Research Article
Received: 11 February 2021 Revised: 16 April 2021 Accepted article published: 13 May 2021 Published online in Wiley Online Library:
(wileyonlinelibrary.com) DOI 10.1002/ps.6478
Horizon scanning to assess the bioclimatic
potential for the alien species Spodoptera
eridania and its parasitoids after pest
detection in West and Central Africa
Ghislain T Tepa-Yotto,a,b* Gérard N Gouwakinnou,c
Johannes R Fagbohoun,a,d,e Manuele Tamòa and May-Guri Sæthref
Abstract
BACKGROUND: The southern armyworm (SAW) Spodoptera eridania (Stoll) (Lepidoptera: Noctuidae) is native to the tropical
Americas where the pest can feed on more than 100 plant species. SAW was recently detected in West and Central Africa, feed-
ing on various crops including cassava, cotton, amaranth and tomato. The current work was carried out to predict the potential
spatial distribution of SAW and four of its co-evolved parasitoids at a global scale using the maximum entropy (Maxent)
algorithm.
RESULTS: SAWmay not be a huge problem outside its native range (the Americas) for the time being, but may compromise crop
yields in specific hotspots in coming years. The analysis of its potential distribution anticipates that the pest might easily
migrate east and south from Cameroon and Gabon.
CONCLUSION: The models used generally demonstrate that all the parasitoids considered are good candidates for the biolog-
ical control of SAWglobally, except they will not be able to establish in specific climates. The current paper discusses the poten-
tial role of biological control using parasitoids as a crucial component of a durable climate-smart integrated management of
SAW to support decision making in Africa and in other regions of bioclimatic suitability.
© 2021 The Authors. Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
Supporting information may be found in the online version of this article.
Keywords: southern armyworm; climate change; biological control; foresight analysis; decision support
1 INTRODUCTION
Invasive alien species (IAS) have negative ecological and eco-
nomic consequences worldwide and the severity of these impacts
is growing.1,2 IAS have becomemajor threats to global agriculture * Correspondence to: GT Tepa-Yotto, Biorisk Management Facility, International
because of their rapid spread across the globe facilitated by Institute of Tropical Agriculture, 08-01000 Cotonou, Benin. E-mail:
g.tepa-yotto@cgiar.org
increased trade and transport.3,4 As an example, the increasing
spread of pests into Africa has caused critical crop losses esti- a Biorisk Management Facility, International Institute of Tropical Agriculture,
mated to be several billions US dollars per annum.5,6 The southern Cotonou, Benin
armyworm (SAW) Spodoptera eridania (Stoll) (Lepidoptera: Noctui-
b Ecole de Gestion et de Production Végétale et Semencière, Université Nationale
dae) is one of these invaders. Shortly after the fall armyworm Spo- d'Agriculture, Kétou, Benin
doptera frugiperda (JE Smith) (Lepidoptera: Noctuidae) outbreaks
were reported in West Africa7 the southern armyworm c Laboratoire d'Ecologie, de Botanique et de Biologie végétale, Faculté d'Agrono-
S. eridania was also detected in December 2016 and in 2017 in mie, Université de Parakou, Parakou, Benin
West (Benin and Nigeria) and Central (Cameroon and Gabon) d Faculty of Biosciences, Norwegian University of Life Sciences, Ås, Norway
Africa.8 SAW was observed for the first time in these countries
feeding on various crops, including cassava, cotton, amaranth e Department for Invertebrate Pests and Weeds in Forestry, Horticulture and
and tomato. Native to the Americas,9 S. eridania is a voracious Agriculture, Norwegian Institute of Bioeconomy Research, Ås, Norway
polyphagous defoliator known to damage major agricultural f Department for Climate, Energy and Environment, Section for Environment
crops.10,11 The most recent report on its host plants range com- and Food Security, Norwegian Agency for Development and Cooperation, Oslo,
prises 106 plants species belonging to 33 plant families.12,13 Norway
© 2021 The Authors. Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and
distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
1
www.soci.org GT Tepa-Yotto et al.
