How accurate is the forage map?

We check every species' map against real, dated finds from GBIF (iNaturalist, herbaria and surveys), the same way, and publish the scores.

The scores

Species	Finds checked	Habitat vs all land	Habitat vs forest	Timing
Matsutake	990	0.83	0.69	0.64
Fall king bolete	1,516	0.81	0.65	0.70
Chanterelle	1,320	0.93	0.85	0.68
Hedgehog	267	0.92	0.83	0.70
Black trumpet	446	0.97	0.94	0.78
Spring king	293	0.94	0.87	-
Burn morels	-	-	-	0.72
Natural morels	1,336	0.66	0.54	0.72
Lobster	1,950	0.87	0.74	0.72
Cauliflower	524	0.85	0.76	0.77
Oyster	1,803	0.82	0.71	0.66
Candy cap	1,129	0.8	0.74	0.81
Chicken of the woods (conifer)	4,458	0.63	0.39	0.83
Bear's head	1,702	0.81	0.68	0.78
Shrimp russula	371	0.85	0.69	0.70
Yellowfoot	942	0.93	0.86	0.74
Chicken of the woods (oak)	-	-	-	0.68

All scores are AUC: 0.5 is a coin toss, 1.0 is perfect. Habitat: how much higher the map rates the places where the species was found than random places (all land, or only forest, the harder test). Timing: how much higher the daily outlook rated the weeks with finds than other weeks of the season, from past weather (2018-2025). Matsutake locations are blurred by about 20 km in public records, so its habitat score is approximate.

What this means

A good score says the map ranks the right places and weeks higher, not that a zone will hold mushrooms. Weather, picking pressure and luck still decide a day out. Every check-in from our beta testers is used to keep improving the model.

The full forage map

250 m habitat zones for 17 species, a 15-day flush outlook that follows the weather and the terrain, public-land picking rules, field mode with offline maps, picking alerts, and photo check-ins. It is in private beta: ask for an invite.

Ask for an invite