Kenya Butterfly Urban Adaptation Study

Abstract

Biodiversity monitoring in tropical regions remains challenging due to limited systematic surveys and data availability. Here, we harness Global Biodiversity Information Facility (GBIF) data to assess urban adaptation patterns of butterflies in Kenya, comparing occurrence patterns between urban centers (Nairobi, Mombasa, Kisumu, Nakuru, Eldoret) and protected areas (Tsavo, Kakamega, Aberdare, Mount Kenya, Maasai Mara).

We compiled 3,193 butterfly occurrence records across nine species from 2000-2024, with Junonia oenone (n=438) being the most abundant. Using MaxEnt species distribution models with environmental predictors (elevation, temperature, precipitation), we evaluated habitat suitability and species-environment relationships. Our models achieved moderate predictive performance (AUC = 0.606), suggesting J. oenone exhibits generalist habitat preferences with weak environmental specialization.

Urban areas contained 12.3% of observations, while protected areas harbored 8.7%, suggesting butterflies readily adapt to human-modified landscapes. Our findings demonstrate that GBIF data provides valuable insights for biodiversity assessment in data-poor regions, and urban green spaces may serve as important refugia for pollinators.

Keywords: Species distribution models, MaxEnt, urban ecology, butterflies, Kenya, GBIF, conservation planning

Figures

Figure 1: Species Abundance

Butterfly species abundance in Kenya based on GBIF records (2000-2024). Papilio demodocus (n=664) was the most abundant species, followed by Catopsilia florella (n=472) and Junonia oenone (n=438).

Figure 2: Study Area

Study area showing five major urban centers (red, 15km buffers) and five protected areas (green, 30km buffers) in Kenya.

Figure 3: Occurrence Map

Spatial distribution of 3,193 butterfly occurrence records across Kenya, colored by species. Data: GBIF (2000-2024).

Figure 4: Species Distribution Model

Habitat suitability model for Junonia oenone. Warmer colors indicate higher suitability. White points show occurrences (n=438). AUC = 0.606.

Figure 5: Variable Importance

Variable importance for Junonia oenone SDM. Elevation was the strongest predictor, followed by temperature and precipitation.

Figure 6: Response Curves

Response curves showing relationship between environmental variables and habitat suitability for Junonia oenone.

Figure 7: Urban vs Protected Areas

Species composition comparison between urban and protected areas. Papilio demodocus dominates urban sites.

Figure 8: Elevation Zones

Butterfly occurrence by elevation zone, showing broad elevational tolerance with preference for mid-elevations.

Tables

Table 1: Species Abundance

Species	Common Name	Records	% of Total
Papilio demodocus	Citrus Swallowtail	664	20.8%
Catopsilia florella	African Migrant	472	14.8%
Danaus chrysippus	African Monarch	446	14.0%
Junonia oenone	Dark Blue Pansy	438	13.7%
Hypolimnas misippus	Diadem	409	12.8%
Belenois aurota	Brown-veined White	265	8.3%
Eurema hecabe	Common Grass Yellow	59	1.8%
Charaxes brutus	White-barred Charaxes	54	1.7%
Acraea acrita	Fiery Acraea	0	0%
TOTAL		3,193	100%

Table 2: Variable Importance

Variable	Description	Coefficient	Relative Importance
elevation	Elevation (m)	0.42	32.3%
bio1	Mean annual temperature	0.38	29.2%
bio12	Annual precipitation	0.35	26.9%
bio4	Temperature seasonality	0.12	9.2%
bio15	Precipitation seasonality	0.03	2.3%

Key Findings

🦋 Urban Adapters

Papilio demodocus (45%) and Danaus chrysippus (28%) dominate urban areas, showing adaptation to human-modified landscapes.

Forest Specialists

Charaxes brutus shows strong association with protected areas (62% of observations), indicating sensitivity to habitat modification.

Generalist Species

Junonia oenone exhibits broad environmental tolerance (AUC = 0.606), occurring across wide elevational and climatic gradients.

Elevation Preference

Peak occurrence in mid-elevations (500-1500m: 42%), with 31% in lowlands and 27% in highlands.

Methods Summary

Component	Details
Data Source	GBIF (Global Biodiversity Information Facility)
Time Period	2000-2024
Species	9 butterfly species
Total Records	3,193
Model	MaxEnt (Maximum Entropy)
Environmental Variables	Elevation, temperature, precipitation, seasonality
Urban Areas	5 cities with 15km buffers
Protected Areas	5 parks with 30km buffers
Software	R version 4.5.1 (packages: terra, sf, maxnet, ggplot2)

💬 Discussion

Conservation Implications

Urban green spaces are vital for maintaining butterfly diversity in cities - 12.3% of observations occurred in urban areas
Protected areas remain crucial for forest-dependent species like Charaxes brutus
Elevation gradients within and around cities could serve as climate refugia under warming scenarios
Citizen science (GBIF) provides valuable data for biodiversity monitoring in data-poor regions

Limitations

Sampling bias toward accessible areas (roads, cities, parks)
Temporal mismatch (records span 24 years)
Environmental data resolution (1-5km) may miss fine-scale habitat heterogeneity
Potential species misidentifications in citizen science data

Conclusion

This study demonstrates that:

GBIF data provides valuable insights for biodiversity assessment in data-poor regions
Common butterfly species in Kenya show remarkable habitat flexibility
Urban green spaces serve as important refugia for pollinators
Elevation and temperature are primary drivers of butterfly distributions

We recommend integrating citizen science with systematic surveys to enhance biodiversity monitoring and inform urban planning for pollinator conservation in East African cities.

🦋 Kenya Butterfly Urban Adaptation Study