Neonate caterpillars are usually found on the lower surface of the 2 MATERIALS AND METHODS
leaves and only feed on the cuticle, while later larval stages (sec- 2.1 Climate suitability modelling
ond to fifth instars) consume the entire leaves, leading to skele- The horizon scanning and bioclimatic potential assessment of the
tonized plants. Control of the pest is possible through pest and parasitoids were performed using a combination of cli-
application of foliar pesticides on immature stages. However, mate suitability modelling (HSM) and spatial analysis. HSM com-
the growing concern about insecticide resistance, and human bines the observed presence records for each species (pest or
and environmental health concerns suggests the need for sus- parasitoid) with environmental data at the observed locations to
tainable approaches like biological control. generate (i) a prediction map for the suitable current climate for
Insects pests and their associated parasitoids are poikilothermic the target species and (ii) potential future distributionmaps based
organisms, their development being affected by temperature varia- on projections of selected global climate models (GCMs).
tions.14,15 Outbreaks of major pests are frequently related to natural
events such as drought, temperature increase, hurricane and
flood.14,16,17 These climate changes will affect, positively or nega- 2.1.1 Species records and environmental data
tively, the suitability of certain regions for insect pests and their nat- Geographic coordinates (longitude and latitude) of observed
ural enemies.18–20 In extensive inventories in the Americas, dozens locations of the pest (S. eridania) in West and Central Africa were
of parasitoids were found to be associated with SAW.21 Among mainly sourced from International Institute of Tropical Agricul-
these, Telenomus remus (Nixon) (Hymenoptera: Platygastridae) and ture (IITA) records. Additional presence points and occurrences
Trichogramma pretiosum (Riley) (Hymenoptera: Trichogrammatidae) of the four modelled parasitoids (Chelonus insularis, Cotesia
were discovered to be the most important naturally occurring egg marginiventris, Trichogramma pretiosum and Telenomus remus)
parasitoids.22,23 In addition to the egg parasitoids, the egg-larval were obtained from the Global Biodiversity Information Facility
and larval parasitoids Chelonus insularis (Cresson) and Cotesia mar- (www.gbif.org) and from published papers. Before the model-
giniventris (Cresson) (both Hymenoptera: Braconidae) emerged ling process, all the records were quality checked and exact
among the most frequent natural enemies and have also proven duplicated records detected were removed for each species
to be efficient against SAW.24 These parasitoid species co-evolved using the R package Environmental Niche Modeling
with SAW and may be potentially relevant for long-term manage- (ENMTools)25 (Table 1).
ment of the pest in areas of invasion or locationswith risks of spread. We used 19 bioclimatic variables from WorldClim version 1.4
The present work aimed to model the global current and future for both present and future climatic conditions. The variables
risks of southern armyworm distribution and habitat suitability of were downloaded from the worldclim (www.worldclim.org)
the four co-evolved SAW parasitoids, namely Chelonus insularis database at spatial resolution of 2.5 arc minutes, ∼4.64 km at
Cresson (Hymenoptera: Braconidae), Cotesia marginiventris equator. Distribution models were calculated for current cli-
(Cresson) (Hymenoptera: Braconidae), Trichogramma pretiosum matic conditions and for two future climate models' representa-
Riley (Hymenoptera: Trichogrammatidae) and Telenomus remus tive concentration pathways (RCP), RCP8.5 ‘a high emissions/
Nixon (Hymenoptera: Platygastridae), through the maximum business as usual scenario’ and RCP6.0 ‘a moderate reduced
entropy (Maxent) algorithm. The result of this horizon scanning emissions scenario’.26,27 Two global climate models (GCMs)
effort will support decision making in the newly invaded from ensemble models were selected (Table 2) for our model-
continent (Africa) and provide global assessments of SAW estab- ling experiments. The first GCM used is one of the warmer
lishment, focusing on the potential role of biological control using CMIP5 models for almost all locations: HadGEM2-ES (4.6 °C cli-
parasitoids as a crucial component of durable climate-smart inte- mate sensitivity). It was coupled with a relatively cool model
grated management of SAW. over much of the land area, GISS-E2-R.27,28
Table 1. Number and sources of species records used for modelling
Total number of records Number of records
Species' name Species' type after data clearing Source of records per source
Spodoptera eridania Pest 238 IITA 63
GBIF 146
Published papers 29
Chelonus insularis Parasitoid 116 IITA 35
GBIF 05
Published papers 76
Cotesia marginiventris Parasitoid 74 IITA 48
GBIF 12
Published papers 14
Telenomus remus Parasitoid 79 IITA 25
GBIF 09
Published papers 45
Trichogramma pretiosum Parasitoid 82 IITA 59
GBIF 23
Published papers 00
wileyonlinelibrary.com/journal/ps © 2021 The Authors. Pest Manag Sci 2021
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
2
Southern armyworm bioclimatic horizon scanning www.soci.org
Table 2. Outline of the two coupled model intercomparison project phase 5 (CMIP5) general circulation models (GCM) tested
