Introduction

Multipurpose trees are cultivated in order to make more than one significant contribution to the production and service functions within agroecosystems. A wide diversity of these native species, which have a high cultural value for society, are distributed in various ecological regions (^{Aguilar and Condit, 2001}). However, the implementation of local taxa in agroforestry systems requires knowledge of their characteristics: phenology, habits and form of growth (^{Wood and Burley, 1995}), productive capacity, ability to recycle nutrients, nitrogen fixation, resistance to pests and diseases, and the preference of the producers that will use it (^{Huxley, 1983}; ^{Owino, 1992}).

The phenology of species is helpful to make decisions regarding the appropriate canopy management to ensure production; furthermore, the canopy is intended to provide varying levels of shade throughout the year in synchronization with the phenological cycle of the crops (^{Somarriba, 2005}). Tree architecture is another important aspect when agroforestry systems are implemented, because the shape and characteristics of the crown determine the way in which the tree will grow, as well as its efficiency in the production of timber (^{Sestras, 2004}) and that of the associated crops (^{Beer et al., 2003}).

Disturb occurs in the low and medium rainforests of central veracruz as a result of agriculture and stockbreeding (^{Pennigton and Sarukhán, 2005}; ^{Gómez et al. 2010}); the implementation of agroforestry systems with multipurpose native species has been proposed in order to restore them. Some of the main species in these ecosystems --Calyptranthes schiedeana O. Berg (Myrtaceae), Lysiloma acapulcense (Kunth) Benth (Fabaceae) and Tabebuia chrysantha (Jacq.) G. Nicholson (Bignoniaceae)-- are characteristic of secondary vegetation and are prevalent in mature forests. They have a high cultural importance and are also essential for wild life, although they are scarce in the region at present (^{Suárez et al., 2012}).

Although the relevance of planting native trees is recognized, there is little information regarding their ecology, biology and performance; therefore, the options of trees for restoration purposes and for the forestry industry are limited (^{Condit et al., 1993}; ^{Wishnie et al., 2007}). Given the need to become familiar with the native vegetation of the region that has a potential to become integrated into agroforestry systems, the purpose of the present study is to describe the phenological patterns and the tree architecture of C. schiedeana, L. acapulcense and T. chrysantha within agroecosystems with dispersed and clustered trees in pasturelands, agricultural plots and living fences where their characteristics may be assessed in similar conditions to those that prevail in the sites for their potential establishment in agroforestry systems.

Materials and Methods

The study was carried out in various localities of the Paso de Ovejas municipality, veracruz (19°1’58.27’’-19°12’54.18’’ N, 96°25’25.28’’-96°30’46.29’’ W), between the altitudes of 84 and 203 masl, with a mean annual temperature of 24 °C and an annual precipitation of 973 mm (^{Inegi, 2009}). The vegetation in this area is a low deciduous forest, and its importance is derived from its structure, floral diversity and number of endangered species (^{Castillo-Campos, 2005}).

T. chrysantha, L. acapulcense and C. schiedeana --belonging to the group of heliophytes-- were considered (^{Pennigton and Sarukhán, 2005}); they were selected for their cultural importance and their scarcity in the study area (^{Bautista, 2009}; ^{Suárez et al., 2012}). Ten mature individuals of each species were chosen in 14 sites used for agriculture and stockbreeding (surface areas of 0.5 to 30 ha), selected within a radius of approximately 18 km. The assessed trees were located in living fences, dispersed in pasturelands and agricultural plots or in copses, in association with other taxa.

Results

Phenology of the species

Calyptranthes schiedeana kept its leaves throughout the year, showing a rapid development of leaf buds in May, until its crown was completely covered in June. The falling of the leaves began in January and became more intense in April and May. Once the tree crown was covered with foliage, it soon began to develop floral buds, in June, and to fruit slowly, in July; the fruiting became more intense in October, with increased fruit dispersal (Figure 1). The precipitation was necessary for this species to develop leaf buds (r = 0.60, P = 0.0004), and the temperature influenced the fruiting ( r = 0.46, P = 0.0108).

Figure 1 Occurrence of seven phenophases of Calyptranthes schiedeana O. Berg, evaluated every 14 days during the 2014-2015 period. 

