Mapping and Environmental Diagnosis in Native Acai Areas in the Amazon

For several decades, the Acai orchards (acaizais) have directly influenced the survival of the families in the Amazonian floodplains. In this period, the production of the Acai fruit for local consumption was ceased and became an export item produced in intensive management, resulting in an increase in orchards in the floodplains and the emergence of dryland plantations, no longer representing a typical extractive activity in the Amazon. The objective of this study was to map the classes of use and coverage, and the occurrences of the Acai orchards massifs, as well as to analyze the physical and chemical parameters of five islands in the municipality of Igarapé-Miri, State of Pará, Brazil, where there is a great occurrence of productive Acai orchards. This work evaluated the following islands: Jarimbu, Mamangal, Itaboca, Mutirão, and Buçu, where geolocalized collections were carried out in the areas with the highest occurrence of Acai orchards, both to assist in the classification of images and for soil sampling. August 2019 Planet images were processed using the unsupervised method, where seven classes of cover use were obtained: hydrography, exposed soil, urban, alluvial, lowland, arboreal, and agriculture areas. Therefore, occurrences of productive orchards were identified and correlated to the good attributes of soil fertility in the floodplains under continuous flooding and sedimentation. The correlation confirmed the higher productivity of Acai in the Alluvial and Lowland classes, which predominate in the evaluated area, presenting soils considered fertile with a loam-clay-silty and loam -silty texture, high base saturation (greater than 50%), high organic matter content, and significant presence of potassium and phosphorus.

However, the Acai fruit trade is an alternative with great economic potential and that essentially contributes to the income of extractive communities in the Amazon  promoting the sustainable exploitation of floodplain forests (Homma et al., 2006) and expanding the local and national economies (Tagore et al., 2018).
In this context, studies on the mapping of large areas of Acai orchards in the Legal Amazon are still scarce. Not even the local community has an estimate of the size of the areas explored there, as well as their transformations and impacts over the years. In the most recent studies by Lima et al. (2018) on environmental degradation identified by coverage analysis, the use of technologies was effective for assessing the behavior of vegetation, allowing to monitor and estimate the expansion of changes in the environment. The use of digital image classification techniques provides a synoptic view and information on the temporal dimension of spatial phenomena, allowing the generation of information on the dynamics and spatial patterns of the landscape in areas of large territorial extensions (De Souza et al., 2019).
The use of high-spatial-resolution sensors, as is the case of the Planet sensor with a spatial resolution of three meters, has proven to be real sources in the evaluation of the characteristics of the mapped areas, consisting of reliable, solid, and active information in the analysis of the dynamics of land use and cover.
Knowledge of the soil quality through its physical-chemical and biological attributes is essential to assess impacted and cropped areas (Colodel et al., 2018), as well as the lowland areas of the Amazon (Fajardo et al., 2009). Therefore, knowing the spatial variability of soil attributes may assist in the management of areas of native Acai in this ecosystem with the use of geo-technologies, which supports the high-productive potential of the areas with the rich-natural fertility of the soils and productivity in the islands.
Thus, Rodríguez-Echeverry et al. (2018) recommend that such strategies for changing land use obtained with the use of geotechnologies can serve as crucial information for planning conservation strategies and also for subsidizing programs and other planning and other tools of territorial and environmental management and planning.
In this context, the objective of this work was to analyze through the unsupervised classification of images the thematic maps of representativeness of the areas with the highest occurrence of Acai by the ISODATA algorithm the five islands of the municipality of Igarapé-Miri/PA and to correlate the generated maps with results of the analysis of physical and chemical attributes of the soils of the areas.

Study Area
The study was carried out in the municipality of Igarapé-Miri, located in the Low Tocantins region in the mesoregion in the northeastern Para State. The municipality is located 78 km from the capital of the state of Pará, Brazil, and has a territorial area of 199.679 ha (IBGE, 2019) ( Figure 1). The climate in the region is of the tropical humid type corresponding to the megathermic type Ami according to Köppen's classification. The annual rainfall is greater than 2,000 mm with an average annual temperature of 27 °C and air relative humidity of 80% (Alvares et al., 2013).
Five islands participating in the Agroextractive Settlement Project (PAE) and that are also more representative for the Acai fruit were selected (Table 1).   For noise elimination and image smoothing, the Majority parameters 3 × 3 filter was applied (Vasconcelos et al., 2016). Then, the classification was edited through visual interpretation according to the spectral behavior of each pixel (Duarte & Silva, 2019). After, the file was transformed into a vector file (shapefile) and quantified, generating tabulated data.
The following methodological actions were used to measure the classification process of Planet images: i) Evaluation of errors of omission and commission (statistics of the percentage of errors and successes of the pixel-by-pixel classification, attributed by the Confusion matrices); ii) Post-classification visual inspection checking the distribution and connection between the generated classes, the existence of isolated pixels, and the need for classification editing; iii) Validation and in loco confirmation through the georeferenced collection in the mapped classes at visiting the study area before and after the image classification process.

