Nativa, Sinop, v. 10, n. 3, p. 333-340, 2022.

Pesquisas Agrárias e Ambientais

DOI: https://doi.org/10.31413/nativa.v10i3.13177 ISSN: 2318-7670

Forest fragment grouping analysis under selective logging in Amazonian biome

Karen Janones da ROCHA1*, Édila Cristina de SOUZA2, Cyro Matheus Cometti FAVALESSA2,

Sidney Fernando CALDEIRA2, Silvo Alves RODRIGUES, Gilvano Ebling BRONDANI4*

1Federal University of Rondonia, Rolim de Moura, RO, Brazil.

2Federal University of Mato Grosso, Cuiabá, MT, Brazil.

4Federal University of Lavras, Lavras, MG, Brazil.

E-mail: karenrocha@unir.br; gilvano.brondani@ufla.br

ORCID: (0000-0002-2165-3081; 0000-0001-5528-8804; 0000-0002-6630-1979;

0000-0001-6042-4313; 0000-0001-9734-4907; 0000-0001-8640-5719)

Submitted on 2021/11/19; Accepted on 2022/07/18; Published on 2022/09/16.

ABSTRACT: The aim of this study was to analyse the existence of floristic groups in Seasonal Semi-deciduous

Forest fragments under the effects of selective logging located in Tapurah-MT. The given clusters were

allocated and measured, and each one had five subunits of 500 m². For each sampling unit, all of the arboreal

and shrub species with a diameter at breast height (DBH) equal to or greater than 10 cm were considered. In

the sampling subunits studied, the vegetation matrix was composed of the density of the 20 species in the

fragment with the greatest importance. The presence of floristic groups was verified by a grouping analysis

through the species association method. Euclidean distance with Hellinger’s transformation was used as a

similarity measure between the groupings, and Ward’s linking method was used for the dendrogram elaboration.

The number of groups was established by Kendall’s coefficient, and correlations between the groups and the

species were assessed. Once the concordance was determined, the number of groups in which the species were

correlated was chosen. The species’ auto-ecological characteristics, mainly the propagating material dispersion

type and average DBH, were the main factors responsible for the species association and similarity inside each

floristic group.

Keywords: similarity between species groups; species association; floristic group; environmental disturbance.

Análise de agrupamento de fragmento florestal sob efeitos da extração seletiva no

bioma amazônico

RESUMO: O objetivo deste estudo foi analisar a existência de grupos florísticos em fragmentos de Floresta

Estacional Semidecidual sob os efeitos da exploração madeireira seletiva localizada em Tapurah-MT. Os

transectos foram subdivididos em cinco subunidades de 500 m². Para cada unidade amostral, foram

consideradas todas as árvores com DAP maior ou igual a 10 cm. Nas subunidades amostrais estudadas, a matriz

de vegetação foi composta pela densidade das 20 espécies de maior importância no fragmento. A presença de

grupos florísticos foi verificada por meio de uma análise de agrupamento pelo método de associação de

espécies. A distância euclidiana com a transformação de Hellinger foi usada como uma medida de similaridade

entre os agrupamentos, e o método de ligação de Ward foi usado para a elaboração do dendrograma. O número

de grupos foi estabelecido pelo coeficiente de Kendall e as correlações entre os grupos e as espécies foram

avaliadas. Uma vez determinada a concordância, optou-se pelo número de grupos em que as espécies estavam

correlacionadas. Foi verificado que as características auto-ecológicas das espécies, principalmente o tipo de

dispersão do material propagado e o DAP médio, foram os principais responsáveis pela associação e

similaridade das espécies dentro de cada grupo florístico.

Palavras-chave: similaridade entre grupos florísticos; associação de espécies; agrupamentos florísticos;

perturbação ambiental.

1. INTRODUCTION

Although there are laws that authorize logging in specific

areas, illegal logging is widespread in Brazil and in several

Amazonian countries. And despite selective logging only

targeting trees with commercial value, when not carried out

through instruments such as the Sustainable Forest

Management Plan (Plano de Manejo Florestal Sustentável –

PMFS), the logging methods cause environmental damage.

Areas that were the object of selective logging are more

likely to be occupied by new residents and to suffer clear cuts

for the cultivation of pasture or grain, promoting the process

of forest fragmentation. And the consequences are serious,

fragmentation affects the organization of natural

communities, reduces biodiversity, increases the risk of

extinction of wild animals and compromises the ecological

services provided by the forest. Furthermore, the mere

presence of these fragments does not guarantee their

maintenance (ROCHA et al., 2017).

Biological diversity conservation strategies require studies

that quantify the existing species and their distribution in the

environment as well as provide knowledge on the

relationships between floristic composition and ecosystem

3Independent consultant in Forestry, Cuiabá, MT, Brazil.

Forest fragment grouping analysis under selective logging in Amazonian biome

Nativa, Sinop, v. 10, n. 3, p. 333-340, 2022.

334

diversity (PRIMACK; RODRIGUES, 2001). Conservation

initiatives, management and restoration of forest fragments

require detailed studies, such as flora inventories and the

ecology of plant communities (FARAH et al., 2017).

Grouping analysis, which is frequently used in scientific

projects, is a useful tool for understanding natural forest by

helping to detect specific floristic associations (SÜHS;

BUDKE, 2011; CHAI; WANG, 2016). The search for

species associations is often based on the non-tested

suppositions that species have nonrandom patterns of

association due to environmental control or biotic

interactions (LEGENDRE; LEGENDRE, 2012).

