|

Received: 29 June 2018    Accepted: 5 October 2018

DOI: 10.1111/btp.12613

PAPER

Cattle affect regeneration of the palm species Attalea princeps

in a Bolivian forest–savanna mosaic

Iris Hordijk1,2  | Fabian Meijer1 | Esther Nissen1 | Tjalle Boorsma3 | 

Lourens Poorter1

1
Forest Ecology and Forest Management
Group, Wageningen University and Abstract
Research, Wageningen, The Netherlands Attalea princeps is an important palm species that shapes the forest–savanna mosaic
2
Institut für Integrative Biologie, ETH Zürich
in Beni, Bolivia, as it dominates the two principal forest landscape elements (forest
(Swiss Federal Institute of Technology),
Zürich, Switzerland islands and gallery forest), and provides a vital microhabitat, food, and nesting
3
Asociación Armonía, Santa Cruz de la source for numerous plant and animal species. The forest–savanna mosaic is used
Sierra, Bolivia
for extensive grazing, and the palm population is declining on the forest islands due
Correspondence to a low regeneration rate, which threatens the maintenance of this landscape. We
Iris Hordijk, Forest Ecology and Forest
therefore examined the (a)biotic factors that influence the population structure of
Management Group, Wageningen University
and Research, Wageningen, Attalea in the centers and edges of forest islands and gallery forests. Ninety-­one
The Netherlands.
0.1-­ha plots were established, and 500 palm adults and 3,700 juveniles were meas-
Email: [email protected]
ured for their size, health condition, and fire damage. For each plot, habitat charac-
Funding information
teristics, such as landscape position, grazing pressure, and soil conditions, were
Christine Buisman; Kronendak foundation;
Bottelier Stipendium; Alberta Mennega measured. Attalea population density was significantly lower on the forest islands
foundation
than in the gallery forests, especially in the juvenile life stage. A structural equation
model showed that juvenile density is positively related to the health condition of
juveniles and amount of fruits present, where the amount of fruits is positively af-
fected by the condition of adults. Juvenile density is negatively influenced by graz-
ing, affecting the health condition of the juvenile, as well as organic matter and
phosphate availability in the soil. Therefore, it is recommended to decrease the
grazing pressure by decreasing livestock densities, fencing off vulnerable forest
­islands, or by rotating cattle.
Abstract in Spanish is available with online material.

KEYWORDS
Beni savanna, Bolivia, landscape mosaic, Llanos de Moxos, Motacú palm

1 |  I NTRO D U C TI O N are influenced by abiotic factors (Loreau, 2010; Naeem, Duffy,
& Zavaleta, 2012). Landscapes and species are also influenced by
A landscape mosaic promotes a high beta biodiversity as different human activity and with the high human population pressure in
ecological communities occur in different patches (Watt, 1947). the Anthropocene these alterations become more visible than ever
Species within and between patches interact with each other and (Brook, Bradshaw, Koh, & Sodhi, 2006; Hansen et al., 2005).

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction
in any medium, provided the original work is properly cited and is not used for commercial purposes.
© 2019 The Authors. Biotropica published by Wiley Periodicals, Inc. on behalf of Association for Tropical Biology and Conservation

Biotropica. 2019;1–11. wileyonlinelibrary.com/journal/btp  |  1

 |
2       HORDIJK et al.

The Beni savanna in Bolivia is a forest–savanna mosaic and has Beni savanna, as it provides and shapes a specific microhabitat and
been identified as a center of plant diversity and endemism (Mayle, a food source for numerous plant, mammal, and bird species (Alvez-­
Langstroth, Fisher, & Meir, 2007). This dynamic ecoregion is a com- Valles et al., 2018; Corrêa, Fischer, & dos Santos, 2012; Rios & Pacheco,
plex mosaic of rivers and lakes, seasonally flooded treeless grass- 2006). For example, Attalea has been identified as the most import-
lands, savanna, herbaceous wetlands, riverine gallery forest, and ant factor for the survival of the critically endangered endemic Blue-­
forest islands (Ibisch, Beck, Gerkmann, & Carretero, 2003). Riverine throated Macaw (Ara glaucogularis) (BirdLife International 2015; Hesse
gallery forest borders rivers on raised fluvial deposits (Lombardo, & Duffield, 2000). In the Beni savanna, three factors are likely to have
2014) and forest islands are small (<50 m width), round-­
shaped strong impacts on the life cycle and regeneration of Attalea: landscape
patches of trees on elevated areas in a treeless savanna landscape, mosaic, disturbance, and environmental factors (see the conceptual
formed by anthropogenic influences or termite mounds (Denevan, diagram in Figure 1).
1966; Erickson, 1995; Lombardo et al., 2013). These patches are
modified by environmental drivers such as human-­
induced and
1.1 | Landscape mosaic
natural fires in the dry season, and inundation in the wet season
(Berkunsky et al., 2016). Island theory predicts that small and isolated forest islands have
The Beni savanna ecoregion is a threatened Amazonian savanna smaller and less viable species populations because of higher ex-
due to a rapid decline in biodiversity caused by large-­scale cattle tinction risks and lower immigration rates (MacArthur & Wilson,
ranching, habitat degradation, and conversion of natural grass- 2015). At the edge of a forest patch there is a different microcli-
land (de Carvalho & Mustin, 2017; Mayle et al., 2007). For the past mate and a higher disturbance frequency compared to the forest
12,000 years, the Beni savanna has been influenced by human ac- center; light availability and wind intensity are higher and trees
tivity (Erickson, 2000; Haase & Beck, 1989; Lombardo et al., 2013; are more exposed to inundation, fire, and cattle grazing (Camargo
Navarro & Maldonado, 2002) as the savanna was inhabited by a & Kapos, 1995; Cochrane & Laurance, 2002; Williams-­
L inera,
dense population of pre-­Columbian Paleo-­Indians who constructed 1990).
earth mounds for agriculture and habitation, and dug canals and
fishing ponds (Denevan, 1966; Erickson, 2000). Nowadays, the main
1.2 | Disturbance
human impacts are protracted and uncontrolled cattle grazing, burn-
ing of savanna by farmers to ensure fresh grass sprout for cattle, nat- Palm regeneration is affected by cattle ranching, and naturally oc-
ural grassland conversion, and forest destruction for cattle ranching curring and human-­
induced fires (Yamashita & de Barros, 1997).
infrastructure (Berkunsky et al., 2016; Killeen, 1991; Killeen et al., Cattle ranching leads to browsing and trampling of juveniles, and
2003; Mayle et al., 2007). increased soil compaction reduces soil aeration and root penetra-
The two principal forest landscape elements in the Beni savanna, tion (Langstroth, 1996; Yamashita & de Barros, 1997). Adult Attalea
forest islands and gallery forest, are dominated and maintained by palms are fire resistant, because the apical meristem is well above the
the palm species Attalea princeps, hereafter referred to as Attalea ground fire, and insulated by leaf bases that protect the palm against
(Yamashita & de Barros, 1997). Attalea is an important species in the high temperatures (McPherson & Williams, 1998; Souza, Martins, &

