Ecology & Biodiversity Research

Reasoning Strategy

1. Species & Taxonomy Questions

When a question involves identifying or comparing species:

1. LOOK UP DON'T GUESS — Use GBIF_search_species to get taxonomy, WoRMS_search_species for marine organisms
2. If the question asks about invasive species impacts, consider: ecological niche overlap, reproductive rate, predator release, and ecosystem engineering effects
3. Use PubMed_search_articles or EuropePMC_search_articles to find studies on specific ecological impacts

2. Invasive Species Impact Assessment

Reasoning framework — when comparing invasive species impacts:

1. Identify the ecosystem: What habitat/biome is affected?
2. Assess impact mechanisms: Competition? Predation? Disease vector? Habitat modification? Hybridization?
3. Scale of impact: Local (single site) vs regional vs continental?
4. Trophic position: Invasives at higher trophic levels (predators) often cause more damage than lower (herbivores)
5. Ecosystem engineering: Species that modify habitats (beavers, earthworms, honeybees displacing native pollinators) cause outsized impacts
6. Look up specifics — don't rely on general knowledge. Search for "[species name] invasive impact [region]" in literature

3. Pollinator Ecology

Reasoning framework for pollination questions:

1. Foraging behavior: Distinguish investigation (approach/assessment) from actual feeding (proboscis insertion)
2. Interaction types: Mutualistic (pollination reward), parasitic (nectar robbing), commensal
3. Observation methods: Camera traps have resolution/FOV limitations — consider what's identifiable at given resolution
4. Statistical considerations: Observer agreement (inter-rater reliability), sampling effort, temporal patterns
5. Ethogram interpretation: Each behavior category has specific start/end criteria — follow them precisely

4. Population Dynamics

Reasoning framework for population ecology questions:

1. Growth models: Exponential (unlimited), logistic (K-limited), Allee effects (low-density problems)
2. Extinction analysis: Distinguish deterministic extinction (r < 0) from stochastic extinction (small population fluctuations)
3. Survival analysis: Time-to-event analysis needs appropriate statistical tests (log-rank, Cox regression, Kaplan-Meier)
4. Microbial ecology: For microbial stressor responses, use survival curve analysis with time-kill kinetics. To compare extinction points between populations, you need time-to-extinction data analyzed with survival statistics (not just endpoint comparisons)

5. Community Ecology & Food Webs

1. Trophic cascades: Removing top predators → mesopredator release → prey decline
2. Keystone species: Disproportionate impact relative to abundance
3. Island biogeography: Species-area relationship, distance-colonization tradeoff
4. Competitive exclusion: Two species cannot stably coexist on single limiting resource (Gause's principle)

6. Evolutionary Ecology

1. Aposematism: Warning coloration signals toxicity/unpalatability
2. Mimicry: Batesian (harmless mimics dangerous) vs Mullerian (dangerous mimics dangerous)
3. Life history tradeoffs: r-selected (many offspring, low investment) vs K-selected (few offspring, high investment)
4. Birth-death models: For phylogenetic questions, identifiability issues arise with time-varying rates. Strategies to resolve: constrain rate variation, add fossil data, use molecular data calibration, or restrict to specific functional forms

Available Tools

Navigating the GBIF taxonomic tree

Resolve a name to a GBIF usageKey once, then navigate the Backbone tree:

key = tu.run_tool("GBIF_match_name", {"name": "Panthera leo"})["data"]["usageKey"]  # 5219404
tu.run_tool("GBIF_get_taxon_parents", {"taxon_key": key})        # Animalia→...→Felidae→Panthera
tu.run_tool("GBIF_get_taxon_synonyms", {"taxon_key": key})       # Felis leo Linnaeus, 1758, ...
tu.run_tool("GBIF_get_vernacular_names", {"taxon_key": key, "language": "eng"})  # Lion, African Lion
# Walk down from a genus key (Panthera = 2435194) to its species:
tu.run_tool("GBIF_get_taxon_children", {"taxon_key": 2435194, "limit": 8})
# Normalize an authored name string without a key:
tu.run_tool("GBIF_parse_name", {"name": "Quercus robur L."})     # canonicalName 'Quercus robur'

All five tools hit the public GBIF API with no key. Get the starting taxon_key from GBIF_match_name or GBIF_search_species.

