Mosquito-borne diseases pose a significant and growing threat to human and animal health worldwide. Two major factors contributing to the spread of exotic mosquito species into new regions are human-mediated dispersal and climate change (Weinstein et al. 1997; Lounibos 2002); Semenza and Suk 2017. Aotearoa New Zealand benefits from its geographic isolation, which helps prevent and manage potential mosquito incursions. However, this same isolation also means that its endemic fauna (mainly bird species), with their immunological naivety, are particularly vulnerable to exotic diseases (Derraik and Calisher 2004; Kramer et al. 2011). As a result, managing the risk of such incursions is especially critical. Comprehensive, long-term data collection is essential for understanding the distribution and phenology of both invasive and native species that serve as disease vectors. Such data are crucial for mitigating negative impacts on human and animal health and for predicting the potential spread of key vector species.

Global mosquito surveillance programs are inherently constrained by jurisdictional boundaries and financial limitations, often requiring the use of modeling and expert intuition to guide sampling or trapping efforts (Weidong et al. 2008; Moise et al. 2020). These challenges are further compounded by a changing climate, as well as the rapid expansion of global freight, transportation, and human mobility—all of which are reshaping previously established species distributions worldwide (Reiter 2001; Becker 2008). These confounding variables can be difficult to incorporate into models, especially in the absence of current, geo-temporally diverse data points. In Aotearoa, sampling and monitoring efforts primarily focus on exotic and introduced species, leaving significant knowledge gaps regarding native mosquito populations and their distribution. Meanwhile, exotic species continue to be introduced and are increasingly thriving in new areas (Ammar et al. 2019). The ranges of various introduced species are also shifting—expanding and contracting in response to factors such as land-use changes and climate variation (Kasper et al. 2023).

Mosquito research in the context of medical entomology and public health has been evident since the early 20^th century, marked by milestones such as the detection of two introduced species (Culex quinquefasciatus and Aedes notoscriptus), the formation of a mosquito control committee in the late 1920s, and the reinforcement of control measures during and after WWII, when New Zealand’s army, navy, and air force became heavily involved. Interest in mosquito monitoring increased in the 1970s with the rise of international air travel and later with the global trade in used tires. The number of mosquito interceptions began to grow in the 1990s, all effectively prevented from establishing (Laird 1996), until citizen reports of nuisance biting prompted a response from the Ministry for Primary Industries (MPI, then MAF-NZ), leading to the detection of the Australian southern saltmarsh mosquito, Aedes camptorhynchus, in Napier in December 1998.

Over the following 11 years, eradication efforts produced substantial surveillance data, leading to the transfer of many monitoring responsibilities to the Ministry of Health (MOH) and its contractors (Kay and Russell 2013) and ultimately to the establishment of a governed database. The data collected during this period improved our collective understanding of population dynamics for both endemic and introduced mosquito species and has since played a significant role in guiding subsequent sampling strategies (Ritchie and Russell 2002).

The New Zealand BioSecure Entomology Laboratory (NZBEL) specializes in entomology, border health, biosecurity, and vector control. In addition to providing entomological identification and recommendations to the Ministry of Health (MOH), NZBEL manages the Online National Mosquito Surveillance Database, which compiles all MOH mosquito sampling results and official public complaints. The NZBEL webpage provides links to specific datasets and information on mosquitoes. However, despite considerable efforts to improve sampling rates, data collections remain fragmented across a hierarchy of organizations. To help reduce the impact of this, NZBEL staff also maintain a close working relationship with Te Papa Tongarewa regarding mosquito research.

Existing surveillance programs in Aotearoa, com­missioned by MPI and MOH, primarily focus on the early detection (interception) of invasive mosquito species at international airports and seaports—so-called points of entry (POEs). The combined monitoring efforts of regional public health units and border workers, together with the National Saltmarsh Mosquito Surveillance Program—often now referred to as “the National Mosquito Surveillance Program”—offer snapshots of mosquito populations in selected habitats across Aotearoa. However, these programs predominantly operate in and around major ports of entry (POEs), transitional and containment facilities (border clearance sites), and identified high-risk saltmarsh areas (Ammar et al. 2019), leaving vast sections of the country unsurveyed.

