Agroforest Syst DOI 10.1007/s10457-017-0137-y
The role of fertile anthropogenic soils in the conservation of native and exotic agrobiodiversity in Amazonian homegardens Nathalia B. de Souza . Andre´ Braga Junqueira . Paul C. Struik . Tjeerdjan Stomph . Charles R. Clement
Received: 26 December 2016 / Accepted: 27 September 2017 Ó Springer Science+Business Media B.V. 2017
Abstract Amazonian dark earths (ADE) are anthropogenic soils mostly created between 500 and 2500 years ago by pre-Columbian populations. ADE are currently used by local people for different agricultural and agroforestry systems. Because of their high fertility they may play an important role in the conservation of non-native agrobiodiversity. This study aimed to investigate the variation in richness and abundance of exotic and native species in
Electronic supplementary material The online version of this article (doi:10.1007/s10457-017-0137-y) contains supplementary material, which is available to authorized users. N. B. de Souza (&) Programa de Po´s-Graduac¸a˜o em Botaˆnica, Instituto Nacional de Pesquisas da Amazoˆnia, Av. Andre´ Arau´jo, 2.936 - Petro´polis, Manaus 69067-375, Brazil e-mail:
[email protected] A. B. Junqueira Department of Soil Quality, Wageningen University and Research, P.O. Box 47, 6700 AA Wageningen, The Netherlands
homegardens along the ADE-background soil continuum. We conducted floristic inventories in 70 homegardens located in 7 riverside communities along the lower and middle Madeira River, Central Amazonia. Each species sampled was classified according to its origin: native Amazonian, American (from outside Amazonia) and non-American, and each individual was classified according to its form of establishment: cultivated or spontaneous. The floristic diversity was significantly related to soil fertility, texture and homegarden size. We found a positive relationship between soil fertility and richness of species and landraces. Homegardens on more fertile soils tended P. C. Struik T. Stomph Centre for Crop Systems Analysis, Wageningen University and Research, P.O. Box 430, 6700 AK Wageningen, The Netherlands C. R. Clement Coordenac¸a˜o de Tecnologia e Inovac¸a˜o, Instituto Nacional de Pesquisas da Amazoˆnia, Av. Andre´ Arau´jo, 2.936 - Petro´polis, Manaus 69067-375, Brazil
A. B. Junqueira International Institute for Sustainability, Estrada Dona Castorina 124, Rio de Janeiro 22460-320, Brazil A. B. Junqueira Department of Geography and the Environment, Centre for Conservation and Sustainability Science (CSRio), Pontifical Catholic University of Rio de Janeiro, Rua Marqueˆs de Sa˜o Vicente, 225 – Ga´vea, Rio de Janeiro 22451-900, Brazil
123
Agroforest Syst
to have a higher richness and abundance of cultivated non-American species, as well as a higher richness and abundance of spontaneously established American species. Homegardens at the fertile end of the fertility gradient provided conditions for the establishment and growth of many species, especially exotic species, that are generally more nutrient-demanding than Amazonian species. Our results show that homegarden agroecosystems on ADE favour experimentation with the introduction of a wide range of species from various regions of the globe. Keywords Soil fertility Amazonian dark earths Geographical origin of species Agroecosystems Spontaneous plants
Introduction Amazonia is renown for its nutrient poor soils, but patches of highly fertile soils also occur. Terra Preta or Amazonian dark earths (ADE) are fertile anthropogenic soils formed by human activities in preColombian times through the localized disposal of food waste, charcoal and other types of organic residues (Glaser and Birk 2012). These archaeological sites with modified soils have been reoccupied in modern times by riverine communities (Adams et al. 2009; Fraser et al. 2011b), who are well acquainted with their characteristics and have developed specific management practices and crop assemblages for these areas (Hiraoka et al. 2003; Fraser et al. 2011b; Junqueira et al. 2016a, b). ADE were probably created under homegardens of Native Amazonians, since homegardens are intimately associated with dump heaps, or ‘middens’ (Schmidt et al. 2014). In these environments people deposited large amounts of organic materials, including plant biomass, charcoal, excrements, animal bones and turtle backs, leading to high levels of phosphorous (P), calcium (Ca) and magnesium (Mg) (Glaser and Birk 2012), contrasting with the nutrient-poor Oxisols and Ultisols that predominate in Amazonian uplands (Quesada et al. 2011). Given the association of ADE with habitation sites, other archaeological remains (ceramic fragments and lithic artifacts) are also commonly found in these soils. ADE are very heterogeneous, both within and among sites, regarding
123
their physical and chemical characteristics (Fraser et al. 2011c). In general, the degree of soil transformation and the density of archaeological remains tend to decrease as one moves from the ‘core’ areas of ADE towards surrounding soils (Fraser et al. 2011c; Schmidt et al. 2014; Junqueira et al. 2016a, b). Thus, it would be more appropriate to consider ADE and adjacent soils as a continuum, rather than as distinct soil categories (Fraser et al. 2011c). Among the most common land-use systems on ADE are homegardens—agroecosystems where trees, shrubs and herbs are cultivated in proximity to the house, providing food, income and multiple other livelihood and ecosystem services (Kumar and Nair 2004; Kim et al. 2016; Caballero-Serrano et al. 2016; Mattsson et al. 2017). Some ethno-ecological studies have found that the diversity of plant species in homegardens is partially related to soil characteristics (Fraser et al. 2011a; Kawa et al. 2011; Junqueira et al. 2016a). Besides soil, other biophysical (e.g., climate, local flora and fauna, soil seed banks and homegarden size and age) and socio-cultural (such as ethnicity, personal preferences of owners and processes of social exchange) factors may influence plant diversity and other characteristics of homegardens (Kumar and Nair 2004; Perrault-Archambault and Coomes 2008; Abizaid et al. 2016; Dı´az-Reviriego et al. 2016; Timsuksai and Rambo 2016; Alcudia-Aguilar et al. 2017). Landuse history also plays an important role: in anthropic environments such as ADE, the occurrence of spontaneous plants related to past human activity influences substantially the current plant community in homegardens (Clement et al. 2003; Lins et al. 2015). Recent studies have investigated the ethno-ecology and agrobiodiversity of homegardens on ADE (Major et al. 2005; Fraser et al. 2011a; Lins et al. 2015; Junqueira et al. 2016a). Junqueira et al. (2016a) showed that the variation in soil texture and fertility significantly influences the structure, diversity and floristic composition of homegardens. Other studies have shown that homegardens on ADE often contain greater floristic diversity and more exotic species compared with adjacent soils (Major et al. 2005; Fraser et al. 2011a). While non-native species are an important component of homegardens and other agroforestry systems throughout the world (Jose 2011), the specific characteristics of ADE, particularly their high fertility, may encourage the introduction of and experimentation with non-native species and
