Tree families and physical structureacross an elevational gradient in a Southern Andean Cloud forest in Ecuador
Introduction Research into how gradients structure ecosystemshas a long history in ecology (1,2,3,4). For example,studies have shownthat species are distributedrandomly across gradients (1,2,5), gradients have various effects across latitudes, longitudes, and elevations (4), and gradients due to climate changehave increasing important impacts on ecosystems(6,7). Among gradients,spatial gradients across elevations have been especially good places to investigate this relationshipthey include abiotic drivers – for example,precipitation, humidity, and temperature – that have been shown to have fundamental effects on mountain ecosystems (Chapter8). Within the Neotropics, a large elevational gradient is created by the Andean mountainsalong the western edge of South America. These Andes consist of “cordilleras”(9) where at their eastern flank they drop down to Western Amazonia (10,11,12) and at their western flank to the Pacific ocean (13).Occurring onall the cordillerasbetween 1000 m and 3000 m at sea level (i.e., a.s.l) are Cloud forests(8,14). These Cloud forests both add to the biodiversityof the Neotropics (e.g., the Andes themselves have 16.4% of all the plant species in the world:15) and significantly contribute toNeotropical hydrological and other biogeochemical cycles (16). Therefore because of the importance of Andean cloud forests to theneotropics,and because of the importance of theelevational gradient in defining those Cloud forests, I take advantage of a well-defined Cloud forest and elevational gradient in the Andes of Ecuador to sample tree(including tree ferns) species, treefamilies and treephysical structure, and then use that data to test these six hypotheses dealing with plant taxa and ecosystem structure: (1) Treespecies and treefamilies are distributed along the Andean elevational spatialgradient individualistically (2,17,18) without any pattern of clumping, (2) Some of these species and families have a mid-elevation peak along that gradientwhich may be due to the overlap of the distributions of the same tree species and treefamilies found at both higher and lower elevations (11), (3) Many of the species and family gradient patterns can be fitted significantly tomathematical models(19,20) with a skewed unimodal patterna plateau pattern most common (21,22,23), (4) Treephysical structure is distributed along the Andean elevational spatial gradient individualistically(2,17,18) without any pattern of clumping, (5) Many of these structural patterns can be fitted significantly to mathematical models (19,20) with a symmetric unimodal (Gaussian) patternmost common (21,22,23,24), and (6) Structural patterns distributed across the Andean elevational gradient are similar to tree structural patterns found in other Cloud forests in the Neotropics(25,26,27). Material and methods Study area The Reserva Biológica San Francisco (RBSF: 3o58’ 30” S, 79o4’ 25” W,16,28)in Southern Ecuador was where the study was conducted. RBSF is in the Andean Cloud forest but not all of it is primary, some of that Cloud forest is secondary due to the past impacts on indigenous peoples (29,30).Soils include Dystrudepts, Haplosaprists, Petraquepts, and Epiaquents(31). Temperaturesspan 9oto17o C and annual precipitation from 2200 to 5000 mm per year(31). Field sampling In January 2019 my field assistants and I randomly chose an elevational gradientwith a south-facing aspect in an Andean Cloud forest across from Rio San Francisco – RBSF – and set up one50m x 50m (¼ ha) plotat 1900m, 2000m, 2100m, 2200m, 2300m, 2400m, 2500m, 2600m, 2700m and 2800m. This plot size and shape have been used successfully to sample floristics and physical structure in this same Cloud forest at RBSF and at several of these same elevations (see Chapter16). In each plot we sampled all trees (including tree ferns) at least 10 cm in diameter at breast height (dbh), measuring at the lowest point where the stem was cylindrical but above the buttresses if the tree was buttressed. We also identified the trees to family, to genus and (if possible) tospecies,using (32) and (33) as taxonomic sources and also consulting the Missouri Botanical Garden website (www.mobot.org) where voucher samplesare kept. No research permits were necessary because the sampling was non-destructive and did not include any removal of biomass. Data analysis The number of stems for each treefamily and the number of stems for each common treespecies (defined as those species having at least 5% of thetotal stems of their family) was first compiled in each sampling plot on the elevational gradient. Next for each plot, these structural parameters were computed (1) treestem density, number of stems in each size class: 10 ≤ 19cm dbh, 20 ≤ 29 cm dbh, 30 ≤ 39cm dbh and ≥ 40 cm dbh, and the mean dbh for all stems combined, (2) family richness, genus richness and species richness, (3) Fishers’ alpha (α) diversity (34: http://groundvegetationdb-web.com/ground_veg/home/diversity_index), (4) total basal area(∑πr2; r = dbh of an individual stem / 2), (5) above-ground biomass (ABG: [35]), and (6) canopy closure (sum of all tree crown areas in a plot divided by the area of that plot, where crown areas are estimated from regressions on dbh [36]). The number of stems at each elevation for each common tree species, for each treefamily, and for each structural parameter was then subjected to a curve-fitting analysis (19, 20, 37) using (1) a symmetric unimodal model (i.e., a standard normal Gaussian distribution: 38),(2) a skewed unimodal model,(3) a linear modeland (4) a plateau model (21,22,23). Each model employed least-squares regression analysis after the appropriate transformation (38,39,40,41) where the stem data did not have an upper bound (42). The independent variable was elevation and the dependent variables were number of stems in a species or a family, or a structural parameter(software @ www.MyCurveFit.com was used). Significant regressionsare expressed in the results as (1) the Y-intercept of the best-fit regression line, (2) the slope of that line, (3) the amount of variation explained by that line (R2), and (4) the p-value of the best-fit line. Results All 29 treefamilies had individualistic distributions, peaking at different elevations (e.g., the families Aquifoliaceae [at 2400 m] and Euphorbiaceae [at 2300 m] peaked at mid-elevation: Table 1). Melastomataceae (with 461 stems), Lauraceae (with 430), Clusiaceae (with 182), and Rubiaceae (with 180) were most common. Solanaceae (with 1 stem), Malvaceae (with 3), Annonaceae (with 4), Cyatheaceae (with 4), and Hypericaceae (with 4) were least common. Clusiaceae, Lauraceae, and Melastomataceae had stems in every … Read more