Warming global temperatures are driving species to shift beyond their historical geographic ranges into novel, expanded ranges, where they may have both positive and negative effects on resident populations. Although no framework currently exists to anticipate these effects, invasion biology theory suggests these effects may be predictable from species’ historical roles and trophic positions. We conducted a global synthesis of over 1,200 population-level effects of range-expanding species reported across 1,071 studies. We hypothesized that, similar to invasive species, expanding species’ impacts could be predicted by (1) their ecological roles in their native regions and (2) their trophic levels. We found that effects in species’ historical ranges reliably predict their impacts in expanded ranges, but that, overall, range expanders tend to be more detrimental—or less beneficial—in novel communities. Higher trophic level expanders had more negative effects on resident species than lower trophic level expanders, and interaction type further modulated outcomes, with the strongest effects arising from species that modify the physical environment. Finally, we found that native producers were more negatively affected than humans, vertebrates, or invertebrates. Together, our results indicate that existing knowledge of species’ roles are key to anticipating which species undergoing climate-driven range expansions are likely to have the largest effects on resident communities.

Identifying Expanding Species and Impacts

We began by compiling a database of species undergoing range expansions. We first collated the 134 species in range-expanding databases published by Sorte et al. (2010) and Henry & Sorte (2022). We then expanded our database to a global extent through an updated literature review. In June 2022, we searched Web of Science using the string: “range” AND “shift*” OR “expan*” OR “exten*”. We then filtered the results using the string: “ocean*” OR “marin*” OR “estuar*” OR “saltwater” OR “reef” OR “subtidal” OR “intertidal” to focus records for marine species. We reviewed the titles and, when necessary, text from the first 600 results and incorporated expanding species based on the conclusions of the authors themselves or through our own evaluation. To focus on species undergoing climate-driven redistributions, species had to meet several criteria to be included in the database. First, we focused on relatively recent shifters (i.e., beginning after ~1850) that have become established in novel ranges, so we did not include species considered to be transient (temporary extralimital appearances, such as those driven by El Niño events as in Sanford et al. [2019]) or the extralimital appearances of very few (<10) individuals. Second, we only included species with observed, rather than forecasted, range expansions. Finally, we excluded species whose range expansions appeared to rely on human activity, such as those passing through the Suez or Panama Canals or via the dumping of ballast water. Our database, therefore, is not intended to be an exhaustive collection of all observed and putative expansions but contains a subset of marine species that have undergone climate-driven expansions and are likely established in their novel ranges.

We then conducted a second set of literature searches to compile evidence of ecological effects, including effects on humans. We searched Web of Science for studies published through December 2024 using the scientific name (and any synonyms) of each species in the database and collected studies reporting ecological effects, which we identified by reviewing titles and abstracts. For species with >800 results, we reviewed the first 400 results and then filtered the rest using the string: “ecol*” OR “impact*” OR “effect*”. For any species with <100 results in Web of Science, we also conducted searches in Google Scholar, again using the species name and all synonyms. In total, we reviewed over 90,000 titles and 5,100 abstracts (including duplicates between Web of Science and Google Scholar) when collecting articles documenting ecological effects.

Assigning Impact Scores

We assessed ecological impacts using a modified format of the EICAT protocol (2). We first classified impacts by mechanism. Mechanisms yielding negative (i.e., detrimental) impacts included: predation, parasitism, herbivory, competition, disease transmission, physical disturbance, interaction with non-native species (NS) that benefit NS, poisoning/toxicity, indirect (negative) interactions with other species, and “other” negative impacts. Mechanisms yielding positive (i.e., beneficial) impacts included: food provisioning, habitat provisioning, nutrient subsidies, interactions with NS that harm NS, positive indirect interactions, and “other” positive impacts. Following the EICAT protocol, we then assigned impact magnitudes to each mechanism using a semi-quantitative gradient ranging from 1 (lowest) through 5 (highest) (Fig. 1B), reflecting the maximum response described for each mechanism in each article. We scored maximum per expanding species per article impacts for two reasons: (1) this allowed us to estimate maximum potential effects and (2) it most closely aligns with the established EICAT protocols (Volery et al. 2020). We focused on assigning scores of observed effects from primary literature. We therefore did not score predicted effects (i.e, from modelling papers), and we did not score effects from review papers or textbooks to avoid double-counting scores from the primary literature. Species with no documented effects were considered “data deficient”.

We collected additional data about the expanding species and the article. For each expanding species, we collected taxonomic information (phylum and class), its trophic level relative to the impacted responding species, and the responding species classified into broad groups. Relative trophic level of the expander was assessed as “higher” if the expanding species consumes the responding species, “lower” if the expander is consumed by the responding species, or the “same” if the expanding species appears to be on a similar trophic level as the native species. If the trophic level was not clear from the article, we assigned trophic level by kingdom (i.e., expanding producers were always “below” native herbivores, but were considered the “same” with native producers). Impacted organisms were broadly classified into: “human”, “vertebrate”, “invertebrate”, “producer” or, for widely distributed impacts, “multiple” species. For each article, we determined the location of the study and whether the study was conducted in the species’ historical, expanded, or—in rare cases—invaded range (when species have undergone both anthropogenic introductions and climate-driven expansions). “Expanded” and “historical” ranges were designated conservatively and were based on both the source reporting the expansion and the article from which we were assigning the score itself. Nearly all studies occurred in only one of a species’ historical or expanded ranges; when an article described impacts in both ranges, we assigned impact magnitudes for each range separately.

Analysis

We first tested the hypothesis that a species’ role in its historical range would predict its effect in its expanded range. To accomplish this, we grouped effects according to range (“historical” vs. “expanded”) and calculated the average value across all maximum effects within each range. We then performed a simple linear regression to examine the relationship between historical- and expanded-range effects and a two-sample t test to compare average effect magnitudes between ranges. Both analyses were conducted at the species and class levels. We also examined how the relationship between historical and expanded-range effects varies with effect sign using a parallel approach. We opted to not use mixed-effects model analysis for two reasons: 1) the vast majority of studies provided effects in only one range, so we could not assign a corresponding value for the missing range and 2) after averaging, each species is represented by a single point; as such, adding random effects of either species or study would have resulted in singular models.

To further understand factors driving effect magnitude, we examined how effects vary by 1) mechanism, 2) expander trophic level, and 3) taxonomic classification of the expanding species, and 4) by responding species group. To assess variation due to mechanism type, we first divided the effects by whether the mechanisms were positive (e.g., habitat or food provisioning) or negative (e.g., competition or predation). For each group of effects, we analyzed EICAT scores as an ordinal response using CLMMs. CLMMs included impact mechanism and range position (historical vs. expanded) as fixed effects, along with their interaction, and species as a random intercept to account for repeated observations. All models were fit using a logit link with flexible thresholds. We evaluated whether the effect of mechanism differed across ranges by comparing a full model including the mechanism × range interaction to an additive model without the interaction using likelihood ratio tests (LRTs). These interaction tests were conducted using only species with documented effects in both ranges. We then assessed overall differences among mechanisms using the full dataset by fitting models with impact mechanism as a fixed effect and comparing them to reduced models using LRTs. Mechanisms not observed in both historical and expanded ranges, or with fewer than four observations in one range (e.g., hybridization, nutrient subsidies, and effects benefiting nonnative species), were excluded from these analyses. To assess how effects varied by expander trophic level, expander phylum, and affected taxon, we followed the same analytical framework. For each variable, we tested for differences across ranges by comparing models with an interaction between the variable of interest and range to additive models using LRTs. We then evaluated the overall effects of each variable by comparing models including the variable of interest as a fixed effect to null models.