UNIVERSIDAD VERACRUZANA

INSTITUTO DE NEUROETOLOGÍA

Doctorado en Neuroetología

Elucidating intrinsic properties of ant-plant mutualistic networks

TESIS QUE PARA OBTENER EL GRADO DE DOCTOR EN NEUROETOLOGÍA

Presenta: M. en C. Wesley Francisco Dáttilo da Cruz

Director de tesis: Dr. Víctor Rico-Gray

Instituto de Neuroetología, Universidad Veracruzana

Xalapa, Veracruz, México, Agosto 2015

2

TUTORIAL COMMITTEE

Dr. Víctor Rico-Gray (Advisor) Instituto de Neuroetología, Universidad Veracruzana Xalapa, Veracruz. Mexico

Dra. Cecília Díaz-Castelazo Red de Interacciones Multifróficas Instituto de Ecología A.C. Xalapa, Veracruz. Mexico

Dra. Laura Teresa Hernández Salazar Instituto de Neuroetología, Universidad Veracruzana Xalapa, Veracruz. Mexico

Dr. Armando Martínez Chacón Instituto de Neuroetología, Universidad Veracruzana Xalapa, Veracruz. Mexico

3

"It is interesting to contemplate a tangled bank, clothed with many plants of many kinds, with birds singing on the bushes, with various insects flitting about, and with worms crawling through the damp earth, and to reflect that these elaborately constructed forms, so different from each other, and dependent upon each other in so complex a manner, have all been produced by laws acting around us." (Charles Darwin, 1859)

4 ACKNOWLEDGMENTS

After three academic degrees, at three different universities (Universidade Estadual do

Norte Fluminense, Universidade Federal de Mato Grosso, and Universidad Veracruzana), in two different countries (Brazil and Mexico), I have learned one thing – I could never have done any of this, particularly the research and writing that went into this thesis, without the support and encouragement of a lot of people. Words can never be enough in expressing how grateful I am to those incredible people in my life that made this thesis possible. I would like an attempt to thank them for making my time during my research in the Institute of

Neuroethology at Universidad Veracruzana an amazing period.

First and foremost I want to thank my advisor PhD. Víctor Rico-Gray, especially by accepting a Brazilian student still unknown at that moment for him. It has been an honor to be his Ph.D. student. He has taught me, both consciously and unconsciously, how the good and old natural history can help us to understand complex patterns of species interactions. I appreciate all his contributions of time and ideas to make my Ph.D. experience productive and stimulating. The enthusiasm he has for his research was contagious and motivational for me, even during tough times in the Ph.D. I am also thankful for the excellent example he has provided as a successful ecologist and professor. I would also like to thank all my co-authors of the papers generated in this thesis, but especially to Dr. Thiago Izzo (Universidade Federal de Mato Grosso, Brazil) and Dr. Kléber Del Claro (Universidade Federal de Uberlândia,

Brazil), that even far were important throughout my training as a scientist in Mexico. I would also like to thank to Dr. Pedro Jordano, I am very appreciative of his generosity with his time, advice, data analysis and all help when I was a visiting student in 2014 into his lab at the

Estación Biológica de Doñana in Seville, Spain.

For this thesis I would like to thank my reading committee members: Dra. Laura

Teresa Hernández Salazar (Universidad Veracruzana), Dr. Armando Jesús Martínez Chacón

(Universidad Veracruzana) and Dra. Cecília Díaz Castelazo (Instituto de Ecología, A.C.) for 5 their time, interest, and insightful questions and comments.

I gratefully acknowledge the funding sources that made my Ph.D. work possible. I was funded by the National Council of Science and Technology of Mexico (CONACYT) through the fellowship no. 489746.

Mexico is a wonderful country, full of culture and food (mainly peppers and tequila) and my time over the years was made enjoyable in large part due to the many friends that became a part of my life. I am grateful also to my friends and brothers Armando Aguirre,

Dulce Rodríguez Morales and Juan C. Serio Silva who over the years have patiently endured my highs and lows and have kept me sane and entertained.

Last but not least, I would like to thank my family for all their love and encouragement. Without your unending support and love from childhood to now, I never would have made it through this process or any of the tough times in my life. And most of all for my loving, supportive, encouraging, and patient wife Jessica Falcão whose faithful support during the final stages of this Ph.D. is so appreciated. Thank you for being there for me from the very beginning since my undergraduate studies.

Thank you everybody for believing in me.

6 CONTENTS

1.) ABSTRACT. …………………………………………………………….. p. 7

2.) RESUMEN. ……………………………………………………………… p. 8

3.) General introduction. ………………………….……………….………. p. 9

4.) Chapter I – Strength of the modular pattern in Amazonian symbiotic ant-plant networks. ………………………………………………………….. p. 19

5.) Chapter II – Different tolerances of symbiotic and nonsymbiotic ant- plant networks to species extinctions. ……………………………………….. p. 27

6.) Chapter III – Soil and vegetation features determine the nested pattern of ant-plant networks in a tropical rainforest. ………………………………. p. 40

7.) Chapter IV – Importance of interaction frequency in analysis of ant- plant networks in tropical environments. ……………………………………. p. 48

8.) Chapter V – Ant dominance hierarchy determines the nested pattern in ant-plant networks. ……………………………………………………...... p. 53

9.) General conclusion………………………………………………………. p. 64

7 ABSTRACT: Ants and plants can interact positively in different ways, from facultative to highly specialized relationships. Over spatial and temporal scales different ant and plant species can interact with each other and generate complex ecological networks of interactions.

In this doctoral thesis, the text was divided into five interrelated chapters to evaluate how different biotic and abiotic factors and the specialization level among species structuring ecological networks involving ants and plants with extrafloral nectaries (nonsymbiotic interactions) and myrmecophytes (symbiotic interactions). Moreover, I also evaluated the robustness of such networks to simulated species extinctions and disturbances caused by human modification of the landscape. I showed for the first time that symbiotic ant-plant networks are highly specialized and with low overlap in the use of partners. Consequently, such symbiotic networks were less robust for both ants and plants species extinction compared to nonsymbiotic networks (i.e, those involving ants and plants with extrafloral nectaries). In addition, soil pH was an important factor structuring the nested pattern in nonsymbiotic networks. However, unlike binary networks, quantitative networks are often significantly non-nested. I also show that although ants have an extremely territorial effect only near their nests, competition among ants is strong enough to structure ant-plant networks. Specifically, I found that ant position within the nested ant-plant network can be predicted only by differences among the competitive ability (numerical dominance and recruitment) of ant species. In short, my results in this doctoral thesis highlight the importance of the level of specialization and abiotic and biotic factors in the maintenance of the structure of ant-plant mutualistic networks.

Keywords: ant-plant interactions, dominance hierarchy, mutualistic networks, modularity, nestedness, and robustness.

8 RESUMEN: Las hormigas y las plantas interactuan positivamente de diferente manera, desde facultativa hasta relaciones altamente especializadas. A lo largo de diferentes escalas espaciales y temporales, diferentes especies de hormigas y plantas pueden interactuar entre sí y generar redes ecológicas complejas de interacciones. En esta tesis de doctorado, el texto se divide en cinco capítulos interrelacionados para evaluar cómo los diferentes factores bióticos y abióticos y el nivel de especialización entre las especies estructuran las redes ecológicas que involucran hormigas y plantas con nectarios extraflorales (interacciones no simbióticas) y mirmecófitas (interacciones simbióticas). Por otra parte, también evalué la robustez de dichas redes para la extinción simulada de especies y para las perturbaciones causadas por la modificación humana del paisaje. Muestro por primera vez que las redes simbióticas hormiga- planta son altamente especializadas y con una baja superposición en el uso de los socios. En consecuencia, este tipo de redes simbióticas fueron menos robustas tanto para la extinción de especies de hormigas y plantas en comparación con las redes no simbióticas (es decir, las interacciones entre hormigas y plantas con nectarios extraflorales). Además, el pH del suelo fue un factor importante para estructurar el patrón anidado en las redes no simbióticas. Sin embargo, a diferencia de las redes binarias, las redes cuantitativas fueron frecuentemente no- anidadas. También muestro que aunque las hormigas tienen un efecto marcadamente territorial sólo cerca de sus nidos, la competencia entre las hormigas es fuerte y lo suficiente para estructurar las redes no simbióticas involucrando hormigas y plantas. Específicamente, muestro que la posición de las hormigas dentro de la red anidada puede predecirse sólo por las diferencias entre la capacidad competitiva (dominancia numérica y reclutamiento) de las especies de hormigas. En resumen, mis resultados en esta tesis doctoral ponen de relieve la importancia del nivel de especialización y de factores abióticos y bióticos en el mantenimiento de la estructura de las redes mutualistas hormiga-planta.

Palabras-clave: interacciones hormiga-planta, jerarquía de dominación, redes mutualistas, modularidad, anidamiento y robustez. 9 1.) General introduction

A central goal of ethology is to understand the causes, functions, development, and evolution of feeding behavior (Dugatkin 2004; Krebs and Davies 2009). Foraging is fundamental is much more than a simple matter of find, obtain and consume food (Stephens et al. 2007). Successful foraging is essential for the survival and reproduction of an organism, mainly because it influences an animal's ability to forage in such a way as to maximize their energy intake per unit time (e.g., Optimal Foraging Theory) (MacArthur and Pianka 1966;

Morse and Fritz 1987). Moreover, trophic interactions affect all attributes of ecosystems and play an important role in the stability and diversity of communities and populations in space and time (Del-Claro and Torezan-Silingardi 2009; Ohgushi et al. 2007; Tylianakis et al.

2007). Since the 1970s, behavioral ecologists have studied a set of models to test their null hypothesis that animals forage randomly (Charnov 1976; Goss-Custard 1977; Krebs et al.

1978). Nevertheless, owing to the need to consider multiple determinants in the structure and dynamics of natural communities, the factors influencing the outcome of feeding behavior is far from fully understood, mainly because it varies within populations and communities

(Bolnick et al. 2002; Petchey et al. 2009).

Traditionally, feeding relationships have been studied as small groups of species

(Mendelson 1975; Coelho et al. 1997). However, some recent studies have used tools based on the theory of complex networks to simultaneously evaluate the trophic interactions among different species (Ings et al. 2009; Donatti et al. 2011; Mello et al. 2011). A network analysis uses a set of mathematical abstractions to identify and connect links among species that are characterized by different feeding patterns, allowing us a very rich and detailed graphical visualization of large-scale datasets (Jordano et al. 2003; Olesen et al. 2007). In these networks, for example, plant and animal species can be represented as nodes and their feeding interactions are depicted by links describing the use of plant species by animal species (Figure

1A) (Pires et al. 2011). In addition, these feeding relationships can be also viewed in an 10 adjacency matrix A, where Aij= 1 if the consumption of a resource from a plant species j by the animal species i was recorded, and zero otherwise (Figure 1B) (Bascompte et al. 2003). In fact, the use of network analysis allows us to describe more clearly different patterns of interactions among trophic levels and to detect non-random patterns in the use of food resources by animals (Bascompte et al. 2003; Thébault and Fontaine 2010). Therefore, this approach identifies the role of each species within a food web based on the roles of all species within a community.

Figure 1. Main ways to view the structure of a network involving two trophic levels, A) bipartite graphs: and B) adjacency matrices. In both graphics, ant and plant species are ordered according to nestedness (i.e., number of links) (see text for more information).

A useful system to study questions on trophic relationships and resource utilization in ecological networks is ant-plant mutualisms (Dáttilo et al. 2014). Ant-plant interactions are 11 common in tropical rainforests, in which more than 94% of arthropods and 86% of the biomass collected in canopies consists of ants (Majer 1991; Tobin 1995). In tropical environments, ant diversity is extremely high, reaching approximately 500 species at local scales (Vasconcelos and Delabie 2000; Longino et al. 2002). Because of both their abundance and diversity, it is extremely common to see ants foraging on plants (Rico-Gray and Oliveira

2007). One explanation for the high frequency of ant foraging on the surface of plants is the good availability of different food and nesting sites within their structures (Andersen 1990;

Blüthgen et al. 2000; Davidson et al. 2003). Ants and plants can interact positively in different ways, from facultative to highly specialized relationships (Rico-Gray and Oliveira

2007; Dáttilo et al. 2009). For instance, some plants with extrafloral nectaries (EFN-bearing plants) produce a liquid rich in carbohydrates and amino acids, which attracts different ant species (Koptur et al. 1998), that in exchange for food, some ants can protect the host plant against potential herbivores (Rico-Gray and Oliveira 2007) (Figure 2A). On the other hand, in symbiotic ant-plant interactions, plants known as myrmecophytes provide nesting sites in cavities called domatia, and often food to their resident ant colonies (Mckey and Davidson

1993; Leroy et al. 2008) (Figure 2B). In contrast to nonsymbiotic ant-plant interactions that have a variety of interchangeable partners, both myrmecophyte and ants are highly specialized, involving only one or a few partners (Blüthgen et al. 2007; Guimarães et al.

2007). Therefore, over spatial and temporal scales different ant and plant species can interact with each other and generate complex ecological networks of interactions (Guimarães et al.

2006).

12

Figure 2. Different kind of ant-plant interactions studied in this doctoral thesis. A) nonsymbiotic interaction: workers of Crematogaster sp. (Myrmicinae) feeding on an extrafloral nectar of Inga sp. (Mimosaceae), and B) symbiotic interaction: workers of

Pseudomyrmex ferrugineus (Pseudomyrmicinae) patrolling on a domatia of Acacia cornigera

(Fabaceae) (Photo credit: Wesley Dáttilo).

Focused on the structure of ant-EFN-bearing plant networks, some studies have recently described several patterns of ant-plant interactions, such as: a nested pattern and an average level of network specialization (Guimarães et al. 2006; Blüthgen et al. 2007; Díaz-

Castelazo et al. 2010; Dáttilo et al. 2013). This indicates that within an ant-plant network there is a core of generalist species (those with the most interactions), which interact among themselves, and specialists species (those with fewer interactions) also interacting with the generalist species in cohesive subgroups (Figure 1A and 1B) (Bascompte et al. 2003).

However, despite the importance and increasing knowledge about the topological structure of ant-EFN-bearing plant networks, little is known about how different biotic and abiotic factors structuring ecological networks involving ants and both EFN-bearing plants and myrmecophytes (Rico-Gray et al. 2012; Lange et al. 2013). Morevoer, no study directly evaluated the robustness of such networks to species extinctions and disturbances caused by 13 human modification of the landscape.

In order to understand the existing gaps in the literature on ant-plant networks, my main aims of this doctoral thesis were, to: 1) describe the topological structure of symbiotic networks involving ants and myrmecophytes, 2) evaluate the tolerances of different kinds of ant-plant networks to species extinctions, 3) test whether or not the soil pH and canopy cover affect the qualitative nested pattern of interactions in ant–plant networks, 4) address the effect of interaction frequency in the nested pattern of ant-plant networks, and 5) evaluate the role of ants competition structuring the nested pattern in an local ant-plant network. My thesis is divided into five interrelated chapters. Initially and in Chapter 1, my prediction was that a modular pattern is indeed present in symbiotic ant–myrmecophyte networks, and that this pattern is an invariant property of the forest environment in which the networks occur. In

Chapter 2, I hypothesized that due to highly specialized and compartmentalized interactions between ants and myrmecophytes, these symbiotic ant-plant networks are more vulnerable to species extinction compared to nonsymbiotic ant-plant networks. In Chapter 3, I hypothesize that once soil pH and canopy cover can directly influence the amount and quality of nectar, both factors can therefore affect nestedness in ant–plant mutualistic networks. In Chapter 4, I addressed the following question: Does the nested pattern in ant-plant networks differ between binary or quantitative data? To try to answer this question, I performed nestedness analysis in ant-plant networks from different ecosystems around the world and compared two different nestedness metrics (qualitative and quantitative). Finally, in Chapter 5, I postulated that due to the important role of competition in structuring many ant communities around the world, competition among ants could also contribute to the structure of ecological networks involving ants and plants with extrafloral nectaries.

14 Literature cited

Andersen, A.N. (1990). The use of ant communities to evaluate change in Australian

terrestrial ecosystems: a review and a recipe. Proceedings of the Entomological

Society of Australia 16: 347–357.

Bascompte, J.; Jordano, P.; Melián, C.J. & Olesen, J.M. (2003). The nested assembly of

plant–animal mutualistic networks. Proceedings of the National Academy of Sciences

USA 100: 9383–9387.

Blüthgen, N.; Verhaagh, M.; Goitía, W.; Jaffé, K.; Morawetz, W. & Barthlott, W. (2000).

How plants shape the ant community in the Amazonian rainforest canopy: the key

role of extrafloral nectaries and homopteran honeydew. Oecologia 125: 229–240.

Blüthgen, N.; Menzel, F.; Hovestadt, T.; Fiala, B. & Blüthgen, N. (2007). Specialization,

constraints, and conflicting interests in mutualistic networks. Current Biology 17:

341-346.

Bolnick, D.I.; Yang, L.H.; Fordyce, J.A.; Davis, J.M. & Svanbäck, R. (2002). Measuring

individual-level resource specialization. Ecology 83: 2936-2941.

Charnov, E.L. (1976). Optimal foraging: the marginal value theorem. Theoretical Population

Biology 9: 129–136

Coelho, M.M.; Martins, M.J.; Collares-Pereira, M.J.; Pires, A.M. & Cowx, I.G. (1997). Diet

and feeding relationships of two Iberian cyprinids. Fisheries Management and

Ecology 4: 83-91.

Dáttilo, W.; Marques, E.C.; Falcão, J.C.F. & Moreira, D.D.O. (2009). Interações

mutualísticas entre formigas e plantas. EntomoBrasilis 2: 32-36.

Dáttilo, W.; Guimaraes, P.R. & Izzo, T.J. (2013). Spatial structure of ant-plant mutualistic

networks. Oikos 122: 1643–1648

Dáttilo, W.; Marquitti F.M.D.; Guimarães, P.R. & Izzo, T.J. (2014). The structure of ant-plant

ecological networks: is abundance enough?. Ecology 95: 475–485. 15 Davidson, D.W.; Cook, S.C.; Snelling, R.R. and Chua, T.H. (2003). Explaining the

abundance of ants in lowland tropical rainforest canopies. Science 300: 969–972.

Del-Claro, K. & Torezan-Silingardi, H.M. (2009). Insect-plant interactions: new pathways to

a better comprehension of ecological communities in Neotropical savannas.

Neotropical Entomology 38: 159-164

Díaz-Castelazo, C.; Guimaraes, P.R.; Jordano, P.; Thompson, J.N.; Marquis, R.J. & Rico-

Gray, V. (2010). Changes of a mutualistic network over time: reanalysis over a 10-

year period. Ecology 91: 793-801.

Donatti, C.I.; Guimaraes, P.R.; Galetti, M.; Pizo, M.A.; Marquitti, F.M.D. & Dirzo, (2011).

Analysis of a hyper-diverse seed dispersal network: modularity and underlying

mechanisms. Ecology Letters 14: 773-781.

Dugatkin, L. A. (2004). Principles of animal behavior. WW Norton Press.

Goss-Custard, J.D. (1977). Optimal foraging and the size selection of worms by redshank

Tringa totanus. Animal Behaviour 25: 10-29.

Guimaraes, P.R.; Rico-Gray, V.; Dos Reis, S.F. & Thompson, J.N. (2006). Asymmetries in

specialization in ant-plant mutualistic networks. Proceedings of the Royal Society of

London B. 273: 2041-2047.

Guimaraes, P.R.; Rico-Gray, V.; Oliveira, P.S.; Izzo, T.J.; Dos Reis, S.F. & Thompson, J.N.

(2007). Interaction intimacy affects structure and coevolutionary dynamics in

mutualistic networks. Current Biology 17: 1797-1803.

Ings, T.C.; Montoya, J.M.; Bascompte, J.; Blüthgen, N.; Brown, L.; Dormann, C.F.; Edwards,

F.; Figueroa, D.; Jacob, U.; Jones, J.I.; Lauridsen, R.B.; Ledger, M.E.; Lewis, H.M.;

Olesen, J.M.; van Veen, F.J.F.; Warren, P.H. & Woodward, G. (2009). Ecological

networks – beyond food webs. Journal of Animal Ecology 78: 253–269.

Jordano, P.; Bascompte, J. & Olesen, J.M. (2003). Invariant properties in coevolutionary

networks of plant–animal interactions. Ecology Letters 6: 69-81. 16 Koptur, S.; Rico–Gray, V. & Palacios–Rios, M. (1998). Ant protection of the nectaried fern

Polypodium plebeium in central Mexico. American Journal of Botany 85: 736–739.

Krebs, J.R.; Kacelnik, A. & Taylor, P. (1978). Test of optimal sampling by foraging great tits.

Nature 275: 27-31.

Krebs, J. R. & Davies, N.B. (2009). Behavioural ecology: an evolutionary approach. John

Wiley & Sons.

Lange, D.; Dáttilo, W. & Del-Claro, K. (2013). Influence of extrafloral nectary phenology on

ant–plant mutualistic networks in a neotropical savanna. Ecological Entomology 38:

463-469.

Leroy, C.; Jauneau, A.; Quilichini, A.; Dejean, A. & Orivel, J. (2008). Comparison between

the anatomical and morphological structure of leaf blades and foliar domatia in the

ant-plant Hirtella physophora (Chrysobalanaceae). Annals of Botany 101: 501-507

Longino, J.T.; Coddington, J. & Colwell, R.K. 2002. The ant fauna of a tropical rain forest:

estimating species richness in three different ways. Ecology 83: 689–702.

MacArthur, R.H. & Pianka, E.R. (1966). On optimal use of a patchy environment. American

Naturalist 100: 603-609.

Majer, J. D. & Camer–Pesci, O. (1991). Ants species in tropical Australian tree crops and

native ecosystems – is there a mosaic? Biotropica 23: 173–181.

Mello, M.A.R; Marquitti, F.M.D.; Guimaraes, P.R.; Kalko, E.K.V.; Jordano, P. & de Aguiar,

M.A.M. (2011). The modularity of seed dispersal: differences in structure and

robustness between bat - and bird - fruit networks. Oecologia 167: 131-140

Mendelson, J. (1975). Feeding relationships among species of Notropis (Pisces: Cyprinidae)

in a Wisconsin stream. Ecological Monographs 45: 199-232.

Mckey, D. & Davidson, D.W. (1993). Ant-plant symbioses in Africa and the Neotropics:

history, biogeography and diversity. In: Goldblatt, P. (ed). Biological relationships

between Africa and South America. Yale University Press, pp. 568-606. 17 Morse, D.H. & Fritz, R.S. (1987). The consequences of foraging for reproductive success. In:

Kamil, A.C.; Krebs, J.R. & Pulliam, H.R. (eds). Foraging behavior. Plenum Press,

pp. 443-455.

Ohgushi, T.; Craig, T.P. & Price, P.W. (2007). Ecological communities. Cambridge

University Press.

Olesen, J.M.; Bascompte, J.; Dupont, Y.L. & Jordano, P. (2007). The modularity of

pollination networks. Proceedings of the National Academy of Sciences 104: 19891-

19896.

