种间互作网络的结构、生态系统功能及稳定性机制研究

生态群落中不同物种间发生多样化的相互作用, 形成了复杂的种间互作网络。复杂生态网络的结构如何影响群落的生态系统功能及稳定性是群落生态学的核心问题之一。种间互作直接影响到物质和能量在生态系统不同组分之间的流动和循环以及群落构建过程, 使得网络结构与生态系统功能和群落稳定性密切相关。在群落及生态系统水平上开展种间互作网络研究将为群落的构建机制、生物多样性维持、生态系统稳定性、物种协同进化和性状分化等领域提供新的视野。当前生物多样性及生态系统功能受到全球变化的极大影响, 研究种间互作网络的拓扑结构、构建机制、稳定性和生态功能也可为生物多样性的保护和管理提供依据。该文从网络结构、构建机制、网络结构和稳定性关系、种间互作对生态系统功能的影响等4个方面综述当前种间网络研究进展, 并提出在今后的研究中利用机器学习和多层网络等来探究环境变化对种间互作网络结构和功能的影响, 并实现理论和实证研究的有效整合。     

Analysing ecological networks of species interactions

Network approaches to ecological questions have been increasingly used, particularly in recent decades. The abstraction of ecological systems – such as communities – through networks of interactions between their components indeed provides a way to summarize this information with single objects. The methodological framework derived from graph theory also provides numerous approaches and measures to analyze these objects and can offer new perspectives on established ecological theories as well as tools to address new challenges. However, prior to using these methods to test ecological hypotheses, it is necessary that we understand, adapt, and use them in ways that both allow us to deliver their full potential and account for their limitations. Here, we attempt to increase the accessibility of network approaches by providing a review of the tools that have been developed so far, with – what we believe to be – their appropriate uses and potential limitations. This is not an exhaustive review of all methods and metrics, but rather, an overview of tools that are robust, informative, and ecologically sound. After providing a brief presentation of species interaction networks and how to build them in order to summarize ecological information of different types, we then classify methods and metrics by the types of ecological questions that they can be used to answer from global to local scales, including methods for hypothesis testing and future perspectives. Specifically, we show how the organization of species interactions in a community yields different network structures (e.g., more or less dense, modular or nested), how different measures can be used to describe and quantify these emerging structures, and how to compare communities based on these differences in structures. Within networks, we illustrate metrics that can be used to describe and compare the functional and dynamic roles of species based on their position in the network and the organization of their interactions as well as associated new methods to test the significance of these results. Lastly, we describe potential fruitful avenues for new methodological developments to address novel ecological questions.     

Ecological integration of alien species into a tree-parasitic fungus network

The diversification of species and their interactions during the course of evolution has produced ecological networks with a complex topology. This topology influences the current functioning of ecosystems. It is therefore important to investigate whether the species introduced recently by human activities have merged seamlessly into recipient ecological networks by developing interactions quantitatively and qualitatively similar to those of native species, or whether their establishment has altered the topology of the networks. We tackled this issue in the case of a well resolved interaction network between 51 forest tree taxa and 154 pathogenic fungal species. We found that alien and native species with similar phylogenetic histories and life-history strategies had similar types and numbers of interactions. Our results also suggest that the clustered architecture of the network has not been altered by the integration of alien species. It therefore seems that a few centuries have been sufficient for the network to assimilate the newly introduced species. This rapid integration was unexpected for a plant-pathogen network, because selection acts continually on plants, favouring the emergence of defences against new pathogens and impeding the development of new interactions. However, it was recently shown that perturbation of the structure of ecological networks might be overlooked if species interactions are not quantified. The tree-parasitic fungus network considered in this study is binary. We might therefore end up with different results by using quantitative data.     

