Functional biogeography of herbaceous assemblages along edaphic and climatic gradients in Amazonian forests

Biogeografia funcional de comunidades de plantas herbáceas ao longo de gradientes edáficos e climáticos em florestas amazônicas

ABSTRACT

Soil, topography, and climate have been identified as strong drivers of functional composition turnover of plant communities, setting the dominant (“optimal”) community weighted mean (CWM) trait values. These same environmental factors may also affect disturbance regimes (e.g., the effect of soil fertility increasing tree turnover rates) as they create high-luminosity patches within dense canopy forest, promoting trait diversity, i.e. variance of the mean community trait values within and among communities. We analysed the drivers of functional trait turnover within and among 451 Zingiberales’ assemblages across Brazilian Amazonia, based on leaf mass per area (LMA), plant maximum height and seed size of 192 species. We detected functional turnover of the dominant assemblage trait values (given by the CWM) towards lower assemblage mean values of LMA, height and seed size, as soil fertility increased (higher sum of exchangeable bases) and local environments became wetter (bottomlands), in accordance with the expectations of the fast-slow growth trade-off along the whole-plant economic spectrum. Variability in seed size and plant height within assemblages increased towards lower topographic positions and low-nutrient soils, and the variance around dominant height and seed size values among assemblages increased towards bottomlands and regions with more seasonal climate. We conclude that, while soil and topography promote the filtering of a predictable functional composition, climate mostly promoted within and among-assemblage functional variance. Thus, it may be difficult to predict the effects of climate change based solely on the mean functional structure of Amazonian understory communities.

KEYWORDS:
ecological gradients; soil fertility; topography; tropical forest; Zingiberales

RESUMO

Solo, topografia e clima tem sido identificados como fortes impulsionadores do turnover da composição funcional de comunidades vegetais, definindo os valores dominantes (média da comunidade, CWM). Esses fatores ambientais também podem afetar os regimes de perturbação (p.ex., a fertilidade do solo aumentando o turnover de árvores), o que cria clareiras com alta luminosidade em florestas com dossel fechado, promovendo a variância dos valores médios de características funcionais dentro e entre comunidades. Neste estudo, avaliamos os determinantes do turnover das características funcionais médias e sua variabilidade dentro e entre 451 assembléias de Zingiberales na Amazônia brasileira, com base na massa foliar por área (LMA), altura máxima da planta e tamanho da semente de 192 espécies. Detectamos um forte turnover dos CWM em direção a valores mais baixos de LMA, altura e tamanho de semente à medida que a fertilidade do solo aumentou e nos baixios, de acordo com as expectativas do trade-off de crescimento rápido vs. lento ao longo do espectro econômico da planta. A variabilidade no tamanho da semente e na altura da planta dentro das assembleias aumentou nos baixios e em solos com poucos nutrientes, e a variância em torno dos valores dominantes de altura e tamanho de semente entre as assembleias aumentou em regiões de clima mais sazonal. Concluímos que, enquanto o solo e a topografia promovem a filtragem da composição funcional, o clima promoveu principalmente a variância funcional. Portanto, pode ser dificil prever os efeitos das mudanças climáticas sobre a estrutura funcional média das comunidades do sub-bosque amazônico.

PALAVRAS-CHAVE:
gradientes ecológicos; fertilidade do solo; florestas tropicais; topografia; Zingiberales

INTRODUCTION

Functional traits represent the phenotypic expression of individuals and relate to population vital rates (growth, recruitment, mortality) and plant life strategies. Direct measurements of life cycle and performance of organisms are not straightforward, especially in highly diverse ecosystems, so ecologists have been using plant functional traits as proxies of plant strategy (Westoby 1998; Reich 2014; Díaz et al. 2016). This field has advanced strongly for tropical palms, lianas and especially trees, but less so for tropical herbs, despite their relevant contributions to forest diversity and food-webs (Perea et al. 2022).

Plant height, leaf mass per area (LMA) and seed size have been suggested as key indicators of plant strategies (Westoby 1998). In closed-canopy forests, plant height is an indicator of light-related niche position since taller species have more access to light. Seed size indicates reproduction and establishment strategies: species may produce fewer but larger seeds with high supply of reserves, enabling establishment under resources scarcity, or a larger quantity of smaller seeds to enable colonising of micro-sites that are rare in space and time (Dalling et al. 1998; Kitajima 2002). LMA summarises the global spectrum of leaf economics, running from fast to slow return of investment in leaf tissues (Wright et al. 2004). Theory predicts that when resources are scarce, species grow slowly and invest in highly durable tissues in leaves, stems and/or roots, and the opposite strategy is expected under high resource availability (Reich 2014), i.e., an integrated ‘fast-slow’ plant economics spectrum aligned with resource availability.

Studies of tree functional turnover patterns in tropical forests have revealed that trait-turnover patterns are governed by climate and edaphic conditions (ter Steege et al. 2025; Fyllas et al. 2009; Fortunel et al. 2014; Muscarella and Uriarte 2016; Joswig et al. 2022), and a recent review focused on the Neotropics concluded that resource-limited environments tended towards conservative traits and strategies (Freitas et al. 2023). Tree species with slow growth strategies (conservative strategy) tend to successfully establish and live in nutrient-poor soils, whereas fast growth species (acquisitive strategy) are dominant in nutrient-rich soils (ter Steege et al. 2025; Fyllas et al. 2009; Joswig et al. 2022). Climate is expected to modulate leaf traits and plant size at the global scale, but a weak relationship has been often documented for leaf traits (Wright et al. 2004, 2017; Maire et al. 2015). Indeed, Fyllas et al. (2009) found a positive relationship between LMA and annual precipitation in Amazonia, whereas Muscarella and Uriarte (2016) found a negative relationship in Central America.

Because organisms do not access the total amount of water provided by rainfall, water availability depends also on the soil capacity to retain water and the proximity to the water table (Costa et al. 2023). Topography influences species variability in hydraulic traits and may be a stronger filter of functional strategies than climate at local (Cosme et al. 2017; Oliveira et al. 2018) or even regional scales (Garcia et al. 2023). Following the theory of an integrated ‘fast-slow’ plant economics spectrum aligned with resource availability (Reich 2014), we could expect the community means of traits to linearly change along the gradients of soil fertility, climatic and topography. The effects of natural disturbances (e.g., tree falls) are expected to interact with the effects of soil, climate and topography, promoting trait diversity as they create high light patches among a matrix of closed forest, each selecting different plant traits (Sterck et al 2011). At the broad regional scale, the expected consequence is the generation of noise in the turnover of the mean plant trait values in communities along those environmental gradients.

Herbaceous plants - and therefore their assemblages - can be expected to follow the same functional trends along environmental gradients as described for other life-forms, even if the absolute trait values are different (Matsuo et al. 2023). However, Swenson (2009) described a “reversal of the functional dicot strategy for monocots” - i.e., that species associated to high-luminosity environments had higher LMA, lamina size and culm density (conservative traits) when compared to their congeneric pairs of shaded environments - the opposite of what is found for dicot herbs and trees. If this pattern of response to light is extended to other resources, the turnover of herbs along soil fertility and topography gradients might not follow the patterns described for other life-forms.

Here we investigated the variation of functional traits of 451 herb-assemblages across the Amazon. We focused on assemblages of Zingiberales (hereafter gingers), a very diverse order of monocotyledonous species (Kress et al. 2001), accounting for more than 50% of the herb cover in many plots in Amazonia (Costa 2004, 2006). Gingers are rhizomatous herbs, varying in habit from small rosulate (10 cm height) to tall palm-like or scandent plants (5 m tall). Our objective was to determine how resource gradients affect the dominant assemblage functional strategy (given by the trait’s community weighted mean, CWM in the inventory plots), the across-assemblage variance of the dominant strategy (variance of CWM across plots), and the within-assemblage variation of traits among individuals (given by the community weighted deviation, CWD, within plots). The traits evaluated were LMA, plant height and seed size, which provide a general description of functional strategies (Westoby 1998). We hypothesised that the dominant functional strategy will change along the resource gradients, with acquisitive strategies (lower CWM of LMA, plant height and seed size) increasing towards high resource levels (fertile soils, wet climate and bottomlands). At the same time, we expected the variance of the dominant functional strategy among and within-assemblages to increase towards nutrient-rich soils, wet climate and bottomlands.

