Coastal urbanization is on the rise, with urban and periurban marine areas currently making up more than 1.5% of the global exclusive economic zones (Airoldi et al., 2021; Bugnot et al., 2021). Urbanization results in the permanent modification of abiotic and biotic conditions around artificial structures (Dafforn et al., 2015; Bugnot et al., 2021), often generating a variety of disturbance sources (Todd et al., 2019). Marine urban areas and other anthropic (i.e. manmade) habitats are in fact typically characterized by high levels of chemical pollution (Lam and Todd, 2013; Gauff et al., 2022), physical disturbance (Atalah et al., 2007), sediment load and turbidity (Airoldi, 2003), and increased temperature variability (Menniti et al., 2020) compared to non-urban areas. This set of specific conditions largely influences the distribution and functionality of organisms in these habitats.

The fouling communities colonizing the most urbanized artificial habitats differ from those of less impacted natural habitats, being characterized by distinct faunal and floral communities, lower species and genetic diversity (Fauvelot et al., 2009; Megina et al., 2016), altered ecological interactions (Klein et al., 2011) and functions (Dafforn et al., 2015) and a notorious prevalence of non-indigenous species (NIS) over native species (Mineur et al., 2012; Rondeau et al., 2022). NIS thrive in marine urban habitats due to their often-higher performance in stressful environments compared to native species (McMahon, 2002; Lenz et al., 2011). The underlying drivers of native and NIS distributions are, however, not yet fully understood (Airoldi et al., 2015; Gauff et al., 2022; Rondeau et al., 2022). Identifying the ecological forces and processes shaping marine biodiversity along gradients of urbanization is considered a priority for a better planning, design, and management of the most urbanized marine spaces, such as ports and city waterfronts (Airoldi et al., 2021), especially since NIS are considered a critical biosecurity issue (Pyšek et al., 2020), causing loss of biodiversity and ecosystem services (Walsh et al., 2016), species extinctions (Blackburn et al., 2019), and significant economic loss (Diagne et al., 2021).

Disturbance in marine urban spaces could be attenuated at the most flushed areas, where the hydrodynamic conditions create dilution effects (Schiff et al., 2007; Menniti et al., 2020). Flush attenuation induces disturbance gradients potentially affecting NIS prevalence (Floerl and Inglis, 2003; Gauff et al., 2022; Rondeau et al., 2022) and propagule pressure, which tends to be correlated to greater settlement and invasion success (Simberloff, 2009). Many marine taxa show active larval behaviors ranging from substrate preferences (Pinochet et al., 2020) to actively seeking out biofilm and/or conspecifics (Scheltema et al., 1981; Von Der Meden et al., 2010), which could vastly impact recruitment patterns. However, whether the distribution of NIS and native species along urban gradients is related to factors affecting the settlement or subsequent recruitment of propagules is not clear yet.

We took advantage of the urban gradients along the channel harbor of Ravenna, North Adriatic, Italy to explore whether the distribution of target adult NIS and native species colonizing artificial structures is related or not to settlement patterns (Walters, 1992). We quantified how the gradient of urbanization affected a range of relevant environmental parameters, and then compared the distributions of adult and settlers of two native and two functionally similar NIS of the same family of bivalves along that environmental gradient. Based on previous knowledge (see above), we expected higher abundances of the native species at the most flushed periurban entrance of the harbor canal and higher abundance of the two NIS at the most confined inner waterfront. We hypothesized that these gradients in adult distribution would be paralleled by settler abundances as environmental constraints should lead to preferential settlement in suitable habitat. Alternatively, a larger settlement potential compared to the final adult distribution would suggest that other post-settlement factors such as predation or competition (often overlooked in marine urban areas) could better explain the NIS adult distribution.

Environmental factors varied across sites and times ( Supplementary Figures 1–11 ). In general, the urban innermost site S1 was characterized by the highest temperature (1 to 3 degrees higher compared to the most external site S5, depending on the time of the year), salinity, and turbidity, which decreased moving towards the outermost periurban site, while nutrient levels tended to peak at the most industrial sites S3 and S4. When all the environmental data were combined, the PCAs for each time showed that overall, a temporally stable environmental gradient structure was present, with a high differentiation across sites from S1 to S5 ( Supplementary Figure 11 ). The PC1, based on the 10 environmental variables, explained between 31.2% and 50.2% of the data distribution, and the boxplot of the PC1 centroid value showed an ordered structure of the sites ( Figure 2 ). The logistic regression, testing for the effect of site on the PC1 centroid value, revealed a significant linear decrease (L: t-val. = -9.06; Pr (> |t|) < 0.001; Figure 2 ) with deceleration moving towards the outermost site (Q: t-val. = 2.714; Pr (> |t|) = 0.01; Figure 2 ). The overall model significance was high (Adj. R² = 0.7; F-value = 23.4; 35 DF; p < 0.001).

Oysters were sparse at all study sites, with O. edulis virtually absent in most samples, and M. gigas sparsely occurring at only few plots at sites S2, S3, and S4 ( Supplementary Figure 12 ). Like adult abundances, oyster settler abundances were unexpectedly low/absent based on molecular data. During the 7 sampling times, a total of only two settlers of M. gigas and no settlers of O. edulis were found on the ASUs, which prevented further analyses for these species.