Petchey, O.L.; Morin, P.J. & Olff, H. (2009). The topology of ecological interaction

networks: the state of the art. In: Verhoef, H.A. & Morin, P.J. (eds). Community

Ecology: Processes, Models, and Applications. Oxford University Press, pp. 7-22.

Pires, M.M.; Guimaraes, P.R.; Araujo, M.S.; Giaretta, A.A.; Costa, J.C.L.; dos Reis, S.F.

(2011). The nested assembly of individual-resource networks. Journal of Animal

Ecology 80: 893-903

Rico–Gray, V. & Oliveira, P.S. (2007). The ecology and evolution of ant–plant interactions.

University of Chicago Press.

Rico-Gray, V.; Díaz-Castelazo, C.; Ramirez-Hernandez, A.; Guimaraes, P.R.; Holland, J.N.

(2012). Abiotic factors shape temporal variation in the structure of an ant-plant

network. Arthropod-Plant Interactions 6: 289-295

Stephens, D.W.; Brown, J.S. & Ydenberg, R.C. (2007). Foraging: behavior and ecology.

University of Chicago Press.

Thébault, E. & Fontaine, C. (2010). Stability of ecological communities and the architecture

of mutualistic and trophic networks. Science 329: 853-856.

Tobin, J.E. (1995). Ecology and diversity of tropical forest canopy ants. In: Lowman, M.D.

and Nadkarni, N.M. (eds). Forest canopies. San Diego Academic Press, pp. 129–

147. 18 Tylianakis, J.M.; Tscharntke, T. & Lewis, O.T. (2007). Habitat modification alters the

structure of tropical host-parasitoid food webs. Nature 445: 202-205.

Vasconcelos, H. & Delabie, J.H.C. (2000). Ground ant communities from central Amazonia

forest fragments. In: Agosti, D.; Majer, J.D.; Alonso, L. & Schulyz, T. (eds).

Sampling ground-dwelling ants: case studies from the world's rain forests. Curtin

University School of Environmental Biology Press, pp. 59–70.

19

Chapter I

Dáttilo, W.; Izzo, T.J.; Vasconcelos, H.L.; Rico-Gray, V. (2013). Strength of the

modular pattern in Amazonian symbiotic ant-plant networks. Arthropod-Plant

Interactions 7: 455-461. Arthropod-Plant Interactions (2013) 7:455–461 DOI 10.1007/s11829-013-9256-1

ORIGINAL PAPER

Strength of the modular pattern in Amazonian symbiotic ant–plant networks

Wesley Da´ttilo • Thiago J. Izzo • Heraldo L. Vasconcelos • Vı´ctor Rico-Gray

Received: 15 November 2012 / Accepted: 2 April 2013 / Published online: 21 April 2013 Ó Springer Science+Business Media Dordrecht 2013

Abstract In natural environments, distinct species inter- species (Links = 0.68 ± 0.15) and network connectance act with differing level of specialization, and thereby (C = 0.31 ± 0.17) among the different types of forest. generate complex interaction networks. Recent studies Furthermore, the composition of ants and myrmecophytes have shown that certain antagonistic and mutualistic net- did not differ among forest types (ANOSIM; P [ 0.05), works have a highly modular structure. This indicates that reinforcing the idea that this symbiotic interaction can be within an ecological network, densely connected semi- resistant to different local and landscape environmental independent compartments occur and reflect different lev- factors. In summary, this study contributes to a better els of the structural and functional system. Despite strong understanding of the way biodiversity is organized in evidence for the existence of a modular structure in obli- obligate and symbiotic ant–plant mutualisms. gate and symbiotic networks involving ants and myrm- ecophytes, it is still uncertain whether these networks are Keywords Ecological networks Ant–myrmecophyte modular. Here, we used the modularity index (M)to interactions Compartment Modularity evaluate whether symbiotic ant–myrmecophyte networks exhibit a modular pattern of interactions. The analyses were based on a large data set of ant–myrmecophyte net- Introduction works collected in mature and secondary forests of the Brazilian Central Amazon. Our results indicate that sym- Within a natural environment different species can interact biotic ant–myrmecophyte networks are highly modular with each other in different ways and generate complex (M = 0.53 ± 0.13) regardless of forest type. Using other ecological networks of interactions (Lewinsohn et al. 2006; network descriptors, we found no difference in links per Da´ttilo et al. 2013). Such interactions may vary in the level of dependence, from facultative to highly specialized relationships (Thompson 2005). Focusing on the topolog- Handling Editor: Kris Wyckhuys. ical structure of these ecological networks, recent studies have provided important insights on non-random patterns & W. Da´ttilo ( ) V. Rico-Gray of species interaction in a wide range of ecosystems Instituto de Neuroetologı´a, Universidad Veracruzana, UV, Av. Dr. Luiz Castelazo s/n, Col. Industrial–Animas: 9119, (Bascompte et al. 2003; Lewinsohn et al. 2006; Hagen 91190 Xalapa, Veracruz, Mexico et al. 2012). e-mail: [email protected] One of these non-random patterns, results from the fact that antagonistic and mutualistic networks are extremely T. J. Izzo Lab. de Ecologia de Comunidades, Departamento de Ecologia e modular (Prado and Lewinsohn 2004; Olesen et al. 2007; Botaˆnica, Universidade Federal de Mato Grosso, Cuiaba´, Kratochwil et al. 2009; Mello et al. 2011). This indicates Mato Grosso 78068-900, Brazil that within an ecological network, there are groups or modules of species of one trophic level that interact more H. L. Vasconcelos Instituto de Biologia, Universidade Federal de Uberlaˆndia, frequently with a group of species of another trophic level, Uberlaˆndia, Minas Gerais 38400-902, Brazil with few interactions among the subgroups of the network 123 456 W. Da´ttilo et al.

(Newman and Girvan 2004; Lewinsohn et al. 2006; Olesen study was to show that symbiotic networks involving ants et al. 2007). Pollination networks are the most notable and myrmecophytes could have a modular pattern of example of modularity in ecological networks (Olesen interaction, and that this pattern is an invariant property of et al. 2007) in which species that are adapted to particular the forest environment in which the networks occur. groups of pollinators (Faegri and van der Pijl 1979) could converge on correlated suites of traits (e.g., flower shape, size, color, reward type, and amount). The particular Materials and methods groups of complementary species can interact more among each other than with other species from other modules in Field observations and data collection the network (Corbet 2000; Thompson 2005; Olesen et al. 2007). Modules primarily reflect habitat heterogeneity, Field work was conducted between August 2002 and July divergent selection regimes, and phylogenetic clustering of 2003 in the 20 km 9 50 km study area of the Biological closely related species (Lewinsohn et al. 2006). Thus, the Dynamics Of Forest Fragments Project (BDFFP) located existence of modules in biological communities can ca. 80 km north of Manaus, Amazonas, Brazil (548500W, describe different levels of structural and functional sys- 028250S). The climate is tropical humid (Afi, Ko¨ppen) with tems (Hintze and Adami 2008). an average annual temperature of 26.7 °C, 85 % humidity Myrmecophytes are plants that provide nesting sites in and 2200 mm of precipitation, with periods of rain between specialized structures (domatia) and often also food to their November and May and a dry season between June and resident ant colonies. The ants, in exchange for the October (Lovejoy and Bierregaard 1990; Laurance 2001). resource provided, often protect their host plants against The main vegetation in this area is mature (or primary) herbivores (Benson 1985; Mckey and Davidson 1993; tropical terra-firme rainforest, characterized by canopy Leroy et al. 2008;Da´ttilo et al. 2009a). In these mutualistic trees reaching 30–40 m in height and some emergent trees interactions, both plants and ants are highly specialized, reaching up to 50 m. The understory is relatively open and usually involving only a few partners (Benson 1985; contains several species of palms (Lovejoy and Bierregaard Blu¨thgen et al. 2007; Guimara˜es et al. 2007; Vicente et al. 1990). In addition to semi-abandoned cattle pastures, two 2012). Moreover, observations of a variety of ant–plant types of secondary (regenerating) forests are also found in systems have shown that the number of unoccupied the BDFFP study area. One secondary forest is dominated myrmecophytes is extremely low, indicating a high sta- by trees in the genus Cecropia (Cecropiaceae), and the bility in this symbiotic system (Bruna et al. 2005; Passmore other is dominated by trees in the genus Vismia (Clusia- et al. 2012). In order to maintain such specificity and avoid ceae). The latter was established in former pasture areas, cheaters over evolutionary time, different filters of mutu- whereas Cecropia-dominated secondary forests grow in alism were developed by both partners of this system to sites where pasture establishment failed. Cecropia-domi- select and recognize their specific hosts (Davidson et al. nated forests have a taller canopy, and the understory has a 1989; Ferdele et al. 1997;Da´ttilo et al. 2009b). much higher species richness of plants than the Vismia- Whether or not networks involving ants and myrmeco- dominated stands (Mesquita et al. 2001). phytes preset a modular structure is still a matter of debate. We established a total 27 transects of 500 m 9 6m Evidence in favor of a modular structure was presented by each. Eight transects were located in mature forest, nine in Fonseca and Ganade (1996). Recently, however, Da´ttilo Cecropia-dominated secondary forests, and 10 in Vismia- (2012) showed that ant–myrmecophyte networks are not dominated secondary forests dominated by Vismia. Tran- significantly modular, although only one network was sects were separated by a minimum distance of 1 km analyzed. Here we used a much larger data set of ant– between them (the maximum distance was 40.7 km). myrmecophyte networks to address the following question: Transects location was determined by accessibility and Are ant–myrmecophyte networks modular? We hypothe- chance, since the area was not previously known before the sized that due to the relatively high specificity in these ant– transect was marked (Fig. 1). Two people (TJI and one plant associations and low overlap in the use of partners, trained field assistant) walked each transect and recorded the interactions remain stable over space and time (Fonseca all myrmecophytes inhabited by ants. As ant nests in and Ganade 1996; Bruna et al. 2005; Guimara˜es et al. myrmecophytes have a stable relationship with the plant 2007; Orivel et al. 2011; Passmore et al. 2012). Therefore, along the year, visits were done only once. For each plant our prediction was that a modular pattern is indeed present we opened approximately seven domatia to confirm that in symbiotic ant–myrmecophyte networks. In order to test the ants present were indeed inhabiting the plant and not this hypothesis, we recorded the occurrences of ants nest- just foraging in the plant foliage. For Azteca spp., the queen ing on myrmecophytes in different forest ecosystems of each colony also was collected to facilitate the identi- within the Brazilian Central Amazon. The main aim of our fication of the ant occupant. Specimens were identified to 123 Modularity in ant–myrmecophyte networks 457 species or morphospecies level. Vouchers were deposited interactions) of plant and ant species (Bascompte et al. in the Entomology Collection of the Brazilian National 2003). Although the M index is used for unipartite net- Institute for Amazonian Research (INPA) in Manaus, works, our null models control any potential effects of Brazil. bipartite structure on modularity (Pires et al. 2011). We also calculated for each network the mean number Network topology and statistical analysis of links for ant and plant species, and the proportion of possible interactions that are actually carried out. We used We used the modularity index (M) based on simulated the connectance index: annealing (SA) (range 0–1) (Guimera` et al. 2004; Guimera` I C ¼ and Amaral 2005) to estimate the degree in which groups ðA PÞ of species (ants and myrmecophytes) interact more among each other than with species from other subgroups in the where I is the total number of interactions observed, A is network (Newman and Girvan 2004). We used the SA the number of ant species, and P is the number of plant algorithm because it is the most effective existing method species (Jordano 1987). 2 to estimate the modularity in ecological networks (Olesen We used maximum likelihood v tests (G tests) in order to et al. 2007). The M index decreases when the fraction of compare the frequency of occurrence of significantly mod- links among modules increases in the total network. High ular networks in ant–myrmecophyte networks. We also used values of M indicate that different ant and plant species an analysis of variance (one-way ANOVA) to test for dif- form modules that are semi-independent from other mod- ferences among the observed values of connectance, links ules (Olesen et al. 2007; Fortuna et al. 2010). The M index per species, modularity, and number of modules in the three is calculated as follows: forest environments studied. In addition, we tested the dif- "# ference in the ant and myrmecophyte species composition XNm l d 2 1 M ¼ s s ; M 2 0; 1 among the three environments through a permutation test L 2L N s¼1 m (10,000 permutations) using Analysis of Similarities (ANOSIM) based on Sørensen’s similarity index. We chose where Nm is the number of modules, L is the total number a qualitative index in order to not overestimate the impor- of links in the network, ls is the number of links between tance of species with lower dispersal abilities. All statistical species in module s, and ds is the sum of the degrees of all analyzes were done using the R–software. species in module s (Olesen et al. 2007). We tested the significance of the observed M index for each network with 1,000 simulated networks generated by a null model (Null Results Model II; Bascompte et al. 2003) using a simple Z test (Santos et al. 2012). This allowed us to assess whether the We collected 583 myrmecophyte individuals from nine value of M observed in the empirical network was higher species in six plant families. These plants were colonized than expected for networks of equal size and with similar by 36 ant species from six subfamilies (Table 1 and 2). heterogeneity in interactions among species. In this null Although overall (i.e., combining data from all transects) model, the probability of an interaction occurring is pro- ant–myrmecophyte networks were significantly modular portional to the level of generalization (mean number of (v2 = 6.259; DF = 1; P = 0.01), only 74.07 % (n = 20)

Fig. 1 Map of the Biological Dynamics of Forest Fragments Project (BDFFP) study area located in the Brazilian Central Amazon showing the spatial arrangement of our 27 transects established in three different environments (Vismia- dominated secondary forests, Cecropia-dominated secondary forests, and mature forests)

123 458 W. Da´ttilo et al. of the individual networks exhibited this pattern. In mature Table 2 Ant species recorded in the three forest environments forests all networks (n = 8) were significantly modular, (VSF = Vismia secondary forest; CSF = Cecropia secondary forest; whereas in Cecropia- and Vismia-dominated secondary MF = Mature forest) within the Brazilian Central Amazon, during the period from August 2002 to July 2003 (see text for more forests 55.5 % (n = 5) and 80 % (n = 8) of the networks information) were significantly modular, respectively. However, these Forest type differences in the frequency of significant modular net- works among the three types of forest studied were not Species (Subfamily) Code VSF CSF MF significant (v2 = 1.128; DF = 2; P = 0.56). Allomerus octoarticulatus ALOC X X X Furthermore, we found no difference in links per species (Myrmicinae) (Links = 0.68 ± 0.15, mean of all networks ± SD) Allomerus septemarticulatus ALSE X X X (ANOVA: F2,24 = 0.294; DF = 2; P = 0.748), network (Myrmicinae) connectance (C = 0.31 ± 0.17) (ANOVA: F2,24 = 2.021; Azteca sp1 (Dolichoderinae) AZT1 X X X DF = 2; P = 0.154), and modularity value Azteca sp2 (Dolichoderinae) AZT2 X X (M = 0.53 ± 0.13) (ANOVA: F2,24 = 1.846; DF = 2; Azteca sp3 (Dolichoderinae) AZT3 X X X P = 0.180) among the three types of forest, even though Azteca sp4 (Dolichoderinae) AZT4 X X X the number of modules within networks was different Azteca sp5 (Dolichoderinae) AZT5 X among them (ANOVA: F2,24 = 3.469; DF = 2; P = 0.04) Azteca sp6 (Dolichoderinae) AZT6 X X X (Table 3). Similarly, we did not find differences in both Azteca sp7 (Dolichoderinae) AZT7 X ants and myrmecophyte composition among the three types Azteca sp8 (Dolichoderinae) AZT8 X of ecosystems studied (ANOSIM; for both myrmecophytes Azteca sp9 (Dolichoderinae) AZT9 X and ant composition P [ 0.05). Azteca sp10 (Dolichoderinae) AZT10 X When we combined all interactions found in each forest Azteca sp11 (Dolichoderinae) AZT11 X type, we found that the interaction between Myrcia madida Azteca sp12 (Dolichoderinae) AZT12 X (Myrtaceae) and Myrcidris epicharis (Pseudomyrmicinae) Azteca sp13 (Dolichoderinae) AZT13 X tends to be more stable when compared with the other Camponotus femoratus (Formicinae) CAFE´ X interactions (Fig. 2). Moreover, our study also indicates Camponotus sp1 (Formicinae) CAM1 X that the few interactions are species specific, and that most Cephalotes sp1 (Myrmicinae) CEP1 X interactions tend to share just a few partners (Fig. 2). Crematogaster brasiliensis CRBR X X X (Myrmicinae) Crematogaster laevis (Myrmicinae) CRLA X X X Crematogaster limata (Myrmicinae) CRLI X X Crematogaster tenuicula (Myrmicinae) CRTE X Gnamptogenys striatula GNST X Table 1 Myrmecophyte species recorded in the three forest envi- (Ectatomminae) ronments (VSF = Vismia secondary forest; CSF = Cecropia sec- Myrcidris epicharis MYEP X X X ondary forest; MF = mature forest) within the Brazilian Central (Pseudomyrmicinae) Amazon, during the period from August 2002 to July 2003 (see text Nylanderia sp1 (Formicinae) NYL1 X X X for more information) Pachycondila unidentata (Ponerinae) PAUN X X Forest type Pheidole minutula (Myrmicinae) PHMI X Species (Family) Code VSF CSF MF Pheidole sp1 (Myrmicinae) PHE1 X Pheidole sp2 (Myrmicinae) PHE2 X Cordia nodosa (Boraginaceae) CONO X X X Pseudomyrmex concolor PSCO X X Duroia saccifera (Rubiaceae) DUSA X X X (Pseudomyrmicinae) Hirtella myrmecophila HIMY X X X Pseudomyrmex nigriceps PSNI X (Chrysobalanaceae) (Pseudomyrmicinae) Hirtella physophora HIPH X X X Pseudomyrmex tenuis PSTE X (Chrysobalanaceae) (Pseudomyrmicinae) Maieta guianensis (Melastomataceae) MAGU X X Solenopsis sp1(Pseudomyrmicinae) SOL1 X X Myrcia madida (Myrtaceae) MYMA X X X Solenopsis sp2 (Pseudomyrmicinae) SOL2 X Palicourea corymbifera (Rubiaceae) PACO X X X Solenopsis sp3 (Pseudomyrmicinae) SOL3 X Tachigalia sp1 (Caesalpiniaceae) TAC1 X X X Wasmannia auropunctata WAAU X X Tococa bullifera (Melastomataceae) TOBU X X X (Pseudomyrmicinae)

123 Modularity in ant–myrmecophyte networks 459

Table 3 Topological structure of ant–myrmecophyte networks studied in three forest environments in the Brazilian Central Amazon Network Vismia Cecropia Mature metrics secondary forest secondary forest forest

No. of plant 4.91 ± 1.19 4.11 ± 1.96 7.13 ± 2.07 species No. of ant 6.12 ± 2.33 4.77 ± 1.64 8.62 ± 3.24 species Links per 0.72 ± 0.22a 0.66 ± 0.12a 0.68 ± 0.11a species Connectance 0.32 ± 0.18a 0.36 ± 0.16a 0.21 ± 0.14a Modularity 0.51 ± 0.08a 0.48 ± 0.21a 0.57 ± 0.11a No. of 2.01 ± 0.66 1.77 ± 0.83 2.75 ± 0.88a modules a Means followed by the same letter are not statistically different using Tukey’s test (5 % probability level)

Discussion

Using the SA algorithm we show for the first time that symbiotic networks involving ants and myrmecophytes are highly modular. Our results agree with previous studies where ant–myrmecophyte networks were formed by iso- lated groups of species (Fonseca and Ganade 1996; Blu¨thgen et al. 2007; Guimara˜es et al. 2007;Da´ttilo 2012). The high modularity found in these networks is probably due to the elevated degree of specialization and compart- mentation found in obligate and symbiotic mutualisms (Thompson 2005; Blu¨thgen et al. 2007; Guimara˜es et al. 2007). Our main goal was to demonstrate that these highly specialized associations result in cohesive associations within modules, regardless of the forest habitat (and therefore of the ecological conditions) where the associa- tions occur. Several studies have shown that regardless of local and temporal variations, the compartments found in symbiotic ant–myrmecophyte networks are partially explained by the occurrence of phylogenetically related species with similar ecological roles (Benson 1985; Davidson et al. 1989; Fig. 2 Graphical representation of ant–myrmecophyte networks Longino 1989; Vasconcelos 1991; Ward 1991; Fonseca found in a Vismia-dominated secondary forests, b Cecropia-domi- and Ganade 1996). Thus, the modules in these networks nated secondary forests, and c mature forests. Each circle represents one ant or myrmecophyte species, which are represented by black and could reflect the different adaptations for selection of white colors, respectively. Lines represent ant–myrmecophyte inter- specialist ants (mutualism filters) that prevent the coloni- actions. These networks were represented as an energy two-mode zation of myrmecophytes by generalist ants (Fonseca and graph (Kamada–Kawai free method) obtained using the program Ganade 1996; Guimara˜es et al. 2007). Different plant Pajek. Plant and ant species codes are presented in Tables 1 and 2, respectively features seem to be responsible for the high levels of compartmentalization observed in ant–plant symbioses. For instance, variations in size and shape of the entrance of (Mean ± SD: 0.29 ± 0.16) when compared with the net- domatia and recognition of chemical volatiles by specialist works generated by interactions between ants and plants ant queens are considered important filters in these mutu- with extrafloral nectaries (varies between 0.13 and 0.17) alistic interactions (Ferdele et al. 1997; Quek et al. 2004; (Dı´az-Castelazo et al. 2010), indicating a higher individual Da´ttilo et al. 2009b). Additionally, we observed that the specialization among the species involved in ant–myrm- ant–myrmecophyte networks are highly connected ecophyte networks. 123 460 W. Da´ttilo et al.

