Sampling was conducted in the upper catchment of the Gellibrand River (Figure 1), the whole catchment being 1,200 km². Grab composite samples were collected (n = 10), and passive samplers were deployed (n = 10), from five locations across four water bodies: the Gellibrand River (GV, R), Loves Creek (M), Charleys Creek (CC) and Boggy Creek (BC) (Figure 2). Gellibrand River was sampled at Valley Road (GV) near small family farms and livestock farming and further downstream (R) in the vicinity of livestock farms. Boggy Creek (BC) sampling point was in the vicinity of softwood (Pinus radiata) and hardwood (Eucalyptus nitens) plantations, native forests, small-scale agricultural, and livestock activities. Loves Creek (M) sampling site was located within a livestock farm, while Charleys Creek (CC) was sampled in a location surrounded by hardwood plantations. Coordinates of the sampling locations are provided in Supplementary Table S1 (Supplementary Material).

At the beginning of the first campaign, participants divided responsibilities among each other for the different sites, so that the same family, or group, collected samples at the same location throughout the seasons. Three monitoring campaigns were conducted within the Gellibrand River catchment in March (Campaign 1), April (Campaign 2) 2024 and November (Campaign 3) 2024. Campaign 1 was conducted during very dry weather (8 mm of rain recorded in March); Campaign 2 and 3 occurred during wet weather with 53.9 mm recorded in April and 60.4 mm recorded in November 2024. Rainfall information was obtained for the nearby measurement station of Irrewillipe (38.45° S, 143.66° E) (Australian Government Bureau of Meteorology, 2025).

During these campaigns, four different sampling methods were implemented (Figure 2). Chemcatcher with SDB-XC disk and Polar Organic Chemical Integrative Sampler (POCIS) passive samplers were deployed over a month in duplicates. 300 mL water samples were collected weekly over the deployment period of the passive samplers to make up a 4-week composite sample of 1 L, to be extracted by Solid Phase Extraction (SPE), and 100 mL, to be extracted by Stir Bar Sorptive Extraction. Stir Bar extracted samples were collected in duplicates for a total of 10 samples, while 1 L SPE samples were collected singularly, as control samples, for a total of 5 samples. It should be noted that, during the second campaign, due to misunderstanding in the collection of grab composite samples’ volumes, 1 L samples to be extracted by SPE were not collected.

A total of ten families were engaged in the water quality monitoring initiative. Most participants were either part of LAWROC, or partners or family members of someone directly involved. Almost all participants were retirees, with few of them having a background in natural sciences (i.e., ecology, chemistry). Interest to participate in a water quality monitoring activity was initiated by LAWROC, due to concern raised by the vicinity of forestry plantations to the catchment of the Gellibrand River and a decline of endemic blackfish populations. After a couple of online meetings, an in-person meeting occurred to jointly define the sampling locations and initiate the first campaign of sampling. Citizen scientists provided land use maps of the area, which aided in the selection of the sampling locations, with the choice driven by presence of forestry, horticultural, livestock activities, and National Parks. Instructions on how to undertake sampling and extraction were provided in written and visual format to facilitate access and understanding (Cacciatori et al., 2023). During the first sampling campaign, a PhD student, together with the staff from AQUEST at RMIT University, trained participants in the deployment and retrieval of passive samplers, in the collection of composite samples, sample extraction and storage. The next two campaigns were conducted by LAWROC members (Figure 3) with sampling materials being delivered to them by courier.

To collect feedback on the engagement of the citizen scientists, a list of survey questions was shared with the group, while additional feedback was collected informally throughout the process. The feedback was instrumental to understand the suitability of protocols and instruction material, the ease in deploying the different sampling and extraction techniques, and whether any improvement could be implemented along the project rollout (i.e., communication strategy, logistics). The survey’s questions are reported in Supplementary Material S2.

The preparation of passive samplers was performed at RMIT AQUEST, where also bottles for grab sampling were pre-cleaned and labelled. Preparation of passive samplers followed protocols by Pettigrove et al. (2023), and further described in Cacciatori et al. (2026). In short, passive samplers’ casings, polyether sulfone (PES) protective membranes and absorbent SDB-XC disks (Chemcatcher) were soaked in methanol and Milli-Q water for cleansing and chemical activation. For grab samples, pre-cleaning of new bottles was performed by acetone rinsing (3x), while used bottles were rinsed with acetone (3x), dish washed, autoclaved and rinsed again with acetone (3x).