F I G U R E   1   Conceptual model showing the hypothesized effects of landscape mosaic, disturbance, and environmental factors (boxes,
left side) on the condition and life cycle of Attalea (boxes, right side). These factors operate at broad to small spatial scales, as indicated by
the vertical arrow. Each of the factors affects a different phase and process in the life cycle. Arrows indicate the expected positive (solid
line) and negative (broken line) effects on the condition and density of individuals at the different life cycle stages, and the effects on the
regeneration processes (italics)
HORDIJK et al. |
      3

Matos, 2000). In contrast, juveniles are fire intolerant because of their 2 | M E TH O DS
superficial meristem position (McPherson & Williams, 1998; Souza
& Martins, 2004) and intense fires can strongly reduce germination 2.1 | Study area
by damaging the fruits (Anderson, May, & Balick, 1991). Additionally,
The Beni savanna is located in the Amazon Basin of northern
large-­seeded palms such as Attalea suffer from density-­dependent
Bolivia and covers an area of 16 million ha (Mayle et al., 2007). The
seed predation by bruchid beetles (Pachymerus cardo) (Salm, 2006),
study was executed in the Barba Azul Nature Reserve (11,000 ha),
and parasites and diseases may also increase Attalea mortality (Gilbert
situated in the center of the Beni savanna (coordinates field sta-
& Webb, 2007).
tion: 13°45′45″S 66°5′54″W). Mean annual temperature is 26°C,
mean annual precipitation is 1,800 mm/year (Herrera-­MacBryde,
1.3 | Environmental factors Dallmeier, MacBryde, Comiskey, & Miranda, 2000), and the wet
season lasts from October to May (Haase & Beck, 1989). The for-
Environmental factors such as soil characteristics, inundation, and
est islands and center of the gallery forests are not flooded during
light availability have a direct influence on the survival and regen-
the wet season, as they are elevated (Mayle et al., 2007). The Beni
eration of Attalea. The Attalea genus is known to be present at a
savanna is classified as a hydromorphic climatic savanna, mean-
nutrient, alkalinity, soil texture, and inundation gradient (Grégoire,
ing that the savanna is mainly flooded by rainwater and receives
2010; Salm, Prates, Simoes, & Feder, 2015). Attalea occurs on dry
limited sediment from the catchment, resulting in a nutrient-­p oor
to seasonally flooded soils; therefore, this species can be expected
soil (Junk et al., 2011).
close to seasonally inundated areas (Moraes & Zenteno-­Ruiz, 2017).
As most palm species, Attalea is a light-­demanding pioneer spe-
cies associated with medium shade tolerance (Fredericksen, 1999), 2.2 | Study species
although shaded, moist environments stimulate seed germination
Recently, a species revision was performed on Attalea phalerata
(Anderson et al., 1991).
in the Beni savanna, resulting in correcting the species to Attalea
princeps throughout most of the ecosystem (Moraes & Zenteno-­
1.4 | Aims and hypotheses Ruiz, 2017; Pintaud, del Castillo, Ferreira, Moraes, & Mejía, 2016).
The habitat of Attalea princeps spreads from 150 up to 1,000
Currently, the population density of Attalea on the forest islands is
meter in the Amazon and Subandean forest in Bolivia, Peru, and
decreasing and the population structure is changing due to a low
West-­B razil (Pintaud et al., 2016). It occurs in high flood-­f ree for-
regeneration rate (T. Boorsma, pers.obs.). If in the next 100 years
ests, seasonally flooded forests, and river edges with fluctuating
the regeneration rate does not recover, the majority of the palms
water levels (Moraes & Zenteno-­Ruiz, 2017). Attalea princeps is a
may disappear with large consequences for the landscape mosaic as
single stemmed palm, has a maximum stem height of 15 m, and
whole, and for the numerous plant and animal species that depend
the crown is formed of 15–20 arching leaves (Pintaud et al., 2016).
on this palm species. This study aims to describe the current popula-
The staminate flowers are spirally inserted on rachillae, and both
tion structure of Attalea and analyze how the landscape mosaic, dis-
staminate and pistillate inflorescences are placed in between the
turbance, and environmental factors affect its population structure
leaves, unisexual, and occur on the same plant. The palm may
and abundance (cf. Figure 1).
flower throughout the year, and the most important pollinators
We hypothesize a positive relationship between palm popu-
are nitidulid beetles of the genus Mystrops. The infructescences
lation density and forest island size, soil nutrient concentration of
are pendulous along the stem with 350–500 fruits, each contain-
the soil and proximity to the river, because we expect that larger
ing 2–5 seeds (Pintaud et al., 2016). The removal of the mesocarp,
islands have a lower extinction rate (MacArthur & Wilson, 2015),
shaded and moist environment stimulates germination, which gen-
soil nutrient concentration positively affects this species (Grégoire,
erally takes 3 months (Anderson et al., 1991).
2010), and this palm is adapted to a seasonally flooded environment
(Moraes & Zenteno-­Ruiz, 2017). We hypothesize a negative relation-
ship between population density and edge effect, seed predation 2.3 | Research design
rate, diseases, grazing pressure, and fire intensity, as we expect that
At the Barba Azul Nature Reserve, Attalea occurs on 25 forest
disturbances are more severe at the edge (Camargo & Kapos, 1995;
islands ranging from 0.02 to 0.8 ha in size (5.7 ha in total) and 8
Cochrane & Laurance, 2002; Williams-­Linera, 1990), seed preda-
patches of riverine gallery forest ranging from 8 to 60 ha in size
tion lowers viable seed availability (Salm, 2006), and disturbances
(245 ha in total). To study the different habitat types of Attalea,
affect the viability and density of the individuals (Gilbert & Webb,
91 plots of 10 × 20 m were established at least 60 m apart in the
2007; Yamashita & de Barros, 1997). Light-­demanding adult palms
center and edge of forest islands and gallery forest: 16 random
are expected to be more abundant with increased light availability
plots in the forest island center, 25 plots in the forest island edge,
(Fredericksen, 1999), and juvenile density is expected to be higher in
25 plots in the gallery forest center, and 25 plots in the gallery
a more shaded environment, as shade stimulates seed germination
forest edge. From the forest edge to 10 m into the forest islands
(Anderson et al., 1991).
 |
4       HORDIJK et al.

