Forest Ecology and Management 339 (2015) 105–111

Contents lists available at ScienceDirect

Forest Ecology and Management

journal homepage: www.elsevier .com/locate / foreco

Efficiency of different planted forests in recovering biodiversityand ecological interactions in Brazilian Amazon

http://dx.doi.org/10.1016/j.foreco.2014.12.0070378-1127/� 2014 Elsevier B.V. All rights reserved.

⇑ Corresponding author at: Universidade Federal de Mato Grosso, Dept. Botânicae Ecologia, Av. Fernando Corrêa da Costa, s/n, CEP: 78060-90091190, Cuiabá, MatoGrosso, Brazil.

E-mail address: izzothiago@gmail.com (T.J. Izzo).

Jéssica C.F. Falcão a, Wesley Dáttilo b, Thiago J. Izzo a,c,⇑a Departamento de Ecologia e Botânica, Universidade Federal de Mato Grosso, Cuiabá, Mato Grosso CEP 78060-900, Brazilb Instituto de Neuroetología, Universidad Veracruzana, Xalapa, Veracruz CP 91190, Mexicoc Centre for Ecology, Evolution and Conservation, School of Environmental Sciences, University of East Anglia, Norfolk, Norwich NR4 7TJ, UK

a r t i c l e i n f o

Article history:Received 4 October 2014Received in revised form 12 December 2014Accepted 12 December 2014

Keywords:Ant–plant networksRestore biodiversityReforestation programs

a b s t r a c t

In tropical forests, several reforestation programs have been created to recover biological diversity indeforested environments. However, most of these studies have focused on the loss of species and ignoredthe loss of ecological interactions and ecosystem services that these species provide in their natural envi-ronment. Here, we assess how reforestation (including both native and exotic species) and the distancefrom the forest matrix could recover the following parameters: richness, composition, and ant–plantinteractions in the southern Brazilian Amazon. For this, we collected data in five different agroforestrylandscapes: primary and secondary forests, reforestations of teak and fig, and pasture. In general, weobserved that the distance from the forest matrix was not an important factor for the recovery of ant–plant interactions at the scale of this study. Ant–plant interactions were more specialized in the pasture,possibly due to the low richness and high dominance in pastures, which generates a high ant–plantdependency. Moreover, none of the strategies employed for the recovery of diversity (teak and fig refor-estation) increased the recovery rate of the studied parameters: richness, composition, and the patternsof ant–plant interactions, when compared to primary forests. These results indicate that regardless ofspecies richness and composition, secondary forests and reforestation lands are functionally differentfrom pastures. In conclusion, our results represent a valuable tool to conservation planners, mainlybecause we show that secondary forests created by natural regeneration can be an efficient and econom-ical method for restoring the disruption of ant–plant interactions in tropical forests.

� 2014 Elsevier B.V. All rights reserved.

1. Introduction

One of the main causes of tropical deforestation is the conver-sion of forest to pasture for cattle ranching (Fearnside, 2006). Thistype of habitat modification not only decreases the diversity ofplant and animal species (Mitja et al., 2008; Leal et al., 2012) butalso reduces ecological interactions. This, in turn, affects all ecosys-tem services provided by such interactions (e.g., pollination andseed dispersal) (Murcia, 1996; Cramer et al., 2007). Thus, there isan urgent need to implement conservation actions that aim toreduce the effects of deforestation on biodiversity as a whole(Melo et al., 2013).

One strategy frequently used to restore biodiversity is the refor-estation of large areas or riversides. In this strategy, native and/or

exotic species are planted to create microenvironments that canfacilitate the establishment of other organisms and promote therecovery of biodiversity (Massad et al., 2011; Milder and Clark,2011). However, the choice of pioneer species is a keystone factorin such reforestation programs, mainly due to the differences inlocal environmental conditions provided by the chosen species(e.g., microclimatic conditions, habitat diversity and leaf litteraccumulation) (Bellingham et al., 2010; Castillo-Núñez et al.,2011; Sansevero et al., 2011).

Among the plant species commonly used in reforestation pro-grams in the tropics, we can highlight Eucalyptus spp. (Myrtaceae)and teak Tectona grandis Linn. f. (Verbenaceae) (Evans, 1999). Bothplant species have a sub-product of great economic value. T. gran-dis is a species primarily planted for timber purposes but has dam-aging effects on plant succession processes; it has been shown toaffect allelopathic properties (Healey and Gara, 2003) and decreasethe attraction of seed dispersers (Tewari, 1992). Consequently, it ispossible that the planting of native species may be a better alterna-tive because these species have been reported to provide better

106 J.C.F. Falcão et al. / Forest Ecology and Management 339 (2015) 105–111

habitats for native animals (Evans and Turnbull, 2004; Assumpção,2008). However, in addition to the taxonomic identity of the spe-cies used in reforestation programs, the distance of the forestmatrix (provider of propagules) can also exert a large influenceon the richness and composition of species in these environments(Tabarelli and Gascon, 2005). This is primarily due to the dispersalability limits of both seeds and animals (Wunderle, 1997; Wagneret al., 2004). Thus, it is expected that both the identity of the pio-neer species and the distance from the forest matrix may be themain mechanisms responsible for the recovery of biodiversity insuch reforestation programs.

