Structure and Diversity of the Migration Habitats of Quetzals (Pharomachrus mocinno, Trogonidae) in Chiapas, Mexico

Abstract

Pharomachrus mocinno breeds in the cloud forests of the El Triunfo Biosphere Reserve, and migrates annually for six months to elevations of 900–1600 m. On the Gulf slope, temperate forests were identified as habitats for migration, but the forests on the Pacific slope have not been similarly described. In this study we described the emergent properties and phenological behavior of the plant communities of five sites identified as migration habitats, in order to test if the number of fruit-bearing species is related to the migration period. At each site, 10,000 m² was sampled, for which PBH (perimeter at breast height) and the height of shrubs and trees were annotated, including the number of palms and ferns included. We identified 25 orders, 41 families, 71 genera, and 94 species; 86.6% of these species produce fleshy fruits or fruits with modified structures that are eaten by Quetzals. During the migration period, 25–43% of these species have fruits. Eight woody species included 49% of the total individuals, which produce Quetzals’ feeding resources. The sites differed in vertical structure, composition and diversity levels. The rarefaction curve indicated that the upper site (1600 m) required more sampling. We identified three plant communities that were distributed either in montane rain forest or in the temperate forest. Since nearly 84% of the plant species are listed in the IUCN (International Union for Conservation of Nature), these forests have an intrinsic importance. The number of fruit-bearing species did not differ between migration and breeding seasons (X² (1, N = 76) 0.57; p = 0.32. Lauraceae did not stand out for the number of fruit-bearing species in any of the migration sites.

1. Introduction

Pharomachrus mocinno (De la Llave 1832) is an emblematic native bird species of the Mesoamerican highland wet–cold forests [1]. The common name of Quetzal was given to this bird by pre-Columbian cultures [2] and in this article this original name is used. Presently, this species is distributed throughout northern Panama, crossing into Costa Rica, Nicaragua, Honduras, El Salvador, Guatemala, and southern Mexico [3]. In the last century, Skutch [4] studied Quetzals in Costa Rica and warned that the destruction of cloud forests (breeding habitats) might become a threat to their long-term survival. In the Mexican state of Chiapas, it was estimated that over 30 years (1970–2000) nearly 70% of the populations were locally extinct [5]. This study concluded that forest loss and forest fragmentation were the two main threats to the long-term survival of this bird. In addition, most of the deforested habitats of the Quetzals have been transformed into lands with agricultural and livestock activities [5]. The main agricultural activities identified were plantations of coffee (Coffea L.) and maize (Zea mays L.) [6]. This land transformation has diminished the number, size, and connectivity of the remnant habitats of Quetzals, while also causing recent genetic isolation [7]. Moreover, illegal harvesting and poaching have been documented as activities that impact natural populations [8]. The ultimate effects of the loss and fragmentation of forests are a decrease in the geographic distribution, number of populations, and size of the remnant populations [2]. Accordingly, these geographic and ecological parameters influence the survival of Quetzals. Presently, the International Union for Conservation of Nature (IUCN) recognizes [9] that P. mocinno is declining in population size and geographic distribution, and it is listed as a near-threatened species; however, the federal normative of Mexico considers this species in danger of extinction [10].

Since the Quetzal is an altitudinal migratory species, its annual reproductive cycle is divided into reproductive and migratory seasons [11]. The altitudinal migration of this bird was attributed to changes in the availability of the fruits of Lauraceae. In Costa Rica, Wheelwright [12] documented that fruits eaten by Quetzals come from 41 plant species, and 48% of them belonged to the family Lauraceae; moreover, 81% of the items carried by parents to nestlings were lauraceous. This author proposed that the scarcity of fruits of Lauraceae in breeding areas caused Quetzals to move to other areas and seek out avocados.

In Mexico, Solórzano [13] examined the relationship between changes in the abundance of fruits and Quetzals. The results showed that Quetzal abundance was correlated with the scarcity of fruits of 32 species grouped into 16 families; however, these were not only lauraceous fruits. This author also analyzed the phenological behavior of seven species documented in one migratory site, and the abundance of fruits was not correlated with the Quetzals’ migratory season [13]. Presently, it has been documented that most of the Quetzals’ diet is composed of fleshy-fruits like drupes (e.g., Lauraceae) and figs (Moraceae) [4,12,13,14,15,16,17], as well as berries (e.g., Solanaceae), and other fleshy modified structures such as arils [13], and also the red fleshy receptacle of the dry cones and the young leaves of the gymnosperm Podocarpus matudae Lundell [2]. This bird also consumes vertebrates such as lizards and frogs [4,12,13,14,15,16,17], and invertebrates of seven arthropod orders including beetles and grasshoppers [4,12,13,14,15,16,17]; butterflies [13]; and terrestrial snails [17]. This bird swallows these animals whole without first tearing them into small pieces. This wide spectrum of feeding resources suggests that fruit scarcity may not be a unique factor that causes altitudinal migration. According to these studies, detailed future studies are necessary in order to clearly determine the factors and processes that drive the altitudinal migration of this omnivorous bird species [2].