Comparing species interaction networks along environmental gradients

Knowledge of species composition and their interactions, in the form of interaction networks, is required to understand processes shaping their distribution over time and space. As such, comparing ecological networks along environmental gradients represents a promising new research avenue to understand the organization of life. Variation in the position and intensity of links within networks along environmental gradients may be driven by turnover in species composition, by variation in species abundances and by abiotic influences on species interactions. While investigating changes in species composition has a long tradition, so far only a limited number of studies have examined changes in species interactions between networks, often with differing approaches. Here, we review studies investigating variation in network structures along environmental gradients, highlighting how methodological decisions about standardization can influence their conclusions. Due to their complexity, variation among ecological networks is frequently studied using properties that summarize the distribution or topology of interactions such as number of links, connectance, or modularity. These properties can either be compared directly or using a procedure of standardization. While measures of network structure can be directly related to changes along environmental gradients, standardization is frequently used to facilitate interpretation of variation in network properties by controlling for some co‐variables, or via null models. Null models allow comparing the deviation of empirical networks from random expectations and are expected to provide a more mechanistic understanding of the factors shaping ecological networks when they are coupled with functional traits. As an illustration, we compare approaches to quantify the role of trait matching in driving the structure of plant–hummingbird mutualistic networks, i.e. a direct comparison, standardized by null models and hypothesis‐based metaweb. Overall, our analysis warns against a comparison of studies that rely on distinct forms of standardization, as they are likely to highlight different signals. Fostering a better understanding of the analytical tools available and the signal they detect will help produce deeper insights into how and why ecological networks vary along environmental gradients.     

Bayesian networks as a novel tool to enhance interpretability and predictive power of ecological models

In today's world, it is becoming increasingly important to have the tools to understand, and ultimately to predict, the response of ecosystems to disturbance. However, understanding such dynamics is not simple. Ecosystems are a complex network of species interactions, and therefore any change to a population of one species will have some degree of community level effect. In recent years, the use of Bayesian networks (BNs) has seen successful applications in molecular biology and ecology, where they were able to recover plausible links in the respective systems they were applied to. The recovered network also comes with a quantifiable metric of interaction strength between variables. While the latter is an invaluable piece of information in ecology, an unexplored application of BNs would be using them as a novel variable selection tool in the training of predictive models. To this end, we evaluate the potential usefulness of BNs in two aspects: (1) we apply BN inference on species abundance data from a rocky shore ecosystem, a system with well documented links, to test the ecological validity of the revealed network; and (2) we evaluate BNs as a novel variable selection method to guide the training of an artificial neural network (ANN). Here, we demonstrate that not only was this approach able to recover meaningful species interactions networks from ecological data, but it also served as a meaningful tool to inform the training of predictive models, where there was an improvement in predictive performance in models with BN variable selection. Combining these results, we demonstrate the potential of this novel application of BNs in enhancing the interpretability and predictive power of ecological models; this has general applicability beyond the studied system, to ecosystems where existing relationships between species and other functional components are unknown.     

Ant-Plant Interaction in a Tropical Savanna: May the Network Structure Vary over Time and Influence on the Outcomes of Associations?

Plant-animal interactions occur in a community context of dynamic and complex ecological interactive networks. The understanding of who interacts with whom is a basic information, but the outcomes of interactions among associates are fundamental to draw valid conclusions about the functional structure of the network. Ecological networks studies in general gave little importance to know the true outcomes of interactions and how they may change over time. We evaluate the dynamic of an interaction network between ants and plants with extrafloral nectaries, by verifying the temporal variation in structure and outcomes of mutualism for the plant community (leaf herbivory). To reach this goal, we used two tools: bipartite network analysis and experimental manipulation. The networks exhibited the same general pattern as other mutualistic networks: nestedness, asymmetry and low specialization and this pattern was maintained over time, but with internal changes (species degree, connectance and ant abundance). These changes influenced the protection effectiveness of plants by ants, which varied over time. Our study shows that interaction networks between ants and plants are dynamic over time, and that these alterations affect the outcomes of mutualisms. In addition, our study proposes that the set of single systems that shape ecological networks can be manipulated for a greater understanding of the entire system.     

Circumventing structural uncertainty: A Bayesian perspective on nonlinear forecasting for ecology

As a consequence of the complexity of ecosystems and context-dependence of species interactions, structural uncertainty is pervasive in ecological modeling. This is particularly problematic when ecological models are used to make conservation and management plans whose outcomes may depend strongly on model formulation. Nonlinear time series approaches allow us to circumvent this issue by using the observed dynamics of the system to guide policy development. However, these methods typically require long time series from stationary systems, which are rarely available in ecological settings. Here we present a Bayesian approach to nonlinear forecasting based on Gaussian processes that readily integrates information from several short time series and allows for nonstationary dynamics. We demonstrate the utility of our modeling methods on simulated from a wide range of ecological scenarios. We expect that these models will extend the range of ecological systems to which nonlinear forecasting methods can be usefully applied.     

