Fig. 3: CNN-derived bird community composition as predictors for birds and insect communities. | Nature Communications

Fig. 3: CNN-derived bird community composition as predictors for birds and insect communities.

From: Soundscapes and deep learning enable tracking biodiversity recovery in tropical forests

Fig. 3: CNN-derived bird community composition as predictors for birds and insect communities.The alternative text for this image may have been generated using AI.

CNN-bird community versus (a) first community axis of total vocalizing vertebrates by experts (the scores of the first NMDs axis in Fig. 2a and all other ordinations were mirrored to create an ascending recovery gradient from pastures to old-growth forests), versus (b) first nocturnal insect community axis by metabarcoding of light trap samples along the forest recovery gradient and versus (c) Expert-derived bird community composition based on 49 species shared in expert and CNN predictions. R² values represent adjusted-R² from a linear regression; polygons represent 95% confidence intervals of the means of the linear regression.

Back to article page