Towards Ecological Science For All By All

Imagine a global community of people who—as a matter of daily routine—observe wildlife, smell the air, discuss the weather and touch the world around them. As sensors, people have evolved the ability to assess environmental change and one might easily conceive an army of such ‘citizen scientists’ ready and willing to advance knowledge. Of course, it is not that straightforward. Although volunteers have supported science for decades, the concept of citizen science that has a more holistic consideration of methods, ethics, philosophy and social science is relatively new. We know citizen science can generate large, high-quality datasets, but as much as this volume of observations demonstrates the potential of citizen science, there are new challenges and opportunities, such as reconnecting people with nature. No matter the discipline, how best to incorporate citizen science into research in an efficient, cost-effective and ethical way that works for both contributing citizen, and professional scientist is still to be fully understood. Indeed, from the perspective of the social sciences, there is the need for citizen science to adapt to a society that demands science to be responsive to rapidly changing concerns. The Oxford English Dictionary defined citizen science in 2014 (OED, 2020), and combined two early definitions that emphasised (a) the responsibility of science to society (Irwin, 1995) and (b) the participatory role of people contributing observations or efforts to scientific endeavours (Bonney, 1996). More recently, Ceccaroni et al. (2017) attempted to reconcile these viewpoints to describe citizen science as work undertaken by civic educators and scientists together with citizen communities to advance science, foster a broad scientific mentality, and/or encourage democratic engagement, which allows society to deal rationally with complex modern problems (Eitzel et al., 2017). As of 12th December 2020, the citizen science hub SciStarter.com lists 1,358 active citizen science projects. Of these, 642 (47.3%) are listed under the topic ‘Ecology and Environment’. Indeed, there has been a rapid growth in such projects since the 1940s, and since 1990 there has been a 10% decadal increase (Pocock et al., 2017). This expansion in citizen science has precipitated a range of supporting infrastructure, including typologies (e.g. Danielsen et al., 2009; Wiggins & Crowston, 2011), best practice principles (e.g. ECSA, 2017) and frameworks for implementation (Chase & Levine, 2016; Resnik et al., 2015; Shirk et al., 2012). It is because of the growth in citizen science activities across the breadth of ecology and environmental studies, and increasing attention in the social sciences, that the BES launched an open call for papers to this Special Feature on citizen science across six of the BES journals in October 2019. In this Editorial, we discuss the papers and topics covered and conclude with a brief outlook on ongoing and future developments. This Special Feature comprises 19 papers, of which six are in Journal of Applied Ecology, five in People and Nature, four in Journal of Animal Ecology, two in Ecological Solutions and Evidence and one each in Journal of Ecology and Methods in Ecology and Evolution. Although wide ranging and varied, as a collection these articles go some way to addressing the two perspectives intended for this Special Feature; the contribution of citizen science to the advancement of ecological knowledge and the contribution of community-based perspectives to citizen science. Among these 19 papers are 16 that present original quantitative or qualitative research, a practice-based article (Bonnet et al., 2021), a perspective (Palmer et al., 2021) and a literature review (Winch et al., 2021). Seven papers address aspects of quality assurance and quality control (QA/QC). Trust is a critical factor for leveraging partnerships between citizen scientists, project coordinators and decision-makers (Freitag et al., 2016). However, although an increasing number of disciplines incorporate citizen science, there remains scepticism of the public as a trusted source of scientific information (Burgess et al., 2017; Tredick et al., 2017) and many projects struggle to meet decision-maker needs (Newman et al., 2016), who require data suitable for reliable inference. Thus, underpinning quality assurance and control is that it is the scientific mode and quality of information that matters, which in the context of citizen science means ensuring that any participation in the data collection by community members may be improved by the development of procedures and protocols to be followed before and during data collection (quality assurance). This is also extended to processes for improving data quality after data collection, or ‘at the back-end’ (quality control). The papers by Pernat et al. (2021) and El Bizri et al. (2021) are two such studies of quality assurance. Pernat et al. (2021) compared trained researcher (or ‘professional’) data to citizen science data on mosquito collections in the German ‘Mückenatlas’, while El Bizri et al. (2021) compared local peoples’ capacity to determine the reproductive status of female pacas (Cuniculus