MATERIAL AND METHODS

Biotic and abiotic data were obtained in 451 plots of 500 m² distributed along non-inundated (terra-firme) lowland forests in Brazilian Amazonia (Figure 1). Data came from two plot designs: (1) 409 plots of 250 x 2 m following the RAPELD protocol (Costa and Magnusson 2010); and (2) 42 plots of 100 x 5 m sub-sampled from 500 x 5 m plots (UTU plots) (Tuomisto et al. 2016). RAPELD plots are placed at 1‐km intervals across grids or modules that are distributed across the Amazon lowland basin. RAPELD plots are installed following elevation contour lines to minimize within-plot environmental variation associated with topography (e.g., soils, vertical distance to the water table). UTU plots are not distributed in grids but placed in specifically designated sites using remote sensing imagery (Tuomisto et al. 2016). They follow a predetermined compass bearing, thus crossing the local topographical variation to maximize environmental variation within plots. A minimum distance of 1 km is kept between UTU plots. Both RAPELD and UTU plots were sampled in areas that vary from tall mature forest to white sand forests with a simpler canopy structure (regionally known in Brazil as campinaranas). We selected the continuous 5-m sub-units with the lowest internal topographical variation from the UTU plots to make the two sampling designs comparable, as previously done in other studies (Moulatlet et al. 2017; Tuomisto et al. 2019; Zuquim et al. 2019).

Figure 1
Location of the 451 plots where ginger community and environmental data were obtained in the Brazilian Amazon. The background map represents the dry season length. The legend denotes the maximum number of consecutive months with less than 100 mm of precipitation along the 1998 - 2004 period. Black triangles represent the clusters of RAPELD plots and blue triangles represent UTU plots located along the Juruá River. Clusters may have 5 to 57 plots spaced by at least 1 km.

We counted every ginger clump with a height > 5 cm rooted inside each plot. Clumps were defined as groups of stems or leaves arising from the soil less than 20 cm from each other or based on our field experience of each species. Representative samples were collected to be further identified. All specimens were identified by the first author (FOGF) based on specialised literature and the fertile material was sent to the specialists Helen Kennedy and Paul Maas for confirmation. Vouchers were deposited in herbaria in Brazil (INPA), Canada (UBC), and the Netherlands (U).

We measured or obtained from the literature three functional traits (LMA, plant height and seed size) for the species found in the plots (Maas 1972, 1977; Andersson 1977, 1985; Hagberg 1990; Horvitz and Schemske 1994) plus other scattered information listed in Costa et al. (2008). LMA is the ratio between the dry leaf mass and fresh (i.e., completely expanded) leaf area. We adapted the original method to estimate LMA (Pérez-Harguindeguy et al. 2013) using small pieces of the lamina of dried material from herbarium specimens. For each individual, a maximum of 4 pieces (on average 3 cm²) was cut from different parts of the lamina (at middle and close to the base and apex), excluding the main vein. We chose the locations carefully from mature leaves and leaf pieces where the lamina tissue had no or very little shrinkage. LMA was calculated as the average of values obtained from a maximum of six individuals per species or only from one individual, when the species was rare and collected only once. For the few direct field observations, height estimation was based on the 95% quantile. Seed size was obtained using the formula of an ellipse based on the two largest measures of seed dimension obtained from the literature. Out of the 192 species, measures of LMA, plant height and seed size were missing for 23, 23 and 109 species, respectively. In these cases, the trait value was assigned as the mean genus trait value considering only species found in the Amazon region (Borgy et al. 2017). Most of the species for which traits could not be measured were rare (< 3 individuals) in the total sampling.

Environmental data was measured in each plot. To represent soil fertility, topographic and climatic gradients we used the sum of exchangeable base cations (Ca⁺², Mg⁺² and K⁺ measured in cmol(+) kg^-1), the vertical distance from drainage (VDD, in metres) and the dry season length (DSL, in months), respectively. The sum of exchangeable bases is a widely used proxy of soil fertility in Amazonia (Quesada et al. 2011) and strongly related to floristic patterns across the basin (Figueiredo et al. 2014; Tuomisto et al. 2019). The protocols for soil sampling and laboratory analyses are described in previous studies (Figueiredo et al. 2014; Tuomisto et al. 2016).

The vertical distance from drainage (VDD) was extracted from a SRTM digital elevation model. VDD have been successfully used to explain floristic patterns in Amazonia (Moulatlet et al. 2014, 2022). Sites with low VDD are located near drainages or areas where the water table is closer to the soil surface. These areas are in bottomlands, where usually the water availability is higher in comparison with hilltops. VDD values were obtained from published articles. For 198 plots located in the interfluve Purus-Madeira, we obtained VDD values from Moulatlet et al. (2014). For 193 plots located north of the Amazon River in central Amazonia and in Pará state, VDD was obtained from Figueiredo et al. (2014). For the remaining plots, VDD was calculated using the Vertical Distance to Channel Network algorithm (Bock and Köthe 2008), implemented in the QGis software through the SAGA-toolbox (new data extracted for this study).

Dry season length (DSL) was estimated from data of the TRMM satellite (Kummerow et al. 1998), available at http://disc.sci.gsfc.nasa.gov. We converted monthly data from 1998 to 2004 of the TRMM product 3B43 V6 at a 0.25° resolution (about 28 km at the equator) into the dry season length variable, defined as the maximum number of consecutive months with less than 100 mm of precipitation along that period. For each plot, the DSL value was extracted after rescaling the data to a 0.05° (~ 5 km) spatial resolution. We used the raster package in R to manipulate and process the TRMM variable (R Core Team 2022, Hijmans 2017).

Data analysis

To evaluate the effect of environmental gradients on functional trait patterns within and among ginger assemblages, we first estimated the community weighted mean (CWM) [Equation 1] to characterise the dominant trait strategy of each assemblage (i.e., each RAPELD or UTU plot) and the community weighted deviance (CWD) [Equation 2] to characterise the within-assemblage trait variability, according to Díaz et al. (2007) and Violle et al. (2007).

$CWM={\sum }_{k=1}^{n_j}{w}_{k,j}\times T_{k,j}$ Equation 1

$CWD=\sqrt{\frac{{\sum }_{k=1}^{n_j}{w_{k,j}(T_{k,i}-CWM)}^2}{{\sum }_{k=1}^{n_j}{w}_{k,j}-1}}$ Equation 2

where w is the relative abundance of the species j in the assemblage (plot) k, and T is the trait value of the species j in the assemblage k.

The effects of environment on within-assemblage variability were evaluated with multiple GAMLSS (Generalized Additive Models for Location Scale and Shape) regressions (Stasinopoulos and Rigby 2008), which allow the joint estimation of environmental gradients effects on the mean (µ) and variance (σ) of assemblage-trait response. Since the CWD is a descriptor of variance, only its mean response to the environment was evaluated. The turnover of the dominant trait strategy of each assemblage along environmental gradients was evaluated with multiple GAMLSS regressions of each trait CWM against soil fertility, DSL and VDD. As in an ordinary least-squares regression, in GAMLSS the change of a response variable as a function of a predictor variable is evaluated by the mean response. The variance around the mean response of communities to each environmental variable indicates divergence among communities (i.e., variance can increase with the environmental predictor, indicating divergence towards more positive values of the predictor, or the opposite, or even no significant trend), and this was evaluated by the variance parameter of the GAMLSS.

We fitted different models assuming that our response variables (CWM or CWD of each trait) can be represented by general distribution families parameterized by the mean and variance (shape parameters, skewness and kurtosis). Our approach consisted in modelling all parameters as linear and/or non-linear responses to the predictor variables (soil, VDD and DSL) using specific link functions for each parameter. The soil fertility and VDD were logarithmically transformed before the analysis to fit the assumptions of normality required in linear models and because the relative change in these variables is more important than the absolute change (i.e., a unit change in base concentration is more strongly perceived by plants in poor than in rich soils).