Under the hypothesis of environmental control, once

associations are verified, it is possible to focus on finding the

mutual ecological demands of most or even all species of an

association instead of having to describe the biology and

habitat of each species individually. Associations can be

better predictors of the environmental quality than the

individual species because associations are less affected by

sampling errors (LEGENDRE, 2005; LEGENDRE;

LEGENDRE, 2012).

In this context, the aim of the present study was to

analyse the existence of floristic groups and to evaluate which

are the main factors responsible for the association and

similarity of species within each floristic group in a fragment

of the Seasonal Semi-deciduous Forest under the effects of

selective logging.

2. MATERIALS AND E METHODS

A floristic and structural survey on a 32.98 ha fragment

was conducted in the municipality of Tapurah, Mato Grosso,

Brazil (12°28'5.67" S; 56°33'32.14" W) (Figure 1). The

fragment is located in the field of the Submontane Seasonal

Semi-deciduous Forest from Amazon Biome (INPE, 2010)

and suffers the effects of selective logging that occurred in

the 1990s. The climate in the region is Am type according to

Köppen’s classification, with a short dry winter, high annual

rainfall of approximately 3,000 mm year-1, and an average

annual temperature of 25°C (ALVARES et al., 2013).

Figure 1. Allocation and scheme of sample units used for the study

of the Seasonal Semideciduous Forest fragment under the influence

of selective logging, Amazonian Biome, 2022.

Figura 1. Localização e esquema das unidades amostrais utilizadas

para o estudo do fragmento de Floresta Estacional Semidecidual sob

influência da extração seletiva de madeira, Bioma Amazônico, 2022.

For systematic sampling, the fixed area method was

applied to 10 m × 250 m clusters with five subunits of 10 m

× 50 m each, and a minimum border of 15 m was considered.

The allocation of the clusters respected the greatest variation

gradient of the forest, West-East direction due to proximity

to the river (1,010 m) and the slope of the terrain (13.6%)

(Figure 1). All living individuals with a diameter at breast

height (DBH) greater than or equal to 10 cm were measured.

Botanical material of the species that could not be identified

in the field by a parabotanist was collected for taxonomic

analysis and identification in the herbarium of the Federal

University of Mato Grosso (UFMT).

To update and confirm the nomenclature of species,

Brazil’s Flora Species List (BFG, 2022 – reflora.jbrj.gov.br)

was used. The delimitation of families followed the APG IV

classification system (THE ANGIOSPERM PHYLOGENY

GROUP, 2016).

The vegetation matrix was composed of the density of 20

species with the greatest importance (VI) in the fragment

(Table 1) in the 25 sampling subunits that were studied, more

details in Rocha (2015). Based on field observations and

literature reviews (CARVALHO, 2006; LORENZI, 2009;

PERES, 2016), the species were classified in ecological

groups and dispersion syndrome types. The succession

classification was based on the terminology by Gandolfi et al.

(1995), and the species were categorized as pioneers, early

secondary, late secondary and due to a lack of information,

non-classified (NC). The dispersion syndromes were based

on the terminology by Van der Pijl (1982) and included

anemochorous, zoochorous, autochorous and, due to a lack

of information, non-classified (NC).

When looking for species associations, Legendre (2005)

suggested transforming the species abundance to control for

differences in total abundance among places, producing more

monotonic correlations between species. Before the

concordance analysis among species, the data were

transformed using Hellinger’s transformation (Equation 1)

(LEGENDRE; GALLAGHER, 2001). The presence of

floristic groups was verified through grouping analysis with

the species association method (LEGENDRE;

LEGENDRE, 2012). Euclidean distance with Hellinger’s

transformation was used as a similarity measure between the

groupings (Equation 2).

𝑦′ =

∑



 (01)

𝐷(𝑦,𝑦) = ∑𝑦′ −𝑦′



  (02)

where: 𝑦′ = the abundance of 𝑗 species in the sampling 𝑥 subunit

transformed; 𝑦 = the abundance of 𝑗 species in the sampling 𝑥

subunit; 𝑦 = the abundance of 𝑗 species in the sampling 𝑥

subunit; 𝑘 = total of species in the sampling 𝑥 subunit.

The Ward’s linking method was used for the dendrogram

elaboration. Ward’s method is a procedure in which the

similarity measure used to unite groupings is calculated as the

sum of the squares between both groupings that present the

lowest increase in the global sum of the squares inside the

groupings (HAIR et al., 2009). According to Dutra et al.

(2004), Ward’s method forms groups based on the minimum

standard deviation among the data of each group.

The data were organized in a data matrix with 20

columns, representing the 20 species with greatest

importance values, and 25 lines, representing the

observations formed by the 25 sampling subunits. The

grouping analysis was processed through R software 2.2-0

(2014) version, mvpart package (DE’ATH, 2007).

Rocha et al.

Nativa, Sinop, v. 10, n. 3, p. 333-340, 2022.

335

Table 1. Species with the greatest importance found in the Seasonal Semideciduous Forest fragment under the influence of selective logging

in Amazonian Biome, with their number of individuals, basal area, phytosociological parameters, succession ecological group and

propagating material dispersion type.

Tabela 1. Espécies de maior importância encontradas no fragmento de Floresta Estacional Semidecidual sob influência do corte seletivo no

bioma Amazônia, com número de indivíduos, área basal, parâmetros fitossociológicos, grupo ecológico de sucessão e tipo de dispersão do

material de propagação.

Species

DoR

Disp.

Qualea paraensis

Ducke

8.4011

5.36

14.46

7.32

9.05

AUTO

Aspidosperma discolor

A.DC.