Tabebuia chrysantha reached its maximum development of leaf buds in May, and its largest leaf cover from June to December, with the highest defoliation in February, being completely leafless in April and May. The development of floral buds was observed from June to March, when flowering occurred; the fruiting and fruit dispersal occurred in April (Figure 2). The decrease of the temperature had an impact on the development of floral buds (r = - 0.58, P = 0.0008), and the falling of the leaves (r = - 0.67, P< 0.0001), while the increase in precipitation was related to the formation of leaf buds (r = 0.45, P = 0.0107).

Lysiloma acapulcense, like C. schiedeana, kept its leaves the year around, developed leaf buds in March --the most intense period of leaf fall (February-March)--, while in April (the month with the lowest precipitation) a quick transition from the development of floral buds to flowering was observed. Fruiting occurred mainly in November, and seed dispersal, in April of the following year (Figure 3). Fruiting in L. acapulcense diminished with low temperatures (r = -0.90, P < 0.0001).

Figure 2 Occurrence of seven phenophases of Tabebuia chrysantha (Jacq.) G. Nicholson, evaluated every 14 days during the 2014-2015 period. 

Figure 3 Occurrence of seven phenophases of Lysiloma acapulcense (Kunth) Benth, evaluated every 14 days during the 2014-2015 period 

Architecture of the species

Calyptrantes schiedeana exhibited a determined, sympodial growth pattern of the main axis and a proleptic merisystematic activity; the branching patter was sympodial, with a plagiotropic orientation and symmetry; the orientation of the first order branches in relation to the main axis is upward (29.5°), with axillary reproductive structures, and the morphology of the inflorescences is monopodial, with total and adaptive reiterations.

Lysiloma acapulcense exhibited a sympodic growth of the main axis, a determined shape and a proleptic merisystematic activity; the branches are sympodial and plagiotropic, and first order branches have a diffused orientation (51.2°). The reproductive structures were in an axillary position, and the morphology of the inflorescences was monopodial, with total and adaptive reiterations. Tabebuia chrysantha exhibited a monopodial growth factor, a determined shape and a proleptic meristematic activity, with a sympodial branching pattern and an orthotropic orientation and symmetry; the first order branches were upward (44.1°). The position of the reproductive structures was apical (terminal), and the inflorescences had a sympodial morphology, with total and adaptive reiterations.

The first order branches of Lysiloma acapulcense and T. chrysantha had the same angle (P = 0.067), which surpassed that of C. schiedeana branches (P < 0.0001); likewise, the angles of the second order branches differed between C. schiedeana and T. chrysantha (P = 0.044), and those of the third order branches were similar in all these species (P = 0.943). Clearly, L. acapulcense is a larger tree, with longer branches (P < 0.0001) than T. chrysantha (P < 0.0001) and C. schiedeana (P< 0.0001), the latter of which had the shortest branches (Table 1).

Table 1 Branch architecture of Calyptranthes schiedeana O. Berg, Lysiloma acapulcense (Kunth) Benth and Tabebuia chrysantha (Jacq.) G. Nicholson pole trees. 

Different letters on the same line indicate statistical differences, Tukey (P<0.05).

All three taxa exhibited the same leaf insertion angle (P = 0.642); the leaves of L. acapulcense were compound (consisting of leaflets and small subleaflets), with a larger leaf area (P = 0.05) and leaf blade length (P = 0.05); however, this species exhibited a lower leaf density index. Like T. chrysantha, C. schiedeana had smaller, simple leaves and a higher leaf density index (Table 2).

Table 2 Leaf architecture and dasometric measures of Calyptranthes schiedeana O. Berg, Lysiloma acapulcense (Kunth) Benth and Tabebuia chrysantha (Jacq.) G. Nicholson pole trees. 

Different letters on the same line indicate statistical differences, Tukey (P < 0.05).

Lysiloma acapulcense stood out for its larger normal diameter (P < 0.0001), diameter at base ( P< 0.0001) and height (P < 0.0001) than those of other taxa; also, its crown is higher (P < 0.0001) and larger (P < 0.0001), while C. schiedeana had the smallest dimensions (Table 2).

The crown architecture varied: the crown of L. acapulcense is parasol-shaped, that of T. chrysantha is cylindrical, and that of C. schiedeana is mostly spherical-ovoid (Table 3). The arrangement of the foilage is evenly light and discontinuous in L. acapulcense, while in C. shiedeana it ranges from light to continuous, and in T. chrysantha, the leaf cover is less well defined. The crown architecture of Lysiloma acapulcense was observed to be tufted, segmented, multiple, and ranging from shallow to deep; the crown of T. chrysantha is star-shaped, unsegmented, ranging from compound to simple and from shallow to deep, and that of C. schiedeana is maily segmented and billowing, simple and deep (Table 3). Due to the management and characteristics of the sites where the trees were located, certain aspects of the architecture turned out to be unclearly defined, and therefore, individuals of the same species exhibited differente patterns.