The adults of the two mussel species (M. galloprovincialis and X. securis) significantly varied across sites, showing opposite patterns along the gradient (Adj. R² > 0.66; F-value > 15.27; 25 DF; p < 0.001; Figures 3A, B ; Table 1 ). A significant increase of the native M. galloprovincialis was observed moving from the innermost waterfront to the outermost periurban areas (L: t-val. = 9.5; Pr (> |t|) < 0.001; Figure 3A ; Table 1 ) accelerating (Q: t-val. = 6.47; Pr (> |t|) < 0.001; Figure 3A ; Table 1 ) towards the outermost periurban site. In this model, the cubic term was significant (C: t-val. = 3.26; Pr (> |t|) = 0.003; Figure 3 ; Table 1 ) indicating the presence of a threshold at which the acceleration starts centered between S3 and S4. For the non-indigenous X. securis, the distribution of adults was opposite to that of M. galloprovincialis, with a significant linear decrease from the innermost waterfront to the outermost periurban areas (L: t-val. = -7.2; Pr (> |t|) < 0.001; Figure 3B ; Table 1 ).

The distribution of M. galloprovincialis settlers over 7 sampling times paralleled the adult distribution, while there was a notable mismatch between adult and settler abundance for X. securis. Like for adults, M. galloprovincialis settlers had an increasing (L: t-val. = 4.78; Pr (> |t|) < 0.001; Figure 3C ; Table 1 ) and accelerating (Q: t-val. = 2.99; Pr (> |t|) = 0.003; Figure 3C ; Table 1 ) abundance towards the outermost periurban site. For X. securis, on the other hand, only the quadratic polynomial had significant effect (Q: t-val. = -2.83; Pr (> |t|) = 0.005; Figure 3D ; Table 1 ) reflecting maximum abundances in the middle portion of the gradient and very low abundances at the periurban site. This notably deviates from the adult abundances, which were highest in the two innermost sites. This mismatch is especially striking for the innermost waterfront site, which had the highest adult abundances, but second lowest settler abundances (after the periurban site).

In the present study, we investigated whether a gradient of increasing urbanization structured the distribution and abundances of adult populations of four target bivalves and their settlers. The periurban-urban gradient observed at the harbor of Ravenna was reflected in a temporally stable, significant gradient in relevant environmental parameters, including temperature, salinity, turbidity, and nutrient concentration, with consistent dissimilarities among the outermost and inner sites ( Figure 2 ). This occurred despite strong temporal fluctuations in the individual parameters ( Supplementary Figures 1–10 ). Only two of the four tested adult bivalve species were significantly influenced by this gradient, however, the distribution profiles align with our hypothesis: the NIS was more abundant in the inner parts of the canal, while the native species dominated the entrance. For settlers, only one species (M. galloprovincialis) exhibited a distribution profile that overlapped with adult distribution, indicating preferential settlement in habitats suitable for recruitment. Xenostrobus securis displayed a notable mismatch between adult and settler abundances, potentially linked to more indiscriminate recruitment owing to its resistance to disturbance.

Environmental conditions in the canal changed between March and July, as expected for both natural and urban areas (Lacroix et al., 2017). However, in accordance with hypotheses focusing on the importance of hydrodynamics and water exchange with the open sea (Floerl and Inglis, 2003; Schiff et al., 2007; Menniti et al., 2020), we observed a gradient structure in environmental conditions. This gradient was found to be temporally stable for individual parameters such as temperature (T S1 > T S5; Supplementary Figure 1 ), but also for the multivariate distribution of combined environmental conditions, with consistently high dissimilarity between S1 and S5 on the PC1 axis ( Figure 2 ; Supplementary Figure 11 ). This gradient structure was however not entirely linear, as a slight deceleration trend for differences between sites was observed between the middle and the outermost site ( Figure 2 ; Table 1 ), indicating that environmental conditions exhibit steeper spatial variations at the innermost part of the canal.

In Ravenna harbor, oysters showed surprisingly low abundance compared to similar marine urban areas in the northern Adriatic (see photographic comparison with an urban substrate from the Venice Lagoon: Supplementary Figure 12 ). In total, only 35 live adult oysters were sampled, with the majority (27) being M. gigas. Oyster settlers were virtually absent during the study, despite the study period coinciding with their spawning period (Bataller et al., 1999; Massapina et al., 1999). Both oyster species have been documented in estuaries, harbors, and marinas (Stagličić et al., 2020), including in the study area and nearby lagoons (Airoldi et al., 2016), and should be able to resist the environmental conditions in the area. Transient disturbances could have nevertheless led to mass mortality of oysters, possibly replaced by faster-growing mussels, X. securis, in the innermost sites, and M. galloprovincialis in the outermost sites respectively.