6.2954

3.83

10.84

4.36

6.34

ANE

Matayba arborescens

(Aubl.) Radlk.

3.2019

4.82

5.53

8.35

6.23

ZOO

Vochysia vismiifolia

Spruce ex Warm.

3.7899

4.08

6.52

6.79

5.80

AUTO

Pouteria guianensis

Aubl.

3.5305

3.83

6.08

5.23

5.04

ZOO

Pseudolmedia laevigata

Trécul

1.4975

3.57

2.58

3.48

3.21

ZOO

Xylopia

spp.

1.1054

3.83

1.90

3.83

3.19

ZOO

Nectandra cuspidata

Nees

0.9396

3.57

1.62

4.36

3.18

ZOO

Ocotea acutangula

(Miq

.) Mez

1.3459

2.55

2.32

3.31

2.73

ZOO

Y10

Tachigali vulgaris

L. G. Silva & H. C. Lima

0.9537

3.06

1.64

3.14

2.61

ANE

Y11

Sterigmapetalum obovatum

Kuhlm.

1.2410

2.81

2.14

2.26

2.40

Y12

Inga

spp.

1.5222

2.03

2.63

1.74

2.13

ZOO

Y13

Erisma uncinatum

Warm.

1.2546

2.30

2.16

1.92

2.12

ANE

Y14

Toulicia guianensis

Aubl.

0.8503

2.04

1.46

2.09

1.87

ZOO

Y15

Tapirira guianensis

Aubl.

0.8480

2.30

1.46

1.74

1.83

ZOO

Y16

Diplotropis purpurea

(Rich.) Amshoff

0.9957

1.79

1.71

1.92

1.81

ANE

Y17

Bellucia grossularioides

(L.) Triana

1.2088

1.53

2.08

1.74

1.78

ZOO

Y18

Licania blackii

Prance

0.9868

1.78

1.70

1.39

1.62

Y19

Helicostylis tomentosa

(Poepp. & Endl.) Rusby

0.8850

1.79

1.52

1.39

1.57

ZOO

Y20

Cheiloclinium cognatum

(Miers) A.C.Sm.

0.5248

2.03

0.91

1.57

1.50

ZOO

Where: Yi = variable that represents the species i; N = number of sampled individuals; G = basal area (m² per ha); FR = relative frequency (%); DoR =

relative dominance (%); DR = relative density (%); VI = importance value (%); EG = ecological group; P = pioneer; SI = early secondary; ST = late

secondary; NC = non-classified; Disp. = propagating material dispersion type; ZOO = zoochorous; ANE = anemochorous; AUTO = autochorous.

Onde: Yi = variável que representa a espécie i; N = número de indivíduos amostrados; G = área basal por hectare (m² por ha); FR = frequência relativa (%);

DoR = dominância relativa (%); DR = densidade relativa (%); VI = valor de importância (%); EG = grupo ecológico; P = pioneira; SI = sencundária inicial;

ST = secundária tardia; NC = não classificada; Disp. = tipo de dispersão de propágulos; ZOO = zoocórica; ANE = anemocórica; AUTO = autocórica.

Concordance analysis, which is based on Kendall’s

concordance coefficient, is used to outline groups of species

that form statistically significant associations. The number of

groupings was established using Kendall’s concordance

coefficient (W – Equation 3), together with the permutation

test – paired average of Spearman’s correlation, which was

suggested by Legendre (2005). Spearman’s correlations

between the groups and the species were analysed, and a

concordance global test was applied among the species of

each group. With a concordance occurrence, the number of

groups was chosen when the species were correlated.

𝑊 =()

 (03)

where: 𝑊 = Kendall’s concordance coefficient; 𝑝 = the number of

variables (judges), among which Spearman’s correlation coefficients

are calculated; 𝑟 = the paired average of Spearman’s correlation.

First, a general all-species independence test is

performed. If the null hypothesis is rejected, groups of

correlated species are searched and, within each group, the

contribution of each species to the global statistics is tested,

using the paired average of Spearman’s correlation. The

method aims at finding the most encompassing assemblages,

i.e. the smallest number of groups containing the largest

number of positively and significantly associated species

(BOCARD et al., 2011).

When two sets of species are specified, the default

correlation analysis includes descriptive statistics for each

specie (𝑟) and pairwise Pearson correlation statistics between

the two sets of variables. For a Spearman correlation, the

Fisher’s z transformation is used to derive its confidence

limits and a p-value under a specified null hypothesis.

The null hypothesis (H0) from Kendall’s concordance

test (W) states that all species are independent, whereas the

alternative hypothesis states that at least one of the species is

concordant with one or some of the other species. The test

was performed for four possibilities: with 3 groups, with 4

groups, with 5 groups and with 6 groups, which guarantees a

minimum of two species per group.

3. RESULTS

Through the Euclidean distance with the Hellinger

transformation, it is possible to verify the similarity between

species and, consequently, between clusters (Table 2). But in

general, the Spearman correlation between the species

studied was low, r ≤ |0.60| (Table 3).

Six floristic groups were determined by Kendall’s

concordance test (W). Although groups 1 and 2 accepted the

null hypothesis (H0) (Table 4), the groups did not have

species with the paired average of Spearman’s negative

correlation (Table 5).

An a posteriori analysis was calculated to determine

which individual species were concordant with one or some

of the other species of the group. By analysing Spearman’s

paired average of each species, six floristic groups were

established (Table 5).

The dendrogram elaborated by Ward’s linking method is

presented in Figure 2 in which the red dashed line represents

the cut-off point cutting for the determination of the number

of groups. None of the groups occurred in all sampling

subunits, and as a Ward’s method characteristic (HAIR et al.,

2009), they presented similar dimensions (Table 6).

