4.1 Environmental variation
Three groups of plots were identified in each of the two environments, influenced by different variables that act together or individually. A significant percentage of the silt content in the igapo forest was found in a group of igapo plots located downstream of the Jari River, near its junction with the Amazon River. This suggests the possibility that alluvial sediments, deposits from whitewater rivers, may have penetrated blackwater rivers in the past, perhaps caused by extreme flooding events (Haugaasen and Peres 2006; Irion et al 2011; Assis et al 2015; Householder et al 2021).
A group of sites in an intermediate position along the Jari river axis (igapó) was strongly associated with high concentrations of micronutrients (Fe, Zn, Mn and Cu), as well as high CEC and Al. Notably, high Fe, Mn, Al concentrations demonstrated high levels of toxicity in the environment (Assis et al 2015). Upstream, higher concentrations of sand, acidity (pH), flooded days and potassium were observed. In any case, generally low soil pH and high Al concentration can indicate intense weathering (Quesada et al 2011).
A large number of plots that were geographically located near the junction of the Amazon and Mazagão rivers formed a distinct cluster in the PCA of environmental factors. Differences in the Ca concentration, SB, CEC and sand at locations downstream can generally be explained by complex river flow patterns driven by interacting daily tidal ebbs and flows from the ocean to the east with flood pulses, and an annual flow of rivers driven by seasonal rainfall patterns in distant reaches of the eastern Amazon basin.
Specifically, high levels of sand may be associated with sediment transport by tidal currents from the Amazon River with sufficient force to invade the Mazagão River, causing a deposition of sandy sediments further downstream (mouth). As the tide enters towards the upstream of the Mazagão River, it loses energy and consequently forms clayey soils. Areas located upstream from the Mazagão River were associated with high levels of organic material, high P availability, and high concentrations of micronutrients; these patterns may be associated with decreased tidal influence in these further upstream areas (Quesada et al 2011).
4.2 Distribution of species along the environmental gradient
CCA was used as a tool to interpret potential relationships between species distribution and measured environmental variables. The analyses revealed that despite the majority of species in both environments not showing a specific pattern associated with the environmental differences presented by the PCA, this does not compromise the results since the correlations between the abundance of species and the variables selected in the final CCA were high and significantly correlated in both environments.
Acidity (pH), Al, flooded days, K and sand, in decreasing order of importance, were the variables in the igapo forest which showed the highest correlations with the two axes of the CCA. However, our results showed high edaphic heterogeneity along the environmental gradient with chemical and texture variations downstream upstream of the Jari River. There was a predominance of sandy soils upstream, positively associated with flooded days, K, pH and Al.
Igapo forest soils are often associated with high acidity and nutritional poverty, as described in this work (Haugaasen and Peres 2006; Quesada et al 2011). Surprisingly, the relationship between species and variables when associated with differences in the environmental gradient along the downstream and upstream axis did not show relevant effects on the distribution of species, which may explain their alignment in the center of the diagram in the CCA. It was identified that 28 of the 66 species in the igapo forest (42.42%) did not show affinity with any of the selected variables.
At any rate, low pH values associated with flooding tend to decrease the availability of other nutrients, which reinforces the thesis of an environment with limiting conditions. This pattern suggests that the availability of various nutrients to plants is indirectly influenced by pH. The distribution of species in an environment of limited resources must be influenced by specific nutrients which may, despite being limiting, favor the release of other nutrients (John et al 2007).
Several soil attributes correlate with flood regime for floodplain forests, as indicated by the position of environmental vectors in the CCA. Plots with fewer days flooded mainly tended to have more silt content, higher pH and higher P concentration. While species-environment associations found here can be interpreted as differential species response to a heterogeneous environment (i.e. habitat filtering), alternative mechanisms can also explain the patterns, such as dispersion limitation and priority effects (Assis et al 2015). It was identified that 21 of the 39 species in the lowland forest (51.3%) did not show affinity with any of the selected variables.
There was also a strong spatial segregation of the sample units in the lowland forest along the Mazagão River with the formation of groups from the mouth towards the head of the river. A large group was formed at the mouth of the Amazon (downstream), consisting of seven plots. The formation of plots in the median and upstream portion towards the head of the Mazagão River is not as clear, as shown in the PCA.
Significant correlations regarding the environmental gradient formed were observed between flooding, CEC, SB and K, strongly influencing the grouping of sampling units downstream. On the other hand, silt, Ca, P and pH show a significant association with related sampling units from the middle third of the Mazagão River. Nonetheless, it seems clear, as stated earlier, that the strength of the Amazon River on the Mazagão River seems to be preponderant in structuring of the environment. Although the effects of nutrients and substrate seem to be determinant in the distribution of species, as described by (Ferreira and Almeida 2005; Haugaasen and Peres 2006; Budke et al 2007; Filizola Jr and Guyot 2009; Piedade et al 2011), this is not evident in this work. The trend towards low habitat specificity could explain the widespread distribution of many floodplain species in the Amazon (Wittmann et al 2011).
The present study demonstrates that some species occurred far from the center of the ordination diagram in both environments, which indicates that most species have a high tolerance in relation to some environmental variables used. As a consequence, the species present a generalist distribution pattern. According to Rodrigues et al. (2007), species gradually have different tolerance intervals for environmental variables. A species reaches its maximum abundance level at some ideal time along resource gradients or habitat conditions (Damasco et al 2013), from which abundance declines to establish more extreme conditions.
However, in order to characterize species with respect to their preferred habitats, it is necessary that the trends demonstrated by a study in species are also observed in other study areas (Dalanesi et al 2004). Most of the species identified in the present work are generalists, occurring throughout the entire soil and flood gradient. For example, Pentaclethra macroloba (Willd.) Kuntze, Hevea brasiliensis (Willd. Ex A. Juss.) Müll. Arg. and Virola surinamensis (Rol. Rottb ex.) Warb. are distributed among almost all locations in both environments (Marinho et al 2013; Braga and Jardim 2016; Dantas et al 2021).
Few species appear to have a high degree of habitat specificity in relation to hydro-edaphic conditions. Root traits are adaptations exhibited by most tree species that are flood tolerant. Specific traits that increase tolerance to flooding conditions may explain the wide distribution of some species along the flood gradient we studied (Parolin and Wittmann 2011). However, the flood period is substantially different across the entire Amazon basin, with variations that depend heavily on topography, ranging up to 240 days/year.
This variability in flood duration could control the distribution of species along the flood gradient of Amazon forests (Wittmann et al 2006; Wittmann et al 2011). However, the interval of days flooded in this study varied by 25 and 80 days for floodplain and igapo forest, respectively. In this sense, a variation in the duration of flooding in our study is relatively similar compared to other regions of the Amazon (Targhetta et al 2015; Householder et al 2021). Thus, it should affect the distribution of species in a different way when compared to a regional scale.
While the distribution of species in our study does not differ with environmental variation between the two environments, the vast majority of studies suggest that hydro-edaphic variability is an important source of habitat heterogeneity, and thus beta diversity, at landscape scales (Wittmann et al 2006; Filizola Jr and Guyot 2009). Ultimately, it is suggested that the species distribution in the floodplain forest studied herein strongly responds to a greater number of environmental variables. In contrast, species distributions in the igapo forest strongly respond to variability in soil acidity and aluminum toxicity.
Our study suggests that small-scale environmental heterogeneity within and between different types of forest wetlands has large differences with respect to location, distribution and community composition (Albernaz et al 2012). Thus, we mainly attribute the differences in composition and species distribution observed in this study to species responses to a heterogeneous environment, as opposed to alternative mechanisms that can produce similar spatial pattern, such as dispersal limitation.