The scale of the task is extraordinary. Latin American highlands stretch from latitude 20 degrees North in Mexico to 55 degrees South in the Chilean-Argentinian Patagonia, forming one of the longest sub-continuous mountain chains on the planet. The 7,500-kilometer Andean cordillera alone accounts for 99 percent of the region’s 550,000 square kilometers of highlands, calculated as the surface above the potential treeline. Additional highlands rise on smaller ranges, some related to Andean orogeny, such as the Cordillera de la Costa in Venezuela, others built by isolated mountain-building events, for example the Trans-Mexican Volcanic Belt. Because these ranges compress and expand their altitudinal belts from tropical to temperate climates along a unique latitudinal gradient, Latin American mountains offer scientists an outstanding natural laboratory for testing alpine ecological theory.

Mountains matter far beyond their surface area. More than half of the human population relies on freshwater bodies with high mountain sources for water supply, and mountain landscapes, shaped by orogeny, geo-environmental factors and land-use history, are distributed along altitudinal belts running from foothills through premontane, montane, subalpine, alpine and subnival-nival zones. A pronounced change in vegetation physiognomy occurs at the timberline, typically located in the subalpine belt, where forest-dominated ecosystems give way to shrublands, grasslands and deserts. Unlike the naturally continuous forests below, mountain highlands often form archipelagos of continental biogeographic islands, a sky island pattern that makes highland floras exceptionally rich in endemics and extremely susceptible to change. Notable exceptions include the high plateaus of Tibet and the Andean Altiplano, where most of the subalpine and alpine belts remain continuous and only the highest peaks prevail.

Scientific interest in Latin American mountain ecology and biogeography has increased rapidly over recent decades, mostly because these mountains contribute to more than half of the 85,000-species pool that forms the Latin American vascular flora, and because they coincide with several noteworthy biodiversity hotspots, including the Tropical Andes, Mesoamerica, Tumbes-Chocó-Magdalena and the Chilean Winter Rainfall-Valdivian Forests. Floristic data contributions are steadily increasing and becoming available in national repositories such as SiB Colombia and on global platforms like GBIF and TRY. Vegetation data, however, remains scarce and difficult to access, even though it is uniquely advantageous, providing information on the distribution of both species and plant communities. VegAndes was conceived precisely to remediate this dramatic gap, emerging from consortium discussions involving expert botanists and vegetation scientists dedicated to highland research in Latin America.

Technically, the database structure was programmed in SQL, following the earlier VegPáramo dataset, and revolves around a primary table containing the basic vegetation plot data, connected to five secondary tables via the VegAndes plot ID. Among the extensive array of data sources consulted for Latin America, only those containing plot vegetation data expertly identified as highland or ecotonal vegetation by the data’s authors were included, and plots were not discriminated between zonal and azonal vegetation as long as they fitted the at- or above-the-timberline rule, or an equivalent in desertic areas. The primary table records species occurrence and cover, using the initial taxon names provided by the plot authors, with certainty particles such as cf., aff. and gr. carefully retained. Non-vascular plants were excluded, and all non-identified taxa above the genus level were discarded. The database supports presence-absence data as well as several cover scales, including the Braun-Blanquet phytosociological scale, transformed versions such as Van der Maarel’s, and percentage cover.

The secondary tables capture rich geo-environmental and structural detail. The geo-environmental table stores plot surface area, elevation, slope and aspect, along with geolocation in decimal coordinates and UTM grid cells in Datum WGS84. Plots precisely georeferenced with an error of 10 meters or less, as provided by modern GPS, are considered exact, while imprecise plots with errors between 10 meters and 10 kilometers are referenced to the centroid of the smallest resolution UTM grid cell available, and plots with errors above 10 kilometers were not registered at all. The vegetation-structure table reports total, vascular and non-vascular vegetation covers, bare soil, and the height and cover of vegetation strata from trees through high and low shrubs to herbs and prostrate layers. It also accounts for three classification systems, including a simple zonal-azonal scheme, the syntaxon names initially given by the plot’s authors, and informal plant community labels combining structural qualifiers with dominant or diagnostic species, for example bunchgrassland with Espeletia grandiflora and Eryngium humboldtii.

Taxonomic standardization relies on two global botanical platforms, TROPICOS as the preferential source and World Flora Online as the secondary source, used to assess each name’s validity. Whenever a taxon is classified as doubtful or unresolved by both platforms, the initial name is retained, and hybrid taxa are kept as accepted names. The taxon list also includes family names and higher clades across angiosperms, gymnosperms, lycophytes and monilophytes, and is expected to undergo annual synonymy revisions. At present, VegAndes encompasses 5,804 taxon names corresponding to 3,858 accepted names, as well as 136 syntaxon names compiled from the data sources, though the complete syntaxonomic classification has not yet been revised and is advised to be handled as highly provisional. The data-source table documents 53 primary sources, 86 when counting VegPáramo’s multi-source input, of which 11 are unpublished datasets such as private datasets and academic theses, carefully and expertly revised beforehand.

