Ecological findings

Is community composition
changing at La Saladita?

Shannon-Wiener diversity and Pielou evenness per taxon per year, 2000–2025. Linear trend tests. Rarefaction where data allow.

Source: GBIF & iNaturalist — 2000–2025  ·  Method: OLS linear regression  ·  June 2026

Every year that passes, citizen-science platforms like iNaturalist and Happywhale push more records into GBIF. That growing coverage is a gift — and a trap. Raw species counts look like they rise over time not because the community is richer, but because more people are photographing it. Separating genuine ecological signal from observation-effort bias is the central challenge of this analysis.

We computed Shannon-Wiener H' and Pielou's J for each taxon in each year using individual GBIF occurrence records. Where year-level species breakdowns existed (sea turtles, cetaceans, lagoon reptiles), we ran OLS linear regressions of H and J against year. For fungi, insects, and plants, the source artifacts hold only pooled totals — per-year trend analysis is not possible for those groups.

Rarefaction — subsampling each year to the smallest qualifying annual count to control for effort differences — was computed but proved uninformative here: minimum annual n across all three record-level taxa fell below 3, making the subsample essentially a coin flip. The raw trends are the working signal; they carry the effort-bias caveat prominently.

Method

Individual GBIF occurrence records were extracted from each climatology artifact's all_observations array. Year was taken directly from the year field. Species identity from species or scientific_name.

H = −Σ pᵢ · ln(pᵢ)   J = H / ln(S)

where pᵢ = count of species i ÷ total observations that year, S = number of species observed that year. Trend: OLS linear regression (scipy.stats.linregress) of H and J against year. p-values are two-tailed. Rarefaction: 200-rep subsample without replacement to the 20th-percentile annual n. Not applied where target n < 3 (all three taxa).

Pooled indices for fungi, insects, and plants use n_obs per named species from the species_calendar entries — a rough proxy for relative abundance across the full 25-year period, not a proper time series.

Observation-effort bias underlies everything here. GBIF record counts for this region roughly tripled between 2015 and 2025 — not because the animals tripled, but because smartphones and citizen-science platforms spread. A year with H = 0 could mean one species dominated, or it could mean only one species was photographed that year. Rarefaction partially corrects for this by standardizing sample size, but it cannot correct for species-specific detectability: charismatic species (humpback whales, Olive Ridley turtles) are systematically over-photographed. Treat these trends as hypotheses about what citizen-science records reveal, not as statements about ecological reality.

Trend summary

What the records show

Sea Turtles
H rising
H slope +0.033 / yr
p-value 0.0008  ·  R² 0.49
J slope +0.032 / yr  ·  p 0.001
19 years  ·  4 species max

From a Lepidochelys olivacea monoculture in 2000–2014 to multi-species years (Chelonia mydas, Dermochelys coriacea, Eretmochelys imbricata) appearing regularly after 2015. The signal is partly real — rarer species are now being recorded — and partly effort: more observers with better cameras since 2015.

Cetaceans
H declining
H slope −0.022 / yr
p-value 0.021  ·  R² 0.32
J slope −0.020 / yr  ·  p 0.032
16 years  ·  3 species present

The Happywhale platform launched in earnest around 2019, and humpback whale records exploded from a handful per year to 70–218 records annually — almost all Megaptera novaeangliae. Bottlenose and common dolphins, recorded sporadically, vanish into statistical irrelevance. This looks like a diversity collapse; it is really a data-source effect. Do not interpret as ecological decline without independent evidence.

Lagoon & Coastal Reptiles
H rising
H slope +0.036 / yr
p-value 0.0004  ·  R² 0.49
J slope +0.023 / yr  ·  p 0.028
21 years  ·  3–5 species

American crocodile, green iguana, and black spiny-tailed iguana are all increasingly co-recorded in recent years (2017–2025), pushing J toward 0.90–0.99 — near-perfect evenness. Early years had only one or two species photographed per year, giving artificially low H. The trend likely reflects both more complete coverage and possibly a real recovery of Ctenosaura pectinata after earlier declines.

Fungi & Lichens
Pooled only
S 102 taxa  ·  N 529 obs
H (pooled) 4.21  ·  J 0.911

High evenness — no single fungal species dominates the record. Per-year trends not computable: the source artifact aggregates monthly totals without species × year breakdowns.

Insects
Pooled only
S 69 named taxa  ·  N 1,445 obs
H (pooled) 3.96  ·  J 0.934

Note: only the top-30 species calendar entries are in the pooled calculation; 1,514 species have GBIF records but no per-species annual count is exposed in the artifact. True H would be higher if all species were included — this is a lower-bound estimate.

Plants
Pooled only
S 30 named taxa  ·  N 1,098 obs
H (pooled) 3.34  ·  J 0.982

Top-30 species only. 671 species recorded in full dataset. J near 1.0 among the top 30 suggests no extreme dominance — but the top algal species (Amphiroa, Hypnea) represent a June/November intertidal survey cluster, not a year-round even community.

Birds
No data yet

Bird climatology build is pending (63k+ GBIF records still being paginated). Once available, birds will likely show the highest raw species richness of any taxon at this site.

Sea Turtles — per year

From a single species to four

H = 0 for every year before 2015 — not because no other turtles were present, but because only Olive Ridley was being photographed on this coast. The three-species years of 2017–2025 represent a genuine expansion of the record, almost certainly driven by more observers rather than more turtles. Slope +0.033/yr, p = 0.0008.