Rapid development of indirect effects in ecological networks

Indirect effects are important components of ecological and evolutionary interactions that may maintain biodiversity, enable or inhibit invasive species, and challenge ecosystem assessment and management. A central hypothesis of Network Environ Analysis (NEA), one type of ecological network analysis, is that indirect flows tend to dominate direct flows in ecosystem networks of conservative substance exchanges. However, current NEA methods assume that these ecosystems are stationary (i.e. time invariant exchange rates), which is unlikely to be true for many ecosystems for interesting time and space scales. For the work reported here, we investigated the sensitivity of the dominance of indirect effects hypothesis to the stationary modeling assumption by determining the development rate of indirect effects and flow intensity, as expressed as the number of transfer steps, in thirty‐one ecosystem models. We hypothesized that indirect effects develop rapidly in ecological networks, but that they would develop faster in biogeochemically based models than in trophically based models. In contrast, our results show that indirect effects develop rapidly in all thirty‐one models examined. In 94% of the models, indirect flows exceeded direct flows by a pathway length of 3. This indicates that ecological systems do not need to maintain a particular configuration for long for indirect effects to dominate. Thus, the dominance of indirect effects hypothesis remains plausible. We also found that biogeochemical models tended to require more of the extended path network than the trophic models to account for 50% and 95% of the total system activity, but that both types of models required more of the power series than is typically considered in engineered systems. These results succinctly illustrate the complexity of ecological systems and help explain why they are challenging to assess and manage.     

结构稳定性: 概念、方法和应用

群落内物种间相互作用的结构是高度组织化的。群落结构对多物种共存的影响机制是群落生态学的核心科学问题之一。目前生态学界在这一问题上存在多种不同的观点。一个可能的原因是, 由于环境因子的复杂性, 大部分研究忽略了环境因子对群落结构和物种共存的重要影响。在这一背景下, 近期发展起来的结构稳定性理论系统地联系了群落结构、环境因子和物种共存, 并在此基础上建立了一个和经验数据紧密结合的理论框架。本文首先简要回顾了当前关于群落结构研究的争鸣, 然后介绍了结构稳定性的理论框架和计算方法, 最后详细介绍了结构稳定性理论在不同生态群落和不同生态学问题中的应用。在全球气候变化的背景下, 结构稳定性理论提供了一种新的视角来理解群落层面的生物多样性维持机制。     

Removing interactions,rather than species,casts doubt on the high robustness of pollination networks

In the last 15 years, a complex networks perspective has been increasingly used in the robustness assessment of ecological systems. It is therefore crucial to assess the reliability of such tools. Based on the traditional simulation of node (species) removal, mutualistic pollination networks are considered to be relatively robust because of their 1) truncated power‐law degree distribution, 2) redundancy in the number of pollinators per plant and 3) nested interaction pattern. However, species removal is only one of several possible approaches to network robustness assessment. Empirical evidence suggests a decline in abundance prior to the extinction of interacting species, arguing in favour of an interaction removal‐based approach (i.e. interaction disruption), as opposed to traditional species removal. For simulated networks, these two approaches yield radically different conclusions, but no tests are currently available for empirical mutualistic networks. This study compared this new robustness evaluation approach based on interaction extinction versus the traditional species removal approach for 12 alpine and subalpine pollination networks. In comparison with species removal, interaction removal produced higher robustness in the worst‐case extinction scenario but lower robustness in the best‐case extinction scenario. Our results indicate that: 1) these two approaches yield very different conclusions and 2) existing assessments of ecological network robustness could be overly optimistic, at least those based on a disturbance affecting species at random or beginning with the least connected species. Therefore, further empirical study of plant–pollinator interactions in disturbed ecosystems is imperative to understand how pollination networks are disassembled.     

The role of biotic interactions in shaping distributions and realised assemblages of species: implications for species distribution modelling