Our results also indicate that the networks found in modules (Guimara˜es et al. 2007). Our study contributes to mature forests tend to have a higher number of modules the understanding of the organization of biodiversity in than those from secondary, regenerating forests. Previous obligate and symbiotic ant–plant mutualisms in tropical studies indicate that mature forests tend to have higher rainforests. myrmecophyte richness, relative to forest fragments sur- rounded by secondary forests (Bruna et al. 2005). Simi- Acknowledgments We would like to thank Je´ssica Falca˜o and three larly, here, although ca. 20 years after land abandonment, anonymous reviewers for providing useful comments on earlier draft of manuscript. We also thank the cooperative agreement between the several myrmecophyte species typical of mature forest Instituto Nacional de Pesquisas da Amazonia (INPA) and the (e.g., Hirtella myrmecophila and Cordia nodosa) were Smithsonian Tropical Research Institute (STRI) on Biological present in the secondary forests studied, overall, mature Dynamics of Forest Fragments Project (BDFFP) for logistical sup- forests contained up to 73.4 % more myrmecophyte spe- port. R. Guimera` provided the SA algorithm software, and Pavel Dodonov helped with the matrix randomization procedure in the R cies when compared to secondary forests. This may help to Language Platform. WD is grateful for financial support by the explain the higher number of modules in mature forests, as Conselho Nacional de Desenvolvimento Cientı´fico e Tecnolo´gico given the high degree of compartmentation in ant–myrm- (CNPq) and Consejo Nacional de Ciencia y Tecnologı´a (CONACYT). ecophyte systems, it is expected that with an increase in the This is publication 617 in the BDFFP technical series. number of species in the network there is also an increase in the number of modules. References A recent study in the BDFFP area (Passmore et al. (2012) indicates that the topological structure of ant– Ayala FJ, Wetterer JK, Longino JT, Hartl DL (1996) Molecular myrmecophyte networks is resistant to changes associated phylogeny of Azteca ants (Hymenoptera: Formicidae) and the with habitat fragmentation. Similarly, our results show that colonization of Cecropia trees. Mol Phylogenet Evol 5:423–428 both modularity and the composition of ants and myrm- Bascompte J, Jordano P, Melia´n CJ, Olesen JM (2003) The nested ecophytes do not differ significantly between mature and assembly of plant–animal mutualistic networks. Proc Natl Acad Sci USA 100:9383–9387 successional forests of different origins and structure. Our Benson WW (1985) Amazon ant–plants. In: Prance GT, Lovejoy TE results suggest that it may be possible that the non-random (eds) Amazonia. Pergamon Press, New York, pp 239–266 pattern of organization of ant–myrmecophyte networks Blu¨thgen N, Menzel F, Hovestadt T, Fiala B, Blu¨thgen N (2007) (see Passmore et al. 2012), including modularity (this Specialization, constraints, and conflicting interests in mutual- istic networks. Curr Biol 17:341–346 study), can be an invariant property of this symbiotic sys- Bruna EM, Vasconcelos HL, Heredia S (2005) The effect of habitat tem. These results reinforce the idea that ant–myrmeco- fragmentation on communities of mutualists: a test with phyte networks can be resistant to different local and Amazonian ants and their host plants. Biol Conserv 124:209–216 landscape environmental factors (Bruna et al. 2005; Pass- Corbet SA (2000) Conserving compartments in pollination webs. Conserv Biol 14:1229–1231 more et al. 2012). Da´ttilo W (2012) Different tolerances of symbiotic and nonsymbiotic Due to morphologically similar species, some studies ant–plant networks to species extinctions. Net Biol 2:127–138 have reported the difficulty of identification at species level Da´ttilo W, Marques EC, Falca˜o JCF, Moreira DDO (2009a) based only on workers of Azteca inhabiting myrmeco- Interac¸o˜es mutualı´sticas entre formigas e plantas. EntomoBras- ilis 2:32–36 phytes (Ayala et al. 1996; Bruna et al. 2005). Species-level Da´ttilo W, Izzo TJ, Inouye BD, Vasconcelos HL, Bruna EM (2009b) taxonomy of Azteca is still unclear (Ferna´ndez 2003), Recognition of host plant volatiles by Pheidole minutula Mayr although, however, our collections were based on both (Myrmicinae), an amazonian ant–plant specialist. Biotropica workers and queens, which made it easier to separate ants 41:642–646 Da´ttilo W, Rico-Gray V, Rodrigues DJ, Izzo TJ (2013) Soil and in morphospecies with the help of experts. Azteca (Doli- vegetation features determine the nested pattern of ant–plant choderinae) was the most common genus found interacting networks in a tropical rainforest. Ecol Entomol (in press) with myrmecophytes (Table 2), indicating the importance Davidson DW, Snelling RR, Longino JT (1989) Competition among of this genus for the topology of ant–myrmecophytes net- ants for myrmecophytes and the significance of plant trichomes. Biotropica 21:64–73 works found in the Brazilian Central Amazon. Dı´az-Castelazo C, Guimara˜es PR, Jordano P, Thompson JN, Marquis In summary, we show that symbiotic ant–myrmecophyte RJ, Rico-Gray V (2010) Changes of a mutualistic network over networks are highly modular, meaning that there are sub- time: reanalysis over a 10-year period. Ecology 91:793–801 sets of species (modules) partially independent that are Dupont YL, Olesen JM (2009) Ecological modules and roles of species in healthland plant–insect flower visitor networks. J highly connected among them. On such modules, a small Anim Ecol 78:346–353 change of traits or exclusion of one partner directly affects Faegri K, Van Der Pijl L (1979) The principles of pollination ecology. the other partner (Guimara˜es et al. 2007; Dupont and Pergamon Press, Oxford Olesen 2009;Da´ttilo 2012). However, the extinction of one Ferdele W, Maschwitz U, Fiala B, Riederer B, Ho¨lldobler B (1997) Slippery ant–plants and skilful climbers: selection and protection species in one module, or even the entire module, probably of specific ant partners by epicuticular wax blooms in Maca- does not affect the dynamics of the species in other ranga (Euphorbiaceae). Oecologia 112:217–224 123 Modularity in ant–myrmecophyte networks 461

Ferna´ndez F (2003) Introduccio´n a las hormigas de la regio´n Goldblatt P (ed) Biological relationships between Africa and Neotropical. Instituto de Investigacio´n de Recursos Biolo´gicos South America. Yale University Press, New Haven, pp 568–606 Alexander von Humboldt, Bogota´ Mello MAR, Marquitti FMD, Guimara˜es PR, Kalko EKV, Jordano P, Fonseca CR, Ganade G (1996) Asymmetries, compartments and null de Aguiar MAM (2011) The modularity of seed dispersal: interactions in an Amazonian ant–plant community. J Anim Ecol differences in structure and robustness between bat–and bird– 65:339–347 fruit networks. Oecologia 167:131–140 Fortuna MA, Stouffer DB, Olesen JM, Jordano P, Mouillot D, Mesquita RCG, Ickes K, Ganade G, Williamson GB (2001) Alter- Krasnov BR, Poulin R, Bascompte J (2010) Nestedness versus native successional pathways in the Amazon Basin. J Ecol modularity in ecological networks: two sides of the same coin? 89:528–537 J Anim Ecol 79:811–817 Newman ME, Girvan M (2004) Finding and evaluating community Guimara˜es PR, Rico-Gray V, Oliveira PS, Izzo TJ, Reis SF, structure in networks. Phys Rev E 69:025103 Thompson JN (2007) Interaction intimacy affects structure and Olesen JM, Bascompte J, Dupont YL, Jordano P (2007) The coevolutionary dynamics in mutualistic networks. Curr Biol modularity of pollination networks. Proc Natl Acad Sci USA 17:1797–1803 104:19891–19896 Guimera` R, Amaral LAN (2005) Functional cartography of complex Orivel J, Lambs L, Male PJG, Leroy C, Grangier J, Otto T, Quilichini metabolic networks. Nature 433:895–900 A, Dejean A (2011) Dynamics of the association between a long- Guimera` R, Sales-Pardo M, Amaral LAN (2004) Modularity from lived understory myrmecophyte and its specific associated ants. fluctuations in random graphs and complex networks. Phys Rev Oecologia 165:369–376 E 70:025101 Passmore HA, Bruna EM, Heredia S, Vasconcelos HL (2012) Hagen M, Kissiling WD, Rasmussen C et al (2012) Biodiversity, Resilient networks of ant–plant mutualists in Amazonian forest species interactions and ecological networks in a fragmented fragments. PLoS ONE 7:e40803 world. Adv Ecol Res 46:89–210 Pires MM, Guimara˜es PR, Arau´jo MS, Giaretta AA, Costa JCL, dos Hintze A, Adami C (2008) Evolution of complex modular biological Reis SF (2011) The nested assembly of individual–resource networks. PLoS Comput Biol 4:e23 networks. J Anim Ecol 80:893–903 Jordano P (1987) Patterns of mutualistic interactions in pollination Prado PI, Lewinsohn TM (2004) Compartments in insect–plant and seed dispersal: connectance, dependence, and coevolution. associations and their consequences for community structure. Am Nat 129:657–677 J Anim Ecol 73:1168–1178 Kratochwil A, Beil M, Schwabe A (2009) Complex structure of Quek SP, Davies SJ, Itino T, Pierce NE (2004) Codiversification in an pollinator–plant interaction webs: random, nested, with gradients ant–plant mutualism: stem texture and the evolution of host use or modules? Apidologie 40:634–650 in Crematogaster (Formicidae: Myrmicinae) inhabitants of Laurance WF (2001) The hyper–diverse flora of the Central Amazon: Macaranga (Euphorbiaceae). Evolution 58:554–570 an overview. In: Bierregaard RO, Gascon C, Lovejoy TE, Santos GMM, Aguiar CML, Genini J, Martins CF, Zanella FCV, Mesquita R (eds) Lessons from Amazonia: The ecology and Mello MAR (2012) Invasive Africanized honeybees change the conservation of a fragmented forest. Yale Universit Press, New structure of native pollination networks in Brazil. Biol Invasions Haven, pp 47–53 14:2369–2378 Leroy C, Jauneau A, Quilichini A, Dejean A, Orivel J (2008) Thompson JN (2005) The geographic mosaic of coevolution. Comparison between the anatomical and morphological structure University of Chicago Press, Chicago of leaf blades and foliar domatia in the ant–plant Hirtella Vasconcelos HL (1991) Mutualism between Maieta guianensis Aubl, physophora (Chrysobalanaceae). Ann Bot 101:501–507 a mirmecophytic Melastome, and one of its ant inhabitants—ant Lewinsohn TM, Prado PI, Jordano P, Bascompte J, Olesen JM (2006) protection against insect herbivores. Oecologia 87:295–298 Structure in plant–animal interaction assemblages. Oikos Vicente RE, Da´ttilo W, Izzo TJ (2012) New record of a very 113:174–184 specialized interaction: Myrcidris epicharis Ward 1990 (Pseudo- Longino JT (1989) Geographic variation and community structure in myrmecinae) and its myrmecophyte host Myrcia madida McV- an ant–plant mutualism: Azteca and Cecropia in Costa Rica. augh (Myrtaceae) in Brazilian Meridional Amazon. Acta Biotropica 21:121–125 Amazon 42:567–570 Lovejoy TE, Bierregaard RO (1990) Central Amazonian forests and Ward PS (1991) Phylogenetic analysis of Pseudomyrmecine ants the minimum critical size of ecosystems project. In: Gentry AH associated with domatia–bearing plants. In: Huxley CR, Cutler (ed) Four neotropical rainforests. Yale University Press, New DF (eds) Ant–plant interactions. Oxford University Press, Haven, pp 60–71 Oxford, pp 335–352 McKey D, Davidson DW (1993) Ant–plant symbioses in Africa and the Neotropics: history, biogeography and diversity. In:

123 27

Chapter II

Dáttilo, W. (2012). Different tolerances of symbiotic and nonsymbiotic ant-plant

networks to species extinctions. Network Biology 2: 127-138. Network Biology, 2012, 2(4):127-138

Article

Different tolerances of symbiotic and nonsymbiotic ant-plant networks to species extinctions

Wesley Dáttilo Institute ofl Neuroethology, Universidad Veracruzana, Av. Dr. Luiz Castelazo s/n, CP 91190, Xalapa, Veracruz, Mexico E-mail: [email protected]

Received 21 August 2012; Accepted 25 September 2012; Published online 1 December 2012 IAEES

Abstract The knowledge of the mechanisms that shape biodiversity–stability relationships is essential to understand ecological and evolutionary dynamics of interacting species. However, most studies focus only on species loss and ignore the loss of interactions. In this study, I evaluated the topological structure of two different ant-plant networks: symbiotic (ants and myrmecophytes) and nonsymbiotic (ants and plants with extrafloral nectaries). Moreover, I also evaluated in both networks the tolerance to plant and ant species extinction using a new approach. For this, I used models based on simulations of cumulative removals of species from the network at random. Both networks were fundamentally different in the interaction and extinction patterns. The symbiotic network was more specialized and less robust to species extinction. On the other hand, the nonsymbiotic network tends to be functionally redundant and more robust to species extinction. The difference for food resource utilization and ant nesting in both ant-plant interactions can explain the observed pattern. In short, I contributed in this manner to our understanding of the biodiversity maintenance and coevolutionary processes in facultative and obligate mutualisms.

Keywords mutualistic interactions; ecological networks; robustness; loss species; modularity.

1 Introduction One of the central goals in biological conservation is to understand how different ecological mechanisms shape biodiversity (Van Jaarsveld et al., 1998; Loreau et al., 2001; Groom et al., 2006). The recent increases in species extinction rates have increased the researcher’s interest as to how these losses may affect ecosystems functioning and help us in management actions towards species and ecosystem conservation (Casey and Myers, 1998; Berglind, 2000; Stuart et al., 2004; Jones et al., 2007). However, most studies have focused only on species loss and ignore loss of species interactions (Janzen, 1974; Memmott et al., 2007; Dyer, 2010; Blüthgen, 2012). Ecological interactions have an important role in the structure and stability of populations and communities over space-time (Janzen, 1974; Burslem et al., 2005; Del-Claro and Torezan-Silingardi, 2009; Dyer et al., 2010; Dormann, 2011; Nedorezov, 2011; Zhang, 2011; Elsadany, 2012; Zhang, 2012a, 2012b). Thus, understanding how and why these loss of species interactions occur is important for our current knowledge about the ecological dynamics of interacting species (Wilmers, 2007; Kaiser-Bunbury et al., 2010; Pocock et al., 2012). Ecological interactions are increasingly at risk from local and global extinction as a consequence of disturbances caused by human activities, including habitat loss, altered land use, introduction of alien species

IAEES www.iaees.org 128 Network Biology, 2012, 2(4):127-138 and climate change (Kearns et al., 1998; Biesmeijer et al., 2006; Zhang et al., 2006; Sayadi and Sayyed, 2011; Zhang and Chen, 2011; Zhang and Liu, 2012; Zhang and Wu, 2012; Zhang and Zhang, 2012). Recently, studies about ecological networks have provided important insights into mechanisms that contribute to the stability and structural organization of species interactions at community level (Medan et al., 2007; Morales and Vázquez, 2008; Nielsen and Bascompte, 2007; Rezende et al., 2007; Stang et al., 2007; Vázquez et al., 2007, 2009; Zhang, 2011; Zhang, 2012a). In mutualistic networks, the extinction of one of the interaction partners can lead to coextinction of the other partner, and it has important consequences for ecological system dynamics (Wilson, 1992; Solé and Montoya, 2001; Dunne et al., 2002; Memmott et al., 2004; Dorman, 2011; Blüthgen, 2012). The extinction risk of an organism depends on the number of interactive partners, as organisms with higher number of partners are more robust to extinction (Ashworth et al., 2004; Memmott et al., 2004). Moreover, the specialization level of the organism also influences the whole ecological and evolutionary dynamics of the system (Ashworth et al., 2004; Memmott et al., 2004; Vázquez and Simberloff, 2002; Stang, 2007). A specialist interacting with a generalist is less prone to extinction (Melian and Bascompte, 2002; Memmott et al., 2004), and the loss of specialized interactions can destabilize the system (Bascompte et al., 2005; Bascompte et al., 2006; May, 1973; McCann et al., 1998; Kokkoris et al., 1999; Neutel et al., 2002). A good system in which to study questions about coextinction in mutualistic networks is the ant-plant mutualism. Ants and plants can interact positively in different ways, from facultative to highly specialized relationships (Rico-Gray and Oliveira, 2007; Dáttilo et al., 2009a). In this paper, I used as a model the symbiotic and nonsymbiotic ant-plant interactions to study the loss of interactions on mutualistic networks. In both kinds of interactions ants defend the plants against potential herbivores (Vasconcelos, 1991; Del-Claro et al., 1996; Oliveira et al., 1999; Rico-Gray and Oliveira, 2007). In symbiotic ant-plant interactions, plants known as myrmecophytes provide nesting sites in cavities called domatia, and, often food to their resident ant colonies (Benson, 1985; Mckey and Davidson, 1993; Leroy et al., 2008). On the other hand, in nonsymbiotic ant-plant interactions, plants produce nutritious liquid in their extrafloral nectaries for ants (Baker et al., 1978; Rico-Gray and Oliveira, 2007). In this case, as the resource offered by plants is seasonal over space-time, the ants do not have "fidelity" of foraging on the same plant, and therefore the interactions tend to be less specialized (Rico-Gray et al., 1998; Díaz-Castelazo et al., 2004; Schoereder et al., 2010). In contrast to nonsymbiotic ant-plant interactions that have a variety of interchangeable partners, both myrmecophyte and ants are highly specialized, involving only one or a few partners (Benson, 1985; Blüthgen et al., 2007; Guimarães et al., 2007). In some cases, the specialization degree between ants and myrmecophytes is so high, that ant queens use volatile cues to discriminate their host-plants from nonmyrmecophytic species at the time of colonization (Edwards et al., 2006; Dáttilo et al., 2009b). Here, I hypothesize that although the interaction between ants and myrmecophytes is extremely specialized and compartmentalized (Benson, 1985; Fonseca and Ganade, 1996; Guimarães et al., 2007), the symbiotic ant-plant networks are more vulnerable to species extinction compared to nonsymbiotic ant-plant networks. Moreover, I expected that the difference in natural history of symbiotic and nonsymbiotic ant-plant interactions could generate differences in the topological structure of both networks. In order to test my hypothesis, I used databases from literature about the frequency of interactions of symbiotic and nonsymbiotic ant-plant networks in two tropical rainforests.

2 Material and Methods 2.1 Datasets In literature, there are few ecological datasets about ant-plant interactions based on the frequency of partner interactions (quantitative data). Because of this, I only used two datasets in this study from literature of

IAEES www.iaees.org Network Biology, 2012, 2(4):127-138 129 symbiotic and nonsymbiotic ant-plant networks. The symbiotic study of ant-plant interactions was carried out by Davidson et al. (1989) from September through November in 1985 and 1986 in the Amazon tropical rainforest at the Parque Nacional Manu, Madre de Dios, Peru (11º52’ S, 71º22’ W). The authors walked 4.800 m of trails and recorded the occurrence of different ant species in all myrmecophyte individuals found. The nonsymbiotic study was carried out by Blüthgen et al. (2004) between September 1999 and May 2002 in the rainforest at the Australian Canopy Crane in Cape Tribulation, Far North Queensland, Australia (16º07’ S, 145º27’ E), including patches of open secondary forest. The authors collected ants consuming nectar in extrafloral nectaries during the day and night. Observed plants were haphazardly selected and irregularly distributed throughout the forest. 2.2 Network topology In order to describe the network topology of both nonsymbiotic and symbiotic ant-plant networks, I calculated the following metrics: links per species, network specialization, modularity and nestedness. I calculated the level of specialization networks using the specialization index (H2’) [ranges from zero (extreme generalization) to one (extreme specialization)]. This index is mathematically derived from the Shannon entropy, and it is based on the deviation from the expected probability distribution of the interactions (Blüthgen et al., 2006). The index is robust to changes in sampling intensity and the number of interacting species (see more details about this index in Blüthgen et al., 2006, 2007). The bipartite graphs and all metrics were made in bipartite packpage (Dormann et al., 2009) using the R-Project software version 2.15.0 (R Development Core Team, 2005). I calculated the modularity of both networks using the modularity index M (range 0-1). This index estimates the degree at which groups of species (ants and plants) interact more among each other than with species in other groups in the network (Newman and Girvan, 2004). High values of M indicate that the ants and plants form modules that are semi-independent of other interactions within the network (Olesen et al., 2007). I tested the significance of index M for each network through 1000 simulated networks generated by Null Model II (Bascompte et al., 2003), in order to assess whether the value of M observed in the empirical network is higher than expected for networks of equal size and with similar heterogeneity in interactions among species. In this null model, the probability of an interaction occurring is proportional to the level of generalization (degree) of plant and ant species (Bascompte et al., 2003). I made the null model network through a routine in MATLAB, and I calculated the M using the software Netcarto (Guimerà and Amaral, 2005). Although this index is used for bipartite network, my null models control any potential effects of bipartite structure on modularity (Pires et al., 2011). I also used the NODF index (Nestedness metric based on Overlap and Decreasing Fill) to estimate the nestedness value of networks, using ANINHADO software (Guimarães and Guimarães, 2006). This metric is a much better nestedness metric than others and less prone to type-I statistical error, since it is based on the nestedness of all pairs of columns and rows in the matrix (Almeida-Neto et al., 2008). To assess if the nestedness value observed was higher than expected by random interaction patterns, I tested the nestedness of each network with 1000 networks generated by Null Model II. 2.3 Robustness to extinctions I calculated the robustness of symbiotic and nonsymbiotic ant-plant networks to species extinction in both trophic levels (plants and ants) based on cumulative removals of species from the network at random (Burgos et al., 2007). Initially, I removed one species from one trophic level (e.g. ants), and when species from the other trophic level (e.g. plants) were connected only to the initial removed species, they was also removed from the network, indicating secondary losses. Afterwards, I was removing randomly all remaining species until all species from the trophic level chosen were removed (Mello et al., 2011). For more information about

IAEES www.iaees.org 130 Network Biology, 2012, 2(4):127-138 this procedure, please see Dormann et al. (2009). Moreover, I calculated the area under the extinction curve (R) proposed by Burgos et al. (2007) to measure the robustness of networks, where R= 1 corresponds to a very slow decrease in the curve until the point at which almost all species of one trophic level are eliminated (more robust network), and R= 0 corresponds to a very fast decrease in the curve as soon as any species is lost (less robust network). I ran 100 randomizations for each network to simulate the species removals. I chose the R index because it is more robust and it is not sensitive to the shape of the curve when compared to the index proposed by Memmott et al. (2004), called Attack Tolerance Curve (ATC).

3 Results Both networks studied here had different numbers of interacting species. The symbiotic network had 8 plant species and 18 ant species, while the nonsymbiotic network had 51 plant species and 41 ant species (see Fig. 1A-B). The nonsymbiotic network also had approximately four times more links per species (3.097) than the symbiotic network (0.807). Additionally, nestedness was higher in the nonsymbiotic network (NODF= 22.11) than in the symbiotic network (NODF= 15.96). Nonsymbiotic networks exhibited a significantly nested topology (P= 0.01). However, the nestedness value observed in symbiotic network was more equal than expected by random patterns of interaction (P= 0.999).

Symbiotic ant-plant network was more specialized (H2’= 0.926) than the nonsymbiotic network (H2’= 0.193). In both networks, I did not observe significantly higher modularity than expected by the heterogeneity of interactions (Nonsymbiotic: P= 0.999; Symbiotic: P= 0.151). However, there was lower modularity in the nonsymbiotic network (M= 0.302) than in the symbiotic network (M= 0.763). The robustness to cumulative extinctions had different patterns in nonsymbiotic and symbiotic ant-plant networks (see Fig. 2). The simulations of cumulative removals of species showed that the nonsymbiotic network is very robust for both removals of plant and ants than the symbiotic network, since their extinction curves declined more slowly. The robustness of the nonsymbiotic network was relatively high, both for plants (R= 0.680) and ants (R= 0.773) (see Fig. 2A-B). However, in the symbiotic network, the robustness was low both for plants (R= 0.449) and ants (R= 0.446) (see Fig. 2C-D).

IAEES www.iaees.org Network Biology, 2012, 2(4):127-138 131

Fig. 1 Bipartite graphs of (A) symbiotic and (B) nonsymbiotic ant-plant networks. Symbiotic network represents mutualistic interaction between ants and myrmecophytes plants (database: Davidson et al., 1989). Nonsymbiotic network represents the mutualistic interactions between ants and plants with floral and extrafloral nectaries (database: Blüthgen et al., 2004). The nodes on the left represent different plant species, and the nodes on the right correspond to ant species that interact positively with the plants. Lines, also called "links", connect positively interacting species.

