Keywords

1 Multitrophic and Indirect Interactions

Species affect each other in countless ways. The most evident and easily visible influences are the direct trophic interactions, those involving feeding relationships between two species. On the other hand, indirect trophic interactions require the presence of a third species as an intermediary (Wootton 1994) and are propagated through feeding relationships. In a trophic cascade, a predator affects the density and/or behaviour of a prey and, therefore, indirectly affects the rate of consumption of a species in the next lower trophic level (Estes and Palmisano 1974). These interaction chains or indirect effects mediated by density (e.g. trophic cascades, keystone species, apparent competition) have been broadly demonstrated (e.g. Wootton 1994, 2002; Morris et al. 2004; van Veen et al. 2006). Indirect trophic interactions play an important role in structuring communities and have generally not been taken into account in traditional studies (Bukovinszky et al. 2008).

Indirect interactions between species can also happen in non-trophic ways (e.g. via habitat, trait or behaviour modification: Wootton 1994). However, these interactions remain largely ignored (Ogushi 2008). Indirect effects can cause changes in species density (e.g., interference competition, facilitation: Prugh and Brashares 2012; Wetzel et al. 2016) or changes in species interactions (interaction modification: Wootton 1994, 2002; Prasad and Snyder 2010; Pagès et al. 2012; Hammill et al. 2015). Species can modify interactions by modifying a trait of one of the interacting species (e.g. fear effects associated with predator presence; plant-mediated interactions: Prasad and Snyder 2010; Hammill et al. 2015) or by changing the biotic or abiotic physical environment where the interaction takes place (Pagès et al. 2012; Barbosa et al. 2019).

Indirect effects have received increasing attention in ecological research, and their role in structuring ecological communities has been demonstrated experimentally (Tompkins et al. 2000; Morris et al. 2004; Sanders et al. 2015; Barbosa et al. 2019). For example, the positive indirect effects that consumer species have on each other, reducing competition among their respective resource species, can help maintain the diversity of consumer species (Sanders et al. 2015). In addition, indirect interactions can be the cause of unexpected results often found in experimental manipulations of direct interactions (Doak et al. 2008). Barbosa et al. (2017), for instance, found that the experimental reduction in abundance of a species affects other unrelated species in a food web. In addition, many of the effects that have been interpreted in the past as the result of competition can be caused by other indirect interactions (Holt 1977). For example, in the so-called apparent competition (Holt 1977; Morris et al. 2004), the negative effect of one herbivore on another, mediated by a common predator, may produce a result that resembles competition. Thus, it has become clear that the structure and dynamics of ecological communities cannot be fully understood without taking indirect interactions into account.

2 The Baccharis dracunculifolia System

The Baccharis dracunculifolia D.C. (Asteraceae) system is a self-contained assemblage of arthropods living on this host shrub species (Figs. 5.1 and 5.2). The plant species is a perennial, evergreen, dioecious shrub, 2–3 m high, widely distributed in the south-central portion of South America (Espírito-Santo et al. 2003). In some parts of Brazil, B. dracunculifolia produces flowers twice a year from March to June and from November to December (Collevatti and Sperber 1997). B. dracunculifolia frequently forms distinct patches that vary from 18 to 12,000 m2 in area (Collevatti and Sperber 1997). This plant is fundamental to natural succession and regeneration (Fernandes et al. 2016) playing an important role in promoting biodiversity and ecosystem functioning (Perea et al. 2019).

Fig. 5.1
figure 1

Species engaged in multitrophic and indirect interactions in the Baccharis dracunculifolia system: (a) B. dracunculifolia; (b) and (c) galls induced by Baccharopelma dracunculifoliae; (d) nymph of the galler B. dracunculifoliae; (e) parasitoid wasp; (f) aphids Uroleucon tucumani invading a gall; (g) ant tending aphids inside a post-emergence gall; (h) aphids inside a post-emergence gall; (i) ants removing a free-feeding herbivore. Photos by Milton Barbosa