GCM Institution Horizontal resolution 2x [CO2] Equilibrium climate sensitivity (°C)
GISS-E2-R* National Aeronautics and Space Association Goddard 2° × 2.5° 2.1
Institute for Space Studies (NASA GISS)
HadGEM2-ES* UK Meteorological Office - Hadley Centre 1.25° × 1.875° 4.6
*HadGEM2-ES (4.6 °C climate sensitivity) is among the warmer CMIP5models for almost all locations, while *GISS-E2-R (2.1 °C) inclines to be relatively
cool over much of the land area.27
2.1.2 Modelling technique (0.75 ≤ AUC ≤ 1) is considered to have a good fit. The TSS is an
The maximum entropy (Maxent) algorithm was used to predict evaluation method of the model's power to detect true presence
the global environmental suitability of S. eridania and its four par- (sensitivity) and true absence (specificity). It is expressed as the
asitoids, namely C. insularis, C. marginiventris, T. pretiosum and sensitivity plus specificity minus one. A TSS > 0.5 indicates good
T. remus. Maxent has been demonstrated to perform well in the predictive power.35
context of developing models using presence data only as
input.29 Its predictions rely on the ability to estimate a distribution 2.2 Determination of the ensemble model for the
of probability based on the physics science principle of ‘maximum parasitoids
entropy’ that satisfies a set of checks from environmental vari- Based on the climate suitability layers obtained for each species of
ables. The output of Maxent is the level of environmental suitabil- parasitoid, we implemented a spatial prioritization using the cur-
ity also considered as potential species ecological niche. Maxent is rent distribution on one hand and the projected future potential
a machine learning approach. It estimates themost uniform distri- distributions on the other hand using Zonation (a decision sup-
bution (maximum entropy) of sampling points compared to back- port system for spatial planning and described in section 1.3
ground locations given the constraints derived from the data.30 above). Using climate suitability layers of parasitoids as features,
Recent developments of the Maxent approach show that the Zonation produces a hierarchical priority ranking across all grid
same maximum likelihood estimates from the Gibbs distribution cells in the study area based on occurrence levels of each species
(an exponential family distribution) used by Maxent can be in each grid cell, while it balances the output simultaneously for all
obtained from an inhomogeneous Poisson process (IPP) model.31 species used in the analysis.34 Core area zonation (CAZ) was
applied to rank areas that have high occurrence levels for a single
2.1.3 Variables selection and models calibration and validation parasitoid species as potentially suitable climates for biocontrol
for the pest and its parasitoids considering the four modelled species.36
To reduce correlations among predictor variables, climatic variables
selection was performed using ENMTools to avoid redundancy, 2.3 Mapping and biorisk analysis
which could affect the accuracy of the model output, especially for 2.3.1 Thresholding and habitat suitability mapping
future climate projections.32,33 The least correlated variables selected The present-day and future layers for the pest and the ensemble
for model calibration had correlation coefficients ρ < 0.8.32 A jack- layers for the parasitoids were imported in ArcMap to map the cli-
knife test was also performed on the selected bioclimatic vari- mate suitability. We converted the continuous predictions of hab-
ables to determine those which contribute best to the models. itat suitability into binary suitability based on a threshold using
All models were run and validated by applying the cross- the SDMtoolbox in ArcGIS.37 The probability of occurrence below
validation method with five replicates (i.e. 5-fold cross-validation). the threshold is considered as unsuitable for the species, while
The method of cross-validation consists of splitting the occur- those with probability greater than the threshold are considered
rence records into five sets where one set is used to evaluate suitable. For the pest species we analyzed habitat suitability based
the model and the four other sets for calibration. The process on two threshold levels: the minimum training presence (MTP)
was iterated five times. Average outputs were used for the pest and the tenth percentile training presence (P10). Applying the
and its parasitoid habitat suitability maps. In addition, average MTP, we assumed that the least suitable habitat at which the spe-
results from individual parasitoid species were combined through cies is known to occur (based on records used to train the model)
a prioritization process in Zonation for identifying combined suit- is the minimum suitability value for the pest species. Using the
ability areas for all parasitoids together. Zonation develops a pri- P10 threshold, we considered that the least suitable climate from
ority ranking. It iteratively ranks sites, at each step removing the the continuous prediction containing 10% of the occurrence
spatial unit that leads to the smallest suitability. In this process, records was not representative for pest habitat suitability. There-
the least suitable climates received the lowest ranks (close to 0) fore, the MTP extends the habitat suitability of the pest while
and the most suitable received the highest ranks (close to 1).34 the P10 threshold minimizes it.