Limitations in understanding native and exotic mosquito distributions and dynamics in New Zealand due to fragmented surveillance led to the Ministry of Health (MOH) funding a citizen science program in 2019. The aim of the census was to:

1. Monitor mosquito populations in Aotearoa New Zealand.
2. Engage with citizen scientists and the broader scientific community.
3. Assist other vector-monitoring initiatives by identifying and reporting suspected interceptions of exotic mosquito species.
4. Collate, maintain, and present data supporting further ecological and science communication research.

The early scoping and development of this project were undertaken by a student research team from Worcester University as part of an exchange program (see research summary in Wood et al. 2019). The team worked closely with experts from Te Papa, MOH, Find-A-Pest, iNaturalist, the New Zealand BioSecure Entomology Laboratory (NZBEL), the University of Otago, and the Ministry for Primary Industries (MPI) to establish the approach, function, and workflow of the NZ Mosquito Census.

The New Zealand Mosquito Census primarily aims to address geographic and taxonomic gaps in data collected by other national mosquito surveillance initiatives. Compared to traditional monitoring programs, citizen-based monitoring is relatively inexpensive and utilizes valuable input from community volunteers. When engagement remains high and consistent, it enables effective surveillance over large areas with high temporal resolution (Dickinson et al. 2010; Sousa et al. 2022). Citizen science projects focused on mapping mosquito diversity, such as the “Mückenatlas” in Germany (Walther and Kampen 2017) and “Mosquito Alert” in Spain (Južnič-Zonta et al. 2022), have demonstrated their value as early warning systems for the emergence of exotic species near human populations. These projects are also highly responsive to sudden or unexpected changes in dispersal patterns, providing valuable data that complement active government surveillance programs. Citizen-collected samples make a unique contribution in that they often represent a much wider spatial extent and incorporate samples that would otherwise not be collected. These provide valuable insights into population distributions and their dynamics, which are crucial for a comprehensive surveillance system.

iNaturalist is a community-driven citizen science platform that allows anyone to submit photos of species they encounter anywhere in the world (Matheson 2014). Each observation is linked to the location where it was taken and identified by the user to the best of their ability. With support from the wider iNaturalist community, users may then confirm the identification or suggest an alternative. The entries contribute valuable taxonomic information on species distributions, biodiversity, and variation (Unger et al. 2020). The platform’s community focus aligns well with the goals of the NZ Mosquito Census, particularly in increasing the quantity and geographic diversity of collected biodiversity data—especially in locations that are difficult to access through traditional surveying methods. However, mosquitoes are notoriously difficult to identify taxonomically, often only distinguishable by microscopic features and requiring considerable expertise. Single photographs, like those used in iNaturalist, especially those with limited angles or low magnification, are usually insufficient for species-level identification of mosquitoes. A notable exception is Aedes notoscriptus, which is readily recognizable by its striped legs and distinctive band on the proboscis. This species is frequently identified to the species level and often reaches “research grade” on iNaturalist. In contrast, field observations of most other mosquito species are typically identified only to the family (Culicidae) or genus level. As a result, iNaturalist data should be interpreted with caution when comparing mosquito presence, absence, or abundance across species.

The census encouraged citizen scientists to proactively collect mosquitoes from their homes, wider suburbs, and domestic holiday destinations to obtain a broad geographical range of specimens. A promotional video was created to provide a step-by-step guide on how to participate in the census and why contributions are valuable to a national surveillance network (Te Papa website). Participants were instructed to carefully collect specimens to avoid physical damage to important morphological traits and to euthanize specimens in a freezer prior to postage (Te Papa website). A simple online form was produced to allow participants to fill out important metadata to submit alongside their specimens and is hosted on Te Papa’s website. The form was designed to incorporate valuable environmental and location information for the collected specimens to support future analysis and any necessary incursion follow-up. The time of day the mosquito was caught, the type of habitat it was collected from, and any further comments or notes from the participants were also included. Participants were also asked to select their preference for location sharing and to provide consent for either the exact or general location to be used for the iNaturalist entry.

Specimens were then mailed by participants, along with the associated submission ID, to Te Papa’s laboratory in Wellington.

