Advancing Amphibian Conservation through Citizen Science in Calgary (Diversity 2021)

Introduction

• Topic: Advancing amphibian conservation through citizen science in urban municipalities, focused on Calgary, Alberta, Canada.
• Authors and context: Tracy S. Lee et al.; case study within urban biodiversity planning and citizen science implementation.
• Rationale: Cities seek to protect urban biodiversity; need species-specific data on urbanization effects to inform planning and policies; amphibians as sensitive indicators of environmental health due to biphasic life cycles.
• Goals of the program:
  • Systematically monitor amphibians in a large city (Calgary) with substantial wetland loss (90% wetlands lost since settlement).
  • Demonstrate how publicly collected data can inform biodiversity monitoring and urban design.
  • Outline design lessons for citizen science programs, including engagement with decision makers, quality control, data/public engagement trade-offs, and conservation messaging.
• Key claims: Citizen science can improve ecological knowledge, engage the public in urban biodiversity monitoring, and influence planning and policy when programs are designed with quality control and decision-maker involvement.
• Definitions and scope:
  • Urban ecology studies: species-specific responses to urbanization; broader public engagement.
  • Amphibians as biodiversity indicators and early signals of environmental degradation.
  • Public engagement benefits: improved eco-literacy, behavioral change, social cohesion, trust, and potential civic actions.
• Data and methods overview: Systematic amphibian surveys with simple methods, quality control via autonomous recording units (ARUs), occupancy modeling, and participant surveys to guide program design.

Methods

2.1. Study Area

• Location: Calgary, Alberta, Canada; over 1.2 million people; area ~824.5 km².
• Context: Heavily urbanized, with two major rivers (Bow and Elbow).
• Wetland loss: ~90% since European settlement; remaining wetlands are largely stormwater structures rather than biodiversity-focused.
• Historically recorded amphibian species in city boundary: boreal chorus frog (Pseudacris maculata), wood frog (Lithobates sylvaticus), tiger salamander (Ambystoma mavortium), northern leopard frog (Lithobates pipiens), Canadian toad (Anaxyrus hemiophrys), western toad (Anaxyrus boreas).
• Study aim within Calgary: Determine which of the six historically observed species persisted and engage the public in conservation.

2.2. Monitoring Amphibians

• Site selection: 52 wetland sites chosen from 200 randomly selected wetlands based on municipal/provincial ownership; sites publicly accessible.
  • One wetland site was lost to development during the first year.
• Wetland types: Constructed stormwater ponds, modified and natural wetlands; permanent wetlands favored for ease of access and relevance to stormwater management.
• Survey design: Simple, systematic surveys conducted April–August over three years (2017–2019).
• Survey schedule: Nine survey time periods per season to ensure systematic effort; combined into five time periods for analysis due to overlaps.
• Survey methods:
  • Daytime visual surveys (eggs, larvae, juveniles, adults).
  • After-dusk auditory surveys (frog/toad calls) for vocal species; tiger salamanders documented via visual observation since they do not vocalize.
  • Data capture: Species observed, observation type, health status, photos/voice recordings uploaded to a smartphone app; opportunistic observations also encouraged.
• Training and participation:
  • Field orientations offered annually for new citizen scientists; advisory committee provided input on methodology and an identification guide with images/ recordings.
  • Participants could sign up for multiple wetlands and survey periods; two independent sign-ups per period; summer intern used for wetlands with low participation.
• Data collection goals: Build presence/absence data across sites and time to support occupancy analyses and habitat associations.

2.3. Quality Control

• Rationale: Ensure data quality for occupancy modeling and to test false positives/negatives.
• Two forms of observation error:
  • False positives: reporting a species not present.
  • False negatives: failing to report a species that is present.
• Quality control approach: Compare citizen science observations to ARU data for vocal species (wood frog and boreal chorus frog).
• ARUs used: Eight wetlands in 2017 and 2018; ARUs included multiple Song Meter models (SM4, SM2+, AF1); recordings every hour for 10 minutes; one site repeated across years.
• Data processing: Recordings classified by student bioacousticians; a professional bioacoustician performed cluster analysis to validate peak calling periods (10:00 p.m. – 1:00 a.m.).
• Data integration: Combine ARU results with citizen science classifications to create a comprehensive acoustic dataset for comparison.
• Validation technique: Confusion matrix comparing ARU detections (actual) with citizen science detections (predicted) across five time periods; used Correct Classification Rate (CCR) as a performance metric.
• CCR definition:
  • ext{CCR} = \frac{\text{Number of correct predictions}}{\text{Total predictions}}
• Correct classifications and findings:
  • Boreal chorus frog CCR = 82%;
  • Wood frog CCR = 44%;
  • False positives: none for either species, which is favorable for occupancy modeling that assumes no false positives.
• Detection probabilities (single survey p):
  • Boreal chorus frog: citizen scientists p = 0.56 \pm 0.07; ARUs p = 0.90 \pm 0.04.
  • Wood frog: citizen scientists p = 0.07 \pm 0.04; ARUs p = 0.59 \pm 0.08.
• Implications: ARUs provide higher detection probability, especially for wood frogs; citizen scientists may miss low-activity species or those with low call rates.
• Additional notes: Wood frog detections occurred at 12 of 16 ARU wetland sites; citizen scientists failed to report wood frogs at nine sites (six with low calls
• Visual representation: Figure 3 illustrates ARU detections vs. citizen science detections across sites.

