Grazing and the vanishing complexity of plant association networks

Submitted by editor on 24 March 2021.Get the paper!

Grazing and the vanishing complexity of plant association networks 

Plants do not grow in isolation in nature but interact with their neighbors – this can determine how the community as a whole will respond to environmental changes. If we are to understand the effect of those changes, we thus need to know which type of interactions are present in natural communities, and how they are affected by disturbance, stress and other environmental factors. One such ubiquitous factor for plant communities is grazing, which can tip the balance between competition and facilitation among plants in grasslands.

Our study site, La Crau (Mediterranean France), has a rich history of such grazing. Archeological remains and roman trading evidence show that sheep herding has been around for no less than two millennia (Saatkamp et al. 2020). This historical activity, along with the poor accessibility of the water table to plants, maintains a unique steppe-like landscape (Figure 1). Diversity of plants there is one of the highest in France (above 45 species/m²), and many endangered birds and insects have most of their population, if not all of it, in the area. This warranted national protection in 2001 and since then, much of the area is managed in collaboration with stakeholders for the long-term preservation of biodiversity and traditional sheep farming practices. This context is particularly opportune for both applied and fundamental ecological research on herbaceous plants.


Figure 1. (a) General view of the steppe-like landscape of La Crau (France), locally named a Coussoul (from Lat. Cursorium, “the space that we cross”). The area is grazed by herds of sheep at various grazing intensities, ranging from intensely grazed zones (b), to areas where grazing is lower and vegetation higher (c). The relative homogeneity of the steppe (apart from variations in grazing pressure) makes it very interesting to isolate and study the effect of grazing on plant communities.


Measuring the effect of grazers on plant-plant interactions requires measuring interaction strengths. A typical approach to do so is based on experiments, in which plants are paired to infer the effect they have on each other under various levels of stress. Such experiments have dramatically advanced plant ecology in the last decades, but can fail to provide an accurate description of plant communities (Engel and Weltzin 2008). One reason may be because isolating plants in pairs can produce measurements of interactions that are disconnected from those that occur in more complex, species-rich natural settings.

To alleviate this shortcoming, an alternative approach is to rely on in situ spatial patterns instead of experiments. In this approach, it is assumed that two plants that grow together frequently are doing so because of a positive interaction (facilitation). Conversely, two plants that segregate in space are assumed to do so because they interfere with each other (competition). Such pairwise spatial patterns are often named “associations”. For a given community, they can be viewed as a network where positive links represent pairs of plants that aggregate in space (positive associations), and negative links represent plants that segregate in space (negative associations). Association networks are not equivalent to interaction networks in ecological communities (Barner et al. 2018). They may nonetheless inform on the general type of interactions present in a given plant community, as evidenced by the common use of this approach in arid ecosystems (Saiz and Alados 2012). Surprisingly, association networks are seldom used outside of those particular ecosystems – a knowledge gap that motivated the work we present in our article, Grazing and the vanishing complexity of plant association networks.  


Figure 2. We recorded in the field the spatial distribution of all plant individuals to a precision of 2 mm using linear transects (a, d, e). From these transects, we derived association networks between plant species (b), which we summarized into network-level metrics. We then investigated how these network-level metrics changed along the grazing gradient (c).


To investigate how association networks change with grazing, we surveyed the spatial structure of the vegetation, built association networks and focused on different network metrics to describe the changes in community structure (Figure 2). We show that under low grazing conditions, plant communities have a highly non-random structure, in which negative associations dominate over positive ones, suggesting that competition processes are dominant under low-grazing conditions in these plant communities. In these conditions, the number of associations per species was highly variable, suggesting that not all plants were equally competitive but instead engaged into competition to various degrees. Higher grazing made those non-random patterns close to their null expectation, hinting at a “neutralizing” effect of grazing on plant-plant interactions.

Our work highlights that association networks provide valuable insights into the structure of plant communities and extend what has been known about plant strategies. Here we focused only on a few aspects of networks for which we had ecological expectations – however, networks are rich descriptions of ecological communities, and network analyses that are common in other areas of ecology could also be applied to plant communities to better understand the effects of anthropogenic disturbance.

This work would not have seen the light without the help from the staff of the RNN des Coussouls de Crau. And because recording the spatial structure of plants takes an awful amount of time (Figure 3), we are grateful for the infinite patience of all the field assistants during the many months of field work: many thanks to them!


Figure 3. One of the co-authors taking a well-deserved break from surveying.


Alexandre Génin (@alexandregen1), Thierry Dutoit, Alain Danet, Alice le Priol and Sonia Kéfi -  Grazing and the vanishing complexity of plant association networks in grasslands.




Barner, Allison K., Kyle E. Coblentz, Sally D. Hacker, and Bruce A. Menge. “Fundamental Contradictions among Observational and Experimental Estimates of Non-Trophic Species Interactions.” Ecology 99, no. 3 (March 2018): 557–66.

Engel, E. Cayenne, and Jake F. Weltzin. “Can Community Composition Be Predicted from Pairwise Species Interactions?” Plant Ecology 195, no. 1 (March 2008): 77–85.

 Saatkamp, Arne, Frédéric Henry, and Thierry Dutoit. “Romans Shape Today’s Vegetation and Soils: Two Millennia of Land-Use Legacy Dynamics in Mediterranean Grasslands.” Ecosystems, November 11, 2020.

 Saiz, Hugo, and Concepción L. Alados. “Changes in Semi-Arid Plant Species Associations along a Livestock Grazing Gradient.” PloS One 7, no. 7 (2012): e40551.

Insights into Oikos papers