LAWROC citizen scientists were involved in deployment and retrieval of passive samplers, as well as in the collection of grab samples and extraction using SBSE. The passive samplers were tied using rope to stable structures such as tree trunks or bridges and from a star picket. During the first campaign, passive samplers and deployment stainless steel casings were provided separately, yet to be assembled. This created some difficulties for the citizen scientists, as attaching the passive samplers onto the stainless-steel cases is a rather meticulous step. Therefore, during the other two campaigns, these samplers were provided already attached on the cases, facilitating deployment greatly. LAWROC citizen scientists were provided with 500 mL, 5 L bottles. 500 mL bottles were used to collect grab samples each week for a total of 300 mL, indicated with a line. 5 L bottles were used for storage of the total sample amount (1.5L: 5 samples x 300 mL), which was eventually separated into 1 L and 100 mL for SPE and SBSE extraction.

All samples processing procedures were conducted at RMIT Laboratory, except for SBSE extraction, which was performed on 100 mL composite grab samples by LAWROC’s members. Procedures of samples processing differed depending on samplers. Before elution, POCIS Hydrophilic Lipophilic Balance (HLB) absorbent material was transferred on an empty SPE cartridge, while Chemcatcher SDB disks were dried on a hot plate for 1–1.5 h at 35 °C. 1 L grab composite samples were filtered and extracted on a HLB cartridge according to the procedure described in (Cacciatori et al., 2025c). Extraction by SBSE consists of adding a stir bar to the water samples and letting it stir on a magnetic plate for 5 h at room temperature. Details of this extraction method are provided in Cacciatori et al. (2025c). After extraction, stir bars, dried SDB disks and HLB cartridges (POCIS and 1 L composite samples) were shipped via courier (DHL P/L) to the JRC Water Laboratory for analysis. All material used for sampling is thoroughly described in Cacciatori et al. (2026).

Quality Assurance and Control was guaranteed deploying field and laboratory blanks for passive samplers’ deployment and retrieval and preparation and processing in the laboratory, respectively. For composite grab samples, blank samples were processed alongside real samples. Furthermore, internal standard solution was added after elution of passive samplers and SPE extracted grab samples, and before SBSE extraction. The internal standard solution consisted of trans-Nonachlor (¹³C₁₀, 98%, Cambridge Isotope Laboratories Inc., United States, quantifier ion: 418.8182; qualifier ion: 420.8154), diluted in acetone (1 mL), for a final concentration of 0.5 ng/L for the samples processed with SBSE and 5 ng/L for the water samples extracted with SPE and passive samplers.

Samples were target screened for 447 pesticides using a Gas Chromatograph 8,890 with Quadruple Time-Of-Flight 7,250 (Agilent Technology, United States). All chemicals were separated and analyzed on HP-5MS UI capillary column (length = 30 m, internal diameter = 0.25 mm, film thickness = 0.25 μm, Agilent Technology, Santa Clara, CA, United States). A thorough description of the analysis is reported in Cacciatori et al. (2026), while GC-QToF-HRMS method parameters are reported in Supplementary Table S3 (Supplementary Material).

Passive samplers provided qualitative information on the presence or absence of pesticides in the water, while grab composite sampling with SPE and SBSE allowed for the determination of their concentrations expressed in ng/L. Analysis and quantification of samples and calculation of the calibration curve was based on computation for 4 calibration levels (L1: 10 ng/L, L2: 25 ng/L, L3: 50 ng/L, L4: 100 ng/L). Calibration curve L1 was injected (SPE-extracted grab-samples) or desorbed (SBSE-extracted grab samples) 6 times to determine Limit of Detection (LOD) following Cacciatori et al. (2025c). Supplementary Table S4 of the Supplementary Material reports the list of all screened compounds, with their categories, and LODs for SPE and SBSE extracted samples. For compounds detected above 10 ng/L, Limit of Detection (LOD) were estimated using the signal-to-noise ratio method, with LOD defined as three times the baseline noise, respectively, following the approach described by Ravisankar et al., (2021).