paca). Pernat et al. (2021) used data from seven years of surveillance to evaluate what kind of information each method of collection provides. They found that systematic monitoring was superior in terms of mapping diversity, but passive monitoring did a better job detecting invasive species. This suggests that citizens can often do better at detecting novel occurrences than systematic approaches. With a 17-year long dataset, El Bizri et al. (2021) found that indigenous knowledge was already highly accurate for pregnancy diagnosis (72.5% correct) in female pacas, which was even better after training (88.2% correct). Thus, depending on the circumstances, citizen science data may not only complement systematic monitoring but can be superior to it by providing novel insights emerging from extensive local or indigenous knowledge. This has been seen in citizen science before: for example, with Hanny's Voorwerp on the Galaxy Zoo project (Cardamone et al., 2009). It is through insights such as these that blur the boundaries between those who are considered citizen scientists or professionals when either may produce sufficiently high-quality data to inform wildlife management. However, such high rates of agreement as found by El Bizri et al. (2021) may only be seen in about 55% of similar studies on the accuracy of citizen science data (Aceves-Bueno et al., 2017). The term ‘extreme’ citizen science, which has emerged in recent years, was borne partly out of internet accessibility, which has led to the development of thousands of web-based or mobile applications to aid the citizen scientists in recording accurate observations. Extreme citizen science involves not only the crowdsourcing of data but also its analysis (see Haklay, 2013). The ‘From Practice’ paper by Bonnet et al. (2021) presents the Pl@ntNet platform, which was initiated in 2009 and as well as being a platform for participatory research, aggregating and disseminating observations, it allows the identification of plants by automatic visual recognition. Bonnet et al. (2021) describe how Pl@ntNet was disseminated to local communities in two different socioeconomic contexts; the Ramières Reserve (France) and the Lewa Conservatory (Kenya). For such a tool to be widely adopted, the key factors they identify reinforce existing research and include good communication (Constant & Roberts, 2017), data quality and validation, and recognition of participant expectations (Dickinson et al., 2012). In line with the concept of extreme citizen science, this includes adopting an open data policy and ensuring technologies are commensurate with local infrastructure. The paper by Petersen et al. (2021) provides a forensic study of biological recording in Norway and finds classic taxonomic biases towards traditionally recorded and charismatic fauna (e.g. birds), and a non-random skew of observations towards anthropogenic land-uses (‘road-side bias’), whether the focus be on any, rare or non-native species. Thus, the study by Petersen et al. (2021) provides useful context for the remaining three papers under the theme of quality control which consider how to account for different forms of bias generated through citizen science data related to site selection and observer retention over time (Dambly et al., 2021), artificial light conditions (Ditmer et al., 2021), and choice of methods in generating less biased SDMs (Steen et al., 2021). With long-term monitoring, there is a concern that frequent turnover of observers can affect accuracy (Dickinson et al., 2010). Dambly et al. (2021) develop a ‘virtual ecologist’ model to test whether and how opportunistic site selection and uneven observer retention over time affect monitoring of bat roosts. They showed that these issues can result in biased trends, affecting the reliability of monitoring projects. Their findings highlight the value of engaging and retaining citizen science observers, a standardised sampling design and the collection of metadata. However, from an interdisciplinary perspective, what is termed ‘bias’ in the natural sciences may be part of the rich social dimensions that shape the science in particular ways; and it is this social shaping of the citizen science that the papers submitted in to People and Nature are largely concerned with. In particular because they alert us to the human and social dimensions related to why the concerns of citizen science may be particularly compelling to observers, who we might like to engage for extended periods of time. Observer bias may also be influenced by environmental conditions. Ditmer et al. (2021) tested whether the incorporation of artificial light at night (ALAN) conditions influenced the detection of American black bears (Ursus americanus) by citizen scientists. Members of the public provided 1,315 observations of black bear across Minnesota, USA. Using an occupancy modelling framework, the authors found that when compared to other commonly used metrics of human footprint (e.g. housing density), artificial light conditions did the best job of accounting for spatial bias showing higher rates of detection with elevated illuminance. Thus, bear abundance may be substantially underestimated in more natural conditions. Such spatial bias in volunteer