Large collections of freely available food webs are commonly reused by researchers to infer how biological or environmental factors influence the structure of ecological communities. Although reusing food webs expands sample sizes for community analysis, this practice also has significant drawbacks. As food webs are meticulously crafted by researchers for their own specific research endeavors and resulting publications (i.e., books and scientific articles), the structure of these webs inherently reflects the unique methodologies and protocols of their source publications. Consequently, combining food webs sourced from different publications without accounting for discrepancies that influence network structure may be problematic. Here, we investigate the determinants of structure in freely available food webs sourced from different publications, examining potential disparities that could hinder their effective comparison. Specifically, we quantify structural similarity across 274 commonly reused webs sourced from 105 publications using a subgraph technique. Surprisingly, we found no increased structural similarity between webs from the same ecosystem nor webs built using similar network construction methodologies. Yet, webs sourced from the same publication were very structurally similar with this degree of similarity increasing over time. As webs sourced from the same publication are typically sampled, constructed, and/or exposed to similar biological and environmental factors, publications likely holistically drive their own webs' structure to be similar. Our findings demonstrate the large effect that publications have on the structure of their own webs, which stymies inference when comparing the structure of webs sourced from different publications. We conclude by proposing different approaches that may be useful for reducing these publication‐related structural issues.

中文翻译：

---

出版物驱动的食物网结构一致性：对比较生态学的影响


研究人员通常重复使用大量免费提供的食物网来推断生物或环境因素如何影响生态群落的结构。尽管重复使用食物网扩大了社区分析的样本量，但这种做法也有明显的缺点。由于食物网是由研究人员为自己的特定研究工作和由此产生的出版物（即书籍和科学文章）精心制作的，因此这些网的结构本质上反映了其来源出版物的独特方法和协议。因此，将来自不同出版物的食物网组合在一起而不考虑影响网络结构的差异可能会有问题。在这里，我们调查了来自不同出版物的免费食物网中结构的决定因素，研究了可能阻碍它们有效比较的潜在差异。具体来说，我们使用子图技术量化了来自 105 篇出版物的 274 个常用网页的结构相似性。令人惊讶的是，我们发现来自同一生态系统的 Web 或使用相似网络构建方法构建的 Web 之间的结构相似性没有增加。然而，来自同一出版物的网站在结构上非常相似，并且这种相似程度随着时间的推移而增加。由于来自同一出版物的 Web 通常是经过采样、构建和/或暴露于相似的生物和环境因素的，因此出版物可能会从整体上推动其自身 Web 的结构相似。我们的研究结果表明，出版物对其自身网络结构有很大影响，这在比较来自不同出版物的网络结构时阻碍了推断。 最后，我们提出了可能有助于减少这些与出版物相关的结构性问题的不同方法。