Busabase
Busabase is an open-source (MIT) database and workspace for AI agents — data, docs, skills, and apps in one place. Run it locally, self-host it inside your network, or use Busabase Cloud, and turn age
翻訳待ち
No updates detected yet. GitHub releases will appear here.
Shiny app that that uses NEON data to compare small mammal capture data across NEON sites.
翻訳待ち
Loading the full README.
Read the full README ↗A Shiny web app for exploring the National Ecological Observatory Network's small-mammal box-trapping data — reconstructing each captured animal's history from its ear-tag and turning 46 field sites of capture records into maps, charts, and individual profiles.
Live: the Posit Connect Cloud app
and the Pages cover. After each merge the
post-deploy smoke waits until Connect serves the merged revision (the ddl-build meta) and checks the
page title and language; docs/project-status.md records the last verified state.
Connect's source points at main, but automation never writes there directly. Validated monthly
refreshes are published to automation/small-mammal-data-refresh as a review PR; an intentional
merge plus a verified Connect republish is the production decision.
The app opens to a national map of every NEON site — tap a dot to dive in (dot size = animals caught there, color = the dominant mammal family), or flip to "by species" to map where a single animal turns up across the country. Each site loads instantly from a per-site data bundle that ships with the app (no network round-trip). From there it reconstructs each animal's capture history from its ear-tag ID, ranks the regulars, profiles individuals, estimates abundance, and maps where they were caught. The map is the site picker; once a site is open, a "change site" link in the top bar takes you back to it.
It is built for two audiences: anyone curious about NEON small-mammal sampling, and new field technicians getting to know the species at their site.
Select-your-site map. A national Leaflet map of all 46 bundled sites — sized by total captures, colored by the most-caught mammal family — with a "by site / by species" toggle, an accessible list fallback, and a one-tap load. A 30-second guided tour points out the rest.
Explore by species. Pick any species caught at two or more sites and the map redraws to just the sites where it's caught, sized by local abundance: a national range map.
Detection-corrected abundance. Closed-capture estimates per trapping bout (Schnabel for ≥3 nights, Chapman for 2) with a per-night detection probability, shown alongside MNKA, gated when recaptures are too few, and clamped to the minimum known alive — defensible, with caveats shown.
Diversity profile. Hill numbers (q0 richness, q1 effective-common, q2 effective-dominant) plus an evenness read, computed over distinct individuals.
Shareable trading cards, compare, and report cards. Export any individual's dossier as a holographic PNG trading card, put two sites head-to-head, or print a one-page site report card to PDF.
Overview — the story of a site. Species ranked by abundance, an automatically written plain-English summary, and quick-jump navigation to every view.
Hall of Fame — rank every individual. A top-3 podium over a leaderboard of every animal, re-sortable by captures, weight, career length, roaming, or weight-for-its-species, with rarity tiers and a Legendary shimmer. Each dossier opens with a computed one-line "story" — the animal's standout stat ranked against its peers ("the most-caught individual at this site") — and count-up stats.
Site map. Species diversity by plot on a satellite basemap; the selected individual's plots are highlighted. An optional recapture-movement layer draws curved arcs between grids where the same tagged animals were recaptured (thicker = more individuals made the move) — the between-grid connectivity the dots can't show, framed honestly as mark-recapture, not telemetry.
Measurements over time. An individual's weight and hind-foot length tracked across captures, against the species' typical range.
Body-size map. Where an animal sits in its species' weight-by-length cloud, with a fitted size–mass line drawn only where the relationship is statistically real.
Size Lab — an interactive QC workbench. Every individual at the site on one body-size map (hind-foot length × weight, coloured by species). Pick a species (and plot) to add its median crosshairs and an adult size–mass fit line — drawn only where length really predicts mass. Tap any dot to pin its card, drag it around, and open the QC history card for that animal: every capture's measurements plus automatic, ranked data-quality flags (phrased "verify, not wrong" — a same-tag-at- two-plots impossibility, a backward life-stage, an implausible weight jump, a sex flip…). Then download the works — the whole map with the pins baked in, the QC card as a PNG, or the animal's full capture history as CSV metadata (an analysis-ready field record). Framed throughout as a QC / morphometric map, not a body-condition index — a dot far from its species' cloud flags an unusual or mistyped record, not "fitness."
Community body-size profile. The weight distribution of every species at a site, lightest to heaviest, with the selected animal marked.
This is the beginning of the README. Read the rest at the original. Read the full README ↗