On the other hand, telemetry studies carried out in Monteverde, Costa Rica [11], Sierra de las Minas, Guatemala [14], and the El Triunfo, Mexico [13,15], documented specific areas and periods of migration. Moreover, these studies recorded altitudinal movements from the breeding territories to the lower surrounding areas. These areas had different types of vegetation, such as montane wet and moist forests [11,13,14], temperate forests and other types of habitats not clearly identified but reported as being dense forests like subtropical humid–cold forests [13,14,15,16]. In Mexico, telemetry studies were carried out on the core area I of the El Triunfo Biosphere Reserve [12,13,15], which is located on the Sierra Madre of Chiapas (Figure 1a).

Chiapas is the southernmost Mexican state, which is recognized for its high levels of biodiversity, particularly of vascular plants, vertebrates and butterflies [18], including 19 types of vegetation [19], 10 of which are located on the El Triunfo Biosphere Reserve [20]. The ecosystem benefits of Sierra Madre of Chiapas spread along its nearly 600 km in length. For example, the two main products exported from Chiapas (bananas and coffee) are produced in the regions of Costa and Socunusco [21], lowlands irrigated by water that descends from the highlands.

In Mexico, Sierra Madre records the highest production of Arabica coffee [22]. This production benefits diverse types of owners, whether communal lands and private farms, as well as small villages and larger urbanized towns. However, the lack of a political system that regulates these economic activities and human settlements has caused a massive loss of original vegetation, which has reduced the habitats of Quetzals on the Sierra Madre of Chiapas [5]. The migration habitats located on the Gulf slope were represented by temperate forests of Liquidambar, Pinus, and Quercus, as well as mixed forests of Pinus-Quercus-Liquidambar, and Pinus-Quercus [13]. On the Gulf slope, some migratory areas located between 900 and 1200 masl were mostly transformed into coffee plantations and the original vegetation was lost [6]; however, some remaining standing trees like pines, oaks and Ulmus mexicana Planch., indicate that the past vegetation was similar to the surrounding temperate remnant forests (SS personal observations).

In contrast, on the Pacific slope, migration habitats were covered by subtropical vegetation [13], that some authors [5,20] have proposed is similar in part to the montane rain forest described by Breedlove [19]. We used the classification of vegetation types that this author proposes for Chiapas, since other classifications (e.g., [23]) are too broad and imprecise for the ultimate purpose that we wish to accomplish, which is to contribute to the knowledge of migration habitats. This author [19] considers that the vegetation in the entire state of Chiapas is technically subtropical since this state is above a latitude of 14° N. In addition, he detected that the most common vegetation classification systems used for Mexico [23] do not include several vegetation types unique to this state [19]. For Chiapas, Breedlove [19] distinguished 4 optimum vegetation formations and 19 seasonal vegetation formations. Seasonal formations were defined as transitional vegetation types found between the optimum formations. Based on the general structural characteristics, genera composition, and elevation range, we propose that our study sites were covered by the montane rain forest sensu Breedlove [19]. For this author, montane rain forest occurs on both slopes of the Sierra Madre of Chiapas, ranging from 900 to 2200 m, and the vertical strata is clearly discontinuous and is often composed of 2–3 strata: (1) a canopy of 25–35 m, (2) a second layer of 5–15 m; and (3) a dense shrubby understory.

For our study sites there is, however, a previous floristic inventory [24], in which all plant forms that occurred on and adjacent to the mule trail that crosses from the Gulf slope, crosses the peaks of the Sierra and descends toward the Pacific slope were sampled. Along this trail, a total of seven plant communities were identified, and four of them were identified solely on the Pacific slope (Figure 3 of [24]). These four distinct plant communities were composed of (1) Cupressus-Pinus (1700–1500 masl), (2) Ficus-Coccoloba-Dipholis-Sapium (1500–1200 masl), (3) Garcinia-lnga-Oesmopsis (1300–1000 masl), and (4) Quercus salicifolia (1100–1000 masl) [24]. Plant communities 2 and 3 were closest to two of our study sites; however, none of the taxa that defined those communities were recognized as key components of the montane rain forest identified by Breedlove (page 12 of [19]). Moreover, the floristic inventory [24] did not provide details of how such plant communities were identified (e.g., relative importance value, basal area, abundance); however, it is a unique floristic reference for our study area. Thus, in this context, it is even more relevant to perform analyses at the level of community ecology in specific geographic areas. We predict that the number of fruit-bearing species is highest during the migratory season. In addition, we expect that our study will contribute to knowledge about the vegetation characteristics of the migration habitats used by the Quetzals, and secondarily also be a contribution to the poorly documented vegetation of the central part of the Sierra Madre of Chiapas. The objective of the present study was to describe the structural traits, taxonomic diversity, and composition of the vegetation at five sites used as migration habitats by Quetzals on the Pacific slope of the core area I of the El Triunfo Biosphere Reserve.

