A Phytosociological Study on Andean Rainforests of Peru, and a Comparison with the Surrounding Countries

This work is a phytosociological approach to the montane rainforests of Peru with the aim of advancing on the diversity of plant communities, which we had already begun in previous research. From 364 phytosociological plots and 3389 species of the South American tropics, we have developed a cluster, using the Sørensen index, to know the similarities between the forests and their parallelism with bioclimatic conditions. After studying the existence of characteristic groups of the Peruvian forests, we have established different communities and phytosociological units for Peru. As a result, we have described seven associations, within three new alliances, which are gathered in the new order Saurauio peruvianae-Condaminetalia corymbosae of the new class Morello pubescentis-Myrsinetea coriaceae. In addition, two associations have been described within the class Pruno rigidae-Oreopanacetea floribundae (mesotropical laurel-like forests), and three for the class Alnetea acuminatae (alder forests and palm groves). The humid forests of Peru are closer to those of Ecuador and to those of the set formed by the three Colombian mountain ranges than to those of Bolivia and Argentina, due to the common flora these share with areas of Paraguay and even of the Parana River region.


Introduction
Tropical montane rainforests are located worldwide in areas where east-west winds (Trade Winds) blow against the mountains [1,2], establishing well-defined vegetation belts that depend on rainfall and humidity condensation at medium elevations.
In South America, winds from Amazonia push the evapotranspiration humidity to the Andean forests. The decrease in temperatures causes condensation of humidity, and consequently downpours, particularly between 1000 and 3500 m, producing forests of enormous diversity [3,4] between Venezuela and NW Argentina. They depend on rainfall oscillation of between 500 and 600 mm above 3500 m and more than 5000 mm at 600 m above sea level [5][6][7].
Group L represents the forests studied throughout Peru. Here, we can distinguish several subgroups: L1 are hyperhumid and ultra-hyperhumid thermo-mesotropical forests from Southern Peru, L2 are forests ranging from humid to hyperhumid and thermo-to mesotropical from Northern and Central Peru, L3 encompasses the subhumid to humid meso-and supratropical forests and shrub formations from Northern Peru, where Oreopanax eriocephalus Harms and Baccharis latifolia are respectively constant, subgroup L4 are the Andean alder forests that grow on hydromorphic soils with constant moisture, although they share a bioclimate that ranges from subhumid to humid and thermoand mesotropical, L5 are high Andean hyperhumid meso-and supratropical forests from the South of Peru with Polylepis incarum (Bitter) M. Kessler & Schmidt-Leb. and Hesperomeles ferruginea, and finally, branch L6 is a unique plot containing Muntingia calabura L. and Hura crepitans L., a flooding forest within a subhumid infratropical bioclimate.
With group M, we return to the forests of the central Andean mountain range of Colombia, but here, there are the infra-and thermotropical humid and hyperhumid forests. These are located next to those of Peru due to common species such as Disterigma acuminatum Using the Sørensen coefficient, a numerical similarity analysis of all forests for each country is presented in Table 1. Under each country name, the number of plots and their alpha diversity are presented. Peruvian forests are more similar to those of Ecuador and the three Colombian mountain ranges than Bolivian forests, because the latter have elements in common with Paraguay and the Parana region.

Describing the New Associations and Plant Communities
Forest dominated by Polylepis incarum, humid supratropical, on rocky and stony areas, with a 40% slope and an average tree canopy of 10 m, spreading through the south of Peru.
Smallantho parvicipitis-Oreopanacetum eriocephali ass. nov. Hyperhumid mesotropical forests, with Oreopanax eriocephalus, installed on a steep slope (60%) and with a forest canopy of about 10 m, on deep sandy soils at the surface, susceptible to strong erosion. We have also studied this association in the San Gaban valley in southern Peru.

Holotypus: Appendix A
Iriartello setigerae-Cinchonetum micranthae ass. nov. A Table A2 Lower thermotropical ultra-hyperhumid forests very rich in Peruvian bark trees, with rainfall over 5000 mm. They are located in soft reliefs on clayey soils with a slope between 0% and 50%. The tree canopy is up to 30 m. They are spread throughout the South of Peru.