Table 3 Frequency-based crown architecture of Calyptranthes schiedeana O. Berg, Lysiloma acapulcense (Kunth) Benth and Tabebuia chrysantha (Jacq.) G. Nicholson. 

Discussion

All three taxa showed characteristic phenological patterns of low deciduous and sub-deciduous forest species, althought they tended to differ in certain aspects. L. acapulcense did not lose absolutely all its leaves at any point, and when it reached 75 %, new shoots began to cover its crown again (in the spring), while most specimens of C. schiedeana and T. chrysantha lost all of their foliage (during the dry seasons), as do many deciduous species whose leaves fall at the onset of hydric stress (^{Wright, 1996}), and which produce new leaves one month before the beginning of the rainy season (^{Opler et al., 1980}). Notably, defoliation lasted longer in L. acapulcense than in other taxa; this indicates differences in the hydric stress tolerance mechanism between the taxa (^{Sánchez et al., 2004}).

The flowering times did not coincide: C. schiedeana flowered during the summer dry season (August), while the other species flowered during the dry season. However, they all had short, clearly defined flowering gaps. Furthermore, the seed dispersal occurred at a different rate: during the dry season (spring) for L. acapulcense and T. chrysantha, and in winter for C. schiedeana. In deciduous species, leaf growth does not coincide with the ripening of the fruits, which occurrs in spring and fall. This is because the distribution of resources (nutrients) for organ production takes place during the most favorable periods of the year (^{Opler et al., 1980}).

In general, the studied species had different phenological patterns although they shared the same physiographic region, and can be identified with one of the three strategies (competitors, stress tolerator and ruderal) proposed by ^{Grime (2001)}. The pioneers bloom throughout the year, while those of late successional phases do so in annually marked periods, or even every two or three years (^{Plana, 2000}). The phase difference of blooming and fruiting peaks between the species reduces the competition for common dispersal agents and pollinators within a vegetal community (^{Snow, 1965}), and furthermore, it defines the diet of the consumers according to the availability of food (^{Van Schaik et al.,1993}). Therefore, the coincidence of biotic factors with the phenological cycles determines the intensity or occurrence of the phenophase peaks; climate cycles define the time in which the biotic processes related to phenology occur. In this study, soil management may have exerted some effect on the phenological patterns described above, although the scope of the research did not allow to discern their magnitude.

In studies carried out in Costa Rica and Mexico, T. chrysantha has been observed to bloom during the dry season (January-May) (^{Borchert et al., 2004}), and, like T. neochrysantha A. H.Gentry and T. rosea (Bertol.) Bertero ex A. D.C., it produces flowers immediately after the falling of its leaves (Borchert, 1983); furthermore, these species coincide with most low forest trees, which bloom during short periods (7-9 weeks) and extreme draughts (^{Opler et al., 1980}; ^{Kochmer and Handel, 1986}).

The phenological pattern of L. acapulcense differed from that cited by ^{Pennington and Sarukhán (2005)}, as its blooming and fruiting were more premature than in the tropical rain forest, where taxa of the Fabaceae family bloom at various times of the year (^{Ochoa-Gaona et al., 2008}). L. watsonii Rose, an evergreen taxon, shares flowering with L. acapulcense (April-May). However, the leaf fall period coincided partially with that observed by ^{Pennington and Sarukhán (2005)}, lasting two months (February-March) in the study area.

Only certain phenologic events were related to climatic variables. The development of leaf buds in C. schiedeana and T. chrysantha exhibited an evident positive relationship with precipitation: leaf buds were formed as the rainy season advanced.

Conversely, L. acapulcense did not seem to depend on precipitation, as its buds emerged during the season with least rain and the highest temperatures. This response may be due to the fact that L. acapulcense has a deeper root system that allows it greater access to the moisture of the soil (^{Meinzer et al., 1999}), and therefore it tolerates draughts well.

Leaf fall, flowering, the development of floral buds and fruiting were all related to the temperature. The fruiting and defoliation of L. acapulcense and T. chrysantha progressed as the temperatures fell; i.e. they occurred in the fall and the winter. The leaf fall rate during the dry season coincided with the reduction of the soil moisture and the increase in the hydric stress of the trees (Reich and ^{Borchert, 1984}).

