Diversity and dominance patterns in

Amazon coast dune forest island tree

communities

Rachel Macedo da Silva, Ângela Cristina

Alves Reis & Ulf Mehlig

Plant Ecology
An International Journal

ISSN 1385-0237
Volume 219
Number 3

Plant Ecol (2018) 219:343-357

DOI 10.1007/s11258-018-0799-1

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Science+Business Media B.V., part of Springer
Nature. This e-offprint is for personal use only
and shall not be self-archived in electronic
repositories. If you wish to self-archive your
article, please use the accepted manuscript
version for posting on your own website. You
may further deposit the accepted manuscript
version in any repository, provided it is only
made publicly available 12 months after
official publication or later and provided
acknowledgement is given to the original
source of publication and a link is inserted
to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.

1 23
 Author's personal copy
Plant Ecol (2018) 219:343–357
https://doi.org/10.1007/s11258-018-0799-1

Diversity and dominance patterns in Amazon coast dune

forest island tree communities
Rachel Macedo da Silva . Ângela Cristina Alves Reis . Ulf Mehlig

Received: 6 July 2017 / Accepted: 12 January 2018 / Published online: 19 January 2018
Ó Springer Science+Business Media B.V., part of Springer Nature 2018

Abstract Aiming to better understand richness, forested dune ridges and surveyed areas outside of
dominance, diversity and community changes of plots for additional species. Richness was lower at
coastal vegetation, we studied woody plant commu- Camarauaçu (18 species vs. 25, 26 and 28 at
nities and soil parameters of equatorial dune forest Bonifácio, Apeu-Salvador and Buraco Beach, respec-
islands. We investigated four sites isolated by man- tively). Fisher’s a ranged between 3.2 and 4.7,
groves (Buraco Beach, Bonifácio, Camarauaçu, Apeu- Pielou’s J’ between 0.70 and 0.80. Rényi profiles
Salvador) and compared data to a paleodune forest confirmed lower diversity for Camarauaçu, dominated
with higher floristic and structural complexity (Salinas by a small group of species with high importance
do Roque) for assessment of the regional species pool. values. Within-site beta diversity was lowest at Buraco
We determined grain size distribution and organic Beach. Species were mainly wide-spread generalists.
matter content at Buraco Beach, Camarauaçu and We found low richness compared to the Salinas do
Apeu-Salvador. Grain size distribution was similar Roque reference site; null model tests indicated that
among sites but upper soil layers at Camarauaçu the species pool of the dune sites was partly shaped by
showed low organic matter contents. We recorded environmental filtering. We attribute differences in
trees and shrubs along several transect plots crossing species composition and diversity/dominance patterns
among the four dune sites to distance to the mainland
Communicated by James D A Millington. and stochastic events with effects on non-standard
long distance dispersal.
Electronic supplementary material The online version of
this article (https://doi.org/10.1007/s11258-018-0799-1) con-
tains supplementary material, which is available to authorized
users.

R. M. da Silva (&)
Instituto de Ciências Exatas e Naturais, Universidade Â. C. A. Reis  U. Mehlig
Federal do Pará, Campus de Belém, Rua Augusto Corrêa Instituto de Estudos Costeiros, Universidade Federal do
01, Guamá, Belém, Pará 66075-110, Brazil Pará, Campus Bragança, Alameda Leandro Ribeiro s/n,
e-mail: [email protected] Bragança, Pará 68600-000, Brazil
e-mail: [email protected]
R. M. da Silva U. Mehlig
Programa de Pós-Graduação em Biologia Ambiental, e-mail: [email protected]
Universidade Federal do Pará, Campus Bragança,
Alameda Leandro Ribeiro s/n, Bragança, Pará 68600-000,
Brazil

123
 Author's personal copy
344 Plant Ecol (2018) 219:343–357

Keywords Brazil  Environmental filtering  Long Vegetation types found on the sandy coastal plains
distance dispersal  Sandy soil  Organic matter  of Brazil reflect niche availability; additionally,
Restinga species composition is influenced by the surrounding
vegetation matrix on the landward side (Scarano
2002). Niche availability may depend on historical and
Introduction geological aspects of the processes forming the sandy
coastal plains; the result is a range of different
Sandy deposits are common features of coastlines environmental settings along the more than 7000 km
around the world (Araujo and Pereira 2004). Their of Brazilian coastline, which spans a latitudinal range
vegetation has long been subject to ecological inves- from approximately 4°N to 34°S (Araujo and Pereira
tigation, for example due to the succession during 2004). The causes for the floristic differences among
dune development and the fact that plant community regions with sandy plains along this extensive coast-
composition and structure on these deposits univer- line (and the differences themselves) are still not
sally undergo shifts from the shoreline in landward satisfactorily known. As the respective areas vary in
direction along gradients of tidal influence, wind size and contiguity, distance-related dispersal limita-
exposure, substrate stability, salt spray, pH, organic tions might play a role in structuring communities (e.g.
matter content and soil humidity, among others (e.g. Scarano 2002). It is, at present, not easy to tell distance
Ellenberg 2009). effects apart from those caused by environmental
However, even when considering a specific zone gradients, which are, to date, not sufficiently well
along the gradient between seashore and mainland documented (Silva et al. 2015). However, among the
within a sandy coastal plain ecosystem complex, environmental parameters influencing plant commu-
species do not necessarily show a homogeneous nities, special significance is attributed to soil charac-
distribution at a given local or regional spatial scale teristics (El-Demerdash et al. 1995; Almeida et al.
(Kim and Yu 2009). Community composition may 2011; Lima et al. 2011; Melo and Boeger 2015).
vary distinctly between sites. Site-to-site species To examine the processes that cause changes in
turnover may reflect deterministic processes like plant community composition, it may be useful to
adaptations of species to differences of climate or observe species beta diversity on a regional scale
substrate (Brunbjerg et al. 2014). It may, however, (Condit et al. 2002). Among the reasons for changes in
also result from limited dispersal abilities (Leibold community composition among sites are environmen-
et al. 2004). Therefore, the presence of a species at a tal filtering, interactions between organisms and
given site on the coastal plain will depend on the dispersal limitations (de Cáceres et al. 2012). How-
ability of the species’ diaspores to reach the area in ever, the degree of influence attributed to filtering by
question as well as on its survival under the local environmental conditions and to stochastically gov-
environmental conditions. Moreover, stochastic erned processes, e.g. in dispersal, has been discussed
events influence floristic composition by affecting, controversially (see, for example, Chase and Myers
e.g. diaspore survival in long distance dispersal. 2011). For sandy coastal plain vegetation, the relative
The processes influencing plant species distribution importance of deterministic and stochastic processes
vary within a wide range of spatial scales, with for the formation of communities has had relatively
different sets of factors shaping vegetation structure little consideration (Silva et al. 2015).
and community composition at each level (Chase and Studies of coastal dune vegetation along the
Myers 2011). However, it is generally accepted that Amazon coast have focused mainly on compiling
both spatial and environmental processes are involved local-scale floristic data (e.g. Amaral et al. 2008; Silva
in determining species distribution and relative abun- et al. 2010). Moreover, the existing data are strongly
dance (Chase and Myers 2011; Garzon-Lopez et al. biased towards a few, easily accessible sites. There-
2014). The balance between stochastic processes and fore, ecological aspects of community changes, phy-
environmental factors may also change with the tosociological associations and species biogeography
observed spatial scale as well as with the degree of remain unclear.
heterogeneity of environmental conditions at the An interesting aspect of coastal dune forest sites in
locality under study (Chase and Myers 2011). the Amazon is that the dune areas are frequently