effort associated with urban infrastructure is well established (Geldmann et al., 2016; Tiago, Ceia-Hasse, et al., 2017), but the added interaction of artificial light improves our understanding of detection bias, and can be incorporated into occupancy models to improve estimation and predictions of organisms’ distributions and abundances. Further error may result from methodological choices made during the production of SDMs based upon presence-only data, which is a frequent characteristic of citizen science. Such data generate class imbalances where one class (e.g. absence data) is far more abundant than another (Robinson et al., 2018). Steen et al. (2021) compare three choices for mitigating class imbalance: spatial thinning, class balancing and majority-only thinning using eBird data for 102 species. Finding that there was no single best approach across all species, and the considerable differences in SDM performance, the authors recommend a series of factors to be considered on a case-by-case basis to guide how to thin or balance data. The involvement of volunteers in recording species observations is one of the most established forms of citizen science (although it may not always have been referred to as such). For example, The Audubon Society sponsored Christmas Bird Count began in 1900 (Bock & Root, 1981) and the UK Butterfly Monitoring Scheme in 1976 (Pollard & Yates, 1994), closely followed by American and European equivalents. Such is the popularity of ornithology in particular, that bird data dominate biodiversity occurrence data on the global biodiversity information facility (Troudet et al., 2017), thanks in part, to the highly successfully citizen science platform eBird (Sullivan et al., 2009). When combined with conventional monitoring, citizen science data have broadened the opportunity for temporal ecological studies, allowing for the study of phenology and temporal population dynamics. It has been demonstrated to improve our understanding of the phenology of animals (Van Der Kolk et al., 2016) and plants (van Vliet et al., 2014) in relation to a changing climate; and of the distribution of invasive species (Crall et al., 2011), amphibians and reptiles (Tiago, Pereira, et al., 2017) and migratory birds (Robinson et al., 2020). In the absence of citizen science data, many studies may be limited to fewer species or a more narrow geographical range, or investigated only within a small portion of an annual cycle or an organism's life cycle. In this collection, the use of eBird data by La Sorte and Graham (2021) has allowed the associations between vegetation and breeding bird migration to be tracked across the entire annual cycle, rather than just focusing on a small portion of the cycle, and across a broad spatial extent. The use of citizen science data has also allowed habitat tracking to be investigated across a large number of species while previous tracking studies have been limited in the number of species they have been able to investigate simultaneously. Similarly, citizen science data from Project FeederWatch allowed Latimer and Zuckerberg (2021) to identify temporal population dynamics of North American birds. The authors used the repeat observation data from this scheme to develop dynamic occupancy models, which explicitly model differences in the recording and detection processes. The recent development and expansion of hierarchical modelling techniques such as these have facilitated a boom in the usability of citizen science data. The combined use of citizen science data from multiple sources or programmes will facilitate greater flexibility and increased scope in the questions that can be answered. For example, Williams et al. (2021) combine data on purple martin (Progne subis) occurrence from eBird, abundance from the Breeding Bird Survey and demographic parameters from Project MartinWatch to investigate the relationship between habitat suitability and species demography. This study highlights how citizen science projects may also record more than just the observations of a species’ presence by asking observers to record data on fecundity such as number of eggs and fledglings per nest. Technologies such as camera traps offer new ways for citizen scientists to collect and contribute conventional data (e.g. species presence) in a more planned and systematic way. This is because traps can be left in fixed locations on site, and over more defined survey periods. Twining et al. (2021) use such data in a study of red (Sciurus vulgaris) and grey squirrel (Sciurus carolinensis), and pine marten (Martes martes) distribution at 332 sites distributed across Northern Ireland. Camera trap data were used to build species occupancy models and demonstrated that the recovery of pine marten was strongly, and positively associated with red squirrel populations, with grey squirrel having the opposite relation. Expanding on the data to develop a model of habitat suitability, the authors also warn that grey squirrel populations are likely to persist in urban areas. With clear policy and ecosystem management implications, this study demonstrates eloquently how careful set-up, handling and verification of citizen science data through the use of camera traps can lead