Full size image
Fig. 5.2
figure 2

Species engaged in multitrophic and indirect interactions in the Baccharis dracunculifolia system: (a) spider using a post-emergence gall as shelter; (b) ladybird; (c) weevil; (d–f) mistletoe Struthanthus flexicaulis and its germinating seeds. (Photos a–c by Milton Barbosa and d-f by António Cruz)

Full size image

The species B. dracunculifolia hosts many species of free-feeding herbivores, 17 gall-inducing species and also many species of predators, including the orders Araneae, Coleoptera, Mantodea and Hymenoptera (Fagundes et al. 2005; Fagundes and Fernandes 2011; Barbosa et al. 2017, 2019; Monteiro et al. 2020a). In Brazil, over the past three decades B. dracunculifolia has been extensively used in community ecology research. The diverse associated fauna, and the wide distribution, usually at high frequency and through gradients (elevational, higrothermal, and habitat disturbance), make the species an attractive study model (e.g. Fernandes and Price 1988; Ribeiro-Mendes et al. 2002; Fagundes et al. 2005; Julião et al. 2005; Barbosa et al. 2017, 2019; Monteiro et al. 2020a, b). As a result, several multitrophic and indirect interactions have been revealed in the system, involving B. dracunculifolia, its associated arthropods and a mistletoe species (Collevatti and Sperber 1997; Fernandes et al. 1999; Espírito-Santo and Fernandes 2002; Fagundes et al. 2005; Neves et al. 2011; Bahia et al. 2015; Monteiro et al. 2020b). Below, we describe the B. dracunculifolia system and its interactions, summarizing them in an interaction network (Table 5.1, Fig. 5.3).

Table 5.1 Empirically demonstrated interactions involving arthropods and plants in the Baccharis dracunculifolia system, in Brazil. Interaction codes correspond to the links in the interaction network (Fig. 5.3)
Full size table
Fig. 5.3
figure 3

Interaction network of arthropods and plants in the Baccharis dracunculifolia system, in Brazil. Links are categorized into direct interactions (solid arrows), indirect interactions (dashed orange arrows) and interaction modifications (dashed blue arrows). The link numbers refer to the interaction descriptions in Table 5.1

Full size image

2.1 Plant–Free-Feeding Herbivore Interaction

The most frequent orders of free-feeding herbivores found on B. dracunculifolia are Hemiptera, Coleoptera and Orthoptera (Interaction 1; see Table 5.1 and Figs. 5.1, 5.2 and 5.3, as for all interactions described hereafter). Among the chewing insects, the most common families are Chrysomelidae and Curculionidae. Chrysomelidae is one of the most numerous herbivorous insect families and has the greatest ecological and economic importance (e.g. monoculture pests: Riley et al. 2002, Chaboo 2007). The sucking insects most commonly found on B. dracunculifolia are from the families Cicadellidae, Membracidae and Psyllidae. These sucking insects often have a trophobiontic relationship with ants, which chase away other free-feeding herbivores. Ant effects depend on the context, the environment and the species of ants involved and influence the presence of other insects (Fagundes et al. 2005; Neves et al. 2011).

Interactions between herbivorous insects and their host plants can be affected by characteristics of the host plants or by changes in environmental conditions (Novotny et al. 2003; Fernandes 2016; Nunes et al. 2016a, b; Peters et al. 2016; Longino and Branstetter 2018). Monteiro et al. (2020a) investigated how the interaction between insect herbivores and B. dracunculifolia changed along an elevational gradient in the campo rupestre (rupestrian grassland) in Serra do Cipó, Brazil. In this study, the composition, diversity of interactions and specialization of insect herbivores changed with altitude. The richness and abundance of chewing insects increased with elevation, while the richness and abundance of sucking insects were higher at intermediate elevations. The change in diversity of insects associated with B. dracunculifolia along the elevation was due to abiotic and biotic factors. Abiotic factors such as increased wind speed and solar radiation and decreased temperature at higher elevations functioned as an environmental filter for many species. On the other hand, biotic factors such as the relationship between the richness of trophobiont ants and herbivores associated with B. dracunculifolia influenced the distribution of insect herbivore species.

