Synthesis Science Course Development

Nick J Lyon

Outline

• Long Term Ecological Research Network Background
• Synthesis Course Motivation
• Strategic Value & Audience
• Development Approach
• Lessons Learned

To check out the course content, visit lter.github.io/ssecr.

LTER Network

• NSF-funded

• Established in 1980

• 26 sites currently in the network
  • & LTER Network Office

• Thousands of investigators

• Hugely productive!
  • >7,000 public datasets
  • >17,000 journal articles

LTER Core Research Areas

• – Primary Production
• – Populations & Communities
• – Disturbance
• – Organic Matter
• – Inorganic Nutrients
• – Land Use/Cover Change
• – Human-Environment Interactions

LTER & Synthesis

• LNO funds synthesis working groups
  • ~2-4 / year

• For LTER, “synthesis” means:

Integrating existing data from many sources in order to test novel hypotheses

• Some (not all) working group data must be from LTER sites

• Groups are immensely enthusiastic, curious, and productive

Working Group Process

Figure from Elmendorf et al. 2026.

Working Group Support

My primary role as the LNO data scientist is to support working groups

• – Offer advice on reproducibility, organization, coding, etc.

• – Create & offer workshops & trainings

• – Write/review code for data harmonization, QC, analysis, visualization, dashboards, etc.

• – Advise on appropriate statistical tests and interpret results

• – Draft relevant pieces of peer-reviewed manuscripts

For more details on data scientist support of working groups, visit lter.github.io/scicomp.

Lyon & Working Groups

• From 2022-26, I supported 180+ researchers across 17 working groups
  • Often warranted inclusion as a co-author

• Over time, developed insight into human dimension
  • Actions happy/successful groups take to remain so
  • Team dynamics obstacles and how to solve them

• Now offer interpersonal support in addition to data science services
  • E.g., discussing intellectual credit policies, navigating within-team conflict, facilitation of tech-adjacent discussions

Working Groups & ECRs

• Nearly all groups include early career researchers
  • Typically graduate students and post-docs
  • Sometimes undergraduates!

• ECRs get a ton out of participating
  • E.g., networking, leadership opportunities, publications, job opportunities

• Groups get a lot out of including ECRs as well!
  • E.g., novel analytical methods, new perspectives, coding/technical literacy

Synthesis Training & ECRs

• ECR demand for synthesis training far outstrips working group capacity

• Inspired creation of a course: Synthesis Skills for Early Career Researchers

• SSECR aims to train ECRs in needed technical and interpersonal skills
  • Synthesis science is inherently collaborative
  • ECRs frequently indicated they weren’t formally trained in either sphere

• Timed to leave students well-prepared to apply for LTER working group funding
  • Scoped to be useful beyond LTER format of working groups

SSECR & LTER Strategic Plan

Audience

Cohort #1 (2024-25)

• 25 students (Master’s / PhD / post-doc)
• Were actively working at an LTER site
  • Or (recent) past work at one
• Commit to 3 hours biweekly for 9 months

Cohort #2 (2026)

• 29 students (Master’s / PhD / post-doc)
• Active/past work at a long-term site
  • Not necessarily “LTER”
• Commit to 2 hours weekly for 4 months

• Wanted students with some prior research experience
  • No requirement for synthesis experience

SSECR Modality

• SSECR is/was offered synchronously (mostly online)
  • But, having materials be free and publicly-available was a priority

• Built to maximize utility for asynchronous & synchronous learners
  • Regardless of LTER Network connection

• This ethos shaped several structural elements of the course
  • Module content and organization
  • Preference for website pages over slide decks
  • Activity & discussion prompt structure

Development Team

• 3 course leads: Marty Downs, Carrie Kappel, Nick Lyon

• 5 additional design team members

• 3 additional contributors (narrower focus on content editing/feedback)

• 19 panelists / guest speakers

Additional contributors & panelists/guest numbers from 2024-25 cohort only; TBD for 2026.
For full contributor information, visit lter.github.io/ssecr/instructors.