Prior to the modelling we tested several distribution families to best fit the distribution of CWD and CWM. For height and seed size, we tested continuous distributions assuming CWM >0 (for either height or seed size), and for CWM-LMA, we also included the distribution families that do not have lower bounds (e.g., normal distribution). The selected family distributions were those with the lowest values of the generalized Akaike information criteria (GAIC) obtained by fitting the available distributions on observed values of CWD or CWM without environmental predictors. Then, we tested the effects of environmental gradients on the CWD or CWM of each trait, modelled with the selected family distribution. If models failed to find an optimal solution after iteractions, the second and third best ranked families were employed.

For plant height and seed size, CWM models were fit using a Box-Cox t with four parameters assuming a log link for mean, variance and kurtosis and identity link for the skewness parameter. For LMA, the CWM model was fit using a t family distribution with 3 parameters using identity link for mean and log link for variance and skewness parameters. The CWD models were fitted using a zero-adjusted Gamma distribution. When plots had only a single species, CWD was assumed to be zero.

After setting the distribution that best fitted our response variables, we performed a model selection procedure to quantify how well each independent variable explained the variation of each parameter of our response variables. For both CWM and CWD, the model selection procedure started with a forward approach (leave-one-out variable starting with the first one) applied first to the mean model and subsequently to the variance model. The procedure finished running the backward selection (leave-one-out variable starting with the last one) from variance to mean. Only linear relationships between the response and predictor variables were allowed. Since for CWD we only evaluated the mean response, the variance was estimated as the intercept of the variance parameter of a zero-adjusted gamma distribution.

Then, we fitted three types of models: one allowing the environmental variable to affect only the mean (environmental pressure towards a dominant value, i.e. functional turnover); the second allowing the environmental variable to affect variance only (environmental pressure towards among-community divergence); and a third model allowing the environmental variable to affect both mean and variance. The relative effect size of each variable was assessed by the beta coefficients (slopes). Model performance was evaluated using the ΔGAIC and generalised R² (Nagelkerke 1991) as implemented for GAMLSS models (Stasinopoulos and Rigby 2008).

GAMLSS were controlled for potential spatial autocorrelation due to plot aggregation within sites by setting site as the random term. Sites were defined as any group of plots within a radius of 10 km. To evaluate the effect of the spatial autocorrelation in distances above 10 km, we computed Moran’s I correlograms for the CWM as measured for each of the three functional traits (LMA, seed size and plant height) and for the environmental variables soil fertility, HAND and DSL (Figure S1). Significant autocorrelation in the same distance class for both dependent and independent variables would indicate the potential for spatial factors not related to the tested hypothesis to be affecting the models. The only distance class that showed consistent overlap between dependent and independent variables was the first one, as it is normally expected given the contagious plant dispersal process. This justifies the inclusion of a random factor designed to tackle the expected correlation of samples that are within a 10 km radius.

Finally, we evaluated if the individually modelled trait responses could be related to a general functional strategy. For that, we used Principal Component Analysis (PCA). The traits were scaled prior to the analyses using the function scale from the base R package. Then we performed linear multiple regressions with the PC1 as the response variable and the soil fertility, HAND and DSL as explanatory variables.

GAMLSS models were performed using the gamlss R-package (Stasinopoulos and Rigby 2008) and model selection was done using the stepGAICAll.A function from the same package. Moran’s I correlograms were performed using the function lets.correl of the R package letsR (Vilela and Villalobos 2015)

RESULTS

We recorded 47,129 ginger individuals from 192 species (Table S1). The minimum, median and maximum abundance per plot were 4, 66 and 897 individuals, respectively. Median species richness per plot was 8, varying from 1 to 23 species. Functional traits were often moderately or weakly intercorrelated at species and community level (Table S1). At the species level, taller species tended to have high LMA (LMA and plant height were positively correlated; Figure 2a). At the community level, the same communities tended to have high variability for both plant height and seed size (the corresponding CWD values were correlated; Figure 2f), and the same communities that had tall species also tended to have large seeds (the corresponding CWM values were correlated; Figure 2i).

Figure 2
Correlations (Pearson’s r) between functional traits of Zingiberales at the species (A - C) and community levels (D - I). Correlations between traits at species level were obtained applying a logarithmic transformation and vertical plot’s axis are presented using that scale.

We found a weak effect of environmental gradients on the local within-community trait variability, given by the CWD (Table 1). The variation in seed size within communities was slightly higher in communities on nutrient-poor soils (b = -0.04) and in bottomlands (b = -0.06). The variation in plant height also tended to increase towards nutrient-poor soils, but the best height-environment model had only weak support when compared with a null model (ΔGAIC < 2). Within-community variability in LMA was not significantly related to the environmental variables.

Table 1
Summary results of mixed-effect GAMLSS models for CWD and CWM of 451 herbaceous communities of Zingiberales along environmental gradients in the Amazon. The CWD models were fitted using the zero adjusted gamma distribution family and include only the relationship between the mean (μ) parameter and predictor variables. The CWM models were fit using the three parameters of the t family distribution for LMA, the four parameters box-cox t for plant height and seed size, and include the relationships between the mean (μ) and/or variance (σ) and the predictor variables. Best models appear in bold and models where predictor variables were not significant (P > 0.05) are not shown. Null models are the first entry for each response variable. Values in brackets denote the scaled β coefficients. SB = sum of exchangeable bases; VDD = vertical distance to the nearest drainage; DSL = dry season length.

Trait	Response variable	Model	ΔGAIC	R2
LMA	CWD	(μ) ~ 1|random(sites)	0	0.35
	CWM	(μ) ~ 1|random(sites)	63.28	0.38
		(μ) ~ (- 2.52) SB + (1.18) VDD + 1|random(sites)	0	0.469
		(σ) ~ 1
		(μ) ~ 1	58.59	0.392
		(σ) ~ (0.14) SB
Plant height	CWD	(μ) ~ 1|random(sites)	1.88	0.381
		(μ) ~ (-0.04) SB + (-0.06) VDD + 1|random(sites)	0	0.386
	CWM	(μ) ~ 1|random(sites)	45.83	0.398
		(μ) ~ (-0.04) SB + (-0.03) VDD + 1|random(sites)	41.73	0.407
		(μ) ~ 1|random(sites)	3.57	0.457
		(σ) ~ (-0.26) VDD + (0.23) DSL
		(μ) ~ (-0.05) SB + 1|random(sites)	0	0.463
		(σ) ~ (-0.25) VDD + (0.23) DSL
Seed size	CWD	(μ) ~ 1|random(sites)	11.73	0.469
		(μ) ~ (-0.04) SB + (-0.06) VDD + 1|random(sites)	0	0.487
	CWM	(μ) ~ 1|random(sites)	75.73	0.503
		(μ) ~ (-0.08) SB + (0.05) VDD + 1|random(sites)	37.54	0.548
		(μ) ~ 1|random(sites)	49.44	0.536
		(σ) ~ (-0.36) VDD + (0.23) DSL
		(μ) ~ (-0.07) SB + (0.08) VDD + 1|random(sites)	0	0.587
		(σ) ~ (-0.58) VDD + (0.29) DSL

We found significant effects of environmental gradients on the turnover of the mean and variance of the functional trait composition among communities. The best model (lowest ΔGAIC and highest R²) included significant effects of the environmental variables on both the CWM mean and variance response, except for LMA, for which only the mean response was associated with the environmental gradients (Table 1). Models that included only the mean response of CWM had better performance than models that included only the variance response, except for plant height (Table 1).

Mean and variance turnover of trait values among communities differed strongly between traits and environmental predictors. The mean community LMA was lower in nutrient rich soils and bottomlands, but there was no change in LMA along the dry season length gradient (Figure 3a-c). The standardised rate of change of mean LMA values along the soil exchangeable bases gradient (b = -2.52) was the highest comparing all models and was twice bigger than the rate of change along the VDD gradient (b = 1.18).

The variation of mean height and seed size along environmental gradients was similar (Figure 3d-i), probably due to the high correlation between these two traits at community level (Figure 3f,i). Height and seed size also decreased as soil fertility increased, but with a lower rate of change than for LMA (Table 1; Figure 3a,d,g). Mean seed size slightly decreased in bottomlands, but with a smaller rate of change than the rate at which the variance increased (Table 1; Figure 3h).