2.2 Plant–Galler Interactions

The genus Baccharis supports the richest galling insect fauna recorded so far in the Neotropics, with 121 galling insect species found on 40 host species (Fernandes et al. 1996, 2014; Carneiro et al. 2009a, b). Among the several host species, B. dracunculifolia has the largest fauna of galling insects, with a total of 17 species recorded considering various locations (Fernandes et al. 1996, 2014; Barbosa et al. 2017). Galling insects represent appropriate model systems as they are usually straightforward to identify and count, which makes their populations easy to manipulate and monitor in the field (Barbosa et al. 2017, 2019). Galling insects also present very high specialization to their host plants and very persistent morphology, and, therefore, taxonomic identification can be indirectly made through the gall’s external morphology and through the identity of their host (Rohfritsch and Shorthouse 1982; Carneiro et al. 2009b). In addition, the fact that parasitoids do not cause immediate damage to the galler allows the quantification of interaction frequency (van Veen et al. 2006).

Baccharopelma dracunculifoliae (Sternorrhyncha: Psyllidae) is by far the most common galling insect, inducing up to 83% of galls on B. dracunculifolia (Araújo et al. 1995; Barbosa et al. 2017) (Interaction 2). This psyllid induces a gall on the central rib of the leaf, which folds over itself until the borders are joined, forming an elliptical, green, glabrous, unicameral gall, which usually houses up to 4 nymphs, although up to 21 have already been recorded (Lara and Fernandes 1994; Arduin et al. 2005). The gall is not completely sealed because the edges of the leaves are joined but not fused. The galls of B. dracunculifoliae are colonized by several inquiline arthropods (species that occupy a living space produced by another species – e.g., a gall), such as Diptera, Hemiptera (Aphididae), Neuroptera (Chrysopidae), Thysanoptera and Acarina immature (Collevatti and Sperber 1997; Espírito Santo and Fernandes 2002; Barbosa et al. 2019). The post-emergence galls can persist on the plant and dry out but are still occupied by inquilines (Espírito-Santo and Fernandes 1998; Barbosa et al. 2019). These post-emergence galls can trigger indirect effects that feedback to the galler modifying its interactions with other species (discussed below in Sect. 2.4; Barbosa et al. 2019).

According to Espírito-Santo and Fernandes (2002), the host plant B. dracunculifolia has strong bottom-up control over the galling population. Plant-mediated mortality in the galling insect B. dracunculifoliae, due to gall drop and plant resistance, represented 40.7% of the initial cohort studied (Interaction 3).

2.3 Galler–Parasitoid Interaction

The performance of the galling psyllid B. dracunculifoliae is strongly controlled by top-down effects. Espírito-Santo et al. (2004) found that parasitoid attack was the main cause (45.2%) of mortality of the galls of B. dracunculifoliae that survived plant defences. At least ten wasp species have been known to parasitize galls of B. dracunculifoliae (Barbosa et al. 2017), the main ones are Lyrcus sp. (Hymenoptera: Pteromalidae), Brasema sp. (Hymenoptera: Eupelmidae), Platygastridae sp. (Hymenoptera: Platygastridae) and Psyllaephagus baccharidis (Hymenoptera: Encyrtidae) (Tavares and Perioto 1993; Sperber and Collevatti 1996; Barbosa et al. 2017).

Psyllaephagus baccharidis is the most common parasitoid of the galling psyllid. In the survey carried out by Espírito-Santo et al. (2004), this parasitoid was responsible for 93% of the parasitism recorded in galls of B. dracunculifoliae (Interaction 4). P. baccharidis is a solitary koinobiont endoparasitoid that consumes the nymph and pupates inside the host’s cuticle, causing a so-called ‘nymph mummification’ (Espírito-Santo et al. 2004). The parasitoid attacks galls in the early stages of development, making them increase in size. Mortality by parasitism is clearly recognizable as parasitized nymphs have a distinct colour and shape (Espírito-Santo et al. 2004).