Course Development Plan

1. Define project management framework for tackling a project of this scope

2. Scope & draft 14, quasi-independent “modules”
   • Cover fundamental elements of team-based synthesis science

3. Plan & support student-lead synthesis projects
   • Facilitate students in coalescing into teams around actionable, exciting projects

2024-25 Module List

Team Science

• Team Sciences Practices
• Inclusive Facilitation
• Project Management
• Supporting Divergent, Emergent, and Convergent Thinking
• Authorship & Intellectual Credit
• Next Steps & Logic Models
• Communicating Findings

Data Science

• Version Control
• Reproducibility Best Practices
• Data Discovery & Management
• Data Visualization & Exploration
• Data Wrangling
• Analysis & Modeling
• Notebooks & Website Creation
• Bonus: Spatial Data
• Bonus: Interactive Apps with Shiny

Lyon helped design; Lyon fully designed; Lyon designed & taught.

2026 Module List

• Move away from data vs. team science dichotomy
  • Massive refactoring effort
  • Forked before editing to preserve live 2024 version

• Still 14 modules but each features a blend of skills
  • Team & data science instructors paired for each

• While refactoring, also changed structure to increase course link stability

To check out the 2024 version, visit lter.github.io/ssecr-2024.
For a granular view of what changed between cohorts, see github.com/lter/ssecr/pull/68.

So, How Did We Do It?

Development Tool Overview

• Git - version control

• Quarto - website development

• GitHub - website hosting, continuous integration for website rendering (GitHub Actions), project management, team coordination

• R - embedded code demos

• Google Drive - internal documentation, course in/outputs

• No specific required integrated development environment for instructors
  • Most used Positron, VS Code, or RStudio

Version Control

• Onboarded design team to Git/GitHub use & best practice
  • Encouraged to adopt Conventional Commits-style messages

• Mostly working in main branch
  • Accessible to largest subset of design team
  • Branch + Pull Request for major updates

• CONTRIBUTING.md for collaboration plan

• NEWS.md + GitHub Releases for DOI (from Zenodo)

Check out the primary course repository at github.com/lter/ssecr.
For more on Conventional Commits, see here.

Project Management

• Extensive adoption of GitHub issues and milestones

• Further organization enabled by a GitHub project

• Team used issues & issue metadata to:
  • Delegate or trade tasks
  • Discuss & document judgment calls
  • Standardize pedagogical approach among modules
  • Standardize format elements among modules

Check out SSECR GitHub issues here; #1 nicely aligns with this slide!

Website Creation/Maintenance

• Quarto is a computational notebook system
  • From makers of Positron; successor to RMarkdown
  • Akin to Jupyter Notebooks

• Supports highly flexible website design
  • Low learning curve

• Combined with GitHub Action for re-deployment of site
  • Allows direct editing of pages in GitHub

• Easily supports embedded code chunks for applied tutorials
  • Cool code annotation features within code chunks

Student Group Support

• First cohort yielded four projects
  • Carbon Cycling Responses Across Spatiotemporal Scales
  • Synchrony of Above & Belowground Responses Across Ecosystems
  • Diversity-Stability Relationships
  • Effects of Environmental Drivers Across Ecological Scales

• Offered working group-level support
  • Create GitHub repository (with custom .gitignore)
  • Advise on organization, reproducibility, statistical approach, etc.
  • More code review, less direct code writing
  • Offer tutorials extending beyond course content

For more on student projects (including team GitHub repositories & websites), check out lter.github.io/ssecr/projects.

Student Group Outcomes

1. Increased student skill and confidence with team synthesis

2. Adoption of version control / reproducibility best practices into non-synthesis work

3. Strong within-cohort sense of community

4. Several peer-reviewed papers in progress

5. One currently-funded LTER working group (so far!)

Lessons Learned

Technical

• Project management schema slightly over-engineered
• Needed increased ‘modularity’ within course modules
• Needed more discussion of AI (esp. generative AI)
  • Guiding principles rather than specific (often quickly-superseded) tools

Interpersonal

• Mutual student-teacher trust was pre-requisite to course success
• Should have paid more attention to design team capacity & availability
  • Could increase emphasis on reproducible, low-friction contribution methods
• Participant feedback invaluable to process

(Hopefully not ironic to separate this slide into a dichotomy that SSECR tries to avoid )

Thanks! Questions?

Photo credit: Nick J Lyon

Handing Over the Reins

In order to transition project leadership, a colleague would need to:

• Explore the process documents in the Git repo & Google Drive
  • E.g., CONTRIBUTING.md, contents of “notes/” folder

• Look at the GitHub issues on the course repository

• Ask questions of the rest of the design team
  • Project management schema developed collaboratively
  • So all design team members are able to onboard new collaborators