5 6
Authors 7
Heloisa Dantas Brum1*, Alexandre F. Souza2
8
1 - Programa de Pós-Graduação em Ecologia, CB, Universidade Federal do Rio Grande do 9
Norte, Campus Universitário, Lagoa Nova, Natal 59072-970, RN, Brazil. E-mail: 10
12
2 - Departamento de Ecologia, CB, Universidade Federal do Rio Grande do Norte, Campus 13
Universitário, Lagoa Nova, Natal 59072-970, RN, Brazil. E-mail: 14 [email protected] 15 16 *Corresponding author 17
Programa de Pós-Graduação em Ecologia, CB, Universidade Federal do Rio Grande do 18
Norte, Campus Universitário, Lagoa Nova, Natal 59072-970, RN, Brazil. 19
Telephone number: +55 (84) 99915-4336 20
102
Abstract
22
Euterpe precatoria (açaí) is the most abundant plant species in the Amazon basin and one 23
of the main non-timber forest products of the continent. A thorough understanding of
24
the ecology of this species is needed to support sustainable management initiatives.
25
Resource availability, disturbance regime, and human management are some of the main
26
factors influencing population structure. Here we described life stages of E. precatoria,
27
evaluated their allometric relationships, and assessed the effects of habitat type
28
(floodplain and unflooded upland) and proximity of human settlements on population
29
size distribution in Central Amazon near the Purus River. The height:diameter relationship
30
increased from Seedlings to Juvenile 2 but decreased from Juvenile 2 to Reproductive 2,
31
indicating changing height investment for any given diameter along these life stages.
32
There was a marked habitat dependency in both the density and population size
33
distribution, with populations in upland forests dominated by juveniles while populations
34
in the floodplains were dominated by reproductive palms. Nearness of human
35
settlements was not related to population structure parameters. The patterns we
36
uncovered have implications for our interpretation of widespread Amazon forest species
37
that inhabit habitats with contrasting disturbance regimes and resource levels like
38
flooded várzea and upland terra firme.
39
40
Key-words: Várzea, Terra firme, Population structure, Forest Management, Skewness, 41 Allometry. 42 43 44 45
103
Introduction
46
Euterpe precatoria Mart. is an iconic palm tree popularly known as açaí or assaí in South
47
America. It is the most abundant species in the Amazon basin, being one of the few 48
hyperdominant species in four of the broad five forest types in the region (ter Steege et 49
al. 2013). It is also one of the main non-timber forest products of the continent (Stoian 50
2004). The species is traditionally used for the extraction of fruits, heart of palm, and 51
timber by indigenous peoples (Anderson 1977; Albert and Tourneau 2007). It produces 52
large quantities of dark single-seeded fruits that, as in other Euterpe species, is dispersed 53
by many birds and mammals (Leite et al. 2012). The fruit is commonly collected and a 54
regional market developed around its use to produce a creamy beverage known as ‘açaí 55
wine’ that is very nutritious and bears superior antioxidant and anti-inflammatory 56
properties (Kang et al. 2012) that have been shown to bring health benefits like memory 57
protection when used as a dietary supplementation (Carey et al. 2017). Euterpe 58
precatoria fruit consumption may be a sustainable alternative to this species’ use in
59
countries like Bolivia, where the traditional consumption of palm hearth implies that 60
palms are killed for resource extraction (Rocha 2004; Velarde and Moraes 2008). Most 61
recently prices for E. precatoria products have increased dramatically due to the global 62
commodization of palm hearts and raw, processed, or lyophilized fruit pulp as nutritional 63
supplement that are increasingly consumed in Brazil as well as exported to countries 64
outside South America, mainly Europe, Canada and USA (Bussmann and Zambrana 2012). 65
Market sales for açaí berry (E. oleracea and E. precatoria combined) amount USD 126.3 66
millon/yr in Brazil alone (Martinot et al. 2017). With increasing trade volumes, E. 67
104
precatoria and many other exploited native species cannot meet demand in a sustainable
68
manner (Mostacedo and Fredericksen 1999; Stoian 2004; Vallejo et al. 2016). A thorough 69
understanding of the biology and ecology of this species is thus needed, in order to 70
support sustainable management initiatives (Velarde and Moraes 2008). 71
One of the most easily measurable population parameter in the field is population 72
structure, which integrates in itself a wealth of demographic information (Avalos et al. 73
2013; Peltzer et al. 2014). Yet, it may be a puzzling tool for the assessment of natural 74
populations, because of two reasons. The first is that the interpretation of population 75