Apparently, the parasitoid P. baccharidis stimulates the nymphs to feed, increasing their size and also the size of the gall induced by B. dracunculifoliae (Espírito-Santo et al. 2004). This can decrease the performance of the plant, since larger galls represent larger nutrient sinks. In addition, unparasitized nymphs of B. dracunculifoliae sharing a gall with parasitized nymphs were larger than the ones found in unparasitized galls. This can result in adults of the gall inducer with greater survival and reproductive success (Espírito-Santo et al. 2004) (Interaction 5). Thus, the parasitoid P. baccharidis can have a positive indirect effect on the host plant, controlling the population of the gall inducer B. dracunculifoliae (Interaction 6), and a negative indirect effect, increasing the sinking of nutrients by the gall and the fitness of the galler (Espírito-Santo et al. 2004) (Interaction 7).

Barbosa et al. (2017) investigated the role of direct and indirect interactions in the structure and robustness (tolerance to species loss) of galler-parasitoid food webs on B. dracunculifolia. Theoretical models from previous works suggest that the loss or reduction in abundance of individual species can lead to secondary and cascading extinctions (Saavedra et al. 2008; Staniczenko et al. 2010). The authors experimentally manipulated quantitative host-parasitoid food webs to reduce the abundance of the main galler, B. dracunculifoliae. The perturbation resonated throughout the food web, affecting the food web structure and robustness (tolerance to species loss). Since there was no possibility for these effects to be propagated directly or indirectly via the documented trophic interactions, the effects must have spread non-trophically and/or through trophic links not included in the webs. The results emphasize that, even for a relatively simple ecological community, indirect interactions (trophic and non-trophic) are fundamental to their structure and dynamics (Fontaine et al. 2011; Kéfi et al. 2012).

2.4 Galler–Aphid Interaction

In addition to parasitoids, galls of B. dracunculifoliae can be occupied by many inquiline arthropods (Collevatti and Sperber 1997; Espírito-Santo and Fernandes 2002; Barbosa et al. 2019). These inquilines use the gall as shelter and sometimes as a food source, consuming silk made by the galler or other plant products. Some inquilines can enter the pre-emergence galls through the longitudinal opening, while others enter the gall through the characteristic exit hole left in the gall wall when parasitoids emerge (Espírito-Santo et al. 2004). Thus, the colonization of parasitized galls by inquilines can be facilitated by parasitoids (MB unpublished data). Post-emergence galls remain attached to the plant, sometimes for a few generations, and gradually become dry and woody (Lara and Fernandes 1994; Espírito-Santo and Fernandes 1998) and are still colonized by many species.

The aphid Uroleucon tucumani (Sternorrhyncha: Aphididae) is the main inquiline species found in galls of B. dracunculifoliae (Collevatti and Sperber 1997; Fagundes et al. 2005) (Interaction 8). This aphid species also feeds and reproduces on the apical meristems of the host plant, forming dense colonies that produce honeydew (sugary secretions). Fagundes et al. (2005) showed that aphids can have a negative indirect effect on gall development: in their absence, galls were heavier. Gall size may be related to the performance of the galling insect (Weis 1988). Therefore, this suggests exploitation competition between the galler B. dracunculifoliae and the aphid U. tucumani for sap assimilates and young leaves from terminal buds (Fagundes et al. 2005) (Interaction 9). Inquilinism by the aphid U. tucumani is also common and can be indirectly responsible for the death of the nymphs of the galler B. dracunculifoliae (Espírito-Santo and Fernandes 2002) (Interaction 10).