or gallery forest was considered edge habitat, as Attalea is one of color and degree of damage (intactness) of the leaves. As a measure
the dominant canopy species and it has an average adult height of for light availability, we used canopy openness which was examined
10 m. The edge plots in the forest island and gallery forest were with a spherical densiometer, Forestry Suppliers, Inc. model A.
randomly placed at the North or South edge to avoid an illumina-
tion bias. In case the diameter of the forest island was <40 m, only 2.6 | Data analysis
an edge plot was established. To quantify the landscape charac-
teristics of forest islands, the diameter of the forest islands was To analyze how habitats differ in population density and (a)biotic vari-
measured with a Nikon Forestry 550 clinometer. ables, a two-­way ANOVA with interaction term was performed with
forest type (i.e., forest island or gallery forest) and position (i.e., forest
2.4 | Plot measurements center or forest edge) as independent variables, using a Tukey HSD as
post hoc test (R package stats). The residuals were tested for normal-
Grazing pressure was inferred by counting the number of dung piles in ity with a Shapiro–Wilk test, and the homogeneity of variances was
each plot. To have a measure for competition, the number of all trees evaluated with a Levene's test. To normalize the data, the variables
(including palms) with a stem diameter (dbh) >5 cm was recorded. Also, juvenile density, adult density, amount of fruits, dung piles, and per-
the overall ground cover was scored in percentages measuring bare soil, centage organic matter were log10-­transformed. The amount of fruits
grass, and shrubs. To have a measure for water availability, the distance produced by each palm was calculated by extrapolating the fallen
from every plot to the river was measured in Google Earth (Google Inc. fruits on 1–6.6 m2. Fruits disperse on average up to 1.5 m from the
2016). To evaluate soil fertility, a soil sample at 30 cm depth in the center stem, and the surface of the dispersal area minus the average DBH
of the plot was taken with a closed hand auger. The soil samples were results in a surface of 6.6 m2 around the stem covered with fruits.
analyzed for organic matter, available phosphate, available nitrogen, and To explore how population characteristics, (a)biotic variables
pH at the laboratory of the Plant ecology and nature conservation group, of forest islands and gallery forest were associated, a Principal
Wageningen University. Percentage organic matter was examined by Component Analysis was performed (Oksanen et al., 2016). The
heating the soil samples for 2 hr at 105°C, followed by heating for 12 hr twelve variables with the highest components loadings on the first
at 550°C. The samples were measured before and after placement in the two axes are visualized in the PCA graph (Figure 3). These variables
oven to determine soil organic matter. The available phosphate and nitro- are habitat, juvenile density, condition juvenile, adult density, condi-
gen were analyzed with a 0.01 M CaCl2 extraction method. The pH was tion adult, amount of fruits, infected fruits, fire damage, dung piles,
measured with a WTW inoLab pH/cond level 1 measurement device. available phosphate, percentage organic matter, and bare soil cover.
Based on the conceptual model (Figure 1) and field observations,
a path model was developed to analyze the mechanism behind the
2.5 | Palm measurements
differences in population density between the forest islands and gal-
Attalea with stem were defined as adults. The following measure- lery forest. Three different models were developed, (a) a model at
ments were taken from the 498 adult palms; height (measured with the plot level, (b) a model at the individual level of the adult, and (c)
a Nikon Forestry 550 clinometer), dbh, number of leaves, number of a model at the individual level of the juvenile. Models on individual
inflorescences, and number of infructescences. Fire intensity was es- adult and juvenile level were developed as it became clear from the
timated by measuring fire scar height and width on adult palms, from path models that the condition of the individual and results of mea-
which the percentage of scarred stem was calculated. Light availabil- surements on individual level were important to explain the juvenile
ity for each adult was estimated with the Dawkins crown illumination density. In the models on individual level, the plot number is used as
classes: (a) no direct light, (b) lateral light, (c) some overhead light, a nesting factor. The variables were analyzed with a generalized lin-
(d) full overhead light, and (e) crown fully exposed (Jennings, Brown, ear model from the Gaussian family. The regression coefficient was
& Sheil, 1999). The health condition of adults was scored as good, used to quantify the direction of the relationship in the model (lm.
medium, or bad based on the condition of the stem and the color and beta function from the R package QuantPsyc; Fletcher, 2012).
degree of damage (intactness) of the leaves. Under each adult palm, a To test the overall correctness of the resulting path models,
1-­m2 subplot was randomly established to examine the total amount structural equation models were developed. With structural equa-
of fallen fruits and the amount of fruits infected by bruchid beetles. tion modeling, it is not only possible to test the hierarchical relations
Attalea without stem were defined as juveniles. Four juvenile sub- between variables, but also test the correctness of the whole model
plots of 5 by 5 m were established in the forest island plots, and two (SEM, as implemented in the R package Lavaan; Rosseel, 2012). A
subplots were established in the gallery forest plots. Two subplots SEM on plot level and a SEM on the individual level of the adults
were representative for juvenile density in the gallery forest, but due were developed. It was not possible to develop a correct SEM on
to larger variation in juvenile abundance in forest islands, four sub- individual level of the juveniles with the restricted number of vari-
plots were established. For each of the 3,699 juveniles, the length ables on individual juvenile level present. As the final path models
of the longest leaf and the form of the leaves were noted as closed, on plot and individual adult level consisted of many variables, the
open, or both (Supporting Information Figure S1). The health condi- SEM's were constructed starting with the significant relationships to
tion of the juveniles was scored as good, medium, or bad based on the juvenile density. Juvenile density is the most important dependent
HORDIJK et al. |
      5