IAEES www.iaees.org 132 Network Biology, 2012, 2(4):127-138

Fig. 2 Robustness to cumulative species removal of (A) plants and (B) ants in symbiotic network (database: Davidson et al., 1989), and (C) plants and (D) ants in nonsymbiotic network (database: Blüthgen et al., 2004).

4 Discussion Currently, thousands of species become extinct each year and the role of most of these species in their ecosystems will hardly ever be understood (Röckstrom et al., 2009; Lee and Jetz, 2011; Wilson et al., 2011). In megadiversity regions (i.e., tropical forests) many species depend on one another during their ontogeny and the loss of one partner can lead to coextinctions, which represents the irreplaceable loss of evolutionary history of species interacting (Boucher et al., 1982; Stachowicz, 2001; Tillberg and Breed, 2004). Here, I showed that symbiotic networks are less robust for both ants and plants species extinction compared to nonsymbiotic networks. They can generate different influences on the ecological and evolutionary dynamics of ant-plant

IAEES www.iaees.org Network Biology, 2012, 2(4):127-138 133 interaction. Moreover, I also showed the topology structure of symbiotic and nonsymbiotic ant-plant networks differ fundamentally in the interactions pattern in tropical forests. Symbiotic network showed a high level of specialization and modularity. These results corroborate with previous studies where the authors showed that symbiotic ant-plant networks are formed by isolated groups of species (Fonseca and Ganade, 1996; Blüthgen et al., 2007; Guimarães et al., 2007). However, the results demonstrated that there was low specialization and modularity in nonsymbiotic networks. The difference in specialization and modularity of these networks is possibly due to difference in the intimacy degree of ant-plant interactions. In symbiotic ant-plant interactions, the ants obligatory inhabit myrmecophytes, and during their ontogeny the number of overlapping partners is practically non-existent (Benson, 1985; Fonseca and Ganade, 1996; Heil and McKey, 2003). In nonsymbiotic interactions nectar is a seasonal resource (Rico-Gray et al., 1998; Díaz-Castelazo et al., 2004; Blüthgen et al., 2007; Schoereder et al., 2010). Thus, when a plant does not secrete nectar, the ants can use other resources available on foliage. Therefore, this kind of ant-plant interaction is less specialized and facultative (Schoereder et al., 2010). The ant's ability to change of a nectar possibly occurs because the physiological and nutritional requirements of ants that feed on EFNs are very similar (Blüthgen et al., 2007). In my results, even where the nestedness was not significant in symbiotic network, the nonsymbiotic network was more nested. It is known that nonsymbiotic ant-plant networks are a lot more nested that symbiotic network, and only 15.38 % of the symbiotic network was significantly nested. (Guimarães et al., 2007). Moreover, according to Bastolla et al. (2009) the nested pattern in mutualistic networks reduces interspecific competition enhancing the number of coexisting species. Thus, as the nonsymbiotic network is more nested than the symbiotic network it is expected that the nested pattern of nonsymbiotic networks could also be generated by the low level of specialization of ant-plant interaction. Several studies have showed the role of functional redundancy on the stability of ecological communities (Walker, 1992; Rosenfeld, 2002; Petchey et al., 2007; Joner et al., 2011). Based on the insurance hypothesis (Yachi and Loreau, 1999), systems with high functional redundancy are more resilient to disturbances (Walker, 1995; Naeem, 1998; Fonseca and Ganade, 2001). This is because different species perform similar roles in ecosystem function, and when a species is extinct other species "dampens" the system (Lawton and Brown, 1993; Rosenfeld, 2002; Mouchet et al., 2010). Here, due to high generalization of the nonsymbiotic network, it is expected that these networks are functionally redundant, because there is low specificity in this interaction. So, when I remove species from nonsymbiotic networks, the extinction curves declined more slowly compared with symbiotic networks, because the deletion of one species does not necessarily cause the deletion of other partner species. Biologically, arboreal ants do not depend exclusively on the food offered by a particular plant species and also supplement their diet with insect exudates (honeydew) (Davidson et al., 2004). Moreover, in tropical regions many honeydew tend to be more productive and spatially more concentrated when compared with extrafloral nectaries (Blüthgen et al., 2004b). Thus, when a plant species with extrafloral nectaries is extinct, the ants can substituted this resource by other liquid resources commonly found in tropical environments. On the other hand, due to the high specialization of partners in ant-plant symbiotic interactions, the change of traits or exclusion of one partner directly affects the ecological maintenance of the other partner (Guimarães et al., 2007). In addition, even using only one network for each ant-plant interaction, I believe that, the pattern found in this study will be repeated in other networks with the same differences in natural history, with more specialized ecological networks being more susceptible to species losses. Here, I showed that specialized ant-plant interactions are highly susceptible to loss of species and this pattern should occur in other plant-animal networks with a high degree of specialization. Thus, the loss of very specialized interactions can lead to a cascade effect of loss of other species and all ecological services provided

IAEES www.iaees.org 134 Network Biology, 2012, 2(4):127-138 by them. In addition, the resilience of biological communities is not based only on high functional redundancy but also on how functionally similar species respond to environmental disturbance (Elmqvist et al., 2003; Folke, 2006; Nyström, 2006; Laliberte et al., 2010). Thus, I suggest as a topic for future studies that one evaluates response diversity in different ant-plant mutualistic networks after disturbances in order to assess whether different tolerance also occurs for these networks as regards species extinction in natural environments.

Acknowledgments I would like to thank Laura Leal, Juan Serio-Silva, Gudryan Barônio, John Bagnall, Jéssica Falcão and Thiago Izzo for valuable comments and discussions on earlier versions of the manuscript. Flávia Marquitti helped with modularity analyses. WD is grateful for financial support by the CNPq and CONACYT.

References Almeida-Neto M, Guimarães-Jr P, Guimarães P, et al. 2008. A consistent metric for nestedness analysis in ecological systems: reconciling concept and measurement. Oikos, 117: 1227-1239 Ashworth L, Aguilar R, Galetto L, et al. 2004. Why do pollination generalist and specialist plant species show similar reproductive susceptibility to habitat fragmentation? Journal of Ecology, 92: 717-19 Baker HG, Opler PA, Baker I. 1978. A comparison of the amino acid complements of floral and extrafloral nectars. Botanical Gazette, 139: 322-332 Bascompte J, Jordano P, Melián CJ, et al. 2003. The nested assembly of plant-animal mutualistic networks. Proceedings of the National Academy of Sciences of USA, 100: 9383-9387 Bascompte J, Melián CJ, Sala E. 2005. Interaction strength combinations and the overfishing of a marine food web. Proceedings of the National Academy of Sciences of USA, 102:5443-5447 Bascompte J, Jordano P, Olesen JM. 2006. Asymmetric coevolutionary networks facilitate biodiversity maintenance. Science, 312: 431-433 Bastolla U, Fortuna MA, Pascual-García A, et al. 2009. The architecture of mutualistic networks minimizes competition and increases biodiversity. Nature, 458: 1018-1021 Benson WW. 1985. Amazon ant-plants. In: Amazonia (Prance GT, Lovejoy TE, eds). 239-266, Pergamon Press, New York, USA Berglind SA. 2000. Demography and management of relict sand lizard Lacerta agilis populations on the edge of extinction. Ecological Bulletins, 48: 123-142 Biesmeijer JC, Roberts SPM, Reemer M, et al. 2006. Parallel declines in pollinators and insect pollinated plants in Britain and the Netherlands. Science, 313: 351-354 Boucher DH, James S, Keeler KH. 1982. The ecology of mutualism. Annual Review of Ecological Systems, 13: 315-347 Blüthgen N, Fiedler K. 2004. Competition for composition: lessons from nectar-feeding ant communities. Ecology, 85: 1479-1485 Blüthgen N, Stork NE, Fiedler K. 2004a. Bottom-up control and co-occurrence in complex communities: honeydew and nectar determine a rainforest ant mosaic. Oikos, 106: 344-358 Blüthgen N, Gottsberger G, Fiedler K. 2004b. Sugar and amino acid composition of ant-attended nectar and honeydew sources from an Australian rainforest. Austral Ecology, 29: 418-429 Blüthgen N, Menzel F, Blüthgen N. 2006. Measuring specialization in species interaction networks. BMC Ecology, 6: 12-18 Blüthgen N, Menzel F, Hovestadt T, et al. 2007. Specialization, constraints, and conflicting interests in

IAEES www.iaees.org Network Biology, 2012, 2(4):127-138 135

mutualistic networks. Current Biology, 17: 341-346 Blüthgen N. 2012. Interação planta-animal e a importância funcional da biodiversidade. In: Ecologia das Interações plantas-animais: uma abordagem ecológico-evolutiva (Del-Claro K, Torezan-Silingardi HM, eds). 261-272, Technical Books Editora, Rio de Janeiro, Brazil Burgos E, Ceva H, Perazzo RPJ, et al. 2007. Why nestedness in mutualistic networks? Journal of Theoretical Biology, 249: 307-313 Burslem DFRP, Pinard MA, Hartley SE. 2005. Biotic Interactions In the Tropics. Their Role in the Maintenance of Species Diversity. Cambridge University Press, Cambridge, UK Casey JM, Myers RA. 1998. Near extinction of a large, widely distributed fish. Science, 28: 690-692 Dáttilo W, Marques EC, Falcão JCF, et al. 2009a. Interações mutualísticas entre formigas e plantas. EntomoBrasilis, 2: 32-36 Dáttilo W, Izzo TJ, Inouye BD, et al. 2009b. Recognition of host plant volatiles by Pheidole minutula Mayr (Myrmicinae), an amazonian ant-plant specialist. Biotropica, 41: 642-646 Davidson DW, Snelling RR, Longino JT. 1989. Competition among ants for myrmecophytes and the significance of plant trichomes. Biotropica, 21: 64-73 Davidson DW, Cook SC, Snelling RR. 2004. Liquid-feeding performances of ants (Formicidae): ecological and evolutionary implications. Oecologia, 139: 255-266 Del-Claro K, Berto V, Réu W. 1996. Effect of herbivore deterrence by ants increase fruit set in an extrafloral nectary plant Qualea multiflora (Vochysiaceae). Journal of Tropical Ecology, 12: 887-892 Del-Claro K, Torezan-Silingardi HM. 2009. Insect-plant interactions: new pathways to a better comprehension of ecological communities in Neotropical savannas. Neotropical Entomology, 38(2): 159-164 Díaz-Castelazo C, Rico-Gray V, Oliveira PS, Cuautle YM. 2004. Extrafloral nectary-mediated ant-plant interactions in the coastal vegetation of Veracruz, México: Richness, occurence, seasonality, and ant foraging patterns. Écoscience, 11: 472-482 Dormann CF. 2011. How to be a specialist? Quantifying specialisation in pollination networks. Network Biology, 1(1): 1-20 Dormann CF, Fründ J, Blüthgen N, Gruber B. 2009. Indices, graphs and null models: analyzing bipartite ecological networks. The Open Ecology Journal, 2: 7-24 Dunne J A, Williams RJ, Martinez ND. 2002. Food-web structure and network theory: the role of connectance and size. Proceedings of the National Academy of Sciences of USA, 99: 12917-12922 Dyer LA, Walla TR, Greeney HF, et al. 2010. Diversity of Interactions: a metric for studies of biodiversity. Biotropica, 42: 281-289 Edwards DP, Hassall M, Sutherland WJ, et al. 2006. Assembling a mutualism: ant symbionts locate their host plants by detecting volatile chemicals. Insectes Sociaux, 53: 172-176 Elsadany AEA. 2012. Dynamical complexities in a discrete-time food chain. Computational Ecology and Software, 2(2): 124-139 Elmqvist T, Folke C, Nystrom M, et al. 2003. Response diversity and ecosystem resilience. Frontiers in Ecology and the Environment, 1: 488-494 Folke C, Carpenter S, Walker B, et al. 2004. Regime shifts, resilience, and biodiversity in ecosystem management. Annual Review of Ecology, Evolution and Systematics, 35: 557-581 Fonseca CR, Ganade G. 1996. Asymmetries, compartments and null interactions in an Amazonian ant-plant community. Journal of Animal Ecology, 65: 339-347 Fonseca CR, Ganade G. 2001. Species functional redundancy, random extinctions and the stability of ecosystems. Journal of Ecology, 89: 118-125

IAEES www.iaees.org 136 Network Biology, 2012, 2(4):127-138

Guimerà R, Amaral LAN. 2005. Functional cartography of complex metabolic networks. Nature, 433: 895-900 Guimarães-Jr PR, Guimarães PR. 2006. Improving the analyses of nestedness for large sets of matrices. Environmental Modelling and Software, 21: 1512-1513 Guimarães PR, Rico-Gray V, Oliveira PS, et al. 2007. Interaction intimacy affects structure and coevolutionary dynamics in mutualistic networks. Current Biology, 17: 1797-1803 Groom MJ, Meffe GK, Carroll CR. 2006. Principles of Conservation Biology. Sinauer Prees, Maachusetts, Heil M, Mckey D. 2003. Protective ant-plant interactions as model systems in ecological and evolutionary research. Annual Review of Ecology, Evolution and Systematics, 34: 425-53 Janzen DH. 1974. The deflowering of Central America. Natural History, 83: 49-53 Joner F, Specht G, Müller SC, et al. 2011. Functional redundancy in a clipping experiment on grassland plant communities. Oikos, 120(9): 1420-1426 Jones ME, Jarman PJ, Lees CM, et al. 2007. Conservation management of Tasmanian devils in the context of an emerging, extinction-threatening disease: devil facial tumor disease. EcoHealth, 4: 326-337 Kaiser-Bunbury CN, Muff S, Memmott J, et al. 2010. The robustness of pollination networks to the loss of species and interactions: A quantitative approach incorporating pollinator behaviour. Ecology Letters, 13: 442-452 Kearns CA, Inouye DW, Waser BN. 1998. Endangered mutualisms: the conservation of plant-pollinator interactions. Annual Review of Ecology, Evolution and Systematics, 29: 83-112 Kokkoris GD, Troumbis AY, Lawton JH. 1999. Patterns of species interaction strength in assembled theoretical competition communities. Ecology Letters, 2: 70-74 Laliberte E, Wells JA, Declerck F, et al. 2010. Land-use intensification reduces functional redundancy and response diversity in plant communities. Ecology Letters, 13: 76-86 Lawton JH, Brown VK. 1994. Redundancy in ecosystems. In: Biodiversity and ecosystem function (Eds. ED Schulze and HA Mooney). Springer Prees, Washington, USA Lee TM, Jetz W. 2011 Unravelling the structure of species extinction risk for predictive conservation science. Proceedings of the Royal Society London B, 278 : 1329-1338 Leroy C , Jauneau A, Quilichini A, et al. 2008. Comparison between the anatomical and morphological structure of leaf blades and foliar domatia in the ant-plant Hirtella physophora (Chrysobalanaceae). Annals of Botany, 101: 501-507 Loreau M, Naeem S, Inchausti P, et al. 2001. Biodiversity and ecosystem functioning: current knowledge and future challenges. Science, 294: 804-808 May RM. 1973. Stability and Complexity in Model Ecosystems. Princeton University Press, Princeton, USA Medan D, Perazzo RPJ, Devoto M, et al. 2007. Analysis and assembling of network structure in mutualistic systems. Journal of Theoretical Biology, 246: 510-517 Melián CJ, Bascompte J. 2002. Complex networks: two ways to be robust? Ecology Letters, 5: 705-708 Mello MAR, Marquitti FMD, Guimarães PE, et al. 2011. The missing part of seed dispersal networks: structure and robustness of bat-fruit interactions. Plos One, 6: e17395 Memmott J, Waser NM, Price MV. 2004. Tolerance of pollination networks to species extinctions. Proceedings of the Royal Society London B, 271: 2605-2611 Memmott J, Craze PG, Waser NM, et al. 2007. Global warming and the disruption of plant-pollinator interactions. Ecology Letters, 10: 710-717 Mckey D, Davidson DW. 1993. Ant-plant symbioses in Africa and the Neotropics: history, biogeography and diversity. In: Biological Relationships between Africa and South America (Ed. P Goldblatt). Yale University Press, New Haven, USA

IAEES www.iaees.org Network Biology, 2012, 2(4):127-138 137

Mccann K, Hastings A, Huxel GR. 1998. Weak trophic interactions and the balance of nature. Nature, 395: 794-798 Morales JM, Vázquez DP. 2008. The effect of space in plant-animal mutualistic networks: insights from a simulation study. Oikos, 117: 1362-1370 Mouchet MA, Villeger S, Mason NWH, et al. 2010. Functional diversity measures: an overview of their redundancy and their ability to discriminate community assembly rules. Functional Ecology, 24: 867-876 Naeem S. 1998. Species redundancy and ecosystem reliability. Conservation Biology, 12: 39-45 Nedorezov LV. 2011. About a dynamic model of interaction of insect population with food plant. Computational Ecology and Software, 1(4): 208-217 Neutel A-M, Heesterbeek JAP, Ruiter PC. 2002. Stability in real food webs: weak links in long loops. Science, 296: 1120-1123 Newman M E, Girvan M. 2004. Finding and evaluating community structure in networks. Physical Review E, 69: 025103 Nielsen A, Bascompte J. 2007. Ecological networks, nestedness and sampling effort. Journal of Ecology, 95:1134-1141 Nyström M. 2006. Redundancy and response diversity of functional groups: implications for the resilience of coral reefs. AMBIO, 35(1): 30-35 Olesen JM, Bascompte J, Dupont YL, et al. 2007. The modularity of pollination networks. Proceedings of the National Academy of Sciences of USA 104: 19891-19896 Oliveira PS, Rico-Gray V, Díaz-Castelazo C, et al. 1999. Interaction between ants, extrafloral nectaries and insect herbivores in Neotropical coastal and dunes: herbivore deterrence by visiting ants increases fruit set in Opintia stricta (Cactaceae). Functional Ecology, 13: 623-631 Petchey OL, Evans KL, Fishburn IS, et al. 2007. Low functional diversity and no redundancy in British avian assemblages. Journal of Animal Ecology, 76: 977-985 Pires MM, Guimarães PR, Araujo MS, et al. 2011. The nested assembly of individual-resource networks. Journal of Animal Ecology, 80: 893-903 Pocock MJ, Evans DM, Memmott J. 2012. The robustness and restoration of a network of ecological networks. Science, 335: 973-977 R Development Core Team. 2010. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria Rezende EL, Jordano P, Bascompte J. 2007. Effects of phenotypic complementarity and phylogeny on the nested structure of mutualistic networks. Oikos, 116: 1919-1929 Rico-Gray V, Garcia-Franco JG, Palacios-Rios M, et al. 1998. Geographical and seasonal variation in the richness of ant-plant interactions in Mexico. Biotropica, 30: 190-200 Rico-Gray V, Oliveira PS. 2007. The ecology and evolution of ant-plant interactions. University of Chicago Press, Chicago, USA Röckstrom J, Steffen W, Noone K, et al. 2009. A safe operating space for humanity. Nature, 461: 472-475 Rosenfeld JS. 2002. Functional redundancy in ecology and conservation. Oikos. 98(1): 156-162 Sayadi MH, Sayyed MRG. 2011. Variations in the heavy metal accumulations within the surface soils from the Chitgar industrial area of Tehran. Proceedings of the International Academy of Ecology and Environmental Sciences, 1(1): 36-46 Schoereder JH, Sobrinho TG, Madureira MS, et al. 2010. The arboreal ant community visiting extrafloral nectaries in the Neotropical cerrado savanna. Terrestrial Arthropod Reviews, 3: 3-27 Sole RV, Montoya JM. 2001. Complexity and fragility in ecological networks. Proceedings of the Royal

IAEES www.iaees.org 138 Network Biology, 2012, 2(4):127-138

Society London B, 268: 2039-2045 Stachowicz JJ. 2001. Mutualism, facilitation, and the structure of ecological communities. BioScience, 51(3): 235-246 Stang M, Klinkhamer PGL, Van Der Meijden E. 2007. Asymmetric specialization and extinction risk in plant-flower visitor webs: a matter of morphology or abundance? Oecologia, 151: 442-453 Stuart SN, Chanson JS, Cox NA, et al. 2004. Status and trends of amphibian declines and extinctions worldwide. Science, 306: 1783-1786 Tillberg CV, Breed MD. 2004. Co-extinctions of tropical butterflies and their hostplants. Biotropica, 36(2): 272-274 Van Jaarsveld A S, Freitag S, Chown SL, et al. 1998. Biodiversity assessments and conservation strategies. Science, 279: 2106-2108 Vasconcelos HL. 1991. Mutualism between Maieta guianensis Aubl., a myrmecophytic melastome, and one of its ant inhabitants: ant protection against insect herbivores. Oecologia, 87: 295-298 Vázquez DP, Simberloff D. 2002. Ecological specialization and susceptibility to disturbance: conjectures and refutations. American Naturalist, 159: 606-623 Vázquez DP, Melián CJ, Williams NM, et al. 2007. Species abundance and asymmetric interaction strength in ecological networks. Oikos, 116: 112-120 Vázquez DP, Blüthgen N, Cagnolo L, et al. 2009. Uniting pattern and process in plant-animal mutualistic networks: a review. Annals of Botany, 103: 1445-1457 Walker BH. 1992. Biodiversity and ecological redundancy. Conservation Biology, 6: 18-23 Walker B. 1995. Conserving biological diversity through ecosystem resilience. Conservation Biology, 9: 747-752 Wilmers CC. 2007. Understanding ecosystem robustness. Trends in Ecology and Evoluttion, 22:504-506 Wilson EO. 1992. The Diversity of Life. Harvard University Press, Cambridge, USA Wilson HB, Joseph LN, Moore AL, et al. 2011. When should we save the most endangered species? Ecology Letters, 14: 886-890 Yachi S, Loreau M. 1999. Biodiversity and ecosystem productivity in a fluctuating environment: the insurance hypothesis. Proceedings of the National Academy of Sciences of USA, 96: 1463-1468 Zhang WJ, Qi YH, Zhang ZG. 2006. A long-term forecast analysis on worldwide land uses. Environmental Monitoring and Assessment, 119: 609-620 Zhang WJ. 2011. Constructing ecological interaction networks by correlation analysis: hints from community sampling. Network Biology, 1(2): 81-98 Zhang WJ. 2012a. Computational Ecology: Graphs, Networks and Agent-based Modeling. World Scientific, Singapore Zhang WJ. 2012b. How to construct the statistic network? An association network of herbaceous plants constructed from field sampling. Network Biology, 2(2): 57-68 Zhang WJ, Chen B. 2011. Environment patterns and influential factors of biological invasions: a worldwide survey. Proceedings of the International Academy of Ecology and Environmental Sciences, 1(1): 1-14 Zhang WJ, Liu CH. 2012. Some thoughts on global climate change: will it get warmer and warmer? Environmental Skeptics and Critics, 1(1): 1-7 Zhang WJ, Wu SH. 2012. Current status, crisis and conservation of coral reef ecosystems in China. Proceedings of the International Academy of Ecology and Environmental Sciences, 2(1): 1-11 Zhang J, Zhang WJ. 2012. Controversies, development and trends of biofuel industry in the world. Environmental Skeptics and Critics, 1(3): 48-55