Future predictions were also averaged over the two selected For the parasitoids, we applied only the P10 threshold option for
GCMs for each species and each climatic scenario before priority the ensemble map using the average of the mean (over the five
area analysis in Zonation. replicates) threshold value across the four parasitoid species.
We assessed model accuracy using the area under the receiver
operating characteristic (ROC) curve (area under curve, AUC) and 2.3.2 Spatial analysis of SAW bioclimatic potential
the true skill statistic (TSS). The AUC provides the probability that We analyzed the global bioclimatic potential of Spodoptera eridania
the predictive power of a model is better than random prediction with a focus on the African continent based on each pest suitability
(AUC = 0.5). A model with an AUC value close to unity index defined by the two selected thresholds (MTP and P10). To
Pest Manag Sci 2021 © 2021 The Authors. wileyonlinelibrary.com/journal/ps
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
3
www.soci.org GT Tepa-Yotto et al.
Table 3. AUC and TSS values for the pest and parasitoids models
AUC TSS
Species Mean Standard error Mean Standard error
Spodoptera eridania 0.955 (0.009) 0.7851 (0.0340)
Chelonus insularis 0.934 (0.026) 0.7399 (0.0459)
Cotesia marginiventris 0.905 (0.042) 0.6185 (0.0381)
Telenomus remus 0.921 (0.026) 0.6593 (0.0348)
Trichogramma pretiosum 0.833 (0.041) 0.5155 (0.0456)
Figure 1. Predicted P10 current and future habitat suitability for Spodoptera eridania: (a) current suitable habitats, (b) and (c) future suitable habitats;
1, global; 2, Africa. Future predictions are based on two climate change scenarios for 2050: (b) RCP6.0 and (c) RCP8.5.
achieve this, we subtracted the binary suitability map of the pest its parasitoids performed better than random and showed a
from the binary suitability map of the parasitoids using raster calcu- good predictive power (Table 3). The best predictive model
lation from spatial analysis tools of ArcGIS 10.1. Doing this, we was that of SAW, followed by C. insularis, T. remus and
obtained for each threshold option for the pest a map with three C. marginiventris.
classes: 0 (pest with parasitoids), 1 (parasitoids without pest) and Predictor variables were species-specific (Table S1) and derived
−1 (pest without parasitoids), where the areas of the study region from an initial selection of 11 uncorrelated bioclimatic variables:
classified as −1 represent those with high risk of S. eridania impact. annual mean temperature (Bio1), mean diurnal range (Bio2), iso-
thermality (Bio3), temperature seasonality (Bio4), minimum tem-
perature of coldest month (Bio6), mean temperature of warmest
3 RESULTS quarter (Bio10), annual precipitation (Bio12), precipitation of dri-
3.1 Model performance est month (Bio14), precipitation of wettest quarter (Bio16), precip-
The performance metrics (AUC and TSS) resulting from the itation of driest quarter (Bio17) and precipitation of coldest
models suggested that Maxent models for both SAW and quarter (Bio19).
wileyonlinelibrary.com/journal/ps © 2021 The Authors. Pest Manag Sci 2021
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
4
Southern armyworm bioclimatic horizon scanning www.soci.org
Figure 2. Predicted MTP current and future habitat suitability for Spodoptera eridania: (a) current suitable habitats, (b) and (c) future suitable habitats;
1, global; 2, Africa. Future predictions are based on two climate change scenarios for 2050: (b) RCP6.0 and (c) RCP8.5.
