Study area

This study was conducted in the forests surrounding the towns of Chigorodó and Mutatá in the Urabá region at north western Colombia (Figure 1). The populations and individuals of both Magnolia species were found between 30 and 600 m asl and an average temperature of 28°C in very wet tropical forest, according to the life zone classification system (Holdridge, 1978).

Data collection

Random walks were carried out along the shores of the Mutatá River (7°14′44.4′′ N 76°25′30.9′′ W) and some of its tributaries, in order to locate trees based on information provided by villagers. Each tree and/or population found was georeferenced and properly marked. Next to each adult tree, the soil moisture and the air and soil temperatures were recorded. In order to compare attributes of the tree species of Magnolia (Total height, basal area) and their environmental variables (soil humidity and temperature, air temperature and altitude), One-way ANOVA and Tukey test were performed using R statistical program (R Development Core Team, 2014).

To characterize the forest communities where these species were found, 10 rectangular plots of 50 × 4 m (200 m²) were sampled for a total 2000 m². In each plot, all individuals with a diameter at breast height (DBH) ≥2.5 cm were recorded. Total height, DBH, phenological status and floral morphology were recorded for each tree. Botanical samples of individual trees were collected using the standard techniques for exsiccates. The specimens were processed at the HUA herbarium (Universidad de Antioquia, Medellín). The Catalogue of Vascular Plants of Antioquia (Idárraga, Ortiz, Callejas, & Merello, 2011) was used for species identification.

In order to rank the species values within the community, a modified importance value index (IVI) (Finol, 1976) was used based on the sum of its relative abundance and dominance. A redundancy analysis (RDA) was performed to establish the relationship among the flora communities associated with each Magnolia species and the measured environmental variables such as air temperature, soil humidity, altitude, transversal and longitudinal slopes. Biotic variables were also included such as species number/plot, IVI_M. katiorum and IVI_M. sambuensis. These analyses were performed using Canoco v. 4.56 (Ter Braak & Smilauer, 2009).

To understand the ecological processes determining the reproductive success of these species, information related to floral visitors, seed availability and germination was also recorded. Floral visitors were collected in traps located on the crown of three flowering laurel arenillo trees. The flowers were alcohol (30%) embedded to allure visitors. Flower buds and fruits ranging from the immature stage to the mature stage were observed between March 2010 and September 2012.

By conducting interviews with the local people, we gathered ethno-botanical information related to the location of the Magnolia trees, as well as their uses, local value and any traditional knowledge related to these species from local people.

Ex situ propagation

Seed germination and the viability of Magnolia were assessed based on field observations. Similarly, two experiments on ex-situ seed germination were conducted at the National University of Colombia in Medellín. In the first experiment, 200 seeds of molinillo were moistened with 1% sodium hypochlorite for 15 minutes, before being sown in a previously sterilized 2:1 mixture of soil and sand, and subsequently grown under two contrasting light conditions: full exposure and darkness. The second experiment included 120 seeds of this species that were hydrated for 12 hours and sown in three different substrates: sand, organic soil and a 2:1 mixture of organic soil and sand. Seeds of laurel arenillo were not available at this time.

Distribution and population size of Magnolia species

During field exploration, 24 adult trees of the two Magnolia species were recorded: 13 of laurel arenillo and 11 of molinillo. However, only two juveniles of laurel arenillo were found. All the trees were found in small forest relicts immersed in a matrix of grasslands and crops located on wide alluvial plains, along streams or river banks, or in relatively well-preserved forests at foothills. A few isolated trees were found along trails or in grasslands.

The adult trees of both species recorded during field exploration showed an average height of 24 m for laurel arenillo and 29 m for molinillo and average basal area of 0.112 m²/ha and 0.397 m²/ha respectively. However, both species shared similar micro-environmental conditions such as soil (25°C) and air (28°C) average temperatures, soil moisture (113%), and slope (25%). No significant differences were found between the environmental features where these species are growing. However, significant differences between the basal area (F_1,22 = 8.51 and p ≤ 0.0079) and tree height (F_1,22 = 5.22, p ≤ 0.0322) were recorded, being higher in the molinillo trees as shown in Figure 2. In addition, these analyses showed that the range of environmental conditions of laurel arenillo is wider.

Figure 2.

Environmental variables at the sampled sites containing M. katiorum (laurel arenillo) and M. sambuensis (molinillo) trees. Significant differences between the two Magnolia species were indicated for the basal area and tree height. Asterisks indicate outliers.

