I really appreciate it when researchers engage in public discussions through their papers, challenging and replying to each other. I think these scientific “dialogues” are a great way to see what is happening in the frontiers of science. Recently, an interesting series of papers was published addressing the use of metawebs of biotic interactions to infer local ecological networks. Is it a good option?
But first things first! What is a metaweb? In this context, a metaweb is defined as a network summarising the potential trophic and non-trophic interactions in a given set of species (Maiorano et al. 2020). The metaweb developed by Maiorano and colleagues, for example, summarises all the trophic interactions between tetrapod species across Europe. The use of metawebs for inference of local networks is easily explained: let’s imagine that, in our sampling unit (e.g. a 10 km2 square grid), we have three species co-occurring, Iberian lynx, fox, and wild rabbit. In the metaweb we see that both, the Iberian lynx and fox predate upon wild rabbit, therefore, in our local network we infer that fox and Iberian linx predate upon wild rabbit.
In a previous post, I addressed a related question: if we are going to study the biogeography of ecological networks (food webs for example), from where are we going to source the data? Because we would need spatially distributed ecological networks in any given large enough area (e.g. a continent or global). Lots of them! Two main options are available: i) databases of local empirical networks and ii) inference of local networks from co-occurrence, species trait data and/or diet. Using metawebs is aligned with the second option, using species-co-occurrence and a metaweb of trophic interactions, we infer local food webs.
This is a simplified version of what Botella et al. (2024) did. These authors investigated the influence of land management intensity on food-web architecture. In particular, they addressed these food web properties: apex and basal species proportions, connectance, omnivory, trophic chain lengths and compartmentalisation. They conclude that the architecture of food webs was influenced by land use and management intensity, with intensification strongly lowering the proportion of apex predators. Additionally, intensification lowered the proportion of basal species, favouring mesopredators, decreasing food web compartmentalisation and increasing connectance.
But how did they reach these conclusions? What were the local food webs they were looking at? They used Maiorano’s metaweb to infer local food webs of co-occurring species present in each cell. As they simply put it in their paper “two species were assumed to interact locally if they are both observed in the cell and if they are known to interact in the metaweb”.
However, Brimacombe et al. (2024) challenged this approach. The authors argue that researchers must test these methods against actual field data to evaluate their validity. The two main criticisms are i) the appropriateness of inferring networks from metawebs; and (2) the interpretation of their results to support their conclusions.
In particular (concerning i), they argue that this approach ignores the fact that these tetrapod local food webs lack any spatial or temporal dimensions. Despite evidence that species interactions reorganize across time and space, the interactions in these local food webs lack a spatial or temporal dimension. These local food webs are artificially confined to a 1 km2 resolution. Additionally, supporting these local food webs on species co-occurrence ignores the temporal dimension, did the potentially interacting species co-occur in time also?
In what concerns ii), Brimacombe et al. (2024) argue that even though land-use intensity contributed, on average, to < 1% of the total explained variation across each local food web structural metrics, the authors still titled their article “Land-use intensity influences European tetrapod food webs” giving relevance to these results.
Botella and colleagues did not leave these remarks unanswered. In their reply, Botella et al. (2024) concede that the temporal co-occurrence is vaguely defined, “… assuming known interactions will inevitably happen at some point, across multiple years and in the vicinity of the pixel.”. They agree that validating local food webs with field data would be the best option. But they rightly assert that field-based data are extremely challenging to obtain considering that many potential interactions might not be observed during the sampling time, which is generally limited. Furthermore, as we have shown elsewhere (Mestre et al. 2022), these field-sampling-based food webs are sampled with different methods, spatial resolution or taxa aggregation, impairing comparability. As such, they go on, inferring from metawebs might be a reasonable starting point, comparable across space, which might be essential for some research questions. On what concerns remarks relative to the magnitude of the effect Botella et al. argue that the effect sizes associated with land use intensity were important, explaining up to 10% of the total variability of food web metrics.
As I mentioned in my previous post, I agree that these approaches, based on co-occurrence and metawebs, are a good enough option, and it might be ideal for some research questions. Which approach should we choose is dependent on the research question we are attempting to answer.
I’ve used both! We used a database of real field-based networks in Mestre et al (2022) and metaweb-based inference in Mestre et al. (preprint). However, I’m sure this ongoing discussion will, at some point, lead us to very interesting results and to new approaches.
References
Botella et al. (2024). Land‐use intensity influences European tetrapod food webs. Global Change Biology, 30(2), e17167.
Botella et al. (2024). Don’t bite the hand that feeds you: Meta food webs help in the face of the Eltonian shortfall. Global Change Biology, 30(6), e17359.
Brimacombe et al. (2024). Applying a method before its proof of concept: A cautionary tale using inferred food webs. Global Change Biology, 30(6), e17360.
Maiorano et al. (2020). TETRA‐EU 1.0: a species‐level trophic metaweb of European tetrapods. Global Ecology and Biogeography, 29(9), 1452-1457.
Mestre et al. (preprint). Wildlife-vehicle collisions simplify regional food webs. Biorxiv.
Mestre et al. (2022) Disentangling food-web environment relationships: a review with guidelines. Basic and Applied Ecology, 61: 102-115.
Mestre et al. (2022). Human disturbances affect the topology of food webs. Ecology Letters, 25(11), 2476-2488.
Geekcologist 