structure may be easier if biologically relevant classes are clearly identified, e.g. through 76
the identification of life stages instead of arbitrary size classes (Gatsuk et al. 1980; Souza 77
et al. 2000, 2003; Caswell 2001). The second is that the patterns of relative abundances 78
of juveniles and adults they portray are no reliable proxy of future population growth and, 79
therefore, of population persistence in any given habitat or set of environmental or 80
human-imposed conditions (Johnson et al. 1994; Condit et al. 1998; Souza 2007; Virillo 81
et al. 2011; Bin et al. 2012). Attempts to infer population persistence from static 82
population structure and the relative abundance of juveniles relative to adults are, 83
therefore, flawed. In fact, population size structures result from the realized growth, 84
mortality, and fecundity rates across different size classes (Caswell 2001; Wright et al. 85
2003). A more fruitful use of population structure is the comparison of population size 86
distributions across habitats with contrasting disturbance histories or between co- 87
occurring species in the same habitat, which allows the detection of functional groups of 88
species with similar life histories (Swaine et al. 1990; Poorter et al. 1996; Wright et al. 89
105
2003; Souza 2007; Souza et al. 2008, 2010). Following the fast-growing resource acquiring 90
vs slow growth stress tolerant life history gradient repeatedly found worldwide (Grime 91
and Pierce 2012; Reich 2014; Díaz et al. 2016), left-skewed size distributions dominated 92
by large individuals characterize chronic recruitment failure of gap-dependent species 93
with large fecundities, high seedling mortality, and high sapling growth, while right- 94
skewed size distributions dominated by small individuals characterize shade-tolerant 95
species with the opposite traits (Lorimer and Krug 1983; Swaine et al. 1990; Poorter et al. 96
1996; Wright et al. 2003; Souza 2007; Souza et al. 2008; Vlam et al. 2014). 97
Among the factors that influence population structure are the effects of resource 98
availability, the disturbance regime, and human management (Mostacedo and 99
Fredericksen 1999; Souza and Martins 2004; Souza 2007; Avalos et al. 2013; Peltzer et al. 100
2014). Habitat differentiation between várzea floodplains and upland terra firme are 101
among the main drivers of both light resource and disturbance levels in the Amazon 102
basin. The Amazon basin harbors a great variety of floodplains that cover nearly 500,000 103
Km2 (Junk et al. 2011) that includes a great variety of habitats such as seasonally
104
inundated forests and swamps, with plant communities adapted to differing flooding 105
regimes (Junk et al. 2012). These hydro-edaphic conditions exert strong control on tree 106
species distributions, connecting floodplain floras even when rivers drain distinct climatic 107
regions (Wittmann and Junk 2003). Environmental variation linked to flooding regime 108
affects demographic parameters of neighboring populations, and may be detectable 109
even over short geographic distances (Otárola and Avalos 2014). The disturbance caused 110
by seasonal flooding involves oxygen deprivation, sedimentation, and mechanic damage 111
106
that frequently kills younger plants (Parolin 2008). It has triggered the evolution of 112
metabolic pathways promoting either endurance or escape strategies (Parolin 2008; 113
Voesenek and Bailey-serres 2015). Contrary to flooded forests, which present lower and 114
more irregular canopies allowing more light penetration (Souza and Martins 2005; 115
Sawada et al. 2015), in uplands the main limitation to plant growth and establishment is 116
the deep shade cast by tall canopies, which severely reduce understory light levels 117
(Svenning 2001, 2002; Myster 2016). 118
The shady understory environments have induced the evolution of ecological strategies 119
that fall along the low growth/size shade stress tolerant – fast growth/height light 120
acquisition trade-off (Grime 1977; Forgiarini et al. 2013; Reich 2014). In palms, the lack 121
of secondary meristems precludes exploitation of light through lateral growth, restricting 122
possible plant responses to shading to increases in allometric height: diameter 123
increments (Avalos and Otárola 2010). The height growth of Euterpe precatoria is 124
supported by stilt roots whose size scales with plant height rather than with topographic 125
variation (Avalos and Otárola 2010), but the effects of habitat variation between upland 126
and floodplains on palm shape and allometry are poorly known. Even less studied are the 127
effects of human management on E. precatoriapopulation ecology. These effects area 128
potentially sizable because the species is considered domesticated or semi-domesticated 129