Agroforest Syst
landraces as these species often have greater nutrient demands than native species (Clement et al. 2003). Most previous studies have considered ADE and surrounding soils as discrete categories, without taking into account the variation and the heterogeneity within these categories (Junqueira et al. 2016a, b). In contrast, in this study we take into account the whole variation in physical and chemical properties across the continuum from ‘core’ ADE to surrounding soils. Based on the floristic inventories of Junqueira et al. (2016a), we assess the links between the geographical origins of the species and soil fertility. We aim to understand the effect of the continuum between adjacent soils and ADE on the richness and abundance of native and exotic species in homegardens, and thereby to contribute to the understanding of the role of ADE in the introduction and conservation of exotic agrobiodiversity.
Materials and methods Study locations The study was undertaken in seven riverside communities located along the middle and lower Madeira River, Amazonas state, Brazil (Fig. 1). All communities were located partly on ADE, although not all households have access to core areas of ADE or transitional soils. The predominant types of upland soils in the region are Oxisols and Ultisols (IBGE 2010). Along the Madeira River, ADE are found mainly on bluffs along the main river, tributaries and lakes, in patches ranging from two to more than 50 ha in size (Fraser et al. 2011b). The natural vegetation is characterized predominantly by evergreen broad-leaf forests on uplands and flooded forests in floodplains (Rapp Py-Daniel 2007). The region is currently inhabited by caboclos—a Portuguese term that designates populations that emerged in Amazonia from the intermarriage between Amerindians and Europeans or migrants from other Brazilian regions; these populations are a biological and cultural mixture that combine elements of these different societies (Witkoski 2010). These populations generally live close to the main rivers and tributaries, and use both uplands and floodplain areas for subsistence and other economic activities. They make their living from the cultivation of manioc (Manihot
esculenta Crantz), from the extraction of forest products [e.g., Brazil nut (Bertholletia excelsa Bonpl.), rubber (Hevea brasiliensis (Willd. ex A. Juss.) Mu¨ll. Arg.)] and from the cultivation of other crops, such as bananas (Musa paradisiaca L.) and cacao (Theobroma cacao L.) (Junqueira et al. 2016a, b). Fishing is also an important subsistence and commercial activity, especially in the nutrientrich waters of the Madeira River. Data collection We selected between 6 and 13 homegardens in each community, with a total of 70 homegardens from 7 communities. Since we were interested in sampling the largest soil variation possible, the choice of homegardens was done in a way that maximized the variation in terms of soil color and in presence of ceramic artefacts on the soil surface. We went through each homegarden with local residents in guided walks (Albuquerque et al. 2008), during which the homegarden was measured (the limits were indicated by the owner), and all species and landraces (i.e., local crop varieties that are named and maintained by farmers) were identified and counted. All plant life forms were included in the sampling (trees, shrubs, palms, climbing vines, herbs). During the guided walks, residents provided information on the age of the homegarden (i.e., how long ago it had been established) and on whether each individual plant had grown spontaneously (hereafter ‘spontaneous’) or if it had been deliberately planted (hereafter ‘cultivated’; the latter information was obtained for 67 out of 70 homegardens). The identification of the majority of species was done in the field, given that most plants in homegardens are well known and widely used. When identification in the field was not possible, we photographed plants and collected samples for later identification in the herbarium of the National Research Institute for Amazonia (INPA). We classified each species according to its origin: native to Amazonia, from the Americas but outside Amazonia (‘American’, including species from North and Central America, the Caribbean, and South America outside the Amazon Basin) and non-American species (introduced from Africa, Europe, Asia or Oceania), following Leo´n (2000) and Clement (1999). When there was no information in these references, we attributed possible origins based on distribution maps
123
Agroforest Syst Fig. 1 Study locations. The squares represent the seven communities where data were collected along the middle and lower Madeira River, Amazonas, Brazil. The map was constructed as a mosaic of LANDSAT images from 2009 to 2010, obtained from the US geological survey
and information related to endemism obtained in the following databases: Lista de Espe´cies da Flora do Brasil (2015), ‘‘Tropicos.org’’ (2013) and ‘‘The Plant List’’ (2013). In each homegarden we collected composite soil samples, composed of five subsamples obtained from the 0–20 cm soil layer. Ceramic fragments in the samples were left on site, and we avoided collecting subsamples in areas that looked eroded, recently burned or enriched. All soil samples were air-dried and sieved with a 2 mm mesh. Chemical [pH-H2O, available phosphorous (P), exchangeable calcium (Ca), exchangeable magnesium (Mg), exchangeable potassium (K), and exchangeable aluminum (Al), and total manganese (Mn), total iron (Fe) and total zinc (Zn)], organic matter (OM) and physical (percentage of sand, silt and clay) soil analyses were subsequently conducted at the Thematic Laboratory of Soils and Plants at INPA, following the EMBRAPA (2011) protocol. Given that ADE are usually associated with ceramic fragments, we also registered the presence or absence of such fragments on the soil surface during the collection of soil samples.