to real impacts for natural conservation (see also Santangeli et al., 2020). Biodiversity extends to diversity within species (UNEP, 1994) and so it is that Aavik et al. (2020) took a citizen-science approach to study the distribution in morph frequencies of flower styles in populations of the heterostylous grassland plant cowslip (Primula veris) across Estonia. A citizen science campaign ‘Eesti otsib nurmenukke’ (‘Estonia is looking for cowslips’) engaged participants from 1,700 localities across Estonia. Analysis of the data indicated deviations from equal morph ratios in areas with high human population density and greater habitat fragmentation, with a higher likelihood for inbreeding and fitness declines (Leigh et al., 2019), a finding that builds on earlier evidence from a much more geographically restricted study (Van Rossum & Triest, 2006). In conclusion, citizen science contributed to monitoring of species movement and distribution through the engagement and mobilisation of technologies and people who could monitor these over a wider spatial and temporal range, in comparison to a traditional study. The broadened spatial and temporal range of data collection also has the potential for allowing for new insights and associations, and indeed raises that potential for citizen science to play a significant role in long-term monitoring. Detecting environmental change at multiple spatial and temporal scales through biotic and abiotic monitoring is not only important for science and scientists but it is also fundamentally important for policymakers and environmental managers because it provides the basis for longer-term planning and resource allocation (Parr et al., 2002; Peters et al., 2014). To successfully detect environmental change at multiple scales requires long-term efforts to disentangle change from background noise (Magurran et al., 2010). Citizen science has been posed as a possible solution to this given its potential to generate data at spatial and temporal scales not achievable by conventional means (Pocock et al., 2018; Thornhill et al., 2016). Alongside funding shortfalls, a limiting factor to long-term monitoring is the successful engagement of volunteers, which may be limited by the amount of survey effort required (Weiser et al., 2019). It may, therefore, be sensible to optimise monitoring to maximise volunteer retention, as well as ecological and economic efficiencies. What then would be the minimum amount of monitoring required to detect robust trends in, for example, woodland bird populations? This is an aspect explored by Prowse et al. (2021), who resample 16 years of monitoring data to simulate different levels of monitoring effort. They find that rather than reducing the number of sites surveyed, trend detection would be retained most effectively by revisiting biennially but that would in turn, risk long-term engagement. A compromise might be met by reducing the number of sites visited annually but retaining the spatial extent; however, this may compromise the ability of a monitoring programme to detect early warning signals, particularly in rare or declining species. Thus, social dimensions, traditionally seen as aspects to be controlled within scientific research, become aspects that critically inform both the success of the data collection and the outcomes of the research. Citizen science is also an approach that places non-scientist experts and communities at the centre of the scientific endeavour. Billaud et al. (2021) engage more than 1,000 French farmers over a period of seven years to disentangle the effect of farm practices upon invertebrate diversity from natural variations. Varied relationships are found between farming practices and the five invertebrate groups investigated. For example, flying taxa are negatively impacted by pesticide use and mineral fertilisation, but the effect upon other groups is mixed. The authors recognise the potential benefit of monitoring being sustained by farmers, which may have the additional benefit of contributing to conservation through an increased awareness of biodiversity, and the role of biodiversity for farming. Prowse et al. (2021) question the social and political acceptability of adapting long-monitoring strategies. Indeed, the goal of any monitoring plan is a subjective decision that should be made by a wide variety of stakeholders under the guidance of social scientists (Hauser et al., 2006). The two papers regarding long-term monitoring in this Special Feature (Billaud et al., 2021; Prowse et al., 2021) further contribute to this discussion. Each reviews a successful approach but poses challenging questions for anyone designing a new monitoring plan resourced by volunteers. What resolution of information is required within the protocol spatially or temporally, for it to be effective? And, what are the requirements of the decision-makers that will use the data? Perhaps more so, they point to questions about drivers of participation, which is the central focus of the next selection of papers. The use of the term citizen science suggests that by including citizens or communities, or individuals, in addressing scientific questions that it is by its very nature, a social or a psychological endeavour. The