The spatial distribution of the plots reveals both the strengths and the biases of Latin American highland research. Most VegAndes data comes from the Northern Andes, almost 60 percent of the plot total, and especially from Colombia and Ecuador, which together sum up to 52 percent. The Central Andes come second with almost 23 percent, essentially from Peru, while Mesoamerica contributes just 2.3 percent and the Southern Andes 16.4 percent. Certain areas contribute disproportionately, for instance the Nevados sector in Colombia with 472 plots in 10,000 square kilometers, and to a lesser extent the Moquegua mountains in Peru with 385 plots in the same area. Differences in sampling effort are usually due to expert location and ease of access. The guerrilla conflict in Colombia substantially limited fieldwork in certain mountain areas over recent decades, many of which have only become accessible after the peace treaty of 2016, while conversely other areas are now becoming difficult of access due to civil tensions or local conflicts related to illegal extractive practices. The remaining spatial data gaps, especially in the Patagonian Andes, central Peru and Central American mountains, should be addressed and filled whenever possible.

From an altitudinal standpoint, the plot data serve as a simple proxy for the current timberline position and highland elevation range throughout Latin America. The plots show a hump-shaped pattern for the Andes, with high elevations concentrated at mid-latitudes and around the equator, and lower elevations towards the edges of the main cordillera around latitude 10 degrees North and 40 degrees South, reflecting alpine ecological theory that altitudinal belts typically occur at higher elevation and expand towards the equator. The data also manifest the biogeographical transition at latitude 5 to 7 degrees South, the Huancabamba depression barrier, with relatively low elevation plots in comparison to the high-elevation Central and Northern Andes, and a clear depression at the transition between the Northern Andes and Mesoamerica. According to the VegAndes data, the approximate timberline by proxy appears to be highest in the Altiplano region of the Central Andes and in the northernmost Mesoamerican mountains, while Central Andes plots reached elevations near the 5,000-meter isoline, presumably close to the local snowline. Highland plots are found at dramatically decreasing elevations towards the Patagonian Andes, down to a timberline at roughly 1,500 meters above sea level near latitude 40 degrees South.

The floristic content of VegAndes contains 3,858 vascular plant species across 151 plant families, dominated by Asteraceae, Poaceae, Orchidaceae, Fabaceae and Cyperaceae, which sum up 40 percent of the total floristic inventory. Compared with World Flora Online input for the same family pool at the global scale, the ranking is similar, although proportions differ, and the database provides an interesting subset of the world’s flora with an evident underlying grassland trend, illustrated by the high relative preponderance of graminoid monocots such as Poaceae and Cyperaceae. The family ranking resembles other tropical mountain floras, such as the Papua New Guinea and Afromontane highlands, and shows good representation of Fabaceae and Lamiaceae, which are very important families in the subtropical and temperate highlands. Species richness patterns are complex, averaging about 20 families per 10,000 square kilometers, with certain areas doubling that number, principally in Ecuador and Colombia, though these family-rich areas could reflect genuine phytodiversity or potential artefacts due to sampling effort or plot altitudinal range.

With the emergence of global vegetation initiatives covering wide spatio-temporal frames, such as sPlot, Latin America has more than ever been put on the map for its floristic richness and uniqueness, and VegAndes is a pioneer dataset that not only fills data gaps but also offers an array of research and collaboration opportunities for regional vegetation scientists. At this early stage, the data is fully available to consortium partners and can be extended to external collaborators for agreed projects. On the short- to mid-term, the consortium is expected to grow, and with it, data contributions from less-sampled areas across a larger timespan. A specific VegAndes webpage based on the University of the Andes server is planned to allow easy data queries and downloads, making the database open-access and connected to other key botanical repositories such as sPlot, VegPáramo and sudamerica. The next project employing VegAndes will build on existing syntaxonomic approximations in the region and classify the complete extent of Latin American highlands into phytoregions, promising results that will complement the current syntaxon list and, more importantly, contribute to the understanding of the fascinating Latin American highland biogeography. Because the VegAndes data can support multi-scale studies in botany, ecology and biogeography, the database makes an essential contribution to biodiversity research and management perspectives in Latin America.

Subject of Research: VegAndes, a standardized vegetation plot database for vascular plants in the Latin American highlands

Article Title: VegAndes: the vegetation database for the Latin American highlands

Article References: Peyre, G., Montesinos, D., Giraldo, D., Galán de Mera, A., Ruthsatz, B., Luebert, F., Ontivero, M., García, N., Álvarez, M., Meneses, R. I., Lozano, P., León, D., Weigend, M., Anthelme, F., Palma, M., & Rodriguez, C. (2022). VegAndes: the vegetation database for the Latin American highlands. Vegetation Classification and Survey, 3, 287-296. https://doi.org/10.3897/vcs.95750

Image Credits: AI Generated

DOI: 10.3897/VCS.95750

Keywords: VegAndes, Latin America, Andes, vegetation plots, biodiversity database, highland flora, biogeography, Páramo, Puna, plant ecology, vegetation, database