2. Materials and Methods

2.1. Study Sites: The El Triunfo Biosphere Reserve

The five study sites were located on the Pacific slope of core area I of the El Triunfo Biosphere Reserve (Figure 1). This reserve was decreed in 1991 and is located on the central portion of the Sierra Madre of Chiapas. This reserve includes 119,177 hectares distributed into five core areas or polygons, surrounded by a buffer zone [25]. Sierra Madre is one of the seven physiographic regions recognized in Chiapas. It is nearly 280 km in length, and the mean elevation is 2000 m; however, the elevation in its northern portion is only 25 masl, but in the southernmost portion it is nearly 4,030 m because of the Tacaná volcano. In addition, this mountain chain is riddled with valleys, canyons, and terrains that are steep and rugged [26].

There are no weather stations in the study sites; the closest one is located in the cloud forest at 2000 masl, and another one is on the Gulf slope of the Finca Prusia. The Gulf slope is drier than the Pacific slope; however, these two localities show similarities to our study sites in periods of rain and the short relative dry season (Figure 2). Our study sites show temperatures slightly warmer than those of cloud forests; however, daily, dense fog is common. To illustrate the behavior of the variations in temperature and rain at our study sites, we present a climograph, which was created with data recorded at the El Triunfo weather station on the cloud forests. These data were recorded daily for 19 years (1991–2010) by the reserve’s forest rangers.

Fieldwork was carried out during January 2010, after the hurricanes Mitch (1998) and Stan (2005) had occurred. These hurricanes caused abundant rain that swept along debris, trees, and large rocks and logs from the upper montane elevations to towns and small villages located on the sides of the Sierra Madre. The severe effects of these hurricanes were recorded in the lowlands, where considerable types of damage were recorded, as well as considerable disturbances to farming and plantations [27]. In our study sites, these hurricanes did not cause severe effects on the canopy and sub-canopy; however, it seems that the terrain was flooded and the herbaceous layer mostly disappeared (SS personal observations). Therefore, we expect that the vertical strata were not severely altered and consequently the emergent properties of the plant communities reported here reflect the characteristics of the habitats used by the Quetzals.

2.2. Sampled Vegetation from the Perspective of Ecological Communities

The five study sites were specific areas where the Quetzals migrated [13] (Figure 1b). In each of the five sites, four plots of 50 × 50 m were sampled; thus, a total area of 10,000 m² per site was sampled. We sampled trees, shrubs, vines and palms, since Quetzals feed on them [11,12,13,14,15,16,17]. The dense shrubby understory stratum was composed mainly of ferns and palms that were only accounted for per plot. During the fieldwork for each woody plant, we recorded its perimeter (P) at breast height (PBH) and its height in meters. For trees, PBH was measured at 1.5 m above the ground, and for shrubs, it was measured at their base; these measurements were transformed to estimate the diameter (D = P/π) per individual and the basal area per species [28]. The height of plants was estimated with a collecting pole for heights up to 5 m, and for those trees of higher heights, a compass with a clinometer was used. Specimens of each different species were collected for their posterior taxonomic identification.

2.3. Community Structure and Altitudinal Changes in Species Composition

Since the plant communities were stratified, we characterized the vertical structure into five categories of height (<3, 3.1–6.0, 6.1–17, 17.1–26; and >26.1). In addition, Gotelli [29] proposed that the term “community structure” mainly involves three emergent properties of the community: richness, composition, and ecologic diversity, which were estimated using the procedures described in Magurran [28]. Ecologic diversity was estimated with the Shannon–Weiner index (HS = −∑pilnpi), and the statistical differences in the HS per site were estimated with a t-test [28]. A rarefaction species curve based on abundances was obtained using the package implemented in the software Past 5.1 [30]. Species dominance was measured with the Simpson index (D = ∑pi2) and the equitability or evenness index (i.e., a measure of how similar species of a community are in their abundances, [28]) was based on the Shannon index (EH = HS′/lnS). The differences in species composition among sites was compared using two procedures: the Sorenson quantitative index (SQI) and the dissimilarity index of Bray–Curtis (BC). The SQI solely considers the number of shared species between two communities. In contrast, BC considers the relative abundance in the comparison of species composition, and this was estimated using the package for ecological analysis implemented in the Past 5.1 software [30]. This coefficient varies from zero (total similarity, or the same species composition) to one (total dissimilarity). The estimated dissimilarity matrix (BC) was transformed with the Unweighted Pair Group Method using Arithmetic averages UPGMA [30] to cluster the sites. Finally, we estimated species turnover between with the Whittaker index (βW = γ/α − 1) [31] to explore changes at the spatial scale or among sites.