F I G U R E   2   The effect of four habitat types (forest island [fi] vs. gallery forest [gf], and forest edge vs. forest center) on Attalea princeps
population density, disturbances, and environmental factors. The boxes indicate the data range of the second and third quartile, with the
horizontal line as median, and the whiskers indicate the entire data range. Boxes accompanied by a different letter are significantly different
(Tukey post hoc test, p < 0.05). The data are analyzed on plot level (N = 91)

variable in this study explaining the difference in regeneration rate. (Tukey = 0.0078). Crown illumination is higher in the forest islands
The resulting SEM's were selected on a chi-­square p-­value >0.05, compared to the gallery forest (ANOVA, p < 0.001). Cattle pressure
lowest Akaike's Information Criterion (AIC), Tucker–Lewis index (ANOVA, p < 0.0001), bare soil surface (ANOVA, p = 0.0287), phos-
(TLI) >0.95, and Root Mean Square Error of Approximation (RMSEA) phate availability (ANOVA, p < 0.0001), nitrogen availability (ANOVA,
<0.01 indicating an excellent model fit. See for an overview of the p = 0.0021), and percentage organic matter (ANOVA, p < 0.0001) are
correlation values between the variables Supporting Information significantly higher in the forest islands. The pH value of the soil does
Table S2. All statistical analyses were carried out using R Studio not show a significant difference between the habitats.
v. 3.3.2 (Development Core Team 2011). The diameter of the forest island did not have a significant
effect on juvenile, adult, or fruit abundance (Pearson correlation
coefficient, p > 0.05). The distance of the gallery forest plots to
3 | R E S U LT S
the river is on average almost 250 m larger than the forest island
3.1 | Habitats differ in population density and (a) plots (ANOVA, p < 0.001), as also gallery forests are found in old,
biotic factors dried up river arms, indicating that they currently occur in a drier
environment.
Significantly more juveniles (ANOVA, p < 0.0001) and fruits (ANOVA,
p = 0.0002) are present in gallery forest compared to forest islands
(Figure 2). Also, juvenile condition is better in gallery forest (ANOVA,
3.2 | (a)Biotic factors influencing the
p < 0.0001), compared to forest islands (ANOVA, p = 0.001). In
Attalea population
the forest island edge and gallery forest center, the adult density is
higher compared to the forest island center (Tukey, p = 0.018). Fire It became clear that the four habitat types show significant dif-
damage of adults is higher in the forest island edge compared to the ferences in population density and (a)biotic factors, the underly-
gallery forest edge (Tukey, p = 0.0468), and adults have a better con- ing mechanisms influencing the Attalea population are tested with
dition in the gallery forest center compared to the forest island edge a continuous data approach, because within each habitat a large
 |
6       HORDIJK et al.

variation as found. Associations between palm characteristics and fitting model on plot level, without a nested design, this relationship
environmental variables were analyzed with a Principal Component was significant (Supporting Information Figure S3).
Analysis (PCA) (Figure 3). The first two axes explain together 44 per-
cent of the variation. The first axis is mostly associated with habitat,
health condition of the juvenile, organic matter, and available phos- 4 | D I S CU S S I O N
phate. The second axis is explained by adult density, fire damage,
amount of infected fruits, and bare soil cover. The first axis sepa- The aim of this study was to describe the current population struc-
rates the gallery forest to the left side and forest islands to right side, ture of Attalea and analyze how the landscape mosaic, disturbances,
while center and edge plots are not clearly separated. and environmental factors affect its population structure and abun-
The relationship between (a)biotic factors and the population dance. First, we will discuss the lifecycle of Attalea, followed by a
structure of Attalea was analyzed using a structural equation model discussion of all components presented in the conceptual model
(Figure 4). The health condition of the adults had a positive influence (Figure 1), explaining the effect of (a)biotic factors on the population
on the fruit production (Figure 5a) and therefore a positive effect on abundance.
juvenile density (Figure 5b). The model shows that more fruit pro-
duction resulted in more fruits being infected by bruchid beetles.
4.1 | Life cycle of Attalea
The juvenile density was positively affected by the health condition
of the juvenile (Figure 5c), but negatively affected by the organic We hypothesized that high adult density increases fruit production,
matter content and phosphate availability of the soil. Cattle pressure which positively affects juvenile density and therefore adult den-
had a negative effect on health condition of the juveniles (Figure 5d) sity. Adult density has indeed a positive, but non-­significant, effect
and therefore indirectly on the juvenile density. The density of the on the fruit production (Supporting Information Figure S3), but fruit
adults had a positive effect on juvenile density, but this relationship production is most strongly influenced by the health condition of
was not significant, resulted in a 10% higher AIC value, and this re- the adults (Figure 3). As expected, the amount of fruits has a positive
lationship is therefore not included in the final model. In the best effect on juvenile density, but contrary to the hypothesis, juvenile
density does not influence adult density (Figure 3).

4.2 | Landscape mosaic
In line with the island theory, we hypothesized that smaller is-
lands and forest fragments have a lower adult palm density, due to
stronger edge effects and a higher probability of palm extinction due
to a smaller population size (Souza & Martins, 2004; Williams-­Linera,
1990). However, we found that the size of the forest island was not
related to the number of fruits, juveniles, or adults present on the
forest island. The variation in the landscape mosaic is expected to
be large enough to detect differences, as the diameter of forest is-
lands varied fivefold (from 20 to 100 m). One reason for the lack of
landscape effects could be explained by edge effect and/or environ-
mental disturbances like grazing to be more important than the size
or position of the habitat fragment (Laurance, 2008).
We hypothesized that forest edges contain lower densities of
juveniles and adults compared to the center, due to an increased
disturbance intensity. However, we found no difference in juvenile
density between the center and edge (Figure 2), which is proba-
bly caused by a stronger competition at the edge due to a higher
grass, shrub and tree cover, resulting in a stronger competition
for nutrient and water resources at the edge (Svenning, 2001;
F I G U R E   3   Principal component analysis (PCA) of eleven Williams-­L inera, 1990), which compensates for the decrease in ju-
variables related to Attalea princeps density and condition, venile density in the center due to grazing pressure. Also, as the
disturbance (fire, cattle), and soil conditions (P and Soil Organic edge of the raised forest descends toward the savanna, the seeds
Matter-­SOM). The first axis explains 29% of the variation, the
or juveniles could become inundated for a long period during the
second axis 15% of the variation between the plots (N = 91).
wet season resulting in mortal anaerobic conditions (Gurnell,
Symbols indicate the four habitat types: forest island plots (circles),
gallery forest plots (squares), center plots (filled symbols), and edge Goodson, Thompson, Mountford, & Clifford, 2007; Veneklaas,
plots (open symbols) Fajardo, & Lozano, 2005).
HORDIJK et al. |
      7

F I G U R E   4   Structural equation model at the individual adult level, with adults (N = 498) nested within plots (N = 91). Results of the best
fitting structural equation model (N = 91) indicate the (a)biotic factors affecting the abundance of juvenile Attalea. The arrows indicate
significant positive (solid line) and negative (broken line) relationships (p < 0.05) on the condition, density of individuals at the different
stages or regeneration processes (italics), the numbers next to the arrows indicate the standardized path coefficients. This model explains
29% of the variation in the amount of juveniles present

F I G U R E   5   Scatterplots with a fitted linear regression line of relationships between variables influencing juvenile density according to the
structural equation model on individual adult palm level. (a) shows the effect of adult condition on fruit production. The health condition is
scaled from 1 (bad) to 3 (good). (b and c) visualize the influence of fruit abundance and health condition on juvenile density per plot. (d) shows
the effect of grazing pressure on the juvenile health condition
 |
8       HORDIJK et al.