Forest fragment grouping analysis under selective logging in Amazonian biome

Nativa, Sinop, v. 10, n. 3, p. 333-340, 2022.

336

Table 2. Euclidean distance with Hellinger’s transformation among the species with the greatest importance values in a Seasonal

Semideciduous Forest fragment under the influence of selective logging, Amazonian Biome, 2014. ns – not significant at the level of 5%

error probability; * – significant at the level of 5% error probability.

Tabela 2. Distância euclidiana com a transformação de Hellinger entre as espécies com maior valor de importância em um fragmento de

Floresta Estacional Semidecidual sob influência da exploração seletiva, Bioma Amazônico, 2014. ns – não significativo ao nível de 5% de

probabilidade de erro; * – significativo ao nível de 5% de probabilidade de erro.

Y10

Y11

Y12

Y13

Y14

Y15

Y16

Y17

Y18

Y19

1.294

0.856

0.987

1.356

0.945

1.080

1.104

1.011

0.531

1.173

1.011

0.942

1.164

0.640

1.116

1.224

0.989

0.921

0.996

0.795

1.097

0.847

1.176

0.803

1.100

0.946

1.127

1.155

1.340

1.073

0.848

0.511

0.933

0.767

0.629

0.972

Y10

1.137

0.887

1.598

0.995

1.358

0.953

0.973

0.849

1.034

Y11

0.932

1.095

1.004

1.481

1.027

1.480

0.924

1.012

1.192

0.941

Y12

0.782

1.308

1.071

0.984

1.488

1.074

0.842

1.167

1.094

0.981

1.108

Y13

0.853

0.868

0.950

1.235

0.815

0.955

1.270

1.297

1.509

1.324

1.217

1.028

Y14

1.052

1.108

0.818

1.124

0.818

0.945

1.123

0.726

0.767

1.154

1.068

1.323

1.273

Y15

1.019

1.184

1.183

1.397

1.097

1.201

0.888

0.864

1.134

0.826

0.564

0.729

1.327

0.880

Y16

1.321

0.708

1.171

1.030

0.900

1.065

1.406

1.187

1.368

0.995

0.901

1.473

0.743

1.180

1.113

Y17

1.290

1.099

1.067

1.003

1.144

0.843

0.975

1.186

1.118

1.265

0.880

1.069

0.925

1.050

0.965

0.631

Y18

0.761

1.372

0.822

0.966

0.797

0.779

0.968

0.665

0.898

1.316

1.020

0.968

0.998

1.203

1.151

1.198

0.844

Y19

1.061

1.019

1.460

1.204

1.060

1.469

1.107

1.073

1.242

0.860

0.943

1.094

0.678

0.914

1.111

0.932

1.119

1.367

Y20

1.092

0.671

0.662

1.046

0.772

0.997

1.183

1.082

1.274

1.414

1.015

1.226

0.565

1.082

1.274

0.673

0.884

0.929

1.081

Table 3. Spearman’s correlation matrix among the species with the greatest importance values in a Seasonal Semideciduous Forest fragment

under the influence of selective logging, Amazonian Biome, 2014. (Underlined values are negative). ns – not significant at the level of 5%

error probability; * – significant at the level of 5% error probability.

Tabela 3. Matriz de correlação de Spearman entre as espécies com os maiores valores de importância em um fragmento de Floresta

Estacional Semidecidual sob influência da exploração seletiva, Bioma Amazônico, 2014. (Valores sublinhados são negativos). ns – não

significativo ao nível de 5% de probabilidade de erro; * – significativo ao nível de 5% de probabilidade de erro.

Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Y11 Y12 Y13 Y14 Y15 Y16 Y17 Y18 Y19 Y20

0.29

0.14

0.36

0.10

0.01

0.22

0.15

0.34

0.14

0.07

0.22

0.15

0.05

0.02

0.32

0.29

0.24

0.06

0.09

0.01ns

0.06ns

0.01ns

0.06ns

0.01ns

0.18ns

0.07ns

0.11ns

0.10ns

0.31ns

0.13ns

0.11ns

0.18ns

0.29ns

0.10ns

0.37ns

0.02ns

0.33ns

0.08

0.47*

0.16

0.08

0.20

0.15

0.60*

0.00

0.07

0.05

0.18

0.17

0.07

0.18

0.46*

0.34

0.17

0.36

0.00

0.10

0.49

0.00

0.48*

0.02

0.23

0.12

0.40*

0.03

0.00

0.03

0.20

0.05

0.12ns

0.20ns

0.05ns

0.07ns

0.36ns

0.03ns

0.49*

0.19ns

0.18ns

0.10ns

0.14ns

0.20ns

0.06ns

0.23ns

0.10

0.13

0.23

0.05

0.48*

0.07

0.05

0.06

0.20

0.06

0.16

0.22

0.47*

0.00

0.16

0.37

0.03

0.08

0.16

0.27

0.12

0.11

0.41*

0.03

0.11

0.18

0.03

0.15

0.01

0.17

0.30

0.27

0.14

0.19

0.33

0.07

0.08

0.03ns

0.19ns

0.09ns

0.51*

0.23ns

0.13ns

0.37ns

0.12ns

0.10ns

0.24ns

0.27ns

Y10

0.06

0.02

0.32

0.15

0.17

0.00

0.27

0.32

0.14

0.41*

Y11

0.11

0.22

0.07

0.44*

0.10

0.12

0.02

0.06

0.01

Y12

0.03ns

0.32ns

0.27ns

0.47*

0.07ns

0.03ns

0.09ns

0.23ns

Y13

0.27

0.33

0.26

0.07

0.00

0.32

0.44*

Y14

0.12

0.18

0.05

0.20

0.09

0.08

Y15

0.11ns

0.04ns

0.15ns

0.11ns

0.27ns

Y16

0.37

0.20

0.07

0.33

Y17

0.16

0.12

Y18

0.37ns

0.07ns

Y19

0.08

Table 4. Results from the Ward concordance test regarding the species with the greatest importance values for six floristic groups established