next set of papers focus on the question of what motivates individuals to take part in a citizen science project. This is a critical question that must be addressed if citizen science is to serve the dual purpose of generating trustworthy information at an appropriate spatial and temporal resolution, as well as societal transformation (Bela et al., 2016). Yet, citizen science research has only recently begun to consider what motivates participants to take part and stay engaged (e.g. Geoghegan et al., 2016), and the extent to which citizen science might enhance public understanding of science or engender behaviour change (Bonney et al., 2016). Winch et al. (2021) make a substantial contribution to this area of research first, with a systematic review of more than 1,000 papers regarding volunteer participation. From a large set of reasons for participating, they find that the most important factor is to design projects that align to the motivations of the participants. More intriguingly, they go on to compare and contrast the demographics and motivations of those volunteers who are a part of the online, UK-based NatureVolunteers community, to projects being advertised by conservation organisations. They do this to identify how such organisations might tailor their projects to appeal more readily to prospective volunteers. The NatureVolunteers community is typically younger, and more interested in physical activity, skills use and habitat restoration than the conventional communities involved in environmental volunteering. Addressing these mismatches may broaden the appeal of conservation projects, including citizen science and diversify the volunteer base. However, as they and other authors identify, the motivations of volunteers are often diverse and meeting scientific goals are only likely to be a priority for a minority of groups. Efforts to understand the social, as well as the political and contextual nature of participation, may help us better understand both why some groups are well represented in citizen science, and why some groups (and spatial areas including Asia and Latin Americas) still remain underrepresented. One way to address such mismatches could be to co-design projects with participants to embed their values and opinions (Bela et al., 2016). MacLeod and Scott (2021) report how the New Zealand Garden Bird Survey (NZGBS) reporting process has been adapted in response to three surveys that focused on different communication channels. An impressive 15,844 responses were received and as a result, the team was able to diversify and refine their communication strategy to ensure that the participant feedback loop was more complete, and that the results of the NZGBS were better disseminated to a larger, and more diverse community of participants. Ensuring that this happens has been frequently cited as an important factor for long-term engagement in citizen science (Geoghegan et al., 2016; Sullivan et al., 2014). However, despite efforts to embed community values in the co-design of the NZGBS project, the authors acknowledge the complexities of co-production, and the success as largely one of promoting the organisations' goals of good governance rather than making ‘a real-world difference in engagement’. While understanding what motivates volunteers to participate in citizen science, one might also wonder what the less tangible or societal benefits are for conservation are (see, e.g. Billaud et al., 2021). There are many ways to understand what these societal benefits may be—from building social cohesion through addressing environmental problems, or providing active forms of education and involvement in local or global issues. Most of the contributions in the Special Feature, however, focused on the interests related to the role of psychological or individual behaviour change. Santori et al. (2021), for example, analyse behaviour change in 148 participants in the Australian turtle mapping project TurtleSAT. Unexpectedly, behaviour and attitude changes were not related to observation rate and were unlikely to be altered by participation. If scalable to other projects, this finding adds important context for the often witnessed long tail of participation, that is when relatively few volunteers generate the majority of the information (e.g. August et al., 2019), and that even minimal participation may have tangible conservation outcomes. An under-researched area of citizen science, and more widely environmental stewardship, is the emotional bond between a person and a place, termed ‘place attachment’ (PAT). Haywood et al. (2021) delineate a revised three-dimensional model of PAT for citizen science (after Raymond et al., 2010) and apply it to the COASST program. The three dimensions comprise personal, community and natural environment components, with seven major constructs across them. Through the analysis of interviewee responses, Haywood et al. (2021) demonstrate that the participants exhibit PAT in all three dimensions and show how the PAT profile of an individual may change over time, with important implications for sustained engagement. These findings also reflect the difficulty in measuring, but undoubtedly important, motivations for citizen science (Tiago, Gouveia, et