Citizen science initiatives document biodiversity baselines at an urban lake

Introduction

Global biodiversity is currently experiencing changes that will impact the future makeup of the planet (Sala et al., 2000). For example, the extinction rate of species now greatly exceeds baseline levels (Pimm et al., 2014), which has been attributed to the increasing fragmentation of natural habitats, continued pollution of the environment, and unsustainable consumption of natural resources (Butchart et al., 2010). One of the main drivers of these impacts to biodiversity is urbanization (Piano et al., 2020), and while urbanized areas account for just 3% of the total land use of the planet, the changes wrought from the construction of buildings, roads, and other infrastructure extends far beyond land use (Chapman, 2003; Faeth, Bang & Saari, 2011). Aquatic habitats in cities are particularly vulnerable as they are often hotspots for local biodiversity (Hill et al., 2017), yet also centers for human use, recreation, and waste disposal (Hassall, 2014). Improvements to enhance the value of urban waterbodies and increase their recreational appeal can have both positive and negative impacts on biodiversity. For example, cosmetic changes to an urban lake can provide new habitat features for some species, such as adding a boardwalk where algae can attach and fish can congregate, thereby increasing their abundance and diversity, which can in turn help to raise local awareness of the importance of conserving the waterbody and its surrounding habitat (Savard, Clergeau & Mennechez, 2000; Qiu, Lindberg & Nielsen, 2013). However, urban improvements can also have negative impacts on biodiversity. For example, wetland management practices, even if intended to aid conservation, can unintentionally impact other species, and potentially public health by providing habitat for organisms that may vector diseases (e.g., Hanford, Webb & Hochuli, 2020). Furthermore, urban improvements may lead to increased human disturbances, noise and light pollution, and removal of key habitat elements, which can disrupt the behavior, reproduction, and migration of wildlife (Ewing et al., 2005). To minimize the negative impacts of urban sprawl on biodiversity, it is important for urban planners to consider the resident ecological communities when designing and implementing these projects (i.e., Smart Growth: Daniels, 2001).

Establishing a baseline of the species diversity that exists at a site prior to change provides a point of reference against which future observations can be compared (Mihoub et al., 2017). It can be an insurmountable challenge to accurately track changes in biodiversity without a baseline (Magurran et al., 2010). By comparing current data with such a baseline dataset, we can identify species that are declining, expanding, or shifting their distribution, as well as changes in community processes and ecosystem functions (Gullison et al., 2015). Landscape changes in urban areas often decrease the amount of habitable land available to local biodiversity, and many such initiatives are implemented without much knowledge of the organisms who resided in a habitat prior to changes (Bobrowiec & Tavares, 2017). Temporal baselines are also needed to establish targets for biodiversity conservation and progress to conservation goals to be evaluated. Performing studies to document the biodiversity in habitats before projects take place is important to be able to verify how much of the diversity found its way back to the habitat once the project was completed. Having reliable baseline data is needed to reconstruct the impacts of human activities, climate change, and other factors on biodiversity, and for developing effective strategies to protect and conserve natural ecosystems. Such standard biodiversity monitoring is also needed to identify meaningful benchmarks for biodiversity (Feest, 2006).

Large ecological datasets are critical to track changes in biodiversity but are logistically challenging to cover the spatial and temporal scales needed for understanding the magnitude of an impact (Costello & Wieczorek, 2014). Citizen science is a cost-effective, rapid, and efficient way to gather data on biodiversity over large areas and long intervals, which can be leveraged for documenting changes in biodiversity (Theobald et al., 2015). In recent years, the use of citizen science has become increasingly popular for monitoring and recording changes in the natural world (Chandler et al., 2017). Online applications such as iNaturalist allow citizens to engage with nature by taking photos of organisms they encounter and upload them to a site for other members of the community to identify. However, there are some limitations to the use of citizen science for documenting changes in biodiversity as the quality of the data may be unreliable, or there may be biases due to the locations where people choose to collect data, or certain species may be considered more interesting to document (Tweddle et al., 2012). Regardless, such biodiversity platforms are the increasingly becoming the sources of data for understanding changes in ecological communities over time, informing conservation efforts, and documenting impacts on biodiversity (e.g., Kirchhoff et al., 2021).

Our study focused on Cedar Lake in Cedar Rapids of eastern Iowa, which is a small, urban lake that is frequently used by people for recreational fishing and has a documented history of pollution. The City of Cedar Rapids enacted a 5-year improvement plan to bolster the flood wall and increase the recreational-use of the lake, with construction starting in 2022. Consequently, in 2021, we set out to create a spatially explicit, temporal baseline of biodiversity data at Cedar Lake through the use of semi-structured citizen-science initiatives. We evaluated the resulting dataset by comparing common community diversity metrics in our work to a dataset containing all past observations from Cedar Lake posted by citizen scientists. Our study provides a baseline for documenting the impact on biodiversity derived from the physical changes to the habitat of Cedar Lake and is relevant to urban studies around the globe that aim to document biodiversity and its changes over time.