Soil Sampling
In the areas with the highest occurrence of Acai, disturbed soil samples were collected in the 0-0.10 m layer, at a distance of 50 to 100 m from the river bank and 1 to 2 km from one collection point to another, totaling 19 samples for the five islands. Then, the samples were air-dried, ground, and passed through a 2-mm opening sieve to obtain the air-dried fine earth (ADFE). This sampling was carried out in August, a period of low rainfall in the region. All collection points were geo-referenced with the aid of GPS (Global Position System).

Soil Laboratory Analysis
The chemical attributes of the soil were determined according to the methodology of Teixeira et al. (2017): pH in water and KCl in the soil:solution (1:2.5) rate. Available P and K, extracted by Mehlich-1 (0.0125 mol L -1 of H 2 SO 4 + 0.05 mol L -1 of HCl), and P determined by colorimetry and K by flame photometry; Al, Ca and Mg extracted with KCl 1 mol L -1 ; Al determined by titration (neutralization volume), while Ca and Mg were determined by complexometry with EDTA; H + Al extracted with 1 mol L -1 calcium acetate solution at pH 7.0 and determined by titration. Organic carbon was determined by the modified method of (Walkley & Black, 1934), based on the principle of oxidation of organic matter, with potassium dichromate in the sulfuric medium.
The particle size distribution was determined using the pipette method using 1M NaOH. Before dispersion, samples with an organic matter content > 5% were pre-treated to remove organic matter with H 2 O 2 (Gee & Bauder, 1986). The sand fraction was separated through sieving, the clay by sedimentation, and the silt fraction calculated by the difference.

Land Use and Soil Cover Classification on the Islands
The results obtained for the unsupervised reclassification of land use and the cover of the five islands using the IsoData method generated the identification of ash level patterns. These patterns are defined as samples grouped by clusters of space units (Olofsson et al., 2014).
It was observed the generation of a new classification by grouping applied in the Planet 2019 Images (Figure 3). The images were selected exactly during the Acai harvest period, which provided a good spatial resolution of the areas and, consequently, allowed the identification of the Acai massifs on each island and the most relevant classes. As for the Agriculture class, only one island displayed this characteristic, the Mutirão island with 16.95 (ha) in its extension, representing 1.1%, as this is located closer to the margin of the municipality of Igarapé-Miri and acts as Acai Road, an important highway which is the outlet for all fruit production from the adjacent islands.
Farming carried out by riverside dwellers occurs through clearing and thinning processes in their areas, as a way of implementing some crops, as reported by Silva et al. (2018) in their studies which show that of the 130 settlers on Island Mamangal, 28% have agroforestry systems (AFSs) in their areas, explained mostly by diversification (production), family food and shading of the Acai (Euterpe oleracea Mart.) crop. This reality coincides with what occurs on the other islands, as these are practices found by the riverside people to implement the AFSs as a form of income for these families.
The Urban Area class characterized in this study is represented by houses, villages, and built areas (Table 2). Mutirão Island is the most representative in this class, with 2.58% of the total area. On the Other hand, Jarimbu, Mamangal, Itaboca, and Buçu islands present, respectively, 0.92%, 1.28%, 1.05%, and 1.57% of the urban area ( Figure 4). It was observed that the urban area in all islands is predominantly on the banks of rivers. In general, traditional populations in the Amazon maintain a strong connection with the river network, since they provide connectivity and mobility for their populations (Amaral et al., 2013).
The class characterized as Hydrography is found on all islands. Canto (2007) reports that the floodplains are large strips that border the rivers and are periodically covered by water. However, Tagore et al. (2018) point out that river courses have been being altered by the flow of the vessels, increasing the width of channels. It is worth mentioning that only the internal areas of each island were classified as Hydrography, therefore its outer edge in the image was not classified as in (Figure 3), which allowed to observe and determine the amount of water within each island.