Although the importance of ecological interactions in maintain-ing biodiversity has been reported since the 70s (Janzen, 1974),most studies have evaluated biodiversity restoration by focusingonly on the recovery of the richness, and eventually, the speciescomposition (Mullah et al., 2011; Azcárate and Peco, 2012; Reidet al., 2012). In fact, studies have often ignored the loss of ecolog-ical interaction diversity (Tylianakis et al., 2010). However, ecolog-ical interactions affect all ecosystem attributes, from primaryproductivity to population dynamics (Tylianakis et al., 2007; Del-Claro and Torezan-Silingardi, 2009), and play an important rolein the structure and stability of biological communities (Janzen,1974; Del-Claro and Torezan-Silingardi, 2009; Dyer et al., 2010).Thus, instead of using only the species approach to understand res-toration processes, we should consider the interactions that thesespecies are involved to understand the process as a whole andimprove the reforestation programs.

Due to the high abundance of ants and plants in tropical for-ests (Longino et al., 2002; Feeley and Silman, 2012), the diver-sity of ecological interactions involving both organisms isextremely remarkable in these environments (Rico-Gray andOliveira, 2007). One of these interactions occurs between antsand plants with extrafloral nectaries (EFN-bearing plants). Inexchange for the energy-rich food provided, ants protect theirhost plant against potential herbivores (Rico-Gray and Oliveira,2007). The frequency of EFN-bearing plants in some BrazilianAmazon landscapes can reach up to 53% of the total plant spe-cies (Rico-Gray and Oliveira, 2007); moreover, ant diversity canbe extremely high, reaching approximately 270 species at localscales (Miranda et al., 2012). Thus, because of the abundanceand importance of the ant–plant interactions in Amazon land-scapes, this can be a useful model for understanding the effi-ciency of reforestation projects in the recovery of ecologicalinteractions.

In this study, we evaluated the recovery of ecological interac-tions involving ants and EFN-bearing plants in five different agro-forestry landscapes in the southern Brazilian Amazon: primaryforest, secondary forest, teak reforestation, fig reforestation, andpasture. For this, we sampled the ecological interactions involvingants and EFN-bearing plants in different agroforestry landscapes inthe southern Brazilian Amazon and compared then with primaryforest and active pastures. To achieve this we answered the follow-ing questions: (1) Does the reforestation of pastures help in recov-ering the richness and composition of ants and plants species? (2)Are the network patterns that describe the interactions involvingthese organisms different between planted forests? (3) Does thedistance between the primary forest and the others landscapesaffect the velocity of recovery? We predict that the native refores-tation (Ficus maxima) will be the landscape with better recovery ofboth the species and network properties. We also predicted thatthe sites that are closer to the forest matrix would be morerestored than the other ones. Overall, our goal was to propose anew, empirical and effective recommendation to evaluate the res-toration of ecological interactions in tropical forests using severalmetrics derived from network theory.

2. Materials and methods

2.1. Study area and sampled environments

We conducted fieldwork in December 2010 and December 2011at the São Nicolau farm (9�480S and 58�150W, elev. 254 m), locatedin the municipality of Cotriguaçu in the north of Mato Grosso State,Brazil. The farm has 10,000 ha, of which 7000 ha are primary for-est, 500 ha are riparian and secondary forests, 1700 ha are refores-tation of several species, and 300 ha are pastures used by cattle.Reforestations and secondary forests were both previously pas-tures before the planting of vegetation and the abandonment ofthe pastures, respectively, which occurred between 1999 and2000 (Rodrigues et al., 2011). The regional climate is AW type,according to the Köppen classification (warm and humid), withan average annual temperature of 24 �C, 85% humidity, and2300 mm of precipitation (Rodrigues et al., 2011).

In this study, we collected data on five different landscapes: (1)Primary forest: defined as a ‘‘terra-firme’’ and closed-canopy forestwithout the influence of seasonal flooding of larger rivers. Theunderstory is relatively open and very biodiverse, and the canopyvaries between 30 and 40 m tall with some trees reaching up to50 m; (2) Secondary forest: natural restoration of nearly 13 yearswithout the presence of cattle or anthropogenic interference. Thislandscape is used in our design as the control of the experimentalplantation of trees. The understory of secondary forests is extre-mely dense, with vegetation higher than three meters; there is alow leaf litter accumulation and light incidence; (3) Teak reforesta-tion: reforestation of T. grandis (Verbenaceae) of nearly 10 years, anexotic tree that loses its leaves annually. During the collection per-iod in this landscape, leaves were green, allowing little light pene-tration in the understory. Moreover, the understory is open and hasan abundant leaf litter accumulation; (4) Fig reforestation: refores-tation of Ficus maxima (Moraceae) of nearly 10 years and per-formed with seedlings obtained from native seeds in our studyarea. This landscape presents some gaps formed by dead individu-als, which are characterized by grasses and shrubs. The understoryhas plants of varying sizes and low leaf litter accumulation; (5)Pasture: landscape with a great predominance of planted grasses(not more than 50 cm tall) used as food for cattle. In addition, thislandscape is characterized by a very high incidence of light and theabsence of leaf litter accumulation. This landscape is used in ourdesign as a ‘‘starting point’’ of forest succession under differentreforestations.