Study site

The study was conducted at Cedar Lake, Cedar Rapids, Linn County, Iowa (Fig. 1). Cedar Lake is a 0.49-km² urban lake in the center of the city and currently serves as a drainage for most of the city’s waterways, a flood barrier for the nearby Cedar River, and as a recreational space for fishing, biking, and kayaking. The climate in Cedar Rapids is characterized by hot summers and cold winters, with monthly average normal temperatures ranging from −5.9 °C in January to 23.6 °C in July (National Oceanic and Atmospheric Administration, 2023). The shoreline is dominated by a mix of vegetation types and urban features, including trees, marshes, buildings, rocks, and paved walkways. From anecdotal observations, we have found that the lake supports a variety of plant species, including emergent species such as cattails and bulrushes, submergent species like pondweed and coontail, and floating species such as duckweed. To our knowledge, the biodiversity of Cedar Lake has never been formally assessed because no surveys have ever been published from there, thus available information concerning almost any aspect of Cedar Lake’s resident biota is nonexistent in the peer-reviewed literature.

Results

Our iNaturalist project (The Biodiversity of Cedar Lake) went online 1 April 2021, and the first observation was uploaded on 4 April 2021 and the last on 18 October 2021, which corresponds to the active growing season in eastern Iowa. During this period, we recorded a total of 1,345 biodiversity observations with 787 of these becoming Research Grade observations from 60 different observers by 31 March 2022 when we downloaded the data for analysis (Table 1). For these observations, 232 species were detected, 200 of which were classified as Research Grade. In the prior dataset from this site, the first observation was uploaded on 6 July 2011 and the last on 25 March 2021. During this period, a total of 257 biodiversity observations were uploaded with 168 of these becoming Research Grade observations from 22 observers. For these observations, 182 species were detected, 86 of which were classified as Research Grade. We found that most Research Grade observations (>90%) were recorded from within 100 m of the lake’s shore in both datasets (Fig. 1; Table 2). Only 41 species were shared between the two time periods, with 51 species unique to the prior dataset (mostly birds) and 159 to our work (mostly insects and plants). The top observer in the prior dataset contributed 43% (72 out of 168) of the total Research Grade observations and documented 43% of the species (37 out of 86), whereas the top observer in our work contributed 35% (272 out of 787) of the Research Grade observations and documented 63% of the species (125 out of 200).

The estimated species richness using the Chao method for our study was more than double that of the prior work at the site (374 species versus 174 species), with an upper bound of nearly 500 species in our work compared to an upper bound of just about 275 in the prior work (Table 1). Citizens detected 200 of the 374 estimated species in our data (53.5%), compared to 86 of the 174 estimated species in the past data (49.4%). A temporal assessment where data in the prior work was restricted to the same months as our work (April to October) revealed an even wider gap between time periods for estimated species richness with the prior work down to just 131.1 ± 36.5 species (95% CI = 85.4–238.9 species). Rarefaction curves indicated that our work not only detected more species with more individuals than the prior work, but extrapolations suggested that continued efforts using our approach may result in a much higher amount of species diversity detected at the site overall (Fig. 2). The Shannon Diversity Indices for the two time periods were both >4, with our work being greater than 4.5, indicating that we detected significantly more species from a wider range of abundances than the past data (t = 19.57, df = 188, P < 0.001). Simpson’s Evenness values showed that the past data detected a more even community (0.3 versus 0.22). Our work demonstrated a much more balanced distribution of observations among major groups of biodiversity—vertebrates (56%), invertebrates (30%), and plants (12%)—whereas the prior work was less balanced because it was dominated by vertebrate observations (>75%). An assessment of college involvement where we excluded the authors’ iNaturalist contributions resulted in a species richness estimate for our data of 249.9 ± 39.5 species (95% CI = 193.7–354.6 species), which produced similar diversity indices between the datasets (t = 0.39, df = 226, P = 0.69).