123
 Author's personal copy
Plant Ecol (2018) 219:343–357 345

Fig. 1 Location of the study areas in Brazil (small map) and within the Bragança-Viseu basin

isolated from each other by extensive mangrove discuss deterministic and stochastic processes which
forests as well as by numerous bays and smaller might control community patterns.
estuaries and are thus, in fact, islands. If environmen-
tal conditions among sites were homogeneous, differ-
ences in species composition among sites should be Materials and methods
explained by (predominantly stochastic) long distance
dispersal events. On the other hand, community Study area
changes consistent with environmental parameters
like soil conditions would hint at environmental We chose four study sites along the Amazon coast,
filtering. The geographical setting thus provides an Pará state, Brazil. All sites are located south of the
interesting testbed for hypotheses regarding the shap- Amazon mouth. The coastline features numerous bays
ing of plant communities. and small estuaries. It crosses different geological
The current study aims to improve our understand- sectors, among them the Pará platform with a
ing of woody plant species richness, diversity patterns comparatively slim fringe of mangroves and beaches
and community changes of tropical coastal dune areas and the Bragança-Viseu basin with more extensive,
by comparing community composition within and mangrove-covered mudflats (Souza-Filho 2005). The
among sites. Likewise, we assess soil conditions region is characterised as meso-macrotidal and semid-
potentially relevant for species distribution patterns at iurnal; the tidal amplitude may reach more than 5 m
both spatial scales. Based on this dataset and addi- (Souza-Filho et al. 2009).
tional floristic data from neighbouring areas, we The dominant ecosystem along the coast is man-
grove forest on the tidal mudflats (Menezes et al. 2008;

123
 Author's personal copy
346 Plant Ecol (2018) 219:343–357

Table 1 Distance to the mainland (direct line of flight), total extension and extension of dune ridges at the study sites along the
Amazon coast
Site Dist. to mainland (km) Total area (ha) Dune ridge area (ha)

Apeu-Salvador 19 796.1 507.5

Bonifacio 20 124 49.5
Buraco Beach 10 58.3 20.6
Camarauaçu 22 460.5 230.6
Salinas Roque 9 490 68.7

Mehlig et al. 2010), located between the high spring 50 mm. The annual mean precipitation is 2470 mm;
tide and the mean tidal level. Along the waterfront, the annual mean temperature is 26 °C, with minimum
mangroves are frequently bordered by beaches. In temperature above 18 °C (INMET 2014).
addition, there are sandy deposits completely sur-
rounded by mangroves; these are usually interpreted Floristic inventory and forest structure
as former barrier islands (Souza-Filho et al. 2009). The
vegetation types of dunes and dune slacks have been Field work was carried out in 2006–2007 (Bonifácio)
described by Silva et al. (2010). Fixed dunes are and 2013–2015 (remaining sites) along transects
usually covered by low forest or shrubby vegetation. perpendicularly crossing dune ridges with dune forest.
All study sites are located within the Bragança- Candidate positions for transects at Buraco Beach,
Viseu basin, Buraco Beach (47°060 4000 W, 0°410 1300 S) Camarauaçu and Apeu-Salvador were selected based
near the division between both sectors and Bonifácio on satellite images (CBERS2b) (Barbosa and Ceballos
(46°370 0900 W, 0°500 1900 S), Camarauaçu (46°250 4200 W, 1995) at random positions within the central portion of
0°530 3100 S) and Apeu-Salvador (46°110 5700 W, the available larger dune ridges of each site, after
0°550 0400 S) further east in the centre of the basin eliminating areas with a high percentage of large gaps.
(Fig. 1). The distance between Buraco Beach and Transect positions at Bonifácio were chosen based on
Bonifácio is about 58 km, from Bonifácio to Cama- Ikonos satellite images in approximately equidistant
rauaçu 22 km and between Camarauaçu and Apeu- intervals along a single large dune ridge. As canopy
Salvador is 26 km. The sites are separated from the height could not be estimated reliably from remote
mainland by extensive mangrove forests. Buraco sensing data, we checked the selected sites on the
Beach, Camarauaçu and Apeu-Salvador can only be ground and excluded those with average canopy height
reached by small boats, Bonifácio also by road. \ 4.5 m. Four transects per site were used at
Distances to the mainland and the extension of the Bonifácio, Camarauaçu and Apeu-Salvador, three at
dune ridges with the surrounding marsh complexes are Buraco Beach. Along each transect at Buraco Beach,
listed in Table 1. The Bonifácio dune site is located Camarauaçu and Apeu-Salvador, 10 m 9 4 m plots
next to Bonifácio village which occupies the south- were established. Original transect plots at Bonifácio
east portion of its dune ridges since the 1990s. At the were larger (10 m 9 10 m) but vegetation data from
beach of Camarauaçu, there are a small number of central 10 m 9 4 m subplots were extracted from a
houses occupied by local fishermen; at Apeu-Sal- database with exact tree positions (Silva, unpub-
vador, there is a small fishing village near the lished). Due to local differences in the extension of
waterfront. Marshland between dune ridges has been dune ridges, transect length and number of plots varied
used as pasture for cattle at Bonifácio, Camarauaçu within and among sites. The total sampling areas were
and Apeu-Salvador. 0.092, 0.176, 0.132 and 0.104 ha at Buraco Beach,
The regional climate is warm and humid, with peak Bonifácio, Camarauaçu and Apeu-Salvador, respec-
rainfalls between January and May and a dry season tively. Within plots, all shrubs and trees with circum-
between September and November. The monthly ference at ground height C 7 cm were recorded.
mean precipitation in the driest months is below Circumference was measured at an approximately

123
 Author's personal copy
Plant Ecol (2018) 219:343–357 347

Table 2 Site-wise averaged sediment grain size distribution parameters according to Folk and Ward (1957)
Site Sorting Kurtosis Skewness Mean size Median size

Apeu-Salvador 0.29 ± 0.02 0.94 ± 0.16 0.16 ± 0.04 3.00 ± 0.05 2.95 ± 0.04
Buraco Beach 0.29 ± 0.03 0.97 ± 0.22 0.14 ± 0.16 2.98 ± 0.06 2.94 ± 0.09
Camarauaçu 0.28 ± 0.04 0.90 ± 0.13 0.07 ± 0.13 3.04 ± 0.06 3.02 ± 0.08
All parameters in u units