in the Seasonal Semideciduous Forest fragment under the influence of selective logging, Amazonian Biome, 2014. W = Kendall’s

concordance coefficient; F = Fisher-Snedecor test statistic; X² = Friedman's Chi-Square test statistic; P = the p-value. * – significant at the

level of 5% error probability.

Tabela 4. Resultados do teste de concordância de Ward quanto às espécies com os maiores valores de importância para seis grupos florísticos

estabelecidos no fragmento de Floresta Estacional Semidecidual sob influência da exploração seletiva, Bioma Amazônico, 2014. W =

coeficiente de concordância de Kendall; F = estatística do teste de Fisher-Snedecor; X² = estatística do teste Qui-Quadrado de Friedman;

P = o valor p. * – significativo ao nível de 5% de probabilidade de erro.

Tests

Group 1

Group 2

Group 3

Group 4

Group 5

Group 6

W 0.6090 0.5698 0.4760 0.3844 0.5695 0.3437

0.1477 0.2530 0.0412 0.0010 0.0023 0.0065



9.2296 27.3487 34.2685 46.1250 41.0045 41.2416

0.1490 0.2250 0.0490* 0.0010* 0.0040* 0.0050*

Rocha et al.

Nativa, Sinop, v. 10, n. 3, p. 333-340, 2022.

337

Table 5. Paired averages of Spearman’s correlation among the species with the greatest importance values, calculated with Kendall’s

concordance test to determine the ideal number of floristic groups in the Seasonal Semideciduous Forest fragment under the influence of

selective logging, Amazonian Biome, 2014.

Tabela 5. Médias pareadas da correlação de Spearman entre as espécies com os maiores valores de importância, calculadas com o teste de

concordância de Kendall para determinar o número ideal de grupos florísticos no fragmento de Floresta Estacional Semidecidual sob

influência de exploração seletiva de madeira, Bioma Amazônico, 2014.

Test with 3 groups

Test with 4 groups

Test with 5 groups

Test with 6 groups

Group Sp.

0.0257

0.2179

Y12

0.0772

Y12

0.0772

Y12

0.2179

Y12

0.2179

Y10

0.0470

Y10

0.0470

Y10

0.0999

Y10

0.1396

Y19

0.0295

Y19

0.0295

Y19

0.0055

Y19

0.1396

0.0070

0.0270

0.0942

Y11

0.0981

Y11

0.0981

Y11

0.1572

Y11

0.2562

Y15

0.1438

Y15

0.1438

Y15

0.1526

Y15

0.2740

0.1091

0.1634

Y13

0.1251

Y13

0.2246

Y13

0.2246

Y13

0.2246

Y16

0.2132

Y16

0.3115

Y16

0.3115

Y16

0.3115

Y17

0.0820

Y17

0.1151

Y17

0.1151

Y17

0.1151

Y20

0.2252

Y20

0.3019

Y20

0.3019

Y20

0.3019

Y14

0.1387

0.0884

0.4245

0.0679

0.2965

0.1788

0.3612

0.1253

0.2563

0.0840

0.2270

0.0646

0.2149

Y18

0.1788

Y18

0.1280

Y18

0.1280

Y18

0.1280

Y14

0.1086

Y14

0.1086

Y14

0.1086

where: 𝑟 = paired averages of Spearman’s correlation; Yi = species with the greatest importance values, ranging from 1 to 20.

onde: 𝑟 = médias emparelhadas da correlação de Spearman; Yi = espécies com maior valor de importância, variando entre 1 e 20.

Figure 2. Classification of the species with the greatest importance values in six floristic groups through “Ward’s” hierarchical method. The

species were located in the Seasonal Semideciduous Forest fragment under the influence of selective logging, Amazonian Biome, 2014.

Figura 2. Classificação das espécies com os maiores valores de importância em seis grupos florísticos através do método hierárquico de

“Ward”. As espécies foram localizadas no fragmento de Floresta Estacional Semidecidual sob influência da exploração seletiva, Bioma

Amazônico, 2014.

Forest fragment grouping analysis under selective logging in Amazonian biome

Nativa, Sinop, v. 10, n. 3, p. 333-340, 2022.

338

Table 6. Descriptive characteristics of the established groups in the Seasonal Semideciduous Forest fragment under the influence of selective

logging, Amazonian Biome, 2014.

Tabela 6. Características descritivas dos grupos estabelecidos no fragmento de Floresta Estacional Semidecidual sob influência da exploração

seletiva, Bioma Amazônia, 2014.