Analyses by taxonomic groups found similar patterns to the full dataset with several important distinctions, including the fact that the observed species accumulation curves did not reach asymptotes. For arthropods (phylum Arthropoda), rarefaction curves estimated higher species diversity with non-overlapping confidence intervals in our dataset compared to the past work (Fig. 2), which was supported by a significantly higher Shannon Diversity Index in our data (3.94 versus 2.51) (t = 9.83, df = 17, P < 0.001). Higher diversity in our work was also projected by curves for plants (kingdom Plantae) with a higher diversity index (3.05 versus 2.73) (t = 2.21, df = 21, P = 0.039) and herpetofauna (classes Amphibia and non-avian Reptilia) also with a higher diversity index (1.82 versus 1.04) (t = 3.17, df = 4, P = 0.034), but both had overlapping confidence intervals for extrapolated data. Birds (class Aves) were the sole exception with slightly higher richness estimates in the past data and a higher diversity index (3.34 versus 2.77) (t = 18.24, df = 145, P < 0.001), but confidence intervals also overlapped in this analysis. Estimated species richness using the Chao method by these four major taxonomic groups showed similar results to the rarefaction curves and community comparisons (Table 3).

An assessment of community composition metrics, as determined by the number of Research Grade observations, between time periods revealed that only the top two species were present in both datasets out of the top 10 most frequently observed species (Table 4). Among the top 10 most frequently observed species in the past time period, all were vertebrates, and included nine species of bird and one mammal. In comparison, the top 10 most frequently observed species in our study included four bird species, three invertebrates, two plants, and one fish. The rank frequency of observations curves between the time periods revealed that many more species in our data were detected more frequently (>10 observations) compared to the past data, which only had a single species detected >10 times (Fig. 3). For example, the top two species in the past data represented 20.2% of all observations in that dataset, whereas the same top two species made up 14.6% of all observations in our data. Quantitative differences between the rank frequency curves for the two time periods detected increases in our data relative to the past data for species richness (0.465), species rank (0.279), species composition (0.367), and overall curve difference (508.6), with only a single decrease which was detected for evenness (−0.245).

Discussion

Datasets built by citizen scientists will be important to understanding how global environmental changes, such as urbanization, will impact biodiversity worldwide. At a single site, we compared opportunistic citizen-science observations posted on iNaturalist (2008–2021) to semi-structured observations (i.e., BioBlitz events) posted over the course of a single active season (April–October 2021), equating to 48 h of directed surveys. In comparison to historical observations at Cedar Lake, our baseline data provided a clear improvement to our understanding of the biodiversity in and around the site. Below we discuss potential biases impacting citizen-science studies with illustrative examples from our own efforts. Ultimately, we emphasize the need for ongoing monitoring that uses systematic data collection for citizen science coupled with structured surveys and complementary census techniques to track future changes in biodiversity at this urban lake.

Our approach engaged more community members and produced higher biodiversity metrics than was found throughout several years of opportunistic observations recorded at Cedar Lake. Our findings showed that adding structure to citizen-science activities, coupled with digital biodiversity tools and college-student involvement, can generate more data than passive approaches, even over short timescales (Kelling et al., 2019). Several diversity metrics in our dataset were higher than in the prior work, indicating that significantly more biodiversity is present in the area than could have been realized with the past data alone. Our efforts, however, are still likely a vast underestimation of the true species diversity of Cedar Lake as nearly all observed species accumulation curves failed to reach an asymptote and citizen scientists observed just slightly more than half of the estimated species richness. Consequently, additional sampling that incorporates multiple techniques (Christie et al., 2019) in addition to spatially and temporally structured surveys (Kamp et al., 2016) is needed to produce a more complete biodiversity estimate of the site. Nonetheless, including semi-structured initiatives in citizen-science studies led by a local college appear to generate biodiversity assessments that are closer to an accurate representation of the community at a site over passive, opportunistic datasets (Gigliotti, Franzem & Ferguson, 2023). Because both datasets were not collected systematically, however, it remains unclear if at least some of the differences could be attributed to annual variation in species occupancy, variation in time frames, number of observers, or even preferences of observers. For example, some locations around the lake were apparently easier for observers to document biodiversity from, such as along the main trail (Fig. 1), which could be remedied in future studies with additional training at the outset of surveys. Lastly, we acknowledge that the contribution from the local college—where both authors were employed—cannot be overlooked as citizen involvement in isolation would not have produced the same diversity metrics, thus higher education can be an important catalyst in societal efforts to track local species.