2.2. Experimental design and data collection

We performed the survey of ant–plant interactions in 62 plots:primary forest (n = 12 plots), secondary forest (n = 10 plots), teakreforestation (n = 14 plots), fig reforestation (n = 14 plots), and pas-ture (n = 10 plots) (Fig. 1). With the exception of the primary forest,we divided the landscapes into two classes of distance from theprimary forest (forest matrix), in which half of the plots werelocated immediately close the forest edge and the other half were250 m from the edge. In these landscapes, plots consisted of rect-angles of 57.5 m � 52 m and totaled 3000 m2; all plots were atleast 200 m apart. The data collection in the primary forest wasconducted in a module of the Brazilian Program for BiodiversityResearch (PPBio) at the São Nicolau farm. This module consists ofsix parallel trails running north/south and two parallel trails run-ning east/west. Each stretch of 1 km on the trails was consideredan independent sample unit of 250 m � 12 m, totaling 12 plots of3000 m2. At each of the 62 plots, we searched for EFN-bearingplants that were accessible to the collector (height between 0.5and 3 m), and tried to reach the total of 40 individuals of plants

Fig. 1. Map showing the spatial arrangement of our 62 plots established in five different landscapes at the São Nicolau farm, municipality of Cotriguaçu, State of Mato Grosso,Brazil. Primary forest (s), secondary forest (4), teak reforestation (h), fig reforestation (I), and pasture (}).

J.C.F. Falcão et al. / Forest Ecology and Management 339 (2015) 105–111 107

for sample site. All the plants collected have at least 1 m betweeneach other. In each EFN-bearing plant found we collected all theants using an adapted method in which we placed a beat sheetunder the branches of a plant, and after shaking the plant we col-lected all the ants that fell on the sheet (Dáttilo and Dyer, 2014).Ants were collected for later identification, as well as a botanicsample of the EFN-bearing plants where the ants were found. Antswere identified using published keys and with the help of a refer-ence collection of the Entomological Collection of the UniversidadeFederal de Mato Grosso (CEMT), Brazil. Plants were identified anddeposited at the Central Herbarium of the same institution.

2.3. Data analysis

Because forested environments are older, we expected a natu-rally greater abundance of individuals in these environments. Asour goal was to compare the diversity independent of abundance,the values of the number of species of ants and plants were rarefiedto the abundance of the plots with the lowest number of individu-als found. For plants, we used rarefaction by individuals. On theother hand, for ants, because more than one species can be foundin the same plant, we used rarefaction by sample.

To evaluate the effect of distance from the forest matrix (near orfar) and the type of landscape (primary and secondary forests,reforestations, and pasture) on rarefied richness and compositionof ants and plants, we performed a two-way ANOVA (for richness)and a two-factor ANOSIM (for species composition). In both analy-ses, the distance and the landscape were used as fixed factors. Fur-thermore, to assess the differences in ant and plant richness in thefive landscapes, we performed a one-way ANOVA using only thelandscape as a fixed factor. Significant differences among land-scapes were tested using a Tukey post hoc test.

We used ANOSIM tests with Bonferroni corrections to deter-mine whether there were effects of the types of landscape on ant

and plant compositions separately. In addition, we created graphsof species composition using scores of nonmetric multidimen-sional scaling (NMDS) built using a Bray–Curtis dissimilarity asthe distance metric.

To assess whether a landscape had more ant species than typi-cal of primary forests, we performed a one-way ANOVA using thefrequency of occurrence of species from primary forests found inother landscapes. It was not possible to do the same analysis forthe plant species because only four species typically found in pri-mary forest were found in the other landscapes. The NMDS wasperformed in the R-software, version 2.13.1 (R Core Team, 2010)and all other analyses described above were performed using thePAST-software version 2.04 (Hammer et al., 2001).

To study ant–plant interactions based on a network theoryapproach, we built an adjacency matrix C for each plot studied,where cij = 1 if the consumption of nectar from a plant species jby the ant species i was recorded; the value is equal to zero other-wise (Bascompte et al., 2003). To calculate the level of specializa-tion of each network, we used the specialization index (H02)(Blüthgen et al., 2006). This index ranges from 0 (extreme general-ization) to 1 (extreme specialization) and is robust to differences insampling intensity and variation in richness (Blüthgen et al., 2006,2007). We also calculated the diversity of ant–plant interactionsaccording to Bersier et al. (2002). This index is derived from Shan-non’s diversity index and ranges from 0 (no diversity) to infinity.We used the package bipartite (Dormann et al., 2009) to calculateboth indexes of network specialization and diversity of ant–plantinteractions in R-software, version 2.13.1 (R Core Team, 2010).

To analyze the effect of distance from the forest matrix and thetype of environment (pasture, teak, fig and secondary forest) onnetwork specialization and diversity of ant–plant interactions, weused three separate two-way ANOVAs for each metric of networkinteractions. Finally, to compare the effect of all types of environ-ment (pasture, teak, fig, secondary forest and primary forest) on

108 J.C.F. Falcão et al. / Forest Ecology and Management 339 (2015) 105–111

network specialization and the diversity of ant–plant interactions,we used one-way ANOVAs for each of the three metrics followedby a Tukey post hoc test. We performed these analyses in PAST, ver-sion 2.04 (Hammer et al., 2001). The distribution of all variables didnot differ from normal distribution (Kolmogorov–Smirnov testP > 0.05).

3. Results

3.1. Richness and species composition

In this study, we recorded 174 EFN-bearing plant species dis-tributed in 41 families. Bignoniaceae was the most representativefamily (20% of the species), followed by Mimosaceae (18%), Caesap-inalceae (8%), and Fabaceae (6%). Only four plant species recordedin the primary forest were also found in the other environments:Bauhinia sp5 (Caesapinalceae), Inga sp1, Inga sp2 and Inga sp4(Mimosaceae). For ants, we recorded 211 species in 29 generaand 8 subfamilies. Interestingly, the ant Pheidole gertrudae (Myrm-icinae) was collected in 100% of the plots in reforestations land-scapes, 98% of the plots in pasture, and 80% of the plots insecondary forest. The primary forest was the only environmentwhere P. gertrudae was not collected.