by indigenous people since pre-Columbian times (Clement et al. 2015; Levis et al. 2017). 130
The intensive management of preferred species by indigenous and current locals has 131
taken place mainly along the floodplain margins of large rivers (McMichael et al. 2012), 132
107
and is thus expected to influence the abundance of the species across different habitat 133
types (Ticktin et al. 2012; Baldauf and Maës dos Santos 2013). 134
Here we describe the macromorphological life stages of E. precatoria, evaluate their 135
allometric relationships, and assess the effects of habitat type (floodplain and unflooded 136
upland) and proximity of human settlements on population size distribution in Central 137
Amazon. Specifically, we tested the following hypotheses: 1) Due to the higher and more 138
closed canopies in upland than in floodplain forests (Myster 2016), established palms 139
should present taller and slenderer trunks in upland forests. Therefore, there should be 140
a significant interaction between height vs. diameter allometric relationship and habitat; 141
2) Because E. precatoria forms seedling banks in upland shaded understory (Peña-Claros 142
and Zuidema 2000; Rocha 2004; Isaza et al. 2017), we infer that the species present a 143
moderate degree of shade tolerance (Condit et al. 1998; Mostacedo and Fredericksen 144
1999) that allows it to wait for canopy opening increases (Avalos et al. 2013; Otárola and 145
Avalos 2014). We thus expect right-skewed size distributions in upland habitats where 146
adult recruitment and productivity are light-limited. Considering that in the floodplains 147
flooding disturbs seedling recruitment by killing seedlings (Parolin 2008), we also expect 148
reduced size distribution skewness or even symmetrical size distributions. This would 149
reflect the prevalence of larger plants in brighter and more productive floodplains, which 150
would be more favorable to adult growth and survivorship once the juvenile recruitment 151
bottleneck is surpassed (Avalos et al. 2013; Otárola and Avalos 2014). 3) Given the 152
rationale above, we expect to find greater seedling and juvenile densities in upland but 153
greater adult densities in floodplains (Rocha 2004; Velarde and Moraes 2008; Otárola and 154
108
Avalos 2014); 4) Due to negative effects of fruit harvesting on population recruitment 155
(Mostacedo and Fredericksen 1999), we expect that proximity to human settlements 156
depress the number of juveniles. Hence, we expect to find a positive relationship 157
between the distance from the nearest human settlement and the skewness of 158
population size distribution (where more positive skewness reflects increased juvenile 159
abundance). The outcomes of these tests can help stakeholders and local users to deal 160
with management decisions. 161
162
Methods
163Study area and species
164
Euterpe precatoria is a single-stemmed palm (but see Avalos and Schneider 2011)that
165
may reach up to 20 m height occurring from Central America to Bolivia (Stoian 2004). 166
Individuals are obligate outcrossing monoecious and present a long flowering time that 167
provide easily accessed resources to thousands of insect flower visitors (Kuchmeister et 168
al. 1997). It’s large seeds (ca. 11 cm diameter Aguiar and Mendonça 2003) germinate 169
immediately after harvesting (Costa et al. 2018). The ability of E. precatoria to germinate 170
in darkness and its preference for relatively mild germination temperatures (20ºC) may 171
favor its establishment in seasonally flooded habitats, while its tolerance to moderate 172
desiccation (Costa et al. 2018) allows germination in upland forests. As show for other 173
Euterpe species, germination is likely facilitated by frugivorous dispersal (Leite et al. 174
2012). Seedlings tolerate shade and do not respond promptly to increased irradiance 175
109
(Coelho et al. 2015). The species attains higher densities, sizes, and fruit productivity in 176
floodplains than in unflooded uplands, where it shows etiolation signs (Velarde and 177
Moraes 2008). 178
The study was carried out in the Piagaçu-Purus Sustainable Development Reserve (SDR- 179
PP), Amazonas State, located in the Purus-Madeira and Purus-Juruá interfluve (Figure 1). 180
The reserve covers 834,245 ha and allows human habitation and the sustainable use and 181
commercialization of natural resources. It harbors 65 riverine communities and more 182
than 5000 people, who live mainly from family agriculture, fishing, hunting, and the 183
extraction of non-timber forest products like açaí. The SDR-PP is located in the 184
municipalities of Coari and Codajás, which together account for 96% of the 50,000 tons 185
of açaí produced by the Amazonas state annually (IBGE 2018). Açaí is collected in the 186