Expectedly, adult distributions reflected their status (NIS vs Native), with M. galloprovincialis (Native) increasing towards the periurban part of the canal (S5), and X. securis (NIS) decreasing. This dichotomous distribution aligns with observations made by Montes et al. (2021) in a Spanish harbor, where a combined effect of spatially organized features (i.e., environmental gradients) and seasonal dispersal barriers were hypothesized as underlying drivers. In our study, for M. galloprovincialis, a linear trend was accompanied by an acceleration setting at a threshold (Quadratic term; Table 1 ), centered between S3 and S4. Between S3 and the canal’s entrance, the abundance of adults increased at a high rate. The distribution pattern of its settlers closely mirrored this trend. Given that the length of the canal falls well within the typical dispersal distance of M. galloprovincialis (Gomes et al., 2016), this may indicate a preferential settlement as it tightly matched the adult distribution. Mytilus galloprovincialis adults and larvae are known to be controlled by their tolerance to environmental parameters, including temperature, salinity (Olabarria et al., 2016; Vasquez et al., 2022), and some sources of pollution (Serrano et al., 1995; Boukadida et al., 2016). Environmental conditions between S1 and S3 may not be favorable to M. galloprovincialis and may constitute a limiting factor for their settlement and survival. The distribution profile of M. galloprovincialis adults and settlers was similar to the PC1 value profile used as a proxy to outline the environmental gradient, with S4 and S5 markedly different from S3, S2 and S1 ( Table 1 ; Figure 1 ). It is likely that M. galloprovincialis distribution was mostly influenced by its inability to thrive in the environmental conditions of inner urban areas and its preference for more suitable environments for settlement.

Decreasing from the innermost waterfront to the periurban site, the abundance of X. securis adults was consistently high at the two innermost sites, while it was almost absent at the outermost ones ( Figure 2 ; Table 1 ). Unlike M. galloprovincialis, where the distribution of settlers closely mirrored that of the adults, X. securis displayed a significant discrepancy between the abundance of settlers and adults. Settlers of X. securis were consistently spread out in the canal, with the highest abundance found in the middle areas, where adult populations were already decreasing. The mismatch between adult and settler abundances could indicate that for this species, post-settlement processes could drive the adult population distribution patterns as settlers colonize a broader range than that to what the adult population was limited. One such post-settlement process could be predation from boring snails, crabs, and fish, which depending on the environmental conditions and characteristics of the mussel aggregates, can reduce X. securis survival (Gestoso et al., 2014), and consequently adult populations. Another post-settlement process affecting X. securis distribution could be competition with the native species M. galloprovincialis at the outermost part of the canal. Despite in aquaria experiments revealing a high resistance to a wide range of temperatures, pH, and salinities (Garci et al., 2007; Gestoso et al., 2016; Olabarria et al., 2016) in situ, X. securis seems restricted to the upper estuarine sites with low salinities (Garci et al., 2007; Montes et al., 2021). The general high resistance of NIS to multiple and variable environmental stressors (Lenz et al., 2011), which also applies to X. securis (Gestoso et al., 2016; Olabarria et al., 2016; Astudillo et al., 2017) could explain its high abundance at the inner canal sites. However, stress resistances come with costly metabolic trade-offs (Pook et al., 2009; Gauff et al., 2022). This metabolic challenge could lead to decreased competitivity in environments with lower disturbance and may explain why numerous NIS thrive in high disturbance habitats like anthropic habitats or estuaries but are far less common in nearby natural sites (Airoldi et al., 2015; Gauff, 2021). Other species could also compete with X. securis, but keystone bivalves were selected for this study because they are the most abundant species in the study area, thus it seems unlikely that other species compete too strongly with X. securis.

Overall, our data support the hypothesis that NIS are more prevalent in disturbed habitats (Piola and Johnston, 2008; Megina et al., 2016; Rondeau et al., 2022), potentially due to unfavorable post-settlement competitive matchups against similar native species in less disturbed environments (Montes et al., 2021; Gauff et al., 2022), and/or due to specialization into resistant life strategy traits implying certain metabolic trade-offs (Gauff, 2021; Gauff et al., 2022). Future work on larval stages and settlers of a greater range of species is worth pursuing to test the generality of this hypothesis and to identify the underlying drivers, by disentangling biotic interactions from abiotic parameters like temperature, salinity, and chlorophyl-a, which considerably varied in our experiment and could have impacted both species (Gestoso et al., 2014; Olabarria et al., 2016; Montes et al., 2021). This should include tests on competition between native and NIS with similar niches, as well as other post-settlement processes like predation or seasonal post-settlement mortality (Montes et al., 2021), which could also lead to the discrepancies observed here.

The datasets are available within the Supplementary Material and scripts for this study will be provided upon reasonable request.

Ethical approval was not required for the study involving animals in accordance with the local legislation and institutional requirements because not needed for these specimens.

RG: Data curation, Formal Analysis, Visualization, Writing – original draft, Writing – review & editing. FM: Data curation, Investigation, Methodology, Writing – original draft, Writing – review & editing. PM: Conceptualization, Methodology, Supervision, Writing – original draft, Writing – review & editing. FP: Conceptualization, Methodology, Writing – review & editing. FC: Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing. LA: Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing.