Focal species selection

We selected 20 invasive forest wood-boring pests as focal species for this study. These belonged to Buprestidae, Cerambycidae, Curculionidae, and Siricidae. The selected species represent a wide range of public detectability, from small and cryptic bark or ambrosia beetles to larger and more noticeable longhorn beetles and buprestids. To confirm their status as invasive forest wood-boring pests, we checked four evidence sources: EPPO Global Database, CABI Compendium, Global Invasive Species Database (GISD), and scientific publications. A species was defined as invasive when at least one authoritative source provided evidence of establishment or invasion (Table 1). From the species that met this invasion criterion, we further applied three inclusion rules: (1) association with forest trees or woody plants (i.e., wood-boring habit); (2) availability of verifiable official first records at the country level; and (3) presence of occurrence records in GBIF (Global Biodiversity Information Facility) that were from a recognised citizen science platform. The resulting dataset therefore represents a set of focal, well-documented invasive forest wood-boring pests, rather than a complete global inventory of this group.

We compiled 129 official first species–country records for 20 invasive forest wood-boring pests across 49 countries (Suppl. material 1: table SS1). After strict filtering of GBIF source fields, we retained 14,889 citizen science occurrence records from 70 countries and areas. These records yielded 234 species–country combinations with citizen science first records. For the paired time-lag analysis, only species-country combinations with both an official first record and a citizen science first record were retained. This produced 79 valid record pairs involving 17 species and 38 countries. A. mali, A. bilineatus, and I. grandicollis had no matched citizen science first records (Fig. 1).

Across the 79 matched pairs, citizen science first records were predominantly later than official first records (Suppl. material 1: fig. S1). Citizen science first records preceded official records in only 2 cases (2.5%), occurred in the same year in 3 cases (3.8%), and followed official records in 74 cases (93.7%). The overall time lag was strongly positive, with a mean of 24.82 ± 3.24 years (SE) and a median of 15 years. This difference gradually decreased when the analysis was restricted to time closer to the widespread use of citizen science platforms. The mean time lag decreased to 8.47 years in the post-2000 subset (Suppl. material 1: fig. S2) and 5.18 years in the post-2010 subset (Fig. 2a) (Table 2). However, positive time lags remained dominant across all temporal subsets, even after restricting the analysis to the post-2000 and post-2010 periods. This temporal pattern coincided with the annual trend in the filtered citizen-science GBIF records (Fig. 2b), which increased markedly after 2010 and especially in recent years. Annual mean ΔT_lag also declined over time and approached zero after 2010, with some negative values in recent years (Fig. 2c), suggesting that the reduction in ΔT_lag partly reflects the increasing availability of citizen science data.

Beyond the three temporal periods shown in Fig. 2, we further examined ΔT_lag as a continuous function of Y_official across all 79 species–country pairs (Fig. 3). This continuous analysis revealed a strong, significant declining trend (Spearman’s ρ = −0.84, p < 0.001; Theil-Sen robust slope = −0.92 years per year, 95% CI: −4.52 to 2.76). This trend remained significant after accounting for non-independence among repeated species using cluster-robust standard errors (slope = −0.93, cluster-robust SE = 0.048, p < 0.001, clustered by 17 species).

We explored species traits that could influence citizen science recording intensity. Recording intensity varied strongly among species, ranging from 15 records in X. glabratus to 4,174 records in A. planipennis (Suppl. material 1: fig. S3). Adult body size was positively associated with citizen science recording intensity, with larger species tending to have more citizen science records (Spearman’s ρ = 0.52, p = 0.022; Table 3). However, adult body size class and adult body coloration were not significantly associated with recording intensity (p = 0.085 and p = 0.051, Table 3). In contrast, adult detectability, a composite trait combining adult body size and adult body coloration, was significantly associated with recording intensity, with high-detectability species having more citizen science records than low-detectability species (Mann–Whitney U = 72.0, p = 0.010; Table 3). Research attention was not associated with citizen science recording intensity, and public attention showed only a non-significant positive trend (p = 0.689, p = 0.085, Table 3). Family-level differences were also detected overall (Kruskal–Wallis H = 6.219, p = 0.045; Table 3), but this pattern was not emphasized because pairwise differences were not significant after Holm correction (Suppl. material 1: fig. S4). Candidate model comparison further identified adult detectability as the best-supported single predictor of citizen science recording intensity (ΔAICc = 0) (Suppl. material 1: fig. S5). In addition, we also found that citizen science records were unevenly distributed among countries, with the largest numbers recorded in the United States (3,969), Sweden (2,656), and China (1,540), whereas substantially fewer records were available from many other countries (Suppl. material 1: fig. S6).

Neither range size (ρ = 0.350, p = 0.142) nor invasion history length (ρ = −0.035, p = 0.886) was significantly associated with citizen science recording intensity (Table 4), indicating that the observed association between adult detectability and recording intensity is unlikely to be confounded by these two factors.