GROUP

N°

Sample

N°

Average

Max

g G

SSub

area

Families

Species

Trees

DBH

1.20

26.67

72.16

0.5393

10.7862

0.80

17.54

41.22

0.1436

2.8729

1.05

17.91

36.76

0.1901

3.8029

1.05

17.31

50.93

0.2807

5.6133

1.05

18.75

50.93

0.3393

6.7855

1.20

123

18.46

55.07

0.4953

9.9054

TOTAL

6.35

392

1.9883

MEDIA

1.06

19.44

51.18

0.3314

6.6277

in which: N° SSub = number of sampled subunits in the group; Sample area = group sampled area in hectares (ha); N° Families = number of families in the

group; N° Species = number of species in the group; N° Trees = number of sampled trees in the group; D = sample density in the group by hectare (ha);

DBH = diameter at 1.3 m (cm); Max DBH = maximum diameter (cm); 𝑔 = group basal area for sample subunit (m2.500 m-2); G = group basal area (m² ha-

1).

em que: N° SSub = número de subunidades amostradas no grupo; Sample area = área amostrada do grupo em hectares (ha); N° Families = número de

famílias no grupo; N° Species = número de espécies no grupo; N° Trees = número de árvores amostradas no grupo; D = densidade amostral no grupo por

hectare (ha); DBH = diâmetro a 1,3 m (cm); Max DBH = diâmetro máximo (cm); 𝑔 = área basal do grupo por subunidade amostral (m².500 m-2); G = área

basal do grupo por hectare (m2.ha-1).

4. DISCUSSION

In at least one group, all of the tested possibilities

accepted the H0 (P>0.05 and F>0.05). That is, there was at

least one independent species in one group that should be

with only concordant species. Legendre’s recommendation

(2005) is that the group that accepted H0 be redefined until

it is composed of concordant species only. However, in the

present study, working with more than six floristic groups

would make the process unfeasible, as it would result in a

high number of groups with only one species.

Spearman’s paired average is an a posteriori test that is

recommended to identify discordant species inside groups.

However, the test does not reveal whether there are one or

many groups of concordant species among those for which

the independence null hypothesis is rejected. This

information is obtained by calculating Spearman’s correlation

among species and by grouping them inside positively

correlated groups (LEGENDRE, 2005).

The low Spearman’s correlation among the 20 species

with the greatest importance values of the studied fragment

– r ≤ |0.60| – can be attributed to the fact that the fragment

suffered from disturbances due to selective logging (Table 3).

In fragments altered by human interference, clearing and

canopy reconstruction are among the most important factors

in the dynamics. The formation of small clearings increases

the levels of luminosity and temperature, which lead the

species to show some degree of physiological and

morphological adaptation in response to sun or shade

environments, and these variations can occur up to the intra-

specific level (FELFILI et al., 2001; BARROS, 2007).

In general, the logging results in species found at the end

and/or beginning of their cycle and opportunistic and/or

generalist species, which changes the typology’s original

characteristic. Over time, in response to interference, these

species obtained a higher importance value (LARA et al.,

2017).

Although groups 1 and 2 accepted the H0 (Table 4), the

species that composed the groups presented positive

correlations (r) (Table 3). Groups 4 and 6 were the only

groups that presented species with negative correlations (r).

These groups were the largest groups and were composed of

five species (Figure 7). Even with the negative correlation of

Aspidosperma discolor A.DC. species with Bellucia grossularioides

(L.) Triana in group 4, and Toulicia guianensis Aubl. species

with Licania blackii Prance in group 6, the Euclidean distances

with Hellinger’s transformation among species were like the

further groups (Table 2).

Groups 1, 4 and 6 were well distributed throughout the

entire fragment, and they occurred in 96% of the sampled

studied subunits (Table 6). Group 1 was composed of Qualea

paraensis Ducke and Inga sp., which were classified as late

secondary and non-classified (NC), respectively (Table 1).

Species that belong to the late secondary ecological group are

species that evolved in undergrowth in conditions with light

or dense shadow. These species are able to remain in these

conditions for all of their life, or they grow until they reach

the canopy or an emergent condition (GANDOLFI et al.,

1995).

Regarding the propagating material dispersion type,

Qualea paraensisis was classified as autochorous (Table 1),

where fruits fall by gravity due to their own weight or with

explosive dispersion (VAN DER PIJL, 1982). The

autochorous syndrome has been frequently recorded in

secondary forests and is considered more advantageous in

open places (LÓPEZ-MARTÍNEZ et al., 2013). This

advantage explains the individual’s occurrence and their high

average DBH (Table 6) in group 1 in the entire fragment.

Although the fragment has some subsequent glades from

selective logging, the fragment also has good coverage of the

Canopy (Table 1).

The distributions of groups 4 and 6 can be attributed to

the fact that they have classified species in every ecological

group (Table 1). Pioneer species are clearly light-dependent

and do not occur in the undergrowth but rather in glades or

forest borders. Moreover, the early secondary species prefer

average shading conditions or low-intensity luminosity, and

they occur in small glades, borders of large glades, forest

borders or in the non-densely shaded undergrowth

(GANDOLFI et al., 1995).

In addition to the succession characteristics, the

distributions of groups 4 and 6 can also be attributed to the

propagating material dispersion type. Group 4 presented a

predominance of the anemochorous dispersal (Table 1),

which is seeds dispersed by the wind (KULMANN;

Rocha et al.

Nativa, Sinop, v. 10, n. 3, p. 333-340, 2022.

339

RIBEIRO, 2016). Researchers highlight that smaller area are

probably more permeable to the arrival of seeds dispersed by

the wind, which are capable of being transported for long

distances (THOMSON et al., 2011). In fact, it is expected

that in more irregular fragments, which have a greater

proportion of borders than regular fragments, anemochorous

species are efficient in their dispersion and settlement

(GOTTSBERGER; SILBERBAUER-GOTTSBERGER,

2018).