Holotypus: Appendix
Acalypho macrostachyae-Cecropietum polystachyae ass. nov. Ultra-hyperhumid infratropical Amazon forests, from the lowest parts, with a hilly relief of the Peruvian Andes (800 m above sea level). They are settled on very yellowish silty soils, with slopes of up to 50%, and a tree canopy of about 30 m. The presence of the palm Chelyocarpus ulei indicates the Amazonian character of this forest, as does Heliconia rostrata and wild forms of Theobroma cacao L.
The ultra-hyperhumid montane forests of southern Peru are joined together in the Serpocaulo dasypleuronis-Alchorneion latifoliae all. nov. (Appendix A, Table A1, orange square)-holotypus alliance: Iriartello setigerae-Cinchonetum micranthae ass. nov.; diagnostic species: Alchornea latifolia Sw., , and in the new phytosociological class Morello pubescentis-Myrsinetea coriaceae cl. nov., whose diagnostic species are those of the order. Both order and class may exceed the territory of Peru, judging by the species common with Venezuela, Colombia, Ecuador, Bolivia, and Argentina.

Relationships among South American Montane Rainforests
According to Table 1, the mountain forests of Peru bear a greater resemblance from a floristic point of view to those of Ecuador and to the sum of the entire Colombian Andes than to those of Bolivia. However, as shown in the dendrogram in Figure 1, the forests of Peru form a well-defined unit, probably because they are confined between the Huancabamba depression in the north and the Abancay deflection in the south [55]. The depression of Huancabamba separates the north from the center of the Andes, while in the deflection of Abancay, the eastern cordillera begins a granitic arch from the basin of the Urubamba River towards the South [56], with the characteristic forests of the alliance Serpocaulo dasypleuronis-Alchorneion latifoliae all. nov. However, many characteristic species of the new class Morello pubescentis-Myrsinetea coriaceae cl. nov. are also found in Ecuador and Colombia (Appendix A, Table A1), so this class could go beyond the Peruvian Andes. On the other hand, we do not know of a phytosociological class from the humid-ultrahyperhumid Andes of Ecuador [46,57]. In the Eastern Cordillera of Colombia, the class Palicoureo leuconerae-Cybianthetea iteoides Rangel, Cleef & Arellano 2008 has been described, but without following the precepts of the Code [44]. In a previous paper [18], we had established the class Nectandro laurel-Licarietea cannellae Izco 2013 [57] in Peru, and although some of its characteristics, such as Licaria canella (Meisn.) Kosterm., Isertia laevis (Triana) Boom, or Guzmania killipiana L.B. Sm., exist in Peru, its presence is not evident in our plots.
The forests of Bolivia are different from those of Peru, due to the elements common with the forests to the south of the Amazonian basin, in Brazil, Paraguay, and Argentina, such as Inga saltensis Burkart, Juglans australis, Myrcianthes mato, Ocotea porphyria, Podocarpus parlatorei, and Schinopsis brasiliensis Engl. [47], and the forests of Argentina, even those of the Parana basin, such as Allophylus edulis

Phytosociological Units Previously Described
According to Appendix A,  [58,59]. In Ecuador, we also find Polylepis forests that belong to the same class [45], but the phytosociological units described with Polylepis incana Kunth, Polylepis pauta Hieron., and Polylepis sericea Wedd., do not follow the Code, although they clearly constitute different associations with the Peruvian ones. The eastern associations in Bolivia also constitute different communities from those of Peru, as they are dominated by Polylepis besseri Hieron. [60], absent from Peru, and Polylepis tomentella Wedd., which forms forests both in Bolivia [61] and in eastern areas of Peru [62], where other associations could be described.

Syntaxonomical Checklist for the Montane Rainforests of Peru
This scheme was ordered according to the plant communities and groups of Appendix A,

Study Area
The mountain rainforests of the western and eastern slopes of the Peruvian Andes were studied (4 • 29 34.1 S to 14 • 37 52.68 S): on the north, those located in the vicinity of the Huancabamba depression (5 • 48 0 S), and on the south, those located in the surroundings of the Abancay deflection (13 • 16 37.22 S) [55], which are joined to those of central Peru ( Figure 2).