Predicting which species will occur together in the future, and where, remains one of the greatest challenges in ecology, and requires a sound understanding of how the abiotic and biotic environments interact with dispersal processes and history across scales. Biotic interactions and their dynamics influence species' relationships to climate, and this also has important implications for predicting future distributions of species. It is already well accepted that biotic interactions shape species' spatial distributions at local spatial extents, but the role of these interactions beyond local extents (e.g. 10 km² to global extents) are usually dismissed as unimportant. In this review we consolidate evidence for how biotic interactions shape species distributions beyond local extents and review methods for integrating biotic interactions into species distribution modelling tools. Drawing upon evidence from contemporary and palaeoecological studies of individual species ranges, functional groups, and species richness patterns, we show that biotic interactions have clearly left their mark on species distributions and realised assemblages of species across all spatial extents. We demonstrate this with examples from within and across trophic groups. A range of species distribution modelling tools is available to quantify species environmental relationships and predict species occurrence, such as: (i) integrating pairwise dependencies, (ii) using integrative predictors, and (iii) hybridising species distribution models (SDMs) with dynamic models. These methods have typically only been applied to interacting pairs of species at a single time, require a priori ecological knowledge about which species interact, and due to data paucity must assume that biotic interactions are constant in space and time. To better inform the future development of these models across spatial scales, we call for accelerated collection of spatially and temporally explicit species data. Ideally, these data should be sampled to reflect variation in the underlying environment across large spatial extents, and at fine spatial resolution. Simplified ecosystems where there are relatively few interacting species and sometimes a wealth of existing ecosystem monitoring data (e.g. arctic, alpine or island habitats) offer settings where the development of modelling tools that account for biotic interactions may be less difficult than elsewhere.     

The nested assembly of plant facilitation networks prevents species extinctions

Facilitation is a positive interaction assembling ecological communities and preserving global biodiversity. Although communities acquire emerging properties when many species interact, most of our knowledge about facilitation is based on studies between pairs of species. To understand how plant facilitation preserves biodiversity in complex ecological communities, we propose to move from the study of pairwise interactions to the network approach. We show that facilitation networks behave as mutualistic networks do, characterized by a nonrandom, nested structure of plant-plant interactions in which a few generalist nurses facilitate a large number of species while the rest of the nurses facilitate only a subset of them. Consequently, generalist nurses shape a dense and highly connected network. Interestingly, such generalist nurses are the most abundant species in the community, making facilitation-shaped communities strongly resistant to extinction, as revealed by coextinction simulations. The nested structure of facilitative networks explains why facilitation, by preventing extinction, preserves biodiversity.     

Consequences of plant invasions on compartmentalization and species’ roles in plant–pollinator networks

Compartmentalization—the organization of ecological interaction networks into subsets of species that do not interact with other subsets (true compartments) or interact more frequently among themselves than with other species (modules)—has been identified as a key property for the functioning, stability and evolution of ecological communities. Invasions by entomophilous invasive plants may profoundly alter the way interaction networks are compartmentalized. We analysed a comprehensive dataset of 40 paired plant–pollinator networks (invaded versus uninvaded) to test this hypothesis. We show that invasive plants have higher generalization levels with respect to their pollinators than natives. The consequences for network topology are that—rather than displacing native species from the network—plant invaders attracting pollinators into invaded modules tend to play new important topological roles (i.e. network hubs, module hubs and connectors) and cause role shifts in native species, creating larger modules that are more connected among each other. While the number of true compartments was lower in invaded compared with uninvaded networks, the effect of invasion on modularity was contingent on the study system. Interestingly, the generalization level of the invasive plants partially explains this pattern, with more generalized invaders contributing to a lower modularity. Our findings indicate that the altered interaction structure of invaded networks makes them more robust against simulated random secondary species extinctions, but more vulnerable when the typically highly connected invasive plants go extinct first. The consequences and pathways by which biological invasions alter the interaction structure of plant–pollinator communities highlighted in this study may have important dynamical and functional implications, for example, by influencing multi-species reciprocal selection regimes and coevolutionary processes.     

A theory for species co-occurrence in interaction networks

The study of species co-occurrences has been central in community ecology since the foundation of the discipline. Co-occurrence data are, nevertheless, a neglected source of information to model species distributions and biogeographers are still debating about the impact of biotic interactions on species distributions across geographical scales. We argue that a theory of species co-occurrence in ecological networks is needed to better inform interpretation of co-occurrence data, to formulate hypotheses for different community assembly mechanisms, and to extend the analysis of species distributions currently focused on the relationship between occurrences and abiotic factors. The main objective of this paper is to provide the first building blocks of a general theory for species co-occurrences. We formalize the problem with definitions of the different probabilities that are studied in the context of co-occurrence analyses. We analyze three species interactions modules and conduct multi-species simulations in order to document five principles influencing the associations between species within an ecological network: (i) direct interactions impact pairwise co-occurrence, (ii) indirect interactions impact pairwise co-occurrence, (iii) pairwise co-occurrence rarely are symmetric, (iv) the strength of an association decreases with the length of the shortest path between two species, and (v) the strength of an association decreases with the number of interactions a species is experiencing. Our analyses reveal the difficulty of the interpretation of species interactions from co-occurrence data. We discuss whether the inference of the structure of interaction networks is feasible from co-occurrence data. We also argue that species distributions models could benefit from incorporating conditional probabilities of interactions within the models as an attempt to take into account the contribution of biotic interactions to shaping individual distributions of species.     