Mosquito specimens were identified using Belkin (1968) and Snell (2005) by a mosquito taxonomist with a Leica M60 stereomicroscope and photographed using a Leica MC170 HD. The identification results were shared with each participant via an email sent manually by the Te Papa team. The collected photos were then uploaded to iNaturalist NZ so that participants could view their specimens online and see their contribution within the broader context of the project. This step was also important for supporting the project’s data accessibility, transparency, and reporting.

The census has been fully operational since 2019 and had received 840 submissions by the end of 2022.

This project took place during an unexpected event: the SARS-CoV-2 (COVID-19) pandemic. In 2020 and the following years, New Zealand faced the impacts of the COVID-19 pandemic, like many other countries around the world.

On 25 March 2020, New Zealand moved to COVID-19 Alert Level 4 in response to the global pandemic, initiating a nationwide lockdown for all non-essential workers. This measure remained in place until 27 April 2020. Various lockdown restrictions continued throughout the remainder of 2020.

Lockdown measures continued through 2021 in response to the ongoing pandemic, and in August 2021, New Zealand again entered a nationwide lockdown. From December 2021 through May 2022, these measures were lifted as the country transitioned to a new COVID-19 “traffic light” system, with public health restrictions progressively eased. The coincidental convergence of the pandemic with the first years of the mosquito census presented both a challenge and an opportunity, offering a chance to examine the pandemic’s impact on mosquito sampling in general and on a citizen science project in particular.

For the purposes of this retrospective report, each data point represents the submission of one mosquito species from one location, not individual specimen counts. The dataset included all supplied metadata from participants during the online submission stage (collection date, time of day, geo-coordinates, and habitat type—indoors, forest/bush, coast, or backyard/urban—for each calendar year). Using the coordinates, we further divided the dataset according to Statistical Area (SA) boundaries as defined by Stats NZ. Any unclear or contradictory site descriptions were manually verified using the provided coordinates in Google Maps. There were 24 submissions that could not be verified due to incomplete coordinates and were excluded from the final dataset. The final dataset comprised 445 verified submissions. Exploratory and descriptive analyses of this dataset were conducted using the R packages ggplot2 (Wickham 2016), treemap (Vitolo 2014), and summarytools (Comtois 2024) in R version 4.0.1.44 (R Core Team 2021). To support spatial analysis, a hexagonal grid with 10 km resolution was applied across all of New Zealand, excluding hexagons where more than 80% of the total area was covered by water. Submissions in each grid cell were tallied to produce a spatial distribution raster. To map species composition, the identified species were clustered and normalized against the “normal resident population” of their respective Territorial Authority (TA) as defined by Stats NZ.

To consider the origin of submissions in more detail, we added an additional descriptor of Statistical Areas (SAs) for each submission. SAs are geographic units used by Stats NZ that provide more detailed information about population characteristics than meshblocks (the smallest geographic unit for which statistical data are collected and processed by Stats NZ). The difference between the two types of SA can be explained as follows:

• SA1 is the output geography with the highest spatial resolution used by Stats NZ. Each SA1 is an aggregation of meshblocks and typically represents a population of approximately 100–200 people, with a maximum of around 500.
• SA2, a higher-level output geography, provides broader population aggregations than SA1. SA2 boundaries are designed to reflect communities that interact socially and economically. In populated regions, SA2 units generally contain similar-sized populations, typically fewer than 1,000 residents in rural areas and more than 5,000 in urban centers.

Using SA classifications, the distribution raster of submissions was divided into urban and rural zones for comparative analysis.

Submission data were analyzed against environmental and anthropogenic predictors (Tables 1, 2), aggregated by grid cells, to explain patterns in submission distribution. These predictors were integrated as spatial raster values.

Key environmental factors chosen for this study included mean annual temperature, precipitation, wind speed, and the presence of water bodies suitable for mosquito breeding (Belkin 1968; Laird 1996; NZBEL 2019). Weather data were sourced from NIWA (National Institute of Water and Atmospheric Research) and hydrographical data from the NZHA (New Zealand Hydrographic Authority). Wind speeds above 4.7 m/s were used as a threshold, as higher speeds reduce mosquito flight (Adeleke et al. 2022). Hydrological features such as wetlands, floodplains, lakes, and rivers were included, but coastal saline environments were excluded, as no saline-breeding mosquito specimens (Opifex fuscus) were recorded. Water presence was classified as a binary variable (0 = no, 1 = yes) to avoid misleading correlations.