Theory suggests that indirect effects may be an important mechanism for community stability and persistence, but empirical data are scarce (Strauss 1991; Kéfi et al. 2012). Barbosa et al. (2019) examined the propagation of indirect effects in the B. dracunculifolia system. The authors investigated whether the indirect effects initiated by ecosystem engineering – physical changes in the biotic or abiotic environment caused by a species – can feedback to the engineer, changing the magnitude and direction of its interactions with other species. Gallers can be seen as ecosystem engineers since galls serve as habitat for other species, particularly aphids, which occupy hatches and unhatched galls of the most common galler on B. dracunculifolia. Barbosa et al. (2017) raised the hypothesis that the hatched galls could generate feedbacks on the galler – e.g. increasing the availability of shelter for aphids and, therefore, positively affecting the galler by decreasing the occupation of live galls by aphids. In a field experiment, they generated treatments with reduced or elevated ecosystem engineering, removing or adding hatched galls. Inquilism by aphids negatively affected parasitism rates (interaction modification) likely by killing parasitized galling larvae (Interaction 11) since they preferentially colonize parasitized galls (MB unpublished data). Post-emergence galls changed the interaction between aphid inquilines and the galler (positively or negatively, depending on gall density), probably by providing extra shelter for aphids (Interaction 12) and also altered (positively or negatively, depending on gall density) the negative effect that aphids had on parasitism (modification of an interaction modification) (Interaction 13). The results show that hatched galls of the dominant galler can trigger indirect interactions that feedback to the galler, modifying its interactions with parasitoids and inquiline aphids. In addition, the results suggest that these interaction modifications are dependent on the context, which changes with species densities.

2.5 Ant–Aphid Interaction

At least 15 species of ants have been recorded foraging on B. dracunculifolia, many of them tend and protect the aphid U. tucumani in a trophobiotic relationship (Fagundes et al. 2005; Neves et al. 2011; MB unpublished data) (Interaction 14). Fagundes et al. (2005) experimentally excluded ants, aphids or both from shoots of B. dracunculifolia. Ants had a direct negative effect on the performance of B. dracunculifoliae. When ants were present, a smaller number of nymphs were found in each gall of B. dracunculifoliae, most likely because ants interfered with the female galler during oviposition (Fagundes et al. 2005) (Interaction 15). Neves et al. (2011) observed that the presence of ants and U. tucumani aphids on B. dracunculifolia also decreased the abundance of other free-feeding herbivores and that the presence of aphids decreased the growth of plant shoots (Interactions 16, 17, and 18).

Monteiro et al. (2020a) studying the community of insect herbivores associated with B. dracunculifolia also found that ant abundance was positively related to the abundance of trophobionts and negatively associated with the richness of chewing herbivores. Ants protecting aphids can directly affect herbivores negatively, preying or interfering with them (Abe 1988; Fernandes et al. 1999). However, aphids on their own can reduce the abundance of fluid-sucking and chewing insects due to exploitation competition or by altering the nutritional quality of the host plant (Fay et al. 1996; Larson and Whitham 1997). Thus, aphids can potentially reduce herbivory and indirectly benefit the host plant. Also, as chewing insects can attack the gall and kill the galling nymph, aphids can also positively affect the galling species B. dracunculifolia (Espírito-Santo and Fernandes 2002).

2.6 Mistletoe–Plant Interaction

The hemiparasitic Struthanthus flexicaulis frequently colonizes B. dracunculifolia (Monteiro et al. 2020b; Fig. 5.2). S. flexicaulis represents an important stress factor for the host plant. Over time S. flexicaulis can cause significant reduction in growth and fitness or irreversible and sublethal damage to the host plants, increasing mortality (Press and Phoenix 2005; Cameron et al. 2008; Bahia et al. 2015; Mourão et al. 2016; Monteiro et al. 2020b) (Interaction 19).

Bahia et al. (2015) showed that S. flexicaulis can reduce the number of leaves of B. dracunculifolia and cause the death of occupied branches or even of the entire plant. The water imbalance caused by this hemiparasite may be a factor responsible for affecting growth and increased mortality of parasitized host plants (Bahia et al. 2015; Mourão et al. 2016; Monteiro et al. 2020b). In addition, when the plant faces water restriction, there is a tendency to decrease leaf size to prevent water loss through transpiration (Lincoln and Zeiger 2013). The effect of lower water availability on the vegetative development of host plants can also decrease the specific leaf area (SLA) of the host or even lead to increased senescence of the leaves (e.g. Figueirôa et al. 2004; Lincoln and Zeiger 2013). This water imbalance can affect the plant’s ability to store water in the leaves and can also compromise leaf succulence (Cruz et al. 2018) due to the lower water supply for the hydration of the leaves (e.g. Scatena and Scremin-Dias 2003, Cruz et al. 2018).