Motifs in bipartite ecological networks: uncovering indirect interactions

Indirect interactions play an essential role in governing population, community and coevolutionary dynamics across a diverse range of ecological communities. Such communities are widely represented as bipartite networks: graphs depicting interactions between two groups of species, such as plants and pollinators or hosts and parasites. For over thirty years, studies have used indices, such as connectance and species degree, to characterise the structure of these networks and the roles of their constituent species. However, compressing a complex network into a single metric necessarily discards large amounts of information about indirect interactions. Given the large literature demonstrating the importance and ubiquity of indirect effects, many studies of network structure are likely missing a substantial piece of the ecological puzzle. Here we use the emerging concept of bipartite motifs to outline a new framework for bipartite networks that incorporates indirect interactions. While this framework is a significant departure from the current way of thinking about bipartite ecological networks, we show that this shift is supported by analyses of simulated and empirical data. We use simulations to show how consideration of indirect interactions can highlight differences missed by the current index paradigm that may be ecologically important. We extend this finding to empirical plant–pollinator communities, showing how two bee species, with similar direct interactions, differ in how specialised their competitors are. These examples underscore the need to not rely solely on network‐ and species‐level indices for characterising the structure of bipartite ecological networks.     

Learning Bayesian network classifiers using ant colony optimization

Bayesian networks are knowledge representation tools that model the (in)dependency relationships among variables for probabilistic reasoning. Classification with Bayesian networks aims to compute the class with the highest probability given a case. This special kind is referred to as Bayesian network classifiers. Since learning the Bayesian network structure from a dataset can be viewed as an optimization problem, heuristic search algorithms may be applied to build high-quality networks in medium- or large-scale problems, as exhaustive search is often feasible only for small problems. In this paper, we present our new algorithm, ABC-Miner, and propose several extensions to it. ABC-Miner uses ant colony optimization for learning the structure of Bayesian network classifiers. We report extended computational results comparing the performance of our algorithm with eight other classification algorithms, namely six variations of well-known Bayesian network classifiers, cAnt-Miner for discovering classification rules and a support vector machine algorithm.     

Inferring species interactions from ecological survey data: A mechanistic approach to predict quantitative food webs of seed feeding by carabid beetles

1. Ecological networks are valuable for ecosystem analysis but their use is often limited by a lack of data because many types of ecological interaction, for example, predation, are short‐lived and difficult to observe or detect. While there are different methods for inferring the presence of interactions, they have rarely been used to predict the interaction strengths that are required to construct weighted, or quantitative, ecological networks.
2. Here, we develop a trait‐based approach suitable for inferring weighted networks, that is, with varying interaction strengths. We developed the method for seed‐feeding carabid ground beetles (Coleoptera: Carabidae) although the principles can be applied to other species and types of interaction.
3. Using existing literature data from experimental seed‐feeding trials, we predicted a per‐individual interaction cost index based on carabid and seed size. This was scaled up to the population level to create inferred weighted networks using the abundance of carabids and seeds from empirical samples and energetic intake rates of carabids from the literature. From these weighted networks, we also derived a novel measure of expected predation pressure per seed type per network.
4. This method was applied to existing ecological survey data from 255 arable fields with carabid data from pitfall traps and plant seeds from seed rain traps. Analysis of these inferred networks led to testable hypotheses about how network structure and predation pressure varied among fields.
5. Inferred networks are valuable because (a) they provide null models for the structuring of food webs to test against empirical species interaction data, for example, DNA analysis of carabid gut regurgitates and (b) they allow weighted networks to be constructed whenever we can estimate interactions between species and have ecological census data available. This permits ecological network analysis even at times and in places when interactions were not directly assessed.

