Biodiversity Data Journal :
Research Article
|
Corresponding author: Ana G. Del Hierro (anagdelhierro@gmail.com)
Academic editor: Diego Aguilar Fachin
Received: 09 Jul 2022 | Accepted: 10 Oct 2022 | Published: 17 Oct 2022
© 2022 Alex Pazmiño-Palomino, Carolina Reyes-Puig, Ana Del Hierro
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Pazmiño-Palomino A, Reyes-Puig C, Del Hierro AG (2022) How could climate change influence the distribution of the black soldier fly, Hermetia illucens (Linnaeus) (Diptera, Stratiomyidae)? Biodiversity Data Journal 10: e90146. https://doi.org/10.3897/BDJ.10.e90146
|
|
The black soldier fly, Hermetia illucens (Linnaeus, 1758), is a saprophagous species used to decompose organic matter. This study proposes a distribution model of H. illucens to illustrate its current and future distribution. The methodology includes data collection from the Global Biodiversity Information Facility (GBIF), complemented with iNaturalist, manual expert curation of occurrence records, six species distribution models algorithms and one ensemble model. The average temperature of the driest annual quarter and the precipitation of the coldest annual quarter were the key variables influencing the potential distribution of H. illucens. The distribution range is estimated to decrease progressively and their suitable habitats could change dramatically in the future due to global warming. On the other hand, current optimal habitats would become uninhabitable for the species, mainly at low latitudes. Under this scenario, the species is projected to move to higher latitudes and elevations in the future. The results of this study provide data on the distribution of H. illucens, facilitating its location, management and sustainable use in current and future scenarios.
BSF, citizen science, Hermetia illucens, iNaturalist, species distribution model
Insects (Insecta) are amongst the most diverse and efficient groups on the planet, capable of performing ecological functions at different levels of food chains (
Hermetia illucens, very likely native to the tropics and subtropics of the Neotropical Region (
Thus, considering the lack of information regarding BSF distribution, our study aims to map BSF's current worldwide potential distribution, identify key climatic factors associated with its distribution and predict future potential distribution under a global climate change scenario.
We collected H. illucens occurrence data (Suppl. material
Only the records with uploaded images were considered in the current analysis because this is a criterion for verifying the insect's identity through visual identification. We found 6,171 occurrence data for the species in GBIF, gathered from 1897 to 2020 (
Distribution models of Hermetia illucens. A Occurrence records (2,988) from iNaturalist and GBIF; B Current global potential distribution C Global potential distribution under RCP 4.5 to 2050; D Global potential distribution under RCP 8.5 to 2050; E Global potential distribution under RCP 4.5 to 2070; F Global potential distribution under RCP 8.5 to 2070. The colour gradient scale represents the probabilities of habitat suitability for the distribution of the species, being blue = 0 and red = 1.
WorldClim version 2.1 (http://www.worldclim.org) provided 19 environmental variables to construct an exploratory model of the potential distribution of the species. The variables have an approximate resolution of 1 кm2 (
Within BAM (i.e. A. environmental conditions; B. biotic interactions; and M. accessible area scheme to generate SDMs) (
With the chosen environmental variables, the SDM package (
The algorithms included 30 replicates per method and bootstrap for data partitioning (Suppl. material
Model performance and accuracy evaluation included the following statistics indicators: Receiver Operating Characteristic (ROC) curve with the Area Under the Curve (AUC), partial-area ROC (as the ratio of AUC to the null expectation), Correlation Coefficient (COR) and the True Skill Statistics (TSS). The AUC-ROC and partial-area ROC are suitable to evaluate the model's ability to predict better than chance, considering the genuinely positive and false-positive rates (
Finally, the best candidate models were modelled as an ensemble model. The cut-off threshold for the potential distribution maps of absence and presence was based on the average of all the models of the AUC and TSS indicators; in general, these indicators tend to favour models with a more significant number of occurrences (
Thirteen out of the nineteen input variables had collinearity problems. The linear correlation coefficients ranged between minimum correlation (bio19 ~ bio8) = 0.1154862 and maximum correlation (bio9 ~ bio8) = 0.6247589. We identified six non-correlated variables. A total of 150 models were built (Suppl. material
All the methods used demonstrated exemplary performance and predictive power (Table
Performance evaluation of the six algorithms used to develop the model through the different statistical indicators.
Methods | AUC | AUC ratio | COR | TSS |
GLM | 0.78 | 1.52 | 0.58 | 0.48 |
SVM | 0.92 | 1.66 | 0.77 | 0.74 |
RF | 0.99 | 1.78 | 0.89 | 0.88 |
BRT | 0.9 | 1.62 | 0.72 | 0.66 |
MARS | 0.9 | 1.65 | 0.73 | 0.67 |
Maxent | 0.9 | 1.71 | 0.72 | 0.66 |
According to the predictions, under current climatic conditions, H. illucens is potentially established in much of the sub-humid to humid tropics and subtropics (Fig.
The model also predicts that the United States (Texas to Virginia), Mexico, Central America and South America (excluding the dry-lands of south-western Ecuador, Peru, Chile and southern Argentina) are optimal habitats for BSF. Our results suggest that suitability decreases towards the Poles, generating marginal zones in other dry-lands of the world: the Middle East, South Africa and central Australia (Fig.
Using a cut-off threshold with habitat suitability, 40,139,897 km2 of the world's land surface is climatically suitable (Fig.
Potential distribution of Hermetia illucens considering a cut-off threshold. Orange represents presence and brown absence. A Current potential distribution; B RCP 4.5 2050 potential distribution; C RCP 8.5 2050 potential distribution; D RCP 4.5 2070 potential distribution; E RCP 8.5 2070 potential distribution.