123
Data analyses Our analytical approach followed the one used by Junqueira et al. (2016a). First, we used a Principal Components Analysis (PCA) to summarize the variation in soil chemical and physical parameters. Prior to ordination, soil variables with skewed distributions were log-transformed (loge) (except soil pH), data in percentages (% of clay, silt and sand) were transformed by the arcsine of the square root divided by 100, and all soil variables were centered and standardized. Only the percentages of sand and clay were included in the analysis, given that the three texture variables are complementary (i.e., they add up to 100%). The first two axes of the PCA, which summarize most of the variation of chemical and physical properties of the soils, were used as predictor variables in subsequent analyses. In order to evaluate the effect of soils (PCA axes), size and age of homegardens on the dependent variables [richness of species and landraces (cultivated and spontaneous) and abundance, classified by geographical origins], we used mixed effect models. The nested sampling design was taken into
Agroforest Syst
consideration by setting ‘‘village’’ as a random factor. For each analysis, we calculated the ‘marginal’ and ‘conditional’ R2 [R2m and R2c , i.e., the proportion of the variance explained, respectively, by the fixed components and by the whole model, following (Nakagawa and Schielzeth 2013)]. In order to test the differences between homegardens ‘with’ and ‘without ceramics’ regarding soil properties, we used mixed effect models, with the groups ‘with’ and ‘without ceramics’ as predictors and with ‘village’ as a random factor, followed by Tukey post hoc tests. PCA was done with Canoco (Ter Braak and Smilauer 2002) and the remaining statistical analyses were done using R (R Core Team 2013).
and 29% (64) Amazonian species. Of the cultivated plants 4.5% were not identified, so their origins are unknown. Homegarden soil characteristics and heterogeneity In the field, we used soil color and the amount of ceramics as visual indicators of soil variation. We used the a priori classification ‘with’ and ‘without ceramics’ in the graphical representation of soil variation (Fig. 2; Supplementary Material Figs. S1 and S2) to highlight the relationship between soil fertility and the presence of ceramics, but we did not use this classification in the statistical analyses. The characteristics that contributed most to distinguish the soils in the homegardens along the first axis of the PCA
Results Homegarden characteristics and species diversity The size of the homegardens ranged from 200 to 15,080 m2, averaging 2303 ± 2324 m2, while their age ranged from 1 to 48 years, averaging 16 ± 10.9 years. Many of these areas had been previously inhabited by indigenous people and/or kin of current residents who cultivated in the past a variety of species, some of which are still being grown in the homegardens today. We found 378 landraces of 269 species from 76 botanical families in the 70 homegardens. The majority of the 269 species are of Amazonian origin (36.2%), followed by those introduced from the Americas (31.1%), and by non-American species (28.5%). Eleven species (4.2%) were not identified, so their origins are unknown. We sampled 9157 individual plants, 2129 of which (23% of the total, belonging to 139 species) were spontaneous plants that were favored/maintained by the homegarden owners, and the remaining 6711 individuals (73% of the total, belonging to 224 species) were cultivated. Among the spontaneous species, 55% (75 species) were native to Amazonia, 29% (41) were from the Americas and 16% (22) were non-American. We found 43 species that only occurred spontaneously, equivalent to 16% of the total species richness. As for the cultivated plants, the majority of individuals were non-American (34%; 77 species), followed by 32.5% (74) from the Americas
Fig. 2 Soil characteristics and relative abundance of nonAmerican species in homegardens. The homegardens were separated based on a principal components analysis (PCA) of the chemical and physical variables of soils collected in 70 homegardens in 7 riverside communities along the middle and lower Madeira River, Amazonas, Brazil. The numbers within parentheses represent the percentage of variation explained by each axis. Black and white circles represent homegardens with (n = 41) and without ceramic fragments (n = 29), respectively. The size of the circles represents the relative abundance of nonAmerican species. The direction of the vectors (arrows) represents the way the variables contribute to the distribution of points, and the length represents the magnitude of the contribution of each variable in the spatial configuration of the points
123
Agroforest Syst
(which explained 51% of the variance in the dataset) were Ca, pH, Mg, P, Mn, OM and Zn (positively correlated with the axis PCA1), and Al and Fe (negatively correlated with the axis PCA1). In general, the homegardens with ceramic fragments (41) had higher levels of P, Ca, Mg, Zn and Mn, lower levels of Fe and Al, and higher pH (Table 1; Fig. 2). The second axis of the PCA (accounting for 18% of the variance) was strongly correlated with physical characteristics of the soil (clay and sand content). The variation in soil texture was not related to fertility nor with the presence of ceramic fragments (Fig. 2), thus this variation is probably unrelated to human actions.