A study by Monteiro et al. (unpublished) found that the rate of fluorescence in B. dracunculifolia is affected by the presence of S. flexicaulis. The photosynthetic rate of B. dracunculifolia was reduced in the most critical period of the day (12:00 noon to 3:00 pm); however, the host plant was unable to recover itself in the early evening. The percentage of chlorophyll was not affected by the presence of the hemiparasite on individuals of the host B. dracunculifolia. The percentage of nitrogen balance (NBI) was higher in non-parasitized individuals than in parasitized individuals. The percentage of flavonoids was higher in parasitized individuals than in non-parasitized individuals. Thus, parasitism affected the N/flavonoid balance of the host plant, representing a source of damage. The S. flexicaulis mistletoe has a much higher rate of transpiration than the host B. dracunculifolia. In addition, the mistletoes have the homeostatic control of their hosts and intensify the withdrawal of water throughout the day (Glatzel and Geils 2009). Plants parasitized by the mistletoe have a higher water potential than non-parasitized plants as a way to guarantee their hydration, since the mistletoe removes a large volume of water at noon.

The effects of parasitism on physiology also affect the architecture of the host B. dracunculifolia. Monteiro et al. (2020b) showed that individuals parasitized by the hemiparasite presented lower crown growth and height. These effects on the host are related to the fact that the hemiparasite obtains water and nutrients directly from the xylem, reducing the resources available for the host’s own metabolism (Press and Phoenix 2005; Westwood et al. 2010; Guerra and Pizo 2014), which can affect their fitness and even lead to their death (Mallams and Mathiasen 2010; Mourão et al. 2016; Monteiro et al. 2020b). In fact, Monteiro et al. (2020b) found a high mortality of parasitized B. dracunculifolia individuals, 61%, after 2 years. The high mortality of abundant species such as the host B. dracunculifolia can lead to a restructuring of the community of woody plants in the environment in which they occur in high frequency, increasing plant diversity.

In addition, the presence of the hemiparasite S. flexicaulis can affect the host in indirect ways. The presence of the parasitic plant alters the interactions of B. dracunculifolia with its associated insect community. Bahia et al. (2015) found that branches occupied by mistletoe showed a greater abundance of galls induced by B. dracunculifoliae (Interaction 20). Although parasitic plants are known to cause deleterious effects on plant communities, a broader view of these relationship reveals several positive direct and indirect effects on the community (e.g. Watson 2009; Watson et al. 2011; Mellado et al. 2016; Ndagurwa et al. 2016; Mellado and Zamora 2017; Hódar et al. 2018). Mistletoes can be an important source of nectar and fleshy fruits, arthropods, nesting sites and perches for birds (Guerra and Pizo 2014; Watson et al. 2011). The abundance of ants and trophobionts can be increased since insects can benefit from the large flow of water and nutrients that come from parasites and, therefore, become more abundant (Guerra et al. 2011; Freitas and Rossi 2015). Mistletoes can also have positive impacts on soil microbial communities, increasing the diversity, quality and quantity of organic matter that enters the soil next to the host (Mellado et al. 2016). This mechanism also promotes an increase in diversity and in leaf cover of herbaceous vegetation next to the host tree (Watson 2009; Ndagurwa et al. 2016), which can lead to the presence of vertebrates in the area (Hódar et al. 2018).

3 Concluding Remarks

The B. dracunculifolia system is well studied and comprehended, as described here. The focal community represents a very appropriate model system for the study of indirect interactions under natural field conditions. The aggregated distribution of the host plant combined with a diverse and fairly specialized fauna of arthropods creates a discrete and highly self-contained multi-trophic community. These characteristics not only facilitate the manipulation and monitoring of species densities but also increase the chances of observing their effects, since they are more likely to be a result of local ecological processes, rather than being entangled with external processes (e.g. compensatory migration). In addition, contrary to experimental manipulations in laboratory conditions, where sets of interacting species are studied in isolation, in this system the propagation of systemic indirect effects is possible. We hope that this work inspires and facilitates the design of further empirical investigations on the role of indirect interactions in community structure and dynamics.