Which Species at La Saladita Track the Weather?

GBIF/iNaturalist 2000–2025, 1,651 records, cross-referenced against ERA5. 9–17 usable years per group for interannual tests — small by climate-science standards. Results are explicit about what that size can and cannot support.

Summary

Climate sensitivity ranking

Mushrooms (all fungi)

Driver: Prior month's rainfall (roughly 30-day lag)

Rainfall explains 72% of monthly fruiting variation (seasonal pattern, not year-to-year shifts).

0.72r² — explains 72% of variation

Humpback whale (Megaptera novaeangliae)

Driver: Year (as a stand-in for long-term ocean warming)

Arrival centered on mid-January; no measurable shift. n=9 years — preliminary.

0.11r² — 11% of variation; not significant

Olive Ridley sea turtle (Lepidochelys olivacea)

Driver: Year (as warming proxy)

Peak month swings widely; no trend. Full-moon clustering real (p≈0.0001) but likely a visibility artifact, not nesting preference.

0.04r² — 4% of variation; essentially null

Analysis 1

Mushroom fruiting
vs cumulative rainfall

Significant — seasonal pattern confirmed

Mushroom fruiting tracks prior-month rainfall with unusual strength

667 records, 2000–2025. Prior-month ERA5 rainfall explains 72% of monthly mushroom activity variation (r²=0.72, p=0.0005). One-month lag gives the best fit.

          0.72
          r² at 30-day lag (explains 72% of variation)
        

p = 0.0005 almost certainly real

0.298 additional observations per mm of rain

n = 12 calendar months tested

Signal weakens at longer lags (2-month: r²=0.585; 3-month: r²=0.509). September–October peak tracks the rainy-season tail; September averages 344 mm prior-month rain. January secondary bump = bracket fungi and lichens, not rainfall-triggered.

Lag window	Lag (approx.)	r²	p-value	Slope
Prior 1 month (~30d)	30 days	0.720	0.0005	0.298
Prior 2 months (~60d)	60 days	0.585	0.0037	0.144
Prior 3 months (~90d)	90 days	0.509	0.0092	0.098

What this finding does and doesn't show: This is a seasonal pattern, not a year-by-year climate signal. It tells us that September has more mushrooms than January because September follows a wet August — which would be true even if the climate never changed. What it cannot tell us is whether hotter or wetter years are shifting the fruiting peak earlier or later over time. Answering that requires year-by-year counts against year-by-year rainfall. With 667 total records spread across 25 years and 220 species, most individual months in individual years have too few observations to support that test. The aggregate seasonal shape is reliable; interannual tracking is not.

Monthly fruiting vs mean prior-month rainfall

Month	Mushroom obs (total)	Mean prior-month rainfall (mm)
January	41	8.1 (Dec)
February	12	11.6 (Jan)
March	11	4.4 (Feb)
April	20	4.7 (Mar)
May	11	0.2 (Apr)
June	32	21.4 (May)
July	107	215.3 (Jun)
August	49	218.9 (Jul)
September	151	265.5 (Aug)
October	91	344.0 (Sep)
November	86	216.2 (Oct)
December	35	36.6 (Nov)

Analysis 2

Humpback whale arrival timing
vs ocean temperature

No detectable shift — with only 9 years of data, take this as preliminary

Humpback arrival timing is stable — we can't yet see any trend

Center-of-mass arrival month computed per year (≥3 records). Regressed against calendar year as a proxy for the ERA5 warming trend (~+0.015°C/yr). n=9 usable years — preliminary.

mid-January typical arrival center

0.11 r² (explains only 11% of variation)

p = 0.37 not significant

n = 9 years with ≥3 sightings — too few to be sure yet

The (weak, insignificant) slope suggests slightly later arrivals — consistent with warmer-ocean delays — but with n=9 the confidence interval is too wide to distinguish trend from noise.

