Caprellid specimens were repeatedly collected at the same site with fouling assemblages from floating docks and on mooring lines of the aquaculture concessions of the Universidad Católica del Norte (UCN) in Bahía La Herradura in Coquimbo, Chile (29°57'58.4"S, 71°21'12.9"W) on August 30^th, 31^st and September 9^th 2022. Several culture lines for scallop aquaculture are established in the concession of UCN, where lantern-nets are suspended from longlines (Bakit et al. 2024). These artificial structures host extensive fouling communities (Dumont et al. 2009) providing habitat to many mobile organisms (including several species of caprellid amphipods) (Astudillo et al. 2009). For the sampling procedure, the buoys and longlines were lifted up from a boat, and the fouling biomass was scraped from these artificial substrata and brought to the lab (approx. transport time: 10 min). For each sampling we collected an approximate volume of about 10 l fouling biomass, which included seaweeds, hydrozoans, bryozoans, tunicates, mussels and other sessile organisms (for species inventory see e.g. Astudillo et al. 2009). Additional material was obtained from samples collected the same way and at the same site on 23 June and 7 July 2023.

In the laboratory, the fouling organisms were immediately placed in large trays (approximately 20 cm x 30 cm surface area) with seawater, and the material was sorted alive. No signs of predation in the samples were observed during the procedure. All amphipods were retrieved and carefully inspected under a dissecting microscope. Caprellid amphipods were identified to the lowest taxonomic level, and counted. Voucher material was fixated in ethanol and deposited in the collection of the UCN.

During the years 2004 to 2023, caprellid amphipods were collected annually for the Invertebrate Zoology laboratories in the Marine Biology program of the Marine Science Faculty in Coquimbo (30°S). For these courses, usually a few hundred live caprellid individuals were brought to the teaching laboratory (on seaweeds, bryozoans and hydrozoans). The collection of caprellid amphipods was conducted in a very similar way as described above by sampling extensive amounts of fouling organisms with the associated caprellid amphipods. The samples were collected a few hours before the course, transported to the nearby lab, and maintained alive for students to observe and document the morphology and behavior of the caprellids. Students quantified the ventilation movements of ovigerous caprellid females and had to identify the particular species for which they recorded these behaviors using Guerra-García and Thiel (2001); the species identifications were usually checked by the course instructors.

In order to characterize the recent survey efforts focusing on the caprellid fauna in Central and South America, we searched the literature using the Web of Science and GoogleScholar. The keywords “Caprella” and “amphipod” were linked with the names of all Central and South American countries. In order to identify additional studies, all studies on caprellids that were published after 2000 were carefully examined for cross-citations. The recovered references were then scanned to identify those that reported on caprellid surveys in their regions or countries. These studies typically included species inventories that were based on targeted samplings of the caprellid fauna. All studies were conducted by invertebrate zoologists, often including amphipod or even caprellid specialists, who were very familiar with the taxonomic literature and species identifications. The investigations focused on shallow habitats up to approximately 20 m water depth, including fouling communities (e.g. Nunez Velazquez et al. 2017; Chunga-Llauce and Pacheco 2021; Chunga-Llauce et al. 2022) and macrophyte or animal reefs (Díaz et al. 2005; Alarcón-Ortega et al. 2017; Cunha et al. 2018). Usually the authors sampled several sites within their study region, where individual sites had distances of a few to > 100 km between them. Most studies covered one or maximally two ecoregions (sensu Spalding et al. 2007). References that focused only on the population or reproductive biology of selected caprellid species were not included.

For comparative purposes, we extracted presence/absence data from each respective study, which is common practice in biodiversity reviews of specific groups or regions (see e.g. Gallardo and Penchaszadeh 2001; Cárdenas-Calle et al. 2020; Durand et al. 2024). Only records on species-level were considered in the current data consolidation. The similar approaches used by all examined studies allowed for direct comparison in the context of the current overview.

Worldwide georeferenced occurrences for C. mutica were downloaded and curated from the Global Biodiversity Information Facility (GBIF, www.gbif.org; downloaded on 06 September 2023). The database was augmented by an exhaustive literature search and further published records were added (i.e. derived from: Schurin 1935; Vassilenko 1967; Arimoto 1976; Locke et al. 2007; Ashton et al. 2008a; 2008b; Willis et al. 2009; Hosono 2011; Almón et al. 2014; Collin and Johnson 2014; Coolen et al. 2016; Peters and Robinson 2017; Heo et al. 2020; Lavrador et al. 2024). A total of 1388 occurrences of C. mutica were used for the model (excluding the current presence in Chile reported here; Fig. 1). The occurrences were thinned to reduce sampling biases (Aiello‐Lammens et al. 2015), leaving only one presence per grid cell (0.08°, see below), resulting in 800 occurrences. We also compiled information on documented absences from sampled localities in South America (n = 170), where previous community-level studies of Caprellidae did not detect any specimens of C. mutica. While these absences were not used in the SDM, they were used to cross-validate the output of the SDM.