Model performance under different RCPs indicated significant habitat suitability reductions across the species' presence range. A suitable habitat loss gradient is seen as the years and RCP increase. Compared to the current ideal habitat extension, the 4.5 model in 2050 (Fig.
Change between the current and future potential distribution of Hermetia illucens. The red areas represent considerable future changes. A changes between current climate and RCP 4.5 2050; B changes between current climate and RCP 8.5 2050; C changes between current climate and RCP 4.5 2070; D changes between current climate and RCP 8.5 2070.
The areas most likely to lose BSF suitability in the future are the interior of South America, especially the Amazon Basin, southern USA, central Africa, Mediterranean coasts, Southeast Asia, insular Asia and coasts of Australia. At the same time, the suitability of the habitat would progressively increase in continents with high latitudes, such as North America, Europe and Asia, especially around the 40° N parallel. The same effect seems to manifest below the 40° S parallel in America, Africa and Australia (Figs
Our models also suggest that BSF will gradually climb in elevation. It is currently known that the species is concentrated from sea level to 2800 m a.s.l. (pers. obs.). According to our projections under the proposed climate scenarios, it could reach up to 4000 m a.s.l. (Fig.
The suitability of the habitats responds to the climatic variables described in our model. We found that the global distribution of Hermetia illucens is mainly influenced by temperature and precipitation. The Mean temperature of the driest annual quarter (Bio9) plays a fundamental role in its dispersion. According to physiological laboratory experiments performed by
The second most influential variable was Precipitation of the coldest annual quarter (Bio19) and Precipitation of the wettest month (bio13), both directly related to humidity.
The current model predicted by this study demonstrates habitat suitability for the BSF in tropical and subtropical areas globally. In this model, BSF presents a marked distribution in the tropical belt of America, Africa, Asia and the Mediterranean coasts. On the other hand, desert areas or low temperatures habitats showed poor suitability for the species. Our model indicates that H. ilucens would not be distributed in dry-lands from Peru to Chile, the Andes Mountains in South America, dry-lands of North America, North Africa, the Middle East, Russia and central Australia. Our model matches the collected historical occurrence records (
Numerous historical records from the late 19th and early 20th centuries document the presence of the species in the New World (
Consistent records of this species in new sites, mainly in the Palaearctic, are found along coasts and islands, suggesting that maritime transport may play a role in possible repeated accidental introductions (
The BSF has been classified as hemi-synanthropic; it lives around human populations in semi-rural and rural landscapes (
Based on a future scenario in which climate change would modify the planet's climatic conditions, the RCP 4.5 and 8.5 scenarios have been taken as reference in 2050 and 2070 (
According to similar models, habitat suitability in South America, Africa, Southeast Asia and Australia would decrease due to increased temperature stress. Similar results were reported for other dipteran crop pests, such as Bactrocera dorsalis (Hendel) (
Our model predicts that the current altitude range of H. illucens would change under climate change scenarios. It is projected that this species could colonise elevations 2800 m, higher than what is currently recorded. This coincides with the projections of coleopterans in the mountains of south-eastern Brazil (
Data on the occurrence of H. illucens were collected during the confinement of COVID-19, a condition that prevented access to biological collections. The restrictions motivated the use of the world's biodiversity registry repositories, such as the GBIF and iNaturalist platforms. The data provided by the citizen-science observations of the iNaturalist platform served as the primary tool to create the presented models. The morphological characters of the species acted as a delimiting parameter for the selection of data occurrence, mainly due to characteristics visually recognisable by photographs, which showed a pattern in the records throughout its distribution (
The present study modelled the potential distribution of H. illucens on a global scale in current and future scenarios. The methodology applied six species distribution modelling algorithms, based on bioclimatic variables. The model exhibited a high prediction performance where the mean temperature of the driest annual quarter and precipitation were the key factors influencing the potential distribution of BSF. According to our model, the possible ranges of H. illucens would decrease by up to 10.5% in the future due to global warming. The distribution of favourable habitats would change to high latitudes and high elevations.
On the other hand, much of the current optimal habitats would become uninhabitable for the species, mainly at low latitudes. This study provides the distribution of BSF to facilitate its location, management and sustainable use in current and future scenarios. Nevertheless, the model presents a strong warning about global warming consequences, which, in addition to the ecological distribution, threatens the growing production of H. illucens as an alternative for waste management. However, the model did not include drought and precipitation indices resulting from climate change, which could reduce H. illucens distribution (
Finally, we should mention that the citizen-science data served to create our models under strict verification and curation filters (
The authors thank to Xaali O'Reilly-Berkeley for the English revision of the text. This research was supported by Universidad San Francisco de Quito USFQ, Museo de Zoología & Laboratorio de Zoología Terrestre and Instituto iBIOTROP granted to CRP (project HUBI: 17475).
Conceptualisation of the project, Alex Pazmiño-Palomino (APP) and Ana G. Del Hierro (ADH); data curation and validation, APP; modelling and statistical analysis, Carolina Reyes-Puig (CRP); writing—original draft preparation, APP, ADH and CRP; writing—review and editing, APP, CRP and ADH; project administration, ADH. All authors have read and agreed to the published version of the manuscript.
The supplementary material includes Hermetia illucens occurrences collected from the Global Biodiversity Information Facility (GBIF) complemented with iNaturalist.
Supplementary material includes manually curated records from Supplementary material 1.1.
Thirty predictions of current potential distribution for each algorithm.
R script used to build the models.