cylindrical stem section as close to the ground as considerably older (Souza-Filho et al. 2009) and
possible. Within each dune ridge, transects passed features more developed soils. Vegetation was sam-
through a sequence of crests and slacks; however, pled along 10 m 9 100 m transect plots (Santos et al.
significant slopes occurred only in a few plots at 2014). To facilitate comparison with the dune sites, we
Bonifácio and Camarauaçu. Therefore, no slope aggregated data in several contiguous units of 160 m2
correction was applied, and reported plot area corre- from all sites by means of our database with tree
sponds to surface area, not to planar projection. coordinates, yielding 4 (Buraco Beach), 10 (Bonifá-
Multi-stemmed trees or bushes with ramifications cio, Camarauaçu), 5 (Apeu-Salvador) and 60 (Salinas
below the point of circumference measurement were do Roque) sampling units.
considered as single units in tree density calculations. To complement the sampling procedure described
Basal area coverage was calculated by summing up the above, we recorded the occurrence of additional
cross-sectional areas of all stems equal to or greater woody species spotted during extensive walks through
than the established minimum circumference. How- the dune forests at each site. Vouchers of all species
ever, it was not possible to estimate the stem-based were deposited at the herbarium of the Institute for
ground cover of dense clusters of young stemless Coastal Research (Instituto de Estudos Costeiros),
Astrocaryum vulgare Mart. palms; presence and Federal University of Pará, Campus Bragança
number of the clusters were, however, recorded. (HBRA; Thiers (continuously updated)). Family cir-
Calculation of forest structure parameters (frequency, cumscription follows APG IV (2016).
relative density, relative dominance, relative fre-
quency, importance value) followed Müller-Dombois Collection and analysis of soil samples
and Ellenberg (1974).
For comparison of the four dune sites’ species pool Along two transects at Buraco Beach, Camarauaçu
with neighbouring environments, we used data from and Apeu-Salvador, soil samples were collected from
Santos et al. (2014). This site (Salinas do Roque; two trenches dug out on dune ridge crests and two
46°400 0900 W, 0°550 3000 S) is a tall, closed-canopy forest trenches in dune slacks. At each trench, samples were
on a paleodune on the same mangrove-covered taken from two depth ranges (5–20, 40–55 cm; total
peninsula as Bonifácio (Fig. 1, Table 1). This paleo- number of samples: 45). To determine grain size
dune is geologically similar to the remaining sites but distribution, oven-dried (\ 70 °C) subsamples of 5 g

Fig. 2 Soil organic matter

content in two depth ranges
(5–20 and 40–55 cm) at
three of the study sites along
the Amazon coast

123
 Table 3 Forest structure parameters for the four study sites along the Amazon coast
348

family/species Burac o Bonifáci o C a m a r a u aç u A p e u- S a l v a d o r S a l. R o .

freq. dens. dom. rel. rel. IV freq. dens. dom. rel. rel. IV freq. dens. dom. rel. rel. IV freq. dens. dom. rel. rel. IV
dens. dom. dens. dom. dens. dom. dens. dom.
A n ac a r di a ce a e
Anacardium occidentale 13,0 32,6 1,1 0,6 5,4 7,6 27,3 96,6 1,7 4,6 14,6 26,8 30,3 204,5 12,5 11,2 47,9 70,1 15,4 76,9 2,7 1,9 16,0 20,9

123
Spondias mombin **
Tapirira guianensis 36,4 204,5 2,2 11,2 8,5 32,8 30,8 153,8 1,3 3,9 8,0 17,8
Apocynaceae
Himatanthus articulatus 51,5 712,1 4,2 38,8 16,2 73,7 57,7 221,2 1,1 5,6 6,9 23,5
A r ec ac e ae
Astrocaryum vulgare 69,6 326,1 2,6 5,9 12,9 27,4 22,7 108,0 1,1 5,2 9,3 20,8 72,7 378,8 2,9 20,7 11,2 58,2 *
A s te r a ce a e
Lepidaploa grisea 8, 7 2 1, 7 0,0 0,4 0,1 1,5 ** **
B ur serace ae
Protium heptaphyllum 56,5 239,1 1,0 4,3 4,8 16,2 11,4 39,8 0,1 1,9 1,0 6,1 12,1 106,1 2,2 5,8 8,4 18,6 57,7 567,3 1,7 14,3 10,0 35,5
Capparaceae
Cynophalla flexuosa **
Celastraceae
Maytenus obtusifolia 4,3 10 ,9 0, 0 0,2 0 ,2 0, 9 ** 1 5, 2 4 5, 5 1 ,0 2 ,5 3 ,9 11,9 3 ,8 1 9 ,2 0, 0 0 ,5 0 ,2 1, 4
Chrysobalanaceae
Chrysobalanus icaco ** ** 6,1 1 5, 2 0, 4 0,8 1 ,3 4 ,4 **
Clus iaceae
Clusia hoffmannseggiana ** ** ** **
Garcinia madruno 3,8 19,2 0,0 0,5 0,1 1,4
Combretaceae
Buchenavia tetraphylla ** **
Erythroxylaceae
Erythroxylum barbatum ** 4,5 11,4 0,0 0,5 0,1 1,9
Erythroxylum sp. 60,9 217,4 0,2 3,9 1,0 12,5
F a ba c e a e
Chloroleucon acacioides 17,4 43,5 0,4 0,8 1,9 4,8 13,6 56,8 0,3 2,7 2,9 9,4 9,1 30,3 0,1 1,7 0,5 5,5 7,7 19,2 0,1 0,5 0,5 2,4
Copaifera sp. ** **
Inga heterophylla 2,3 5,7 0,0 0,3 0,1 1,0 3,8 9,6 0,0 0,2 0,1 1,1
H u mi r ia c e a e
Humiria balsamifera 9 ,1 5 1 ,1 1 ,0 2, 5 8 , 8 13, 8 ** 7, 7 3 8, 5 0, 6 1, 0 3,7 6, 2
Malpig hi ace ae
Byrsonima coccolobifolia **
Byrsonima crassifolia ** 4 0 , 9 2 0 4, 5 1,2 9 ,8 1 0, 6 3 1, 8 ** 11,5 28,8 0,6 0 ,7 3,6 6,6
Melastomataceae
Mouriri guianensis ** 3, 0 7,6 0, 0 0 ,4 0,2 1,7 **
Mo r a c e a e
Ficus citrifolia ** 3,0 7 ,6 0,1 0,4 0,4 1, 9
Myrtaceae
Eugenia protenta **
Eugenia sp. 87,0 608,7 2,0 11,0 9,7 31,5 57,7 730,8 0,7 18,4 4,1 33,6
Myrcia cuprea 43,5 282,6 1,1 5,1 5,2 15,8 61,4 471,6 1,0 22,6 8,3 48,0 61,5 442,3 0,9 11,2 5,1 28,1
Myrcia sp. 8,7 21,7 0,0 0,4 0,1 1,6
Myrcia splendens 69,6 782,6 2,3 14,1 11,4 34,2 3,8 9,6 0,0 0,2 0,1 1,0
Nyctaginaceae
Guapira opposita 4,3 10,9 0,0 0,2 0,0 0,8 15,9 39,8 0,1 1,9 0,8 7,1 6,1 15,2 0,0 0,8 0,1 3,1 7,7 19,2 0,0 0,5 0,1 2,0
O ch na c e a e
Author's personal copy