Floristic composition and environmental variables of the communities

In the 10 sampling plots, 523 trees with a DBH ≥ 2.5 belonging to 120 species and 46 plant families were recorded with an average of 54 trees/plot. Despite the environmental similarity, the RDA analysis showed that each Magnolia species grew in a different floristic community. The first two axes of the RDA with self-values of 0.369 and 0.235 explained 60.4% of the data variance. The IVI of laurel arenillo was positively related to altitude, longitudinal slope, soil humidity, and was negatively related to air temperature. On the other hand, molinillo was positively related to air temperature and transversal slope and was negatively related to longitudinal slope (Figure 3, Table 1).

Figure 3.

Biplot of the redundancy analysis (RDA) showing the species grouping along the first two axes and the environmental variables related to the species groups.

Table 1.

Results of the redundancy analysis.

The communities of both species showed contrasting characteristics. While the laurel arenillo was part of a more diverse community, it did not show a clear species dominance. On the other hand, molinillo was the most dominant species in its community which was less diverse (Table 2). However, the molinillo community showed a higher variability since in the RDA two plant communities could be identified as shown in Figure 3.

Table 2.

Features of communities associated to each Magnolia species group.

Laurel arenillo was associated with species such as Alibertia patinoi, Brosimum guianense, Erythroxylum panamense, Guarea glabra, Matisia cordata, Miconia trinervia, Orphanodendron bernalii, Pentaclethra macroloba, and Socratea exorrhiza, among others. This species composition suggests relatively preserved forests. Molinillo, on the other hand, was associated with species such as Bunchosia macrophylla, Cecropia tessmanii, Chrysochlamys dependens, Huberodendron patinoi, Hyeronima scabrida, Ossaea sp., Psychotria acuminate, Pouteria multiflora and Tovomita weddeliana. These species as well as molinillo can be found in open canopy forests, suggesting that are tolerant to forest disturbances.

However, several species were common to both plant communities, such as Amphirrhox longifolia, Castilla tunu, Cynometra martiana, Dendropanax arboreum, Euterpe oleracea, Minquartia guianensis, Simira cordifolia, Simaruba amara and Unonopsis colombiana. The forest composition of each community and their IVI values are shown in Appendix 1.

Some insights on reproductive biology and propagation

Based on a nearly two-year period observations, both species displayed large, fragrant, eye-catching flowers throughout most of the year. Molinillo began displaying buds in June and ended the flowering phase in September as ripening fruits began to appear. Ripe fruits and seeds were observed in October. However, in some cases, ripe fruits without dehiscence were found, which resulted in seed putrescence. Pieces of fruits were observed on the forest floor and were occasionally colonized by a fungus of the genus Xylaria (Figure 4).

Figure 4.

Magnolia sambuensis (molinillo): a) flower bud, b) flower bud and gynoecium, c) unripe fruit, d. Xylaria sp. on a fruit. M. katiorum (laurel arenillo): e) gynoecium, f) female stage, g) male stage, h) flower closed after female stage, i) gynoecium attacked by an unknown insect, j) unripe fruit.

Buds and flowers of laurel arenillo were observed from March through September. For this species, a much smaller fruit production was observed, and seeds could not be collected for propagation, despite having observed the flowers. One fruit was found in March 2010, and a fruit receptacle was observed in January 2012. Only remnants of a single ripe fruit were observed long after its dehiscence in April 2012 (Figure 4).

Regarding flower visitors, bees of the Halictidae family were the most obvious. Nocturnal moths of the families Noctuidae and Geometridae were also collected near the flowers of laurel arenillo and molinillo trees.

Data regarding the anthesis and pollination of both Magnolia species remain scarce, despite their importance for understanding the reproductive cycles of both species. Trees producing flowers and buds year round are very rare, suggesting that the flowers may fall off or are aborted prior to the fruit formation stage. Insects that attack the gynoecium (Figure 4) or immature fruits (Yepes, 2007) could also influence fruit formation. When fruits are formed as in the case of molinillo, they might reach maturity without opening.

Seeds from fruits picked from the ground beneath 25- to 30- m-tall mother trees of molinillo were used in the two germination experiments. In the first experiment with 200 seeds, only three seeds germinated: two at full exposure and the other in the dark. In the second experiment with 120 seeds and three different substrates, 33 seeds germinated as follows: 13 in the sand substrate, 11 in the soil-sand mixture, and nine in the forest soil substrate (Figure 5).

Figure 5.

Percentage of seed germination in two experiments. a. Full exposure and dark treatments (n = 200), b. Substrate treatments using seeds soaked for 12 hr (n = 40 per treatment).