The axis PCA1 can be interpreted as a fertility axis (higher values—more fertile soils), while PCA2 can be interpreted as a texture axis (higher values—soils with higher sand content). Effects of soil on plant diversity The richness of species and landraces was positively related to soil fertility (PCA1) and sandy texture (PCA2), as well as to homegarden size (Table 2). Regarding the origin of species, the richness of nonAmerican species was positively related to soil fertility, sandy texture, homegarden age and size. This
Table 1 Average (± standard deviation), minimum and maximum values of the chemical and physical characteristics of the soils found in 70 homegardens of seven communities along the middle and lower Madeira River, Central Amazonia Variable
Unit
pH (H20) Ca
cmolc kg-1
Mg
cmolc kg-1
Al K P
cmolc kg-1 cmolc kg-1 mg kg-1
Fe
mg kg-1
Zn
mg kg-1
Mn
mg kg-1
Organic matter
g kg-1
% clay
%
% silt % sand
% %
Without ceramics (n = 29)
With ceramics (n = 41)
Avg ± SD
4.62 ± 0.39b
5.36 ± 0.65a
Min–max
3.71–5.30
3.76–6.28
Avg ± SD Min–max
0.79 ± 1.02b 0.07–4.13
6.09 ± 4.26a 0.16–14.41
Avg ± SD
0.24 ± 0.23b
0.71 ± 0.43a
Min–max
0.08–0.94
0.08–1.84
Avg ± SD
3.51 ± 2.55b
1.01 ± 1.75a
Min–max
0.25–9.65
0.00–7.50
Avg ± SD
0.18 ± 0.11a
0.16 ± 0.07a
Min–max
0.06–0.50
0.06–0.33
Avg ± SD
38.1 ± 38.1b
312.3 ± 323.6a
Min–max
2.7–185.8
14.5–1167.4
Avg ± SD
271.7 ± 120.5b
102.6 ± 95.8a
Min–max
99.9–529.3
30.9–503.3
Avg ± SD
2.59 ± 2.94b
8.09 ± 6.79a
Min–max
0.5–14.9
0.6–26.3
Avg ± SD
11.76 ± 16.63b
40.58 ± 21.82a
Min–max
1.3–62.2
3.3–78.9
Avg ± SD Min–max
31.05 ± 11.54b 14.66–73.56
43.07 ± 16.96a 19.58–93.30
Avg ± SD
28.5 ± 22.0a
18.8 ± 15.5a
Min–max
2.0–81.0
1.0–81.0
Avg ± SD
35.5 ± 22.4a
36.2 ± 14.3a
Min–max
3.9–79.7
6.5–64.4
Avg ± SD
36.0 ± 26.7a
45.0 ± 16.6a
Min–max
5.2–92.1
3.6–71.5
‘With ceramics’ corresponds to homegardens where we found ceramic fragments on the soil surface. Lowercase letters indicate significant differences between means (p \ 0.05), calculated with mixed effects models (using village as a random factor) followed by Tukey post hoc tests
123
Agroforest Syst Table 2 Effects of soil fertility and texture, homegarden age and size on the number of species, landraces and of individuals (abundance) sampled in 67–70 homegardens along the middle and lower Madeira River, Amazonas, Brazil
Individual plants are classified as ‘cultivated’ (i.e., deliberately planted by the homegarden owner) or ‘spontaneous’ (i.e., not planted or sown but maintained/favored by the owner), and species are classified according to their geographical origin [‘Amazonian’, ‘American’ (from outside Amazonia) and ‘non-American’]. Values in the table are standardized regression coefficients and asterisks indicate their significance (* p B 0.05; ** p B 0.01; *** p B 0.001). Values of ‘marginal’ (R2m ) and ‘conditional’ (R2c ) R2 are the proportion of the variance explained by the fixed components and by the whole model, respectively (Nakagawa and Schielzeth 2013)
Variable
PCA1 (fertility)
PCA2 (texture)
Age
Size
R2m
R2c
Richness Total Species (all)
0.22*
0.30**
0.14
0.51***
0.49
0.49
Landraces (all)
0.19*
0.33***
0.11
0.54***
0.50
0.51
Total species Non-American
0.34**
0.25*
0.26*
0.23*
0.36
0.47
American
0.25*
0.34***
0.04
0.36**
0.32
0.32
Amazonian
0.00
0.15
0.13
0.62***
0.46
0.49
Cultivated Non-American
0.29**
0.26*
0.28**
0.28**
0.40
0.48
American
0.09
0.32**
0.04
0.33**
0.23
0.23
0.25*
0.14
0.47***
0.35
0.40
0.10 0.33**
0.05 0.04
0.17 0.24*
0.08 0.26
0.17 0.36
0.20
0.07
0.53***
0.35
0.35
0.34***
0.09
0.43***
0.46
0.48
0.39***
- 0.10
0.41***
0.34
0.36
0.08
0.59***
0.45
0.45
Amazonian
- 0.03
Spontaneous Non-American American Amazonian
0.18 0.25* - 0.07
Abundance Total Non-American American Amazonian
0.33*** 0.20 - 0.21*
0.22*
Cultivated Non-American
0.30**
0.31**
0.09
0.44***
0.44
0.46
American
0.00
0.34**
- 0.11
0.41***
0.23
0.23
- 0.14
0.26**
0.04
0.60***
0.44
0.45
Amazonian Spontaneous Non-American
0.18
0.15
American
0.29*
0.38**
Amazonian
- 0.13
same pattern was observed for American species (except for the effect of homegarden age). The richness of native Amazonian species was positively correlated only with homegarden size (Table 2). Among cultivated individuals, non-American species richness was positively correlated with soil fertility, sandy texture, homegarden age and size; as for the American and Amazonian species richness, both were positively correlated with sandy texture and homegarden size, but not with soil fertility or age (Table 2). Among the spontaneous individuals, the richness of American species was positively related to soil fertility, sandy texture and homegarden size; the native Amazonian species richness was positively related only to homegarden size; and the non-
0.15
0.04
0.28*
0.14
0.14
- 0.01
0.25*
0.29
0.35
0.53***
0.32
0.32
0.04
American species richness was not related to any of the predictor variables (Table 2). The total number of individuals of non-American species was positively correlated with soil fertility, sandy texture, and with the size of homegardens. The same pattern occurred for the total number of individuals of species from the Americas, but the effect of soil fertility was marginally significant (p = 0.06). The total number of individuals of native Amazonian species was negatively correlated with soil fertility and positively correlated with sandy texture and homegarden size (Table 2). The number of cultivated individuals belonging to non-American species was positively correlated with soil fertility, sandy texture and homegarden size, while the number
123
Agroforest Syst
of cultivated individuals belonging to American and native Amazonian species was positively correlated with sandy texture and homegarden size (Table 2). Lastly, the number of spontaneous individuals belonging to American species was positively correlated with soil fertility, sandy texture and homegarden size. The number of spontaneous individuals belonging to nonAmerican and to native Amazonian species was positively correlated with homegarden size (Table 2).