Ouratea cassinifolia **
Ouratea caudata 73,9 326,1 0,7 5,9 3,5 18,6 34,1 181,8 0,5 8,7 4,4 22,6 18,2 53,0 0,1 2,9 0,4 9,9 65,4 528,8 0,7 13,3 4,0 30,0
Ouratea cearensis 8,7 21,7 0,0 0,4 0,1 1,5
Ouratea sp. 3,0 7,6 0,0 0,4 0,2 1,7 11,5 28,8 0,1 0,7 0,6 3,5
P e ra c e a e
Pera glabrata 13,0 43,5 1,8 0,8 9,0 11,4
P o l yg o n a c e a e
Coccoloba ramosissima 78,3 641,3 1,7 11,6 8,5 29,8 40,9 340,9 0,7 16,3 6,3 34,1 15,4 86,5 1,4 2,2 8,6 13,7
Ru b ia ce a e
Alibertia edulis **
Alibertia sp. 91,3 1467,4 1,9 26,5 9,2 47,1 34,6 182,7 0,1 4,6 0,8 12,1
Faramea sp. 4,3 10,9 0,0 0,2 0,0 0,8 3,8 19,2 0,0 0,5 0,1 1,3
Guettarda angelica ** 2,3 5,7 0,0 0,3 0,0 0,9 9,1 45,5 0,2 2,5 0,8 6,6
Tocoyena brasiliensis 30,4 108,7 0,1 2,0 0,6 6,3 13,6 39,8 0,0 1,9 0,3 6,0 3,8 19,2 0,0 0,5 0,2 1,4
Sapotaceae
Manilkara triflora 60,9 315,2 3,3 5,7 16,4 29,7
Pouteria ramiflora ** 59,1 431,8 3,8 20,7 32,6 69,8 53,8 740,4 4,6 18,7 27,3 56,4

Frequ. frequency (percentage of plots with occurrence of the species), Dens. density (ind. ha-1), Dom. dominance based on stem basal area (m2 ha-1), Rel. dens. relative density
(%), Rel. dom. relative dominance (%), IV importance value (sum of relative frequency, relative density, relative dominance), Sal. Ro. species occurs at Salinas do Roque
reference site (complete list of species provided as online supplement)
*Species present in plots but plants without measurable stripes
**Species seen outside of plots
Plant Ecol (2018) 219:343–357
 Author's personal copy
Plant Ecol (2018) 219:343–357 349

Species accumulation curves were calculated using

sample-based rarefaction (‘‘exact’’ method of specac-
cum) (Colwell et al. 2012), and the total (seen ? un-
seen) number of species was obtained by Chao,
jackknife and bootstrap estimators (function specpool
of vegan). Diversity among sites was compared by
means of Rényi profiles (Kindt et al. 2006); addition-
ally, Fisher’s a and Pielou’s evenness measure J’ were
calculated. Community structure patterns were
inspected by non-metric multidimensional scaling
with a quantitative Jaccard distance matrix applied
to basal area data derived from stem circumference at
Fig. 3 Species–area curves for the four study sites. Thin dotted
lines indicate confidence envelopes ground height and Jaccard and Raup–Crick distance
matrices applied to presence/absence data. Hetero-
were sieved in a sieving column (1/4 U intervals) on a geneity of species composition within sites was
mechanical shaker for 7 min. Material from each sieve analysed by multivariate dispersion around group
was weighed and classified according to Folk and centroids (Anderson et al. 2006) using presence–
Ward (1957) and Wentworth (1922). The percentage absence data; as measure of beta diversity we used
of organic matter in soil samples was determined Whittaker’s bw and bz (based on the species–area
gravimetrically through weight loss of initially oven- relation) (Koleff et al. 2003). For procedures with
dried subsamples (104 °C) after combustion at presence–absence data, we ran alternative analyses
450 °C. Soil samples from Camarauaçu and Apeu- with stemless Astrocaryum clusters added to the
Salvador were tested for salinity by gravimetric dataset. The influence of distance to the mainland,
determination of the water content of the fresh sample, dune ridge area and their interaction on the commu-
followed by measuring salinity with an electronic nities was tested by PERMANOVA (adonis2 of
conductivity metre (WTW 304i) in a sediment-water package vegan). For assessing the stochastic or
extract and referring the outcome to the initial water deterministic character of community changes, we
content (Schwendenmann et al. 2006). used a null model-based approach with a modified
Raup–Crick procedure (see Chase et al. 2011) on the
Data analysis dataset including the Salinas do Roque site. This
procedure estimates the degree of change among
All calculations were done with R 3.4.2 (R Core Team communities by chance alone, correcting for dissim-
2017). R packages vegan (Oksanen et al. 2017) and ilarity due to different a diversities, and permits
rysgran (Gilbert et al. 2014) were used for community interpretation e.g. regarding environmental filtering
analyses and for grain size data classification, respec- and dispersal limitations (Chase et al. 2011).
tively. Differences in grain size parameters and soil
organic matter were tested by Analysis of Variance
(ANOVA) with main factors soil depth range, site and Results
the respective interaction. Influence of sampling point
location (crest/slack) had been found to be insignif- Soil characteristics
icant in preliminary analyses and corresponding data
were pooled. Quality of ANOVA models was assessed All samples were well-sorted and consisted of pre-
by examining diagnostic plots of residuals. If neces- dominantly fine and very fine sand (99.6–99.8%).
sary, data were transformed according to the Box–Cox Grain size distribution parameters are listed in
procedure using routines from R package car (Fox and Table 2. The mean of the samples’ average grain size
Weisberg 2011). In case of significant differences in the upper and the lower soil depth range was
between sites, Tukey Honest Significant Differences significantly different between Buraco Beach and
(HSD) tests were used for site-wise comparison. Camarauaçu (Analysis of variance and subsequent
Tukey HSD test, p = 0.04. However, differences

123
 Author's personal copy
350 Plant Ecol (2018) 219:343–357

Fig. 4 Relative dominance based on stem coverage at the four study sites

between layers and layer–site interaction were not Buraco Beach and Apeu-Salvador showed widely
significant; ANOVA: p = 0.2 and p = 0.6, overlapping confidence envelopes; however, the cor-
respectively). responding curves for Bonifácio and Camarauaçu
Organic matter content (Fig. 2) was generally levelled off at a much lower number of species and
higher within the subsurface depth range (ANOVA were less steep in the initial section (Fig. 3). Species
of Box–Cox-transformed data with k = - 0.5: seen during plot sampling accounted for 87, 84, 78 and
p \ 0.0001); Camarauaçu samples had a significantly 71% of the least conservative estimate for total
lower organic matter content that those from the other (seen ? unseen) species richness at Buraco Beach,
sites (Tukey HSD with p = 0.004 and p = 0.02 for Bonifácio, Camarauaçu and Apeu-Salvador, respec-
comparison with Buraco Beach and Apeu-Salvador, tively. 15% of the species occurred at all three sites,
respectively). 67% at at least two sites. If species encountered
Pore water salinity was zero in all tested samples. outside of plots were considered, the respective
percentages were 23 and 70%. Including these addi-
Forest structure and floristic inventory tional species, species counts raised to 28, 25, 18 and
26 for Buraco Beach, Bonifácio, Camarauaçu and
In the plots, we found 33 tree species from 28 genera/ Apeu-Salvador, respectively (Table 3). Including the
21 families (Table 3). When species observed outside species seen outside of plots, 7 species occurred
plots were included, the total number of species raises exclusively at Buraco Beach, versus 4, 0 and 2
to 42. Total species richness in plots was lower at exclusive species at Bonifácio, Camarauaçu and
Bonifácio and Camarauaçu (15 and 14 species, Apeu-Salvador, respectively. Among the 109 species
respectively) than at Buraco Beach (20 species) and of the Salinas do Roque site (Santos et al. 2014 and
Apeu-Salvador (22 species). Species–area curves for own, unpublished data; Online Supplement 1) are 62%