Distribution and population size of Magnolia species

Magnolias in the Urabá region are rare and difficult to find, not only due to their characteristic rareness, but also because of forest fragmentation. Similar situations have been reported for other Colombian species such as M. guatapensis, M. jardinensis and M. polyhypsophylla in Andean cloud forests (Gómez, 2011). An additional difficulty for locating trees arose from the common names given to these species by local people. According to herbaria records, the common name “almanegra” actually refers to the species O. bernalii (Serna, 2011). According to sawyers in the area, the common name of M. katiorum is “guacharaco” or “laurel arenillo”; the latter name is used more often. Interestingly, forestry workers complain that when this species is cut with the chainsaw, it damages (wears out) the chain saw blades. Obviously, this is not well appreciated by the loggers. However, other inhabitants have stated that these species were widespread in the past but are currently rare due to wood overharvesting.

Regarding species composition of the communities, all sampling sites were located in moderately preserved forests except for the highly disturbed plot 6, which was associated with laurel arenillo and was characterized by several pioneer species such as Urera baccifera, Piper auritum and S. cordifolia. This suggests that this area had been recently disturbed, as these dominant species belong to the Urticaceae, Rubiaceae and Piperaceae families, respectively (Guariguata & Ostertag, 2002).

Of the 10 most important species, according to the IVI of both communities, at least three belong to Rubiaceae family. This is one of the most abundant families in Colombia (Mendoza & Jiménez, 2004) and is common not only in preserved forest ecosystems but also in disturbed areas. In contrast, species such as C. biflora (Moraceae), C. martiana (Fabaceae), S. exorrhiza (Arecaceae) and M. guianensis (Olacaceae) are typical of well-preserved forests with low light availability. These differences in species composition might indicate that laurel arenillo inhabits forests that are relatively less disturbed compared to molinillo. However, the floristic composition suggests a high successional process in both forests.

As with several American (Cruz, Vega, & Jiménez, 2008; Dieringer & Espinosa, 1994; Dillon & Sánchez, 2009; Sánchez & Pineda, 2006; Tobe, 2000; Vásquez-García et al., 2013; Weaver, 1997) and Asian (Si, 2000; Xie, Fu, Zeng, Lui, Wen, & Zhong, 2012) magnolias, these two species grow in diverse forests, but are not dominant in them. They also typically exhibit low population densities and are considered rare. However, in some disturbed broad leaf forests in the United States, M. grandiflora may be the dominant species (Batista & Platt, 2003; Kwit & Platt, 2003).

Some insights on reproductive biology and propagation

Regarding the pollination process, studies have shown that Magnolia flowers are protogynous (Thien, 1974) in order to avoid or minimize self-pollination, such that on day 1 of functional flowering (female stage), the stigmas are receptive for only some hours before and upon the opening of the first flower, while the stamens remain immature. Overnight, the flower closes (i.e., the tepals reflex over gynoecium). On day 2 (male stage), the flower re-opens with ripe stamens that usually detach from the androecium as they dehisce pollen, while the stigmas are no longer receptive. In most Magnolia species, female and male stages are separated by c. 24 hours (Dieringer & Espinosa, 1994).

Although it is known that individual flowers of temperate species can last from two to four days, the stigmatic receptivity and the pollen maturity and viability for several Magnolia species remain unknown (Dieringer & Espinosa, 1994; Matsuki, Tateno, Shibata, & Isagi, 2008; Thien et al., 1998).

For many Magnolia species, bees have been reported to be effective pollinators (Wang, Wang, Liu, Wang, & Shen, 2005). However, in previous studies of magnolias from other countries, most insects were found consuming the flowers rather than pollinating them. Conversely, moths have not been reported to be pollinators but have been reported to be fruit consumers (Yepes, 2007). The role of insects in Magnolia pollination in Colombia remains unclear, and knowledge of pollination systems remains limited, especially in tropical America. This limitation may be associated with difficulties of observing pollinators, especially their movements from plant to plant (Thien et al., 1998), due to the difficulty of climbing high trees. Additionally, the existence of a few sparse individuals complicates the observation process as well as the capture of pollinators. Moreover, these species may be self-compatible, as reported for other Magnoliaceae (Thien et al., 1998).

The lack of dehiscence in fruits of molinillo may have been associated with heavy rains and/or long periods of rain that characterized the year 2010 compared to 2011 (METEOMANZ, 2015).

Although we do not know the exact cause of the low germination percentage for seeds of molinillo, we suspect that it is related to the short-term viability that has been recorded for some seeds of Magnolia or to improper storage (it took approximately 8 hours to transport the seeds from the field to the lab, under uncontrolled environmental conditions). Nevertheless, the germination percentage of magnolias may vary significantly depending on the species and/or treatment used, as shown in Appendix 2 (Cao et al., 2012; Chalermglin, 2012; Li, Zhou, Chen, & Liu, 2000).