2.4. Occupancy Modeling

• Objective: Compare detection probabilities between citizen scientists and ARUs for two species (wood frog and boreal chorus frog).
• Modeling approach: Single-season occupancy models (PRESENCE 2.13.10) using 15 wetlands; first-year data used for the one site with two-year data.
• Data aggregation: Nine survey periods per season collapsed into five monthly periods (April–August).
• Spatial buffering: Sites within 500 m considered as a unit unless a major road/residential development intersects the buffer; total sites used = 42 for occupancy modeling.
• Exclusions: Tiger salamander not analyzed due to too few observations.
• Habitat covariates (occupancy ψ):
  • Wood frogs: distance to forest as a covariate on occupancy ψ.
  • Boreal chorus frogs: proportion of manicured land within a 20-m buffer as a covariate on occupancy ψ.
• Data transformations:
  • Proportion manicured land transformed with a logit transformation:
  • \text{logit}(\text{manicured land proportion}).
  • Distance to forest standardized: \text{distance}_{\text{std}} = \frac{\text{distance} - \mu}{\sigma}.
• Detection covariate on probability of detection (p): method of observation (ARU vs. citizen science).
• Occupancy model results (key findings):
  • Wood frog occupancy higher when forests are near (log-likelihood statistic: -2 \times \log L = 100.47, K=4).
  • Boreal chorus frog occupancy higher with greater manicured land in the 20 m buffer (log-likelihood -2 \times \log L = 116.14, K=4).
• Data interpretation: Occupancy models help identify habitat associations and support habitat connectivity models for Calgary; follow-up analyses planned/published separately.
• Cumulative detection probability: For comparing survey effort across methods, the study uses:
  • P^* = 1 - (1 - p)^K
  • where P^* is the cumulative probability of detection on a site over K surveys, and p is single-survey detection probability from the occupancy model.
• Model comparison: Models including survey type (ARU vs. citizen science) had higher AICc; specifically, >12 ΔAICc compared to models without survey type, indicating detection method significantly influences detection probability and sample effort required.
• Practical implication: To achieve comparable detection for boreal chorus frogs, roughly 3 citizen-science surveys are needed to match ARU detection; for wood frogs, approximately 12 surveys are needed.

2.5. Participant Engagement

• Objectives: Engage the public in amphibian conservation; measure engagement via smartphone app downloads, data collection, and program events; enhance knowledge and civic participation in biodiversity conservation.
• Recruitment: Through partner networks (conservation groups, universities) and immigrant organizations; events and resource sharing used to sustain engagement.
• Participant metrics (2017–2019):
  • Total engaged: 546 individuals (measured by app downloads over the project duration).
  • Average annual contributions: about 51 citizen scientists per year (2017: 48; 2018: 59; 2019: 47).
  • Estimated total data contributors per year: at least 100 (participants typically worked in pairs).
  • Observations submitted: 1116 amphibian and null observations across three years.
  • Wetland surveys per site: average 6.4 surveys annually (target was 9 surveys per wetland site per year).
• Survey participation details: 40 respondents to a 2018 participant survey (target 150 responses for 5% margin of error); results not fully representative of all participants but provided insights.
• Key motivational findings (from survey):
  • Motivations to participate included a desire to exercise, meet others, learn, volunteer, connect with nature, contribute to science, and personal interest in the topic.
  • Most respondents (65%) were already interested in environmental issues and used participation to act on that interest.
  • 95% reported speaking to others about the program and recruiting others; sharing of specific knowledge about urban wetlands or municipal planning issues was less common.
• Information sharing and messaging: Participants frequently shared general program information but less often shared detailed conservation messaging about wetlands, amphibian conservation, or municipal planning issues.
• Conservation messaging integration: Messaging and civic actions were incorporated into public events and materials in the final year to address urban biodiversity loss and to clarify pathways for citizen influence on planning processes.
• Data availability: Amphibian data available on request; acknowledgments to volunteers and interns; IRB not applicable; no informed consent required.