Group 6 presented only zoochorous species (Table 1),

where the propagating material dispersion is intrinsically

related to fauna maintenance (MELO et al., 2016). The

animals necessary for seed dispersion inhabit the lake and

river borders and the forest track around the fragment part

(Figure 1). Zoochorous dispersion is the most important

dispersion mechanism in rainforests (GUILHERME et al.,

2021) and is significantly influenced by the isolation effect

and fragment connectivity (MELO et al., 2016).

The species from groups 2 and 3 were concentrated in

the sampled subunits that were close to the agricultural areas

(Figure 1). The species in both groups were classified as early

secondary species (Table 1). They are also classified as

zoochorous, but group 3 presented more developed

individuals, with higher values for DBH and basal areas

(Table 6). In group 5, the species preferred the subunits that

were closer to the forest tracks (Figure 1), and there was a

predominance of late secondary and zoochorous species

(Table 1).

5. CONCLUSIONS

The species association method through Kendall’s

concordance test (W) is suitable for determining the floristic

groups in the Seasonal Semideciduous Forest fragment under

the influence of selective logging.

The species association and similarity inside each

determined floristic group showed similar auto-ecological

characteristics of the species, mainly the propagating material

dispersion type, and the average DBH of each species.

6. ACKOWLEDGEMENTS

Thanks to CAPES (Coordenação de Aperfeiçoamento de

Pessoal de Nível Superior, Brazil) for providing a scholarship

to the first author and the forester Diogo Cavazzini for

enabling us to carry out this research. Finally, the authors are

grateful to CNPq (Conselho Nacional de Desenvolvimento

Científico e Tecnológico, Brazil).

7. REFERÊNCIAS

ALVARES, C. A.; STAPE, J. L.; SENTELHAS, P. C.;

GONÇALVES, J. L. M.; SPAROVEK, G. Köpen’s

climate classification map for Brazil. Meteorologische

Zeitschrift, v. 22, n. 6, p. 711-728, 2013. DOI:

10.1127/0941-2948/2013/0507.

BARROS, M. G. de. Regeneração natural de espécies

lenhosas e suas correlações com fatores ambientais

na mata de galeria do córrego Pitoco. 138p.

Dissertação [Mestrado em Ciências Florestais] –

Universidade de Brasília, Brasília, 2007.

BFG Brazilian Flora 2020: leveraging the power of a

collaborative scientific network. Taxon, v. 71, n. 1, p.

178-198, 2022. DOI: 10.1002/tax.12640

BOCARD, D.; GILLET, F.; LEGENDRE, P. Kendall’s W

Coefficient of Concordance. In: Numerical ecology

with R. New York: Springer, 2011. p. 92-95.

CARVALHO, P. E. R. Espécies arbóreas brasileiras. Vol.

2. Brasília: Embrapa Informação Tecnológica; Colombo,

PR: Embrapa Florestas, 2006. 627p.

CHAI, Z.; WANG, D. A comparison of species composition

and community assemblage of secondary forests between

the birch and pine-oak belts in the mid-altitude zone of

the Qinling Mountains, China. PeerJ, v. 21, n. 4, e1900,

2016. DOI: 10.7717/peerj.1900

DE’ATH, G. 2007. mvpart: multivariate partitioning.

rpart by Terry M Therneau and Beth Atkinson. R port of

rpart by Brian Ripley. Some routines from vegan – Jari

Oksanen. Extensions and adaptations of rpart to mvpart

by Glenn De’ath. Available from cran.r-

project.org/package=mvpart.

DUTRA, R. M. O.; SPERANDIO, M.; COELHO, J. O

método Ward de agrupamento de dados e sua

aplicação em associação com os mapas auto-

organizáveis de Kohonen. 2015.

<http://inf.unisul.br/~ines/workcomp/cd/pdfs/2308.

pdf>. Accessed 10 jan 2015.

FARAH, F. T.; MUYLAERT, R. L.; RIBEIRO, M. C.;

RIBEIRO, J. W.; MANGUEIRA, J. R. S. A.; SOUZA, V.

C.; RODRIGUES, R. R. Integrating plant richness in

forest patches can rescue overall biodiversity in human-

modified landscapes. Forest and Ecology

Management, v. 397, p. 78–88. 2017. DOI:

https://doi.org/10.1016/j.foreco.2017.03.038

FELFILI, J. M.; FRANCO, A. C.; FAGG, C. W.; SOUSA-

SILVA, J. C. Desenvolvimento inicial de espécies de

matas de galeria. In: J.F. Ribeiro; C.E.L. Fonseca & J.C.

Sousa-Silva. Cerrado: caracterização e recuperação de

Matas de Galeria. Planaltina: Ed. Embrapa Cerrados,

2011. p. 779-811.

GANDOLFI, S.; LEITÃO FILHO, H. F.; BEZERRA, C. L.

F. Levantamento florístico e caráter sucessional das

espécies arbustivo-arbóreas de uma Floresta Semidecídua

no município de Guarulhos, SP. Brazilian Journal of

Biology, v. 55, n. 4, p. 753-767, 1995.

GOTTSBERGER, G.; SILBERBAUER-GOTTSBERGER,

I. How are pollination and seed dispersal modes in

Cerrado related to stratification? Trends in a cerrado

sensu stricto woodland in southeastern Brazil, and a

comparison with Neotropical forests. Acta Botanica

Brasilica, v. 32, n. 3, p. 434-445, 2018. DOI:

10.1590/0102-33062018abb0186

GUILHERME, F. A. G.; FERREIRA, W. C.; SILVA, G. E.;

MACHADO, D. L. Floristic and structure of different

strata in an urban Semideciduous Forest in Jataí, Goiás

state, Brazil. Ciência Florestal, v. 31, n. 1, p. 456-474,

2021. DOI: https://doi.org/10.5902/1980509847868

HAIR, J. F.; BLACK, W. C.; BABIN, B. J.; ANDERSON, R.