At the municipal scale in China, we collected 36 official species-city records of invasive forest wood-boring pests across 11 provinces (Suppl. material 1: table S7). These records included D. valens (16), S. noctilio (13), H. ligniperda (6), and P. semipunctata (1). In contrast, strict filtering of GBIF citizen science records retained 1,318 occurrences from nine focal forest wood-boring pests in China, including 5 native and 4 non-native species. These records were highly uneven among species and were dominated by three conspicuous longhorn beetles: A. chinensis (614 records, 46.6%), A. bungii (371 records, 28.1%), and A. glabripennis (252 records, 19.1%), which together accounted for 93.9% of all retained China citizen science records. Spatially, these records were concentrated in eastern China, including Beijing, Zhejiang, Shandong, Shanghai, and Jiangsu, which together accounted for 757 records, representing 57.5% of the records with province-level information. In contrast, many inland or northwestern regions had few records. However, of the four invasive species documented by official records in China (Fig. 4), only P. semipunctata (five records) and D. valens (one record) were represented in citizen science data, with five and one record, respectively. No citizen science records were available for H. ligniperda or S. noctilio. Notably, the citizen science first record of P. semipunctata was from 2022, preceding its official first record in 2024, and records were found in both Guangdong and Yunnan, whereas our compiled official records documented the species only from Guangdong.

We analyzed Coleoptera citizen science records in China from 2010 to 2025. Annual Coleoptera citizen science records increased markedly, especially after 2021 (Fig. 5a). Spatially, Coleoptera records were highly uneven across China, with records concentrated mainly in eastern coastal provinces and parts of southwestern China, whereas northwestern China had relatively few records (Fig. 5b, Table 5). We further compared province-level Coleoptera records with cumulative HS 4403 wood-in-the-rough import value during 2015–2025 and found a significant positive correlation (Spearman’s ρ = 0.62, p < 0.001, n = 30), suggesting partial spatial overlap between Coleoptera citizen science records and log-import trade value. However, Moran’s I tests indicated significant residual spatial autocorrelation in this relationship (Moran’s I = 0.24, p = 0.007), suggesting that the conventional p-value may be anti-conservative due to spatial clustering among neighboring provinces. Their spatial patterns were not fully matched. Yunnan had high Coleoptera records but relatively low HS 4403 import value. Conversely, Heilongjiang, Tianjin, Chongqing, and Inner Mongolia had relatively high HS 4403 import value but lower Coleoptera recording effort. These results suggest that eastern coastal provinces and major import-associated regions remain important areas requiring continued biosecurity attention for invasive forest wood-boring pests.

Following the research approach of González‐Moreno et al. (2025), we treated the time lag as the difference between the official first records and the citizen science first records at the country level. This metric should be interpreted as a country-level evidence lag rather than as a true lag from actual introduction, as a species may already have been present and spreading undetected before either type of record was made. Our results indicate that, at the national scale, citizen science records of invasive forest wood-boring pests did not systematically precede official first records, and only in a few cases did they detect species earlier than official monitoring. Across the 79 matched species-country pairs, citizen science first records preceded official first records in only two cases. Although the mean time lag decreased in the post-2000 and post-2010 subsets, positive lags remained dominant. Therefore, for forest wood-boring pests, the primary role of citizen science is unlikely to be early warning; rather, it is more likely to function as a spatial complement to existing monitoring systems, thereby broadening our understanding of their distribution patterns. For example, after X. crassiusculus was first detected in New Zealand through routine surveillance, Biosecurity New Zealand asked the public to report possible sightings and signs to help assess whether a wider population was present (Biosecurity New Zealand 2019; Ministry for Primary Industries 2019). However, the few negative-lag cases should nevertheless not be overlooked. Citizen science first records preceded official first records for P. semipunctata in China and T. campestris in Hungary. In China, citizen science records of P. semipunctata further supplemented official records by adding Yunnan to the known occurrence information. This suggests that citizen science records can sometimes supplement spatial information. Nevertheless, because our analysis was restricted to GBIF-indexed citizen science platforms, it did not include non-platform public reports submitted directly to biosecurity agencies, diagnostic services, or local reporting schemes. Such records may also contribute to general surveillance and biosecurity responses but are often not represented in large biodiversity platforms (Pawson et al. 2020). The continuous decline in time lag with more recent official records (Fig. 3) likely reflects the growing adoption of citizen science platforms since around 2010, rather than a fixed or steadily improving detection capacity; we therefore avoid extrapolating this trend beyond the observed data.