Figure 3
Changes in the community weighted mean (CWM) of 451 herbaceous communities of Zingiberales along environmental gradients in the Amazon. Continuous and dashed lines represent the mean (μ) ± variance (σ) of CWM, respectively. Predicted mean is the 0.5 quantile and predicted variance denotes the interval between the 0.25 and 0.75 quantiles of the t family distribution (A - C) and Box-Cox t distribution (D - L). For each variable the predicted mean and variance were obtained keeping all other variables constant at their mean values. The μ (mean) and/or σ (variance) symbols displayed in the upper-right corner of panels indicates which of these parameters had a significant association with the predictor variables and ‘ns’ denotes non-significant association. Vertical axes from E to L are in logarithmic scale. Dots denote the partial residual, and light grey crosses represent the original values.

We found a significant effect of VDD and DSL on the among-community variance of plant height and seed size, but not of LMA. As expected, the variance on the mean community values of plant height and seed size increased towards bottomlands (Figure 3e,h). We found a wide variation of CWM for plant height and seed size in bottomlands (but not for LMA), while on uplands there was a predominance of communities with intermediate mean stature and medium mean seed size. The variance of the CWM of plant height and seed size increased towards drier conditions, while in wet areas the dominant species were those with medium height and seed size (Figure 3f,i).

The principal component analysis captured 60% of the variation of the LMA, plant height and seed size CWM in the first component and 24.5% in the second component (Figure S2). PC1 was significantly related (adjusted R² = 0.16, p < 0.001) to the sum of bases (b = 0.5, p = 0.04), to the vertical distance from the drainage (b = 0.1, p < 0.001) and to the dry season length (b = 0.1, p = 0.05), with the same trends found for the individual trait CWM (Figure S3).

DISCUSSION

Our findings reveal that environmental gradients affected both within and among-assemblage functional patterns of Amazonian gingers. We detected functional turnover of the dominant assemblage trait values (given by the CWM) towards lower assemblage mean values of LMA, plant height and seed size, as soil fertility increased and local environments became wetter (bottomlands), in accordance with the expectations of the fast-slow growth trade-off along the whole-plant economic spectrum (Reich 2014). Functional divergence was detected both within and among assemblages: within-community variability (given by the CWD) in seed size and plant height tended to increase towards bottomlands and nutrient-poor soils, and the variance around dominant plant height and seed size values among communities increased towards bottomlands and more seasonal regions. As the combination of these later environmental conditions lead to more dynamic forests (Quesada et al. 2012), our results suggest that higher disturbance levels might drive an increase in the functional divergence among assemblages at regional scales.

We found stronger trends of functional turnover for the assemblage mean trait values (given by the CWM) than for the variance of these values (CWM variance among assemblages), with soil fertility gradients having the strongest filtering effect. Trait values observed along the soil gradient were in strong accordance with the fast-slow growth trade-off, which predicts that acquisitive traits (low LMA, low plant height and small seed size) dominate in highly productive habitats, while conservative traits predominate in low resource conditions (Reich 2014). Indeed, the turnover of assemblage weighted trait means was higher along the soil than along topographic or climatic gradients, reinforcing previous observations that soil fertility is the key driver of regional functional composition across lowland Amazonia (Fyllas et al. 2009).

The mean functional trait values also turned over along topographical gradients following the expectation of the acquisitive-conservative strategy trade-off along water-resource gradients (Hoffmann et al. 2005; Reich 2014). This relationship depends not only on climate but also on its interaction with topography, as topography is a surrogate for soil moisture (Fan et al. 2013), a condition that directly affects plant performance. Hills tend to suffer more from water deficit than bottomlands, favouring the predominance of conservative strategies (high LMA and larger seed size) in the former. A similar trend was observed at a local scale for trees, with species associated with high topographic positions having higher LMA than species associated with bottomlands (Cosme et al. 2017; Schmitt et al. 2020). Topographic variation has been disregarded in regional scale studies of community functional patterns, which may explain why tight links between changes in functional composition and water deficit have not been found.

We did not find turnover in the mean assemblage trait values along the climate gradient. Previous results on functional changes among tree communities along a rainfall gradient indicate that LMA is higher in wetter regions (Fyllas et al. 2009). The lack of a climate effect may be a consequence of at least two factors. First, most trait variation was explained by soil and the topographic gradient, and the local water availability might be more relevant in filtering the traits related to the water-strategy. The absence of a climate effect on LMA could also arise by the mixture of deciduous and non-deciduous species in the seasonally dry regions. Deciduous species are often associated with dry regions and tend to have lower LMA than evergreens (Poorter et al. 2009), which could counteract the trend of increasing LMA (a conservative strategy) towards dry regions. Although we have not directly evaluated the phenological habits of species in our dataset, our field observations suggest that some short-statured species of Goeppertia and Chamaecostus generally associated to drier regions, can exhibit deciduous behaviour. However, as LMA decreased with increasing soil nutrients, we hypothesise that deciduousness should be modulated jointly by drier conditions and high concentrations of nutrients in soils.

The variance of functional composition among assemblages changed along topographic and climatic gradients, but not along the soil nutrient gradient. We found that bottomlands and seasonally dry regions contained assemblages that varied more among each other in plant height and seed size, which are aligned with the expectation that higher disturbance rates should lead to higher functional differences among communities under the same soil and climate conditions. Those differences should arise mostly from the colonization of disturbed patches by the local fast-growth acquisitive functional types.

Higher frequency of disturbances also led to higher within-assemblage trait variance (given by the CWD) of seed size and plant height in bottomlands. Bottomlands tend to be more susceptible to natural disturbance than higher ground, given terrain instability and soil physical constraints, leading to higher rates of tree mortality (de Toledo et al. 2011). Higher tree mortality rates lead to higher spatial and temporal light heterogeneity in the understory, which would allow the occupation of these sites by species with more variable seed size and plant height, and their co-occurrence in local communities within bottomlands.

Nutrient-rich soils are associated with higher tree turnover (Quesada et al. 2012), which is expected to increase light heterogeneity and lead to higher within-community trait variance. Contrary to this expectation, within-community variance of plant height and seed size slightly increased as soil fertility decreased. A potential explanation is that light may not be a limiting factor in nutrient-poor soils. Although forests with high fertility have potentially more gaps, maybe the gaps close too fast to allow the persistence of a large variety of plant-height/seed strategies. Thus, nutrient-poor soils were found to promote more local variation in plant height through the higher contrast between gaps and closed canopies at the local scale. Alternatively, the strong convergent filtering of species LMA by soil fertility may carry together the filtering of plant height, since these traits are correlated at the species level.

CONCLUSIONS

Our findings revealed that, while soil and topography promote the filtering of a predictable functional composition, climate mostly promoted within and among-assemblage functional variance across Amazon forests. An obvious extension of this study would be the evaluation of the combined effects of these predictors on multiple plant taxa, which would help understanding how environmental gradients affect functional strategies of and how these strategies will influence on the flora survivorship in climate change scenarios.

ACKNOWLEDGMENTS

We acknowledge the administrative and logistical support of the Instituto Nacional de Pesquisas da Amazonia (INPA), Programa de Pós-Graduação em Ecologia do INPA, Programa de Pesquisas em Biodiversidade (PPBio), the Long-term Ecological Research Program (PELD), the Biological Dynamics of Forest Fragments (BDFFP), Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio), and Instituto Brasileiro Meio Ambiente e Recursos Renováveis (IBAMA). We acknowledge the following funding sources: Brazilian Program for Biodiversity Research - (PPBio), Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) grant #47890820/2-1 (led by FC); CNPq (fellowship to F.F); CNPq-PPG7 grant (#556988/2005-1), FINEP/Projeto Integrado MCT-EMBRAPA (led by Ana L. K. M. Albernaz); BDFFP (grant to FF); Hidroveg Project - FAPEAM/FAPESP (#1428/2010 led by FC and Javier Tomasella). Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Finance Code 001; Academy of Finland grants #139959 and 273737 to HT; CLAMBIO consortium funded through the BiodivERsA 2019-2020 Joint COFUND Call on “Biodiversity and Climate Change” (Academy of Finland grant #344733 to HT); Independent Research Fund Denmark grant #9040-00136B to Henrik Balslev. This is publication number 883 of the BDFFP technical series.

REFERENCES

• Andersson, l. 1977. The genus Ischnosiphon (Marantaceae). Opera Botanica 43: 1-114.
• Andersson, l. 1985. Revision of Heliconia subgen. Stenochlamys (Musaceae-Heliconioideae). Opera Botanica 82: 5-123.
• Bock, M.; Köthe, R. 2008. Predicting the depth of hydromorphic soil characteristics influenced by ground water. Hamburger Beiträge zur Physischen Geographie und Landschaftsökologie 19: 13-22.
• Borgy, B.; Violle, C.; Choler, P.; Garnier, E.; Kattge, J.; Loranger, J.; et al 2017. Sensitivity of community-level trait-environment relationships to data representativeness: A test for functional biogeography. Global Ecology and Biogeography 26: 729-739.
• Cornwell, W.K.; Ackerly, D.D. 2009. Community assembly and shifts in plant trait distributions across an environmental gradient in coastal California. Ecological Monographs 79: 109-126.
• Cosme, L.H.M.; Schietti, J.; Costa, F.R.C.; Oliveira, R.S. 2017. The importance of hydraulic architecture to the distribution patterns of trees in a central Amazonian forest. New Phytologist 215: 113-125.
• Costa, F.R.C. 2004. Structure and composition of the ground-herb community in a terra-firme Central Amazonian forest. Acta Amazonica 34: 53-59.
• Costa, F.R.C.; Espinelli, F.P.; Figueiredo, F.O.G. 2011. Guide to the Zingiberales of PPBio Sites in the Brazilian Western Amazonia Átemma Editora, Manaus, 287p.
• Costa, F.R.C.; Magnusson, W.E. 2010. The Need for Large-Scale, Integrated Studies of Biodiversity - the Experience of the Program for Biodiversity Research in Brazilian Amazonia. Natureza & Conservação 08: 3-12.
• Costa, F.R.C.; Magnusson, W.E.; Luizão, R.C. 2005. Mesoscale distribution patterns of Amazonian understorey herbs in relation to topography, soil and watersheds. Journal of Ecology 93: 863-878.
• Costa, F.R.C.; Schietti, J.; Stark, S.C.; Smith, M.N. 2023. The other side of tropical forest drought: do shallow water table regions of Amazonia act as large-scale hydrological refugia from drought? New Phytologist 247: 714-733.
• Dalling, J.W.; Hubbell, S.P.; Silvera, K. 1998. Seed dispersal, seedling establishment and gap partitioning among tropical pioneer trees. Journal of Ecology 86: 674-689.
• de Freitas, G.V.; Da Cunha, M.; Vitória, A.P. 2024. A systematic review of leaf and wood traits in the Neotropics: environmental gradients and functionality. Trees 38: 551-572.
• de Toledo, J.J.; Magnusson, W.E.; Castilho, C.V.; Nascimento, H.E.M. 2011. How much variation in tree mortality is predicted by soil and topography in Central Amazonia? Forest Ecology and Management 262: 331-338.
• Díaz, S.; Lavorel, S.; de Bello, F.; Quétier, F.; Grigulis, K.; Robson, T.M. 2007. Incorporating plant functional diversity effects in ecosystem service assessments. Proceedings of the National Academy of Sciences 104: 20684-20689.
• Díaz, S.; Kattge, J.; Cornelissen, J.H.C.; Wright, I.J.; Lavorel, S.; Dray, S.; et al 2016. The global spectrum of plant form and function. Nature 529: 167-171.
• Fan, Y.; Li, H.; Miguez-Macho, G. 2013. Global patterns of groundwater table depth. Science 339: 940-943.
• Ferry, B.; Morneau, F.; Bontemps, J.-D.; Blanc, L.; Freycon, V. 2010. Higher treefall rates on slopes and waterlogged soils result in lower stand biomass and productivity in a tropical rain forest. Journal of Ecology 98: 106-116.
• Figueiredo, F.O.G.; Costa, F.R.C.; Nelson, B.W.; Pimentel, T.P. 2014. Validating forest types based on geological and land-form features in central Amazonia. Journal of Vegetation Science 25: 198-212.
• Fortunel, C.; Paine, C.E.T.; Fine, P.V.A.; Kraft, N.J.B.; Baraloto, C. 2014. Environmental factors predict community functional composition in Amazonian forests. Journal of Ecology 102: 145-155.
• Fyllas, N.M.; Patiño, S.; Baker, T.R.; Bielefeld Nardoto, G.; Martinelli, L.A.; Quesada, C.A.; et al 2009. Basin-wide variations in foliar properties of Amazonian forest: phylogeny, soils and climate. Biogeosciences 6: 2677-2708.
• Garcia, M.N.; Domingues, T.F.; Oliveira, R.S.; Costa, F.R.C. 2023. The biogeography of embolism resistance across resource gradients in the Amazon. Global Ecology and Biogeography 32: 2199-2211.
• Goulamoussène, Y.; Bedeau, C.; Descroix, L.; Linguet, L.; Hérault, B. 2017. Environmental control of natural gap size distribution in tropical forests. Biogeosciences 14: 353-364.
• Grime, J.P. 2006. Trait convergence and trait divergence in herbaceous plant communities: Mechanisms and consequences. Journal of Vegetation Science 17: 255-260.
• Hagberg, M. 1990. The genus Monotagma (Marantaceae). Doctoral thesis, University of Gothenburg, Gothenburg, Sweden, 90p. (http://hdl.handle.net/2077/11351).
  » http://hdl.handle.net/2077/11351
• Horvitz, C.C.; Le Corff, J. 1993. Spatial scale and dispersion pattern of ant-and bird-dispersed herbs in two tropical lowland rain forests. In: Fleming, T.R.; Estrada, A. (Eds.). Frugivory and Seed Dispersal: Ecological and Evolutionary Aspects Springer Netherlands, Dordrecht, p.351-362.
• Hijmans, R.J. 2017. Package raster: Geographic Data Analysis and Modeling. R package.
• Hoffmann, W.A.; Franco, A.C.; Moreira, M.Z.; Haridasan, M. 2005. Specific leaf area explains differences in leaf traits between congeneric savanna and forest trees. Functional Ecology 19: 932-940.
• Joswig, J.S.; Wirth, C.; Schuman, M.C.; Kattge, J.; Reu, B.; Wright, I.J.; et al 2022. Climatic and soil factors explain the two-dimensional spectrum of global plant trait variation. Nature Ecology & Evolution 6: 36-50.
• Kitajima, K. 2002. Do shade-tolerant tropical tree seedlings depend longer on seed reserves? Functional growth analysis of three Bignoniaceae species. Functional Ecology 16: 433-444.
• Kraft, N.J.B.; Valencia, R.; Ackerly, D.D. 2008. Functional traits and niche-based tree community assembly in an Amazonian Forest. Science 322: 580-582.
• Kress, W.J.; Prince, L.M.; Hahn, W.J.; Zimmer, E.A. 2001. Unraveling the evolutionary radiation of the families of the Zingiberales using morphological and molecular evidence. Systematic Biology 50: 926-944.
• Kummerow, C.; Barnes, W.; Kozu, T.; Shiue, J.; Simpson, J. 1998. The tropical rainfall measuring mission (TRMM) sensor package. Journal of Atmospheric and Oceanic Technology 15: 809-817.
• Maas, P.J.M.1972. Costoideae (Zingiberaceae). Flora Neotropica 8: 1-139
• Maas, P.J.M.1977. Renealmia (Zingiberaceae - Zingiberoideae) Costoideae (additions) (Zingiberaceae). Flora Neotropica 18: 1-218
• Maire, V.; Wright, I.J.; Prentice, I.C.; Batjes, N.H.; Bhaskar, R.; van Bodegom, P.M.; et al 2015. Global effects of soil and climate on leaf photosynthetic traits and rates. Global Ecology and Biogeography 24: 706-717.
• Matsuo, T.; van der Sande, M.T.; Amissah, L.; Dabo, J.; Mohammed Abdul, S.; Poorter, L. 2024. Herbaceous species and dry forest species have more acquisitive leaf traits than woody species and wet forest species. Functional Ecology 38: 194-205.
• Moulatlet, G.M.; Costa, F.R.C.; Rennó, C.D.; Emilio, T.; Schietti, J. 2014. Local hydrological conditions explain floristic composition in lowland Amazonian forests. Biotropica 46: 395-403.
• Moulatlet, G.M.; Rennó, C.D.; Figueiredo, F.O.G.; Ruokolainen, K.; Banon, L.; Emilio, T.; et al 2022. The role of topographic-derived hydrological variables in explaining plant species distributions in Amazonia. Acta Amazonica 52: 218-228.
• Moulatlet, G.M.; Zuquim, G.; Figueiredo, F.O.G.; Lehtonen, S.; Emilio, T.; Ruokolainen, K.; et al 2017. Using digital soil maps to infer edaphic affinities of plant species in Amazonia: Problems and prospects. Ecology and Evolution 7: 8463-8477.
• Muscarella, R.; Uriarte, M. 2016. Do community-weighted mean functional traits reflect optimal strategies? Proceedings of the Royal Society B: Biological Sciences 283: 20152434.
• Nagelkerke, N. 1991. A note on a general definition of the coefficient of determination. Biometrika 78: 691-692.
• Oliveira, R.S.; Costa, F.R.C.; van Baalen, E.; de Jonge, A.; Bittencourt, P.R.; Almanza, Y.; et al 2019. Embolism resistance drives the distribution of Amazonian rainforest tree species along hydro-topographic gradients. New Phytologist 221: 1457-1465.
• Perea, R.; Schroeder, J.W.; Dirzo, R. 2022. The herbaceous understory plant community in the context of the overstory: An overlooked component of tropical diversity. Diversity 14: 800. doi.org/10.3390/d14100800
  » https://doi.org/10.3390/d14100800
• Pérez-Harguindeguy, N.; Díaz, S.; Garnier, E.; Lavorel, S.; Poorter, H.; Jaureguiberry, P.; et al 2013. New handbook for standardised measurement of plant functional traits worldwide. Australian Journal of Botany 61: 167-234.
• Poorter, H.; Niinemets, Ü.; Poorter, L.; Wright, I.J.; Villar, R. 2009. Causes and consequences of variation in leaf mass per area (LMA): a meta-analysis. New Phytologist 182: 565-588.
• Quesada, C.A.; Lloyd, J.; Anderson, L.O.; Fyllas, N.M.; Schwarz, M.; Czimczik, C.I. 2011. Soils of Amazonia with particular reference to the RAINFOR sites. Biogeosciences 8: 1415-1440.
• Quesada, C.A.; Phillips, O.L.; Schwarz, M.; Czimczik, C.I.; Baker, T.R.; Patiño, S.; et al 2012. Basin-wide variations in Amazon forest structure and function are mediated by both soils and climate. Biogeosciences 9: 2203-2246.
• R Core Team. 2022. R: A Language and Environment for Statistical Computing R Foundation for Statistical Computing, Vienna, Austria.
• Reich, P.B. 2014. The world-wide ‘fast-slow’ plant economics spectrum: a traits manifesto. Journal of Ecology 102: 275-301.
• Schietti, J.; Martins, D.; Emilio, T.; Souza, P.F.; Levis, C.; Baccaro, F.B.; et al 2016. Forest structure along a 600 km transect of natural disturbances and seasonality gradients in central-southern Amazonia. Journal of Ecology 104: 1335-1346.
• Schmitt, S.; Hérault, B.; Ducouret, É.; Baranger, A.; Tysklind, N.; Heuertz, M.; et al 2020. Topography consistently drives intra- and inter-specific leaf trait variation within tree species complexes in a Neotropical forest. Oikos 129: 1521-1530.
• Stasinopoulos, D.M.; Rigby, R.A. 2008. Generalized Additive Models for Location Scale and Shape (GAMLSS) in R. Journal of Statistical Software 23: 1-46.
• Sterck, F.; Markesteijn, L.; Schieving, F.; Poorter, L. 2011. Functional traits determine trade-offs and niches in a tropical forest community. Proceedings of the National Academy of Sciences 108: 20627-20632.
• Swenson, N.G. 2009. Herbaceous monocot plant form and function along a tropical rain-forest light gradient: a reversal of dicot strategy. Journal of Tropical Ecology 25: 103-106.
• ter Steege, H.; Poorter, L.; Aguirre-Gutiérrez, J.; Fortunel, C.; Magnusson, W. E.; Phillips, O. L.; et al 2025. Functional composition of the Amazonian tree flora and forests.Communications Biology, 8: 355. doi:10.1038/s42003-025-07768-8
  » https://doi.org/10.1038/s42003-025-07768-8
• Tuomisto, H.; Moulatlet, G.M.; Balslev, H.; Emilio, T.; Figueiredo, F.O.G.; Pedersen, D.; et al 2016. A compositional turnover zone of biogeographical magnitude within lowland Amazonia. Journal of Biogeography 43: 2400-2411.
• Tuomisto, H.; Van doninck, J.; Ruokolainen, K.; Moulatlet, G.M.; Figueiredo, F.O.G.; Sirén, A.; et al 2019. Discovering floristic and geoecological gradients across Amazonia. Journal of Biogeography 46: 1734-1748.
• Vilela, B.; Villalobos, F. 2015. letsR: a new R package for data handling and analysis in macroecology.Methods in Ecology and Evolution6: 1229-1234.
• Violle, C.; Navas, M.-L.; Vile, D.; Kazakou, E.; Fortunel, C.; Hummel, I.; et al 2007. Let the concept of trait be functional! Oikos 116: 882-892.
• Westoby, M. 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant and Soil 199: 213-227.
• Wright, I.J.; Reich, P.B.; Westoby, M.; Ackerly, D.D.; Baruch, Z.; Bongers, F.; et al 2004. The worldwide leaf economics spectrum. Nature 428: 821-827.
• Wright, I.J.; Dong, N.; Maire, V.; Prentice, I.C.; Westoby, M.; Díaz, S.; et al 2017. Global climatic drivers of leaf size. Science 357: 917-921.
• Zuquim, G.; Stropp, J.; Moulatlet, G.M.; Van Doninck, J.; Quesada, C.A.; Figueiredo, F.O.G.; et al 2019. Making the most of scarce data: Mapping soil gradients in data‐poor areas using species occurrence records. Methods in Ecology and Evolution 10: 788-801.

• CITE AS:
  Figueiredo, F.O.G.; Moulatlet, G.M.; Zuquim, G.; Emilio, T.; Tuomisto, H.; Silveira, M.; Rodrigues, D.; Costa, F.R.C. 2025. Functional biogeography of herbaceous assemblages along edaphic and climatic gradients in Amazonian forests. Acta Amazonica 55: e55bc24271.

Data availability

The data that support the findings of this study are available in the GitHub repository of the corresponding author. New data is attached to the “Supplementary Material” section.

SUPPLEMENTARY MATERIAL

Figueiredo et al. Functional biogeography of herbaceous assemblages along edaphic and climatic gradients in Amazonian forests

Figure S1
Moran’s I correlograms calculated for the community weight means (CWM) of each functional trait (LMA, plant height and seed size) of Zingiberales and also for the environmental variables tested for their effects on the CWM (soil fertility, HAND - the height above the nearest drainage, and climate - measured as the dry season length). Moran’s I correlograms were calculated for 100 classes of distance, roughly representing every 10 km in the geographic space with equidistant classes.

Figure S2
Biplot of the PCA performed with the CWM of three functional traits of Zingiberales in Amazonian sampling plots used in this study (LMA, seed size and plant height). This step is necessary prior to the analysis when variables have different magnitudes. The traits were scaled prior to the analyses using the function scale from the base R package.

Figure S3
Partial regression from the multiple linear regression model performed with the PC1 (derived from the functional traits; Figure S2) and the explanatory variable sum of bases (logSB), dry season length (dsl) and vertical distance to the drainage (logHAND). The regression was significant at p = 0.001 (Adjusted R2 = 0.16).

Table S1
List of the 192 ginger species and their functional trait values. LMA denotes the leaf mass per area. Species not fully identified have morphotype names as reference for the local voucher collection.