al., 2017). However, the authors highlight that more research is needed to investigate whether the unique PAT profile of participants is a function of personal, social or programmatic variables pre- and post-programme participation. In recent years, there has been an increasing focus on ethics within citizen science, both in the engagement of volunteers (Resnik et al., 2015), the contribution and status of social knowledge, and, although perhaps less so, the ethical treatment of wildlife. The latter reflects the setting of the ground around environmental ethics in scientific research, and the potential shaping of citizen science by other regulatory forces, in this case animal welfare policies (e.g. Drinkwater et al., 2019). This regulatory context, in turn, is being shaped by an increasing public concern for animal welfare (McMahon et al., 2012) and, for example, in 2019 Citizen Science: Theory and Practice presented a special issue on ethical issues in citizen science (Rasmussen & Cooper, 2019). In this Special Feature, Palmer et al. (2021) ask ‘What (is the role) for regulation?’ in particular, for those wildlife-focused citizen science projects that disturb animals, such as involving mark–recapture methods or trapping. The authors provide a comprehensive and thoughtful overview of UK legislation pertaining to animal research and citizen science and offer three key discussion points: (a) take stock of wildlife-focussed citizen science, (b) assess the state of formal regulations and (c) consider the integration of informal regulations. The paper opens out its perspective on the need for citizen science to engage with the political, social and historical regulatory frameworks that shape society as well as science. There have been huge developments in the application of citizen science, and we are now moving beyond its use simply to observe or model where species occur. Citizen science data are now being used to study species demographics, phenology and biodiversity change to name just a few applications, as well as the behaviour of both participant and subject. All the evidence, of which this Special Feature provides an excellent snapshot, points towards further growth and expansion of the field with goals to engage more people from diverse backgrounds, generate better and more research, expand into new topic areas and to become more democratic; giving the participant a voice. However, one of the limitations of this Special Feature is that many of the theoretical and conceptual assumptions about citizen science have not yet been addressed. What does it mean, for example, to include a diversity of participants in citizen science, if citizen science remains unwilling to be shaped by the real challenges that this may offer the natural sciences? What different roles can citizen science play across the broad temporal and spatial scales that these studies have identified, and to what extent can citizen science play in elevating local or indigenous knowledge in the scientific endeavour, potentially transforming our understanding of what science can do and how it can contribute to solving our contemporary environmental problems? From this Special Feature, care could be considered a golden thread necessary to maximise the citizen science opportunity. Care extends to every aspect of the journey of a citizen science project from inception to dissemination. The use of citizen science should be carefully considered and not all projects are suitable for public engagement (Pocock et al., 2014). Thereafter, the design of projects must consider many facets including procedures for minimising and mitigating errors, the spatial and temporal resolution of data required, how that data will be handled and how best to communicate the results. These ideas are not new for ecosystem monitoring (Vos et al., 2000), but in a field so dependent on volunteer effort like citizen science, they take on extra gravitas. Indeed, the emerging knowledge areas shift the focus away from data acquisition to the complex but fundamental world of the participant. Care here extends to understanding the audience at a granular level, their attachment to place and their expectations. Several papers provide new insights into participant motivations, and others have adapted elements of their projects to increase engagement to good effect. Furthermore, there is potential for the general public to gain a greater understanding of and hence build greater trust in science overall. Indeed, there are many contributions to best practice among the collection. In conclusion, the future of citizen science remains very promising. As an interdisciplinary scientific community, we are beginning to understand how it operates and are in the process of formalising the approach in an ethical way. Challenges remain in each of the themes highlighted here, as well as opportunities for refinement. For ecological research, citizen science can assist from a local to international scale, and can focus not only on species occurrences but also on a wide range of ecosystem components. And, just as technology provides so much support to citizen science, as that technology continues to advance, so too will the capacity of citizen science to deliver.