Figure 1.

World oceans with mean sea surface temperature (SST) and confirmed reported presences and absences of Caprella mutica before and after the year 2009 (i.e. the survey of Boos et al. 2011) considered for this study. ‘No findings’ refers to sites where previous surveys had examined the caprellid fauna (see also Table 1) without finding Caprella mutica.

We used 13 oceanographic variables (Table 2) from the BioOracle database v.2.2. (Assis et al. 2018), with a 0.08° (~9.2 km²) resolution. These variables have commonly been used by previous studies and covered a wide range of biophysical and geochemical conditions in the ocean (Bosch et al. 2018), also reflecting relevant ecophysiological stressors for C. mutica as proven by experimental studies (Cook et al. 2007; Lim and Harley 2018). Rasters were masked to include only coastal grid cells, as the species is restricted to shallow waters. The degree of collinearity of environmental predictors was examined by using a variance inflation factor (VIF) analysis where values of VIF > 10 have traditionally been used to claim high collinearity. VIF analyses were carried out using the library ‘usdm’ (Naimi et al. 2014) in R (ver. 4.1.0; R Core Team 2024). Two variables (mean and range of phytoplankton concentration) showed a high degree of collinearity and were removed from further analyses.

The SDM was built using recommended methodological protocols (Bosch et al. 2018; Feng et al. 2019; Zurell et al. 2020). We created 10,000 random pseudo-absences obtained from all coastal grid cells. We used a Maxent modeling approach, a robust machine-learning algorithm successfully applied to implement SDMs (Elith et al. 2011; Phillips et al. 2017). The model fit was evaluated using the Area Under the Curve (AUC) of the Receiver Operating Characteristic Curve, where values close to 1 indicate a perfect fit. Analyses were conducted using the library ‘SDMtune’ (Vignali et al. 2020). The model’s accuracy was maximized by hyper-parameter tuning and different combinations of the regularization parameter and feature classes. We used a genetic algorithm to assess 150 possible combinations of parameters, evaluating 15 populations in two generations. Genetic algorithms are computational optimization techniques inspired by the process of natural selection (Goldberg and Holland 1988; Alhijawi and Awajan 2024), enhancing model performance, selecting relevant variables, or optimizing parameters when predicting species distributions based on environmental data (Vignali et al. 2020). To ensure the robust spatial transferability of SDMs, we used a four-fold spatial cross-validation scheme based on a checkerboard pattern, implemented in the library ENMeval (Kass et al. 2021) in R. We evaluated the importance of all oceanographic variables in terms of percent contribution and permutation importance and estimated the functional relationship between the occupancy probability and the top predictors using partial dependence plots to isolate the effect of each predictor. A Multivariate Environmental Similarity Surfaces (MESS) analysis was carried out to evaluate areas with non-analog oceanographic conditions. MESS analyses were carried out using the library ‘predicts’ (Hijmans 2024) in R. Finally, we projected the probability of species occurrence onto the global coasts using ArcGIS Pro (ver. 3.3.0; ESRI Inc.).

In total, seven individuals of C. mutica (6 adult males and 1 ovigerous female) were found on August 30^th, 31^st and September 9^th 2022. Besides this newly recorded NIS for this area, the 4 caprellid species Caprella equilibra, Caprella verrucosa Boeck, 1871, Caprella scaura and Deutella venenosa as well as the ischyrocerids Jassa marmorata Holmes, 1905, Jassa slatteryi Conlan, 1990 and Ericthonius cf. rubricornis (Stimpson, 1853), the maerid Elasmopus rapax A. Costa, 1853 (sensu Hughes and Lowry 2010), the aorid Aora typica Krøyer, 1845, the dexaminid Paradexamine cf. pacifica (Thomson, 1879) and a stenothoid Stenothoe sp. were found coexisting in the amphipod fouling communities of Bahía La Herradura.