Results

3.1. Amphibian Ecology

• Documented species in Calgary during the study: boreal chorus frog, wood frog, tiger salamander.
• Species not observed from historical six: northern leopard frog, Canadian toad, western toad.
• Observed prevalence across sites:
  • Boreal chorus frog at 85% of surveyed sites.
  • Wood frog at 36% of surveyed sites.
  • Tiger salamander at 21% of surveyed sites.
• Species co-occurrence: Most sites had 1–2 species; only 3 sites showed co-occurrence of all three species.
• Breeding evidence: 12 sites reported eggs/tadpoles or larvae, representing 29% of the urban wetland sites surveyed.
• Occupancy modeling implications: The amphibian dataset enabled habitat associations for boreal chorus frog and wood frog and contributed to city-scale connectivity models; results supported planning and conservation prioritization. Specific occupancy results:
  • Wood frog occupancy higher near forests.
  • Boreal chorus frog occupancy higher with greater manicured land proportion in a 20 m buffer.
• Figure reference: Figure 2 maps amphibian species richness per wetland across Calgary (2017–2019).

3.2. Quality Control and Assurance

• Summary of key results:
  • CCR for boreal chorus frog: 82%
  • CCR for wood frog: 44%
  • False positives: none for either species
  • Wood frog detections: ARUs detected at 12 of 16 wetlands; citizen scientists missed at 9 sites (6 with
• Details from Table 1 (confusion matrix):
  • Boreal chorus frog: True Positives 11; True Negatives 2; False Positives 0; False Negatives 3
  • Wood frog: True Positives 3; True Negatives 4; False Positives 0; False Negatives 9
• ARU vs. citizen science comparison: ARUs had higher single-survey detection probabilities for both species, especially wood frogs.
• Site-level patterns: Wood frog detected at 12 ARU sites; citizen scientists failed to report at several sites with low calls; one site had high ARU activity but no citizen detections.
• Interpretation: Citizen science is valuable for boreal chorus frogs (higher CCR) but underestimates wood frogs due to detection challenges; highlights need for mixed-method approaches and sufficient survey effort, especially for low-call species.
• Figure 4: Cumulative probability of detection comparing citizen scientists (circles) and ARUs (squares) for both species.

3.3. Participant Engagement Survey

• Engagement metrics: 546 participants; average annual contributions ~51; at least 100 participants contributed data each year; 1116 observations across three years.
• Survey response results:
  • Motivations to register: exercise, meeting others, learning, networking/career, volunteering, time in nature, contributing to science, personal interest.
  • Information shared: various topics; open-ended responses coded for themes. Participants commonly shared general program information; specifics on urban wetlands or municipal planning were less common.
• Interpretation: The program successfully attracted environmentally interested individuals and created avenues for public engagement, but targeted messaging and broader diffusion of conservation knowledge required further refinement.

Discussion

4.1. Enhanced Amphibian Ecological Knowledge

• Contribution to urban ecology knowledge: Documented amphibian species diversity in Calgary, including breeding evidence in urban wetlands and habitat associations.
• Absence of some historic species: Absence of northern leopard frog, Canadian toad, and western toad highlights shifts in urban amphibian communities and potential reintroduction considerations.
• Habitat associations: Wood frogs strongly associated with forests; preserving forests near wetlands is important for maintaining wood frog populations.
• Broader implications for urban planning:
  • Calgary’s expansion and land-use changes threaten remaining natural landscapes; occupancy and habitat models can inform prioritization for protection and restoration.
  • Habitat connectivity models at city scale can guide planning and conservation actions.
• Practical relevance: Models and data support prioritization of wetland protection, restoration, and forest conservation within urban landscapes.

4.2. Lessons on Citizen Science Program Development

4.2.1. Engage Decision Makers Early

• Early involvement of municipal staff was critical to align program objectives with planning, park management, water management, and biodiversity policy.
• Examples of possible integrations: Citywide plans, biodiversity strategies, urban forest plans, wetland conservation plans, riparian action programs, and environmental reserve guidelines.
• Outcome: Enhanced likelihood that amphibian data would inform existing plans and management actions.

