruit between October and March (Lorenzi, 2008). The fruits are capsule-shaped, linear-oblong, where the seeds are dispersed by the wind since it is a winged seed (Carvalho, 2003;Kolb and Joly, 2010).

Study area and fruit collection
We collected seeded fruits in a flooded high restinga forest fragment, located in Núcleo Picinguaba, Serra do Mar State Park, Ubatuba Municipality, São Paulo State, Brazil (23° 21' 10.57" S and 44° 51' 5.699" W, 10 m elevation). The region yearly average temperature of the region is 22.6 ºC, and ranging between a minimum average of 17.8 °C and a maximum average of 27.3 °C (CEPAGRI, 2017). The yearly average rainfall is 2.154 mm, with a rainy season from November to April (Rolim et al., 2007).
In the study area, we identified five individuals of T. cassinoides with the largest diameters at breast height (DBH) and collected their fruits with seeds. We collected on December 5, 2018, with the aid of high pruning shears to reach the highest branches. We incorporated two vouchers in Herbarium Amélia Vera Guimarães de Sousa (MACK) of Mackenzie Presbyterian University under numbers 3199 and 3200. Following Ramos (1980), we visually qualified the seeds searching for pathogens presence or predation signs, then underwent kiln drying at 42 °C for 3 h and stored in multiwall paper bags at room temperature and humidity controlled with the aid of a hygrometer during 60 days.

Seed germination
We carried out the seed germination tests in germitest paper, vermiculite, and sand substrates placed in Gerboxes, which are a germination box made of transparent plastic (11 x 11 cm). We sterilized all Gerboxes with 70% alcohol. We kept the sand and germitest paper substrates for 12 h in kiln drying at 100 °C to minimize any contamination, while we opened the vermiculite substrate package before the experimentation.
We weighed each substrate with a semi-analytical electronic balance (Gehaba, model BK 3000, São Paulo, Brazil) to properly attain the field capacity, which varies according to each substrate type (Brasil, 2009). Is indicated a water volume an amount equivalent to 2.0-3.0 times of the weight of germitest paper, hereafter named only paper, to attain the field capacity, while 60% of the weight of sand and vermiculite (Brasil, 2009). Therefore, according to field capacity calculation, we initially moistened the seeds with 1 mL of water in vermiculite, paper 3 mL, and sand 4.5 mL.
We placed ten seeds in each Gerbox of the three types of substrates performed in duplicates; thus, a total of 60 seeds in six Gerboxes. We placed the Gerboxes randomly in the germination chamber (Tecnal model TE Piracicaba,Brazil). The size of the seeds is often the factor that might render unviable a higher number of tested seeds due to lack of space in small gerboxes or the germination chamber (Ferraz and Calvi, 2010), which were our scenario. We irrigated every two days, respecting the field capacity of each substrate. We set the germination chamber at 30°C and a photoperiod of 12 h, as indicated by Bianchetti et al. (1995). We periodically counted the number of germinated seeds for 35 days, and we considered the germinated seeds those with approximately 2 mm of radicle emergence (Brasil, 2009).

Data analysis
To test whether the total number of germinated seeds differed between substrates, we performed at the 5% significance level the goodness-of-fit Chi-square test (x²) on Past 3 software (Hammer et al., 2001). We analyzed the seed germination percentage according to 15, 25, 35 days elapsed after planting, which we identified as the days that the treatments had more distinction in the number of germinated seeds.
For each substrate, we calculated germination percentages, the germination speed indexes (GSI), and mean germination time (MGT). We calculated the GSI by the sum of seeds number germinated in each day evaluated and divided by the number of days elapsed between sowing and germination (Maguire, 1962). We considered as partial GSI the number of germinated seeds divided by the number of respective intervals of 15, 25 and 35 days. We calculated the MGT by the sum of germinated seeds number per day multiplied by respective incubation time divided by the sum of germinated seeds number (Ranal and Santana, 2006).

Results
At the end of 35 days of the experiment, out of 60 seeds of Tabebuia cassinoides placed in the germination chamber, 24 seeds germinated (40%). Sand substrate obtained the highest percentage of germinated seeds (80%) significantly when compared to paper (25%) and vermiculite (15%; Table 1) (X² = 12.25; p=0.002). Evaluating through time, in the first 15 days of germination test, only one seed in sand substrate germinated, but after 25 days, T. cassinoides seeds germination reached levels of 50%, 20% and 5% on the sand, paper, and vermiculite substrates, respectively (Figure 1). At the end of the test, at 35 days, the germination proportion was maintained. The type of substrate also showed differences in seed germination speed. According to GSI, sand substrate had the highest germination velocity (0.60), followed by paper (0.19) and vermiculite (0.10). Analyzing the partial calculations of GSI, a peak of germination velocity was on day 25 in paper and sand substrates (Figure 2). The mean germination times were similar between treatments, and paper substrate presented the shortest mean time of germination (27 days), followed by sand (28 days) and vermiculite (32 days). The germination test for up to 35 days allowed the observation of a higher percentage of germination but, consequently, increased the MGT.