Family	Species	LMA (gm-2)	Plant height (m)	Seed size (cm2)
Costaceae	Chamaecostus acaulis	26.59	0.2	0.074
Costaceae	Chamaecostus congestiflorus	71.93	0.8	0.074
Costaceae	Chamaecostus fragilis	33.46	0.6	0.074
Costaceae	Chamaecostus fusiformis	45.89	0.7	0.074
Costaceae	Chamaecostus sp1	19.34	0.6	0.074
Costaceae	Chamaecostus sp1 pec	17.67	0.6	0.074
Costaceae	Chamaecostus sp2	29.26	0.6	0.074
Costaceae	Chamaecosuts fragilis	27.18	0.6	0.074
Costaceae	Costus acreanus	53.52	1.7	0.074
Costaceae	Costus amazonicus	59.98	2.0	0.074
Costaceae	Costus arabicus	42.23	1.8	0.074
Costaceae	Costus fusiformis	57.44	0.7	0.074
Costaceae	Costus guanaiensis var guanaiensis	57.09	2.0	0.074
Costaceae	Costus lasius	40.67	1.0	0.074
Costaceae	Costus scaber	54.11	1.5	0.074
Costaceae	Costus spiralis	65.52	2.0	0.074
Costaceae	Costus sprucei	40.58	1.5	0.074
Costaceae	Costus zingiberoides	28.81	1.5	0.074
Costaceae	Costus sp1 acre	43.45	1.5	0.074
Costaceae	Costus sp1 cunia	42.29	1.5	0.074
Costaceae	Costus sp1 m01	40.58	1.5	0.074
Costaceae	Costus sp1 pec	19.36	1.5	0.074
Costaceae	Costus sp1 virua_maraca	46.11	1.5	0.074
Costaceae	Costus sp2	27.85	1.7	0.074
Costaceae	Costus sp2 m01	40.58	1.5	0.074
Costaceae	Costus sp3 pec	33.07	1.5	0.074
Costaceae	Costus sp4	23.16	1.6	0.074
Costaceae	Costus sp5	31.79	1.8	0.074
Costaceae	Costus sp6	31.61	1.7	0.074
Costaceae	Costus sp7	40.58	1.5	0.074
Costaceae	Costus sp7 pec	59.03	1.8	0.074
Costaceae	Costus sp8	54.98	1.7	0.074
Costaceae	Dimerocostus strobilaceus	49.30	2.5	0.074
Heliconiaceae	Heliconia acuminata	43.78	0.9	0.296
Heliconiaceae	Heliconia aemygdiana	30.92	2.2	0.821
Heliconiaceae	Heliconia bihai	48.09	2.0	0.350
Heliconiaceae	Heliconia brachyantha	52.14	1.4	0.236
Heliconiaceae	Heliconia chartacea	44.46	2.2	0.334
Heliconiaceae	Heliconia densiflora var angustifolia	36.10	1.5	0.361
Heliconiaceae	Heliconia densiflora var densiflora	30.73	1.0	0.361
Heliconiaceae	Heliconia hirsuta	25.05	1.7	0.334
Heliconiaceae	Heliconia juliani	53.20	2.0	0.304
Heliconiaceae	Heliconia juruana	88.56	2.2	0.334
Heliconiaceae	Heliconia lasiorachis	42.89	1.7	0.247
Heliconiaceae	Heliconia lourtegiae	51.42	0.6	0.334
Heliconiaceae	Heliconia psittacorum	40.71	1.3	0.334
Heliconiaceae	Heliconia rostrata	49.69	2.5	0.334
Heliconiaceae	Heliconia schumanniana	52.14	1.6	0.227
Heliconiaceae	Heliconia spathocircinata	42.73	1.8	0.334
Heliconiaceae	Heliconia stricta	44.27	1.7	0.371
Heliconiaceae	Heliconia tenebrosa	48.72	0.8	0.334
Heliconiaceae	Heliconia velutina	46.80	1.8	0.350
Heliconiaceae	Heliconia sp1 chandless	43.03	0.8	0.334
Heliconiaceae	Heliconia sp1 pec	44.09	1.5	0.334
Heliconiaceae	Heliconia sp 46	39.79	2.0	0.334
Marantaceae	Calathea crotalifera	45.10	1.7	0.225
Marantaceae	Calathea lutea	52.20	2.5	0.250
Marantaceae	Calathea striata	44.49	0.9	0.237
Marantaceae	Ctenanthe eriacae	37.86	2.0	0.212
Marantaceae	Ctenanthe sp1 acre	35.66	0.2	0.212
Marantaceae	Ctenanthe sp1 pime	32.03	0.5	0.212
Marantaceae	Goeppertia altissima	63.53	1.3	0.454
Marantaceae	Goeppertia attenuata	53.23	1.0	0.234
Marantaceae	Goeppertia cannoides	70.44	1.5	0.234
Marantaceae	Goeppertia capitata	42.48	2.0	0.217
Marantaceae	Goeppertia comosa	46.39	2.0	0.217
Marantaceae	Goeppertia curaraya	35.19	0.2	0.234
Marantaceae	Goeppertia elliptica	50.49	0.7	0.234
Marantaceae	Goeppertia exscapa	29.13	0.4	0.234
Marantaceae	Goeppertia fragilis	50.65	1.7	0.234
Marantaceae	Goeppertia fucata	42.34	0.2	0.106
Marantaceae	Goeppertia hopkinsii	25.45	0.7	0.234
Marantaceae	Goeppertia loeseneri	37.89	0.8	0.212
Marantaceae	Goeppertia maassiorum	50.50	0.5	0.234
Marantaceae	Goeppertia metallica	25.97	0.4	0.196
Marantaceae	Goeppertia micans	36.56	0.2	0.144
Marantaceae	Goeppertia microcephala	43.84	0.2	0.147
Marantaceae	Goeppertia neblinensis	42.38	0.8	0.234
Marantaceae	Goeppertia ovata	27.44	0.5	0.234
Marantaceae	Goeppertia panamensis	19.98	0.3	0.234
Marantaceae	Goeppertia picturata	40.80	0.3	0.234
Marantaceae	Goeppertia polytricha	29.03	0.3	0.234
Marantaceae	Goeppertia straminea	82.77	0.8	0.234
Marantaceae	Goeppertia taeniosa	35.12	0.3	0.234
Marantaceae	Goeppertia variegata	44.55	1.7	0.454
Marantaceae	Goeppertia zingiberina	62.30	0.8	0.279
Marantaceae	Goeppertia sp m2	36.17	0.5	0.234
Marantaceae	Goeppertia sp m3	36.17	0.5	0.234
Marantaceae	Goeppertia sp m4	36.17	0.5	0.234
Marantaceae	Goeppertia sp m5	36.17	0.5	0.234
Marantaceae	Goeppertia sp m9	36.17	0.5	0.234
Marantaceae	Goeppertia sp1 acre	30.34	0.4	0.234
Marantaceae	Goeppertia sp brev m10	36.17	0.5	0.234
Marantaceae	Goeppertia sp brev uat	35.93	0.7	0.234
Marantaceae	Goeppertia sp1 pec	24.65	0.3	0.234
Marantaceae	Goeppertia sp11 alt	52.17	1.5	0.454
Marantaceae	Goeppertia sp15	20.26	0.3	0.234
Marantaceae	Goeppertia sp16	30.22	0.4	0.234
Marantaceae	Goeppertia sp17	28.02	0.2	0.234
Marantaceae	Goeppertia sp18	19.33	0.3	0.234
Marantaceae	Goeppertia sp19	37.08	0.3	0.234
Marantaceae	Goeppertia sp2 orna virua	65.39	1.8	0.234
Marantaceae	Goeppertia sp20	11.44	0.4	0.234
Marantaceae	Goeppertia sp21	35.31	0.3	0.234
Marantaceae	Goeppertia sp22	36.17	0.5	0.234
Marantaceae	Goeppertia sp24	36.17	0.5	0.234
Marantaceae	Goeppertia sp25	25.69	0.3	0.234
Marantaceae	Goeppertia sp26	35.59	0.6	0.234
Marantaceae	Goeppertia sp28	36.09	0.3	0.234