Data from the 2018 NZ Census, at the SA1 level, provided anthropogenic variables: population size, population age, income, and COVID-19 Alert Levels during submissions. This allowed for further exploration of engagement patterns and the potential impacts of lockdown restrictions. While simplified, these predictors likely influence both mosquito occurrence and participant collection behavior. For instance, adverse weather conditions such as high wind or rainfall are likely to reduce both mosquito activity and human collection efforts. Similarly, densely populated areas may offer more breeding habitats for container-breeding mosquitoes. Raster data were processed using the R packages sf (Pebesma 2018), leaflet (Graul 2016), raster (Hijmans 2024), rgdal (Bivand et al. 2023), and spatstat (Baddeley and Turner 2005).

Survey data often suffer from overdispersion, leading to greater variability than expected under standard count models. To mitigate this, our models were tested using the Online National Mosquito Database (2012–2018) before being applied to the Mosquito Census data. Hurdle models were used to account for zero inflation, separating the probability of a submission occurring from the count of submissions per grid cell. Variance inflation factors (VIF) led to the removal of “mean age” and “temperature” (VIF > 5).

Predictor combinations were tested with automated model selection (AMS) using the Akaike Information Criterion (AIC) to identify the best-fit model. Models were run in R using the packages car (Fox and Weisberg 2019), countreg (Kharrat et al. 2019), MuMIn (Bartoń 2024), MASS (Venables and Ripley 2002), and pscl (Jackman 2024). Spatial dependence was accounted for using a simultaneous autoregressive (SAR) model, where lambda values indicated correlation strength.

Of the total dataset, 445 (53%) of submissions were positively identified as mosquitoes; 270 were identified as non-mosquitoes, arrived without a specimen, or were too damaged to be identified (31%); and 125 submissions were never received at the laboratory (15%) (Fig. 2).

For the majority of submissions, participants selected to share the exact location (66.5%) of their finds, with approximately one-third choosing to obscure the exact location (33.5%) (Fig. 3).

Figure 1. 

Stacked bar plot showing the number of identified mosquito submissions from 2020, 2021, and 2022. The mosquito species are identified in the plot as either native (green) or exotic (purple). Dotted lines indicate different COVID-19 alert levels across these years.

Figure 2. 

Pie chart showing the proportions of submissions that were mosquitoes (purple), non-mosquitoes (grey), empty or too damaged to identify (yellow), and never received at the laboratory (blue).

Figure 3. 

Pie chart showing the proportions of submissions in which participants chose to have the exact location of their collection obscured (yellow) versus exactly given (grey).

2020 census: During the census’s first year of operation, from January through May, 136 submissions were received and identified, representing 89 unique localities with 113 identifiable mosquito specimens (Fig. 1). During the period of COVID-19 Alert Level 4 (25 March–27 April), with the nationwide lockdown for all non-essential workers, only four specimens were collected and submitted. Between June and December, 51 submissions from 19 locations resulted in the identification of 35 mosquito specimens during periods of lockdown restrictions. There were 50 recorded census submissions that did not arrive at the laboratory, and four specimens arrived too damaged for species-level identification. In such cases, a follow-up email was sent to the contributor. In 2020, the Mosquito Census was approached by national news media (1News 2020; RNZ 2020) on two occasions, which helped support the initial public awareness campaign.

2021 census: Various lockdown measures continued through 2021, and in August 2021, Aotearoa again entered a nationwide lockdown. Between January and May, 139 submissions were received and identified, representing 36 unique locations and 111 mosquito specimens. Between June and December, 21 submissions were received from 10 unique locations and 15 mosquito specimens. There were 18 collected specimens recorded in the online database that did not arrive at the laboratory, and six arrived too damaged to process.

2022 census: When lockdown measures were lifted and public health restrictions were progressively eased between January and 31 May, 292 online submissions were received and identified, representing 107 unique locations and resulting in 150 mosquito specimens. There were 41 recorded submissions that did not arrive, and seven specimens arrived too damaged for species-level identification.