4.2.2. Data Quality Assessment

• Rationale: Traditional validation (comparing citizen data to professional scientists) has limitations; ARU data provide an independent benchmark of detection.
• Approach: Use ARU detections to assess false negatives/positives and thus data quality for occupancy modeling.
• Findings: ARUs yielded higher detection probabilities; citizen science data were reliable for boreal chorus frogs but less reliable for wood frogs due to low call rates and detection challenges.
• Implications for program design:
  • Mixed-method monitoring (including ARUs) improves data quality and helps calibrate citizen science effort.
  • Determine appropriate sampling intensity and observer training to achieve monitoring objectives.

4.2.3. Balancing Data and Engagement Tensions

• Tension: Flexible citizen science participation aids engagement but can reduce sampling uniformity and data representativeness.
• Design response: Consider assigning participants to fixed wetlands and requiring commitment to multiple surveys to improve sampling effort; otherwise, uneven survey effort across wetlands emerges.
• Resource strategies: Hire interns to cover wetlands with low participation; adjust survey periods to five time blocks to satisfy data analysis needs.
• Outreach and inclusion: Diversify wetlands (location types) to broaden participation; address access and language barriers to recruit underrepresented groups.
• Community involvement: Frameworks should involve community priorities and include multiple knowledge types to enhance legitimacy and relevance.

4.2.4. Conservation Messaging

• Recruitment observation: Early participation attracted individuals already interested in amphibians or biodiversity; messaging did not always emphasize conservation impacts.
• Strategy adjustments: After year two, targeted outreach to new immigrant communities; embed conservation messaging into events and materials; provide clear civic actions and guidance on navigating municipal planning processes affecting wetlands.
• Outcomes and limitations: While citizen science participation supports data collection, measuring direct conservation actions from participation remains challenging; evidence from other studies suggests benefits in civic engagement can follow participation in citizen science.
• Recommendations for future programs: Explicitly integrate conservation messaging into communications, identify target audiences, and invest in relationship-building with identified communities.

Formulas, parameters, and key equations

• Occupancy-detection framework:
  • Occupancy probability: \psi
  • Detection probability (per survey): p
  • Single-season cumulative detection probability over K surveys:
  • P^* = 1 - (1 - p)^K
• Covariates and transformations:
  • Wood frog occupancy covariate: distance to forest (standardized) → \text{distance}_{\text{std}}
  • Boreal chorus frog occupancy covariate: proportion of manicured land in a 20 m buffer → transformed via logit: \text{logit}(\text{manicured land proportion})
• Standardization and transformations:
  • Distance to forest: \text{distance}_{std} = \frac{\text{distance} - \mu}{\sigma}
  • Manicured land proportion: apply \text{logit}(x) = \log\left(\frac{x}{1 - x}\right) where x is the proportion.
• Model selection metric:
  • AICc comparison; models including survey type (ARU vs. citizen science) had significantly higher ΔAICc compared to models without survey type, indicating detection method effects.

Connections and implications

• Real-world relevance: Demonstrates how urban citizen science can generate defensible, policy-relevant data to inform city planning, biodiversity strategies, and wetland management.
• Foundational principles: Public engagement in science can complement traditional monitoring when quality control and decision-maker involvement are integrated.
• Ethical and practical implications:
  • Balance between broad public participation and rigorous data quality must be managed.
  • Early stakeholder engagement increases likelihood that outputs influence policy and planning.
  • Messaging should be integrated into outreach to encourage civic action and understanding of urban biodiversity challenges.

Key takeaways

• Urban citizen science can yield valuable data on amphibian ecology and habitat associations when coupled with robust quality control (ARU validation) and occupancy modeling.
• Decision maker engagement from the outset improves translation of results into planning and policy.
• There is a clear trade-off between maximizing public engagement (more flexible participation) and achieving rigorous, uniform data collection; mixed-method approaches and targeted design can help balance these goals.
• Conservation messaging should be embedded in outreach to mobilize civic support and actionable steps for urban biodiversity protection.

References to figures and tables (mentioned in the transcript)

• Figure 1: Map of wetlands surveyed by citizens in Calgary (2017–2019).
• Figure 2: Map depicting the number of amphibian species observed at Calgary wetlands (2017–2019).
• Figure 3: ARU detections for wood frog at 12 wetlands; sites where citizen scientists missed detections indicated.
• Figure 4: Cumulative probability of detection for boreal chorus frogs and wood frogs comparing citizen scientists vs ARUs.
• Figure 5: Motivations to register for the program (survey data).
• Figure 6: Topics shared by participants about the program (open-ended responses).
• Table 1: Confusion matrix categorizing true positives/negatives and false positives/negatives for boreal chorus frog and wood frog comparisons between citizen scientists and ARUs.
• Appendix/References: Includes detailed methodology sources (e.g., Sensitive Species Inventory Guidelines, occupancy modeling references) and program materials.