Bidimensional transfer of effects among organisms: An overlooked concept in community ecology Transferencia bidimensional de efectos entre organismos: Un concepto desatendido en ecología de comunidades

Although geneticists and population ecologist have considered in their explanations the importance of both horizontal –interspecifi c– and vertical –intraspecifi c– paths for the fl ow of information, energy or matter among organisms, community ecology and particularly network ecology have been strongly biased towards considering only horizontal effects. In this article our aim is to put on view of ecologists the concepts of vertical and horizontal transfer of disturbance effects in communities, and discuss the relevance of the interplay between both processes for ecological research. We support that incorporating this view into community ecology could be a promising new research avenue, and may improve our understanding about the responses of communities to natural and anthropogenic disturbances.


INTRODUCTION
The development of genetic engineering has allowed geneticists identifying that horizontal gene transfer (i.e.without involving reproduction) among neighbor microorganisms occurs indeed more frequently that ver tical (parental) transfer (McDaniel et al. 2010).Although in genetics the vertical intraspecifi c dimension of gene transfer has been the hegemonic paradigm, it is now recognized that horizontal transfer provides a rapid means of adaptation to environmental change and, unlike vertical transfer, often occurs among organisms of dif ferent species and indeed belonging to dif ferent kingdoms (Aoki & Syõno 1999, Kondo et al. 2002, Brown 2003).Therefore, in genetics it is widely recognized the importance of both vertical -intraspecifi c-and horizontal -interspecifi c-transfer as major means of gene propagation and source of evolutionary novelty.In ecology, conversely, the transfer of effects among organisms has been thought to be mainly a horizontal process, among organism that interact sharing space and time.
More recently, population ecologists have incorporated the ver tical transmission of effects (particularly maternal effects) and its interplay with horizontal transfer of ef fects (direct density dependence) in their theories for explaining the shape of population dynamics (Ginzburg & Taneyhill 1994).Never theless, we stress that this has been not the case in community ecology.Fur thermore, we guess that incorporating this view into community ecology could be a promising new research avenue, and may improve our understanding about the responses of communities to natural and anthropogenic disturbances.
Community ecology, and par ticularly its refreshed face of network ecology, emphasizes the str ucture of connections among the or ganisms or gr oups of or ganisms as determinant of the dynamics and robustness of communities to different kinds of disruption.In this view, form and function of ecological systems are tightly bound (see Gross & Blasius 2008) and the collective behavior of communities has less to do than previously thought with the identity and functioning of their isolated parts.Nevertheless, the recent vigorous development of network ecology has been strongly biased towards considering only horizontal effects.
Our aim in this communication is to put on view of ecologists the concepts of vertical and horizontal transfer of disturbance ef fects in communities.We briefl y discuss the relevance of the interplay between both processes for basic ecological research and its signifi cance for environmental studies.

EFFECTS
Following the advances of statistical causal inference (Pearl 2009, Ar unKumar andVenkatesan 2011), an ef fect can be defined as the capacity to transmit changes between variables.Thus, from an ecological point of view, an ef fect occurs whenever changes in an organism's trait as a response to an environmental stimulus, is transmitted to other organisms by mechanisms other than genetic inheritance.It is clear that, for such effect to occur from organism A to organism B, both A and B must be linked ecologically.More precisely, fi tness of B should depend on the state of a trait of A (e.g., size, physiological condition, etc.)