We found that within the forest islands, adult density is higher in We hypothesized that high juvenile density would lead to an
the edge than the center, whereas in the gallery forest, adult density increase in diseases and decreased health, but found instead that
does not differ between the edge and the center (Figure 2). A higher juvenile density led to an increase in juvenile health (Figure 4d). A
adult density in the edge of the forest island can be explained by possible explanation for this positive relationship is that healthy ju-
higher cattle pressure in the center. Cattle use mainly the center of veniles may lead to a better survival and higher juvenile abundance,
the forest island to find shelter and rest, where they browse and or, that both vitality and density are positively affected by the same
trample Attalea regeneration (Yamashita & de Barros, 1997) result- beneficial environmental conditions (Alvarez-­Clare & Kitajima,
ing in less juveniles and therefore adult individuals over time. The 2009). In contrast, adult health condition is not related to adult den-
lower vigor of adults in the forest island edge can be explained by sity, probably due to an observed high resilience of the adult indi-
the fact that the lower situated edge is inundated for a longer period, viduals. The health condition of the adult has a positive effect on
resulting in anaerobic conditions that may increase palm mortality fruit production, therefore indirectly increasing the juvenile density
(Junk et al., 2011). (Figure 4b), and the health condition of the adult is only better in the
To summarize, the size of the forest fragment does not affect the gallery forest center compared to the forest island edge.
population density. However, the landscape context is important, To summarize, grazing pressure negatively affects the juvenile
as in the gallery forest edge the juvenile density is slightly lower density through biomass removal, while fire disturbance does not
than in the center, because of higher competition and flooding at have a significant effect on population density or the condition of
the edge. In contrast, adult density is higher in the forest island edge individuals. Bruchid beetles do affect seed viability, but overall, we
compared to the center, probably because of lower cattle pressure assume that diseases do not influence the population density.
in the edge.

4.4 | Environmental factors
4.3 | Disturbance
We hypothesized that reduced stress (less competition) and a high re-
We hypothesized that grazing pressure would have a negative ef- source availability (nutrients, water, light) would increase the density
fect on the regeneration of Attalea through trampling, grazing, and of juveniles and adults, as the Attalea genus is thought to have high nu-
soil compaction by cattle (Yamashita & de Barros, 1997) (Figure 1). trient requirements (Grégoire, 2010). The amount of bare soil surface
Grazing has indeed a negative influence on juvenile density by dimin- had, contrary to the expectations, a negative effect on juvenile den-
ishing their health condition (Figure 4c) (Eaton, Keuroghlian, Maria sity, probably because the seeds germinate in a shaded environment
do Carmo, Desbiez, & Sada, 2017; Endress, Gorchov, & Noble, 2004; (circumstances which are more likely with vegetation cover) (Anderson
Langstroth, 1996; Mandle & Ticktin, 2012; Shepherd & Ditgen, et al., 1991), and growing in bare soil could increase the chance at des-
2005). Although grazing pressure is higher in forest island compared iccation (Padilla & Pugnaire, 2007). Alternatively, bare soil surface is
to gallery forest, it does not have a significant effect on adult density an indicator of cattle pressure, as grazing and trampling by cattle lead
(Figure 2), probably because adult leaves are out of reach for cattle, to a reduced vegetation cover (see Supporting Information Table S2,
or because adults are more robust and resilient, which allows them Figure S3). In contrast to the expectation, an increase in soil organic
to recover quickly from browsing. matter and soil phosphate content resulted in a decreased juvenile
We hypothesized that fire intensity would have a negative effect density, while nitrogen did not have a significant effect. This result is
on juvenile density and adult health condition. Instead, we found probably an indirect effect of cattle pressure, as defecation by cattle
that fire does not affect adult health condition nor juvenile density. pressure increases soil organic matter, and nitrogen and phosphate
There is a weak negative correlation (Pearson correlation coefficient availability in the forest islands (Haynes & Naidu, 1998; Parham, Deng,
−0.21, p = 0.08) between fire and juvenile health condition (see Raun, & Johnson, 2002; Whalen, Chang, & Olson, 2001). No differ-
Supporting Information Table S2), which indicates a possible indi- ences in soil pH were found between the four habitat types, indicat-
rect effect of fire on juvenile density. However, this fire effect was ing that cattle pressure is not influencing the alkalinity of the soil,
not included in the final structural equation model as it resulted in a although a higher soil organic matter would be expected to result in
lower model fit. Attalea is a fire-­adapted species, because the apical a higher pH (Walker, Clemente, & Bernal, 2004). Neither competition
meristem is covered and well insulated by leaf bases that protect the nor resource availability had an effect on adult density.
palm against high temperatures (Feil, 1996; McPherson & Williams, We hypothesized that the proximity to the nearest river has a
1998; Souza et al., 2000). positive effect on palm abundance, as the adult is adapted to a sea-
We hypothesized that seed predation rate and diseases have a sonally flooded environment and requires sufficient water. We found,
negative effect on the viability of seeds, juvenile, and adult density however, that distance to river is not related to the density or health
of Attalea (Figure 1). Seed viability is indeed negatively affected by condition of juveniles or adults. Distance to river alone is probably
density-­dependent predation, as we found that the number of fruits not a good proxy for water availability, due to the height differences
increased the number of fruits infected by bruchid beetles, resulting caused by the landscape relief in the Beni savanna (Erickson, 2000;
in a lower juvenile density (Figure 3) (Jansen, Visser, Joseph Wright, Mayle et al., 2007) and inundated of the savanna is caused by rainfall
Rutten, & Muller-­L andau, 2014; Salm, 2006). and not solely river overflooding (Junk et al., 2011).
HORDIJK et al. |
      9

