Progress and future directions of biogeographical comparisons of plant-fungal interactions in invasion contexts

Abstract

Plant invasions are biogeographical phenomena that may involve shifts in belowground plant-fungal interactions, such as the release from fungal pathogens or more beneficial interactions with mutualists in nonnative ranges. However, native and nonnative ranges are not uniform but environmentally heterogeneous, and plant-fungal interactions are strongly shaped by spatio-environmental context. Intense discussion at the 45^th New Phytologist Symposium revealed that we lack information on how well spatio-environmental variation within ranges has been considered in samplings and analyses of studies comparing plant-fungal interactions between ranges. Through a systematic review, we assessed the sampling quality of recent biogeographical studies. We found that the majority relied on a limited population sampling within each range, often covering only a small fraction of the species' spatial distribution and macroclimatic niche. Additionally, low similarity between the sampled climatic gradients in the native and nonnative ranges might have introduced false-positive differences across ranges. These sampling deficiencies may undermine the robustness and representativeness of range comparisons, thereby restricting our ability to accurately assess the role of plant-fungal interactions in invasion success. We recommend that future research incorporate broader and more comparable spatio-environmental variation in both ranges, and we provide practical guidelines for improving sampling designs.

Keywords: ecological sampling design; environmental heterogeneity; native vs nonnative range comparisons; plant invasion; plant–fungal interactions; sampling bias; sampling quality; spatio‐environmental variation.

Fig. 1

Conceptual overview of the selection of articles and studies included in our review. (a) Our review targeted articles that compared fungal community composition or their function (e.g. their effect on plant performance) across the ranges of the study plant species. Common features among these studies included the collection of soil or root samples to assess fungal associations, measurements of plant performance to evaluate fungal effects, or the collection of seeds for use in experiments. The studies varied in design and approach, encompassing three different study types: field surveys, field experiments, and greenhouse experiments. (b) In total, we identified 41 relevant articles, including 79 individual studies. If articles involved multiple study types or multiple target plant species, they were counted as individual studies. Otherwise, they were counted as single studies. Further details on the process of the systematic literature review and the entire literature screening are provided in Supporting Information Notes S1, Figs S1 and S2.

Fig. 2

Schematic overview of the calculation of sampling quality parameters. The example illustrates a field survey by Sheng et al. (2022) on fungal associations with Conyza canadensis populations. The map displays Global Biodiversity Information Facility (GBIF) occurrence data for C. canadensis in its native (blue dots) and nonnative (magenta dots) ranges. Range classification of the GBIF occurrences followed the Invasive Species Compendium (ISC; Diaz‐Soltero & Scott, 2014). Sampling locations from Sheng et al. (2022) are shown as orange triangles in both ranges. Three parameters estimate the sampling quality. (1) Spatial coverage: GPS coordinates of GBIF occurrences and sampling locations were used to calculate dynamic match coefficients (DMC; Sporbert et al., 2019). Dynamic match coefficients represent a measure of cell matches between the sampling locations and the species' global distribution. (2) Climatic coverage: Bioclim variables were extracted to describe the sampled climatic space (sampling locations) and the realised climatic niche (GBIF occurrences). Using dynamic range boxes (DRB; Junker et al., 2016), we calculated the overlap between the sampled climatic spaces (orange squares; light blue, native; light magenta, nonnative) and the overall realised niches (larger squares; blue, native; magenta, nonnative). The left overlap (light blue) indicates the proportion of the overall native climatic space covered by the native‐range sampling, while the right overlap (light magenta) represents the proportion of the nonnative climatic space covered by the nonnative‐range sampling. (3) Climatic similarity: DRBs were used to calculate the climatic overlap between samplings in both ranges. The left overlap (light blue) indicates how much of the nonnative climatic space was covered by the native‐range sampling, whereas the right overlap (light magenta) indicates how much of the native climatic space was covered by the nonnative‐range sampling. Further details on the calculation of sampling quality parameters, including a discussion on the limitations of both the GBIF and ISC databases, are provided in Supporting Information Notes S2 and Fig. S4.

Fig. 3

Sampling quality parameters in relation to the range, the number of species studied, and the year of its publication. Relationships are shown for (a–c) spatial coverage within ranges (SpatCov_WR), (d–f) climatic coverage within ranges (ClimCov_WR), and (g–i) climatic similarity between ranges (ClimSim_BR). The climatic similarity boxplots (g) represent the overlap between the sampled native and nonnative climatic spaces: native climatic similarity is displayed by the overlap of the native climatic space with the nonnative, and vice versa for the nonnative climatic similarity. The colouring in all panels is based on the range (light blue, native; magenta, nonnative). Boxplots (a, d, g) show the interquartile range with the horizontal line indicating the median, and whiskers extending to the min and max values without outliers. The dashed lines represent regression lines (b, c, e, f, h, i), indicating nonsignificant relationships. The confidence intervals of the lines are presented as shadings in grey. Results are derived from linear mixed‐effect models with the variable paper set as a random effect. Details on the calculation of sampling quality parameters are provided in Fig. 2 and Supporting Information Notes S2. Individual values for each parameter and study are listed in Table S1.

Fig. 4

Correlation between the sampling quality and the impact factor of journals that published the reviewed studies. Correlations with impact factors are presented for (a) spatial coverage within ranges (SpatCovWR), (b) climatic coverage within ranges (ClimCovWR), and (c) climatic similarity between ranges (ClimSimBR). The current impact factor of the respective journals was retrieved from the Journal Citation Reports (clarivate.com). The solid lines represent regression lines, indicating significant relationships. The confidence intervals of the lines are presented as shadings in grey. Results are derived from linear models. Details on the calculation of sampling quality parameters are provided in Fig. 2 and Supporting Information Notes S2. Individual study values for each parameter are listed in Table S1.