At the start of 2022, the Mosquito Census was featured in two mainstream news stories—one highlighting mosquito biting behavior (Stuff 2022a) and the other focusing on the census itself and framing it through a climatic perspective (Stuff 2022b). The latter supported the campaign’s goal for the 2022 summer/autumn season to increase participation from the South Island and regions outside major urban centers. To this end, targeted calls to action were shared, where permitted, on community group forums and social media channels.

The census also introduced a summer research scholarship to support an internship in collaboration with Te Herenga Waka – Victoria University of Wellington, fulfilling a key outreach and engagement objective for the project and contributing resources for this report.

As of the end of 2022, a total of 445 mosquito specimens had been submitted and identified through the census, representing 10 different mosquito species (Figs 4, 5, and Suppl. material 1: fig. S1). All three introduced species (Culex quinquefasciatus, Aedes notoscriptus, and Aedes australis) were detected within their previously established ranges, as described in the literature, confirming the increased abundance of Cx. quinquefasciatus analyzed by Kasper et al. (2023).

The native species Aedes subalbirostris was recorded in North Canterbury for the first time. None of the rarer species—Aedes arundinariae, Culex rotoruae, Culex astilae, Culiseta novaezealandiae, and Opifex chathamicus—had been submitted.

Figure 4. 

Radial dot plot of identified mosquito species submitted to the Mosquito Census. Different colors represent different mosquito species. Red and purple colors represent introduced species, and green and yellow colors represent native species.

Figure 5. 

Distribution of submissions by species across New Zealand. Different colored dots correspond to different mosquito species.

The majority of mosquito submissions came from urban areas, with most collected from Auckland (12%) and Christchurch (12%), followed by Wellington/Lower Hutt (7%). Similarly, most rural submissions were from the outskirts of these cities, with the largest number from Auckland (7%) (Fig. 6). There were no mosquito specimens submitted or received from Napier or the wider Hawke’s Bay region. The number of identified specimens from small urban areas was twice that from medium urban areas; interestingly, the medium urban areas showed double the number of native species compared to introduced ones.

When comparing urban and rural mosquito communities, most specimens collected from urban areas were introduced species (Aedes notoscriptus and Culex quinquefasciatus), which made up 53% of all submissions. Specimens collected from rural areas were nearly equal, with 15% being native and 17% introduced (Fig. 7). A broad comparison of environment type for all successfully identified specimens showed that the majority were collected indoors (68%) and in people’s backyards (22%). This pattern was similar across both introduced and native mosquito species (Fig. 8). Only 0.2% of the specimens were collected from coastal environments, identified as Coquillettidia iracunda. Interestingly, Opifex fuscus had not been submitted.

The dominant native species in both urban and rural areas was Culex pervigilans.

Our goal in 2022 to increase extra-urban engagement and sampling appears to have been successful, particularly in South Island submissions. There was a threefold increase in mosquito specimens submitted, rising from approximately 11.7% of total specimens in 2020 and 2021 to 35.9% in 2022. This increase is notable given that less than a quarter of Aotearoa’s population resides in the South Island, even though the island comprises well over half of the country’s land area. Christchurch, which accounts for roughly one-third of the South Island’s population, contributed 50.6% of mosquito specimens from the South Island in 2022, compared to 70.9% in the previous two years. This suggests an increase in participation from extra-urban areas (Fig. 9). Interestingly, no submissions have been received from Napier (Fig. 3).

Figure 6. 

Identified mosquito submissions from 2020 to 2022. Colored bars represent different regions around the country and indicate whether the specimen was collected from an urban or rural area.

Figure 7. 

Pie chart showing the proportions of mosquito specimens identified as native or introduced species and whether they were collected from an urban or rural environment. Segments show urban introduced (dark red), rural introduced (pink), urban native (dark green), and rural native (light green).

Figure 8. 

Stacked bar graph showing the number of mosquito specimens submitted and their identified species. Different colors within the bars denote the environment type from which each species was collected.

Figure 9. 

Temporal trends in identified mosquito submissions from (A) the North Island and (B) the South Island.