123
 Author's personal copy
Plant Ecol (2018) 219:343–357 351

Fig. 5 Rényi diversity profiles for the four study sites (a: scale parameter). The dotted lines indicate extremes and median of the
dataset, open circles show the profile for the respective site

of the species encountered at the dune sites (Table 2). principally between Bonifácio and Apeu-Salvador
Clusters of stemless Astrocaryum palms were com- plots. Usage of Jaccard and Raup–Crick distance
mon at all sites (present in 47.8, 54.5 and 65.4% of matrices on presence–absence data (stress 0.20 and
plots for Buraco beach, Camarauaçu and Apeu- 0.21, respectively) led to a clearer picture; however,
Salvador, respectively) and occurred with high cluster Bonifácio and Apeu-Salvador plots did still not
densities at Camarauaçu and Apeu-Salvador (250, 561 separate neatly (Fig. 6). PERMANOVA indicated
and 606 clusters ha-1 at Buraco Beach, Camarauaçu significant effects (p \ 0.001) of distance to the
and Apeu, respectively). The set of species with high mainland, size of the dune ridge area and the
relative dominance varied distinctly among sites respective interaction. NMDS analysis of ‘‘aggre-
(Fig. 4). gated’’ plots including the Salinas do Roque site
Fisher’s a was 4.2, 3.1, 3.2 and 4.7 for Buraco showed that spread among plots from the tall paleo-
Beach, Bonifácio, Camarauaçu and Apeu-Salvador, dune forest was much larger than among most of the
respectively. Corresponding values for Pielou’s J’ dune site plots. However, most plots from Camarauaçu
were 0.77, 0.80, 0.70 and 0.76. Comparison of Rényi differed in their distribution along the second NMDS
profiles showed (Fig. 5) that Buraco Beach, Bonifácio axis from those of the remaining dune sites (Fig. 6).
and Apeu-Salvador could not be readily distinguished The null model approach according to Chase et al.
from each other in terms of diversity, while Cama- (2011) indicated that, on average, plots from the dune
rauaçu maintained the lowest values over the whole sites shared more species than expected when pro-
range of the profile. Species turnover within site cesses governing species distribution are random
(multivariate dispersion) was lower at Buraco Beach (modified Raup–Crick metric is on average negative;
regardless of the beta diversity parameter chosen Fig. 7). However, in comparison to the Salinas do
(Tukey HSD: pbw \ 0.001, pbw \ 0.001 and Roque reference site, Camarauaçu and Apeu-Salvador
pbw = 0.008 for differences to Bonifácio, Cama- tended to share a number of species so that the
rauaçu and Apeu-Salvador, respectively; metric was, on average, about 0 (expectation under
pbz \ 0.001, pbz \ 0.001 and pbz = 0.003 for differ- random conditions), while Buraco Beach shared less
ences to Bonifácio, Camarauaçu and Apeu-Salvador, species with Salinas do Roque than expected (mod-
respectively). ified Raup–Crick metric positive; Fig. 7). The same
Species turnover among sites was highest between was true for Bonifácio; however, the distribution of the
Buraco Beach and Camarauaçu (bw 0.59, bz 0.67) and Raup–Crick metric had a long tail towards negative
lowest between Bonifácio and Apeu-Salvador values (Fig. 7); i.e. a number of pairs of plots from
(bw 0.33, bz 0.42); the same order was maintained if both sites shared more species than expected.
stemless Astrocaryum clusters were considered (data
not shown).
Non-metric multidimensional scaling (NMDS) Discussion
based on species abundances with a quantitative
Jaccard distance matrix (stress 0.22) did show site- While species lists and descriptions of vegetation
wise grouping, however with considerable overlap types of coastal dune areas in the Amazon have been

123
 Author's personal copy
352 Plant Ecol (2018) 219:343–357

Fig. 6 NMDS ordination plots based on quantitative Jaccard, based on dune forest data only, the last is based on ‘‘aggregated’’
Jaccard and Raup–Crick distance matrices with basal area and plots (see text) and additional data from the Salinas do Roque
presence–absence data, respectively. The first three plots are site

available for some time (Amaral et al. 2008; Silva consistent with those reported by Souza-Filho et al.
et al. 2010; Serra et al. 2016), our dataset is the first to (2009) for Bonifácio and surrounding areas. Souza-
provide information on diversity and species turnover Filho et al. (2009) dated the corresponding barrier
as well as baseline data on corresponding soil island with 2800 cal. years BP. Considering sedimen-
conditions. tological data and geomorphological similarities, it
In spite of some minor differences, study sites were seems likely that the remaining sites and the barrier
very similar in terms of grain size and grain size island around Bonifácio were formed by equivalent
distribution. Specific geological surveys for our study geological processes and at similar times.
sites (except Bonifácio and Salinas do Roque) are The organic matter content of the upper soil layer at
currently not available. However, our data are our sites showed considerable variation both within

123
 Author's personal copy
Plant Ecol (2018) 219:343–357 353

Fig. 7 Boxplots showing the distribution of the modified indicated higher, positive values lower number of shared species
Raup–Crick metric for comparisons between pairs of plots from than expected by random chance (0)
the four dune sites and Salinas do Roque. Negative values

and among sites. A significant portion of the subsur- layers at least at the margins of the dune (Santos et al.
face organic matter is certainly derived from decom- 2014).
posing leaf litter and root material and will therefore Sandy substrates have low water and nutrient
expose micro-scale variations due to differences in the retention abilities, and short periods without precip-
vegetation and its growth dynamics in the past. itation may lead to arid conditions on dunes (Kutiel
However, we currently do not have a sound explana- et al. 2016); due to the high temperatures in combi-
tion for the consistently low organic matter content of nation with a seasonal rainfall regime at the Amazon
subsurface soils at Camarauaçu. Overall tree density at coast (Fisch et al. 1998; Moraes et al. 2005), the
Camarauaçu was distinctly lower than at the other sites coastal dune areas are certainly stressful environments
(Table 3) while basal area was rather high due to the for tree species. It is therefore expected that the more
presence of trees with large diameters, especially demanding or sensitive species from the regional
Anacardium occidentale L. It might be that the mainland species pool suffer exclusion by environ-
corresponding canopy is less productive. Similarly, mental filtering. Given the harsh but then relatively
the importance value of palms (A. vulgare) at this site homogeneous environmental conditions we encoun-
is high, and it might be that the decay of the big leaves tered at the study sites, we do not, however, consider
of the usually clumped palms leads to a more patchy environmental filtering the most convincing reason for
distribution of organic matter in the soil. In this case, species turnover within the pool of hardy species
sampling effort may have been insufficient to give a found at our dune sites. For example, more than 100
correct picture of the site’s organic matter content. tree species were found at the Salinas do Roque
However, such explanations are conjectural in the paleodune site, situated on a separate barrier island
absence of data on organic matter production, decom- landward of Bonifácio (Santos et al. 2014 and own,
position and storage. At any rate, we feel reluctant to unpublished data). This geologically similar but much
relate the differences in richness and diversity between older (5900 cal. years BP; Souza-Filho et al. 2009)
Camarauaçu and the remaining sites (see below) with barrier island features patches of anthropogenic
the observed differences in organic matter content, in archaeological black earth soils (Silveira et al. 2011;
spite of the great importance of organic matter for the Santos et al. 2014). The community at this site
maintenance of nutrient levels and water balance in includes 62% of the species encountered at the four
soils (Blume et al. 2016). dune sites but the dune species account for \ 25% of
The absence of salt in dune ridge soil samples is not the Salinas dos Roque species pool. Many of the
surprising, as the ridges are not exposed to tidal species of this more complex tree community are
inundation and are largely sheltered from salt spray by probably excluded by the less favourable conditions
the surrounding tall mangrove forests and, generally, a (e.g. less mature soils, reduced substrate stability) at
belt of herbaceous vegetation. A saline water the dune sites (see also discussion on null model tests
table may, however, be present in deeper sediment with the modified Raup–Crick metric below).