IAEES www.iaees.org 40

Chapter III

Dáttilo, W.; Rico-Gray, V.; Domingues, D.J.; Izzo, T.J. (2013). Soil and vegetation features determine the nested pattern of ant-plant networks in a

tropical rainforest. Ecological Entomology 38: 374-380. Ecological Entomology (2013), 38, 374–380 DOI: 10.1111/een.12029

Soil and vegetation features determine the nested pattern of ant–plant networks in a tropical rainforest

WESLEY DATTILO,´ 1 V ICTOR´ RICO-GRAY,1 DOMINGOS J. RODRIGUES2 andTHIAGO J. IZZO3 1Instituto de Neutoetolog´ıa, Universidad Veracruzana, Xalapa, Veracruz, Mexico, 2Nucleo´ de Estudos da Biodiversidade da Amazoniaˆ Mato–Grossense, Universidade Federal de Mato Grosso, Sinop, Brazil and 3Departament of Ecology and Botany, Community Ecology Laboratory, Universidade Federal de Mato Grosso, Cuiaba,´ Brazil

Abstract. 1. Recently, several studies have focused on structural properties of ant–plant networks. However, little is known about the role of abiotic factors on these networks. 2. As a result of different abiotic factors that can affect the patterns of ant–plant interactions, it was tested whether soil pH and canopy cover contribute to variation in the nestedness of mutualistic (plants with extrafloral nectar–EFN) and neutral (plants without EFN) ant–plant networks. 3. It was shown that only mutualistic networks were affected by soil pH. It was suggested that this may occur because the variation in soil pH directly influences the secreted nectar, and as there is a preference for nectar composition by ants, this could change the patterns of interaction in mutualistic networks. As prey availability is possibly the main factor influencing ants’ presence on plants without EFN, soil pH should have little or no influence on the patterns of interaction in neutral networks. 4. On the other hand, nestedness was not affected by canopy cover in mutualistic and neutral networks. In spite of that canopy cover (light availability) is directly related to the amount of nectar secreted, the volume of nectar may not be important for the structure of the networks. However, canopy cover varied little in this study site. This small variation could not be enough to change the nested pattern in mutualistic and neutral networks. 5. In short, the present results show that the abiotic factors that affect the availability and quality of food resources may have important effects on the structure of trophic interactions in non-symbiotic ant–plant networks. Key words. Abiotic factors, ant–plant interactions, Brazilian Amazon, ecological networks, nestedness

Introduction (Horvitz & Schemske, 1984; Del-Claro et al., 1996; Oliveira et al., 1999; Rico-Gray & Oliveira, 2007). However, there is Many species of plants in the tropics have extrafloral nectaries a large variation in EFN quality, particularly among sites, and (EFN) (Elias, 1983; Koptur et al., 1998; Rico–Gray & different abiotic factors can change the quantity and quality of Oliveira, 2007) which secrete a liquid rich in carbohydrates the nectar secreted by plants in a community of EFN-bearing and amino acids (Baker et al., 1978; Koptur et al., 1998; plants (Heil et al., 2000; Díaz-Castelazo et al., 2004; Herrera Rico-Gray & Oliveira, 2007). This liquid attracts ants that, in et al., 2006; Rico-Gray & Oliveira, 2007). Thus, ants do show exchange for food, protect the host plant against herbivores preference for nectar with higher concentrations of carbohy- drates and amino acids (Lanza & Krauss, 1984; Bluthgen¨ & Correspondence: Wesley Dattilo,´ Instituto de Neuroetología, Uni- versidad Veracruzana–UV, Av. Dr. Luiz Castelazo s/n, Col. Industrial- Fiedler, 2004a,2004b; Rico-Gray & Oliveira, 2007). Animas: 91190, Xalapa, Veracruz, Mexico. E-mail: wdattilo@ Soil features have been suggested as one of the main factors hotmail.com affecting nectar composition (Campbell & Halama, 1993;

374 © 2013 The Royal Entomological Society Effect of abiotic factors on ant–plant networks 375

Gardener & Gillman, 2001; Heil et al., 2001; Burkle & Irwin, homopterous insects, using the plant only as foraging substrate 2008; Farkas et al., 2012). Some previous studies have shown (Andersen, 1990; Bluthgen¨ et al., 2000). These interactions that the soil nutrients directly influence carbohydrate and are different from agonistic interactions involving leaf-cutting amino acid nectar concentrations (Gardener & Gillman, 2001; ants and plants, and therefore, the presence of ants on these Mevi-Schutz¨ et al., 2003; Burkle & Irwin, 2008; Baude et al., plants reflects the spatial abundance of these species in the 2011). Nectar production by plants located in environments vegetation without the aggregation caused by the resource with soil close to a neutral pH tends to have nectar with (Bluthgen¨ et al., 2000; Yamamoto, & Del-Claro, 2008). Some higher concentrations of sugars and amino acids (Gardener & previous studies have suggested that, within a biological Gillman, 2001; Burkle & Irwin, 2008; Yao et al., 2009). This community, the difference in species abundance can generate mainly occurs because soil characteristics affect the energy nested patterns (Fischer & Lindenmayer, 2002; Lewinsohn allocated for growth and reproduction, resulting in an increase et al., 2006; Bluthgen,¨ 2010). It is expected that owing to in plant fitness and productivity (Sperens, 1997; Monaco et al., non-determinist and neutral foraging on plants without EFN, 2003). Moreover, environments with higher nutrient availabil- most abundant ant species will tend to find individuals of ity and microbial activity tend to have higher values of soil other abundant plant species more often than individuals of pH (Ekenler & Tabatabai, 2003; Yao et al., 2009). Therefore, rare species, generating the nested pattern. it is expected that soil pH will also influence nectar quality. Here we hypothesised that soil pH and canopy cover may Additionally, the amount of nectar secreted by plants directly influence the amount and quality of nectar, both of tropical rainforest understory is influenced by the light factors can therefore affect nestedness in ant–plant mutualistic availability (Szabo, 1980; Kersch & Fonseca, 2005). In dense networks. We also postulate that owing to non-deterministic tropical forests, less than 2% of solar radiation incidence on foraging ants in search of prey on plants without EFNs, the soil the forest canopy reaches the understory (Chazdon & Fetcher, pH and canopy cover does not affect nestedness in these neutral 1984; Oberbauer et al., 1988; Canham et al., 1990). Thus, as networks. Therefore, we expected that the factors that affect a result of the importance of light in the photosynthetic and the availability of prey on plants are possibly the main factors other metabolic processes of plants, light availability can also determining the neutral networks. We studied only nestedness affect the amount of the nectar secreted (Szabo, 1980; Kersch as a dependent variable, once small changes in the patterns & Fonseca, 2005; Radhika et al., 2010). EFN plants inserted in of ant–plant interaction is reflected in this network descriptor environments with high luminosity tend to increase the amount (Rico-Gray et al., 2012). of nectar secreted, mainly because nectar is a carbon-based defence derived from photosynthesis (Radhika et al., 2010; Yamawo & Hada, 2010). So, as the amount and quality of EFN Materials and methods secreted by plants in a community can vary according to local Study area ecosystem features, as soil pH or canopy cover, the trophic interactions between EFN and their visitors can also change We conducted this study in a non-disturbed dense terra-firme ◦ (Frazee & Marquis, 1994; Kersch & Fonseca, 2005; Munoz˜ rainforest within the southern Brazilian Amazon (9 48Sand ◦ et al., 2005; Burkle & Irwin, 2010; Cardoza et al., 2012). 58 15W, elev. 254 m), municipality of Cotriguac¸u, north of Within a natural environment different species can interact Mato Grosso State, Brazil. Terra-firme forests are tropical rain with each other and generate complex ecological networks of forests inserted in environments that do not suffer the influence interactions (Lewinsohn et al., 2006). Focused on the structure of periodic and annual cycles of flooding of the great rivers of these ecological networks, some studies have recently shown of Amazon. The reserve area covers 7000 ha of continuous that mutualistic networks between ants and EFN-plants are forest, surrounded by a much larger area also of continuous highly nested (Guimaraes˜ et al., 2006; Chamberlain et al., forest. The climate is wet tropical (Am) with annual means of ◦ 2010; Díaz-Castelazo et al., 2010; Sugiura, 2010; Dattilo,´ 24 C temperature, 85% humidity, and 2300 mm precipitation 2012; Rico-Gray et al., 2012). This shows that species with (Koppen¨ classification). The region has two distinct seasons, a few links interact with a subset of interactive species with November–April rainy season and a May–October dry season. several interactions (Bascompte et al., 2003; Thompson, 2005). Canopy trees range 30–40 m high, with some emergent trees Biologically, nestedness describes the organisation of niche reaching 50 m. The understory is relatively open, with a high breadth of a biological community, in which more nested frequency of Orbignya phalerata Mart. (Arecaceae) (Camargo networks tend to have the highest niche overlap (Bastolla et al., et al., 2010; Dattilo´ et al., 2012). 2009; Bluthgen,¨ 2010). In spite of the importance of nectar for ant colony fitness (Byk & Del-Claro, 2011), little is known about how nectar composition can affect ant–plant networks. Experimental design and abiotic factors Only one previous study has been conducted showing that the number of nectaries is not an important factor contributing to Ant–plant interactions were sampled, and soil and vegeta- the nested pattern in ant–plant networks (Chamberlain et al., tion features in a module managed by the Brazilian Research 2010). However, it is not known how the factors that affect the Program in Biodiversity (PPBio) (http://ppbio.inpa.gov.br). amount and quality of nectar can structure ant–plant networks. The module consists of two parallel east–west 5-km trails, Furthermore, ants can also randomly forage for prey 1 km apart. Along each trail, we marked one sampling plot of on plants without EFNs or without honeydew secreted by 250 m × 25 m (6250 m2) every km, totaling 12 sampling points.

© 2013 The Royal Entomological Society, Ecological Entomology, 38, 374–380 376 Wesley Dattilo ´ et al.

The central trail in each plot was located to minimise varia- we used general linear models (GLM). We used nestedness as tions of altitude and soil features, increasing the precision of dependent variable, network type (mutualistic or neutral), soil estimates for predictor variables (soil pH and canopy cover) pH and canopy cover and interactions between network type (Magnusson et al., 2005). and soil pH and canopy cover as a fixed factors, and network To measure soil pH, we collected six soil samples (50-m size as a cofactor. In the cases that the interaction was signifi- equidistant points) in the centre of each 12 sampling points cant, we used a posteriori simple regressions to best describe to a depth of 5 cm. We obtained the pH value for each the differences on tendencies between neutral and mutualis- sample from a solution of dry ground soil with distilled water tic networks. Note that we only used GLM because both soil with a pH meter, according to the protocol of the Brazilian pH and canopy cover were not correlated factors (Pearson’s Agricultural Research Corporation (Embrapa-Solos, 1999). In correlation: r = 0.11, P = 0.28). We performed all statistical addition, along the centre of each plot, we also measured analyses using R-software version 2.13.1 (R Development Core the canopy cover at six equidistant points (50 m) using a Team, 2012). concave densiometer. At each point we made four measures according to geographical references (north, south, east, and west). The value representing the canopy cover of each plot is Results the average of the measurements of the geographical references and subsequently the average of six points. We recorded 238 plant species (72 with EFNs) and 149 ant species. The mean number of plant species per sampling plot on mutualistic networks was lower (mean ± SD: 21.4 ± 3.77) Data collection of ant–plant interactions than plant species on neutral networks (27.2 ± 3.97, t =−3.093, d.f. = 11, P = 0.011). However, the mean We recorded ant–plant interactions in December 2010 and number of ant species on mutualistic networks per sampling January 2011, always between 09.00 and 15.00 hours. At each point (23.2 ± 5.85) did not differ from the number on neutral of the 12 sampling points, we collected ants foraging on all networks (23.3 ± 4.11) (t =−0.0647, d.f. = 11, P = 0.949). EFN–plants that were accessible to the collector (from 0.5 to In addition, soil pH and canopy cover did not influence ant 3 m high). We recorded all occurrences of ants collecting liq- or plant richness in both mutualistic and neutral ant–plant uids on EFN of each plant. For each EFN-bearing plant where networks (All P-value > 0.05) (Table 1). ants were collected, we selected a plant without EFNs with Both mutualistic and neutral ant–plant networks show a similar structure (height, width, and number of branches) nested patterns of interactions (all P < 0.05). However, average nearby (neutral networks). No plants with Hemiptera or other values of nestedness are different between mutualistic and neu- visible liquid sources were included when sampling plants tral networks (GLM: F = 3.625, P = 0.04). NODF values were without EFNs. We did not use the agonistic interactions higher in mutualistic networks (mean ± SD: 34.71 ± 10.21) involving leaf-cutting ants and plants. In plants without EFNs, than in neutral networks (15.76 ± 3.34). First, we observed that resources cannot be predicted, and ants forage randomly, using both explanatory variables varied significantly (t-test; soil pH: the plant only as substrate (Bluthgen¨ et al., 2000; Yamamoto, P = 0.03; and canopy cover: P = 0.04), indicating that these & Del-Claro, 2008). Therefore, our neutral networks are possi- variables had a sufficient magnitude to affect the nestedness bly generated by differences in species abundance. All selected of networks. Posteriorly, we observed that small changes on plants were at least 10 m apart to minimise the possibility of soil pH values can exert related changes on the NODF values collecting the same ant colony foraging on different plants. of ant–plant networks (GLM: F = 4.371, P = 0.04). However, there is an interaction between network type (mutualistic and neutral) and soil pH (GLM: F = 8.879, P = 0.01). In mutu- Nestedness and statistical analyses alistic networks, an increase in soil pH is positively related to NODF values (r2 = 0.37; F = 5.912; P = 0.03). But, this To estimate nestedness of the networks, we used the tendency was not observed on neutral networks (r2 = 0.131; NODF index (Nestedness Overlap and Decreasing Fill) using F = 1.506; P = 0.248) (Fig. 1a). Moreover, nestedness was not ANINHADO software (Guimaraes˜ & Guimaraes,˜ 2006). To affected by canopy cover in mutualistic or neutral networks assess if the nestedness value observed was higher than (GLM: mutualistic: F = 0.119; P = 0.738, neutral: F = 0.196; expected by random interaction patterns, we tested the P = 0.668) (Table 1) (Fig. 1b). The network size did not affect nestedness of each network with 1000 networks generated the nestedness of networks (GLM: F = 0.041, P = 0.84). Addi- by Null Model II (CE). In this null model, the probability tionally, the 24 networks studied (both mutualistic and neutral) of an interaction occurring is proportional to the level of tended to not have groups or modules of species of one trophic degree (mean number of links or interactions) of plant and level that interact more frequently with a group of species of ant species (Bascompte et al., 2003). To test the difference of another trophic level (Fig. 2a,b). nestedness between ecological networks of ants and plants with and without EFNs, we used a paired t-test (paired per plot). In order to evaluate the statistical variation of our Discussion explanatory variables (pH and canopy cover) around their own mean, we used a Student’s t-test for one sample. To test how Our results suggest that, in the Brazilian Amazonian rainfor- different factors affect the nested pattern in ant–plant networks, est, different abiotic factors can affect the nested pattern of

© 2013 The Royal Entomological Society, Ecological Entomology, 38, 374–380 Effect of abiotic factors on ant–plant networks 377

Table 1. Values of soil pH, canopy cover (%), ant and plant richness, and nestedness (NODF metric) from mutualistic and neutral ant–plant networks in 12 sampling points in the Brazilian Amazon (see text for more information).

Mutualistic networks Neutral networks Plot pH Canopy (%) Ants Plants NODF Ants Plants NODF

1 4.40 39.90 25 28 18.33 21 31 15.70 2 4.10 41.60 16 22 38.92 13 29 22.15 3 4.13 41.18 31 19 28.26 26 32 13.43 4 4.33 43.47 27 16 39.00 27 30 15.52 5 4.06 41.18 28 21 27.23 23 28 12.06 6 4.36 40.35 30 23 33.92 27 22 15.91 7 4.23 41.18 25 22 29.89 25 26 14.83 8 4.36 40.35 18 28 37.41 21 23 20.91 9 4.40 40.14 20 17 44.50 23 33 13.20 10 4.20 41.18 14 21 21.64 20 22 17.99 11 4.86 40.35 17 18 53.98 26 27 10.94 12 4.33 42.01 27 22 43.43 27 23 16.44

60.00 (a) Mutualistic network Neutral network (a) 50.00

40.00

30.00

NODF value 20.00

10.00 (b) 0.00 44.24.44.64.85 Soil pH

60.00 (b)

50.00

40.00

30.00 Fig. 2. Graphical representation of one (a) mutualistic network and

NODF value 20.00 one (b) neutral network out of the 24 ant–plant networks studied in a tropical rainforest in the southern Brazilian Amazon (December 2010 10.00 to January 2011). Each circle represents one ant or plant species, which are represented by grey and black colours, respectively. Lines represent 0.00 ant–plant interactions. The ant–plant networks are represented as an 39 40 41 42 43 44 energy two-mode graph (Kamada–Kawai free method) obtained using Canopy cover (%) the program Pajek.

Fig. 1. Regressions between (a) soil pH and nestedness (NODF) of = = 2 = mutualistic (r 0.37; P 0.03) and neutral ant–plant networks (r EFNs compared with ants that used the plants only as sub- 0.01; P = 0.248), and (b) canopy cover and nestedness of mutualistic strate. Soil pH was an important factor for the nested pattern (r 2 = 0.01; P = 0.738) and neutral (r 2 = 0.01; P = 0.668) ant–plant networks. only for the mutualistic networks. In addition, we did not find an association between canopy cover and nestedness in both mutualistic and neutral ant–plant networks. According interactions independently of ant–plant network size. Local to Rico-Gray et al. (2012), abiotic factors probably influence features of soil differentially influence the patterns of interac- the availability of resources on plants, which in turn affect the tions in mutualistic and neutral ant–plant networks possibly number of interactions and thus the nested pattern of ant–plant because soil pH differently influences the ants that feed on networks. Vegetation is one of the main factors affecting the

© 2013 The Royal Entomological Society, Ecological Entomology, 38, 374–380 378 Wesley Dattilo ´ et al. structure of ant communities in an environment (Andersen, environments, e.g. foraging periods, competitive interactions, 1990; Retana & Cerda,´ 2000; Wang & Strazanac, 2001; Lassau and variation in daily in resource availability (Rico-Gray & Hochuli, 2004), therefore, factors that influence vegetation & Oliveira, 2007). Therefore, a fundamental next step is to features can be an important mechanism determining the struc- understand all the main mechanisms and factors that shape ture of ant–plant networks. the structure of ecological networks involving ants and plants. Although soil pH did not influence ant and plant richness, this soil feature significantly altered nestedness in mutualistic networks. The present results suggest that soils with a pH Acknowledgements closer to neutral generate highly nested mutualistic networks. This possibly occurs because plants growing on soils with We are grateful to Jessica´ Falcao˜ for his help during the a higher neutral pH and nutrient availability tend to have fieldwork and two anonymous reviewers for providing useful nectar with higher concentrations of sugars and amino acids comments on earlier draft of manuscript. We thank the (Gardener & Gillman, 2001; Burkle & Irwin, 2008; Yao et al., staff of the Central Herbarium of Universidade Federal de 2009). This change in the nectar quality could influence the Mato Grosso for identification of plant specimens. We also thank Office National des Foretsˆ Brazil and the Brazilian patterns of ant–plant interactions and reflect on the nestedness ◦ of networks. Furthermore, although the ants strongly compete Research Program in Biodiversity (PPBio Project) (CNPq n for better quality nectar (Dreisig, 2000; Bluthgen¨ & Fiedler, 558225/2009–8) for logistical and financial support. W.D. is grateful for financial support by the National Counsel of 2004a), in environments with higher neutral pH, it is possible ◦ that all plants tend to have better resources, and therefore Technological and Scientific Development, Brazil (CNPq n 237339/2012-9) and National Council on Science and Tech- we should expect lower competition for resources. Thus, ◦ as nestedness reflects the degree of generalisation of a net- nology, Mexico (CONACYT n 489746). This is publication work (Bascompte et al., 2003), minimising competition and 26 in the Nucleo´ de Estudos da Biodiversidade da Amazoniaˆ increasing the number of coexisting species (Bastolla et al., Mato-Grossense technical series. 2009), mutualistic networks inserted in highly neutral pH environments would then be highly nested. On the contrary, environments where soil pH is more acidic, it is possible References that a few plant species could produce high nectar quality, Andersen, A.N. (1990) The use of ant communities to evaluate generating more competition for resources and less nestedness change in Australian terrestrial ecosystems: a review and a recipe. in these sampling points. Proceedings of the Ecological Society of Australia, 16, 347–357. Although some plants secrete larger amounts of nectar Baker, H.G., Opler, P.A. & Baker, I. (1978) A comparison of the in environments with higher light availability (Szabo, 1980; amino acid complements of floral and extrafloral nectars. Botanical Kersch & Fonseca, 2005), we found no effect of canopy cover Gazette, 139, 322–332. on the nestedness in mutualistic networks. Two main factors Bascompte, J., Jordano, P., Melian,´ C.J. & Olesen, J.M. (2003) The nested assembly of plant–animal mutualistic networks. Proceedings not mutually exclusive could explain this pattern. First, the of the National Academy of Sciences of the United States of America, amount of secreted nectar can not be an important factor 100, 9383–9387. affecting the structure of mutualistic networks. Some previous Bastolla, U., Fortuna, M.A., Pascual-García, A., Ferrera, A., Luque, studies have shown that ants may prefer more concentrated B. & Bascompte, J. (2009) The architecture of mutualistic networks nectar instead of large amounts of nectar (Bluthgen¨ & Fiedler, minimizes competition and increases biodiversity. Nature, 458, 2004a,2004b; Lach, 2005). It is possible that the presence of 1018–1021. ants on plants with EFN is not determined by the amount of Baude, M., Leloup, J., Suchail, S., Allard, B., Benest, D., Meriguet,´ J., nectar available, but by nectar quality. Additionally, as there et al. (2011) Litter input and plant interactions affect nectar sugar was little variation in canopy cover in our study area, this content. Journal of Ecology, 99, 828–837. Bluthgen,¨ N. (2010) Why network analysis is often disconnected from could not have influenced the amount of nectar secreted in community ecology: a critique and an ecologist’s guide. Basic and environments with ’higher light availability’. Applied Ecology, 11, 185–195. In summary, our results show that soil pH and canopy cover Bluthgen,¨ N. & Fiedler, K. (2004a) Competition for composi- can differently structure mutualistic and neutral ant–plant tion: lessons from nectar–feeding ant communities. Ecology, 85, networks. It is possible that nectar quality is the main factor 1479–1485. that influences the nested pattern in mutualistic ant–plant Bluthgen,¨ N. & Fiedler, K. (2004b) Preferences for sugars and networks. We also show that soil pH and canopy cover do not amino acids and their conditionality in a diverse nectar–feeding influence nestedness in neutral networks. In fact, ants forage ant community. Journal of Animal Ecology, 73, 155–166. randomly for prey on plants without EFN, using the plant only Bluthgen,¨ N., Verhaagh, M., Goitía, W., Jaffe,´ K., Morawetz, W. & as a substrate (Andersen, 1990; Bluthgen¨ et al., 2000). We Barthlott, W. (2000) How plants shape the ant community in the Amazonian rainforest canopy: the key role of extrafloral nectaries suggest that prey availability on plants may be the main factor and homopteran honeydew. Oecologia, 125, 229–240. structuring neutral ant–plant networks. Thus, it is possible Burkle, L.A. & Irwin, R.E. (2008) The effects of nutrient addition on that variations in soil pH and canopy cover could have little floral characters and pollination in two subalpine plants, Ipomopsis effect on prey availability and not reflect on nestedness of aggregate and Linum lewisii. Plant Ecology, 203, 83–98. neutral networks. Finally, different biotic and abiotic factors Burkle, L.A. & Irwin, R.E. (2010) Beyond biomass: measuring the may influence the patterns of ant–plant interactions in tropical effects of community–level nitrogen enrichment on floral traits,