3. Results

3.1. Species Richness and Composition in the Whole Study Area

Adding the data of the five sites resulted in an estimated Shannon index of 3.4, and the Simpson index indicated that in the whole studied area there was not one dominant species (D = 0.07). With respect to taxonomic composition, 25 orders were documented that included 41 families, 73 genera and 94 species. The woody plants (shrubs and trees) included only 23 orders with 39 families, 70 genera and 91 species (

Table 1. Plant taxonomic composition documented in the study area arranged by family name. The taxonomic authority was referred according to World Flora Online [32]. For those species included in the risk category according to IUCN [33], data are presented in the last column: NT = near threatened; EN = endangered; LC = least concern; VU = vulnerable.

Family	Species	Risk Category
Woody-taxa
1. Acanthaceae	1. Aphelandra schiedeana Schltdl. & Cham.	NT
	2. Spathacanthus parviflorus Leonard	EN
2. Actinidiaceae	3. Saurauia kegeliana Schltdl.	LC
3. Apocynaceae	4. Alstonia longifolia (A.DC.) Markgr.	LC
4. Araliaceae	5. Dendropanax pallidus M.J.Cannon & Cannon	VU
	6. D. populifolius (Marchal) A.C.Sm.	EN
	7. Oreopanax echinops Decne. & Planch.	LC
	8. O. peltatus Linden	LC
	9. O. sanderianus Hemsl.	LC
	10. O. xalapensis Decne. & Planch.	LC
5. Asteraceae	11. Koanophyllon pittieri (Klatt) R.M. King & H.Rob.	LC
	12. Pseudogynoxys chenopodioides (Kunth) Cabrera
	13. Senecio cobanensis J.M.Coult.	NT
6. Berberidaceae	14. Berberis hemsleyi Donn.Sm.	LC
7. Burseraceae	15. Bursera simaruba (L.) Sarg.	LC
8. Cannabaceae	16. Celtis caudata Planch.	LC
	17. Trema micranthum (L.) Blume	LC
9. Capparaceae	18. Crateva tapia L.	LC
10. Celastraceae	19. Crossopetalum standleyi (Lundell) Lundell
	20. Zinowiewia matudae Lundell	EN
11. Clethraceae	21. Clethra mexicana DC.	LC
	22. C. occidentalis (L.) Kuntze
12. Euphorbiaceae	23. Acalypha macrostachya Jacq.	LC
	24. Croton reflexifolius Kunth	LC
	25. C. xalapensis Hook.f.	LC
13. Fabaceae	26. Calliandra houstoniana Standl.	LC
	27. Inga calderonii Standl.	LC
	28. Pithecellobium Mart
14. Fagaceae	29. Quercus candicans Née
	30. Q. ocoteifolia Liebm.
	31. Quercus L.
15. Clusiaceae	32. Clusia salvinii Donn.Sm.	LC
16. Hamamelidaceae	33. Matudaea trinervia Lundell	LC
17. Lauraceae	34. Beilschmiedia hondurensis Kosterm.	LC
	35. Licaria capitata (Cham. & Schltdl.) Kosterm.	LC
	36. Damburneya patens (Sw.) Trofimov	LC
	37. D. salicifolia (Kunth) Trofimov & Rohwer	LC
	38. D. salicina (C.K.Allen) Trofimov & Rohwer	LR
	39. Nectandra reticulata Mez	LC
	40. Ocotea botrantha Rohwer	LC
	41. O. macrophylla Kunth	LC
18. Malvaceae	42. Malvaviscus lanceolatus Rose
19. Melastomataceae	43. Conostegia volcanalis Standl. & Steyerm.	LC
	44. Miconia argentea (Sw.) DC.	LC
	45. M. glaberrima (Schltdl.) Naudin	LC
20. Moraceae	46. Ficus aurea Nutt.	LC
	47. Trophis chiapensis Brandegee
	48. T. mexicana (Liebm.) Bureau	LC
21. Myrtaceae	49. Myrcia chytraculia var. pauciflora (O.Berg) G.P.Burton & E.Lucas	LC
	50. Eugenia capuli (Schltdl. & Cham.) Hook. & Arn	LC
	51. E. capulioides Lundell	VU
22. Pentaphylacaceae (Theaceae)	52. Cleyera theoides (Sw.) Choisy	LC
	53. Ternstroemia lineata subsp. chalicophila (Loes.) B.M.Barthol.	LC
23. Piperaceae	54. Piper hispidum Sw.	LC
	55. P. sanctum (Miq.) Schltdl. ex C. DC.	LC
24. Podocarpaceae	56. Podocarpus matudae Lundell	VU
25. Polygonaceae	57. Coccoloba montana Standl.	NT
26. Primulaceae (Myrsinaceae)	58. Ardisia compressa Kunth	LC
	59. A. mexicana subsp. siltepacana (Lundell) Pipoly & Ricketson	LC
	60. Gentlea tacanensis (Lundell) Lundell	EN
	61. Parathesis chiapensis Fernald	LC
	62. P. multiflora Lundell	EN
	63. P. subulata Lundell	EN
27. Proteaceae	64. Roupala montana Aubl.	LC
28. Rosaceae	65. Prunus tetradenia Koehne	LC
29. Rubiaceae	66. Arachnothryx laniflora (Benth.) Planch.	LC
	67. Deppea inaequalis Standl. & Steyerm.
	68. Faramea occidentalis (L.) A.Rich.
	69. Glossostipula concinna (Standl.) Lorence	LC
	70. Hoffmannia psychotriifolia (Benth.) Griseb.
	71. Palicourea hebeclada (DC.) Borhidi
	72. Psychotria costivenia Griseb.	LC
	73. P. subsessilis Benth.	LC
	74. Psychotria L.
	75. Randia aculeata L.	LC
	76. Rogiera cordata (Benth.) Planch.	LC
30. Salicacea	77. Casearia corymbosa Kunth	LC
31. Sapindaceae	78. Dodonaea viscosa (L.) Jacq.	LC
	79. Matayba Aubl.
	80. Talisia macrophylla (Mart.) Radlk.	LC
32. Sapotaceae	81. Chrysophyllum mexicanum Brandegee	LC
	82. Sideroxylon portoricense Urb.	LC
33. Simaroubaceae	83. Picramnia antidesma Sw.	LC
34. Solanaceae	84. Cestrum luteo-virescens Francey
35. Styracaceae	85. Styrax glabrescens Benth.	LC
36. Symplocaceae	86. Symplocos hartwegii A. DC.	NT
	87. S. limoncillo Humb. & Bonpl.	LC
37. Thymelaeaceae	88. Daphnopsis selerorum Gilg	LC
38. Tiliaceae	89. Luehea seemannii Triana & Planch.	LC
39. Urticaceae	90. Phenax hirtus (Sw.) Wedd.
	91. Urera caracasana (Jacq.) Gaudich. ex Griseb.	LC
Not-woody taxa
40. Arecaceae	92. Chamaedorea elegans Mart
41. Polypodiaceae	93. Campyloneurum C. Presl
	94. Pleopeltis Humb. & Bonpl. ex Willd.