We hypothesized that sufficient light enhances adult density foundation to I.H. We thank Asociación Armonía to send the soil sam-
(Fredericksen, 1999), while a more shaded environment enhances ples to the Netherlands and giving the opportunity to execute fieldwork
juvenile density, as the seeds germinate in a shaded environment in the Barba Azul Nature Reserve. We also would like to thank Rosario,
(Anderson et al., 1991). We found that light availability is not related Rene, and Hernan for the logistical support and good care in the reserve.
to adult health condition or adult density. Light is probably not a lim-
iting factor for adults in this ecosystem, as the adult crowns form
DATA AVA I L A B I L I T Y
the forest canopy. Canopy openness did not affect the density or
health condition of juveniles according to the results of the SEM Data available from the Dryad Digital Repository: https://doi.
(Figure 4), but a significant negative correlation was detected be- org/10.5061/dryad.8824sd0 (Hordijk, Meijer, Nissen, Boorsma, &
tween the condition of the juvenile and canopy openness, probably Poorter, 2018).
because a higher canopy openness increases the risk of desiccation
(Supporting Information Table S2). Mean canopy openness above
ORCID
the juveniles is 30 percent (standard deviation = 20%) indicating that
the majority of the seedlings occurred in fairly light conditions. Iris Hordijk  https://orcid.org/0000-0002-6302-6254
To summarize, only juvenile Attalea density is negatively affected
by soil fertility, which is probably not a direct causal effect, but re-
flects increased cattle pressure and defecation. Neither juvenile REFERENCES
density nor adult density is affected by light availability, insufficient Alvarez-Clare, S., & Kitajima, K. (2009). Susceptibility of tree seed-
water availability, or inundation by flooding rivers. lings to biotic and abiotic hazards in the understory of a moist
tropical forest in Panama. Biotropica, 41, 47–56. https://doi.
org/10.1111/j.1744-7429.2008.00442.x
4.5 | Management implications Alvez-Valles, C. M., Balslev, H., Carvalho, F. A., Garcia-Villacorta, R.,
Grandez, C., & Menini Neto, L. (2018). Endemism and conservation
Our results suggest that a decreased grazing pressure will improve of Amazon palms. Biodiversity and Conservation, 27, 765–784. https://
the condition of the juveniles which will lead to an increased juve- doi.org/10.1007/s10531-017-1463-0
Anderson, A. B., May, P. H., & Balick, M. J. (1991). The subsidy from na-
nile density. Fencing of individual islands with a low juvenile density
ture: Palm forests, peasantry, and development on an Amazon frontier.
would be desirable to protect the most vulnerable islands and give Columbia University Press.
the time to the population to recover. Reducing cattle density or ro- Berkunsky, I., Cepeda, R. E., Marinelli, C., Simoy, M. V., Daniele, G.,
tation of cattle to diminish the overall grazing pressure in the long Kacoliris, F. P., … Gilardi, J. D. (2016). Occupancy and abundance
term is recommended as well (Giroldo & Scariot, 2015), where the of large macaws in the Beni savannahs, Bolivia. Oryx, 50, 113–120.
https://doi.org/10.1017/S0030605314000258
rotation regime will depend on the observed recovery time of the
BirdLife International (2015). Ara glaucogularis. The IUCN Red List of
population and the goals of the farmers. Threatened Species 2015.
Brook, B. W., Bradshaw, C. J. A., Koh, L. P., & Sodhi, N. S. (2006). Momentum
drives the crash: Mass extinction in the tropics. Biotropica, 38, 302–
305. https://doi.org/10.1111/j.1744-7429.2006.00141.x
5 | CO N C LU S I O N S Camargo, J. L. C., & Kapos, V. (1995). Complex edge effects on soil
moisture and microclimate in central Amazonian forest. Journal
This study shows that Attalea princeps, a robust species adapted to of Tropical Ecology, 11, 205–221. https://doi.org/10.1017/
a dynamic ecosystem of fire and flooding, is negatively affected by S026646740000866X
Cochrane, M. A., & Laurance, W. F. (2002). Fire as a large-­scale edge ef-
human-­induced disturbances like cattle ranching. Currently, palm den-
fect in Amazonian forests. Journal of Tropical Ecology, 18, 311–325.
sity is almost 8 times higher in the gallery forest in comparison with https://doi.org/10.1017/S0266467402002237
forest islands, caused by a higher juvenile density. Juvenile density is Corrêa, C. E., Fischer, E., & dos Santos, F. A. M. (2012). Seed banks on
positively related to health condition of the juveniles and the amount Attalea phalerata (Arecaceae) stems in the Pantanal wetland, Brazil.
Annals of Botany, 109, 729–734. https://doi.org/10.1093/aob/mcr317
of fruits. In contrast, juvenile density is negatively influenced by graz-
de Carvalho, W. D., & Mustin, K. (2017). The highly threatened and lit-
ing, affecting the health condition of the juvenile, as well as organic tle known Amazonian savannahs. Nature Ecology & Evolution, 1, 100.
matter and phosphate availability in the soil. Management practices https://doi.org/10.1038/s41559-017-0100
that decrease the grazing pressure on forest islands, such as fencing Denevan, W. M. (1966). The aboriginal cultural geography of the Llanos de
mojos of Bolivia. National Academy of Sciences - National Research
off vulnerable forest islands and the implementation of a cattle rota-
Council Washington DC Foreign Field Research Program.
tion system, could result in Attalea population recovery, and the long-­ Development Core Team R (2011). R: A language and environment for
term maintenance of the landscape mosaic of the Beni savanna. statistical computing. Vienna, Austria: R Foundation for Statistical
Computing.
Eaton, D. P., Keuroghlian, A., Maria do Carmo, A. S., Desbiez, A. L. J., &
AC K N OW L E D G M E N T S Sada, D. W. (2017). Citizen scientists help unravel the nature of cattle
impacts on native mammals and birds visiting fruiting trees in Brazil's
This study is funded by grants of the Alberta Mennega foundation, southern Pantanal. Biological Conservation, 208, 29–39. https://doi.
Bottelier Stipendium, Dr. Christine Buisman foundation and Kronendak org/10.1016/j.biocon.2016.09.010
 |
10       HORDIJK et al.