The past opportunistic data from Cedar Lake on iNaturalist did exhibit a more even community than our dataset, however, all other diversity metrics were greater in our work. The greater evenness was because more species were observed only once in the past data (64% were singletons) compared to our work (52% were singletons). Regardless, several aspects of the community composition captured in the past data suggested that opportunistic approaches to biodiversity monitoring with citizen scientists are plagued with biases, often associated with a lack of formal ecological training, that may render such datasets difficult to use (Courter et al., 2013; Kamp et al., 2016; Callaghan et al., 2020). For example, nearly 70% of the Research Grade observations and 55% of the species in the past dataset were of birds, whereas birds represented about 49% of the observations and just 21% of the species in our data. Furthermore, analyses indicated that bird diversity was higher in the past data compared to our work, which is likely due to missing migratory or otherwise uncommon species. For example, the past data included 14 more duck species (family Anatidae) than our data (17 versus three species), most of which were winter migrants whose temporal occupancy of the lake would not have overlapped with our sampling period. Birds are often the most conspicuous vertebrates in many habitats and these results suggest that unstructured citizen-science observations tend to be biased towards such taxonomic groups (Horns, Adler & Şekercioğlu, 2018; Callaghan et al., 2021). Less conspicuous vertebrate groups, such as herpetofauna, tend to receive much less attention unless surveyors are explicitly targeting these taxa (Troudet et al., 2017), an observation that is borne out in our data. Less than 2.5% of observations in the past data were of reptiles and amphibians, which included just three species, whereas these groups represented more than 5% of the observations in our data and included nine species. Despite the popularity of fishing at Cedar Lake, only a single observation of one fish species was recorded in the past data compared to 57 observations of 14 species in our data. One of the largest increases in biodiversity representation for animals derived from our semi-structured citizen-science activities were found among the invertebrates, especially mollusks and arthropods. For example, not a single mollusk observation was recorded in the past data, compared to 10 in our work (albeit of a single species). Similarly, among the arthropods, class Insecta was just 10% of the past observations compared to this group representing nearly 30% of our observations. Plants were similarly represented in terms of proportion of overall observations in both datasets. Nevertheless, only 16 species of plants were detected in the past data compared to 41 in our data. Overall, directed activities incorporated into citizen-science surveys appear to facilitate improved biodiversity data compared to passive observations as they may help reduce some of the biases inherent to citizen science (Stevenson, Merrill & Burn, 2021), especially related to the tendency of individuals to overemphasize observations of conspicuous taxonomic groups (Mair & Ruete, 2016; Troudet et al., 2017).

In 2019, the city of Cedar Rapids and ConnectCR articulated master plans (https://connectcr.org/cedar-lake-master-plan) to convert Cedar Lake from an industrial cooling site into a recreational hub through improvements to address water quality, including mitigating stormwater runoff, addressing sediment buildup, installing flood control features, and adding new trails and bridges (Morelli, 2019). Other suggested improvements for Cedar Lake included accessible boat launches, fishing piers, an obstacle course, enhanced fishing amenities, a boardwalk over the lake, and an enhanced wetland area (White, 2021). Plans to transform the lake were carried out under provisions to assess the quality of the water with respect to studying the lake floor sediment for toxins (Morelli, 2023) and redirecting flow regimes (Iowa Department of Natural Resources, 2022). Moreover, an environmental assessment of Cedar Lake included in the Cedar Rapids flood risk management project concluded a finding of “no significant impact”, with most biodiversity concerns focused on four endangered species that may have occurred in the region and the productivity of altered wetlands for aquatic species (US Army Corps of Engineers, 2019). The groundbreaking ceremony for this renovation occurred 7 October 2021, around when our project was completed. Using the time tool on Google Earth, we were able to visualize clear physical changes to habitats contiguous with the lake in satellite imagery nearly one year after the groundbreaking ceremony (1 September 2022) (Fig. 4). In particular, construction activities in the northwest corner of Cedar Lake resulted in the removal all woody vegetation in the area, the production of a 2,200 ft levee, and a complete alteration the lake’s outflow through McLoud Run, a creek that serves as drainage for the lake (Fig. 4, inset). These satellite images were captured to show that changes to the lake began after our project, and they can be used by future researchers to document spatial variation in biodiversity changes in relation to such habitat changes. It remains unknown what long-term biodiversity impacts these significant structural changes will have, but such urban land-use changes generally decrease non-avian vertebrate diversity while increasing plant diversity from the importation of non-native species (McKinney, 2008). For birds and arthropods, urban modifications often reduce richness and diversity, but increase abundance due to the dominance of a few synanthropic species (McKinney, 2008; Faeth, Bang & Saari, 2011).

At the moment, we do not know the resilience of the local species we documented, nor how they will respond to changes such as a new hydrological regime, and we know very little about their ability to migrate to other areas or even recolonize the lake while its habitats regenerate. However, we can leverage the baseline data herein, combined with the historical data, to predict what changes may occur to the biodiversity of Cedar Lake and then conduct follow-up studies at different time intervals post construction to test whether survey data support or reject such predictions. We note that management plans to modify natural or even semi-natural areas could be improved by conducting more rigorous preliminary surveys before construction to establish the baseline data needed to monitor changes over time and to determine potential impacts on resident biodiversity (Underwood, Francis & Gerber, 2011; Rojas et al., 2022), many of the consequences of such changes can be analyzed in the future. Ultimately, without reliable baseline data collected before major disturbances, it could never be possible to understand such impacts on biodiversity, let alone test predictions about what could happen in the future.

Supplemental Information