Our environmental and demographic analysis was limited by insufficient specimen numbers. As a result, attempts to explore socio-economic and environmental or climatic influences on participation rates did not yield statistically significant insights. Despite the absence of significant findings, we have chosen to mention this exploratory analysis for future reference and potential refinement.

The majority of submissions reinforced previously established distributions for the three most common species in New Zealand—Aedes notoscriptus, Culex pervigilans, and Culex quinquefasciatus—all of which are well adapted to urban environments (Kasper et al. 2023). All three introduced species (Culex quinquefasciatus, Aedes notoscriptus, and Aedes australis) are relatively well documented because of their public health relevance and their frequent detection in surveillance programs (Derraik and Snell 2004; Cane and Disbury 2010; Cane et al. 2017). In contrast, the distribution of many native species remains poorly understood. For some, there are no recent records from formal monitoring, and historical literature provides the only available range estimates (NZBEL Newsletters). Many native species are rarely encountered and are not well represented in current surveillance data. Their detection in the census highlights the potential value of citizen science for supplementing formal monitoring efforts and updating species distribution records.

For example, one species that fell outside this range was Aedes subalbirostris, with three specimens found notably farther north and inland than previously documented (Marks and Nye 1963; Nye and McGregor 1964; NZBEL Newsletters). Historically, its known range followed coastal areas of the lower eastern South Island. Although it had occasionally been detected in the Otago region, this new observation is located approximately 150 km north of Otago’s northern boundary. All three specimens originated from the same location in central Canterbury, Rangitata High country (Suppl. material 1: fig. S1g). The fact that all individuals were found at this new location offers more confidence in a true new locality record, and it is hoped that future observations in the area will better characterize this new community. While Ae. subalbirostris is not considered a species of public health concern, this finding highlights the census’s capacity to generate novel records, monitor range shifts, and refine our understanding of species distributions.

For the rarer native species, more targeted sampling efforts in rural areas are required. Unfortunately, due to the nature of this project, in which physical specimens are necessary rather than in situ observations, the requirement for sampling permits on public conservation land poses a significant limitation. This bias restricts the dataset’s utility for estimating population size, but it may still prove valuable for future simulations, particularly when combined with ecological characteristics associated with the samples. There is currently no clear explanation for the lack of submissions from Napier. Strikingly, surveillance data (NZBEL Newsletters) indicate that Napier is a hotspot for Culex quinquefasciatus, a species known to frequently use humans as hosts (Kasper et al. 2023).

The overall higher participation rate from urban areas was anticipated. There is a well-established positive relationship between total human population and the number of records generated by citizen science projects (Geldmann et al. 2016). This relationship is reflected in the census data, with over 65% of the population residing in large and major urban areas (Stats NZ 2018), which also include many of the already monitored POEs (Fig. 10).

Similarly, the higher proportion of introduced species submitted from urban areas was expected, as introduced species are well adapted to urban environments (Kasper et al. 2023). With the limited data available for many native species, analyzing species composition across the 2,237 SA2s at this resolution is not feasible.

Consequently, it remains difficult to determine whether the changes observed in 2022 are due to natural seasonal variability or if they reflect the success of the census’s efforts to increase engagement—particularly in light of the evolving impacts of the COVID-19 pandemic.

Figure 10. 

Submitted mosquito locations where circle size indicates the approximate number of mosquitoes per 5,000 residents. Green-colored fill indicates native species, and purple-colored fill indicates introduced species. Different shades indicate different species within these two main groupings.

This project took place during the SARS-CoV-2 (COVID-19) pandemic. The first global sweep of COVID-19 and the associated lockdown measures in Aotearoa presented both practical and administrative challenges for the census but also, serendipitously, offered an opportunity to survey the national mosquito population at a time when human movement and international travel were at unprecedented lows.

A sharp decline in participation coincided with the onset of the lockdown, followed by moderate engagement throughout 2021—a period during which no media support was provided (Fig. 1).

Lockdowns meant many people spent more time than ever in and around their home environments, but, being focused on personal health, work-from-home arrangements, or caregiving, fewer citizens were likely to be interested in mosquito sampling. On the flip side, the shift to online engagement may have encouraged more people to participate. Although many online submissions were received during the March 2020 lockdown, 50 specimens were ultimately not sent in. As daily routines and access to outdoor spaces changed—and as the pandemic accelerated the use of technology—restrictions on movement and limitations on posting items made participation in the census less appealing as a lockdown activity.