     

Defining invasiveness and invasibility in ecological networks

The success of a biological invasion is context dependent, and yet two key concepts—the invasiveness of species and the invasibility of recipient ecosystems—are often defined and considered separately. We propose a framework that can elucidate the complex relationship between invasibility and invasiveness. It is based on trait-mediated interactions between species and depicts the response of an ecological network to the intrusion of an alien species, drawing on the concept of community saturation. Here, invasiveness of an introduced species with a particular trait is measured by its per capita population growth rate when the initial propagule pressure of the introduced species is very low. The invasibility of the recipient habitat or ecosystem is dependent on the structure of the resident ecological network and is defined as the total width of an opportunity niche in the trait space susceptible to invasion. Invasibility is thus a measure of network instability. We also correlate invasibility with the asymptotic stability of resident ecological network, measured by the leading eigenvalue of the interaction matrix that depicts trait-based interaction intensity multiplied by encounter rate (a pairwise product of propagule pressure of all members in a community). We further examine the relationship between invasibility and network architecture, including network connectance, nestedness and modularity. We exemplify this framework with a trait-based assembly model under perturbations in ways to emulate fluctuating resources and random trait composition in ecological networks. The maximum invasiveness of a potential invader (greatest intrinsic population growth rate) was found to be positively correlated with invasibility of the recipient ecological network. Additionally, ecosystems with high network modularity and high ecological stability tend to exhibit high invasibility. Where quantitative data are lacking we propose using a qualitative interaction matrix of the ecological network perceived by a potential invader so that the structural network stability and invasibility can be estimated from the literature or from expert opinion. This approach links network structure, invasiveness and invasibility in the context of trait-mediated interactions, such as the invasion of insects into mutualistic and antagonistic networks.     

传粉网络的研究进展:网络的结构和动态

植物与传粉者之间相互作用,构成了复杂的传粉网络。近年来,社会网络分析技术的发展使得复杂生态网络的研究成为可能。从群落水平上研究植物与传粉者之间的互惠关系,为理解群落的结构和动态以及花部特征的演化提供了全新的视角。传粉网络的嵌套结构说明自然界的传粉服务存在冗余,而且是相对泛化的物种主导了传粉。在多年或者多季度的传粉网络中,虽然有很高的物种替换率,但是其网络结构仍然保持相对稳定,说明传粉网络对干扰有很强的抗性。尽管有关网络结构和动态的研究逐渐增多,但传粉网络维持的机制仍不清楚。网络结构可以部分由花部特征与传粉者的匹配来解释,也受到系统发生的制约,影响因素还包括群落构建的时间和物种多样性,以及物种在群落中的位置。开展大尺度群落动态的研究,为探索不同时间尺度、不同物种多样性水平上的传粉网络的生态学意义提供了条件。但已有的研究仍存在不足,比如基于访问观察的网络无法准确衡量传粉者的访问效率和植物间的花粉流动,以及结果受到调查精度区域研究不平衡的制约等。目前的研究只深入到传粉者携带花粉构成成分的水平,传粉者访问植物的网络不能代表植物的整个传粉过程。因此,研究应当更多地深入到物种之间关系对有性生殖的切实影响上。     

Sensitivity and specificity of inferring genetic regulatory interactions from microarray experiments with dynamic Bayesian networks

MOTIVATION: Bayesian networks have been applied to infer genetic regulatory interactions from microarray gene expression data. This inference problem is particularly hard in that interactions between hundreds of genes have to be learned from very small data sets, typically containing only a few dozen time points during a cell cycle. Most previous studies have assessed the inference results on real gene expression data by comparing predicted genetic regulatory interactions with those known from the biological literature. This approach is controversial due to the absence of known gold standards, which renders the estimation of the sensitivity and specificity, that is, the true and (complementary) false detection rate, unreliable and difficult. The objective of the present study is to test the viability of the Bayesian network paradigm in a realistic simulation study. First, gene expression data are simulated from a realistic biological network involving DNAs, mRNAs, inactive protein monomers and active protein dimers. Then, interaction networks are inferred from these data in a reverse engineering approach, using Bayesian networks and Bayesian learning with Markov chain Monte Carlo. RESULTS: The simulation results are presented as receiver operator characteristics curves. This allows estimating the proportion of spurious gene interactions incurred for a specified target proportion of recovered true interactions. The findings demonstrate how the network inference performance varies with the training set size, the degree of inadequacy of prior assumptions, the experimental sampling strategy and the inclusion of further, sequence-based information. AVAILABILITY: The programs and data used in the present study are available from http://www.bioss.sari.ac.uk/~dirk/Supplements