Marantaceae	Goeppertia sp29	61.06	0.5	0.234
Marantaceae	Goeppertia sp3	36.17	0.5	0.234
Marantaceae	Goeppertia sp3 acre	26.56	0.3	0.234
Marantaceae	Goeppertia sp30	37.00	0.6	0.234
Marantaceae	Goeppertia sp31	23.40	0.4	0.234
Marantaceae	Goeppertia sp32	31.53	0.3	0.234
Marantaceae	Goeppertia sp33	43.49	0.1	0.234
Marantaceae	Goeppertia sp35	30.01	0.3	0.234
Marantaceae	Goeppertia sp36	24.56	0.2	0.234
Marantaceae	Goeppertia sp38	36.17	0.5	0.234
Marantaceae	Goeppertia sp39	36.17	0.5	0.234
Marantaceae	Goeppertia sp4	48.43	0.7	0.234
Marantaceae	Goeppertia sp40	28.12	0.3	0.234
Marantaceae	Goeppertia sp41	44.18	0.3	0.234
Marantaceae	Hylaeanthe hexantha	20.19	0.4	0.157
Marantaceae	Hylaeanthe unilateralis	19.41	0.6	0.118
Marantaceae	Ischnosiphon arouma	58.24	2.0	0.825
Marantaceae	Ischnosiphon cannoideus	43.40	1.3	0.200
Marantaceae	Ischnosiphon cerotus	43.01	1.5	0.670
Marantaceae	Ischnosiphon crassispicus	150.87	3.0	0.512
Marantaceae	Ischnosiphon gracilis	47.50	1.7	0.503
Marantaceae	Ischnosiphon grandibracteatus	49.00	3.0	0.962
Marantaceae	Ischnosiphon hirsutus	32.14	0.8	0.261
Marantaceae	Ischnosiphon killipii	29.82	3.0	0.746
Marantaceae	Ischnosiphon lasiocoleus	47.20	1.6	0.393
Marantaceae	Ischnosiphon leucophaeus	58.62	0.8	0.234
Marantaceae	Ischnosiphon longiflorus	42.92	4.0	1.292
Marantaceae	Ischnosiphon martianus	57.32	1.2	0.503
Marantaceae	Ischnosiphon obliquus	27.44	2.5	0.605
Marantaceae	Ischnosiphon paryrizinho	60.17	2.0	0.512
Marantaceae	Ischnosiphon petiolatus	53.69	0.9	0.548
Marantaceae	Ischnosiphon puberulus var puberulus	35.55	2.5	0.512
Marantaceae	Ischnosiphon puberulus var verruculosus	48.63	2.5	1.060
Marantaceae	Ischnosiphon sp1	29.74	3.0	0.512
Marantaceae	Ischnosiphon sp10	48.24	1.9	0.512
Marantaceae	Ischnosiphon sp2	55.22	3.0	0.512
Marantaceae	Ischnosiphon sp4	50.29	3.0	0.512
Marantaceae	Ischnosiphon sp5	59.72	1.5	0.512
Marantaceae	Ischnosiphon sp6	48.24	1.9	0.512
Marantaceae	Ischnosiphon sp7	65.71	1.8	0.512
Marantaceae	Ischnosiphon sp8	50.65	2.0	0.512
Marantaceae	Ischnosiphon sp9	44.16	3.0	0.512
Marantaceae	Ischnosiphon surumuensis	55.39	2.0	0.512
Marantaceae	Ischnosiphon ursinus	47.18	1.3	0.512
Marantaceae	Maranta ciclophylla	16.97	0.3	0.314
Marantaceae	Maranta humilis	20.07	0.4	0.314
Marantaceae	Maranta sp1 acre	23.29	0.3	0.314
Marantaceae	Maranta sp1 jurua	21.55	0.3	0.314
Marantaceae	Maranta sp1 pec	20.22	0.3	0.314
Marantaceae	Maranta sp1 pime	25.78	0.3	0.314
Marantaceae	Maranta sp2 acre	9.77	0.2	0.314
Marantaceae	Maranta sp2 jurua	24.94	0.4	0.314
Marantaceae	Maranta sp2 pime	18.75	0.3	0.314
Marantaceae	Maranta sp3 pime	15.04	0.3	0.314
Marantaceae	Monophyllanthe araracuarensis	59.57	0.7	0.118
Marantaceae	Monophyllanthe oligophylla	57.17	0.1	0.118
Marantaceae	Monotagma angustissimum	44.98	1.0	0.396
Marantaceae	Monotagma aurantispathum	29.64	0.8	0.227
Marantaceae	Monotagma breviscapum	35.96	0.7	0.385
Marantaceae	Monotagma contractum	47.59	0.7	0.205
Marantaceae	Monotagma contrariosum	44.15	1.0	0.271
Marantaceae	Monotagma densiflorum	55.37	1.3	0.227
Marantaceae	Monotagma exile	41.66	0.4	0.248
Marantaceae	Monotagma flavicomum	30.30	0.3	0.428
Marantaceae	Monotagma floribundum	63.34	1.6	0.286
Marantaceae	Monotagma humile	29.77	0.4	0.225
Marantaceae	Monotagma juruanum	42.57	0.7	0.440
Marantaceae	Monotagma laxum	50.52	1.0	0.302
Marantaceae	Monotagma lilacinum	46.24	0.7	0.221
Marantaceae	Monotagma plurispicatum	48.76	0.8	0.271
Marantaceae	Monotagma secundum	53.25	0.7	0.324
Marantaceae	Monotagma tomentosum	51.75	0.6	0.319
Marantaceae	Monotagma tuberosum	26.04	0.2	0.226
Marantaceae	Monotagma ulei	38.73	0.5	0.211
Marantaceae	Monotagma vaginatum	44.93	0.7	0.402
Marantaceae	Monotagma sp1	57.18	0.7	0.286
Marantaceae	Monotagma sp1 acre	54.93	0.7	0.286
Marantaceae	Monotagma sp2 pime	51.91	0.7	0.286
Marantaceae	Monotagma sp3 m10	35.78	0.3	0.286
Marantaceae	Monotagma sp3 pime	46.19	0.4	0.286
Marantaceae	Monotagma sp4 pime	42.75	0.6	0.286
Marantaceae	Monotagma sp5 cunia	43.50	0.6	0.286
Marantaceae	Monotagma sp5 jurua	42.99	0.7	0.286
Strelitziaceae	Phenakospermum guyannense	64.07	5.0	0.605
Zingiberaceae	Renealmia acreana	50.38	1.5	0.059
Zingiberaceae	Renealmia alpinia	44.14	2.0	0.063
Zingiberaceae	Renealmia aromatica	48.93	1.7	0.067
Zingiberaceae	Renealmia breviscapa	52.62	0.8	0.063
Zingiberaceae	Renealmia cernua	58.16	1.5	0.057
Zingiberaceae	Renealmia floribunda	66.37	1.5	0.054
Zingiberaceae	Renealmia monosperma	40.56	0.3	0.076
Zingiberaceae	Renealmia thyrsoidea	39.24	1.5	0.108
Zingiberaceae	Renealmia sp1	38.09	1.5	0.067
Zingiberaceae	Renealmia sp1 br319	45.85	1.0	0.067
Zingiberaceae	Renealmia sp1 pec	45.85	1.0	0.067
Zingiberaceae	Renealmia sp1 pime	45.85	1.0	0.067
Zingiberaceae	Renealmia sp10	49.30	1.5	0.067
Zingiberaceae	Renealmia sp4 jurua	32.88	0.4	0.067
Zingiberaceae	Renealmia sp5 jurua	51.48	0.4	0.067
Zingiberaceae	Renealmia sp7 jurua	39.76	0.6	0.067
Zingiberaceae	Renealmia sp8	40.76	1.0	0.067
Zingiberaceae	Renealmia sp9 jurua	45.85	1.0	0.067

Edited by

• ASSOCIATE EDITOR:
  Natália Ivanauskashttps://orcid.org/0000-0001-6643-8476

Publication Dates

• Publication in this collection
  08 Dec 2025
• Date of issue
  2025

History

• Received
  06 Aug 2024
• Accepted
  16 June 2025