In 2022, as life began to return to normal, interest increased significantly in January, likely triggered by media coverage and targeted Facebook callouts. When comparing this trend with the surveillance data collected by the MOH, we observe a very similar pattern (Fig. 11), even in the case of suspected interceptions (Fig. 12).

Because mosquito trapping requires staff to be physically present in the field, sampling was temporarily paused or reduced. Some areas where mosquito samples were traditionally collected became harder to access due to travel restrictions and social distancing measures, limiting the number and range of data collected.

One might assume that heightened public awareness of viruses and disease transmission would have sparked more interest in mosquito-borne disease prevention and mosquito biodiversity, potentially giving the project an unexpected boost in public engagement. However, evidence from around the globe shows that the COVID-19 pandemic had a significant impact on the implementation of conventional mosquito vector control measures. For instance, in Asia, efforts such as the distribution of mosquito nets and community education campaigns were notably reduced (Khan et al. 2021; Lu et al. 2023). To date, studies examining how the pandemic may have altered the environmental dynamics of mosquito populations or other insect species are sparse (Fontúrbel et al. 2024; Zheng et al. 2024).

Figure 11. 

Numbers of adult mosquitoes sampled by the national surveillance program during the peak season between December of the previous year and May of each year are shown.

Figure 12. 

Numbers of suspected mosquito interceptions during peak seasons (December of the previous year to May).

One of the greatest opportunities this project has offered is the insight it provides into the complexity of citizen science. It highlights its approaches, challenges, and benefits. Comparing different projects and engaging with participants and experts (Wood et al. 2019) has helped illuminate best practices for future initiatives, such as identifying gaps, finding the right balance of tools, minimizing bias, and ensuring data quality. A prelaunch questionnaire involving both experts and the public proved crucial for effective planning. Now, after three sampling seasons, we can begin to evaluate the outcomes against the key aspects that should be considered when running a citizen science project.

Specific goals and outcomes: The Mosquito Census had a clear objective that complements targeted mosquito surveillance in Aotearoa New Zealand. To design and implement the project effectively, we needed to clearly define our specific goals and desired outcomes. Our primary aim was to collect data on the distribution and ecology of mosquito populations across the country. This would support future entomological research, enhance public outreach, inform the development of natural history data collection projects using citizen science, and help prevent or respond more effectively to mosquito-related biosecurity threats.

We developed an intuitive collection process and a user-friendly website interface to facilitate the submission of physical specimens for accurate identification and long-term preservation. This design was informed by interviews with mosquito experts from platforms and institutions including iNaturalist, NZBEL, Find-A-Pest, Mückenatlas, the University of Otago, and MPI (Wood et al. 2019).

Another central goal was to facilitate open data sharing and interoperability with other databases. We ensured our platform was compatible with iNaturalist for seamless integration, better supporting collaborative collection efforts.

Simple and user-friendly: Using findable, well-known, highly frequented, and trustworthy systems and platforms worked very well. While iNaturalist does not satisfy the input portal and back-end database needs of the Mosquito Census, this platform provides a way for the data collected and identified through the Mosquito Census to be shared with an existing citizen science community. We developed an additional user-friendly front end and a back-end system for recording and managing specimen data, hosted on the Te Papa website. This had the great benefit of displaying the census among the top three results when entering mosquito-related terms into web search engines.

Furthermore, a survey of museum visitors—later expanded to Reddit and Twitter for demographic data—helped assess public knowledge of biosecurity. We also determined essential fields for the submission form to ensure the data’s usefulness, with a test website complementing the surveys and interviews to further reduce user confusion (Wood et al. 2019).

Manageable engagement: Sending physical specimens is a level of engagement that seems higher than the normal contribution to a citizen science project. On the other hand, participants did not need to take complicated images and upload them to iNaturalist.

The survey prior to the launch revealed that more people thought they would be more likely to contribute to the project if collection kits and free shipping were offered to minimize effort and cost. Although the distribution of sample kits did not prove logistically feasible, providing hints about suitable collecting containers in a video on our website, combined with free postage, turned out to be an effective approach. This was reflected in the variety of creative and innovative containers we received.

Handling the submitted specimens, however, was very time-consuming despite the lower numbers of specimens than expected. Scholarship students and volunteers helped during peak seasons to manage correspondence and uploads.

Participant recruitment: We discussed methods of marketing the Mosquito Census with experts in public programming, exhibitions, design, marketing, and digital outreach, which helped shape marketing, public health, and funding strategies while addressing logistics, privacy, and media concerns.

Prior to the launch, we identified families, people with a previous interest in nature, and those annoyed or concerned by mosquitoes as potential participants. Participants were found directly via the Te Papa website and iNaturalist, but mainly through media releases and radio interviews.

Later, we reached out to communities and segmented the audience via social media after realizing gaps in participation across the South Island and rural areas in general. We facilitated talks and small events in collaboration with local organizations.

In the future, we recommend leveraging existing networks even more through national organizations and schools to facilitate further outreach.

Maintaining commitment: Citizen scientists can often lose interest in projects as the novelty fades, and it requires continuous effort to sustain engagement. While the census was not aiming for long-term participants submitting from the same locations but rather encouraging as many participants from different locations as possible, we still had to advertise the project frequently. Regular media presence was a great opportunity to raise awareness of New Zealand’s insects in general—and mosquitoes and citizen science in particular. Our survey prior to the launch showed a lack of public knowledge about mosquitoes. Having accessible information to link to, such as the Te Papa interface with a well-produced video, was extremely helpful and saved time in the long run. The video was an invaluable investment.

The media presence directed people to the Mosquito Census website, which provided comprehensive information about native and introduced species. We used various formats (videos, brochures, workshops) to cater to different learning preferences while also promoting the Mosquito Census.

We used social media for targeted calls but could have utilized these channels more effectively to reach a wider audience by sharing engaging content and success stories to pique interest.

Educational and hands-on workshops—a good channel to reach people—were held mainly within Te Papa, Zealandia, and Otari-Wilton’s Bush to communicate the need to protect local ecosystems and foster ownership and responsibility for New Zealand’s biosecurity.

Using iNaturalist gave us the opportunity to build a sense of community and gamification within the project, including friendly competition and participant rankings, which was reflected in an increase in direct mosquito submissions on iNaturalist NZ.

Apart from a single colouring competition for children, with book prizes for the best three pictures, we did not offer incentives such as prizes. However, we found that feedback and personal communication with participants about their results—and the display of their contributions on the iNaturalist map—were rewarding. We created a clear communication channel in the form of a census email address. Additionally, personalized communication ensured that we could provide adequate support for citizen scientists so that they understood the tasks and could contribute meaningfully to the project.

Avoiding bias: Citizen science can have strong spatial biases toward areas where human population density is highest, toward biodiversity hotspots, and/or toward areas related to recreational activities such as national parks. Such spatial biases can challenge the comprehensiveness of the data and limit interpretations of changing spatial trends, as well as the likelihood of detecting species in under-visited areas.

While we tried to target groups from underrepresented regions, we did not specifically aim for contributions from underrepresented demographic groups. The aim was to identify those gaps through the anthropogenic drivers in this study.

With more data, this analysis can be repeated in the future, and tailored messages that resonate with specific demographics could be developed, highlighting how participation aligns with their values or interests—whether conservation, education, or community engagement.

Data management: Data storage for the census is ensured via a secure database on the Te Papa website’s backend. We have developed a robust plan for data quality, as the dataset consists of a partly automatically generated suite of covariates composed of information from the submission form. Species identification is conducted by experts, eliminating the need to train citizen scientists. Since no quality check is required, this process does not burden participants, who are valued for their contribution and rewarded with the results. Quantitative data were verified using statistical areas (SA) provided by Stats NZ. Unclear or contradictory site descriptions with provided geo-coordinates were verified manually using Google Maps. Samples that could not be verified were excluded.

We have also agreed with the Ministry of Health (MOH) to check any results indicating potential exotic mosquitoes and for the MOH and Te Papa communications teams to collaborate on a response plan for the media in case an unwanted organism is found.