A standard solution was not available for 12 compounds, namely 2,4,6-tribromoanisole, 2-chloronaphthalene, 2-phenylphenol, 4,4′-methylenebis(N,N-dimethylaniline), acetophenone, dichlorbenzamide, chlorocresol, epoxiconazole, isoxadifen, phthalimide, quinoline, resorcinol. They were excluded from quantitative analysis and are indicated with suspect target screening in Table S. Where possible, concentrations were compared to existing guideline values from the Australian and New Zealand Guidelines for Fresh and Marine Water Quality (ANZECC, 2020). The 99% species protection Default Guideline Value has been reported in the text when available, considering the vicinity of some of the sampling locations to National Parks. Acquisition data were analyzed with Quantitative Agilent Software (version 12.1).

An ecological risk assessment was conducted on a subset of the detected pesticides depending on the availability of data on the IUPAC Pesticide Properties Database (PPDB) (Agriculture & Environment Research Unit AERU, 2018). The toxic effects were calculated as acute risks to the freshwater crustacean Daphnia magna using Toxic Units (TU). TU are calculated by dividing the measured concentrations of each pesticide by the concentration that causes a 50% effect on a defined biological endpoint (Equation 1), indicated by the acute half-maximal effective concentration (EC50), or the median lethal concentration (LC50) (EFSA et al., 2019; SCHER and SCCS, 2011). Toxic units obtained for each compound were summed up for each sample, assuming concentration addition (EFSA et al., 2019) to provide information on the cumulative eco-toxicological risks at each sampling locations. EC(LC)50 values used for the calculations are reported in Supplementary Table S6 of the Supplementary Material. T U m i x t u r e = ∑ i n C i E C L C 50 i (1)

This approach allows for the comparison of the effects of individual chemicals or their addition to an overall potential effect. Values exceeding 0.01 indicate potential toxicity to D. magna (EFSA Panel on Plant Protection Products and their Residues PPR, 2013). Only those pesticides with available toxicity data for D. magna were included in this assessment.

During the first monitoring campaign, 18 pesticides or their metabolites of the 447 screened were detected (Table 1). The second and third monitoring campaigns yielded 8 and 15 pesticides respectively. Of all detected compounds, nine pesticides were found exclusively in Campaign 1, including six fungicides (azoxystrobin, boscalid, difenoconazole, fluquinconazole, imazalil, mefenoxam), two insecticides (deltamethrin, indoxacarb), and one metabolite (atrazine-desisopropyl). In contrast, only the insecticide carbaryl was detected uniquely in Campaign 2. During Campaign 3, 10 compounds were detected that had not previously been detected in previous campaigns. These pesticides were mainly fungicides and including fenhexamid, penconazole, propiconazole and tebuconazole.

The detected pesticides fall into several categories: fungicides, insecticides, herbicides, and other types such as intermediates, breakdown products and metabolites (Table 1). Throughout the three campaigns, most pesticides detected belonged to the class of fungicides, while most pesticides screened belonged to the class of insecticides (n = 176). Fungicides have been mentioned as an often overlooked pesticides class during surface water monitoring, although they are widely and intensively applied both in agricultural and urban environments (Zubrod et al., 2019).

There was not necessarily concordance in terms of individual pesticides detected in the three different monitoring campaigns. Different uses, point sources application and rainfall patterns throughout the year might explain these variations. Trace concentrations of some legacy pesticides might be linked to groundwater recharge because of low rainfall during March 2024. During Campaign 2 spikes in concentrations of carbaryl, piperonyl butoxide and biphenyl, which were not detected during the other two exercises, might hint to recent application in nearby orchards and farms. Combined use of carbaryl with synergist piperonyl butoxide has been mentioned to enhance insecticide’s efficacy (US EPA, 2021). Indeed, the area is characterized by orchards and horticulture, including blueberries, for which carbaryl is registered for use in Australia (APVMA, 2024). The second campaign occurred in April, beginning of the autumn season in Australia. According to literature, pesticides use on blueberries is greatest during the wet season (summer) when summer crops are in fruit, although some blueberry farms produce winter crops which may be sprayed to protect the fruit (Laicher et al., 2022).