Horizontal effects
The study of interactions among organisms and populations has been a foundational issue in ecology.Pairwise interactions refer to a mutual infl uence between two entities, say A and B, and can be decomposed into two directional effects: A → B and B → A. These effects can be direct, i.e. with no intermediaries, or indirect in which one or more entities mediate the effect.Net effects account for all possible direct and indirect effects of one entity on another one.
Ecological horizontal effects (Fig. 1) denote the direction and magnitude in which the presence of organisms of one type alters the fi tness of the others.Assessing the sign and strength of effects is not an easy task (Wootton & Emmerson 2005), especially if indirect effects are considered and the populations are not assumed to be in steady state (Abrams 2001).Never theless, valuable advances have been made in linking the distribution of interactions into ecological networks to their dynamics (McCann et al. 1998, Bascompte et al. 2003, A disturbance exerted on species 1 can be transmitted via direct effects to species 2, 3 and 5; and via indirect effects to the rest of species in the community.Olesen et al. 2007, Thebault & Fontaine 2010).The importance of identifying the distribution and strength of interspecifi c interactions is that those properties are considered to determine the propagation of environmental disturbances over the community (Bascompte & Stouffer 2009, González et al. 2011).A number of theoretical tools have been used for assessing the propagation and outcomes of disturbances in ecological networks.For example, a static approach (i.e.without population dynamics) has been used for studying the responses of communities to the loss of species (Dunne et al. 2003), by means of dynamic qualitative analyses it has been possible to study the consequences of pollution (Ramos-Jiliberto et al. 2012a) and through numerical simulation of dynamical models we have studied the expected consequences of removing exotic (Valdovinos et al. 2009) or native (Ramos-Jiliberto et al. 2009) groups of species.In spite of the potential usefulness of these approaches, all of them disregard transgenerational effects of disturbances on demographic traits.We stress that adding this temporal (i.e.vertical) dimension to future models of interaction networks would represent a step for wards in the search of fundamental determinants of community dynamics and responses of ecosystems to environmental disturbances.

Vertical effects
Life histor y and demographic responses to environmental change can be transmitted "ver tically", across successive generations (Fig. 2).These responses to stimuli perceived by parents but obser ved in the of fspring are known as transgenerational ef fects.Maternal ef fects are a well-known kind of transgenerational effects, and they are defi ned as the causal infl uence of the parental genotype or phenotype on the offspring phenotype (Wolf & Wade 2009).These developmental changes incorporate phenotypic variation that cannot be justifi ed by genetic variation (Uller 2008).The mechanism often implied in this process is the environment (in the broadest sense) provided to offspring by their parents.Transgenerational ef fects driven by natural or anthropogenic disturbances have been reported for many traits in many taxa, e.g.egg size of fi sh (Einum & Flemming 1999), seed size, germination timing and success, leaf production, and early growth in plants (Schuler & Orrock 2012), life-history shifts in Daphnia (Car vajal-Salamanca et al. 2008), increased metabolic rate in Daphnia (Fernández-González et al. 2011), immune defenses to copper in blow fl y (Pölkki et al. 2012).Thus, vertical -transgenerational-effects lead to the transmission of changes in parents' traits driven by an environmental stimulus to changes in offspring's traits when the latter may not be yet subjected to the stimulus.This constitutes a source of delayed life-histor y and demographic ef fects (Beckerman et al. 2002).Vertical effects and specifi cally parental effects -either adaptive or not-have been show to influence the dynamics of populations, explaining oscillator y behavior better than other hypothesized alternative mechanisms (Kendall et al. 2005 promoted by maternal effects (Benton et al. 2001, 2005, Beckerman et al. 2002).Although there is evidence that maternal effects may indirectly affect growth rates by changing the competitive environment of offspring (Benton et al. 2005), this way of thinking that is well considered in genetics and population ecology has not been adequately transferred to the study of ecological interactions as determinants of the structure and dynamics of ecological communities.