Discussion Most smallholder farmers along the middle and lower Madeira were shown to cultivate a substantial number of plant species in their homegardens. The 269 species that we found in 70 homegardens constitute one of the highest diversities reported in the Amazon Basin, comparable to the study of Perrault-Archambault and Coomes (2008) in the Peruvian Amazon, where 309 species were found in 300 homegardens. This high agrobiodiversity likely results from the variety of strategies for plant use, management and cultivation developed by Amazonian populations, combined with the soil heterogeneity associated with ADE (Junqueira et al. 2016a), which in turn favors the co-occurrence of native and exotic agrobiodiversity. Still little explored in the literature about homegardens, the inclusion of spontaneous species in our sampling also substantially contributed to species richness (16% of the species only occurred spontaneously), highlighting the important role of spontaneous plants for homegarden agrobiodiversity. The diversity of plants and other characteristics of homegardens are influenced by complex interactions between a range of factors (Kehlenbeck and Maass 2006; Alcudia-Aguilar et al. 2017), including sociocultural and socioeconomic characteristics of the owners (Dı´az-Reviriego et al. 2016; Abizaid et al. 2016; Timsuksai and Rambo 2016; Kawa 2016), agroecological factors (such as elevation, climate, soil fertility; Kehlenbeck et al. 2007; Haile et al. 2017) and market proximity (Major et al. 2005; Kawa et al. 2011). This is probably why the variables considered in this study (soil, size and age of homegardens) explain only part of the variation in the data set (demonstrated by the R2 values in the models). Despite this complexity, we have shown that soil physical and chemical characteristics—which are important factors
123
for plant establishment and growth—are indeed important in explaining patterns in homegarden agrobiodiversity. The size and age of homegardens are factors that influence plant diversity in general, especially species richness. Large and old homegardens support more species than small and young ones (Lamont et al. 1999; Kumar and Nair 2004; Kehlenbeck and Maass 2006). We also found more species and individuals in larger homegardens (Table 2). WinklerPrins (2002) showed that older homegardens in Santare´m (Para´) are more diverse and structurally more complex, and Wezel and Ohl (2005) found a positive relation between plant diversity and age of homegardens in the Peruvian Amazon; our results, however, revealed a positive correlation between species richness and age only for non-American species (Table 2). One of the reasons why age may be less important in our analyses is that many of the homegardens included plants left by previous inhabitants, so the reported ages often did not reflect the true age since establishment of the homegarden. Furthermore, some residents, particularly those who lived in the same place for many years, occasionally cleared part of their homegardens to build a new house for newly-married members of the family, which may have led to a reduction in species diversity. We observed that homegardens on sandy soils tend to have more species and more individuals in general, but especially of non-native species (Table 2), although in our approach soil texture (PCA2) represents a smaller proportion (18.6%) of the soil variation compared with soil fertility (51.8%; Fig. 2). The overall higher richness and abundance on sandy soils may be due to their lower mechanical resistance than clayey soils (Smith et al. 1997; Vaz et al. 2011), which hinders root growth (Restom and Nepstad 2004). The increase in non-native species with sand content is likely because native species are more adapted to the clayey Oxisols that predominate in central Amazonian uplands (Quesada et al. 2011). Kehlenbeck and Maass (2006) observed that species diversity in homegardens can be influenced by soil fertility, but that this has seldom been studied in detail. Junqueira et al. (2016a) observed that farmers tend to plant more species and landraces on more fertile soils when compared to less fertile soils. Fraser et al. (2011a) also found higher species richness in homegardens on ADE, but Kawa et al. (2011) did not
Agroforest Syst
observe significant differences in species richness between homegardens on ADE and adjacent soils. ADE offer favorable conditions for the growth of numerous plant species due to their higher pH, higher concentrations of organic matter and of important nutrients for plant growth (Sombroek et al. 2002). The higher richness and abundance of exotic species— both American and non-American species—in more fertile soils (Table 2) seems at least partly due to the fact that farmers take advantage of the fertility of ADE to cultivate nutrient-demanding species that are not well adapted to the low fertility conditions found in other upland soils in the region. Junqueira et al. (2016a) reported that exotic species, such as coffee [Coffea robusta (L.) Linden] and bananas, tend to be cultivated more often in homegardens in more fertile soils, while native palms and trees, such as cupuac¸u [Theobroma grandiflorum (Willd. ex Spreng.) K. Schum.], and ac¸aı´ (Euterpe oleracea