In the course of the Marine Biology program of the Marine Science Faculty in Coquimbo, the dominant species in the samples varied between the years, but the most common species were Caprella equilibra, C. scaura and C. verrucosa, and on rare occasions Deutella venenosa; the species identified by the students (using Guerra-García and Thiel 2001) were frequently verified by one of the authors (MT). Prior to 2023, no Caprella mutica were found, but on 23 June 2023 a few caprellid amphipods examined by the students did not match any of the species reported in Guerra García and Thiel (2001). After closer examination, these individuals were confirmed to belong to C. mutica. In addition, two weeks later (7 July 2023), several individuals (adult males and females) of C. mutica were collected during a workshop on marine invasive species. The collected individuals have been deposited in the Biological Collection of the UCN (SCBUCN-5533 1 female + 1 male adult; SCBUCN-5537 1 female + 4 male adults; SCBUCN-5561 5 male adults).

The specimens of Caprella mutica collected in Coquimbo could be easily distinguished from its two sympatric congeners C. verrucosa and C. scaura by the absence of a projection on the head. Further, the individuals of C. mutica were characterized by numerous spiny projections on the dorsal surface of the pereonites (pereonites 1–7 in females, 3–7 in males), which distinguished them clearly from co-occurring Caprella equilibra (Fig. 2). In addition, hyperadult males exhibited dense setation on pereonites 1 and 2, and on gnathopod 2, leading to a conspicuous ‘hairy’ appearance, which is unique among the known Caprella species of the world (Platvoet et al. 1995 as ‘Caprella macho’; Guerra-García and Thiel 2001; Beermann and Franke 2011; Boos et al. 2011; Daneliya and Laakkonen 2012; Heo et al. 2020).

Figure 2.

Individual of Caprella mutica, collected in Bahía La Herradura (Coquimbo, Chile) on 09 September 2022. Habitus of adult male A lateral view B dorsal view. Scale bars: 5 mm.

The Asian species Caprella acanthogaster Mayer, 1890 shares some morphological characteristics with C. mutica that may cause confusion, such as the dorsal spination on the pereonites and the hairy appearance of adult males (Faasse 2005; Daneliya and Laakkonen 2012; Heo et al. 2020). However, the specimens of C. mutica found in Bahía La Herradura were characterized by a dense hairy setation all over pereonites 1, 2 and gnathopod 2, whereas the hairy setation in C. acanthogaster is restricted to gnathopod 2 only. Further, C. acanthogaster bears a pair of two tiny tubercles on the head whereas C. mutica specimens from Chile had no tubercles or projections on the head.

Over the course of the past 20–30 years, several surveys of the local caprellid fauna had been conducted in several countries of Central and South America (Table 1). These surveys documented a total of 25 caprellid species (of 27 taxa in total) on the Atlantic coast (between 21°N and 38°S), and 16 (of 17 taxa in total) species on the Pacific coast (between 23°N and 30°S). Only four of those species (Caprella equilibra, C. penantis, C. scaura and Paracaprella pusilla) were recorded on both Atlantic and Pacific coasts. No findings of Caprella mutica were reported in any of these surveys.

Following the initial survey of the local caprellid fauna by Guerra-García and Thiel (2001) and Thiel et al. (2003), the biota growing on aquaculture buoys in the Coquimbo region were again sampled and examined in 2007/08, and all previously identified caprellid species were recorded, but no C. mutica was found in that survey (Astudillo et al. 2009).

The SDM exhibited a high accuracy (AUC = 0.96), and the MESS analyses showed that the model could be extrapolated to ~96% of the coastal grid cells. The model predicted a high probability of occupancy around the native area in Northeastern Asia, and the already invaded areas in Europe and North America (Fig. 3). In general, there was a lower probability of occupancy in the southern hemisphere, except for some areas in South Africa, South Australia, New Zealand, and Chile. Areas with confirmed absences were characterized by low occupancy probabilities (Fig. 3). Along the Chilean coast, the model predicted elevated probabilities (0.30–0.68) of occupancy between 32–42°S, which is 200 to 1,400 km south of the newly confirmed occurrence in the Coquimbo area reported here (Fig. 4). In contrast, the SDM predicted a relatively low occupancy probability (0.07) in Bahía La Herradura.

Figure 3.

Probability of occupancy of Caprella mutica in coastal regions worldwide according to a calibrated SDM. Values closer to 1 (red) indicate higher occupancy probabilities, whereas values close to 0 (yellow) suggest lower occupancy probabilities. The SDM was calibrated at a 0.08° resolution, but is displayed here at a 1° resolution aggregation scale to improve visualization.

Figure 4.

Probability of occupancy of Caprella mutica in A the Southeastern Pacific and B in Coquimbo, Chile according to a calibrated SDM. Values closer to 1 (red) indicate higher occupancy probabilities, whereas values close to 0 (yellow) suggest lower occupancy probabilities. The SDM was calibrated at a 0.08° resolution, but is displayed in A at a 0.5° resolution aggregation scale to improve visualization. Asterisks mark the location of Coquimbo in northern central Chile.