We observed that the rarefied richness of ants and plants didnot vary in relation to distance from the forest matrix (ants:F4,57 = 1.065, p = 0.36; plants: F4,57 = 0.83, p = 0.51). Moreover, wealso did not observe any change in the ant composition in relationto distance from the forest matrix (ants: r = 0.007, p = 0.49). How-ever, the distance influenced the plant composition (r = 0.15,p = 0.01). Despite this result, only four plant species that were col-lected in the primary forest were also collected in the other land-scapes; the influence of distance did not reflect a similarity withthe primary forest.

We also observed that the type of landscape affects the plantrichness (F4,57 = 20.07, p < 0.01) (Fig. 2A). The primary forest plotshad the largest number of plant species recorded (mean ± SD:12.95 ± 0.91) (F4,57 = 20.07, p = 0.02), and fig reforestation areashad a larger number of species than pasture areas (9.02 ± 1.81)(F4,57 = 20.07, p = 0.02). In the other reforestation pairwise

Fig. 2. Descriptors of the ant–plant interactions studied in five landscapes in the southinteractions; (D) network specialization; for primary forest (PF), secondary forest (SF), tesame letter are not significantly different using Tukey’s test (5% probability level).

comparisons, we did not observe any difference in the plant rich-ness (F4,57 = 20.07, p > 0.05) (Fig. 2A). However, although the plantcomposition differed among all landscapes (ANOSIM: r = 0.72,p < 0.01), we observed a higher similarity between primary andsecondary forests (NMDS: stress = 0.2163, Fig. 3A). Only the teakand fig reforestation areas did not differ between one another inrelation to the plant composition (r = 0.72, p > 0.05).

For ants, we observed that the type of landscape influencedboth the rarefied richness (F4,57 = 22, p < 0.01) and species compo-sition (ANOSIM: r = 0.49, p < 0.01) (Fig. 2B). We observed the larg-est ant rarefied richness in the primary forest (mean ± SD:7.38 ± 1.14) (F4,57 = 22, p < 0.01) and the lowest in the pasture(3.55 ± 0.88) (F4,57 = 22, p < 0.01) (Fig. 2B). Furthermore, only theprimary and secondary forests had ant compositions different fromthe other landscapes (r = 0.49, p < 0.05). Indeed, the secondary for-est was the landscape with largest number of species in commonwith the primary forest (F4,57 = 84.42, p < 0.05), which reflects thegreater similarity in ant composition between these two land-scapes (NMDS: stress = 0.1354, Fig. 3B).

3.2. Ant–plant interactions

We observed that the distance in relation to the forest matrix didnot affect network specialization (F3,48 = 0.41, p = 0.52) or the diver-sity of ant–plant interactions (F3,48 = 0.07, p = 0.79). Moreover, boththe diversity of ant–plant interactions and network specializationdid differ among the landscapes (diversity of interactions:F4,56 = 22.5, p < 0.01; network specialization: F4,56 = 3.61, p = 0.01)(Fig. 2C and D). As expected, the primary forest had the highestdiversity of interactions (mean ± SD: 4.08 ± 0.19) (F4,56 = 22.5,p < 0.01) and the pasture had the lowest (2.45 ± 0.76) (F4,56 = 22.5,p < 0.01). Other landscapes showed similar patterns of diversity inant–plant interactions among each other (F4,56 = 22.5, p > 0.05,Fig. 2C). Finally, the primary forest had lower network specializa-tion (mean ± SD: 0.08 ± 0.05) compared to fig reforestation(0.18 ± 0.07) and pasture (0.18 ± 0.12) (F4,56 = 3.61, p < 0.05). Noneof the other comparisons of the network specializations differedsignificantly when we evaluated in pair-to-pair comparisons(F4,56 = 3.61, p > 0.05) (Fig. 2D).

ern Brazilian Amazon: (A) richness of plants; (B) richness of ants; (C) diversity ofak reforestation (TR), fig reforestation (FR), and pasture (PE). Means followed by the

Coo

rdin

ate

21.

00.

50.

0-0

.5-1

.0-1

.5

-1.0 -0.5 0.0 0.5 1.0 1.5

Coordinate 1

0.5

0.0

-0.5

-0.5 0.0 0.5 1.0

Coordinate 1

Coo

rdin

ate

2

(A)

(B)

Primary forest

Secondary forest

Teak reforestation

Fig reforestation

Pasture environment

Fig. 3. Non-metric multidimensional scaling (NMDS) showing the relative similar-ities of (A) plant and (B) ant communities among the five studied landscapes:primary forest, secondary forest, teak reforestation, fig reforestation, and pasture.