Mart. and E. precatoria Mart.), are cultivated more often in poor soils. Major et al. (2005) found greater numbers of exotic species in ADE in comparison with adjacent soils and, in the lower Madeira River, Kawa et al. (2011) found that farmers usually cultivate exotic species for commercial purposes on ADE. According to Clement et al. (2003), exotic crops seem to be better adapted to ADE than to other soil types because they suffer from the high levels of aluminum typically present in upland soils. Our findings confirm this hypothesis in quantitative terms and highlight the relevance of ADE for the cultivation of exotic crops in homegardens. Many plant species emerge spontaneously in agroecosystems and, depending on their usefulness and/or the cultural preferences of the community or of the individual (Kumar and Nair 2004), they may be favored or removed by farmers. Among the 139 species that we found growing spontaneously in homegardens, 118 of them were found in soils with ceramics (41 homegardens), which is equivalent to what Lins et al. (2015) found (119) in 46 homegardens on ADE. Spontaneous useful plants are common in anthropogenic landscapes (Miller et al. 2006), but are not well studied in homegardens. The occurrence of spontaneous species in homegardens may result from natural ecological processes (e.g., seed dispersal) or anthropic processes [discard of seeds in the homegardens of fruits consumed by the residents (Pinho et al. 2010)]. Our results show that useful species frequently
grow spontaneously and are favored in homegardens, indicating that they play an important role in the maintenance of plant diversity in these ecosystems. Many plants found in the homegardens on ADE descend from populations dating back to pre-Colombian indigenous people (Lins et al. 2015). Before the arrival of the Europeans, Amazonian people certainly used ADE sites to cultivate plants from other regions in the Americas, such as avocado (Persea americana Mill) and papaya (Carica papaya L.) (Fraser et al. 2011b). In our study, both the abundance and the richness of spontaneous plants belonging to American species were greater in homegardens on more fertile soils (Table 2). Farmers also tended to use ADE to cultivate non-American exotic species (Table 2), which were present relatively early in the colonial history of Amazonian landscapes (Homma 2003). During the first two centuries of colonization many species from other continents were introduced to Brazil, and thereafter homegardens were enriched with new species, such as Mangifera indica, bananas and others (Miller et al. 2006). Through time, these species were incorporated into local cultures and practices due to their ornamental, medicinal or food values, and many are currently important commercially.
Conclusions Among the multiple factors that might influence floristic diversity in homegardens, in this study we have shown that differences in soil properties (specifically soil fertility and soil texture) and size of the homegardens played major roles in explaining the richness and abundance of native and exotic species. The high fertility of ADE enables local residents to cultivate more species and landraces in their homegardens, especially exotic species that do not thrive in Amazonia’s low fertility upland soils. Species from the Americas (outside Amazonia) also benefit from more fertile soils, where they are often found to establish spontaneously. Our results highlight the role of ADE for homegardens as experimentation sites, since they allow the establishment and growth of many species, both cultivated and spontaneous, from various geographical regions and with different nutrient demands. It can be argued, therefore, that the use of enriched soils has always allowed people to
123
Agroforest Syst
experiment more freely, especially with exotic species, and has thus systematically contributed to the creation, enhancement and maintenance of agrobiodiversity in agroforestry systems across the world. Management interventions involving the improvement of soil properties can widen farmers’ choices to combine native and non-native species and thereby potentially increase the socio-economic and environmental benefits of agroforestry. Acknowledgements We are grateful to all the people in the ´ gua Azul, Terra Preta do Atininga, Puruzinho, communities of A Vila Espı´rito Santo, Lago do Piauı´, Sa˜o Fe´lix and Vila Gomes who participated in this research and shared their knowledge and experiences. We thank the Institute for Sustainable Agriculture and Forestry Development of Amazonas (IDAM), the National Institute of Amazonian Research and the Post-Graduate Program in Botany for logistic support. We thank the technicians at the Thematic Laboratory of Soils and Plants for performing the soil analyses. NBS received a Masters’ scholarship from the Coordination for the Improvement of Higher Education Personnel (CAPES). ABJ received PhD scholarships and funding for fieldwork from the Interdisciplinary Research and Education Fund of Wageningen University (INREF), via the Terra Preta Program, and from the Netherlands Organization for International Cooperation in Higher Education (NUFFIC). Compliance with ethical standards can be found in the Supplementary Material.