The mean water temperature (i.e., sea surface temperature) was the top predictor explaining the occupancy of C. mutica with a 44% contribution and 74% of the permuted importance (Table 2). The remaining predictors reached much lower contribution and permuted importance, often by one order of magnitude lower and < 10% (Table 2). The partial dependence plot revealed that the effect of the mean temperature was hump-shaped, with maximum occupancy probabilities around 11.3 °C, declining at lower and higher temperatures (Fig. 5).

Figure 5.

Functional relationships between occupancy probability and mean water temperature according to the species distribution model. Each empty dot represents the occupancy probability estimated for each global georeferenced occurrence. The red line shows the partial dependence plot of the isolated effect of mean water temperature on the occupancy probability. The blue dot indicates the new occurrence site in La Herradura Bay, Chile.

The current finding of C. mutica in Chile represents the first record of this caprellid in South America. Native to the north-east Pacific and introduced to the coasts of North America, Europe, New Zealand and South Africa, C. mutica seems to prefer cold-temperate waters (e.g. Arimoto 1976; Ashton et al. 2008b; Willis et al. 2009; Peters and Robinson 2017). Based on the known temperature tolerances of C. mutica and given its invasion history, Boos et al. (2011) predicted the species’ potential to extend its range to southern Pacific and Atlantic coasts of South America which is now corroborated by our recent finding.

The finding of several adult males and an ovigerous female in Coquimbo in 2022, and the collection of additional adult individuals in 2023 suggests the successful establishment of a population in Bahía La Herradura. However, the observed abundances were quite low compared to the known mass occurrences of C. mutica in other introduced ranges (e.g. Buschbaum and Gutow 2005; Peters and Robinson 2017). This could be due to (a) competition with other well-established local caprellid species along with (b) suboptimal environmental conditions for C. mutica, or (c) simply be the result of a very recent arrival of this invader in the region. Since C. mutica had previously never been observed in Coquimbo despite annual scans of the local caprellid fauna, it is indeed likely that this species has arrived relatively recently. Furthermore, the fact that C. mutica has not been reported from other regions in Central and South America, where extensive surveys of the caprellid fauna had been conducted by experts (see Table 1 and references therein), also suggests that this species has only recently arrived in South America. Most of these other studies have surveyed several sites within a country or ecoregion, and explicitly focused on the caprellid fauna (e.g. Díaz et al. 2005; Guerra-García et al. 2006; Paz-Ríos et al. 2014; Chunga-Llauce et al. 2023b), and thus the absence of the highly characteristic C. mutica in these surveys strongly suggests that it had not been present in those previous surveys. Since many of these surveys included taxonomical experts for the crustacean family Caprellidae who examined hundreds of specimens, it is considered very unlikely that C. mutica would have been overlooked. The population development of C. mutica in northern-central Chile must thus be monitored carefully, also with regards to any negative impacts on the local fauna such as the endemic Deutella venenosa.

Overall, the predicted global occupancy probabilities reflected well the known native range of C. mutica as well as its occurrence in areas where it has been introduced (i.e. northern Europe and North America). The modelled predictions of our quantitative approach presented here are roughly in accordance with the “potential range” of C. mutica depicted by Boos et al. (2011). In direct comparison to the predicted probabilities in the northern hemisphere, the southern hemisphere seems to be less favorable for this caprellid species. The highest occupancy probabilities for C. mutica along the southeastern Pacific coast were observed around 32° and 42°S and were comparatively low in other areas such as Ecuador, Peru, as well as northern and southern Chile. Surprisingly, the SDM predicted only low occupancy probabilities for the Coquimbo Bay (0.07), well below other areas with a similar temperature, which is seemingly in contrast to the recent finding reported here. This new population might thus be living under near-suboptimal conditions that may prevent excessive population growth. A possible explanation could be that the original point of introduction of C. mutica to South America may have been located in central-south Chile with its large ports (i.e., San Antonio, Valparaíso and San Vicente) at 33° and 36°S, respectively, where predicted occupancy probabilities increased to up to 0.65. The species may, therefore, already have built undetected populations elsewhere that remain to be found. Further, the local population of C. mutica in Bahía La Herradura may be at its physiological limit, reducing the probability of a northward expansion towards northern Chile and Peru. Nonetheless, also if the original point of introduction was indeed in Bahía La Herradura, a further southward expansion to areas where oceanographic conditions could be more favorable, seems likely.