Year n obs S H' (raw) J (raw) H bar
20001110.0000.000
20018410.0000.000
20034420.1090.157
20061010.0000.000
2010110.0000.000
2011210.0000.000
2013210.0000.000
2014310.0000.000
2015720.4100.592
2016210.0000.000
2017830.7360.670
2018730.7960.725
20191020.5000.722
20201030.6390.582
20211020.5000.722
2022710.0000.000
20231030.9500.865
20241030.6390.582
20251830.9340.850
Cetaceans — per year

A diversity collapse that is mostly Happywhale

Pre-2019 records are sparse and multi-species. Post-2019 records are dense but almost entirely Megaptera novaeangliae from the Happywhale photo-ID platform. H collapses to near zero not because dolphins disappeared, but because the record is now dominated by one species from one data source. Slope −0.022/yr, p = 0.021.

Year n obs S H' (raw) J (raw) H bar
20001730.9560.870
20031720.3620.523
2005510.0000.000
2006930.8490.773
2007110.0000.000
2012110.0000.000
2014110.0000.000
2017210.0000.000
2018620.4510.650
20197110.0000.000
20209010.0000.000
202114310.0000.000
20229130.1210.110
202310520.0940.136
202411220.0510.074
202521820.0290.042
Lagoon & Coastal Reptiles — per year

Three species becoming equally present

After 2013, every year with at least 5 records finds American crocodile, green iguana, and black spiny-tailed iguana all represented — J consistently 0.88–0.99. The early zeroes (2002, 2004, 2015) are single-species observation years, not absences. Slope +0.036/yr, p = 0.0004.

Year n obs S H' (raw) J (raw) H bar
2002110.0000.000
2003220.6931.000
2004310.0000.000
20061130.6000.546
2009220.6931.000
2010220.6931.000
2011520.5000.722
2012320.6370.918
2013631.0110.921
2014930.9370.853
2015110.0000.000
2016830.9740.887
20171131.0900.992
20182431.0510.956
20192230.9670.880
20202531.0880.990
20214131.0430.949
20224931.0800.983
20234630.9840.896
20246030.8960.816
20258931.0230.931
Fungi, insects & plants — pooled

High evenness, no trend possible

For these three groups, the source artifacts contain only monthly aggregates — individual occurrences are not tagged by year and species simultaneously. A pooled 2000–2025 diversity index can be computed, but temporal trend analysis requires per-year species counts that are not available without reprocessing the original GBIF/iNaturalist data.

Taxon S (named taxa) N (obs in calc) H' (pooled) J (pooled) Note
Fungi & Lichens 102 529 4.21 0.911 220 total taxa in GBIF record; 102 in pooled calc (species-level entries with n_obs > 0)
Insects 69 1,445 3.96 0.934 1,514 species in full GBIF record; 69 in pooled calc (top-30 species_calendar + notable_groups). H is a lower bound.
Plants 30 1,098 3.34 0.982 671 species in full GBIF record; 30 in pooled calc (species_calendar top-30). Heavy algal component may reflect a coastal survey event cluster rather than the full floristic picture.
What this means

Three signals, three cautions

Signal 1: Sea turtle diversity records are growing — cautiously positive

In the first decade of this dataset, almost every turtle observation was Olive Ridley — the dominant nesting species on the Guerrero coast. Since 2015, Green turtles, Hawksbills, and occasional Leatherbacks appear in multi-species years. Whether this reflects actual increase in the rarer species, improved observer coverage, or both is unknown. The statistically significant rising trend in H (p = 0.0008) is real as a property of the records. Ecologically, it is a reason for optimism but not a confirmed population recovery signal.

Signal 2: Cetacean apparent decline is almost entirely data-platform driven

The plummeting H trend among cetaceans (p = 0.021) is dramatic in the table. Pre-2019 records are patchy but include multiple species. Post-2019, Happywhale dominates the record with photo-ID sightings of individual humpbacks — a single platform optimized for a single charismatic species. Bottlenose and common dolphins were never abundantly photographed here; they do not disappear from the sea, they disappear from a record that now overwhelmingly reflects one data contributor. Interpreting this as an ecological homogenization would be wrong.

Signal 3: Lagoon reptiles show the cleanest trend

The lagoon community (American crocodile, green iguana, black spiny-tailed iguana) has only three species, so maximum H is ln(3) ≈ 1.099. After 2013, observed H consistently runs 0.90–1.09, and J consistently runs 0.88–0.99. This means all three species are being observed in roughly equal proportions — a near-perfectly even community record. The early-year spikes to J = 1.0 are artifact (two-species years are perfectly even by definition). But the consistent high-J pattern after 2017 is notable and suggests all three species are now reliably present and detectable in any active observation session at this site.

Data sources: GBIF Occurrence Search API v1 and iNaturalist. Species names as reported in source artifacts; no taxonomic harmonization applied. Observations represent GBIF occurrence records, not independent individuals — photo-ID platforms (Happywhale) may have multiple records per individual encounter.

Shannon-Wiener H and Pielou J are computed using natural logarithm throughout. Rarefaction applied at 200 replicates, subsampling without replacement. Trend significance assessed via two-tailed OLS p-values (scipy.stats.linregress); no multiple-comparison correction applied.

Analysis script: scripts/analyze_diversity_trends.py. Data endpoint: functions/api/_findings_diversity_trends.js.