study region non-destructively by climbing. Although the Purus river presents up to 10 m 187
annual level variation (ANA 2018), the várzea we studied is located at a high várzea area, 188
with seasonal flooding of up to 3 m height and during less than 50 days.year-1 (Wittmann
189
et al. 2010). 190
110 192
Figure 1. Lower Purus river area in Central Amazon, showing the location of riverine 193
communities and the 10 blocks where individuals of E. precatoria (açaí) were marked and 194 measured. 195 196 Data collection 197
Sampling was directed at sites selected after a quick participatory mapping with local 198
residents, who indicated areas used for açaí fruit harvesting with differing intensities. 199
Data were collected in 200 20 x 20m permanent plots (total 8 ha) distributed in 10 blocks, 200
five in the floodplain and five in the upland (Figure 1). In each block, plots were arranged 201
in four 100-m long transects, each one subdivided into five plots, and stratified into two 202
topographic positions. In each block, two transects were located in low-lying terrain, thus 203
subjected to more frequent and long-lasting flooding, and two were located in elevated 204
111
terrain, thus subjected to less frequent and short-lasting flooding (approximately with 205
one month of difference). Because the number of newly germinated seedlings was very 206
large in the floodplain plots, we subsampled this life stage only through four 1 x 1 m 207
subplots located at the corners of each 20 x 20m plot. From February to March 2016, all 208
individuals of E. precatoria were tagged and measured for diameter at soil level, total 209
height, and number of leaves in all plots. Individuals in early development that lacked the 210
aerial stem had diameter and height measured at the base, which correspond to the 211
group of leaf sheaths. For seedlings, tags were tied to a wooden (on uplands) or metal (in 212
the floodplains) stick, in order to avoid damaging the plants. For plants with stem base 213
enlargement or with aerial roots, diameter measurement was made above these 214
structures. Presence and height of reproductive structures were also measured. We used 215
the distance to the nearest human settlement as a proxy to the effects of human activities 216
(Duvall 2007; Vandam et al. 2013; Sumarga 2017). Distance to the nearest human 217
settlement was measured as sum of linear distances actually traveled by land and water 218
by the inhabitants of each settlement to reach the location of each transect using a 219
Garmin GPSmap 62s and the QGIS software (QGIS Development Team 2009). 220
Statistical analysis
221
Euterpe precatoria life stages were established based on macromorphological traits
222
(Gatsuk et al. 1980). Further subdivisions within life stages were established based on 223
breakpoints in the relationship between stem diameter and height. This was assessed 224
through a loess regression between diameter and height after iterative adjustment of 225
alpha (α) and lambda (λ) parameters (Jacoby 2000). All statistical analyses were 226
112
performed in the open software R (R Core Team 2017). We performed an analysis of 227
covariance to assess the effect of life stages on the allometric relationship between 228
diameter and height. We used the natural logarithm of height to quantify palm size, and 229
the coefficient of skewness (g1, Zar 1996) to summarize the skewness of the height 230
distributions and, thus, population structure (Wright et al. 2003) at the transect (i.e., 231
blocks of 5 20 x 20 m plots) scale (n = 40). When g1 > 0, it indicated right-skewed height 232
distributions with few tall and many short palms, g1 = 0 indicated symmetrical height 233
distributions, and g1 < 0 indicated left-skewed height distributions with many tall and few 234
short palms (Zar 1996). Seedlings were excluded from population structure analyses 235
because their short-lived duration, mainly in the floodplains, could distort the results 236
(personal observation). 237
We used Generalized Linear Models (GLM) to evaluate habitat (floodplain x upland) and 238
human (distance to the nearest human settlement) on the skewness of population height 239
distribution. Following Zuur et al. (2009) and Plant (2012), different models were fitted in 240
order to decide which model structure best described the error structure in the data. In 241
order to decide whether a random term was necessary, we used the second-order Akaike 242
Information Criterion (AICc, used for small sample sizes) and ANOVA to compare models. 243
Models whose ΔAIC < 2 were regarded as equally plausible model (Burnham and 244
Anderson 2002). We fitted a GLM without a random term (i.e., containing fixed terms 245
only), a Generalized Linear Mixed Model (GLMM) using blocks as a random intercept 246