© 2013 The Royal Entomological Society, Ecological Entomology, 38, 374–380 Effect of abiotic factors on ant–plant networks 379

pollinator visitation and plant reproduction. Journal of Ecology, 98, Fischer, J. & Lindenmayer, D. (2002) Treating the nestedness 705–717. temperature calculator as a black box can lead to false conclusions. Byk, J. & Del-Claro, K. (2011) Ant–plant interactions in the Oikos, 99, 193–199. Neotropical savanna: direct beneficial effects of extrafloral nectar Frazee, J.E. & Marquis, R.J. (1994) Environmental contribution on ant colony fitness. Population Ecology, 53, 327–332. to Xoral trait variation in Chamaecrista fasciculata (Fabaceae: Camargo, F.F., Costa, R.B., Resende, M.D.V., Roa, R.A.R., Rodrigues, Cesalpinoideae). American Journal of Botany, 81, 206–215. N.B., Santos, L.V., et al. (2010) Variabilidade genetica´ para Gardener, M.C. & Gillman, M.P. (2001) The effects of soil fertilizer on caracteres morfometricos´ de matrizes de castanha-do-brasil da amino acids in the floral nectar of corncockle, Agrostemma githago Amazoniaˆ Mato-grossense. Acta Amazonica, 40, 705–710. (Caryophyllaceae). Oikos, 92, 101–106. Campbell, D.R. & Halama, K.J. (1993) Resource and pollen limitations Guimaraes,˜ P.R. Jr. & Guimaraes,˜ P.R. (2006) Improving the analyses to lifetime seed production in a natural plant population. Ecology, of nestedness for large sets of matrices. Environmental Modelling 74, 1043–1051. and Software, 21, 1512–1513. Canham, C., Denslow, J.S., Platt, W.J., Runkle, J.R., Spies, T.A. & Guimaraes,˜ P.R., Rico-Gray, V., Reis, S.F. & Thompson, J.N. White, P.S. (1990) Light regimes beneath closed canopies and tree- (2006) Asymmetries in specialization in ant–plant mutualistic fall gaps in temperate and tropical forests. Canadian Journal of networks. Proceedings of the Royal Society: Biological Sciences, Forest Research, 20, 620–631. 273, 2041–2047. Cardoza, Y.J., Harris, G.K. & Grozinger, C.M. (2012) Effects of Heil, M., Fiala, B., Baumann, B. & Linsenmair, K.E. (2000) Temporal, soil quality enhancement on pollinator–plant interactions. Psyche, spatial and biotic variations in extrafloral nectar secretion by ID 581458, 8 pp. URL http://www.hindawi.com/journals/psyche/ Macaranga tanarius. Functional Ecology, 14, 749–757. 2012/581458 [accessed on 14 January 2013]. Heil, M., Hilpert, A., Fiala, B. & Linsenmair, K.E. (2001) Nutrient Chamberlain, S.A., Kilpatrick, J.R. & Holland, J.N. (2010) Do availability and indirect (biotic) defence in a Malaysian ant–plant. extrafloral nectar resources, species abundances, and body sizes Oecologia, 126, 404–408. contribute to the structure of ant–plant mutualistic networks? Herrera, C.M., Perez,´ R. & Alonso, C. (2006) Extreme intraplant Oecologia, 164, 741–750. variation in nectar sugar composition in an insect–pollinated Chazdon, R.L. & Fetcher, N. (1984) Photosynthetic light environments perennial herb. American Journal of Botany, 93, 575–581. in a lowland tropical rain forest in Costa Rica. Journal of Ecology, Horvitz, C.C. & Schemske, D.W. (1984) Effects of ants and 72, 553–564. ant–tended herbivore on seed production of a neotropical herb. Dattilo,´ W. (2012) Different tolerances of symbiotic and nonsymbiotic Ecology, 65, 1369–1378. ant-plant networks to species extinctions. Network Biology, 2, Kersch, M.F. & Fonseca, C.R. (2005) Abiotic factors and the 127–138. conditional outcome of an ant–plant mutualism. Ecology, 86, Dattilo,´ W., Martins, R.L., Uhde, V., Noronha, J.C., Florencio,ˆ F.P. & 2117–2126. Izzo, T.J. (2012) Floral resource partitioning by ants and bees in a Koptur, S., Rico-Gray, V. & Palacios-Rios, M. (1998) Ant protection jambolan Syzygium jambolanum (Myrtaceae) agroforestry system in of the nectaried fern Polypodium plebeium in central Mexico. Brazilian Meridional Amazon. Agroforestry Systems, 85, 105–111. American Journal of Botany, 85, 736–739. Del-Claro, K., Berto, V. & Reu,´ W. (1996) Effect of herbivore Lach, L. (2005) Interference and exploitation competition of three deterrence by ants increase fruit set in an extrafloral nectary plant nectar–thieving invasive ant species. Insectes Sociaux, 52, Qualea multiflora (Vochysiaceae). Journal of Tropical Ecology, 12, 257–262. 887–892. Lanza, J. & Krauss, B.R. (1984) Detection of amino acids in Dreisig, H. (2000) Defense by exploitation in the Florida carpenter ant, artificial nectars by two tropical ants Leptothorax and Monomorium. Camponotus floridanus, at an extrafloral nectar resource. Behavioral Oecologia, 63, 423–425. Ecology and Sociobiology, 47, 274–279. Lassau, S.A. & Hochuli, D.F. (2004) Effects of habitat complexity on Díaz-Castelazo, C., Rico-Gray, V., Oliveira, P.S. & Cuautle, Y.M. ant assemblages: can we generalise across scale? Ecography, 27, (2004) Extrafloral nectary-mediated ant–plant interactions in the 157–164. coastal vegetation of Veracruz, Mexico:´ richness, occurrence, Lewinsohn, T.M., Prado, P.I., Jordano, P., Bascompte, J. & Olesen, seasonality, and ant foraging patterns. Ecoscience, 11, 472–482. J.M. (2006) Structure in plant–animal interaction assemblages. Díaz–Castelazo, C., Guimaraes,˜ P.R., Jordano, P., Thompson, J.N., Oikos, 113, 174–184. Marquis, R.J. & Rico–Gray, V. (2010) Changes of a mutualistic Magnusson, W.E., Lima, A.P., Luizao,˜ R., Luizao,˜ F., Costa, F.R.C., network over time: reanalysis over a 10–year period. Ecology, 91, Castilho, C.V., et al. (2005) RAPELD: a modification of the Gentry 793–801. method for biodiversity surveys in long-term ecological research Ekenler, M. & Tabatabai, M.A. (2003) Effects of liming and tillage sites. Biota Neotropica, 5, 19–24. systems on microbial biomass and glycosidases in soils. Biology and Mevi-Schutz,¨ J., Goverde, M. & Erhardt, A. (2003) Effects of Fertility of Soils, 39, 51–61. fertilization and elevated CO2 on larval food and butterfly nectar Elias, T.S. (1983) Extrafloral nectaries: their structure and distribution. amino acid preference in Coenonympa pamphilus L. Behavioral The Biology of Nectaries (ed. by B. Bentley and T. S. Elias), pp. Ecology and Sociobiology, 54, 36–43. 174–203. Columbia University Press, New York, New York. Monaco, T.A., Johnson, D.A., Norton, J.M., Jones, T.A., Connors, K.J., EmbrapaSolos (1999) Manual of chemical analysis of soils, plants and Norton, J.B., et al. (2003) Contrasting responses of intermountain fertilizers. Editora Embrapa, Brasília, Distrito Federal. west grasses to soil nitrogen. Journal of Range Management, 56, Farkas, A., Molnar,´ R., Morschhauser, T. & Hahn, I. (2012) Variation 282–290. in nectar volume and sugar concentration of Allium ursinum L. Munoz,˜ A.A., Celedon-Neghme, C., Cavieres, L.A. & Arroyo, ssp. ucrainicum in three habitats. The Scientific World Journal, M.T.K. (2005) Bottom-up effects of nutrient availability on flower ID 138579, 7 pp. URL http://www.hindawi.com/journals/tswj/2012/ production, pollinator visitation, and seed output in a high-andean 138579 [accessed on 14 January 2013]. shrub. Oecologia, 143, 126–135.

© 2013 The Royal Entomological Society, Ecological Entomology, 38, 374–380 380 Wesley Dattilo ´ et al.

Oberbauer, S.F., Clark, D.B., Clark, D.A. & Quesada, M. (1988) Sugiura, S. (2010) Species interactions–area relationships: biological Crown light environments of saplings of two species of rain forest invasions and network structure in relation to island area. Proceed- emergent trees. Oecologia, 75, 207–212. ings of the Royal Society: Biological Sciences, 277, 1807–1815. Oliveira, P.S., Rico-Gray, V., Díaz-Castelazo, C. & Castilho-Gevara, Szabo, T.I. (1980) Effect of weather factors on honeybee a flight C. (1999) Interaction between ants, extrafloral nectaries and insect activity and colony weight gain. Journal of Apicultural Search, 19, herbivores in Neotropical coastal and dunes: herbivore deterrence 164–171. by visiting ants increases fruit set in Opuntia stricta (Cactacea). Thompson, J.N. (2005) The Geographic Mosaic of Coevolution.The Functional Ecology, 13, 623–631. University of Chicago Press, Chicago, Illinois. Radhika, V., Kost, C., Mithofer,¨ A. & Boland, W. (2010) Regulation Wang, C. & Strazanac, J.S. (2001) Association between ants of extrafloral nectar secretion by jasmonates in lima bean is light (Hymenoptera: Formicidae) and habitat characteristics in oak- dependent. Proceedings of the National Academy of Sciences of the dominated mixed forests. Environmental Entomology, 30, 842–848. United States of America, 107, 17228–17233. Yamamoto, M. & Del-Claro, K. (2008) Natural history and foraging R Development Core Team (2012) R: A Language and Environment behavior of the carpenter ant Camponotus sericeiventris Guerin,´ for Statistical Computing. R Foundation for Statistical Computing, 1838 (Formicinae, Campotonini) in the Brazilian tropical savanna. Vienna, Austria[WWW document]. URL http://www.R–project.org Acta Ethologica, 11, 55–65. [accessed on 21 April 2012]. Yamawo, A. & Hada, Y. (2010) Effects of light on direct and indirect Retana, J. & Cerda,´ X. (2000) Patterns of diversity and composition of defences against herbivores of young plants of Mallotus japonicus Mediterranean ground ant communities tracking spatial and temporal demonstrate a trade-off between two indirect defence traits. Annals variability in the thermal environment. Oecologia, 123, 436–444. of Botany, 106, 143–148. Rico-Gray, V. & Oliveira, P.S. (2007) The Ecology and Evolution Yao, H.Y., Bowman, D., Rufty, T. & Shi, W. (2009) Interactions of Ant–Plant Interactions. Chicago, Illinois, The University of between N fertilization, grass clipping addition and pH in turf Chicago Press. ecosystems: implications for soil enzyme activities and organic mat- Rico-Gray, V., Díaz-Castelazo, C., Ramírez-Hernandez,´ A., ter decomposition. Soil Biology and Biochemistry, 41, 1425–1432. Guimaraes,˜ P.R. & Holland, J.N. (2012) Abiotic factors shape temporal variation in the structure of an ant–plant network. Accepted 5 March 2013 Arthropod–Plant Interactions, 6, 289–295. First published online 15 May 2013 Sperens, U. (1997) Long-term variation in, and effects of fertilizer addition on, flower, fruit and seed production in the tree Sorbus aucuparia (Rosaceae). Ecography, 20, 521–534.

© 2013 The Royal Entomological Society, Ecological Entomology, 38, 374–380 48

Chapter IV

Dáttilo, W.; Sánchez-Gálvan, I.; Lange, D.; Del-Claro, K.; Rico-Gray, V. (2014).

Importance of interaction frequency in analysis of ant-plant networks in tropical

environments. Journal of Tropical Ecology 30: 165-168. Journal of Tropical Ecology (2014) 30:165–168. © Cambridge University Press 2013 doi:10.1017/S0266467413000813

SHORT COMMUNICATION Importance of interaction frequency in analysis of ant-plant networks in tropical environments

Wesley Dattilo´ ∗,1, Ingrid Sanchez-Galv´ an´ †, Denise Lange‡, Kleber Del-Claro‡ and V´ıctor Rico-Gray∗

∗ Instituto de Neuroetolog´ıa, Universidad Veracruzana, Xalapa, Veracruz, 91190, Mexico † CIBIO, Universidad de Alicante, San Vicente del Raspeig (Alicante), 03080, Spain ‡ Laboratorio´ de Ecologia Comportamental e Interac¸oes,˜ Instituto de Biologia, Universidade Federal de Uberlandia,ˆ Uberlandia,ˆ Minas Gerais, 38400–058, Brazil (Received 15 August 2013; revised 13 November 2013; accepted 14 November 2013; first published online 13 December 2013)

Abstract: Several studies have shown that qualitative (binary) ant-plant networks are highly nested in tropical environments, in which specialist species (with fewer interactions) are connected with generalists (with the most interactions) in cohesive subgroups. Interactions occur in both qualitative and quantitative networks, however, how their frequency may structure the nestedness in ecological networks involving these organisms is, we believe, unknown. Based on this perspective, we used nestedness analysis to address the effect of interaction frequency on ant- plant networks (n = 14 networks). Unlike binary networks, quantitative networks are often significantly non-nested. In addition, species with a higher interaction frequency have a higher number of links, indicating that these species are possibly more abundant and/or competitive. Moreover, different biological parameters can change the nature of ant-plant interactions, as a plant can be a good resource for one ant and a ‘bad’ resource for another. Thus, this suggests a new perspective for the study of interaction networks in the tropics, since species with lower interaction frequency are not necessarily subsets of species with higher frequency, and consequently generate the non-nested pattern in quantitative networks.

Key Words: Ant-plant interactions, ecological networks, nestedness, qualitative networks, quantitative networks

Recent studies have focused on ecological networks Ant-plant associations are a useful system to study and found a highly nested pattern for different systems questions of the importance of interaction frequency in and habitats (Bascompte et al. 2003). In a nested interaction networks, because ant-plant interactions are network, specialist species (with fewer interactions) are relatively easy to quantify and are extremely common connected with generalists (with the most interactions) in most terrestrial environments (Rico-Gray & Oliveira in cohesive subgroups (Bascompte et al. 2003). Metrics 2007). Perhaps the most well-documented association based on binary (or qualitative) matrices have been between ants and plants in the tropics are those with developed to describe such nested patterns (Ulrich et al. extrafloral nectaries (EFN-bearing plants) (Rico-Gray & 2009). In these matrices, all interactions are ecologically Oliveira 2007). In these associations, common in more equivalent, and individual differences of interactions are than 100 plant families, the nectar is a liquid rich not taken into account (Almeida-Neto & Ulrich 2011). in carbohydrates (Rico-Gray & Oliveira 2007) which However, in the natural environment, free-living species attracts ants. In exchange for this food, ants defend interact quantitatively generating a complex system of plants against natural enemies (Rico-Gray & Oliveira interactions, which allows the study of species preferences 2007). Recent studies focusing on the structure of (Bluthgen¨ 2010), and some authors have criticized binary networks involving ants and EFN-bearing plants nestedness analysis in binary matrices (Almeida-Neto & have shown that these networks are highly nested Ulrich 2011). (Chamberlain et al. 2010, Dattilo´ et al. 2013a). However, it is not known how the frequency of these associations may structure the ecological networks involving these 1 Corresponding author. Email: [email protected] organisms. 166 WESLEY DATTILO´ ET AL. ,

Here we postulate that due to differences in the b I availability and quality of nectar, the strength of the interaction between ants and plants may change due metric. (2000) to the physiological requirements of ants, including (2004)

their foraging ability. We hypothesized that if there WNODF et al. et al. are nectar-composition preferences and competition for best resources among ant species (Bluthgen¨ & Fiedler ¨ ¨ uthgen 2004), species with high interaction frequency do uthgen not necessarily have more links (binary links), or the interaction distribution among partners may not be

equitable (different interaction patterns). So, the species’ Mean number of links per species; , position in the nestedness ranking may be different pa L when compared with binary matrices, in which links ); /PA

of specialists are not necessarily subsets of links of b I generalists. Specifically, we addressed the following , nestedness value observed using question: Does the nested pattern in ant-plant networks WNODF differ between binary or quantitative data? To try to N answer this question, we performed nestedness analysis in ant-plant networks from different ecosystems around metric; the world and compared two different nestedness metrics Ecosystem (country) Database (qualitative and quantitative). NODF , network connectance ( We used quantitative ant-plant networks from the C literaturealongwithourextensivedatabasefromdifferent tropical ecosystems. The 14 networks used belong to six distinct ecosystems (Table 1). Our ant-plant networks were collected in Amazon tropical rainforest (plot: 250 × 25 m) (9º48S, 58º15W, Brazil), Neotropical savanna (two transects: 1 km × 5m)(18º58S, 48º17W, Brazil) and dry tropical lowland coastal vegetation (six transects:   WNODF N 1km× 5m)(19º6 N, 96º22 W, Mexico) (unpubl. data). , nestedness value observed using We recorded all occurrences of ants collecting liquids on NODF

EFN of each plant. N Using ANINHADO we calculated the NODF metric to estimate nestedness value for binary networks (Almeida-

Neto et al. 2008). To assess if the observed nestedness , number of plant and ant species, respectively; A value was higher than expected by random interaction NODF N patterns, we tested nestedness using Null Model II and P (Bascompte et al. 2003). For quantitative networks we used a recent metric for nestedness analysis based on quantitative matrices called WNODF (Weighted q Nestedness Metric Based on Overlap and Decreasing I Fill), and tested the WNODF significance using the Null Model RC (Almeida-Neto & Ulrich 2011). Both b I nestedness metrics vary from zero (no nestedness) to 100 , number of interactions (quantitative data); q (perfect nestedness). Note that while the NODF metric I independently computes the sequence of decreasing pa marginal totals and the overlap of pairs, the WNODF metric considers the same NODF principles, but weighted by abundance ranking (Almeida-Neto & Ulrich 2011). To compare the frequency of occurrence of significantly nested networks in binary and quantitative matrices, we used maximum likelihood χ 2 tests (G tests). A Wilcoxon signed rank test was used to evaluate the difference of Network properties of 14 ant-plant networks studied. nestedness value between the two matrices since our number of interactions (binary data); 212218 2823 1619 17 0.099 30 0.179 31 0.17 0.859 0.12 0.60 159 0.12 0.64 12 578 0.63 18 0.40 6332 0.69 19 0.38 54 73 14 0.40 0.44 84 45 74 0.35 0.50 72 65 27.2 0.34 0.40 (nested) 54 0.14 94 38.9 0.46 (nested) 42 86 53.9 0.56 (nested) 66 0.37 101 33.9 (nested) 60 8.84 (non-nested) 102 28.2 (nested) 39 208 8.88 (non-nested) 197 56.6 (nested) 15.2 (non-nested) Amazon 119 59.8 tropical (nested) rain forest 1254 (Brazil) 7.42 (non-nested) Amazon 61.1 tropical (nested) rain forest 71 (Brazil) 5.38 (non-nested) Amazon 55.0 tropical (nested) rain forest 39.5 (Brazil) (nested) 25.2 (non-nested) Amazon tropical rain forest 53.7 (Brazil) (nested) 26.7 (non-nested) Amazon tropical rain forest (Brazil) 31.7 (non-nested) Neotropical savanna (Brazil) 30.8 (non-nested) Neotropical savanna (Brazil) 21.76 (nested) Neotropical savanna 21.1 (Brazil) (non-nested) Neotropical savanna (Brazil) Our database Neotropical savanna Our Dry (Brazil) database tropical lowland forest (Mexico) Our database Our database Our database Our database Our database Our database Our database Our database Our database 56 17 79 0.47 0.28 0.55 0.62 40 137 596 298 61.1 (non-nested) 35.8 (nested) 42.1 (nested) 16.7 (nested) Submontane tropical rain forest (Papua New Guinea) Amazon tropical rain forest (Venezuela) Whalen & Mackay (1988) Bl resultsdidnothaveanormaldistribution.Finally,inorder Table 1. PAC L 51 41 0.13 3.09 285 644 43.5 (nested) 23.3 (non-nested) Australia’s tropical rain forest (Australia) Bl Nestedness in ant-plant networks 167 to evaluate whether the number of ant and plant species is consequence, competition has been identified as an proportional to their interaction frequency, we compared important factor in the structure of ant communities the number of links and the frequency of interactions (Bluthgen¨ & Fiedler 2004). In ant-plant interactions, divided by the number of links, using a Spearman rank some ant species can monopolize a particular EFN- correlation. This approach was used because the number plant over a long period of time and exhibit a of links and frequency of species interactions were not massive recruitment of workers (Bluthgen¨ & Fiedler independent and could be affected by sampling effects, in 2004). Moreover, due to differences in physiological which more links imply higher frequency of interactions and ontogenetic conditions of each ant species, one when more plants are sampled. plant species can be a good resource (better quality and The ant-plant networks studied (Table 1) exhibited amount of nectar) for an ant species and a ‘bad’ resource different nested patterns when analysed using binary or for another. Species with lower interaction frequency quantitative matrices (g = 14.6, df = 1, P < 0.001). are not necessarily subsets of species with higher Binary networks were often significantly nested (93% of frequency, which generates the non-nested pattern in the networks) whereas most quantitative networks were quantitative networks. Such ‘deviating’ species are called non-nested (only 21% of the networks were nested). Only idiosyncratic species because they show different patterns two networks (14%) were significantly nested in both of interactions from a perfectly nested pattern (Almeida- binary and qualitative matrices (Amazon,Venezuela, Neto & Ulrich 2011). Therefore, we show that ant- and dry lowland, Mexico). Moreover, binary matrices plant networks are a much more complex system than were more nested (NODFobs: 46.3 ± 12.3, mean ± previously known in the literature based only on binary SD) than quantitative matrices (WNODFobs: 20.4 ± data. 10.7) (Wilcoxon signed rank test: z = 3.29, P < Several authors have studied the coevolutionary 0.001) (Table 1). Also, we observed that in all ant-plant dynamics in ecological networks (Guimaraes˜ et al. 2011). networks the number of links was positively correlated The focus of these studies was mainly in the core of with frequency of species interactions (Spearman rank highly generalist species, where the interaction strength correlation: Ants: rs = 0.52 ± 0.11; Plants: rs = 0.49 ± among interacting partners is symmetrical, with the 0.09, all P < 0.05). potential to drive the coevolution of the whole network As hypothesized, the nested pattern found in ant-plant (Guimaraes˜ et al. 2011). We suggest that nestedness networks was different when analysed using binary or analyses using quantitative data could increase the quantitative data. Unlike binary networks, quantitative strength of coevolutionary processes involving ant- networks were often significantly non-nested. These plant networks, since the interdependence among findings suggest that the ecological interpretation of species based on interaction frequency is taken into such ant-plant networks can be remarkably different account. However, we did not find the robust core of when we evaluate the type of network (qualitative or generalist species in quantitative networks, probably quantitative). For binary ant-plant networks, we know due to the random organization of this community. thatrelativespeciesabundanceisoneofthebestpredictors Furthermore, we observed that species with higher to explain the origin of the nested pattern (Chamberlain interaction frequency exhibited a higher number of et al. 2010). Such is the case because species tend to links and that some specific ant-plant interactions were find individuals of other abundant species more often more frequent. We hypothesized that this result is not than individuals of rare species (Krishna et al. 2008). If simply a sampling effect, mainly because these highly ant-plant interactions depend only on ant abundance, it interactive species are ant species that have behavioural implies no actual competition for the resource, mainly and ecophysiological adaptations to use liquid food and because the individual differences of ants and plants are possibly competitively superior ant species (massive are not taken into account, and therefore all species recruitment and/or aggressive behaviour) (Dattilo´ et al. are ecologically equivalent. However, different biological 2013b). Therefore, if such interactions are positive and parameters can change the patterns in which ants and stable over space and evolutionary time, we should find EFN-bearing plants interact in a natural environment convergence and complementarity of traits on both sides and affect the frequency of ants’ presence on such plants. of the ant-plant interaction (Dattilo´ et al. 2013a). For instance, competition, abundance and quality of In conclusion, we show that the interaction frequency resources, seasonality of nectar production, nutritional between ants and plants is an important characteristic to status and dispersal ability of the colony, and other biotic biologically describe the complex networks among these and abiotic factors (Bluthgen¨ & Fiedler 2004, Dattilo´ et al. organisms in the tropics. Our results contribute to our 2013a). understanding of network topology of free-living species. Most ant species are central place foragers, making However,itstillremainsuncertainhowthesequantitative foraging trips to food resources but consistently returning ecological networks are stable over space and time. In to their colony (Yamamoto & Del-Claro 2008); as a summary, our results show a new perspective of biological 168 WESLEY DATTILO´ ET AL.