Other pesticides have been detected consistently throughout the three campaigns, although at concentration <100 ng/L, except for biphenyl spike in Campaign 2. Those pesticides are wood preservative 2,4,6-TCP, fungicide biphenyl, herbicide hexazinone, simazine and terbacil. Except biphenyl, which frequent detection can be explained with its wide natural occurrence, all other compounds can be linked to forestry applications. According to survey results published by the Australian Government (Jenkin and Tomkins, 2006), at the time of publishing, hexazinone covered 19.8%, simazine 13.1% and terbacil 1.9% of active ingredients’ use in forestry in the country. A traditional cycle of pesticides spraying in forestry includes using herbicides for site clean-up from previous land use, pre-plant cleanup, application during the first 2 years for post-planting weed control (Jenkin and Tomkins, 2006). In a similarly old research looking at pesticides contamination in waterways near forestry in Victoria, Wightwick and Allinson (2007) detected 2,4,5-TCP and hexazinone nearby areas of aerial spraying, while a study in Tasmania detected 2,4-D, MCPA, hexazinone and simazine (WIST, 2014).

Grab samples coupled with SPE detected a total of 19 different pesticides, while SBSE extracted 15. Both sampling and extraction techniques yielded 11 unique pesticides, not recovered by the other methods (Figure 8). Therefore, the two techniques appear complementary. Passive samplers, POCIS and Chemcatcher, detected respective 9 and 6 different compounds, with POCIS detecting only 1 unique compound, while Chemcatcher none. The similarity in absorbent material (HLB, SDB) and target polarity range of POCIS, Chemcatcher and SPE sorbent might explain the overlap in detected compounds. During the second campaign when no SPE was performed on grab samples, therefore, POCIS guaranteed detection of a wider range of compounds, than if SBSE would have been deployed alone. Certainly the use of multiple techniques during the campaigns expanded the type and number of compounds detected (Cacciatori et al., 2026).

The survey on the participation in the initiative received one collective response by the group. Although the overall level of satisfaction was rated with the highest score, some issues and improvements have been identified. In particular, the respondents hinted at how the online details were not necessarily consistent with the apparatus set up in the field. The participants particularly valued the in-person demonstration which facilitated the understanding of the protocols and clarified potential disagreement between written material and actual set up. The participants highlighted, through informal feedback, the difficulties in assembling the passive samplers in the first campaign, a preparation step which was modified for the two following campaigns to provide already set up samplers. As general feedback on the different sampling and extraction techniques, participants found passive samplers initially as the hardest, because of the handiness required during deployment and retrieval, and later “less engaging”, due to the limited participation envisioned. Collection of composite samples was not deemed particularly challenging, although it created misunderstanding in the second campaign regarding the volume to be sampled. SBSE was considered exciting because of the extraction procedure, which allowed for a more analytical experience. As lessons learnt, participants voiced the importance of understanding and slowly following each part of the process, for which continuous training, even when deemed unnecessary, remains fundamental. Finally, participants said they were satisfied with the project, although it did not necessarily answer the questions they had on the impact of forestry on declining populations of blackfish. Relevant was the possibility to share this experience in a team with common goal and enthusiasm and, to socialize with members of the same community.

Local citizen engagement was instrumental in the collection of pesticides monitoring data in the Gellibrand catchment and in the assessment of applicability of different sampling and extraction techniques. Although results did not necessarily respond to the expectations of participants on the link between pesticides use for forestry and a decreasing endemic blackfish population, they contributed to increasing the knowledge on the water quality status of local water bodies, sharing information on the types of pesticides used and found in the area, while facilitating the communication with researchers and scientists, which might be seen with reservations and doubts. The implemented techniques resulted not only to be complementary in terms of pesticides detected, but also in terms of applicability in citizen science initiatives. The choice should therefore be based on the time volunteers can invest (i.e., once a month, every week, …), the information to be obtained (quantitative or qualitative), the financial and human resources available.