123
 Author's personal copy
354 Plant Ecol (2018) 219:343–357

Many of the tree species we encountered are in the way dune ridge area and distance to the
generalists with a wide distributional range or species mainland affected communities. Buraco Beach is by
commonly found in seasonally dry vegetation (Flora far the smallest of the three sites; however, the
Brasil/2020 2017). Only Coccoloba ramosissima mainland is only about half as distant as at the other
Wedd. is apparently restricted to coastal restinga sites, enhancing the chances of colonisation by new
forests in Brazil (Flora Brasil/2020 2017). Rarity or species by long distance dispersal. Therefore, disper-
absence of endemics in tropical dune forests has been sal from the mainland may play a major role in shaping
observed elsewhere (Scarano 2002; Silva et al. 2015). the composition of dune forest tree communities.
Scarano (2002) attributes the reduced number of Approximately 70% of the species we recorded have
endemic species to lack of time for speciation due to relatively small, fleshy fruits probably dispersed by
the relatively young age of the sandy coastal plain in birds or bats (Wehncke et al. 2004; Francisco et al.
south-east Brazil; the same may be true for the 2007; Gomes et al. 2008; Silva and Pedroni 2014;
Amazon region with its even younger dune Mendoza et al. 2015; Camargo et al. 2016) that could
environments. cross the surrounding mangrove forests. All species
Many species at our study sites belong to the Rosid found exclusively at Buraco Beach share this dispersal
clade (APG IV 2016); within this clade, the order syndrome. Only two species are wind-dispersed
Malpighiales sensu APG IV (2016) is with nine (Lepidaploa grisea (Baker) H.Rob., H. articulatus).
species especially well represented. Species of this None of the species possesses obviously hydrochoric
large order are ubiquitous in the region. We do not fruits but dispersal via fresh and seawater has been
provide here information on phylogenetic diversity documented, e.g. for Chrysobalanus icaco L. (White
(Webb et al. 2002) due to the lack of a reliable 1983). Interesting is the universal presence of A. vul-
reference dataset for null model tests. The low gare. The medium-sized, non-buoyant fruits and seeds
percentage of arboreus Fabaceae may, however, be a of this palm are relatively heavy; only short-range
characteristic feature of the tree communities of our dispersal by big rodents and larger mammals has been
coastal study sites. More interesting in respect to observed in similar species (Brewer and Rejmánek
environmental filtering is the fact that most genera in 1999; Henry et al. 2000). However, for this and other
our dataset are represented by just one species (rarely species with edible fruits, dispersal by man may play a
two); however, there was no evidence of mutual role; human settlements along the Amazon coastline
exclusion among the few congeneric species. date back to [ 2000 years (Silveira et al. 2011).
The tree communities at the four sites shared a Apart from the narrowly oligarchic community at
considerable number of species; ordination did, how- Camarauaçu, a second aspect of the observed domi-
ever, indicate that sites featured similar but distinct nance pattern calls attention. As an example, only
communities. At Buraco Beach, we found the highest three species at each site reached a relative domi-
number of exclusive species; species turnover within nance [ 10%; of these, only two (A. occidentale,
site was low. On the other hand, evenness at Buraco A. vulgare) maintained this level at more than one site.
Beach was intermediate compared to Bonifácio and While dominant species belonged in part to the pool of
Apeu-Salvador; diversity and richness (with and species shared between all three sites and most of them
without species seen outside of plots) were similar. were shared among at least two sites, the actual set of
The low diversity at Camarauaçu resulted from a dominant taxa at each site was a unique combination
comparatively low richness in combination with the (‘‘regional evenness’’ sensu Hillebrand et al. 2008). It
occurrence of three species with unusually high is likely that this combination is the result of stochastic
importance values (A. occidentale with almost 50% events, especially affecting long distance dispersal.
relative dominance; Himatanthus articulatus (Vahl) For example, the normal dispersal range of Himatan-
Woodson; A. vulgare). Remarkable is the absence of thus is apparently small (Baldauf et al. 2014). H. ar-
species of the Myrtaceae family at this site; Myrtaceae ticulatus is a common element of secondary woody
are usually typical components of coastal dune forests vegetation on the mainland and is frequently found in
in Brazil (Scherer et al. 2005). Unexpectedly, richness the Salinas do Roque palaeodune forest (Santos et al.
and diversity did not correlate positively with dune 2014). It does, however, not occur at Bonifácio, at
area extension and there was a significant interaction about 10 km distance, in spite of its obvious ability to