Study Area
The mountain rainforests of the western and eastern slopes of the Peruvian Andes were studied (4°29′34.1′′ S to 14°37′52.68′′ S): on the north, those located in the vicinity of the Huancabamba depression (5°48′0 S), and on the south, those located in the surroundings of the Abancay deflection (13°16′37.22′′ S) [55], which are joined to those of central Peru ( Figure 2). The forests were studied at altitudes between 575 and 3786 m above sea level, with some of them located in lowland Amazonian forest at an elevation between 120 and 150 m. In general, they are situated in the eastern Andean mountain range, on Paleozoic granite and metamorphic rocks, although Jurassic limestone outcrops are frequent. However, in the north, Tertiary volcanism abounds in the western territories, as well as Quaternary sediments and marine volcanic sedimentary facies in the area of the Marañón River [70].
From a biogeographical point of view, Peruvian montane rainforests belong to Yungenian Province (Tropical Subandean Region, Neotropical-Austroamerican Kingdom) with a tropical pluvial bioclimate [27]. To examine altitudinal relationships between climate and forest plant communities, the bioclimatic system of Rivas-Martínez [71] was applied, using data for at least 30 years [72], from several meteorological stations as near as possible to the studied forests using the location coordinates of the forests and the stations found on Google Earth Pro© (Supplementary The forests were studied at altitudes between 575 and 3786 m above sea level, with some of them located in lowland Amazonian forest at an elevation between 120 and 150 m. In general, they are situated in the eastern Andean mountain range, on Paleozoic granite and metamorphic rocks, although Jurassic limestone outcrops are frequent. However, in the north, Tertiary volcanism abounds in the western territories, as well as Quaternary sediments and marine volcanic sedimentary facies in the area of the Marañón River [70]. From a biogeographical point of view, Peruvian montane rainforests belong to Yungenian Province (Tropical Subandean Region, Neotropical-Austroamerican Kingdom) with a tropical pluvial bioclimate [27]. To examine altitudinal relationships between climate and forest plant communities, the bioclimatic system of Rivas-Martínez [71] was applied, using data for at least 30 years [72], from several meteorological stations as near as possible to the studied forests using the location coordinates of the forests and the stations found on Google Earth Pro© (Supplementary Table S1). We used this method because it precisely reflects the correspondence between bioclimatic belts and vegetation associations.

Plots and Flora of Peru
As it is very difficult to calculate the minimum phytosociological area in a mountain rainforest to carry out vegetation plots, we have established plots of 100 m 2 , as proposed by Dengler [74], up to 0.1 ha, as in Gentry's methods, according to the forest complexity from a subhumid area to an ultra-hyperhumid area. As Gentry pointed out, the 0.1 ha transect method is ideally suited to collect data from multiple sites, in order to generate comparative data on the taxonomic composition in rainforests [75].
Plant names were updated using the database The Plant List [90].

Relationships among South American Rainforests
Taking a total of 364 plots and 3389 species, we built a matrix (Supplementary Table S2), in which 106 plots were carried out by the authors in Peru, and 258 came from the bibliography: 14 from Venezuela [38], 96 from Colombia [41][42][43][44], 103 from Ecuador [45,46], 24 from Bolivia [47], and 21 from Argentina [48] (Figure 2). To obtain a synthetic matrix (Supplementary Table S3) grouping species, we performed a cluster analysis with the unweighted pair-group average (UPGMA) using the Sørensen index [91] in order to observe the similarity between the columns and their linkage with precipitation intervals and bioclimatic belts. To find a numerical similarity between the mountain rainforest flora, we synthesized all plots into a single column per country with JUICE software [92], and then computed the Sørensen index. UPGMA and Sørensen index were performed using PAST 4.03 [93] software.

Vegetation Classification
We used the phytosociology method of Braun-Blanquet [30] for vegetation classification, the aim of which is to define vegetation units by grouping plots with similar species compositions together and arrange these units into a hierarchical system for comparing the qualitative and quantitative floristic compositions of different geographic spaces.
To group characteristic species into phytosociological units, and to identify fidelity among plant species on plots from South America, we used JUICE software [92], with the phi coefficient as a fidelity measure [94]. This coefficient is a standard method in phytosociological studies because the phi coefficient is independent of the number of plots in the dataset. JUICE standardizes all plot groups to an equal size, and we introduced the conditions of >30% of frequency percentage for each species, including a phi measure of >0.2. Species whose concentration in groups was not significant at p < 0.01 were disregarded [95].
The phytosociological names of the vegetation units are given according to the International Code of Phytosociological Nomenclature [96].