Lab 7: Ecological Interactions;
What are the current categories methods used to extract ecological information to know their ecological interactions?
“The first law of ecology is that everything is connected to everything else” is a quote by Barry Commoner. According to Branchet (2020), “Co-occurrence analysis is the study of interactions between species distributions, and as such, it has been at the centre of community ecology for more than 100 years” (p.1). Different species influence each other, and their population affects their demography. Branchet (2020) also stated that recent co-occurrence analysis gained attention from different fields such as ecology and microbiology and systems like gut microbiome and boreal forests due to the accumulation of observational data and the arrival of new statistical methods. “There are several theoretical and statistical reasons explaining why there is only a weak relationship between co-occurrence and interactions” (Branchet, 2020, p.1). Based on presence-absent data, ecologists have proposed a statistical method to determine species relationships (Branchet, 2020). “As early as 1907, Forbes proposed a systematic analysis of pairwise co-occurrences using the ratio between the number of observed and expected co-occurrences to determine the degree of association among pairs of fishes” (Branchet, 2020, p.1). Thirteen years later, a more advanced plea was conducted by Michael which showed that there were drawbacks in Forbes’s coefficient such as “the importance of the spatial scale of sampling unit to draw meaningful conclusions about the underlying ecological relationships inferred from it” (Branchet, 2020, p.1). Similar approaches to the Forbes’ approach were developed independently and were grounded on a similar rationale. Pielou developed statistical methods “to discriminate mechanisms of co-existence among Diptera species on a bracket fungus by determining whether the frequencies of certain assemblages departed from random expectations” in 1967 and the consecutive year, 1968 (Branchet, 2020, p.1). There was an ongoing debate about the relationship between the species interactions and co-occurrence data because some ecologists criticized the lack of expectations of the statistical method used (Branchet, 2020). New technologies were developed to improve the extraction of ecological data from the co-occurrence information. The current categories of methods to extract ecological information include matrix-level approaches, species distribution models and network analysis methods.
One current category of the method used to extract the ecological method is the matrix-level approach. According to Branchet (2020), “the matrix-level approaches aim at determining the main drivers of species’ distribution for a given community based on the entire incidence matrix properties” (p.2). To do this, a comparison is made between the indices based on the data derived from the observation data and the random expectations derived from the null models (Branchet, 2020). The vertex-edge incidence matrix and the edge-face incidence matrix are the two common types of incidence matrix used to represent graphs (Branchet, 2020). A hierarchical approach was proposed based on “coherence, species turnover and clumping to characterize the spatial structure of the community and hence determine the role played by colonization and niche partitioning” (Branchet, 2020, p.3). Thus, one current category of the method used to extract the ecological method is the matrix-level approach which led to more advanced results of the null models as they analyze graphs.
The second current category of the method used to extract ecological method is species distribution models, and they predict the geographical distribution of abiotic variables in species. According to Zurell et al. (2020), SDMs consist of common ecology models, conservation and evolution. “The advent of ready-to-use software packages and increasing availability of digital geoinformation have considerably assisted the application of SDMs in the past decade, greatly enabling their broader use for informing conservation and management and quantifying impacts from global change” (Zurell et al., 2020, p.1). The developments of the model’s aspects and their applications should be adequately considered. “Despite the widespread use of SDMs, standardization and the documentation of the modelling protocols remain limited, which limits the assess development steps which can be are appropriate for end users” (Zurell et al., 2020, p.1). “In particular, the standard protocol provides a fast and good guide to the common guidelines of fitting SDMs; This complement and integrates recent work defining range model metadata standards” (Zurell et al., 2020, p.3). According to Branchet (2020), from 1990 to 2000, SDMs were very criticized because they neglected biotic interactions. “In most cases, these models provide individual species responses to the abiotic environment together with a covariance matrix whose elements capture the correlations in the incidence matrix that are not explained by the abiotic factors” (Branchet, 2020, p.5). Thus, the second current category of the method used to extract ecological method is species distribution models, and they predict the geographical distribution of abiotic variables in species.
The third current category used to extract ecological methods is network analysis methods. According to Douma et al. (2019), “any type of observations in ecology and evolution can be most conveniently expressed and compared as fractions and graphs, and it has been estimated that over a third of publications in ecology analyze some kind of proportional data” (p.2). In abiotic factors and species abundance, each species have two vectors of an optional set of covaries and an occurrence (Branchet, 2020). In the graphs, rows represent the vertices, and the columns of the graph represent the edges. The approaches are near the Forbes coefficient proposed over a century ago, “but recent approaches now focus on the entire set of the significant co-occurrence associations, i.e. the co-occurrence network” (Branchet, 2020, p. 6). A dividing line should exist in these methods because some approaches interpret the changes in the interactions in ecology. According to Cecchini et al. (2018), the most frequently used modelling paradigms are networks for dynamical systems. According to Delma et al. (2019), “Ecological networks can efficiently be represented using the mathematical formalism of graph theory” (p.2). Cecchini et al. (2018) also stated that some common methods in graph theory include clustering coefficient analysis, degree distribution analysis, and modularity analysis. Network analysis methods infer the ecological relationships derived from the incidence matrix (Cecchini et al., 2018). Thus, network analysis methods are the third current category of method used to extract ecological methods.
Thus, the categories of methods have a major shift in interpreting the essential associations in the data derived from co-occurrence as proof of ecological interactions. It is essential to infer ecological interactions from the present absence of data. One should never forget that this is only possible if ecological interactions leave essential information that statistical methods can detect. Recently, examinations have been developed in statistical approaches used to examine species, and this information enables one to detect fundamental interactions in the statistical techniques. The low performances derived from the observations in the investigations done by specific systems. The results show the limitations in the statistical approaches that more advanced technology can address. Ecologist should use examinations because they determine if the present-absence data could leave a signal and if the signal detected could be adequately interpreted.
Blanchet, F. G., Cazelles, K., & Gravel, D. (2020). Co‐occurrence is not evidence of ecological interactions. Ecology Letters, 23(7), 1050-1063.
Bruelheide, H., Dengler, J., Purschke, O., Lenoir, J., Jiménez-Alfaro, B., Hennekens, S. M., … & Jandt, U. (2018). Global trait–environment relationships of plant communities. Nature ecology & evolution, 2(12), 1906-1917.
Cecchini, G., & Schelter, B. (2018). Analytical approach to network inference: Investigating degree distribution. Physical Review E, 98(2), 022311.
Delmas, E., Besson, M., Brice, M. H., Burkle, L. A., Dalla Riva, G. V., Fortin, M. J., … & Poisot, T. (2019). Analyzing ecological networks of species interactions. Biological Reviews, 94(1), 16-36.
Douma, J. C., & Weedon, J. T. (2019). Analyzing continuous proportions in ecology and evolution: A practical introduction to beta and Dirichlet regression.
Rousseau, Factors influencing transferability in species distribution models. Ecography, 2022(7), e06060.
Zurell, D., Franklin, J., König, C., Bouchet, P. J., Dormann, C. F., Elith, J., … & Merow, C. (2020). A standard protocol for reporting species distribution models. Ecography, 43(9), 1261-1277.