Classification of the Island Soil
The results of soil chemical attributes are shown in (Table 4). The soils on the islands Mamangal, Itaboca, and Mutirão had the lowest mean pH values in water and KCl, in relation to the soil of the islands Jarimbu and Buçu. The more acidic pH in the soils of the islands of Mamangal, Itaboca, and Mutirão is the outcome of the ferrolysis process, which is common in soils subject to hydromorphism and rich in iron (Van Breemen & Buurman, 2002).
A work of soil genesis with flooding cycles has shown that high acidity has been observed in superficial horizons in lowland soils due to ferrolysis (Barbiero et al., 2010;Coringa et al., 2012;Da Silva et al., 2019). Studies in Gleisoil under native vegetation on the banks of the Guamá River attested acid pH in water and KCl associated with the alternation between flood and ebb tide (Lopes et al., 2006).
The nutrient content in the evaluated soils showed wide distribution and high variation within each area, with base saturation > 50% in the topsoil and also higher values of organic carbon (OC) and (pH) for the Jarimbu and Buçu islands. Freitas et al., (2015 b) in a floodplain study, reports that high values of base saturation and pH are widely used in the indication of soil fertility. While for the other three islands, Mamangal, Itaboca, and Mutirão, base saturation was < 50% and OC with lower values.    Note. OC: organic carbon; H + Al: potential acidity; SB: sum of bases; CEC: cation exchange capacity; V%: base saturation; SD: standard deviation; CV: variation coefficient.
The sum of bases highly contributed to the five soils, which showed high concentrations of Ca 2+ , where the highest mean value was shown in Jarimbu island., 8.38 cmol c dm -3 , which is the island with a high concentration of (OC), therefore corroborating with studies of Neto et al. (2018) indicating that in secondary forests, the natural dynamics promotes the incorporation and maintenance of organic matter in the soil (Delarmelinda et al., 2017) concentrating high fertilization in native areas.
The OC with a highly considerable value for Jarimbu island of 102.48 g kg -1 is justified because on the margin of this there are many sawmills where all the tailings are discharged directly into the waters, and with the movement of flood and ebb from the river, these sediments are deposited, resulting in high values of organic matter in the soil, which was also found by Freitas et al. (2015b) reporting that large scale deposition of fresh organic material on the soil surface periodically without time for humification, explains the highest concentration of organic carbon in the first layer.
The highest average concentrations of Mg 2+ were for the soils on the islands Mamangal and Mutirão (2.21 and 2.38 cmol c dm -3 ), respectively, and the lowest concentration value was found for the soils of the islands Jarimbu, Itaboca, and Buçu, 1.31 cmol c dm -3 ; 1.77 cmol c dm -3 and 1.98 cmol c dm -3 , respectively; the contents of K + ; Al + showed a small dispersion in the five soils in relation to the other chemical variables, showing a lower K + value (0.39 cmol c dm -3 ) for Mamangal island and lower Al + value for Mutirão island (0.19 cmol c dm -3 ) and higher mean values of K + and Al + (0.47 and 0.23 cmol c dm -3 ) only on the Jarimbu island.
K + is found in large amounts in plant tissues with energetic functions, and water absorption by cells (Forster et al., 2019), where all the evaluated elements favored good spatialization of the Acai palm areas in these islands, presenting higher alluvial classes in this study.
The high P concentration in the soils on the Jarimbu islands, 90.61 mg dm -3 ; Mutirão, 69.20 mg dm -3 , and Buçu 86.27 mg dm -3 , has a great contribution on organic matter, Where P is a fundamental element for the initial development of Acai plant, promoting a greater increase in the aerial part and biomass when in high availability (Araújo et al., 2018).
The cation exchange capacity and the potential acidity had a high contribution to the exchange complex in all five soils. Salviano et al. (1998) also report the similarities between CEC and potential acidity (H + Al) and how much they exert direct influence on the CEC calculation. This was also found by Freitas et al. (2015b), in which high values reflect the natural conditions of this soil and have a high capacity to retain cations.
The soils on Jarimbu, Mamangal, Itaboca, and Mutirão islands presented a silty clay-loam textural class. The soils of Buçu Island, on the other hand, presented a silt-loam textural class. However, in the soil of Jarimbu island., sand contributed more, 135.74 g kg -1 in comparison to the other islands, while the soil of Itaboca island, presented a lower proportion of silt/clay, 1.91 g kg -1 (Table 5).   Although studies for floodplain soils in the islands in the Low Tocantins region are scarce and incipient , the results of this experiment are relevant, as all the five soils showed similar characteristics. According to Guedes et al. (2018), they are fertile soils with fragile peculiarities, in which the soil texture particles under flood are sedimented in the lowest part of the landscape, which induces changes in texture, culminating in siltier soils (Wang et al., 2017).

Conclusions
Based on the results of this experiment, we concluded that in the areas of the Igarapé-Miri islands, large Acai orchards massifs were consolidated, expanded through favorable soil fertility conditions and efficient management practices in the floodplain areas. The zoning and spatial distribution of Acai orchards in the five studied islands (Jarimbu, Mamangal, Itaboca, Mutirão, and Buçu) through the processing and use of image classification algorithms, allowed the correlation of the mappings carried out with the major physical-chemical attributes of the soils.
The mapping established seven classes of use and cover (Hydrography, Exposed Soil, Urban Area, Alluvial, Lowlands, Arboreal, and Agriculture) where it was identified occurrences of productive orchards, correlated to the soil fertility attributes naturally imposed under flooding and sedimentation in the lowest parts in the floodplains. This correlation confirmed the indications of higher productivity of the Acai fruit in the Alluvial and Lowland classes. In these classes of greater predominance in the studied area, the soils are considered fertile and presented similar physical-chemical characteristics with a loam-clay-silty and loam-silty texture, high base saturation (greater than 50%), high levels of organic matter, and significant presence of potassium and phosphorus.
The local community started to combine efficient management practices with environmental sustainability, with positive socio-environmental and cultural returns, resulting in promoting employment and income in the periods of Acai harvest (July to December) and off-season (January to July), supported by other local economic activities (fishing and other family farming activities).