123
 Author's personal copy
Plant Ecol (2018) 219:343–357 355

thrive and become dominant in very similar coastal (Coordenação de Aperfeiçoamento de Pessoal de Nı́vel
dune forests at Camarauaçu. We therefore believe that Superior, CAPES) under the code 3290/2013. The authors are
grateful to Nils E. Asp Neto for helpful comments on the
accidental, non-standard dispersal mechanisms (Hig- interpretation of soil analyses. Luciana O. Santos and Leiliane O.
gins et al. 2003) may be responsible for divergent Santos Rocha granted access to Salinas do Roque forest structure
species composition and dominance patterns at our data. We further wish to thank Colin R. Beasley and two
study sites. anonymous reviewers for comments on a previous version of the
manuscript. Marlana C. Q. da Silva, Vando Gomes, Débora P.
The modified Raup–Crick metric (Chase et al. O. Lima, Vitor A. N. Bragança, Marcelo V. S. Pereira and
2011) provided additional insight regarding the pro- Amanda R. Silva provided valuable support during field and
cesses shaping the communities of the Amazon coast laboratory work.
dune forests. When compared within the joint dataset
including the Salinas do Roque site, the four dune
forest sites share more species than expected when References
random processes shape communities. This can be
Almeida EB, Santos-Filho FS, Araújo EL, Zickel CS (2011)
interpreted as an effect of shared environmental filters
Structural characterization of the woody plants in restinga
(Chase et al. 2011), consistent with our observation of Brazil. J Ecol Nat Environ 3(3):95–103
that the dune forest species belong to a subset of the Amaral DD, Prost MT, Bastos MNC, Neto SVC, Santos JUM
regional species pool that are able to tolerate the dune (2008) Restingas do litoral amazônico, estados do Pará e
Amapá, Brasil. Boletim do Museu Paraense Emilio Goeldi,
environment. On the other hand, relations between sér. Botânica 3(1):35–67
plots from the dune forest and from Salinas do Roque Anderson MJ, Ellingsen KE, McArdle BH (2006) Multivariate
vary. The clearly positive Raup–Crick metric for the dispersion as a measure of beta diversity. Ecol Lett
comparison Buraco Beach/Salinas do Roque indicates 9:683–693. https://doi.org/10.1111/j.1461-0248.2006.
00926.x
that there are less shared species than expected; this is Apg IV (2016) An update of the Angiosperm Phylogeny Group
probably the consequence of spatial distance between classification for the orders and families of flowering
both sites. Salinas do Roque is probably is not a good plants: APG IV. Bot J Linn Soc 181:1–20
proxy for the mainland species pool influencing this Araujo D, Pereira M (2004) Sandy coastal vegetation. Ency-
clopedia of life support systems. UNESCOEolss Publish-
site. On the other hand, the surprising tendency ers. http://www.eolss.net/ebooks/Sample%20Chapters/
towards positive Raup–Crick metrics with plot com- C20/E6-142-TB-08.pdf. Accessed 1 June 2017
parisons between Salinas do Roque and neighbouring Baldauf C, Ciampi-Guillardi M, Aguirra TJ, Corre CE, Santos
Bonifácio and the wide span encountered for compar- FAM, Souza AP, Sebbenn AM (2014) Genetic diversity,
spatial genetic structure and realised seed and pollen dis-
isons between Camarauaçu and Salinas do Roque may persal of Himatanthus drasticus (Apocynaceae) in the
be caused by ‘‘atypical’’ plots (openings in the canopy, Brazilian savanna. Conserv Genet 15:1073–1083. https://
forest border) in the Salinas do Roque area where doi.org/10.1007/s10592-014-0600-5
conditions favour locally uncommon species which Barbosa MN, Ceballos DC (1995) The Brazilian observation
satellites. Acta Astronaut 37:503–512
became dominant in the dune areas. Blume H-P, Brümmer GW, Fleige H, Horn R, Kandeler E,
In conclusion, we observed that the tree communities Kögel-Knabner I, Kretzschmar R, Stahr K, Wilke B-M
of coastal dune forests embedded within the great (2016) Soil organic matter. In: Scheffer/Schachtschabel
mangrove forest belt of the Amazon coast are composed soil science. Springer, Heidelberg
Brewer SW, Rejmánek M (1999) Small rodents as significant
of mainland generalist and dry area species that are able dispersers of tree seeds in a Neotropical forest. J Veg Sci
to survive the harsher conditions on sandy soils next to 10:165–174
the sea. While many of the occurring species seem Brunbjerg AK, Cavender-Bares J, Eiserhardt WL, Ejrnæs R,
reasonably well adapted to long distance dispersal, Aarssen LW, Buckley HL, Forey E, Jansen F, Kattge J,
Lane C, Lubke RA, Moles AT, Monserrat AL, Peet RK,
stochastic events may play an important role in deter- Roncal J, Wootton L, Svenning JC (2014) Multi-scale
mining community composition and dominance patterns. phylogenetic structure in coastal dune plant communities
across the globe. J Plant Ecol 7(2):101–114. https://doi.org/
Acknowledgements Part of the expenses of the field surveys 10.1093/jpe/rtt069
was financed by the project Fluxos (Água, Sedimentos, Camargo PHSA, Martins MM, Feitosa RM, Christianini AV
Nutrientes e Plâncton) Amazônicos ao longo do Continuum (2016) Bird and ant synergy increases the seed dispersal
Rio-Estuário-Costa e Implicações para a Biodiversidade effectiveness of an ornithochoric shrub. Oecologia
Vegetal Costeira Amazônica financed by the Brazilian Agency 181(2):507–518. https://doi.org/10.1007/s00442-016-
for Coordination of Capacitating Advanced-degree Personnel 3571-z

123
 Author's personal copy
356 Plant Ecol (2018) 219:343–357

Chase JM, Myers JA (2011) Disentangling the importance of Higgins SI, Nathan R, Cain ML (2003) Are long-distance dis-
ecological niches from stochastic processes across scales. persal events in plants usually caused by nonstandard
Philos Trans R Soc B 366:2351–2363. https://doi.org/10. means of dispersal? Ecology 84(8):1945–1956
1098/rstb.2011.0063 Hillebrand H, Bennett DM, Cadotte MW (2008) Consequences
Chase JM, Kraft NJB, Smith KG, Vellend M, Inouye B (2011) of dominance: a review of evenness effects on local and
Using null models to disentangle variation in community regional ecosystem processes. Ecology 89(6):1510–1520
dissimilarity from variation in a-diversity. Ecosphere INMET (2014) Banco de dados Meteorológicos para Ensino e
2(2):1–11. https://doi.org/10.1890/ES10-00117.1 Pesquisa (BDMEP), Rede INMET. Instituto Nacional de
Colwell RK, Chao A, Gotelli NJ, Lin S-Y, Mao CX, Chazdon Meteorologia. http://www.inmet.gov.br/projetos/rede/
RL, Longino JT (2012) Models and estimators linking pesquisa. Accessed 2 Sep 2014
individual-based and sample-based rarefaction, extrapola- Kim D, Yu KB (2009) A conceptual model of coastal dune
tion and comparison of assemblages. J Plant Ecol ecology synthesizing spatial gradients of vegetation, soil,
5(1):3–21 and geomorphology. Plant Ecol 202:135–148. https://doi.
Condit R, Pitman N, Leigh EG Jr, Chave J, Terborgh J, Foster org/10.1007/s11258-008-9456-4
RB, Núnez P, Aguilar S, Valencia R, Villa G, Muller- Kindt R, Damme PV, Simons AJ (2006) Tree diversity in
Landau HC, Losos E, Hubbell SP (2002) Beta-diversity in western Kenya: using profiles to characterise richness and
tropical forest trees. Science 295(5555):666–669. https:// evenness. Biodivers Conserv 15:1253–1270. https://doi.
doi.org/10.1126/science.1066854 org/10.1007/s10531-005-0772-x
De Cáceres M, Legendre P, Valencia R, Cao M, Chang LW, Koleff P, Gaston KJ, Lennon JJ (2003) Measuring beta diversity
Chuyong G, Condit R, Hao Z, Hsieh CF, Hubbell S, for presence-absence data. J Anim Ecol 72:367–382
Kenfack D, Ma K, Mi X, Noor MNS, Kassim AR, Ren H, Kutiel PB, Katz O, Ziso-Cohen V, Divinsky I, Katra I (2016)
Su SH, Sun IF, Thomas D, Ye W, He F (2012) The vari- Water availability in sand dunes and its implications for the
ation of tree beta diversity across a global network of forest distribution of Artemisia monosperma. CATENA
plots. Glob Ecol Biogeogr 21:1191–1202. https://doi.org/ 137:144–151. https://doi.org/10.1016/j.catena.2015.09.
10.1111/j.1466-8238.2012.00770.x 007
El-Demerdash MA, Hegazyt AK, Zilay AM (1995) Vegetation- Leibold MA, Holyoak M, Mouquet N, Amarasekare P, Chase
soil relationships in Tihamah coastal plains of Jazan JM, Hoopes MF, Holt RD, Shurin JB, Law R, Tilman D,
region, Saudi Arabia. J Arid Environ 30:161–174 Loreau M, Gonzalez A (2004) The metacommunity con-
Ellenberg H (2009) Vegetation ecology of Central Europe. cept: a framework for multi-scale community ecology.
Cambridge University Press, Cambridge Ecol Lett 7(7):601–613. https://doi.org/10.1111/j.1461-
Fisch G, Marengo JA, Nobre CA (1998) Uma revisão geral 0248.2004.00608.x
sobre o clima da Amazônia. Acta Amazonica Lima RAF, Oliveira AA, Martini AMZ, Sampaio D, Souza VC,
28(2):101–126 Rodrigues RR (2011) Structure, diversity, and spatial pat-
Flora Brasil/2020 (2017) Flora do Brasil 2020 em construção. terns in a permanent plot of a high restinga forest in
Jardim Botânico do Rio de Janeiro. http://floradobrasil.jbrj. Southeastern Brazil. Acta Botanica Brasilica
gov.br/reflora/floradobrasil/FB10913. Accessed 21 June 25(3):633–645
2017 Mehlig U, Menezes MPM, Reise A, Schories D, Medina E
Folk RL, Ward WC (1957) Brazos River bar: a study in the (2010) Mangrove vegetation of the Caeté Estuary. In:
significance of grain-size parameters. J Sediment Petrol Saint-Paul U, Schneider H (eds) Mangrove dynamics and
27(1):3–26. https://doi.org/10.1306/74d70646-2b21-11d7- management in North Brazil. Springer, Heidelberg
8648000102c1865d Melo JCF Jr, Boeger MRT (2015) Riqueza, estrutura e inter-
Fox J, Weisberg S (2011) An R Companion to applied regres- ações edáficas em um gradiente de restinga do Parque
sion. Sage, Thousand Oaks, CA Estadual do Acaraı́, Estado de Santa Catarina, Brasil.
Francisco MR, Lunardi VO, Galetti M (2007) Bird attributes, Hoehnea 42(2):207–232
plant characteristics, and seed dispersal of Pera glabrata Mendoza I, Martin G, Caubère A, Châtelet P, Hardy I, Jouard S,
(Schott, 1858), (Euphorbiaceae) in a disturbed cerrado Forget P-M (2015) Does masting result in frugivore sati-
area. Braz J Biol 67(4):627–634 ation? A test with Manilkara trees in French Guiana. J Trop
Garzon-Lopez CX, Jansen PA, Bohlman SA, Ordonez A, Olff H Ecol 31:553–556. https://doi.org/10.1017/
(2014) Effects of sampling scale on patterns of habitat S0266467415000425
association in tropical trees. J Veg Sci 25:349–362. https:// Menezes MPM, Berger U, Mehlig U (2008) Mangrove vegeta-
doi.org/10.1111/jvs.12090 tion in Amazonia: a review of studies from the coast of Pará
Gilbert ER, de Camargo MG, Sandrini-Neto L (2014) rysgran: and Maranhão States, north Brazil. Acta Amazonica
grain size analysis, textural classifications and distribution 38(3):403–420
of unconsolidated sediments. https://CRAN.R-project.org/ Moraes BC, Costa JMN, Costa ACL, Costa MH (2005) Variação
package=rysgran. Accessed 1 June 2017 espacial e temporal da precipitação no estado do Pará. Acta
Gomes VSdM, Loiselle BA, Alves MAS (2008) Birds foraging Amazonica 35(2):2017–2214
for fruits and insects in shrubby restinga vegetation, Müller-Dombois D, Ellenberg H (1974) Aims and methods of
southeastern Brazil. Biota Neotrop 8(4):21–31 vegetation ecology. The Blackburn Press, New Jersey
Henry O, Feer F, Sabatier D (2000) Diet of the Lowland Tapir Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P,
(Tapirus terrestris L.) in French Guiana. Biotropica McGlinn D, Minchin PR, O’Hara RB, Simpson GL,
32(2):364–368 Solymos P, Stevens MHH, Szoecs E, Wagner H (2017)