Our results showed that citizen science first records generally lagged behind official first records. Many invasive forest wood-boring pests have cryptic life stages, with larvae developing within the xylem or phloem, and adults often being small, seasonal, or short-lived (Avtzis and Lakatos 2021). These traits reduce the likelihood that such species are encountered, photographed, and uploaded compared with more conspicuous foliar-feeding pests or flower-visiting insects. In addition, this group includes many closely related species (Justesen et al. 2023), and formal morphological identification may require subtle diagnostic characters, specialist knowledge, or microscopic examination (Armendáriz-Toledano and Zúniga 2017). However, identification difficulty should not be interpreted as an absolute limitation of citizen science. Highly engaged amateur naturalists and expert contributors can detect and recognize cryptic or unusual species, particularly when they have strong local knowledge of the fauna. Our analysis of species traits supports this interpretation. Adult detectability, a composite variable combining adult body size and adult body coloration, was significantly associated with citizen science recording intensity (p = 0.010). This association was unlikely to be driven by species range size or invasion history length, as neither factor was significantly associated with recording intensity (Table 4), supporting a genuine link between adult detectability and citizen science engagement. In contrast, research attention (p = 0.689) and public attention (p = 0.085, positive trend) were not associated with citizen science recording intensity. This result complements the findings of González‐Moreno et al. (2025), who showed that species traits were strongly associated with the time lag between official first records and citizen science first records. While their time lag metric captures how quickly a species is first detected by the public, our recording intensity metric captures the overall volume of citizen science engagement with a species over time. Nevertheless, both metrics appear sensitive to species traits such as detectability, suggesting that trait-driven biases may act across multiple stages of the citizen science process, from initial detection to sustained recording effort. However, our study shows that even within invasive forest wood-boring pests, detectability differences can strongly constrain citizen science recording.

We also considered urban pest status and host tree symptom visibility as potential confounding factors, but found no consistent basis for classifying species this way: even cryptic species were sometimes recorded in urban settings, and symptom-based detection typically requires expert knowledge that most citizen scientists lack. We therefore consider direct visual detectability of the insect itself the more relevant factor for citizen science records specifically.

Citizen science records showed marked spatial heterogeneity at the national scale. In this study, citizen science records were highly concentrated in only a few countries, whereas many others had very few or no records (Suppl. material 1: fig. S6). This pattern indicates that citizen science records do not directly reflect the true occurrence and spread of these pests, but instead largely capture differences in public participation, platform activity, and monitoring contexts among countries (Arazy and Malkinson 2021; Pocock et al. 2024). Consequently, areas lacking records cannot be simply interpreted as areas not yet invaded or as supporting only small populations. Likewise, large numbers of records in some countries do not necessarily imply higher levels of pest occurrence.

Rather, these spatial differences more likely reflect variation in the socio-technical coverage of observation systems: countries with dense records often have more mature citizen science platforms, more active user communities, and greater public attention to the focal taxa, whereas record-poor regions may reflect weaker participation, lower platform use, or poorer data integration (Beck et al. 2014; Arazy and Malkinson 2021). Thus, interpreting citizen science records requires careful distinction between biological absence and observational absence (Johnston et al. 2023).

Accordingly, for forest wood-boring pests closely associated with forestry production and quarantine regulation, the value of citizen science lies less in early warning than in complementing official monitoring through broader spatial coverage and continued tracking of established populations. Its contribution should therefore be understood not primarily in terms of earlier detection, but in terms of spatial supplementation and spread tracking (De Groot et al. 2023; Pocock et al. 2024). For this group, citizen science may be particularly useful for extending surveillance beyond the long-term reach of official monitoring and for supplementing occurrence information in urban and residential environments.

Nevertheless, citizen science may still capture risk signals earlier for some taxa. For example, in invasive mosquito surveillance, the Mosquito Alert system, which combines public reporting with expert validation, has contributed to the first detection of invasive mosquito species at the municipal scale and supported early warning and rapid response (Palmer et al. 2017). This suggests that citizen science can indeed advance detection timing under appropriate conditions. However, such temporal advantages are not universal and are more likely to emerge for organisms that are frequently encountered by the public, easy to photograph, and relatively easy to recognize (González‐Moreno et al. 2025). These contrasting outcomes further highlight that the effectiveness of citizen science depends on the interaction between species traits, data biases, and public awareness or familiarity with the species.

Several limitations of this study should also be acknowledged, which can be grouped into two broad categories: limitations related to data coverage, and limitations related to using GBIF-indexed records as a proxy for citizen science contributions.

Regarding data coverage, citizen science records were sourced exclusively from GBIF records and excluded other citizen science platforms, such as local platforms, social media, or direct reports to biosecurity agencies. In addition, platform use differs markedly among countries, meaning that our data may overrepresent regions with active GBIF participation and underrepresent others.

Regarding the use of GBIF as a proxy for citizen science contributions, our framework has two related shortcomings. First, our species selection relies on species already indexed in GBIF, which inherently favors relatively detectable species; highly cryptic species that are rarely reported through any monitoring channel, including GBIF, are likely underrepresented, meaning the overall contribution of citizen science to wood-borer detection may be even more limited than observed within this subset. Second, equating citizen science contributions with GBIF-indexed records likely underrepresents the true extent of public involvement in pest detection more broadly: many detections originate from members of the public noticing symptomatic or declining trees and reporting them directly to forestry or biosecurity authorities, after which the causal pest is formally identified and the case is registered as an official record rather than a citizen science record. Such community-initiated detections represent a genuine form of citizen participation but are effectively absorbed into official records rather than captured in databases such as GBIF. As a result, the relative contribution of public participation to pest detection, as measured in this study, is likely a conservative estimate of its true role.

In addition, record absence should not be interpreted as biological absence, and record numbers should not be treated as direct measures of abundance or invasion extent. Therefore, our conclusions primarily apply to GBIF-indexed citizen science records and should not be interpreted as representing all forms of public participation in surveillance.