, Figure 3), and non-woody plants (1516 individuals) were represented by 3 genera of palms and ferns (Table 1).

The highest taxonomic diversity was identified in the order Ericales that included the largest number of taxa (Figure 3). This order included 7 families (Actinidiaceae, Clethraceae, Myrsinaceae, Sapotaceae Styracaceae, Symplocaceae and Theaceae), 11 genera and 16 species. The families with the largest number of species were Rubiaceae (11), Lauraceae (8), and Araliaceae and Myrsinaceae (6). Nineteen families were represented by one genus; and fifty-five genera included only one species (Table 1, Figure 3).

The 91 woody species included 3314 individuals that were not evenly distributed among these species (EHs = 0.15); in fact, only eight species from six families represented 48.5% of the total individuals recorded across the five sites. The most abundant species were Psychotria costivenia (10.20%, Rubiaceae), Symplocos limoncillo (8.19%, Symplocaceae), and Eugenia capuli (7.89%, Myrtaceae). In contrast, two species were singleton (Pithecellobium (Fabaceae) and Chrysophyllum mexicanum (Sapotaceae)) and eight taxa were doubleton (Arachnothryx laniflora (Rubiaceae), Bursera simaruba (Burseraceae), Clusia salvinii (Clusiaceae), Luehea seemannii (Tiliaceae), Oreopanax peltatus (Araliaceae), Quercus candicans (Fagaceae), Styrax glabrescens (Styracaceae), and Ternstroemia lineata subsp. chalicophila (Theaceae)). Across the five sites, the highest values of relative importance were estimated for six species: Psychotria costivenia (18.4%), Eugenia capuli (16%), Quercus ocoteifolia (13%), Daphnopsis selerorum (12%), Symplocos limoncillo (10.3%), and Ardisia compressa (10%).

Across the five study sites, the highest trees (>26 m) belonged to the species Coccoloba montana, which was not recorded in Limonar and Sicilar, Ficus aurea (only recorded in Cañada Honda), Podocarpus matudae that was not recorded in Cresta, and Oreopanax xalapensis, which was recorded in the five sites, whereas the species with the largest basal areas were F. aurea, Oreopanax sanderianus (not recorded in Sicilar), O. xalapensis and Quercus ocoteifolia, which was absent in Sicilar (Appendix A).

3.2. Ecological Characteristics per Site

The results showed that the studied sites were similar in terms of the number of taxa, though the highest values were recorded in Sicilar, which was the upper site (1600 masl). A pattern in diversity levels was identified, since diversity estimators tended to increase with the elevation gradient, and Sicilar accounted for the highest number of species (62) and Limonar the lowest number (46) (

Table 2. Number of taxa recorded and ecological indices of diversity, dominance and equitability (evenness) estimated per site.

Descriptor/Site	Limonar	Mirador	Cresta	Cañada Honda	Sicilar
Richness	46	55	60	60	62
Family	29	29	33	33	32
Genera	41	42	48	48	50
Density per site	1019	1210	912	1010	679
Ecologic diversity	2.95	2.83	3.11	2.96	3.36
Equitability	0.77	0.70	0.76	0.72	0.81
Dominance D	0.075	0.12	0.091	0.10	0.057

). However, the rarefaction curve indicates that the Sicilar could still have a larger number of species (Figure 4). In addition, Mirador was denser than the other four sites; in particular, it was 1.78 times denser than Sicilar (Table 2).

The palms and ferns contributed to the dense structure of the understory in each of the five sites; however, ferns of the genus Campyloneurum were not recorded in Limonar (Appendix A), which was the lowest sampled site (~1000 masl). The three most abundant species differed among the sites: in Limonar these were S. limoncillo (0.12), P. costivenia (0.10) and E. capuli (0.07), in Mirador they were S. limoncillo (0.06), E. capuli (0.06), and A. compressa (0.05), in Cresta they were P. costivenia (0.09), S. limoncillo (0.08), and D. selerorum (0.06), in Cañada Honda they were Psychotria subsessilis (0.15), D. selerorum (0.04), and Rogiera cordata (0.04), and in Sicilar they were P. costivenia (0.09), Matudaea trinervia (0.09), and E. capuli (0.07). With respect to relative importance, at the Limonar, Mirador and Cresta sites, the species Daphnopsis selerorum, Eugenia capuli, Symplocos limoncillo and Psychotria costivenia showed the highest values (20–21%), but in Cañada Honda, the values of Oreopanax xalapensis (36%), Psychotria subsessilis (29%) and Saurauia kegeliana (25%) were highest, whereas at Sicilar the levels of Psychotria costivenia (29%), Matuda trinervia (24%) and Podocarpus matudae (18%) were highest. Interestingly, most of these species (76) produce fleshy fruits or fruits with modified fleshy structures that the Quetzals eat (Appendix B). However, the number of fruit-bearing species was similar (X² (1, N = 76) 0.57; p = 0.32) between the periods of migration (59 species) and breeding (56) (Appendix B).

On the other hand, the vertical structure showed differences among the sites: Cañada Honda and Sicilar were more similar to each other, and Mirador was similar to Cresta; however, Limonar showed a shorter forest (Figure 5). In the sites, the highest proportion of plants were woody individuals in the juvenile stages (<3 m) of species that in the mature stage are part of the canopy and sub-canopy, whereas the second category corresponded to the juvenile stages of species that in the mature state will be in the height class of 3.1–6 m (Figure 5).

With respect to the details of the species composition, only 17 woody species from 13 families were recorded across the five sampled sites. The other 13 species were only recorded in one of the sampled sites; however, Mirador did not include any exclusive species (Appendix A). Consequently, the change in species composition was also shown by the turnover index, which estimated that nearly two new species appeared between sites (βW = 1.71). The taxa Bursera simaruba and Pithecellobium were recorded only in Cañada Honda, whereas Celtis caudata, Chrysophyllum mexicanum, Deppea inaequalis, and Matayba were exclusive to Cresta, and finally Clusia salvinii, Matudae trinervia, Miconia glaberrima, Oreopanax echinops, and Ternstroemia lineata subsp. chalicophila were recorded only in Sicilar. Among the sites, the number of singular species varied and Cañada Honda had the highest number (14), followed by Sicilar (12), Limonar (10), Mirador (9) and Cresta (7). This heterogeneity in the geographic distribution of the species indicates that the composition among the five sites differed (

Table 3. Results of the paired comparisons to test similitude in the species composition based on the quantitative Sorenson index; the number of shared taxa between sites is shown in parentheses (above diagonal). Below of the diagonal values of paired t-test (tα = 0.05, df_∞ = 1.960) to estimate significant differences in values of HS (Table 2); an asterisk indicates significant differences.

Site	Limonar	Mirador	Cresta	Cañada Honda	Sicilar
Limonar	0	0.65 (33)	0.64 (34)	0.63 (33)	0.59 (32)
Mirador	0.91	0	0.75 (43)	0.80 (46)	0.67 (39)
Cresta	1.10	1.95 *	0	0.77 (46)	0.64 (39)
Cañada Honda	0.062	0.91	1.02	0	0.68 (41)
Sicilar	2.47 *	3.29 *	1.42	2.42 *	0

, Figure 5).

The paired comparison based only on the presence/absence of the species showed that the closest geographic sites were more similar. Thus, Cañada Honda and Mirador had the highest similitude (0.80), and only 32 taxa were shared between Limonar (~1000 masl) and Sicilar (~1600 masl) (Table 3, above). In contrast, if the relative abundances of each species were considered in the Bray–Curtis index (BC) in order to compare species composition, the results showed that Mirador and Cresta were more similar, and Sicilar was the most distinct site (Figure 6).

4. Discussion

4.1. Structural Characteristics of the Migration Sites on the Pacific Slope

The results showed that the five sites used during the migration season were characterized by dense forested vegetation; however, they exhibited different structural and diversity levels. The five sites were composed of broad leaved forested vegetation; however, among them we documented differences in species composition, abundance, tree density, vertical structure and height of the canopy. In addition, in each of the five sites the most abundant species and the species with the highest values of relative importance also differed. Accordingly, the plant communities change along the altitudinal gradient studied. Based on the genera composition and the structural characteristics described in Breedlove [19], we consider that four of our studied sites (Mirador, Cresta, Cañada Honda and Sicilar) are similar to the vegetation that he described as montane rain forest, and the lowest site (Limonar) probably corresponded to a distinct type of vegetation with similarities to a semi-deciduous oak forest. Our results disagreed with the national vegetation maps (e.g., [34]), in which all vegetation ranging from 900 to 2500 masl across the entire reserve of the El Triunfo has been reported as “Bosque Mesofilo de Montaña” or as “Tropical Rain Forest”, a typical warm-humid vegetation found in lowlands < 500 masl (e.g., see reference cited in [13]). We consider that these incorrect classifications are due to the lack of fieldwork necessary to corroborate the vegetation types that are extant in the Sierra Madre region.

In the floristic inventory of Long and Heath [24], two of the plant communities they reported, (1) Garcinia-lnga-Desmopsis and (2) Ficus-Coccoloba-Dipholis-Sapium, were assigned to the vegetation type of montane rain forest, following Breedlove [19,35]. Of the four plant communities reported in [24] for the Pacific slope, three of them that ranged from 1000 to 1500 masl were close to our study sites. However, based on the relative importance values we identified distinct plant communities. In three of our sites (Limonar (~980–100 masl); (Mirador 1200–1300 masl); and Cresta (1350–1400)) the plant community identified was composed of Daphnosis selerorum-Eugenia capuli-Symplocos limoncillo-Psychotria costivenia. Cañada Honda (1440–1500 masl) was defined by Oreopanax xalapensis-Psychotria subsessilis-Saurauia kegeliana, and in Sicilar (1510–1600 masl) a third plant community was found composed of Psychotria costivenia-Matuda trinervia-Podocarpus matudae. We already mentioned that these differences might be caused by different measurement methods, but they might also be caused by the high heterogeneity that the Pacific slope has in Sierra Madre. A similar result between our study and that of Long and Heath [24] was that we documented fig species of Ficus (we reported as F. aurea and probably is the same that was reported as F. cookii [24]) as one of the tallest bulky species, which is an emblematic component of the Tropical Rain Forest according to Breedlove [19,35]; however, this author mentioned that some species of vegetation from the lowlands might be found at higher montane elevations [19] and another possible cause is that the area where Ficus species were documented (only in the Cañada Honda site) could in fact be a transitional zone between the montane rain forest and lower montane rain forest. The fig species were not ubiquitous in the study area, and the high trees over 20 m indicate that their presence is not new in these areas.

Like Cañada Honda, the lowest site (Limonar) was also a sui generis site. This site has a short forest with a semi-deciduous physiognomy. In Limonar the species with the highest relative importance values were Daphnosis selerorum-Eugenia capuli-Symplocos limoncillo-Psychotria costivenia, and though Quercus ocoteifolia Liebm. was in fifth place in this value, it defines the vegetation in this altitudinal range (900–1100 masl). In the sampled plots the vegetation had the appearance of an oak forest. However, Breedlove [19] only recognized that the genus Quercus is associated with Liquidambar and/or Pinus to form a temperate forest on the Gulf slope. However, since Liquidambar and Pinus were not documented in any of the five sampled sites, these types of temperate forests were not documented in the present study. The vertical structure grouped a large number of plants into two height classes < 6 m, which determine the physiognomy of a forest with a canopy of 15–20 m in height, and some isolated taller species over 20 m. Similarly, Long and Heath [24] reported a plant community composed of Q. salicifolia Benth (synonym of Quercus laurina Bonpl., according [32]) at altitudes of ~1000 m. Therefore, despite the nomenclatural differences between the species of Quercus, on the Pacific slope there is a temperate forest that is used as a habitat during the migration season. The temperate forest documented in Limonar was not recorded at elevations over 1500 m (Appendix A). This temperate forest provides Quetzal with feeding resources (see Appendix A and Appendix B), and this site was also the lower-most altitudinal limit that the Quetzals reached during their migration [13]. Thus, it would be interesting in future studies to analyze if those lower drier forests have plant species that provide fleshy fruits for this bird, or if there are other factors (e.g., warmer and drier environment) that restrict the altitudinal range of their migration.

On the other hand, based on the classification of Holdridge [36], our results suggested that we sampled three types of vegetation: (1) premontane moist forest (Limonar and Cresta); (2) wet montane forest (Mirador and Cañada Honda); and montane rain forests (only in Sicilar). Future studies may refine this classification of the types of vegetation, but undoubtedly our results addressed the heterogeneity in vegetation composition, structure and richness. Most of the species recorded have a biogeographic affinity to the southern subtropical forests of Central America, which might indicate that the montane rain forest is the native vegetation of Chiapas, and it has not reached higher latitudes; it is perhaps for this reason it has gone unnoticed by many Mexican specialists that have not visited Sierra Madre of Chiapas.

4.2. Threats and Possible Management Actions for Quetzals in Migration Areas

The results showed that the Quetzals migrated to forested areas that corresponded to wet–moist vegetation types. On the Gulf slope, the temperate forests of Liquidambar, Pinus, Quercus-Pinus, and Pinus Liquidambar were already recognized as migration habitats [13]. Therefore, it is important to observe that Quetzals may visit some of the coffee plantations that replaced the original vegetation. However, in these plantations there are neither the structural characteristics nor the feeding resources required. Our results showed that nearly 85% of the documented plant species produce fleshy fruits or structures eaten by Quetzals. During the months of migration (July—ending November) 25–43% of the species had fruits (Appendix B). In addition, of the eight species of Lauraceae documented in our study, only two of them, had fruits in July–August, which suggested that Quetzals visit the forests irrespective of the abundance of avocados. In addition, our results showed that in our study sites there was no difference in the number of fruit-bearing species between migration and breeding seasons. However, we consider that quantitative phenological studies are pertinent to examine available resources in breeding and migration habitats. Based on these results, we recommend that the tenants of plantations over 900 masl may support the conservation of Quetzals if they reintroduce native plant species of trees, shrubs and vines that produce fleshy fruits or fruits with modified structures such as arils and soft carpels. Another alternative action is to maintain remnant trees of native species along the plantations by using them as living-walls or to mark the boundaries between properties. The authorities of the reserve may guide the formal design of corridors composed of native plant species, with an emphasis on tree species, which might maintain the ecological connectivity among distinct breeding sites reported in other studies [5].

In the reserve, on the Pacific slope, the level of deforestation is minor compared to that on the Gulf slope [6], where intense economic activities are recorded [21,22]. However, some threats loom over the Sierra Madre of Chiapas, since the federal governments in the years 2000–2015 authorized dozens of mining concessions to exploit minerals along this Sierra [37], despite three biosphere reserves being decreed to guarantee the conservation of species, natural resources and ecosystem services [25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40]. Moreover, there are many claims to the land at the boundaries of the El Triunfo Biosphere Reserve territories [41]. In addition, our results revealed that 84% of the taxa documented are listed in the IUCN [33], and this is an additional factor in maintaining the protected territories located on the Pacific slope of the reserve. Previous studies have also addressed the relevance of the Pacific slope as a biological reservoir of emblematic species such as the rare bird the Azure-rumped tanager (Tangara cabanissi) [38], the endangered Jaguar (Panthera onca), and Geoffroy’s spider monkey (Ateles geoffroyi) [20].

5. Conclusions

Our study contributes to the knowledge about the habitats used during the migration season of Quetzals. We identified different plant communities that offer feeding resources for this bird. The new vegetation types reported here include temperate forests of oak, with abundant species that produce fleshy fruits, montane rain forest, as described by Breedlove [19], and wet–moist forests as described by Holdridge [36]. These forests provide feeding and water resources, and have architectural and structural characteristics that form a dense vegetation that offers oikos for this canopy-forest bird. Today, large portions of original forests have been transformed into deforested lands, pasturelands, and coffee or maize plantations. However, Quetzals do not seek out these transformed areas, and they maintain their behavior, seeking out forests to down to elevations of around 900 m (in Sierra Madre). Migration habitats are part of the sites used throughout the annual reproductive cycle of the Quetzals. Conservation plans have to consider migration habitats if the survival of Quetzals is to be guaranteed. In addition, these forests provide well-being and ecosystem services for the local villages and humans that inhabit the sides the Sierra Madre of Chiapas, and maintaining the original vegetation in the montane elevations also offers benefits to humans.