Discussion
The highest percentage and germination speed index was obtained on the sand substrate, showing 80% germination at the end of the experiment and the highest GSI. Also, in the first 15 days after sowing, it was observed that only seeds on sand substrate germinated, while seeds on paper and vermiculite substrates not started. The results indicated sand as the best substrate to Tabebuia cassinoides germination since all duplicates of each substrate placed in the germination chamber were submitted to the same temperature at 30 °C, same photoperiod of 12 h, and the proper field capacity of each substrate.
Our results were similar to those obtained by Bianchetti et al. (1995), which indicated the optimal germination condition of Tabebuia cassinoides seeds was in the sand substrate at 30 °C. This temperature was an optimal germination condition for other Bignoniaceae seeds, found for Handroanthus impetiginosus (Mart. ex DC.) Mattos and Handroanthus serratifolius (Vahl) S. Grose (Oliveira et al., 2005). For other related species, the results indicated similar temperatures to attain an optimum germination rate, such as Handroanthus serratifolius, between temperature ranges of 20 °C and 30 °C and Tabebuia roseoalba (Ridl.) Sandwith, at an ideal temperature of 30 °C (Santos et al., 2005;Stockman et al., 2007). Maciel et al. (2013), analyzing the seedling length of species Jacaranda mimosifolia D. Don obtained better results when the seeds were at temperatures of 25 °C to 30 °C.
Regarding substrates types, Bovolini et al. (2015) found for Jacaranda micrantha (Cham.) that sand substrate combined with the temperature 25 °C-30 °C was adequate to provide higher values of germination speed index and for seedling development. According to Borba-Filho (2006), Handroanthus serratifolius seeds on the sand substrate, between five and ten days, the specie obtained the highest values of germination percentage (94%) and germination speed. For Tabebuia cassinoides, Fowler et al. (1999) describe that the highest germination percentage was in sandy soils (77.3%), values very similar to our results.
Considering the germination speed index (GSI), Santos et al. (1994), studying Mimosa caesalpiniifolia Benth. (Fabaceae), and Bocchese et al. (2008), studying Handroanthus heptaphylla (Vell.) Mattos, obtained that sandy substrate has higher GSI due to greater ease of emergency of seedlings and lower difficulty to break the physical barrier of soil through its root system. This conclusion corroborates our results for Tabebuia cassinoides in the sand substrate, while the other substrates used in our study do not impose a high physical barrier for the seedling.
The peak of germination at 25 days, indicated by partial GSI, and about 27-28 days by MGT, evidenced that both paper and sand substrates presented similar germination behaviors, however, the germination even lasted until 35th day with good results. Bianchetti et al. (1995) performed germination analyses until the 28th day, indicates that germination of T. cassinoides prolonged for at least four weeks for final evaluation. Thus, we highlighted that the germination must be performed until four to five weeks to obtain a reasonable total amount of germinated seeds.
According to Martins et al. (2012), the species Handroanthus chrysotrichus (Mart. ex DC.) Mattos, which occurs in the Atlantic Rainforest and gallery forest of the Cerrado domain, presented a higher germination percentage on the vermiculite substrate, which has similar characteristics of clay soil where this species occurs. Temperature and soil have a grand influence on species germination, and these factors should limit the distribution of plant species along the restinga vegetation (Sabonaro, 2011). Since T. cassinoides occur dominantly in high restinga forest with high water infiltration due to the high proportion of sand and low clay content, the germination experiment showed better results on the sand substrate, corroborating our hypothesis.
The importance of knowledge about the germination of species is crucial to establish the best germination success rate regarding the influence of abiotic factors as soil and temperature. By improving the understanding of Tabebuia cassinoides germination, it is possible to develop better sustainable forest management aiming conservation and reintroducing new individuals in nature, since this species has high levels of exploitation due to its high economic value.

Conclusion
Our results indicated that different substrates types had a direct influence on the germination of caixeta. The best germination condition for Tabebuia cassinoides seeds is on the sand substrate, at 30 °C, for at least 35 days in the germination chamber, with a photoperiod of 12 h, and water irrigation reaching 60% of substrate weight. These conditions are similar to those of their natural environments.