J.C.F. Falcão et al. / Forest Ecology and Management 339 (2015) 105–111 109

4. Discussion

4.1. Richness and species composition

As expected, we found greater richness of ants and plants in theprimary forest and lower richness in the pasture landscape. How-ever, contrary to our initial hypothesis, we did not observe anyeffect of distance from the primary forest on the richness and com-position of ants and plants. This is surprising and indicates that theenvironment has a strong categorical effect that acts as a filter tothe entrance of seedlings from primary forests (Holl, 1999). Thisstrong environmental filter effect acted by removing the non-adapted propagules; this effect is constant or varies little spatially.Thus, the premise that distance strongly limits the colonization ofnew habitats is not true or has little influence on the establishmentof plants and ants. In this study, we show that with the exceptionof the primary forest, the fig reforestation plots showed greatest

richness of EFN-bearing plants. We already know that many EFN-bearing plant species seem to have some physiological advantagein environments with higher luminosity (Radhika et al., 2010;Dáttilo et al., 2013a). Obviously, the landscape that had a higherincidence of light in our study was the pasture.

However, pastures are subjected to the strong effect of cattletrampling and grazing, which decreases plant recruitment (Hallet al., 1992; Dunne et al., 2011). Among the other landscapes stud-ied, fig reforestations (F. maxima) are the landscapes with a higherlight incidence. F. maxima is an evergreen species and does nothave a closed canopy when compared with teak. A study per-formed by Healey and Gara (2003) in Costa Rica showed that theshade produced by teak’s large leaves may partially allow lessrecruitment of native trees in relation to secondary forests. There-fore, it is possible that T. grandis adversely affects the germinationrate of several EFN-bearing plants species within teak reforesta-tions. Therefore, light availability could act as a strong environ-mental filter (Holl, 1999), which could therefore generate thedifference in the richness of EFN-bearing plants among the sam-pled landscapes. In addition, the highest richness found in the figreforestation areas does not reflect the plant composition, whichwere quite different between the reforestations and primary for-ests. This indicates that most EFN-bearing plant species found inthe secondary forests could be from the forest edge or treefall gaps(environments not sampled in this study). Indeed, abiotic factorsand the ability of dispersers to move among sites are the main fac-tors that influence the richness and composition of plants(Wunderle, 1997; Holl, 1999).

Here, ant composition was different in relation with the othersampled landscapes only in the primary and secondary forests.Therefore, it is possible that secondary forests, treated here asthe control to restore biodiversity, were ‘‘one step ahead’’ of recov-ering biodiversity due to allowing the entry of ants species typicalof primary forests. Considering the higher similarity in ant compo-sition between the primary and secondary forests, it is possiblethat the recolonization by primary forest ant species was favoredin secondary forests. Furthermore, the ant with the highest numberof interactions in the pasture (P. gertrudae) was also remarkablydominant in teak and fig reforestations. Moreover, although P. ger-trudae can also be found in secondary forests, this ant speciesshares the dominance of interactions with other species in theselandscapes (e.g., Wasmannia auropunctata). A recent study con-ducted in this same study area by Dáttilo et al. (2013b) showedthat W. auropunctata is extremely common in ant–plant networkscollected in primary forests. Moreover, this study also showed thatW. auropunctata has highly generalized interactions because thisant species interacts with a large proportion of plant species. Thus,the decrease in dominance of P. gertrudae and the similarity ofhighly generalist species (i.e., those with highly generalized inter-actions) between secondary and primary forests indicate that bothlandscapes are more similar between one another than comparedwith other sampled landscapes. Moreover, the high structural com-plexity of secondary forests could increase the availability of nest-ing sites and other food resources, and, consequently, ant richness.

4.2. Ant–plant interactions

Prior studies have shown that the diversity of ecological inter-actions tends to increase over the regeneration of an environment(Albrecht et al., 2007; Tylianakis et al., 2007, 2010). In support ofprevious studies, we observed that the primary forest has a greaterdiversity of ant–plant interactions. In addition, we also found thatthe secondary forests and reforestation (teak and fig) environ-ments have a greater diversity of ant–plant interactions in relationto the pasture. These results indicate that regardless of speciesrichness or difference in species composition, secondary forests

110 J.C.F. Falcão et al. / Forest Ecology and Management 339 (2015) 105–111

and reforestation areas are functionally different from pastureareas. These results are important because it is known that envi-ronments with a greater diversity of ecological interactions havemore species that could perform similar roles in ecosystem func-tion, and therefore, when a species is absent other species could‘‘dampen’’ the system (Yachi and Loreau, 1999; Elmqvist et al.,2003; Mouchet et al., 2010). Thus, we believe the diversity ofant–plant interactions may actually be an efficient tool to evaluatethe recovery of ecological interactions along a gradient of environ-mental disturbance.

We already know that in environments with higher levels ofenvironmental complexity, the distribution of ecological interac-tions tend to be more homogeneous, as there is greater functionalredundancy in these sites (Kokkoris et al., 1999; Neutel et al., 2002;Bascompte and Melián, 2005; Bascompte et al., 2006; Tylianakiset al., 2010). Here, we found that although the number of generalistant species increased in more disturbed environments (Ward,2000; Dáttilo et al., 2011), ant–plant interactions were more spe-cialized in the pasture. This can be explained because the environ-mental generalist P. gertrudae was extremely dominant in thepasture, a landscape with low richness of species, which leads tothe high specialization in these ant–plant networks because thisant can interact with mostly plant species. This is a ‘false’ special-ization, since that the plant species interact with almost only withP. gertrudae because this species is super dominant in this environ-ment. The genera Pheidole is hyperdiverse and often abundant anddominant in most of the environments where is found (Wilson,2003). P gertrudae is broadly distributed in Brazil, occurring in dif-ferent biomes, occasionally in environments with some perturba-tion (Santos et al., 2006; Monteiro, 2008; Dias et al., 2008; Lopeset al., 2010; Costa-Milanez et al., 2014; Montine et al., 2014) andoften dominant and abundant (Silvestre et al., 2003; Monteiro,2008; Dáttilo et al., 2012; Soares, 2013). In the pasture, the ant P.gertrudae represented approximately 70% of all specimens inter-acting with EFN-bearing plants in this environment, generating ahigh dependence of this ant species by plants. In this case, it isexpected that the specialization level of ant–plant networks tendsto be greater in disturbed environments because there are fewredundant species. In other words, primary forests have a morestable interactive community, mainly due to greater functionalredundancy generated by the high diversity of generalized ant–plant interactions.