123
 Author's personal copy
Plant Ecol (2018) 219:343–357 357

vegan: community ecology package. https://CRAN.R- ecological drift in the heath vegetation of South America.
project.org/package=vegan. Accessed 1 June 2017 Ecosphere. https://doi.org/10.1890/es14-00548.1
R Core Team (2017) R: a language and environment for sta- Silveira MI, Oliveira ER, Kern DC, Costa ML, Rodrigues SFS
tistical computing. R Foundation for Statistical Comput- (2011) O sı́tio Jabuti, em Bragança, Pará, no cenário
ing, Vienna, Austria. https://www.R-project.org/ arqueológico do litoral amazônico. Boletim do Museu
Santos LO, Santos LO, Menezes MPM, Beasley CR, Mehlig U Paraense Emilio Goeldi, sér. Ciências Humanas
(2014) Composition and structure of a diverse tree com- 6(2):335–345
munity at the edges of a Brazilian Amazon rainforest island Souza-Filho PWM (2005) Costa de manguezais de macromaré
surrounded by marshes and mangroves. Plant Ecol da Amazônia: cenários morfológicos, mapeamento e
215(12):1469–1481. https://doi.org/10.1007/s11258-014- quantificação de áreas usando dados de sensores remotos.
0407-y Revista Brasileira de Geofı́sica 23(4):427–435
Scarano FR (2002) Structure, function and floristic relationships Souza-Filho PWM, Lessa GC, Cohen MCL, Costa FR, Lara RJ
of plant communities in stressful habitats marginal to the (2009) The subsiding macrotidal barrier estuarine system
Brazilian Atlantic rainforest. Ann Bot 90:517–524. https:// of the eastern Amazon coast, Northern Brazil. In: Dillen-
doi.org/10.1093/aob/mcf189 burg SR, Hesp PA (eds) Geology and geomorphology of
Scherer A, Maraschin-Silva F, Baptista LRM (2005) Florı́stica e holocene coastal barriers of Brazil. Springer, Heidelberg
estrutura do componente arbóreo de matas de Restinga Thiers B (continuously updated) Index herbariorum: a global
arenosa no Parque Estadual de Itapuã, RS, Brasil. Acta directory of public herbaria and associated staff. New York
Botanica Brasilica 19(4):717–726 Botanical Garden’s Virtual Herbarium. http://sweetgum.
Schwendenmann L, Riecke R, Lara RJ (2006) Solute dynamics nybg.org/ih/. Accessed 28 June 2017
in a North Brazilian mangrove: the influence of sediment Webb CO, Ackerly DD, McPeek MA, Donoghue MJ (2002)
permeability and freshwater input. Wetlands Ecol Manage Phylogenies and community ecology. Ann Rev Ecol Syst
14:463–475 33:475–505. https://doi.org/10.1146/annurev.ecolsys.33.
Serra FCV, Lima PB, Almeida EB (2016) Species richness in 010802.150448
restinga vegetation on the eastern Maranhão State, North- Wehncke EV, Valdez CN, Domı́nguez CA (2004) Seed dis-
eastern Brazil. Acta Amazonica 46(3):271–280. https:// persal and defecation patterns of Cebus capucinus and
doi.org/10.1590/1809-4392201504704 Alouatta palliata: consequences for seed dispersal effec-
Silva GBMd, Pedroni F (2014) Frugivoria por aves em área de tiveness. J Trop Ecol 20:535–543. https://doi.org/10.1017/
cerrado no municı́pio de Uberlândia, Minas Gerais. Revista S0266467404001865
Árvore 38(3):433–442 Wentworth CK (1922) A scale of grade and class terms for
Silva RM, Mehlig U, Santos JUM, Menezes MPM (2010) The clastic sediments. J Geol 30(5):377–392. https://doi.org/
coastal restinga vegetation of Pará, Brazilian Amazon: a 10.1086/622910
synthesis. Revista Brasileira de Botanica 33(4):563–573 White F (1983) Long distance dispersal, overland migration and
Silva JLA, Souza AF, Jardim JG, Goto BT (2015) Community extinction in the shaping of tropical African floras.
assembly in harsh environments: the prevalence of Bothalia 14(3/4):395–403

123