Year	n obs	Centroid month
2005	5	1.20
2018	5	1.00
2019	71	1.62
2020	90	1.54
2021	143	1.48
2022	89	0.94
2023	103	1.46
2024	111	1.42
2025	217	1.65

Key limitation: We don't have actual ocean temperature readings for each individual year stored in the data artifact — only the long-term average warming trend. This means we're missing the year-to-year swings driven by El Niño and La Niña, which substantially affect humpback feeding-ground productivity. Using actual per-year sea temperatures from the Open-Meteo ERA5 archive would improve this test considerably. Also: iNaturalist sightings have grown rapidly since around 2019, meaning recent years have more records simply because more people are reporting — not because more whales are here. The hypothesis — warmer ocean = later arrivals — is biologically plausible but cannot be confirmed or ruled out from this data.

Analysis 3

Olive Ridley turtle nesting peak
vs ocean temperature and moon phase

Ocean temperature: no signal | Full-moon clustering: real, but likely a visibility artifact

Peak nesting month is unpredictable year to year — a lunar signal exists but may not mean what it seems

Peak month per year (17 years with ≥2 records). Tests: (1) trend over time as ocean temperature proxy; (2) lunar-phase clustering via Rayleigh test.

0.04 r² — 4% of variation explained (essentially null)

p = 0.46 not significant — no ocean-temperature trend detected

Rayleigh R = 0.25 lunar clustering strength

p ≈ 0.0001 full-moon clustering is almost certainly non-random

Peak month swings Feb–Dec with no trend. Half the years show Sep–Nov peak, consistent with the Jul–Dec nesting window for the Guerrero coast.

Moon-phase distribution

Rayleigh test on 146 dated observations: R=0.25, p≈0.0001. Records concentrate around first-quarter to full moon:

New (0–0.25) · 16 obs

Waxing (0.125–0.375) · 35

Full (0.375–0.625) · 62 obs

Waning (0.625–0.875) · 33

62 of 146 records (42%) fall in the full-moon window vs an expected 25%. The clustering is real. The cause is unresolved: GBIF records don't separate nesting females from at-sea sightings, and turtles are simply easier to see on moonlit nights. Observer visibility alone could produce this pattern.

Year	n obs	Peak month
2000	11	September
2003	43	September
2006	10	November
2011	2	October
2013	2	July
2014	3	February
2015	6	November
2016	2	March
2017	6	December
2018	5	April
2019	8	September
2020	8	November
2021	8	April
2022	7	October
2023	6	July
2024	8	March
2025	11	November

The majority of Olive Ridley records come from just two survey years: 43 in 2003 and 11 in 2000. These two years dominate the aggregate statistics. Annual peak-month estimates based on 2–3 records should be read as "this species was here in this month" — not as reliable phenology data. Resolving both the ocean-temperature and moon-phase questions properly would require data from monitored nesting beaches along the broader Guerrero coast — systematic beach-patrol records, not opportunistic wildlife sightings submitted to GBIF.

Analysis 4 (deferred)

Birds vs rainfall & season

Deferred — data not yet available

A bird climatology artifact (_bird_climatology.js) was not present at time of analysis. When available, the planned test is: monthly observation abundance vs ERA5 monthly precipitation and temperature, to identify whether resident vs migratory species show differential rainfall-tracking. This section will be updated automatically when the artifact lands.

Synthesis

Mushrooms: rainfall explains 72% of monthly fruiting variation (seasonal pattern, not year-to-year climate signal). Humpbacks: arrival stable at mid-January; no significant shift; n=9 is too small to resolve. Turtles: peak month varies widely, no trend; full-moon clustering real but most likely a visibility artifact. Shore-based nesting data from the Guerrero coast would resolve the lunar question.

Sources

GBIF/iNaturalist, ~150 km bounding box, 2000–2025, georeferenced only. ERA5 monthly rainfall and temperature 1979–2025. Mushrooms: 667 records, OLS monthly totals vs ERA5 prior-month rainfall, n=12. Humpbacks: 838 records, 9 years ≥3 sightings, center-of-mass arrival vs calendar year as warming proxy. Turtles: 146 records, 17 years ≥2 observations, modal peak month; Rayleigh test on lunar phase. Per-year SST not stored; year used as proxy — misses ENSO variance. p-values 0.37–0.46 are genuine nulls. Source artifact: _findings_phenology_climate.js generated by scripts/analyze_phenology_climate.py.