References Abizaid C, Coomes OT, Perrault-Archambault M (2016) Seed sharing in Amazonian indigenous rain forest communities: a social network analysis in three Achuar villages, Peru. Hum Ecol 44:577–594. doi:10.1007/s10745-016-9852-7 Adams C, Murrieta R, Neves W, Harris M (2009) Amazon peasant societies in a changing environment: political ecology, invisibility and modernity in the rainforest. Springer, New York Albuquerque UP, Lucena RFP, Alencar NL (2008) Me´todos e te´cnicas para coleta de dados. In: Albuquerque UP, Lucena RFP, Cunha LVFC (eds) Me´todos e te´cnicas na pesquisa etnobotaˆnica. Comunigraf/NUPEEA, Recife, pp 41–72 Alcudia-Aguilar A, Van der Wal H, Sua´rez-Sa´nchez J, Martı´nez-Zurimendi P, Castillo-Uzcanga MM (2017) Home garden agrobiodiversity in cultural landscapes in the tropical lowlands of Tabasco. Agrofor Syst, Me´xico. doi:10. 1007/s10457-017-0078-5 Caballero-SerranoV V, Onaindia M, Alday JG, Caballero D, Carrasco JC, McLaren B, Amigo J (2016) Plant diversity and ecosystem services in Amazonian homegardens of Ecuador. Agric Ecosyst Environ 225:116–125. doi:10. 1016/j.agee.2016.04.005 Clement CR (1999) 1492 and the loss of Amazonian crop genetic resources. I. The relation between domestication
123
and human population decline. Econ Bot 53(2):188–202. doi:10.1007/BF02866499 Clement CR, McCann JM, Smith NJH (2003) Agrobiodiversity in Amazonia and its relationships with dark earths. In: Lehmann J, Kern D, Glaser B, Woods W (eds) Amazonian dark earths: origin, properties and management. Kluwer Academic, Dordrecht, pp 159–178 Dı´az-Reviriego I, Gonza´lez-Segura L, Ferna´ndez-Llamazares ´ , Howard PL, Molina JL, Reyes-Garcı`a V (2016) Social A organization influences the exchange and species richness of medicinal plants in Amazonian homegardens. Ecol Soc 21:1. doi:10.5751/ES-07944-210101 EMBRAPA—Empresa Brasileira de Pesquisa Agropecua´ria (2011) Manual de me´todos de ana´lise de solo, 2a edn. Centro Nacional de Pesquisa de Solos, Rio de Janeiro Fraser J, Junqueira AB, Clement C (2011a) Homegardens on Amazonian dark earths, non-anthropogenic upland, and floodplain soils along the Brazilian middle Madeira river exhibit diverging agrobiodiversity. Econ Bot 65:1–12. doi:10.1007/s12231-010-9143-y Fraser J, Junqueira AB, Kawa N, Moraes C, Clement C (2011b) Crop diversity on anthropogenic dark earths in central Amazonia. Hum Ecol 39:395–406. doi:10.1007/s10745011-9405-z Fraser J, Teixeira W, Falca˜o N, Woods W, Lehmann J, Junqueira AB (2011c) Anthropogenic soils in the Central Amazon: from categories to a continuum. Area 43(3):264–273. doi:10.1111/j.1475-4762.2011.00999.x Glaser B, Birk JJ (2012) State of the scientific knowledge on properties and genesis of anthropogenic dark earths in Central Amazonia (terra preta de ´Indio). Geochim Cosmochim Acta 82:39–51. doi:10.1016/j.gca.2010.11. 029 Haile G, Lemenih M, Senbeta F, Itanna F (2017) Plant diversity and determinant factors across smallholder agricultural management units in Central Ethiopia. Agrofor Syst 91:677–695. doi:10.1007/s10457-016-0038-5 Hiraoka M, Yamamoto S, Matsumoto E, Nakamura S, Falesi IC, Baena ARC (2003) Contemporary use and management of Amazonian dark earths. In: Lehmann J, Kern D, Glaser B, Woods W (eds) Amazonian dark earths: origin, properties and management. Kluwer Academic, Dordrecht, pp 387–406 Homma AKO (2003) Histo´ria da agricultura na Amazoˆnia: da era pre´-colombiana ao Terceiro Mileˆnio. Embrapa Informac¸a˜o Tecnolo´gica, Brası´lia IBGE (2010) Mapa de solos do estado do Amazonas. Instituto Brasileiro de Geografia e Estatı´stica. http://mapas.ibge. gov.br/tematicos/solos. Accessed 17 Sept 2015 Jose S (2011) Managing native and non-native plant species in agroforestry. Agrofor Syst 83:101–105. doi:10.1007/ s10457-011-9440-1 Junqueira AB, Souza NB, Stomph TJ, Almekinders CJM, Clement CR, Struik PC (2016a) Soil fertility gradients shape the agrobiodiversity of Amazonian homegardens. Agric Ecosyst Environ 221:270–281. doi:10.1016/j.agee.2016. 01.002 Junqueira AB, Stomph TJ, Clement CR, Struik PC (2016b) Variation in soil fertility influences cycle dynamics and crop diversity in shifting cultivation systems. Agric
Agroforest Syst Ecosyst Environ 215:122–132. doi:10.1016/j.agee.2015. 09.015 Kawa NC (2016) How religion, race, and the weedy agency of plants shape Amazonian home gardens. Cult Agric Food Environ 38:84–93. doi:10.1111/cuag.12073 Kawa NC, Rodrigues D, Clement CR (2011) Useful species richness, proportion of exotic species, and market orientation on Amazonian dark earths and Oxisols. Econ Bot 65(2):169–177. doi:10.1007/s12231-011-9158-z Kehlenbeck K, Maass BL (2006) Are tropical homegardens sustainable? Some evidence from Central Sulawesi, Indonesia. In: Kumar BM, Nair PKR (eds) Tropical homegardens: a time-tested example of sustainable agroforestry. Springer, Dordrecht, pp 339–354 Kehlenbeck K, Arifin HS, Maass BL (2007) Plant diversity in homegardens in a socio-economic and agro-ecological context. In: Tscharntke T, Leuschner C, Zeller M, Guhardja E, Bidin A (eds) Stability of tropical rainforest margins: linking ecological, economic and social constraints. Springer, Dordrecht, pp 297–319 Kim DG, Kirschbaum MUF, Beedy TL (2016) Carbon sequestration and net emissions of CH4 and N2O under agroforestry: synthesizing available data and suggestions for future studies. Agric Ecosyst Environ 226:65–78. doi:10. 