SEARCH

SEARCH BY CITATION

Keywords:

  • Co-kriging interpolation;
  • hawkmoths;
  • Lepidoptera;
  • sampling effort;
  • spatial pattern;
  • Sphingidae;
  • Sub-Saharan Africa

Abstract

Aim

Many taxa, especially invertebrates, remain biogeographically highly understudied and even baseline assessments are missing, with too limited and heterogeneous sampling being key reasons. Here we set out to assess the human geographic and associated environmental factors behind inventory completeness for the hawkmoths of Africa. We aim to separate the causes of differential sampling from those affecting gradients of species richness to illustrate a potential general avenue for advancing knowledge about diversity in understudied groups.

Location

Sub-Saharan Africa.

Methods

Using a database of distributional records of hawkmoths, we computed rarefaction curves and estimated total species richness across 200 km × 200 km grid cells. We fitted multivariate models to identify environmental predictors of species richness and used environmental co-kriging to map region-wide diversity patterns. We estimated cell-wide inventory completeness from observed and estimated data, and related these to human geographic factors.

Results

Observed patterns of hawkmoths species richness are strongly determined by the number of available records in grid cells. Both show spatially structured distributions. Variables describing vegetation type, emerge as important predictors of estimated total richness, and variables capturing heat, energy availability and topographic heterogeneity all show a strong positive relationship. Patterns of interpolated richness identify three centres of diversity: Cameroon coastal mountains, and the northern and southern East African montane areas. Inventory completeness is positively influenced by population density, accessibility, protected areas and colonial history. Species richness is still under-recorded in the western Congo Basin and southern Tanzania/Mozambique.

Main conclusions

Sampling effort is highly biased and controlling for it in large-scale compilations of presence-only data is critical for drawing inferences from our still limited knowledge of invertebrate distributions. Our study shows that a baseline of estimate of broad-scale diversity patterns in understudied taxa can be derived from combining numerical estimators of richness, models of main environmental effects and spatial interpolation. Inventory completeness can be partly predicted from human geographic features and such models may offer fruitful guidance for prioritization of future sampling to further refine and validate estimated patterns of species richness.