Skip to content

Home

Jump to bottom
Shai Pilosof edited this page Jul 24, 2023 · 3 revisions

Introduction

The EMLN package contains a collection of multilayer data sets. Because these data sets came from different sources and are different from each other (e.g., pollination vs food webs), we needed to standardize the data structures and develop pipelines that would allow us to then convert these data structures to the more common tables researchers use. In this Wiki we describe this process.

You can find more details in our paper linktopaper and the accompanying online guide: https://ecological-complexity-lab.github.io/emln_package/index.html

Overview

We obtained network data from multiple sources, including Web of Life [1], Interaction Web Database, mangal [2] and publications. All the data can be used according to their license. We also took care to reference the data sources in the network objects. For instance try:

library(emln)
d20 <- load_emln(20)
d20$references
d20$description

For data sets that were not originally collected to construct EMLNs we followed stringent criteria to determine the nature and extent of the EMLN, as follows. For temporal EMLNs, we considered as layers networks that were collected in the same place, and, to the extent possible, in even intervals and with the same methodology and effort (e.g., [3,4]). For spatial EMLNs we considered as layers networks that were collected using similar methodology across locations and as part of the same study or group of studies (e.g., [5,6]). For multiplex EMLNs, we considered network sets that have previously been analysed as multiplex networks (e.g., [7,8]). When available, we included node, layer and link attributes.

Prior to inclusion in the package, we validated each data set and standardised species names using the National Center for Biotechnology Information (NCBI) database via the R package taxize. We formatted each EMLN to a standard data structure that included the following data frames:

  • Nodes: Information on the physical nodes in the network and their attributes.
  • State nodes: Information on specific instances of nodes in layers. For example, in network no. 28 state node attributes were given in the original paper as the degree and strength of each node in its respective layer.
  • Layers: Information on the layers present in the multilayer network, such as coordinates, type of layer (e.g., time point or patch).
  • Extended: An ELL of the network. Each interaction is given a unique id.
  • Description: A general description of the network. Fields include the source, the type of ecological interactions (e.g., pollination), and the type of multilayer (e.g., spatial or temporal).
  • References: Information on the original publication(s) describing the network.

Most of the data sets were not originally collected to be multilayer networks. For these data sets, we provided the layer names and details according to the original study.

Because networks differed in their types and attributes, we could not use the same "wide format" data frame for all networks. For instance, some layers contained information on coordinates while others did not. Hence, a data frame with the fields: layer_id, longitude, latitude is not adequate for all data sets. The same issue occurs when node and link attributes are present. Therefore, we used a "long format" that included id, attribute and value fields. We provide a complete description of the formats in the Wiki that accompanies the development of the package at https://github.com/Ecological-Complexity-Lab/emln_package/wiki/Standard-formats-for-EMLN-data-sets. While a long format is adequate for standardisation of a database, it is less convenient for network analysis. Therefore, the function load_emln transforms each data set to its wide format before including it in the package.

References

  1. Fortuna, M. A., Ortega, R., & Bascompte, J. The Web of Life (2014).
  2. Poisot, T., et al. mangal - making ecological network analysis simple. Ecography 39, 384–390. doi:10.1111/ecog.00976 (2016).
  3. Chacoff, N. P., et al. Interaction frequency, network position, and the temporal persistence of interactions in a plant-pollinator network. Ecology 99, 21–28. doi:10.1002/ecy.2063 (2018).
  4. Ponisio, L. C., et al. Proximity of restored hedgerows interacts with local floral diversity and species’ traits to shape long-term pollinator metacommunity dynamics. Ecol. Lett. doi:10.1111/ele.13257 (2019).
  5. Carstensen, D. W., et al. Beta diversity of plant-pollinator networks and the spatial turnover of pairwise interactions. PLoS One 9, e112903. doi:10.1371/journal.pone.0112903 (2014).
  6. Trøjelsgaard, K., et al. Geographical variation in mutualistic networks: similarity, turnover and partner fidelity. Proc. Biol. Sci. 282, 20142925. doi:10.1098/rspb.2014.2925 (2015).
  7. Pocock, M. J. O., et al. The robustness and restoration of a network of ecological networks. Science 335, 973–977. doi:10.1126/science.1214915 (2012).
  8. Kéfi, S., et al. How structured is the entangled bank? The surprisingly simple organization of multiplex ecological networks leads to increased persistence and resilience. PLoS Biol. 14, e1002527. doi:10.1371/journal.pbio.1002527 (2016).
Clone this wiki locally