Additional Taxonomy & Biodiversity Sources

GBIF/WoRMS/BOLDSystems/iDigBio (above) are the primary sources. These eight add authority-specific, phylogenetic, and observational coverage GBIF alone doesn't provide — use them when GBIF is ambiguous, when you need a phylogenetic tree rather than a flat classification, or when you need citizen-science occurrence density rather than museum-specimen records.

Choosing among the identity/taxonomy sources (EOL, ITIS, CoL, GBIF all answer "what is this species"):

• ITIS if you need the North American regulatory-standard identifier (tsn).
• CoL for the broadest global checklist coverage when GBIF's backbone misses a match.
• EOL when you want an aggregated page (images, multiple hierarchies, curated collections) rather than a bare taxonomic record.
• Open Tree of Life only when the question is genuinely phylogenetic (MRCA, branch order, a Newick subtree) — none of the others return evolutionary relationships.
• Cross-check IDs are NOT interchangeable: the same species has a different key in each system (GBIF usageKey, ITIS tsn, CoL taxon_id, EOL page_id, Open Tree ott_id, iNaturalist taxon_id, OBIS AphiaID) — always resolve within one system, don't mix an ID from one database into another's lookup call.

Example: cross-checking a species across four identity systems (verified live, real IDs):

tu.run_tool("EOL_search_species", {"query": "Panthera leo"})            # page_id 1270491
tu.run_tool("ITIS_search_by_scientific_name", {"scientific_name": "Panthera leo"})  # tsn 183803
tu.run_tool("CoL_search_species", {"q": "Panthera leo"})                # 293 CoL-database matches
tu.run_tool("OpenTree_match_names", {"names": "Panthera leo,Panthera tigris"})
# -> ott_id 563151 (leo), 42314 (tigris); note `names` is a comma-separated
# STRING, not a JSON array, even though it takes multiple names

Example: phylogenetic MRCA and subtree (verified live):

tu.run_tool("OpenTree_get_mrca", {"ott_ids": "417950,770315"})
# -> mrca_name "Homininae", mrca_ott_id 312031, num_tips 19
tu.run_tool("OpenTree_get_induced_subtree", {"ott_ids": "417950,770315,312031"})
# -> real Newick tree with supporting_studies citations (pg_2741@tree6645, ...)

Example: marine species + region (verified live):

tu.run_tool("OBIS_search_taxa", {"scientificname": "Rhincodon typus"})   # whale shark occurrence-resolved AphiaID
tu.run_tool("MarineRegions_search_by_name", {"name": "Mediterranean Sea"})  # MRGID 25180

Example: citizen-science occurrence density (verified live):

tu.run_tool("iNaturalist_search_taxa", {"query": "Panthera leo"})
# -> genus-level hit "Panthera" (59,036 observations) if the exact binomial
# isn't the top match — check the returned `rank` before assuming species-level

LOOK UP DON'T GUESS

Ecology questions often have counter-intuitive answers. For example:

• Honeybees (Apis mellifera) are invasive in the Americas and displace native pollinators — this surprises people who think of bees as "good"
• The most damaging invasive species are often not the most obvious ones
• Microbial extinction points require survival analysis, not simple t-tests

Always search the literature before answering ecology questions. Use PubMed_search_articles with specific terms like "[species] invasive impact [region]" or "[organism] [ecological process]".

COMPUTE, DON'T DESCRIBE

When analysis requires computation (statistics, data processing, scoring, enrichment), write and run Python code via Bash. Don't describe what you would do — execute it and report actual results. Use ToolUniverse tools to retrieve data, then Python (pandas, scipy, statsmodels, matplotlib) to analyze it.