processes for the study of ecological networks taking into BLUTHGEN,¨ N., VERHAAGH, M., GOIT´IA, W., JAFFE,´ K., MORAWETZ, consideration the interaction frequency. W.&BARTHLOTT,W.2000.Howplantsshapetheantcommunityin the Amazonian rainforest canopy: the key role of extrafloral nectaries ACKNOWLEDGEMENTS and homopteran honeydew. Oecologia 125:229–240. BLUTHGEN,¨ N., STORK, N. E. & FIEDLER, K. 2004. Bottom-up control We are grateful to Mario´ Almeida–Neto, Thiago Izzo, and co-occurrence in complex communities: honeydew and nectar and two anonymous reviewers for valuable comments on determine a rainforest ant mosaic. Oikos 106:344–358. earlierversionsofthemanuscript.WealsothanktheONF- CHAMBERLAIN, S. A., KILPATRICK, J. R. & HOLLAND, J. N. 2010. Brasil and the PPBio for logistical and financial support. Do extrafloral nectar resources, species abundances, and body sizes WD is grateful for financial support by the CNPq and contribute to the structure of ant–plant mutualistic networks? CONACYT. K.D.C. and D.L. thank CNPq for a research Oecologia 164:741–750. ´ ˜ grant. This is publication 38 in the NEBAM technical DATTILO, W., GUIMARAES, P. R. & IZZO, T. J. 2013a. Spatial structure series. of ant-plant mutualistic networks. Oikos 122:1643–1648. DATTILO,´ W., MARQUITTI, F. M. D., GUIMARAES,˜ P. R. & IZZO, T. J. 2013b. The structure of ant–plant ecological networks: is abundance LITERATURE CITED enough? Ecology (in press). GUIMARAES,˜ P. R., JORDANO, P. & THOMPSON, J. N. 2011. Evolution ALMEIDA–NETO, M. & ULRICH, W. 2011. A straightforward com- and coevolution in mutualistic networks. Ecology Letters 14:877– putational approach for measuring nestedness using quantitative 888. matrices. Environmental Modelling & Software 26:173–178. KRISHNA, A., GUIMARAES,˜ P. R., JORDANO, P. & BASCOMPTE, J. ALMEIDA–NETO, M., GUIMARAES,˜ P. R., GUIMARAES,˜ P., LOYOLA, R. 2008. A neutral–niche theory of nestedness in mutualistic networks. D. & URLICH, W. 2008. A consistent metric for nestedness analysis Oikos 117:1609–1618. in ecological systems: reconciling concept and measurement. Oikos RICO–GRAY, V. & OLIVEIRA, P. S. 2007. The ecology and evolution of 117:1227–1239. ant–plant interactions. University of Chicago Press, Chicago. 331 pp. BASCOMPTE, J., JORDANO, P., MELIAN,´ C. J. & OLESEN, J. M. 2003. The ULRICH, W., ALMEITA-NETO, M. & GOTELLI, N. J. 2009. A consumer’s nested assembly of plant–animal mutualistic networks. Proceedings guide to nestedness analysis. Oikos 118:3–17. of the National Academy of Sciences USA 100:9383–9387. WHALEN, M. A. & MACKAY, D. A. 1988. Patterns of ant and herbivore BLUTHGEN,¨ N. 2010. Why network analysis is often disconnected from activity on five understorey euphorbiaceous saplings in submontane community ecology: a critique and an ecologist’s guide. Basic and Papua New Guinea. Biotropica 20:294–300. Applied Ecology 11:185–195. YAMAMOTO, M. & DEL-CLARO, K. 2008. Natural history and foraging BLUTHGEN,¨ N. & FIEDLER, K. 2004. Competition for composition: behavior of the carpenter ant Camponotus sericeiventris Guerin,´ 1838 lessons from nectar-feeding ant communities. Ecology 85:1479– (Formicinae, Campotonini) in the Brazilian tropical savanna. Acta 1485. Ethologica 11:55–65. 53

Chapter V

Dáttilo, W.; Diaz-Castelazo, C.; Rico-Gray, V. (2014). Ant dominance hierarchy

determines the nested pattern in ant-plant networks. Biological Journal of the

Linnean Society 113: 405-414. bs_bs_banner

Biological Journal of the Linnean Society, 2014, 113, 405–414. With 3 figures

Ant dominance hierarchy determines the nested pattern in ant–plant networks

WESLEY DÁTTILO1,*, CECILIA DÍAZ-CASTELAZO2 and VICTOR RICO-GRAY1

1Instituto de Neuroetología, Universidad Veracruzana, Xalapa, Veracruz, C.P. 91190, Mexico 2Red de Interacciones Multitroficas, Instituto de Ecología A.C., Xalapa, Veracruz, C.P. 91070, Mexico

Received 21 April 2014; revised 12 May 2014; accepted for publication 13 May 2014

Extrafloral nectar (EFN) is a predictable and renewable resource for many ant colonies, and different ant species compete strongly to obtain and monopolize this highly nutritious food resource. Despite the importance of competition in structuring patterns of ant–plant interactions, this biological mechanism has been largely ignored in studies involving ant–plant networks. In this study we investigate the role of ant dominance hierarchy in structuring an ecological network involving ants and EFN-bearing plants in a tropical coastal environment in Mexico. We show that within a nested ant–plant network, ant species found in the central core of highly interacting species were competitively superior, showing massive recruitment and resource domination, compared with peripheral species with fewer interactions. Moreover, we also observed that both central and peripheral ant species have the ability to quickly find the food resource. However, after 2 h of observation, central ant species are more frequently collected on the food resource when compared with peripheral species. We hypothesize that the existence of a central core of competitive ant species may indicate that most plant species found within ant–plant networks could be better protected against herbivory by these dominant ant species. In short, our results highlight the importance of competition and monopolization in the resource use by ants in the maintenance of the nested pattern in ant–plant mutualistic networks. © 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414.

ADDITIONAL KEYWORDS: ant–plant interactions – competition – ecological networks – food webs – nestedness – numerical dominance.

INTRODUCTION action networks (e.g. abundance, phylogeny, spatial process and species richness) (Medan et al., 2007; One of the main goals in ecology is to identify how Rezende, Jordano & Bascompte, 2007; Vázquez et al., different species interact among themselves, and how 2009). these interactions persist over space and time (Ings Although knowledge about the organization of eco- et al., 2009). Recent studies have used tools derived logical networks has increased in many study systems from network theory to investigate the complex (e.g. plant–pollinator and plant–disperser) (Memmott organization of these interactions at the ecological et al., 2007; Santamaría & Rodríguez-Gironés, 2007), community level (see Hagen et al., 2012 and refer- only recently have studies focused on the biological ences therein). Under this unified framework of factors that contribute to structuring interaction species interactions, studies have focused on the networks involving ants and plants with extrafloral structural properties of these networks on different nectaries (EFN-bearing plants) (Chamberlain, systems and habitats throughout the Earth (Olesen & Kilpatrick & Holland, 2010; Lange, Dáttilo & Jordano, 2002; Schleuning et al., 2012). Moreover, Del-Claro, 2013). In these systems, plants secrete a some studies have highlighted different factors that liquid rich in carbohydrates (predominantly sucrose, contribute to the organization of these complex inter- glucose and fructose), amino acids and other com- pounds that attract and benefit ants (Rico-Gray, 1989; Koptur, Rico-Gray & Palacios-Rios, 1998; Heil et al., *Corresponding author. E-mail: [email protected] 2000; Byk & Del-Claro, 2011). These EFN-bearing

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 405 406 W. DÁTTILO ET AL. plants are protected against herbivores as a conse- in individual-based networks (Araújo et al., 2010; quence of nectar foraging by ants (Del-Claro, Berto & Pires et al., 2011). For instance, if individuals with Réu, 1996; Rico-Gray & Oliveira, 2007). It is our different competitive abilities exhibit identical prefer- understanding that ecological networks involving ences for different food resources, the diets of more ants and EFN-bearing plants show a highly nested competitive individuals represent subsets of the pattern (Guimarães et al., 2006; Rico-Gray et al., diets of other less competitive individuals, generating 2012; Dáttilo et al., 2013a). This indicates that within nestedness (Araújo et al., 2010; Tinker et al., 2012). an ant–plant network, a central core of species has Here we investigated the role of ant dominance many interactions among themselves, and peripheral hierarchy structuring an ant–plant network. We pos- species with few interactions interact with a proper tulated that due to the important role of competition subset of the central core of generalists with the most in structuring many ant communities around the interactions (Bascompte et al., 2003; Guimarães et al., world (Parr et al., 2005; Hölldobler & Wilson, 2008; 2006). Baccaro et al., 2012; Soares, 2013), competition Recent studies have focused on the ecological and among ants could also contribute to the structure of evolutionary dynamics of the central core of general- ecological networks involving ants and EFN-bearing ist ant species in ant–plant networks (Lange et al., plants. Specifically, we hypothesized that ants 2013; Dáttilo et al., 2013a; Dáttilo, Guimarães & Izzo, found in the central core of generalists are competi- 2013b). These studies show that ant species present tively superior to other peripheral ant species, in the central core interact with plants more than mainly because these species have the ability to expected by their natural abundances on the vegeta- maintain stable interactions with EFN-bearing plants tion (Dáttilo et al., 2014a). Thus, it is considered (Sánchez-Galván et al., 2012; Lange et al., 2013; that the spatial abundance outside of the ecological Dáttilo et al., 2013a, 2013b). To test our hypothesis, network only partially explains the probability in we collected data of ant–plant interactions in a tropi- which these generalist ant species find their food cal environment in coastal Veracruz, Mexico, and resource (Dáttilo et al., 2014a). Moreover, ant compo- empirically determined ant dominance hierarchy sition in the central core can be stable over large based on the resource use by ants through field geographical distances, phenological phases of nectar observations. Finally, to describe other biological secretion or even after perturbations caused by tropi- parameters of this ant–plant interaction, we also cal hurricanes (Sánchez-Galván, Díaz-Castelazo & investigated ant recruitment behaviour, discovery Rico-Gray, 2012; Lange et al., 2013; Dáttilo et al., and monopolization abilities between central core and 2013a). Based on the stability of these central ant peripheral species. species in monopolizing the resource, we can expect that such ants could be competitively superior (i.e. by massive recruitment and/or aggressive behaviour) MATERIALS AND METHODS when compared with the other species found in the STUDY AREA network (Dáttilo et al., 2013b, 2014b, 2014c). Research was conducted at the Centro de Investi- Inter- and intra-specific competition among ants is gaciones Costeras La Mancha (CICOLMA), located on considered one of the main mechanisms structuring the central coast of the state of Veracruz, Mexico local ant assemblages (Andersen, 1992; Parr et al., (19°36′N, 96°22′W; < 100 m a.s.l.). The climate is 2005; Hölldobler & Wilson, 2008; Soares, 2013). This warm and sub-humid and experiences three well- strong role of competition possibly occurs because defined seasons: the dry season from February to most ant species are central place foragers and have May, rainy season from June to September, and similar requirements (nesting site and food supplies), ‘Nortes’ or cold front season from October to January. resulting in less overlapping of their foraging areas Annual precipitation is about 1500 mm, and mean (López, Serrano & Acosta, 1994). In ant–plant inter- annual temperature is between 22 and 26 °C actions, competitively superior and territorial ant (Moreno-Casasola, 2006). The major vegetation types species can limit access to the resource by submissive in our study area are tropical dry and deciduous species mainly due to nectar being a highly nutritious forests, and sand dune scrub (Rico-Gray, 1993). and predictable resource (Heil & McKey, 2003; Blüthgen & Fiedler, 2004a,b; Rico-Gray & Oliveira, 2007). In fact, based on the importance of competition SAMPLING ANT–PLANT INTERACTIONS for ant community structure, we can expect that it We observed ant–plant interactions in April and May is also a very important force shaping ecological 2013 (08:00–13:00 h) along six arbitrarily selected, network structure. Theoretical studies based on the but representative 2-km trails that sampled different shared-preferences model (Svanbäck & Bolnick, 2005) environmental types: Trail 1, sand dune pioneer suggest that competition can lead to a nested pattern species; Trail 2, deciduous forest; Trail 3, deciduous

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 DOMINANCE HIERARCHY IN ANT-PLANT NETWORKS 407 forest–dry forest ecotone; Trail 4, dry forest and sand To quantify the numerical dominance of ants, we dune scrub; Trail 5, sand dune scrub; and Trail 6, collected data of competitive ability for each ant sand dune–freshwater lagoon ecotone. Each trail was species regardless of our interaction network. For covered twice. We sampled different habitats mainly this, in each of the six environments in the study site, due to low diversity of EFN-bearing plant species we determined 20 sampling points, 15 m apart. At within each habitat (see Supporting information each sampling point, we put two paired baits (< 5cm Table S1). Distance between trails varied depending apart), totalling 240 baits (6 environments × 20 sam- on how clustered the habitats were (approximately pling points × 2 baits). Each bait was made of a mix 100 and 800 m). For instance, the dune pioneer of honey (75%) with sardine (25%) placed on a plastic species were located along the beach, whereas the card (7.5 × 12.5 cm) ad libitum. We used plastic cards forest trails were perpendicular to the previous and not paper cards to avoid the co-dominance of ants trail. The sand dune scrub was located behind the in the top and bottom of baits. We used honey and dune pioneers (c. 100 m distance). In each visit, two sardine baits because ants that feed on nectar show researchers walked the trails and recorded all ant preferences for food resources rich in carbohydrates species feeding on EFNs present on the spike, pedicel, and proteins (Koptur & Truong, 1998; Blüthgen & bud, calyx, leaves, shoots, petioles, bracts or stems Fiedler, 2004a). At each sampling point, the first bait (based on Rico-Gray, 1993). Our research covered was left for 15 min and the second bait was left for most of those, if not all, habitats present. Also, even 2 h. This sampling procedure enabled us to quantify though it is a quite small area (as mentioned above), whether there is a difference in the ants’ ability to the variety of habitats found within the area is such discover and monopolize the resource. Ant species that one ‘needs’ to sample all places. Many studies visiting the baits were collected and preserved in 70% have been published from this study area by alcohol for subsequent identification. Ants were iden- our research group using this same method (see tified in the laboratory using the database of the Rico-Gray, 1993; Díaz-Castelazo et al., 2010, 2013; Instituto de Ecología, A.C, Mexico. Rico-Gray et al., 2012; Sánchez-Galván et al., 2012). The number of interactions of an ant species could be an artefact of its spatial abundance. In other words, abundant species could interact most fre- SAMPLING DOMINANCE HIERARCHY AND quently with each other and with other less abundant SPATIAL ABUNDANCE species (Vázquez et al., 2007; Dáttilo et al., 2014a). To Studies have evaluated ant competition in different evaluate the role of ant spatial abundance structuring ways (Parr & Gibb, 2012). Here, we used numerical ant–plant networks, we also determined the spatial dominance (Andersen, 1992; Cerdá, Retana & Cros, abundance of ants found in our study area. For this, 1997) to evaluate the role of ant dominance hierarchy we placed pitfall traps at the same sampling points on the structure of our ant–plant network. In this used in the experiment described above, totalling 120 case, numerically dominant ant species are those that traps (20 sampling points × 6 environments). We occur and monopolize many of the baits within an chose soil pitfall traps and not arboreal pitfall traps environment (Parr, 2008). This is an effective method for two reasons. First, most environments in our when the aim is to assess which ant species wins study area have low and open vegetation, indicating the competition by resource, and not how they win that ants that feed on EFN-bearing plants are (Andersen, 1992; Parr & Gibb, 2012). We chose this ground-nesting. Second, using this method we had a method because numerical and behaviour dominance way to standardize sampling in all environments. To are extremely closely related, and therefore such a avoid the aggregation caused by the food resource method has been widely accepted and used in the ant (bait), we conducted sampling of the spatial abun- literature (Dejean & Corbara, 2003; Santini et al., dance of ants 2 days before sampling of competition. 2007; Parr, 2008; Parr & Gibb, 2012). Moreover, we observed few competitive interactions among ants in the field, making it difficult to build a valid model for NETWORK ANALYSIS AND STATISTICS all ant species; also, due to differences in physiological Based on all ant–plant interactions collected in our and ontogenetic conditions at each colony, one-on-one study area we built an adjacency matrix A, in which interactions may not be the best way to quantify aij = 1 if the consumption of EFN from a plant species competition in ant communities (Hölldobler & Wilson, j by an ant species i was recorded, and zero otherwise 2008; Wittman & Gotelli, 2011). Moreover, based on (Bascompte et al., 2003). Subsequently we tested for the ‘ghost of competition past’, if competition among nestedness, a basic property used in studies involving ants is really strong in our study area, we would ecological networks. In a nested interaction network, expect that competitive interactions are rarely seen in species with a higher number of interactions tend to the field (Connel, 1980). interact with each other, while those species with few

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 408 W. DÁTTILO ET AL. interactions tend to interact with highly interactive To evaluate the role of dominance hierarchy struc- species in cohesive subgroups. We computed the turing the nested pattern in ant–plant networks, we nestedness of our ant–plant network using the NODF initially ranked each ant species according to its metric (Nestedness based on Overlap and Decreasing position in an ordered matrix for nestedness, which is Fill) (Almeida-Neto et al., 2008) in ANINHADO proportional to their number of interactions (Dáttilo (Guimarães & Guimarães, 2006). The NODF values et al., 2014a). We then correlated the ants’ position in range from 0 (non-nested) to 100 (perfectly nested). the nestedness ranking with their numerical domi- This metric is less sensitive to matrix size and shape nance index using a Spearman rank correlation test. and also less prone to type I error when compared Once we found 17 ant species (see Results), our with other metrics available in the literature, mainly nestedness ranking ranges from 1 (totally central core because this metric tends to overestimate the degrees species and with the largest number of interactions) of nestedness in the empirical matrix (Almeida-Neto to 17 (totally peripheral species and with the lowest et al., 2008). To verify the significance of nestedness, number of interactions). However, the results of we tested the empirical values of the ant–plant numerical dominance analysis could just be an arte- network against null distributions of these values. fact of the ants’ spatial abundance. Therefore, we For this, we computed 1000 simulated networks gen- also tested whether there was a relationship between erated by Null Model II (Bascompte et al., 2003). In the ant’s spatial abundance and its position in the this null model, the probability of an interaction nestedness ranking (Spearman correlation). In order occurring is proportional to the number of interac- not to overestimate the ant species with more efficient tions of both ants and plants, but also on the habitat systems for recruiting, we calculated the spatial where they occur. In addition, we also defined central abundance of ants based on the frequency of species core or peripheral ant species components of our occurrence in the pitfall traps and not based on the network using the following formula: number of individuals (Gotelli et al., 2011). We also used t-tests to test whether the numerical dominance − index and the number of individuals recruited per = kkimean Gc ( ) bait is greater in ant species of the central core than σ k in peripheral ant species. We used Pearson’s chi-

(Dáttilo et al., 2013b), where ki is the mean number square to test whether ant species of the central core of links for a given ant species i, kmean is the mean discover (bait left for 15 min) and/or monopolize food number of links for all ant species in the network and resources (bait left for 2 h) more frequently than

σk is the standard deviation of the number of links for peripheral species. Finally, based our sampling ant species. Gc > 1 are ant species presents on the design, in which the number of partners that one central core, and Gc < 1 are peripheral ant species. species can interact with could be directly related to To assess the competitive ability of each ant the number of habitats in which they occur, once a species, we calculated the numerical dominance (N) higher number of habitats is occupied by the same only for those ant species found in the ant–plant species, the probability to interact with all partner network. We calculated the numerical dominance species in the network will also be higher. Therefore, index for each ant species i using the formula: we performed a simple linear regression (SLR) using the number of environments that a species was col- lected as an independent variable and their number = Di N ( ), of links within network. We performed Spearman DSii+ correlations, t-tests, chi-square tests and SLR using

where Di is the number of monopolized baits by ant the R-software version 3.0.1 (R Development Core species i; Si is the number of baits that ant species i Team, 2013). used but did not monopolize; these were therefore classified as submissive ant species. Baits were con- RESULTS sidered monopolized when > 10 individuals (workers or soldiers) of the same species were using the We observed 48 interactions involving 16 plant and resource without the presence of other species. There- 17 ant species (Fig. 1). Plant species were included fore, dominant species were those that occur and in ten families, and the Fabaceae was the most well- monopolize a large proportion of food resources in an represented family (N = 5 species). For ants, we environment. Our index ranges from 0 (totally sub- recorded four subfamilies, and Myrmicinae was the missive species) to 1 (totally dominant species), and most well represented (N = 7 species) (see Table S1). is similar to the ‘monopolizing index’ used in other Moreover, the Myrmicinae was the most frequent studies (Fellers, 1987; Santini et al., 2007; Parr & subfamily based on baits (N = 83 baits, 34.5% of the Gibb, 2012). total baits).