The taxa exhibited very different architectures. The Troll model adopted by L. acapulcense is prevalent among the Fabaceae family and is characterized by the presence of horizontal axes with continuous overlap (^{Hallé et al., 1978}); its morphometric characteristics allow inferring structural attributes related to the environment: pioneer tree, pinnate laminae, secondary and tertiary branches flexible enough to reduce the aerodynamic resistance to strong gusts of wind; these taxa develop a limited number of small leaves, held by thin and more flexible petioles (^{Vogel, 1989}; ^{Ennos, 1997}).

Tabebuia chrysantha acquired the Leeuwenberg model, having erect axes (orthotrops), a determined growth and terminal inflorescences, characteristic of secondary vegetation species and disturbed sites (^{Hallé et al., 1978}). The Attims model was associated with C. schiedeana, with a main axis and orthotropic branches, with continuous growth and branching, and lateral inflorescences.

The model of L. acapulcense resembled the one proposed by ^{Vester (2002)} for L. latisiliquum (the Troll model). The Leeuwenberg model representing T. chrysantha has been attributed to T. rosea (^{Borchert and Tomlinson, 1984}; ^{Vester, 2002}). However, ^{Vester (2002)} attributes the Roux model to C. pallens, which differs from the Attims model of C. schiedeana by the horizontal orientation of its branches.

Tree architecture is the result of the influence of ontogenetic and morphogenetic factors affecting all levels of the organization of the organism, in each stage of its development and during its entire lifetime (^{Barthélémy and Caraglio, 2007}). Thus, the space-time combination of the nature and condition, life cycle and habitat determine the various types of branching (^{Weberling et al., 2002}). However, the architecture of the trees examined in this study differed from that of the trees growing in undisturbed sites because the competition conditions in agricultural soils are different from those existing in vegetal communities.

In agroforestry there is a wish to manage and shape the shared woody components, rather than to compete for the environmental resources (^{Huxley, 1999}). For this reason, it is necessary to group the utilized woody plants based on their functionality, for which purpose those shapes that best adapt to a particular cirumstance in order to optimize production must be take into account (^{Donald, 1968}). The three species considered in the present research may be regarded as competent and complementary ideotypes, whose structural attributes are compatible with other cultivated plants and with the livestock under the particular conditions of the studied sites.

All three taxa exhibited adequate phenological qualities and tree architecture for inclusion in agroforestry systems. The presence of leaves in L. acapulcense and C. schiedeana during the rainy season (summer) does not interfere with the development of rainfed crops that are highly important in the region (corn and beans), because they have small leaves that ensure less interference to the passage of light. The attributes of the tree architecture of C. schiedeana --its narrow crown, its leaf density index (1.75 m² m^-3) and the soil covered by its canopy (only 7.5 % less than in full sunshine)-- are useful for establishing this species in living fences around pastures.

Lysiloma acapulcense --having leaflets and small subleaflets, horizontal branches that improve the capture of light (^{King and Maindonald, 1999}) and a lower leaf density during the rainy season (1.17 m² m^-3)-- provides an equal or larger cover to the soil under its canopy than under the full sunshine (24.6 %), a feature that distinguishes it from the other two species and which is probably due to the fact that it belongs to the Fabaceae family. Thus, it can be exploited as dispersed trees in pastures, in association with grasses or other fodder taxa, as it favors their development due to the symbiosis established by Fabaceae with soil bacteria (^{Bloom, 2015}). Tabebuia chrysantha --with a higher leaf area index (2.01 m² m^-3) than the other species, its influence on the soil cover (55.8 % less under the canopy) and its broad crown-- is ideal for establishment in living fences, as boundary trees for timber and as dispersed trees in pastures to provide shade for the livestock and protection from water currents and sources (^{Beer et al., 2003}).

Conclusions

The three arboreal species studied herein have a phenology and an architecture that confer upon them the potential to be planted in agroforestry systems, with some differences that involve assigning them different spaces and functions within these systems. The attributes of L. acapulcense make it best suited to be associated with pastures or other crops, dispersed or with a spatial arrangement. C. schiedeana and T. chrysantha may be utilized in spaces such as living fences and as boundary trees for timber so as to avoid interfering with the crops, and at the same time, they may provide environmental services, such as the embellishment of the landscape, reforestation and improvement of the soils, while contributing to the rescue of culturally important native species with value in use.