term, and a GLMM using both blocks and habitat types as random terms (Zuur et al. 247
2009). All models were fit by maximizing the restricted log-likelihood using the functions 248
113
gls or lme of the nlme package (Pinheiro et al. 2017). The random effect in the GLMM 249
accounted for the possible lack of independence of the skewness values estimated for 250
the blocks. The 20 x20 m plot-level effects of habitat type and distance to human 251
settlement on palm density were assessed through a GLMM using Poisson error and 252
blocks and transects as nested random terms (Zuur et al. 2009). 253
254
Results
255We marked and mapped 3613 individuals of açaí in different life stages in both the 256
floodplains and upland habitats. Observation of external macromorphological structures 257
allowed the distinction of two pre-reproductive stages: seedlings and juveniles. Seedlings 258
were defined as individuals without the presence of an aerial stem. Seedling leaves 259
present fan-shaped leaves (Figure 2). Juveniles were all individuals with an aerial stem 260
and diameter at soil level below the minimum diameter found for reproductive palms, 261
which was 10.0 cm (Figure 2). Juveniles presented a marked discontinuity in their height 262
vs. diameter relationship at ca. 1.5 m height (Figure 3A) and were thus subdivided into 263
Juvenile 1 (height < 1.5 m) and Juvenile 2 (1.5 ≤ height < 10 m). The diameter x height 264
relationship was nonlinear (loess regression: α = 0.35, λ = 2, Figure 3) with two marked 265
slope changes. The first change was located at the transition between Juveniles 1 and 2, 266
and the second among reproductive palms at ca. 23.0 cm diameter, a threshold from 267
which higher diameter values were not accompanied by higher height values (Figure 3). 268
Reproductive palms were thus divided into Reproductive 1 (height ≥10 m and diameter 269
114
< 23 cm) and Reproductive 2 (height ≥ 10 m and diameter ≥ 23 cm). The dependency of 270
the relationship between diameter and height on life stages was further confirmed 271
through an ANCOVA (R2 = 0.99; F = 2.64× 104; df = 3598; P = 2.2 × 10-16) (Figure 3B). On 272
the one hand, the declivity of the linear fit increased from Seedlings to Juvenile 1, and 273
from these to Juvenile 2, indicating increasing heights for any given diameter in these life 274
stages. On the other hand, the same declivity decreased from Juvenile 2 to Reproductive 275
1, and from these to Reproductive 2, indicating decreasing height investment for any 276
given diameter in these life stages. A second ANCOVA tested for the effects of habitat 277
type (upland x flooded forest) on the height:diameter relationship, this time excluded 278
both seedlings and Reproductive 2 palms because these two life stages did not occur in 279
upland forests. Habitat type, however, did not alter the height:diameter relationship 280
significantly either as an isolated factor (F = 2.66, df = 1, P = 0.103) or in interaction with 281
life stages (F = 1.479, df = 1, P = 0.2241). 282
115 284
Figure 2. Life stages of Euterpe precatoria (Arecaceae). A. Fruit bunch; B. Detail of fruits; 285
C. Detail of the seed with initial development of seedling; D. Seedling; E. Juvenile 1; F. 286
Reproductive. See text to description of stages characteristics. Drawing by Eliziane Garcia. 287
288
289
Figure 3. A) Overall relationship between height and diameter of 3613 açaí palms 290
(Euterpe precatoria) measured in upland and floodplain forests, Central Amazon. The 291
A B
116
fitted line is a Loess regression. B) Height × diameter relationship on a log scale with fitted 292
linear model adjusted by life stage through ANCOVA. 293
294
Seedlings were by far the denser life stage, with average 470.75 ± 486.21 plants ha-1 in
295
the unflooded areas and 4025.0 ± 5086.31 plants ha-1 in the floodplains. Seedlings apart,
296
overall palm density was greater in unflooded forests (66.88 ± 129.38 plants ha-1) than in
297
flooded forests (31.13 ± 79.65 plants ha-1). Palm density changed significantly with
298
habitat as well as with life stage. The best GLMM model included the interaction between 299
habitat and life stage, as well as blocks and transects as random terms (Table 1). This 300
indicates that population density varied between transects within blocks as well as 301
between blocks in a random yet sizable way. The habitat vs.× life stage interaction term 302
indicates that the density of different stages varied between flooded and unflooded 303
forests in different ways. Unflooded forests showed to be dominated by juveniles, with 304
Juvenile 1 and Juvenile 2 palms having the highest densities. On the contrary, flooded 305
forests were dominated by Reproductive 1, which was the densest life stage (Fig. 4).