1016/j.agee.2016.04.011 Kumar BM, Nair PKR (2004) The enigma of tropical homegardens. Agrofor Syst 61–62:135–152. doi:10.1023/B: AGFO-0000028995-13227-ca Lamont SR, Eshbaugh WA, Greenberg AM (1999) Composition, diversity, and use of homegardens among three Amazonian villages. Econ Bot 53(3):312–326. doi:10. 1007/BF02866644 Leo´n J (2000) Bota´nica de los cultivos tropicales. Editorial Agroame´rica, Instituto Interamericano de Cooperacio´n para la Agricultura, San Jose´ Lins J, Lima HP, Baccaro FB, Kinupp VF, Shepard GH, Clement CR (2015) Pre-Columbian floristic legacies in modern homegardens of central Amazonia. PLoS ONE 10(6):e0127067. doi:10.1371/journal.pone-0127067 Lista de Espe´cies da Flora do Brasil (2015) Jardim Botaˆnico do Rio de Janeiro http://floradobrasil.jbrj.gov.br/. Accessed 28 July 2015 Major J, Clement CR, DiTommaso A (2005) Influence of market orientation on food plant diversity of farms located on Amazonian dark earth in the region of Manaus, Amazonas, Brazil. Econ Bot 59(1):77–86. doi:10.1663/0013-0001 (2005)059[0077:IOMOOF]2.0.CO;2 Mattsson E, Ostwald M, Nissanka SP (2017) What is good about Sri Lankan homegardens with regards to food security? A synthesis of the current scientific knowledge of a multifunctional land-use system. Agrofor Syst. doi:10.1007/ s10457-017-0093-6 Miller RP, Penn JW Jr, van Leeuwen J (2006) Amazonian homegardens: their ethnohistory and potential contribution to agroforestry development. In: Kumar BM, Nair PKR (eds) Tropical homegardens: a time-tested example of sustainable agroforestry. Springer, Dordrecht, pp 43–60 Nakagawa S, Schielzeth H (2013) A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol Evol 4(2):133–142. doi:10.1111/j. 2041-210x.2012.00261.x
Perrault-Archambault M, Coomes OT (2008) Distribution of agrobiodiversity in home gardens along the Corrientes River, Peruvian Amazon. Econ Bot 62(2):109–126. doi:10. 1007/s12231-008-9010-2 Pinho RC, Alfaia SS, Miller RP, Uguen K, Magalha˜es LD, Ayres M, Freitas V, Trancoso R (2010) Islands of fertility: soil improvement under indigenous homegardens in the savannas of Roraima, Brazil. Agrofor Syst 81:235–247. doi:10.1007/s10457-010-9336-5 Quesada CA, Lloyd J, Anderson LO, Fyllas NM, Schwarz M, Czimczik CI (2011) Soils of Amazonia with particular reference to the RAINFOR sites. Biogeosciences 8:1415–1440. doi:10.5194/bg-8-1415-2011 R Core Team (2013) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna Rapp Py-Daniel L (2007) Caracterizac¸a˜o da a´rea amostrada. In: Rapp Py-Daniel L, Deus CP, Henriques AI, Pimpa˜o DM, Ribeiro OM (eds) Biodiversidade do me´dio Madeira: bases cientı´ficas para propostas de conservac¸a˜o. Instituto Nacional de Pesquisas da Amazoˆnia, Manaus, pp 35–42 Restom TG, Nepstad DC (2004) Seedling growth dynamics of a deeply rooting liana in a secondary forest in eastern Amazonia. Fort Ecol Manag 190:109–118. doi:10.1016/j. foreco.2003.10.010 Schmidt MJ et al (2014) Dark earths and the human built landscape in Amazonia: a widespread pattern of anthrosol formation. J Archaeol Sci 42:152–165. doi:10.1016/j.jas2013-11-002 Smith CW, Johnston MA, Lorentz S (1997) The effect of soil compaction and soil physical properties on the mechanical resistance of South African forestry soils. Geoderma 78:93–111. doi:10.1016/s0016-7061(97)00029-3 Sombroek W, Kern D, Rodrigues T, Cravo MS, Jarbas TC, Woods W, Glaser B (2002) Terra preta and terra mulata: pre-Columbian Amazon kitchen middens and agricultural fields, their sustainability and their replication. In: Proceedings 17th World Congress of Soil Science, Bangkok Ter Braak CJF, Smilauer P (2002) CANOCO reference manual and CanoDraw for Windows user’s guide: software for canonical community ordination (version 4.5). Microcomputer Power, Ithaca The Plant List (2013) Version 1. http://www.theplantlist.org/. Accessed 28 July 2015 Timsuksai P, Rambo AT (2016) The influence of culture on agroecosystem structure: a comparison of the spatial patterns of homegardens of different ethnic groups in Thailand and Vietnam. PLoS ONE 11:e0146118. doi:10.1371/ journal.pone.0146118 Tropicos.org (2013) Missouri Botanical Garden. http://www. tropicos.org/. Accessed 28 July 2015 Vaz CMP, Manieri JM, de Maria IC, Tuller M (2011) Modeling and correction of soil penetration resistance for varying soil water content. Geoderma 166:92–101. doi:10.1016/j. geoderma.2011.07.016 Wezel A, Ohl J (2005) Does remoteness from urban centres influence plant diversity in homegardens and swidden fields? A case study from the Matsiguenka in the Amazonian rainforest of Peru. Agrofor Syst 65(3):241–251. doi:10.1007/s10457-005-3649-9
123
Agroforest Syst Winklerprins AMGA (2002) Recent seasonal floodplain-upland migration along the Lower Amazon River. Geogr Rev 92(3):415–431. doi:10.1111/j1931-0846-2002-tb00151-x
123
Witkoski AC (2010) Terras, florestas e a´guas de trabalho: os camponeses amazoˆnicos e as formas de uso de seus recursos naturais. Annablume, Sa˜o Paulo