Endress, B. A., Gorchov, D. L., & Noble, R. B. (2004). Non-­timber for- and seedling recruitment in a Neotropical palm. Ecology Letters, 17,
est product extraction: Effects of harvest and browsing on an un- 1111–1120. https://doi.org/10.1111/ele.12317
derstory palm. Ecological Applications, 14, 1139–1153. https://doi. Jennings, S. B., Brown, N. D., & Sheil, D. (1999). Assessing forest can-
org/10.1890/02-5365 opies and understorey illumination: Canopy closure, canopy cover
Erickson, C. L. (1995). Archaeological methods for the study of ancient and other measures. Forestry, 72, 59–74. https://doi.org/10.1093/
landscapes of the Llanos de Mojos in the Bolivian Amazon. Cambridge, forestry/72.1.59
UK: Cambridge University Press. https://doi.org/10.1017/ Junk, W. J., Piedade, M. T. F., Schöngart, J., Cohn-Haft, M., Adeney, J. M.,
CBO9780511521188 & Wittmann, F. (2011). A classification of major naturally-­occurring
Erickson, C. L. (2000). An artificial landscape-­scale fishery in the Bolivian Amazonian lowland wetlands. Wetlands, 31, 623–640. https://doi.
Amazon. Nature, 408, 190–193. https://doi.org/10.1038/35041555 org/10.1007/s13157-011-0190-7
Feil, J. P. (1996). Fruit production of Attalea colenda (Arecaceae) in coastal Killeen, T. J. (1991). The effect of grazing on native Gramineae in
ecuador—an alternative oil resource? Economic Botany, 50, 300–309. Concepción, Santa Cruz, Bolivia. Tropical Grasslands, 25, 12–19.
https://doi.org/10.1007/BF02907337 Killeen, T. J., Siles, T. M., Grimwood, T., Tieszen, L. L., Steininger, M.
Fletcher, T. D. (2012). QuantPsyc: Quantitative psychology tools. R package K., Tucker, C. J., & Panfil, S. (2003). Habitat heterogeneity on a
version 1.5. forest-savanna ecotone in Noel Kempff Mercado National Park
Fredericksen, T. S. (1999). Regeneration status of important tropical for- (Santa Cruz, Bolivia): Implications for the long-term conserva-
est tree species in Bolivia: Assessment and recommendations. Forest tion of biodiversity in a changing climate. In How Landscapes
Ecology and Management, 124, 263–273. Change (pp. 285–312). Springer, Berlin: Heidelberg. https://doi.
Gilbert, G. S., & Webb, C. O. (2007). Phylogenetic signal in plant patho- org/10.1007/978-3-662-05238-9
gen–host range. Proceedings of the National Academy of Sciences of the Langstroth, R. P. (1996). Forest islands in an Amazonian savanna of north-
United States of America, 104, 4979–4983. https://doi.org/10.1073/ eastern Bolivia. Madison: University of Wisconsin.
pnas.0607968104 Laurance, W. F. (2008). Theory meets reality: How habitat fragmen-
Giroldo, A. B., Nascimento, A. R. T., Silva, P. P. F., & Pinho, G. V. (2012). tation research has transcended island biogeographic theory.
Population structure and density of Attalea phalerata Mart. Ex Biological Conservation, 141, 1731–1744. https://doi.org/10.1016/j.
(Arecaceae) in a semideciduous forest. Revista Árvore, 36, 637–645. biocon.2008.05.011
Giroldo, A. B., & Scariot, A. (2015). Land use and management affects the Lombardo, U. (2014). Neotectonics, flooding patterns and landscape
demography and conservation of an intensively harvested cerrado evolution in southern Amazonia. Earth Surface Dynamics, 2, 493–511.
fruit tree species. Biological Conservation, 191, 150–158. https://doi. https://doi.org/10.5194/esurf-2-493-2014
org/10.1016/j.biocon.2015.06.020 Lombardo, U., Szabo, K., Capriles, J. M., May, J.-H., Amelung, W., Hutterer,
Google I. (2016). Google earth (version 7.1.7.2606). R., … Veit, H. (2013). Early and middle Holocene hunter-­gatherer
Grégoire, B. (2010). Le rôle des propriétés du sol dans la distribution des occupations in Western Amazonia: The hidden shell middens. PLoS
palmiers du genre Attalea (Palmae) et des cations échangeables de la ré- ONE, 8, e72746. https://doi.org/10.1371/journal.pone.0072746
gion du Bas Tapajos (Amazonie brésilienne). Loreau, M. (2010). From populations to ecosystems: Theoretical founda-
Gurnell, A., Goodson, J., Thompson, K., Mountford, O., & Clifford, N. tions for a new ecological synthesis (MPB-46). Princeton, NJ: Princeton
(2007). Three seedling emergence methods in soil seed bank studies: University Press. https://doi.org/10.1515/9781400834167
Implications for interpretation of propagule deposition in riparian MacArthur, R. H., & Wilson, E. O. (2015). Theory of Island biogeography
zones. Seed Science Research, 17, 183–199. https://doi.org/10.1017/ (MPB-1). Princeton, NJ: Princeton University Press.
S0960258507774718 Mandle, L., & Ticktin, T. (2012). Interactions among fire, grazing, har-
Haase, R., & Beck, G. (1989). Structure and composition of savanna vege- vest and abiotic conditions shape palm demographic responses
tation in northern Bolivia: A preliminary report. Brittonia, 41, 80–100. to disturbance. Journal of Ecology, 100, 997–1008. https://doi.
https://doi.org/10.2307/2807594 org/10.1111/j.1365-2745.2012.01982.x
Hansen, J., Nazarenko, L., Ruedy, R., Sato, M., Willis, J., Del Genio, A., Mayle, F. E., Langstroth, R. P., Fisher, R. A., & Meir, P. (2007). Long-­term
… Menon, S. (2005). Earth's energy imbalance: Confirmation and forest–savannah dynamics in the Bolivian Amazon: Implications
implications. Science, 308, 1431–1435. https://doi.org/10.1126/ for conservation. Philosophical Transactions of the Royal Society
science.1110252 B Biological Science, 362, 291–307. https://doi.org/10.1098/
Haynes, R. J., & Naidu, R. (1998). Influence of lime, fertilizer and manure rstb.2006.1987
applications on soil organic matter content and soil physical con- McPherson, K., & Williams, K. (1998). Fire resistance of cabbage
ditions: A review. Nutrient Cycling in Agroecosystems, 51, 123–137. palms (Sabal palmetto) in the southeastern USA. Forest Ecology
https://doi.org/10.1023/A:1009738307837 and Management, 109, 197–207. https://doi.org/10.1016/
Herrera-MacBryde, O., Dallmeier, F., MacBryde, B., Comiskey, J. L., & S0378-1127(98)00243-6
Miranda, C. (2000). La Reserva de la Biosfera Estación Biológica del Moraes, M., & Zenteno-Ruiz, F. S. (2017). El género Attalea (Arecaceae)
Beni, Bolivia. Biodiversidad, Conserv. y Manejo la Región la Reserv. la de Bolivia: Afinidades con sistemas ecológicos regionales. Revista
Biosf. Estac. Biológica del Beni. Boliv. SI/MAB, 4, 21–25. Peruana de Biología, 24, 273–282. https://doi.org/10.15381/rpb.
Hesse, A. J., & Duffield, G. E. (2000). The status and conservation of the v24i3.13913
blue-­throated macaw Ara glaucogularis. Bird Conservation International, Naeem, S., Duffy, J. E., & Zavaleta, E. (2012). The functions of biological
10, 255–275. https://doi.org/10.1017/S0959270900000216 diversity in an age of extinction. Science, 336, 1401–1406. https://
Hordijk, I., Meijer, F., Nissen, E., Boorsma, T., & Poorter, L. (2018). Data doi.org/10.1126/science.1215855
from: Cattle affect regeneration of the palm species Attalea princeps Navarro, G., & Maldonado, M. (2002). Geografía ecológica de Bolivia:
in a Bolivian forest-­savanna mosaic. Dryad Digital Repository. https:// vegatación y ambientes acuáticos. Centro de Ecología Simón I.
doi.org/10.5061/dryad.8824sd0 Patiño.
Ibisch, P. L., Beck, S. G., Gerkmann, B., & Carretero, A. (2003). Ecoregiones Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn,
y ecosistemas. P. L. Ibisch and G. Mérida (Eds.). Ministerio de Desarrollo D., … Wagner, H. (2016). vegan: Community ecology package. R pack-
Sostenible y Planificación, Editorial FAN, Santa Cruz de la Sierra, Bolivia. age version 2.4-0.
Jansen, P. A., Visser, M. D., Joseph Wright, S., Rutten, G., & Muller- Padilla, F. M., & Pugnaire, F. I. (2007). Rooting depth and soil mois-
Landau, H. C. (2014). Negative density dependence of seed dispersal ture control Mediterranean woody seedling survival during
HORDIJK et al. |
      11

