Floristic diversity, composition and dominance across Amazonian forest types respond differently to latitude

The latitudinal biodiversity gradient is considered a first-order biogeographical pattern for most taxonomic groups. Latitudinal variation in plant diversity not always consistent, and this could be related to the particular characteristics of different forest types. In study, we compare changes floristic (alpha diversity), composition (beta diversity) dominance across tropical types: floodplain, terra firme submontane forests. Western Amazonia (Ecuador, Peru Bolivia). Woody plants. We inventoried 1978 species 31,203 individuals vascular plants with diameter at breast height ≥ 2.5 cm 118 0.1-ha plots over an 1800 km three relationships between alpha diversity, latitude type were analysed using generalised linear mixed models. Semi-parametric permutational multivariate analysis variance was used investigate effects on beta diversity. Dominant abundances correlated non-metric multidimensional scaling ordination axes reflect their contributions shaping Alpha increased towards equatorial latitudes forests but remained relatively constant floodplains. Beta all types changed latitude, although less clearly Also, floodplain forests, there fewer dominant contributing more homogeneous along gradient. patterns are manifested since summarizes climatic edaphic changes. However, found responses each type. inundation regime stronger predictor than limiting composition. Changes abundance gradients explained composition, harboured well drained It key study environmental trends habitat understand patterns. El gradiente representa un patrón biogeográfico de gran importancia para muchos grupos taxonómicos. No obstante, las variaciones latitudinales en la diversidad vegetal no siempre coinciden, debido características físicas y biológicas los diferentes tipos bosque. En este estudio, comparamos cambios florística (diversidad alfa), composición beta) dominancia bosque (inundables, tierra submontanos) lo largo del latitudinal. occidental Perú, Plantas leñosas. Inventariamos especies individuos plantas vasculares leñosas con diámetro altura pecho ≥2.5 parcelas 0.1 ha, tres La relación entre alfa, latitud tipo fue analizada modelos mixtos lineales generalizados (GLMMs). Se utilizó el análisis multivariante varianza semi-paramétrico permutaciones investigar efectos sobre diversidad. abundancia dominantes se correlacionó ejes no-métrico (NMDS) explicar contribución dichas patrones observados. alfa bosques submontanos incrementó hacia ecuador, mientras que inundables mantuvo más constante todos varió latitud, aunque menos pronunciadamente inundables. Además, encontraron contribuyendo distribuidas homogéneamente gradiente. Los manifiestan beta, ya resume climáticos edáficos. Sin embargo, encontramos respuestas cada inundables, régimen inundación es fuerte limitando riqueza florística. gradientes explicaron especies, pero albergaron homogéneas su bien drenados. Es fundamental comprender tendencias ambientales hábitat entender divesidad dominancia. Understanding causes spatial variations central issue biogeography. Diversity distribution mechanisms have been studied by biogeographers ecologists several decades, thereby yielding many insights into measurement quantification components, including (Chave, 2008). local/within-habitat (Melo et al., 2009), whereas originally defined as degree community differentiation (Whittaker, 1960). has recently wider context describe compositional any scale, communities, habitats or (Jost 2011), regardless involved (McKnight 2007). Variations influence large scales (Condit 2002) both crucial understanding it good example how influences (Rull, 2020). itself mechanistic factor (ter Steege, 2010), generally summarises conditions, providing unimodal (Stropp 2009; ter Steege & Hammond, 2001). Globally, exhibits increasing trend Equator 2008; Willig Presley, 2013), peak ~2.5°S 2010). This increase