5. Conclusions

Here, we observed that the distance from the forest matrix wasnot an important factor for the recovery of ant–plant interactionsat the scale of this study. Moreover, none of the strategiesemployed for the recovery of diversity (teak and fig reforestation)increase the rate of recovery of the studied parameters: richness,composition and the patterns of ant–plant interactions in relationto primary forests. However, when we evaluated the same ecolog-ical parameters in the absence of human intervention (secondaryforests), the recovery of old abandoned pastures toward primaryforests tends to be more efficient than areas reforested with teakor fig monocultures. Thus, it is possible that secondary forestsquickly and efficiently promote the restoration of ants, plantsand interactions as a whole in the southern Brazilian Amazon.However, we do not know the effectiveness of the other systemsof reforestation (e.g., those involving different plant species withmosaics of successional stages) in the restoration of this ecologicalinteraction involving ants and EFN-bearing plants. In short, ourstudy provides useful information about the true role of thecurrent reforestation programs in promoting biodiversity in theBrazilian Amazon.

Acknowledgements

We are grateful to H. Vasconcelos, T. Gonçalves-Souza, D.J.Rodrigues, L.A.F. Mateus and two anonymous referees for valuablecomments and discussions on earlier versions of the manuscript.We also are grateful to R.E. Vicente for the help in the fieldworkand to R. Feitosa for identifying the ant species Pheidole gertrudae.We thank ONF-Brazil and the PPBio Project for logistical and finan-cial support and the staff of the Central Herbarium of UFMT foridentification of plant specimens. J. Falcão and W. Dáttilo are grate-ful for financial support from the CNPq (237339/2012-9). TJI is sup-ported by CAPES (BEX 2548-14/3). This is publication 46 in theNEBAM technical series.

References

Albrecht, M., Duelli, P., Schmid, B., Müller, C.B., 2007. Interaction diversity withinquantified insect food webs in restored and adjacent intensively managedmeadows. J. Anim. Ecol. 76, 1015–1025.

Assumpção, J.V.L., 2008. Desenvolvimento inicial de Ficus maxima Mill. emreflorestamento puro e misto em Cotriguaçu/MT. MSc. Dissertation,Universidade Federal de Mato Grosso.

Azcárate, F.M., Peco, B., 2012. Abandonment of grazing in a mediterraneangrassland area: consequences for ant assemblages. Insect Conserv. Divers. 5,279–288.

Bascompte, J., Melián, C.J., 2005. Interaction strength combinations and theoverfishing. Proc. Natl. Acad. Sci. USA 102, 5443–5447.

Bascompte, J., Jordano, P., Melián, C.J., Olesen, J.M., 2003. The nested assembly ofplant–animal mutualistic networks. Proc. Natl. Acad. Sci. USA 100, 9383–9387.

Bascompte, J., Jordano, P., Olesen, J.M., 2006. Asymmetric coevolutionary networksfacilitate biodiversity maintenance. Science 312, 431–433.

Bellingham, P.J., Wiser, S.K., Wright, A.E., Cameron, E.K., Forester, L.J., 2010.Disperser communities and legacies of goat grazing determine forestsuccession on the remote Three Kings Islands, New Zealand. Biol. Conserv.143, 926–938.

Bersier, L.F., Banasek-Richter, C., Cattin, M.F., 2002. Quantitative descriptors of food-web matrices. Ecology 83, 2394–2407.

Blüthgen, N., Menzel, F., Blüthgen, N., 2006. Measuring specialization in speciesinteraction networks. BMC Ecol. 6, 9.

Blüthgen, N., Menzel, F., Hovestadt, T., Fiala, B., 2007. Specialization, constraints,and conflicting interests in mutualistic networks. Curr. Biol. 17, 341–346.

Castillo-Núñez, M., Sánchez-Azofeifa, G.A., Croitoru, A., Rivard, B., Calvo-Alvarado, J.,Dubayah, R.O., 2011. Delineation of secondary succession mechanisms fortropical dry forests using LiDAR. Remote Sens. Environ. 115, 2217–2231.

Costa-Milanez, C.B., Lourenço-Silva, G., Castro, P.T.A., Majer, J.D., Ribeiro, S.P., 2014.Are ant assemblages of Brazilian veredas characterised by location or habitattype? Braz. J. Biol. 74, 89–99.

Cramer, J.M., Mesquita, R.C.G., Williamson, G.B., 2007. Forest fragmentationdifferentially affects seed dispersal of large and small-seeded tropical trees.Biol. Conserv. 137, 415–423.

Dáttilo, W., Dyer, L., 2014. Canopy openness enhances diversity of ant–plantinteractions in the Brazilian Amazon rain forest. Biotropica 46, 712–719.

Dáttilo, W., Sibinel, N., Falcão, J.C.F., Nunes, R.V., 2011. Mirmecofauna em umfragmento de Floresta Atlântica urbana no município de Marília, Brasil. Biosci. J.27, 494–504.