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 DOMINANCE HIERARCHY IN ANT-PLANT NETWORKS 409

Figure 1. Ecological network involving EFN-bearing plants (circles) and ants (squares) sampled at La Mancha on the coast of Veracruz, Mexico. Each node represents one plant or ant species, and lines represent ant–plant interactions. *Those species that were present in the generalist core of the network.

We observed a significant nested pattern in our per bait in relation to peripheral species (Core = ant–plant network (observed matrix: NODF = 33.40; 16.64 ± 4.87; Periphery = 6.56 ± 3.41) ((t = −4.355; mean ± SD of simulated matrices: NODF = 24.61 ± d.f. = 15; P < 0.001). 4.55; P = 0.02). Three plant species (18.75% of the Additionally, we did not find a relationship between total species recorded) and three ant species (17.64% the ants’ spatial abundance and their position in the of the total species recorded) were found as part of nestedness ranking (rs = −0.024; t = −0.09; P = 0.92) the central core: (i) plant species: Opuntia stricta (Fig. 2B), indicating that ant abundance had little or (Cactaceae), Macroptilium atropurpureum (Fabaceae) no influence on the observed nested pattern. Finally, and Mansoa hymenaea (Bignoniaceae); (ii) ant species: we observed that the ability of ant species in the Forelius pruinosus (Dolichoderinae), Camponotus central core to discover baits after 15 min of observa- planatus (Formicinae) and Monomorium ebeninum tion is equal to the ability of peripheral species (Myrmicinae). Moreover, we observed that the number (χ2 = 0.890; d.f. = 1; P = 0.345). However, after 2 h of of partners that one species can interact with is not observation, ant species of the central core were related to the number of habitats in which they occur 33.9% more frequently collected on baits when com- (r2 = 0.11; P = 0.216). These findings indicate that pared with peripheral species (χ2 = 4.082; d.f. = 1; species that have the highest centrality network P = 0.04). (i.e. those with the most interactions) values are not the same species observed in most of the sampled DISCUSSION habitats. Ant species found in the central core were more Nestedness is a non-random property in mutualistic dominant than peripheral species (numerical domi- networks (Bascompte et al., 2003; Lewinsohn et al., nance index: mean ± SD, Core: N = 0.81 ± 0.18; 2006; Guimarães et al., 2007; Hagen et al., 2012). Periphery: N = 0.31 ± 0.27) (t = −2.602; d.f. = 15; However, little is known about the biological mecha- P = 0.02). Moreover, the level of generalization of nisms that predict this pattern of interaction in the ant species (position in the nestedness ranking) ant–plant networks (Chamberlain et al., 2010; was negatively related to their numerical dominance Rico-Gray et al., 2012; Lange et al., 2013). Here we

(rs = −0.497; t = −2.22; P = 0.02) (Figs 2A, 3). Ant show that although ants have an extremely territorial species of the central core recruited more individuals effect only near their nests, competition among ants is

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 410 W. DÁTTILO ET AL.

chy. We found a strong relationship between ant competitive hierarchy and their number of interac- tions within network (i.e. position in the nestedness ranking). Corroborating with the shared-preferences model, we observed that ants that interact with a large number of plant species and are part of the central core of the network are competitively superior to submissive ants that interact with few plant species and are part of peripheral species. In fact, extrafloral nectar is a predictable and highly nutritive resource for ant colonies, and dominant species use diversified strategies to discover and maintain this resource, which is aggressively protected (Dreisig, 2000; Blüthgen & Fiedler, 2004a, b; Falcão, Dáttilo & Izzo, 2014). Therefore, as most ground-dwelling ant species forage a few metres away from their colonies and strongly defend their territories and food resources (Fellers, 1987; López et al., 1994; Baccaro & Ferraz, 2013), we show that competition might also be one of the most important factors shaping ant–plant networks. When interacting with EFN-bearing plants, ants have as a main functional role to defend their host plants (and their food source: EFN and honeydew- producing insects) against potential natural enemies (Becerra & Venable, 1989; Fiala, 1990; Del-Claro & Oliveira, 1993; Rico-Gray & Oliveira, 2007). Defence efficiency correlates directly with the number of workers recruited (Heil & McKey, 2003; Rico-Gray & Oliveira, 2007). Here, we show that ants in the central core recruit a larger number of workers on the food resource when compared with peripheral and submissive species of the network. Therefore, we hypothesize that ants in the central core can more Figure 2. Spearman correlation between the position of effectively defend their host plants. Additionally, each ant species in the rank of nestedness and their (A) several studies have shown that submissive ant numerical dominance (rs = −0.497; P = 0.02) and (B) spatial species have the ability to find food resources and abundance (rs = −0.024; P = 0.92). remove them more quickly before the arrival of the dominant species (Wilson, 1971; Fellers, 1987; strong enough to structure ant–plant networks. Spe- Perfecto, 1994; Davidson, 1998). However, we found cifically, we show that ant position within the nested no difference in the frequency with which the ant–plant network can be predicted only by differ- dominant (central core of the network) and submis- ences among the competitive ability (numerical domi- sive (periphery of network) species find their food nance and recruitment) of ant species. Interestingly, resources. These findings corroborate a recent review this association was not contingent upon the relative by Parr & Gibb (2012), with only a few ant commu- spatial abundance of ants in the environment, which nities having such a dominance-–discovery trade-off indicates that spatial abundance could not be the (ability to discover versus the ability to monopolize main factor explaining the patterns of ant–plant the resource), mainly because many other factors interactions. could play a major role in structuring such a foraging Competition as a biological process within ecologi- strategy. By contrast, after 2 h of observation, central cal networks has been suggested mainly in theoretical ant species are more frequently collected on the food studies (Bastolla et al., 2009; Araújo et al., 2010; Pires resource, which indicates that central ant species et al., 2011; Tinker et al., 2012). Here, we empirically have the ability both to discover and to monopolize show for a Mexican tropical forest that the nested food resources. In fact, ants specialized to feed on pattern found in ant–plant networks can be gener- liquid diets rich in carbohydrates can often break ated by the difference in the ants’ dominance hierar- this trade-off, as they are highly territorial and can

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 DOMINANCE HIERARCHY IN ANT-PLANT NETWORKS 411

Figure 3. Adjacency matrix representing ant numerical dominance. Ant and plant species are ordered according to nestedness (number of links). Each column represents one ENF-bearing plant species, each row represents an ant species and filled cells represent interactions among species. Darker cells represent the higher ant numerical dominance. *Those species that were present in the generalist core of the network. discover and monopolize the resource (Davidson, also thank Alex Wild (University of Illinois) for kindly 1998). Thus, these results support the mechanism in providing the ant photo used in Figure 1. We grate- which ant species of the central core quickly discover fully acknowledge funding from the CNPq (no. the resource and then may exclude peripheral and 237339/2012-9) and CONACYT (no. 489746) to W.D. submissive species. Although competition has been largely ignored in REFERENCES studies of mutualistic networks involving ants and EFN-bearing plants (but see Chamberlain & Holland, Almeida-Neto M, Guimarães PR Jr, Guimarães P, 2009; Dáttilo et al., 2014c), our results show that Loyola RD, Urlich W. 2008. A consistent metric for dominance hierarchy plays a major role in structuring nestedness analysis in ecological systems: reconciling the nested pattern in ant–plant networks, and that concept and measurement. Oikos 117: 1227–1239. ants found in the central core of generalists are com- Andersen AN. 1992. Regulation of ‘momentary’ diversity by petitively superior to peripheral and submissive ant dominant species in exceptionally rich ant communities of species. The next step would be to evaluate more the Australian seasonal tropics. American Naturalist 140: 401–420. accurately how the variation in the quality and quan- Araújo MS, Martins EG, Cruz LD, Fernandes FL, tity of nectar could explain the reciprocal functional Linhares AX, Dos Reis SF, Guimaraes PR. 2010. Nested effects involving ants and EFN-bearing plants. diets: a novel pattern of individual-level resource use. Oikos 119: 81–88. ACKNOWLEDGEMENTS Baccaro FB, De-Souza JLP, Franklin E, Landeiro VL, Magnusson WE. 2012. Limited effects of dominant ants We thank Paulo Guimarães, Fabricio Baccaro, Thiago on assemblage species richness in three Amazon forests. Izzo, Laura Hernández, Armando Chacón and two Ecological Entomology 37: 1–12. anonymous reviewers for helpful discussions and/or Baccaro FB, Ferraz G. 2013. Estimating density of ant nests comments on earlier versions of the manuscript. We using distance sampling. Insectes Sociaux 60: 103–110.

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 412 W. DÁTTILO ET AL.

Bascompte J, Jordano P, Melián CJ, Olesen JM. 2003. forests: spatio-temporal dynamics and resource use in the The nested assembly of plant–animal mutualistic networks. canopy. Cambridge: Cambridge University Press, 341–347. Proceedings of the National Academy of Sciences of the Del-Claro K, Berto V, Réu W. 1996. Effect of herbivore United States of America 100: 9383–9387. deterrence by ants increase fruit set in an extrafloral Bastolla U, Fortuna MA, Pascual-García A, Ferrera A, nectary plant Qualea multiflora (Vochysiaceae). Journal of Luque B, Bascompte J. 2009. The architecture of Tropical Ecology 12: 887–892. mutualistic networks minimizes competition and increases Del-Claro K, Oliveira PS. 1993. Ant–homoptera interaction: biodiversity. Nature 458: 1018–1020. do alternative sugar sources distract tending ants? Oikos Becerra JX, Venable DL. 1989. Extrafloral nectaries: a 68: 202–206. defense against ant–Homoptera mutualisms? Oikos 55: 276– Díaz-Castelazo C, Guimarães PR, Jordano P, Thompson 280. JN, Marquis RJ, Rico-Gray V. 2010. Changes of a Blüthgen N, Fiedler K. 2004a. Competition for composition: mutualistic network over time: reanalysis over a 10-year lessons from nectar-feeding ant communities. Ecology 85: period. Ecology 91: 793–801. 1479–1485. Díaz-Castelazo C, Sánchez-Galván IR, Guimarães PR, Blüthgen N, Fiedler K. 2004b. Preferences for sugars and Raimundo RLG, Rico-Gray V. 2013. Long-term temporal amino acids and their conditionality in a diverse nectar- variation in the organization of an ant–plant network. feeding ant community. Journal of Animal Ecology 73: Annals of Botany 111: 1285–1293. 155–166. Dreisig H. 2000. Defense by exploitation in the Florida Byk J, Del-Claro K. 2011. Ant–plant interaction in the carpenter ant, Camponotus floridanus, at an extrafloral Neotropical savanna: direct beneficial effects of extrafloral nectar resource. Behavioral Ecology and Sociobiology 47: nectar on ant colony fitness. Population Ecology 53: 327–332. 274–279. Cerdá X, Retana J, Cros S. 1997. Thermal disruption of Falcão JCF, Dáttilo W, Izzo TJ. 2014. Temporal variation in transitive hierarchies in Mediterranean ant communities. extrafloral nectar secretion in different ontogenetic stages of Journal of Animal Ecology 66: 363–374. the fruits of Alibertia verrucosa S. Moore (Rubiaceae) in a Chamberlain SA, Holland JN. 2009. Body size predicts Neotropical savanna. Journal of Plant Interactions 9: 137– degree in ant–plant mutualistic networks. Functional 142. Ecology 23: 196–202. Fellers JH. 1987. Interference and exploitation in a guild of Chamberlain SA, Kilpatrick JR, Holland JN. 2010. Do woodland ants. Ecology 68: 1466–1478. extrafloral nectar resources, species abundances, and body Fiala B. 1990. Extrafloral nectaries versus ant–Homoptera sizes contribute to the structure of ant–plant mutualistic mutualisms: a comment on Becerra and Venable. Oikos 59: networks? Oecologia 164: 741–750. 281–282. Connel JH. 1980. Diversity and the coevolution of competi- Gotelli NJ, Ellison AM, Dunn RR, Sanders NJ. 2011. tors, or the ghost of competition past. Oikos 35: 131–138. Counting ants (Hymenoptera: Formicidae): biodiversity Dáttilo W, Fagundes R, Gurka CAQ, Silva MSA, sampling and statistical analysis for myrmecologists. Vieira MCL, Izzo TJ, Díaz-Castelazo C, Del-Claro K, Myrmecological News 15: 13–19. Rico-Gray V. 2014c. Individual-based ant-plant networks: Guimarães PR, Rico-Gray V, Dos-Reis SF, Thompson diurnal-nocturnal structure and species-area relationship. JN. 2006. Asymmetries in specialization in ant–plant PLoS ONE 9: e99838. mutualistic networks. Proceedings of the Royal Society of Dáttilo W, Guimarães PR, Izzo TJ. 2013b. Spatial structure London B 273: 2041–2047. of ant–plant mutualistic networks. Oikos 122: 1643– Guimarães PR, Sazima C, Dos-Reis SF, Sazima I. 2007. 1648. The nested structure of marine cleaning symbiosis: is it like Dáttilo W, Marquitti FMD, Guimarães PR, Izzo TJ. flowers and bees? Biology Letters 3: 51–54. 2014a. The structure of ant–plant ecological networks: is Guimarães PR Jr, Guimarães PR. 2006. Improving abundance enough? Ecology 95: 475–485. the analyses of nestedness for large sets of matrices. Envi- Dáttilo W, Rico-Gray V, Rodrigues DJ, Izzo TJ. 2013a. ronmental Modelling and Software 21: 1512–1513. Soil and vegetation features determine the nested pattern Hagen M, Kissiling WD, Rasmussen C, de Aguiar MAM, of ant–plant networks in a tropical rainforest. Ecological Brown L, Carstensen DW, Alves-dos-Santos I, Dupont Entomology 38: 374–380. YL, Edwards FK, Genini J, Guimaraes PR, Jenkins Dáttilo W, Sánchez-Galván I, Lange D, Del-Claro K, GB, Jordano P, Kaiser-Bunbury CN, Ledger M, Maia Rico-Gray V. 2014b. Importance of interaction frequency KP, Marquitti FMD, McLaughlin O, Morellato LPC, in analysis of ant–plant networks in tropical environments. O’Gorman EJ, Trojelsgaard K, Tylianakis JM, Vidal Journal of Tropical Ecology 30: 165–168. MM, Woodward G, Olesen J. 2012. Biodiversity, species Davidson DW. 1998. Resource discovery versus resource interactions and ecological networks in a fragmented world. domination in ants: a functional mechanism for breaking Advances in Ecological Research 46: 89–210. the trade-off. Ecological Entomology 23: 484–490. Heil M, Fiala B, Baumann B, Linsenmair KE. 2000. Dejean A, Corbara B. 2003. A review of mosaics of dominant Temporal, spatial and biotic variations in extrafloral nectar ants in rainforests and plantations. In: Basset Y, Novotny secretion by Macaranga tanarius. Functional Ecology 14: V, Miller SE, Kitching RL, eds. Arthropods of tropical 749–757.

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 DOMINANCE HIERARCHY IN ANT-PLANT NETWORKS 413

Heil M, McKey D. 2003. Protective ant–plant interactions environment for statistical computing. Vienna, Austria: R as model systems in ecological and evolutionary research. Foundation for Statistical Computing. Annual Review of Ecology, Evolution and Systematics 34: Rezende EL, Jordano P, Bascompte J. 2007. Effects of 425–453. phenotypic complementarity and phylogeny on the nested Hölldobler B, Wilson EO. 2008. The superorganism: the structure of mutualistic networks. Oikos 116: 1919–1929. beauty, elegance, and strangeness of insect societies. New Rico-Gray V. 1989. The importance of floral and circum- York: W.W. Norton. floral nectar to ants inhabiting dry tropical lowlands. Bio- Ings TC et al. 2009. Ecological networks – beyond food webs. logical Journal of the Linnean Society 38: 173–181. Journal of Animal Ecology 78: 253–269. Rico-Gray V. 1993. Use of plant-derived food resources Koptur S, Rico-Gray V, Palacios-Rios M. 1998. Ant pro- by ants in the dry tropical lowlands of coastal Veracruz, tection of the nectaried fern Polypodium plebeium in central Mexico. Biotropica 25: 301–315. Mexico. American Journal of Botany 85: 736–739. Rico-Gray V, Díaz-Castelazo C, Ramírez-Hernández A, Koptur S, Truong N. 1998. Facultative ant–plant interac- Guimarães PR, Holland JN. 2012. Abiotic factors tions: nectar sugar preferences of introduced pest ant shape temporal variation in the structure of an ant–plant species in south Florida. Biotropica 30: 179–189. network. Arthropod–Plant Interactions 6: 289–295. Lange D, Dáttilo W, Del-Claro K. 2013. Influence of Rico-Gray V, Oliveira PS. 2007. The ecology and evolution extrafloral nectary phenology on ant–plant mutualistic net- of ant–plant interactions. Chicago: University of Chicago works in a Neotropical Savanna. Ecological Entomology 38: Press. 463–469. Sánchez-Galván IR, Díaz-Castelazo C, Rico-Gray V. Lewinsohn TM, Prado PI, Jordano P, Bascompte J, 2012. Effect of Hurricane Karl on a plant–ant network Olesen JM. 2006. Structure in plant–animal interaction occurring in coastal Veracruz, Mexico. Journal of Tropical assemblages. Oikos 113: 174–184. Ecology 28: 603–609. López F, Serrano JM, Acosta FJ. 1994. Parallels between Santamaría L, Rodríguez-Gironés MA. 2007. Linkage the foraging strategies of ants and plants. Trends in Ecology rules for plant–pollinator networks: trait complementarity and Evolution 9: 150–153. or exploitation barriers? PLoS Biology 5: 354–359. Medan D, Perazzo RPJ, Devoto M, Burgos E, Santini G, Tucci L, Ottonetti L, Frizzi F. 2007. Competi- Zimmermann MG, Ceva H, Delbue AM. 2007. Analysis tion tradeoffs in the organisation of a Mediterranean ant and assembling of network structure in mutualistic assemblage. Ecological Entomology 32: 319–326. systems. Journal of Theoretical Biology 246: 510–521. Schleuning M et al. 2012. Specialization of mutualistic Memmott J, Craze PG, Waser NM, Price MV. 2007. Global interaction networks decreases toward tropical latitudes. warming and the disruption of plant–pollinator interac- Current Biology 22: 1925–1931. tions. Ecology Letters 10: 710–717. Soares SA. 2013. The role of competition in structuring ant Moreno-Casasola P. 2006. Entornos Veracruzanos: La Costa communities: a review. Oecologia Australis 17: 81–91. de La Mancha. Xalapa: Instituto de Ecología AC. Svanbäck R, Bolnick DI. 2005. Intraspecific competition Olesen JM, Jordano P. 2002. Geographic patterns in plant– affects the strength of individual specialization: an optimal pollinator mutualistic networks. Ecology 83: 2416–2424. diet theory model. Evolutionary Ecology Research 7: 993– Parr CL. 2008. Dominant ants can control assemblage 1012. species richness in a South African savanna. Journal of Tinker MT, Guimaraes PR, Novak M, Marquitti FMD, Animal Ecology 77: 1191–1198. Bodkin JL, Staedler M, Bentall G, Estes JA. 2012. Parr CL, Gibb H. 2012. The discovery–dominance trade-off Structure and mechanism of diet specialisation: testing is the exception, rather than the rule. Journal of Animal models of individual variation in resource use with sea Ecology 81: 233–241. otters. Ecology Letters 15: 475–483. Parr CL, Sinclair BJ, Andersen AN, Gaston KJ, Chown Vázquez DP, Blüthgen N, Cagnolo L, Chacoff NP. 2009. SL. 2005. Constraint and competition in assemblages: a Uniting pattern and process in plant–animal mutualistic cross continental and modeling approach for ants. American networks: a review. Annals of Botany 103: 1445–1457. Naturalist 165: 481–494. Vázquez DP, Melián CJ, Williams NM, Blüthgen N, Perfecto I. 1994. Foraging behavior as determinant of asym- Krasnov BR, Poulin R. 2007. Species abundance and metric competitive interaction between two ant species in a asymmetric interaction strength in ecological networks. tropical agroecosystem. Oecologia 98: 184–192. Oikos 116: 1120–1127. Pires MM, Guimarães PR, Araújo MS, Giaretta AA, Wilson EO. 1971. The insect societies. Cambridge: Belknap Costa JCL, Dos Reis SF. 2011. The nested assembly of Press. individual-resource networks. Journal of Animal Ecology Wittman SE, Gotelli NJ. 2011. Predicting community struc- 80: 896–903. ture of ground-foraging ant assemblages with Markov R Development Core Team. 2013. R: a language and models of behavioral dominance. Oecologia 166: 207–219.

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 414 W. DÁTTILO ET AL.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Table S1. Plant and ant species recorded in April and May 2013 at La Mancha on the coast of Veracruz, Mexico. Habitat occurrence: SDP, sand dune pioneer species; DFO, deciduous forest; DDF, deciduous forest–dry forest ecotone; DFS, dry forest and sand dune scrub; SDS, sand dune scrub; SDF, sand dune–freshwater lagoon ecotone.

© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113, 405–414 64 9.) General conclusion

In this doctoral thesis I showed for the first time that symbiotic ant-plant networks involving ants and myrmecophytes are highly modular. This indicates that within a symbiotic ant-plant network, there are groups or modules of ant species that interact more frequently with a group of myrmecophyte species, with few interactions among network subgroups. The high modularity found in these networks is probably due to the elevated degree of specialization and compartmentation found in obligate and symbiotic mutualisms (Chapter 1).

Consequently, symbiotic networks were less robust for both ants and plants species extinction compared to nonsymbiotic networks (i.e., those involving ants and plants with extrafloral nectaries). Thus, the loss of very specialized ant-plant interactions can lead to a cascade effect of loss of other species and all ecological services provided by them (Chapter 2).

When evaluating abiotic factors affecting nonsymbiotic ant-plant networks, I found that soil pH was an important factor structuring the nested pattern in such networks. This possibly occurs because soil pH directly influences carbohydrate and amino acid nectar concentrations, which in turn affect the number of ant-plant interactions and thus the nested pattern of ant–plant networks (Chapter 3). However, unlike binary networks, quantitative networks are often significantly non-nested. Thus, species with a higher interaction frequency have a higher number of links, indicating that these species are possibly more abundant and/or competitive. These findings suggest a new perspective for the study of interaction networks in the tropics, since species with lower interaction frequency are not necessarily subsets of species with higher frequency, and consequently generate the non-nested pattern in quantitative networks (Chapter 4).

Finally, I also found in a non-symbiotic ant-plant network, that the number of interactions of an ant species can be explained only by differences among the competitive ability of the species. Specifically, ant species found in the central core of highly interacting 65 species were competitively superior, exhibiting massive recruitment and resource domination, compared to peripheral species with fewer interactions. I hypothesized that the existence of a central core of competitive ant species may indicate that the most plants species found within ant-plant networks could be better protected against herbivory by these dominant ant species

(Chapter 5). In short, my results in this doctoral thesis highlight the importance of the level of specialization and abiotic and biotic factors in the maintenance of the structure of ant-plant mutualistic networks.