drought. Functional Ecology, 21, 489–495. https://doi.org/10.1111/ Veneklaas, E. J., Fajardo, A., & Lozano, J. (2005). Gallery forest types and their
j.1365-2435.2007.01267.x environmental correlates in a Colombian savanna landscape. Ecography,
Parham, J. A., Deng, S. P., Raun, W. R., & Johnson, G. V. (2002). Long-­term 28, 236–252. https://doi.org/10.1111/j.0906-7590.2005.03934.x
cattle manure application in soil. I. Effect on soil phosphorus levels, Walker, D. J., Clemente, R., & Bernal, M. P. (2004). Contrasting effects
microbial biomass C, and dehydrogenase and phosphatase activities. of manure and compost on soil pH, heavy metal availability and
Biology and Fertility of Soils, 35, 328–337. growth of Chenopodium album L. in a soil contaminated by pyritic
Pintaud, J. C., del Castillo, A. R., Ferreira, E. J. L., Moraes, M., mine waste. Chemosphere, 57, 215–224. https://doi.org/10.1016/j.
& Mejía, K. (2016). Towards a revision of Attalea in Amazonia. chemosphere.2004.05.020
Palms, 60, 57–77. Watt, A. S. (1947). Pattern and process in the plant community. Journal of
Rios, R. S., & Pacheco, L. F. (2006). The effect of dung and dispersal Ecology, 35, 1–22. https://doi.org/10.2307/2256497
on postdispersal seed predation of Attalea phalerata (Arecaceae) Whalen, J. K., Chang, C., & Olson, B. M. (2001). Nitrogen and phospho-
by bruchid beetles. Biotropica, 38, 778–781. https://doi. rus mineralization potentials of soils receiving repeated annual cat-
org/10.1111/j.1744-7429.2006.00209.x tle manure applications. Biology and Fertility of Soils, 34, 334–341.
Rosseel, Y. (2012). Lavaan: An R package for structural equation model- https://doi.org/10.1007/s003740100416
ing. Journal of Statistical Software, 48, 1–36. Williams-Linera, G. (1990). Vegetation structure and environmental con-
Salm, R. (2006). Invertebrate and vertebrate seed predation in the ditions of forest edges in Panama. The Journal of Ecology, 356–373.
Amazonian Palm Attalea maripa1. Biotropica, 38, 558–560. https:// https://doi.org/10.2307/2261117
doi.org/10.1111/j.1744-7429.2006.00163.x Yamashita, C., & de Barros, Y. M. (1997). The blue-­throated macaw Ara
Salm, R., Prates, A., Simoes, N. R., & Feder, L. (2015). Palm community glaucogularis: Characterization of its distinctive habitats en savan-
transitions along a topographic gradient from floodplain to terra nahs of the Beni, Bolivia. Revista Brasileira de Ornitologia, 5, 10.
firme in the eastern Amazon. Acta Amazonica, 45, 65–74. https://doi.
org/10.1590/1809-4392201401533
Shepherd, J. D., & Ditgen, R. S. (2005). Human use and small mam- S U P P O R T I N G I N FO R M AT I O N
mal communities of Araucaria forests in Neuquén, Argentina.=.
Mastozoología Neotropical, 12, 217–226. Additional supporting information may be found online in the
Souza, A. F., & Martins, F. R. (2004). Population structure and dynam- Supporting Information section at the end of the article. 
ics of a neotropical palm in fire-­impacted fragments of the Brazilian
Atlantic Forest. Biodiversity and Conservation, 13, 1611–1632. https://
doi.org/10.1023/B:BIOC.0000029326.44647.7f
Souza, A. F., Martins, F. R., & Matos, D. M. S. (2000). Detecting ontoge- How to cite this article: Hordijk I, Meijer F, Nissen E,
netic stages of the palm Attalea humilis in fragments of the Brazilian Boorsma T, Poorter L. Cattle affect regeneration of
Atlantic forest. Canadian Journal of Botany, 78, 1227–1237. the palm species Attalea princeps in a Bolivian
Svenning, J.-C. (2001). On the role of microenvironmental heterogene-
forest–savanna mosaic. Biotropica. 2019;00:1–11.
ity in the ecology and diversification of neotropical rain-­forest palms
(Arecaceae). Botanical Review, 67, 1–53. https://doi.org/10.1007/ https://doi.org/10.1111/btp.12613
BF02857848