Mean water temperature was the most important variable driving the environmental niche of C. mutica. The predicted thermal tolerance according to the SDM, however, is much lower compared to estimations based on ecophysiological experiments (Ashton et al. 2007; Hosono 2011). The median lethal temperature for adults was estimated at 28.3 ± 0.4 °C, while no mortalities occurred at 2 °C, even surviving below-zero temperatures (Ashton et al. 2007). Nevertheless, occupancy probabilities fell below 0.05 at temperatures lower than 5 °C or above 20 °C (Fig. 5). Interestingly, rearing experiments revealed that early stages of C. mutica reach maturity in the range of 10–20 °C, but not at 5 °C (Hosono 2011). All things considered, this suggests that the geographic spread of the species is not only driven by water temperature and it may be strongly co-dependent on the life stage of the animals.

The high densities of caprellid amphipods on aquaculture installations and especially on buoys indicate that aquaculture activities might contribute to the dispersal of caprellids along the Chilean and also the Peruvian coast (Thiel et al. 2003; Chunga-Llauce et al. 2023b). In fact, these buoys frequently become detached and are often found floating in coastal waters (Astudillo et al. 2009). The fouling assemblages previously identified on these lost aquaculture buoys contained all caprellid species currently known for the coasts of the SE Pacific (Astudillo et al. 2009). Now C. mutica is also found on these highly buoyant substrata, which likely will facilitate its future establishment and spread.

High densities and species richness of caprellids were also found on boat hulls in Peru (Chunga-Llauce et al. 2023b), indicating that small boats also might contribute to the transport of caprellids and other species along the SE Pacific coast. The recent finding of Deutella venenosa, a species that previously had only been reported from Coquimbo (30°S) in Chile, from aquaculture structures and boat hulls in Peru (Chunga-Llauce et al. 2023a), indicates that these substrata contribute to the dispersal of caprellids. Rafting dispersal on detached aquaculture structures is also supported by another recent finding of D. venenosa on a rope stranded at Ritoque Beach at 33°S (Rech et al. 2023), which could also be expected for C. mutica in the future.

Recent records of Paracaprella pusilla from Mexico, Costa Rica and Peru (Alarcón-Ortega et al. 2015; Alfaro-Montoya and Ramírez Alvarado 2018; Chunga-Llauce et al. 2022), which had earlier been confirmed at multiple sites near the Pacific entrance of the Panama canal (Ros et al. 2014), suggest another ongoing caprellid expansion along the East Pacific coasts. While most of these findings were made on suspended aquaculture structures, all authors consider transport in/on ships as a more likely cause for the recent appearance of P. pusilla.

Several other NIS have recently been reported along the Chilean coasts, including the sea anemones Diadumene lineata (Verrill, 1869) (Häussermann et al. 2015), Metridium senile (Linnaeus, 1761) (Molinet et al. 2023), and the tunicate Asterocarpa humilis (Heller, 1878) (Pinochet et al. 2017). In many of these cases, dispersal on ship hulls is considered most likely (Pinochet et al. 2023). For several seaweeds, aquaculture activities and intentional introductions are considered likely causes for recent introductions or range expansions along the Chilean coast (Camus et al. 2022; Jofré Madariaga et al. 2023). Many NIS thrive on floating structures (including aquaculture floats and ship hulls), which facilitates their dispersal and establishment in harbors (Leclerc et al. 2020). All this suggests that shipping activity might have led to the initial introduction of C. mutica to the coast of Coquimbo, and that abundant floating structures have then allowed the establishment of a local population.

The authors have declared that no competing interests exist.

No ethical statement was reported.

Financial support for a research visit of JB to Coquimbo was provided by ANID-FONDECYT N°1190954 to MT. MMR was funded by ANID/Centros Regionales R20F0008 (CLAP), and ANID-FONDECYT N°1251475. JB was further financially supported by the German Federal Agency for Nature Conservation (grant no. 53202 and 3519532201, LABEL project). JB acknowledges support by the Open Access publication fund of the Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research.

Conceptualization: JB, MT. Data curation: JB. Formal analysis: MMR, JB. Funding acquisition: MT. Investigation: JB. Resources: MT. Visualization: JB. Writing - original draft: MT, JB, MMR. Writing - review and editing: JB, MMR, MT.

Jan Beermann https://orcid.org/0000-0001-5894-6817

Marcelo M. Rivadeneira https://orcid.org/0000-0002-1681-416X

Martin Thiel https://orcid.org/0000-0001-7535-3888

All of the data that support the findings of this study are available in the main text