, see also Inchausti & Ginzburg 2009). Population trajectories could show cycles due to density-dependent lags
THINKING BIDIMENSIONAL Given the occurrence of both horizontal and ver tical transfer, the likelihood of finding a bidimensional transfer of ecological ef fects arises as a logical consequence.This could occur, for example, since a horizontally transmitted change could be a result of a change that was promoted vertically on the emitter.i.e. the horizontal emitter could have been a vertical receptor (Fig. 3).Conversely, the emitter of a vertical effect could also be a previous receptor of a horizontal ef fect.To our knowledge, there is no empirical demonstration of this kind of complex effect in real communities, but we suspect that indeed there has not been a serious intent to fi nd it.
T h e e c o l o g i c a l s i g n i f i c a n c e o f this bidimensional transfer of ef fects, in the case that their occur rence in nature were demonstrated, is not minor.First, it represents an unexplored path of propagation of disturbances, and therefore it is intricately linked with the stability pr oper ties of communities.Second, it represents a source of complexity in ecological networks, whose potential impact on community dynamics should be elucidated by assessing patterns of occurrence, distribution over the network structure, characteristic diameter of infl uence and timescales in which they act.Third, this kind of ef fects could challenge our ability to predict community responses to natural or anthr opogenic disturbances.Last, bidimensional ef fects should explain delayed and far-acting ef fects of pollutants and other environmental disturbances on complex communities.Overall, we guess that bidimensional ef fect ar e likely to occur in nature, and its discover y should promote opening a new research agenda with Fig. 3: Representation of horizontal and vertical transfer of effects within an ecological community.Nodes are populations linked by ecological interactions (among nodes with different labels) and parental relationships (among nodes with the same label).The populations and interactions that do not participate in the route of effects of this example are shown in light tones.Population1 is being affected by an environmental disturbance, but its effects are transmitted vertically and they are only expressed in the next generation.Next, the changes in demographic traits of population 1 are transmitted horizontally to population 2, which also transmit the effect vertically to the next generation.At the fi nal time frame, the effect received vertically by 2 is transferred horizontally to population 3.
interesting implications for both basic and applied ecological research.
Some interesting patterns are expected to emerge using this approach.Bidimensional transfer could extend the timescale in which indirect ecological ef fects take place.By adding delays in the interspecifi c interactions, the vertical dimension may alter the stability proper ties of the networks and thus their patter ns of responses to environmental perturbations.In addition, adding the vertical dimension to ecological dynamics translates to short-time inheritance of trait values, which could lead to loss of stability.In sum, we expect the relationship between (horizontal) network structure and network dynamics being modulated by the vertical structure.
Nevertheless, things are not so simple.Both taxonomic and temporal resolution of networks is highly relevant for detecting bidimensional ef fects.The taxonomic aggregation in the building of ecological networks, although unavoidable to a cer tain degree, has been recognized to alter the (horizontal) structure and therefore the prediction of the functioning of networks (Abarca-Arenas & Ulanowicz 2002, Jordan 2003, Thompson et al. 2012).This should be also tr ue for the ver tical structure, since transgenerational links in a functional group could belong to a variable and undefi ned number of species within the group.On the other hand, temporal resolution of horizontal effects has been addressed since years (Menge 1997), but it is crucial to consider them in the context of bidimensional effects.Too broad temporal grain will collapse many generations of some organisms, thus yielding a picture not much different from the usual static (horizontal) approach.In turn, too high temporal resolution could loss the signal of the longer-lived species.Thus, the network resolution in terms of topology, time scale and spatial scale of research should be carefully chosen according to the biology of the species at hand.M o r e o v e r, i d e n t i c a l e n v i r o n m e n t a l per turbations applied to populations with different age or size structures will lead to different population responses, since we could expect that the ef fects will be propagated dif ferently both ver tically and horizontally.Interspecific interactions are not fixed but vary over time.On the one hand, interaction strengths are dynamic (Abrams 2001).This implies that the magnitude and scope of bidimensional ef fects should also be highly variable.On the other hand, some interactions among species appear and disappear through time in response to environment stimuli, typically the trophic environment of the organisms (Beckerman et al. 2006, Petchey et al. 2008).The network consequences of this fl exibility or plasticity of interactions are under study presently (Kaiser-Bunbur y et al. 2010, Thierr y et al. 2011, Ramos-Jiliberto et al. 2012b), but considering that such plastic interactions could participate in bidimensional paths of ef fects adds a level of complexity that makes the study of the str ucture/ dynamics interplay in ecological networks more challenging and stimulating.

Fig. 1 :
Fig. 1: Representation of horizontal effects.Nodes represent populations of different species related by interactions composed of positive (links ending in arrowheads) and negative (links ending in circles) effects.A disturbance exerted on species 1 can be transmitted via direct effects to species 2, 3 and 5; and via indirect effects to the rest of species in the community.

Fig. 2 :
Fig.2: Representation of ver tical ef fects.Nodes represent the same population over successive generations (1 to 3).Organisms belonging to generation 1 receive the impact of an environmental disturbance, which affect organisms of the next generation, which also transmit the effect to generation 3. Links ending in circles represent negative effects, as an example.