E.; TATHAM, R. L. Análise multivariada de dados. 6.

ed. Porto Alegre: Bookman, 2009. 688p.

INPE. Projeto PANAMAZÔNIA-II: zoneamento da

cobertura vegetal do Estado do Mato Grosso

(ortoretificado). INPE: DSR

(http://www.dsr.inpe.br/panamazon/img/MT_Valdete

1.pdf). 2010.

KUHLMANN, M.; RIBEIRO, J. F. Evolution of seed

dispersal in the Cerrado biome: ecological and

phylogenetic considerations. Acta Botanica Brasilica, v.

Forest fragment grouping analysis under selective logging in Amazonian biome

Nativa, Sinop, v. 10, n. 3, p. 333-340, 2022.

340

30, n. 2, p. 271-282, 2016. DOI:

https://doi.org/10.1590/0102-33062015abb0331

LARA, R. de O.; PEREIRA, I. M.; FERREIRA, E. A.;

PEREIRA, G. A. M.; SILVA, D. V.; SILVA, E. de B.;

ARAÚJO, F. V. de; OLIVEIRA, P. A. Análise de

cobertura, levantamento florístico e fitossiológico de uma

área em recuperação com topsoil na Serra do Espinhaço,

Brasil. Revista Espacios, v. 38, n. 39, p. 31-46, 2017.

LEGENDRE, P. Species Associations: The Kendall

coefficient of concordance revisited. Journal of

Agricultura, Biological and Environmental

Statistics, v. 10, n. 2, p. 226-245, 2005. DOI:

https://doi.org/10.1198/108571105X46642

LEGENDRE, P.; GALLAGHER, E. D. Ecologically

meaningful transformations for ordination of species

data. Oecologia, v. 129, p. 271-280, 2001. DOI:

https://doi.org/10.1007/s004420100716

LEGENDRE, P.; LEGENDRE, L. Numerical Ecology. 3.

ed. USA: Elsevier, 2012. 1006p.

LÓPEZ-MARTÍNEZ, J. O.; SANAPHRE-VILLANUEVA,

L.; DUPUY, J. M.; HERNÁNDEZ-STEFANONI, J. L.;

MEAVE, J. A.; GALLARDO-CRUZ, J. A. β-Diversity of

Functional Groups of Woody Plants in a Tropical Dry

Forest in Yucatan. PLoS ONE, v. 8, n. 9, e73660, 2013.

DOI: 10.1371/journal.pone.0073660

LORENZI, H. Árvores Brasileiras: manual de

identificação e cultivo de plantas arbóreas nativas do

Brasil. Vol. 3. Nova Odessa, SP: Instituto Plantarum,

2009. 384p.

MELO, M. M.; SILVA, C. M.; BARBOSA, C. S.; MORAIS,

M. C.; D’ANUNCIAÇÃO, P. E. R.; SILVA, V. X.;

HASUI, E. Fragment edge and isolation affect the food

web: effects on the strength of interactions among

trophic guilds. Biota Neotropical, v. 16, n. 2,

e20150088, 2016. DOI: 10.1590/1676-0611-BN-2015-

0088.

PERES, M. K. Estratégias de dispersão de sementes no

bioma Cerrado: considerações ecológicas e

filogenéticas. 360p. Tese [Doutorado em Botânica] –

Universidade de Brasília, Brasília, 2016.

PRIMACK, R. B.; RODRIGUES, E. Biologia da

conservação. Londrina: Editora Planta, 2001. 338p.

R Core Team (2014). R: A language and environment for

statistical computing. R Foundation for Statistical

Computing, Vienna, Austria. URL https://www.R-

project.org/.

ROCHA, K. J. da; SOUZA, E. C. de; FAVALESSA, C. M.

C.; CALDEIRA, S. F.; MARTINEZ, D. T.;

BRONDANI, G. E. Effect of selective logging on

floristic and structural composition in a forest fragment

from Amazon Biome. Acta Scientiarum. Agronomy, v.

39, n. 2, p. 191-199, 2017. DOI:

10.4025/actasciagron.v39i2.32543.

ROCHA, K. J. Composição e estrutura de grupos

florísticos em fragmento de floresta secundária. 2015.

179f. Dissertação [Mestrado em Ciências Florestais e

Ambientais] – Universidade Federal de Mato Grosso,

Cuiabá, 2015.

SÜHS, R. B.; BUDKE, J. C. Spatial distribution, association

patterns and richness of tree species in a seasonal forest

from the Serra Geral formation, southern Brazil. Acta

Botanica Brasilica, v. 25, n. 3, p. 605-617, 2011.

THE ANGIOSPERM PHYLOGENY GROUP. An update

of the Angiosperm Phylogeny Group classification for

the orders and families of flowering plants: APG IV.

Botanical Journal Linnean Society, v. 181, n. 1, p. 1-

20, 2016. DOI: https://doi.org/10.1111/boj.12385

THOMSON, F. J.; MOLES, A. T.; AULD, T. D.;

KINGSFORD, R. T. Seed dispersal distance is more

strongly correlated with plant height than with seed mass.

Journal of Ecology, v. 99, p. 1299-1308, 2011. DOI:

10.1111/j.1365-2745.2011.01867.

VAN DER PIJL, L. Principles of dispersal in higher

plants. Berlin: Springer, 1982. 218p. DOI:

https://doi.org/10.1007/978-3-642-87925-8