richness determined combined geographical (Baselga, 2008) temperature, reduction precipitation seasonality (Brown, 2014) soil complexity (Higgins resulting higher productivity around belt. particular, stable wet warm climate western characterised lower extinction immigration rates (Vasconcelos 2022), contributed diversification, peaks areas seasonal (Hoorn 2010) high topographical relief (Antonelli 2018). parallel, some studies also pointed out globally (Nishizawa southern (Koleff 2003; Toledo 2011) northern hemispheres (Dyer 2007; Qian Ricklefs, relationship inconsistent, driven climate, properties, ecological tolerance (Legendre, 2014), constraints affecting dispersal limitation fact, that factors other might shape such elevation flooding regimes 2007), extent (Soininen (Emilio differential organisms 2009). Many Amazonian ruled ‘few common, rare’ law, where small number account (Macía Svenning, 2005; Pitman 2001; 2013). known rare represented only few 1965) contribute common species. contradictory results obtained regarding importance determining Some indicate poor proxy entire due broad niche breadths (Arellano wide (Pitman addition, heterogeneity possible when these shared represent (Arellano, Jørgensen, 2016). By contrast, authors suggested turnover representative (Draper 2019), thus can which still poorly understood. (floodplain, forests) Amazonia. region mosaic nutrient-poor soils from Miocene sedimentary basins recent nutrient-rich formed sediments Andean uplift 2011; Hoorn 2010; Quesada Tuomisto 2019; Val 2022). productivity, heterogeneity, fast dynamics turnover, leading speciation evolutionary time noted its heterogeneous types, harbour amounts Each own processes (Guayasamin 2022; physiognomy, flooding, geomorphological complexity, seed limitations (Dambros These attributes responsible previous differentiated analyses Emilio Oliveira-Filho 2021; Stropp others (Assis 2017; Bredin 2020; Myster, 2017), (Macía, Macía 2005). showed despite explaining usually had effect result interactions conditions large-scale trends, history local physiognomy (Bicudo Brown, 2014; Vasconcelos Since present similar latitudes, considering add further information drivers easily recognised To best our knowledge, first focus relative Our specific objectives (1) hypothesised would types; (2) analyse differences whether vary similarly (3) roles will community. type, lead turnover. expected improve rainforests, analysis, expect discern unique characterising driving selected 13 regions ranging Ecuador Bolivia (Figure 1b). encompassed one two comprising floodplains, classification based combination elevation, shown diverse flooded world (Myster, 2017). Floodplain very (ca. 5000 years) fertile basin. They fed white-water rivers runoff water material Andes slopes (Quesada 2011). typically elevations below 500 m (Bayley 1992; Macía, marked flat slopes, stream river systems (Wittmann Flooding leads oxygen deficiency low conductance richness, growth rates, wood density, phenological strategies fruit/seed morphology (Hawes Peres, Terra slope inclinations equivalent those floodplains; nevertheless, they never 2002). variable soils, (more 300 10 (DBH) per hectare (Gentry, 1988)), presence Submontane located 1100 m, (Berni, montane lowland coexist (La Torre-Cuadros irregular topography geologically younger rich (Toledo greater same limited set 2005), weaker oligarchies acts Regions throughout region, six 14 established 1997 2019 according standard sampling protocol Cala, Plots (50 × 20 m) least apart, avoiding affected human activities natural disturbances create gaps canopy. Overall, sampled (n = 27), 51) 40) (Table S1). plot, measured individual woody (trees, palms, tree ferns lianas) DBH rooted within plot limits. total, 90% identified level, 10% morphospecies, excluded analyses. Taxonomic standardisation names Plant List R package ‘Taxonstand’ version 2.4 (Cayuela 2012). Voucher specimens classified stored herbaria (QCA, QCNE), (LPB) (USM, MOL). variables affect (Bañares-de-Dios retrieved regional values BIOCLIM V1.1 layers global ‘CHELSA’ model 1 km2 resolution (Karger data (pH, organic carbon fine earth g/kg, clay, sand, silt, volume fraction coarse fragments (>2 mm)) ‘SoilGrids’ database 15 depth (Hengl Pearson's correlation coefficients (r > ± 0.7); therefore, imply variables. deeply covary performed principal component (PCA) mentioned PCA dimension 51% total variance. Pearson (rL) axis (rP): temperature (°C) (rL −0.93, rP −0.68) (mm) −0.90, −0.63) variables, pH −0.87, −0.53) content (g/kg) 0.71, 0.70) correlations found, run explanatory variable, facilitate interpretation. For calculated seven indices rarefied Fisher's alpha, Shannon, Simpson, inverse Simpson evenness. All highly 0.8) provided information, so simplest measure models (GLMMs) negative binomial error log-link function fixed richness. Region 13) random potential autocorrelation among (Zuur fitted eight effects. complex included interaction predictors, expressed quadratic terms non-linear relationships, predictor. compared Akaike's criterion corrected sample sizes (AICc). Models difference AICc >2 indicated worst virtually support omitted. If model, 2 units lowest AICc. Model residuals explored simulation-based approach readily interpretable scaled GLMMs (Hartig, components R2 GLMMs: marginal (R2m) variability conditional (R2c) accounted supported (Nakagawa Schielzeth, glmer.nb ‘lme4’ 1.1–27-1 (Bates 2015). definition partitioning independence thoroughly debated (see Baselga, 2010, 2013; Jost, Legendre, Tuomisto, Here consider similarity species' identities conducted semi-parametric (PERMANOVA) (Anderson, 2001) 999 permutations Bray–Curtis dissimilarity distances (Bacaro visualise compositions gradient, metaMDS ‘vegan’ 2.5–7 (Oksanen 2020) (k 2, stress 0.19, tries 45). Data square-root transformed subjected Wisconsin double standardisation. better NMDS (floodplain: k 0.14, 20; firme: 0.17, submontane: 0.11, 20). coordination (PCoA) dissimilarity. Then, determine (PCoA 1). factor. four combinations (i.e. type). included. sensu al. (2013) 50% trees Following definition, 129 S2). inhibit interpretation Thus, abundant obtain clearer distributions visualisations easier interpret. selection comprised 25 species, 22.2% studied. overall change pattern, log-transformed envfit important structure metric (Dornelas because communities differ explain stream-graph visualisation ‘ggstream’ 0.1.0 (Sjoberg, 2021), round edges baseline categories (dominant species) continuous (latitude). Missing interpolated cubic splines smooth. spline ‘stats’ 4.1.1. v4.0.1 (R Development Core Team, contained cm, belonged 557 genera 112 families. Species index 1a,c) 2a). Comparisons alternative yielded best-fit Both consistent differed predicted 2b; Figure S3). Results PERMANOVA 2). full 21% unable separate predictors. intermingled clear clustering detected 3a). PERMANOVA, effects, S3; S4). 2) striking general, overlap occurred latitude. Three groups them, indicating 3b). divided partitions group clusters particularly Among corresponded 24% 14.3% 20% positions 3b), corners significant axes, subset plots, centre evenly distributed Table 3). Their five (Coussarea brevicaulis K. Krause, Iriartea deltoidea Ruiz Pav., Otoba parvifolia A.H. Gentry, Pentagonia spathicalyx Schum., Phytelephas tenuicaulis A.J. Hend.) peaked mostly medium Five axes: Euterpe precatoria Mart., Guarea macrophylla Vahl, pterorhachis Harms, Rinorea viridifolia Rusby Socratea exorrhiza (Mart.) H. Wendl. significantly except Siparuna decipiens (Tul.) A.DC. (p 0.013) pubiflora (Benth.) Sprague Sandwith 0.132) Leonia glycycarpa Pav. 0.13) Similar (sensu 2013) instead 58.9% 68% 75% S2; tended decrease 3c). whole decreased Equator, decreases noticeable. pronounced, (Wettinia augusta Poepp. Endl. salazarii H.E. Moore) Equator. confirmed well-known finding predictions (e.g., 2006; 2019). implies positively ecosystem (Rohde, 1992). A