Dáttilo, W., Martins, R.L., Uhde, V., Noronha, J.C., Florêncio, F.P., Izzo, T.J., 2012. Floralresource partitioning by ants and bees in a jambolan Syzygium jambolanum(Myrtaceae) agroforestry system in Brazilian Meridional Amazon. Agrofor. Syst.85, 105–111.

Dáttilo, W., Rico-Gray, V., Rodrigues, D.J., Izzo, T.J., 2013a. Soil and vegetationfeatures determine the nested pattern of ant–plant networks in a tropicalrainforest. Ecol. Entomol. 38, 374–380.

Dáttilo, W., Guimarães, P.R., Izzo, T.J., 2013b. Spatial structure of ant–plantmutualistic networks. Oikos 122, 1643–1648.

Del-Claro, K., Torezan-Silingardi, H.M., 2009. Insect-plant interactions: newpathways to a better comprehension of ecological communities inNeotropical savannas. Neotropical Entomol. 38, 159–164.

Dias, N.S., Zanetti, R., Santos, M.S., Louzada, J., Delabie, J., 2008. Interação defragmentos florestais com agroecossistemas adjacentes de café e pastagem:respostas das comunidades de formigas (Hymenoptera, Formicidae). Iheringia98, 136–142.

Dormann, C.F., Fründ, J., Blüthgen, N., Gruber, B., 2009. Indices, graphs and nullmodels: analysing bipartite ecological networks. Open Ecol. J. 2, 7–24.

Dunne, T., Western, D., Dietrich, W.E., 2011. Effects of cattle trampling onvegetation, infiltration, and erosion in a tropical rangeland. J. Arid Environ.75, 58–69.

Dyer, L.A., Walla, T.R., Greeney, H.F., Stireman, J.O., Hazen, R.F., 2010. Diversity ofinteractions: a metric for studies of biodiversity. Biotropica 42, 281–289.

Elmqvist, T., Folke, C., Nystrom, M., Peterson, G., Bengston, J., Walker, B., Norberg, J.,2003. Response diversity and ecosystem resilience. Front. Ecol. Environ. 1, 488–494.

J.C.F. Falcão et al. / Forest Ecology and Management 339 (2015) 105–111 111

Evans, J., 1999. Planted forests of the wet and dry tropics: their variety, nature, andsignificance. New Forest. 17, 25–36.

Evans, J., Turnbull, J., 2004. Plantation Forestry in the Tropics. Oxford UniversityPress, New York.

Fearnside, P.M., 2006. Deforestation in Amazonia: dynamics, impacts and control.Acta Amazonica 36, 395–400.

Feeley, K.J., Silman, M.R., 2012. Extinction risks of Amazonian plant species. Proc.Natl. Acad. Sci. USA 106, 12382–12387.

Hall, L.M., George, M.R., McCreary, D.D., Adams, T.E., 1992. Effects of cattle grazingon blue oak seedling damage and survival. J. Range Manage. 45, 503–506.

Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: paleontological statisticssoftware package for education and data analysis. Palaeontol. Electron. 4, 9.

Healey, S.P., Gara, R.I., 2003. The effect of a teak (Tectona grandis) plantation on theestablishment of native species in an abandoned pasture in Costa Rica. For. Ecol.Manage. 176, 497–507.

Holl, K.D., 1999. Factors limiting tropical rain forest regeneration in abandonedpasture: seed rain, seed germination, microclimate, and soil. Biotropica 31,229–242.

Janzen, D.H., 1974. The deflowering of Central America. Nat. History 21, 620–637.Kokkoris, G.D., Troumbis, A.Y., Lawton, J.H., 1999. Patterns of species interaction

strength in assembled theoretical competition communities. Ecol. Lett. 2, 70–74.

Leal, I.R., Filgueiras, B.K.C., Gomes, J.P., Iannuzzi, L., Andersen, A.N., 2012. Effects ofhabitat fragmentation on ant richness and functional composition in BrazilianAtlantic forest. Biodiversity Conserv. 21, 1687–1701.

Longino, J.I., Coddington, J.A., Colwell, R.K., 2002. The ant fauna of a tropicalrainforest: estimating species richness three different ways. Ecology 83, 689–702.

Lopes, D.T., Lopes, J., do Nascimento, I.C., Delabie, J.H., 2010. Diversidade de formigasepigéicas (Hymenoptera, Formicidae) em três ambientes no Parque EstadualMata dos Godoy, Londrina, Paraná. Iheringia 100, 84–90.

Massad, T.J., Chambers, J.Q., Rolim, S.G., Jesus, R.M., Dyer, L.A., 2011. Restoration ofpasture to forest in brazil’s Mata Atlântica: the roles of herbivory, seedlingdefenses, and plot design in reforestation. Restor. Ecol. 19, 257–267.

Melo, F.P.L., Arroyo-Rodríguez, V., Fahring, L., Martínez-Ramos, M., Tabarelli, M.,2013. On the hope for biodiversity-friendly tropical landscapes. Trends Ecol.Evol. 28, 462–468.

Milder, J.C., Clark, S., 2011. Conservation development practices, extent, and land-use effects in the United States. Conserv. Biol. 25 (697), 707.

Miranda, P.N., Oliveira, M.A., Baccaro, F.B., Morato, E.F., Delabie, J.H.C., 2012. Checklist of ground-dwelling ants (Hymenoptera: Formicidae) of the eastern Acre,Amazon, Brazil. CheckList 8, 722–730.

Mitja, D., Miranda, I.S., Velasquez, E., Lavelle, P., 2008. Plant species richness andfloristic composition change along a rice-pasture sequence in subsistence farmsof Brazilian Amazon, abstract influence on the fallows biodiversity. Ecosyst.Environ. 124, 72–84.

Monteiro, A.F.M., 2008. Diversidade de formigas (Hymenoptera: Formicidae) emsistemas de cultivo de algodoeiro no Distrito Federal. MSc. Dissertation,Universidade de Brasília.

Montine, P.S.M., Viana, N.F., Almeida, F.S., Dáttilo, W., Santanna, A.S., Martins, L.,Vargas, A.B., 2014. Seasonality of epigaeic ant communities in a BrazilianAtlantic Rainforest. Sociobiology 61, 178–183.

Mouchet, M.A., Villeger, S., Mason, N.W.H., Mouillot, D., 2010. Functional diversitymeasures: an overview of their redundancy and their ability to discriminatecommunity assembly rules. Funct. Ecol. 24, 867–876.

Mullah, C.J.A., Totland, O., Klanderud, K., 2011. Recovery of plant species richnessand composition in an abandoned forest settlement area in Kenya. Restor. Ecol.20, 462–474.

Murcia, C., 1996. Forest fragmentation and the pollination of neotropical plants. In:Schelhas, J., Greenberg, R. (Eds.), Forest Patches in Tropical Landscapes. IslandPress, Washington, pp. 19–36.

Neutel, A.M., Heesterbeek, J.A.P., Ruiter, P.C., 2002. Stability in real food webs: weaklinks in long loops. Science 296, 1120–1123.

R Development Core Team, 2010. R: A Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna.

Radhika, V., Kost, C., Mithöfer, A., Boland, W., 2010. Regulation of extrafloral nectarsecretion by jasmonates in lima bean is light dependent. Proc. Natl. Acad. Sci.USA 107, 17228–17233.

Reid, J.L., Harris, J.B.C., Zahawi, R.A., 2012. Avian habitat preference in tropical forestrestoration in southern Costa Rica. Biotropica 44, 350–359.

Rico-Gray, V., Oliveira, P.S., 2007. The Ecology and Evolution of Ant–PlantInteractions. The University of Chicago, Chicago.

Rodrigues, D.J., Izzo, T.J., Battirola, L.D., 2011. Descobrindo a Amazônia Meridional:Biodiversidade da Fazenda São Nicolau. Pau e Prosa Comunicação, Cuiabá.

Sansevero, J.B.B., Prieto, P.V., de Moraes, L.F.D., Rodrigues, J.P.R., 2011. Naturalregeneration in plantations of native trees in lowland Brazilian Atlantic forest:community structure, diversity,and dispersal syndromes.Restor. Ecol. 19, 379–389.

Santos, M.S., Louzada, J.N.C., Dias, N., Zanetti, R., Delabie, J.H.C., Nascimento, I.C.,2006. Riqueza de formigas (Hymenoptera, Formicidae) da serapilheira emfragmentos de floresta semidecídua da Mata Atlântica na região do Alto do RioGrande, MG, Brasil. Iheringia 96, 95–101.

Silvestre, R., Brandão, C.R.F., Rosa da Silva, R., 2003. Grupos funcionales dehormigas: el caso de los gremios de cerrado. In: Fernández, F. (Ed.),Introducción a las Hormigas de la Región Neotropical. Instituto Humboldt,Bogotá.

Soares, S.A., 2013. Mecanismos de coesxistência interespecífica em guildas deformigas do gênero Pheidole Westwood (Formicidae: Myrmicinae). PhD thesis,Universidade Federal do Paraná.

Tabarelli, M., Gascon, C., 2005. Lessons from fragmentation research: improvingmanagement and policy guidelines for biodiversity conservation. Conserv. Biol.19, 734–739.

Tewari, D.N., 1992. Monograph on Teak (Tectona grandis). International BookDistributors, Dehradun.

Tylianakis, J.M., Tscharntke, T., Lewis, O.T., 2007. Habitat modification alters thestructure of tropical host-parasitoid food webs. Nature 445, 202–205.

Tylianakis, J.M., Laliberté, E., Nielsen, A., Bascompte, J., 2010. Conservation of speciesinteraction networks. Biol. Conserv. 143, 2270–2279.

Wagner, S., Wälder, K., Ribbens, E., Zeibig, A., 2004. Directionality in fruit dispersalmodels for anemochorous forest trees. Ecol. Model. 179, 487–498.

Ward, P.S., 2000. Brood-scale patterns of diversity in leaf litter ant communities. In:Agosti, D., Majer, J., Alonso, L.E., Schultz, T.R. (Eds.), Ants: Standard Methods forMeasuring and Monitoring Biodiversity. Smithsonian Institution Press,Washington, pp. 99–121.

Wilson, E.O., 2003. Pheidole in the New World: A Dominant, Hyperdiverse AntGenus. Harvard University Press.

Wunderle, J.M., 1997. The role of animal seed dispersal in accelerating native forestregeneration on degraded tropical lands. For. Ecol. Manage. 99, 223–235.

Yachi, S., Loreau, M., 1999. Biodiversity and ecosystem productivity in a fluctuatingenvironment: the insurance hypothesis. Proc. Natl. Acad. Sci. USA 96, 1463–1468.