3.2 SAW habitat suitability northerly is higher with the MTP's models. The Congo basin, Mad-
The P10models support that the pest can establish in coastal eco- agascar, most of East Africa except Somalia and easternmost parts
systems of West African regions, including Guinea and Sierra of southern Africa might be prone to the establishment of
Leone in current climates (Fig. 1). The models predict the south- SAW (Fig. 2).
ernmost parts of Central African Republic and Sudan, and the
west of Ethiopia and Kenya as suitable for SAW establishment. 3.3 Parasitoid habitat suitability
Portions of northern Congo and Uganda, and the southern Dem- Many parts of west, central and east Africa, including a large land
ocratic Republic of Congo can be suitable ecoregions. Likewise, area of Madagascar and eastern coasts of Southern Africa, are suit-
eastern parts of Madagascar and northern states of Nigeria from able for the egg parasitoids T. remus and T. pretiosum (Figs S1 and
Niger to Bauchi offer bioclimatic conditions for the establishment S2). Similarly, southern, eastern and southeastern Asia are suitable
of S. eridania. The Benin Republic is totally unsuitable indepen- regions for both egg parasitoids. Our models suggest that
dent of climate conditions (current, RCP6.0 and 8.5). The whole T. remus and T. pretiosum can establish in northern and eastern
of West Africa will become unsuitable in the event of climate coasts of Australia. Southern and western Europe can be particu-
change (RCP6.0 and 8.5) except parts of Guinea, Liberia and Sierra larly suitable for T. pretiosum. The suitability status of all these
Leone. Small portions of southern andwestern Europe, southeast- regions will not change despite global warming (RCP6.0 and
ern China and Australia will become suitable in the event of cli- 8.5), except for additional portions of Australia becoming suitable
mate change (RCP6.0 and 8.5). for T. pretiosum. The models predict that the egg-larval and larval
MTP models predict larger habitat suitability of SAW globally parasitoids C. insularis and C. marginiventris have reduced suitabil-
with greater parts of southern and western Europe, southern, ity coverage compared to the two egg parasitoids (Figs S3
eastern and southeastern Asia, and Australia being particularly and S4). In west Africa, only small areas of Liberia, Guinea and
suitable for the pest independent of climate scenario (Fig. 2). West Sierra Leone will be suitable. C. marginiventris can well establish
Africa except Sahelian countries can sustain S. eridania. The estab- in the Congo basin and Asia compared to C. insularis. Almost all
lishment potential in northernmost parts of Africa, such as por- of Madagascar is suitable for the egg-larval parasitoid
tions of Algeria, Morocco and Tunisia, and Egypt and Libya C. insularis. It can also establish in northeastern Australia whereas
Pest Manag Sci 2021 © 2021 The Authors. wileyonlinelibrary.com/journal/ps
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
5
www.soci.org GT Tepa-Yotto et al.
Figure 3. Georeferenced records for Spodoptera eridania and combined predicted current habitat suitability for its parasitoids Telenomus remus, Tricho-
gramma pretiosum, Chelonus insularis and Cotesia marginiventris.
the eastern parts of the country are suitable for C. marginiventris. matches perfectly that of SAW (Figs 1, 3, 4 and S1–S5) but not
Our models indicate that Europe is totally unsuitable for those of MTP (Figs 2 and S6).
C. insularis while southern Europe can sustain C. marginiventris,
particularly in climate change situations (RCP6.0 and 8.5). Consid-
ered all together, the parasitoid suitability niche can decrease 4 DISCUSSION
with global warming, particularly in the Congo basin and South- 4.1 Model performance
ern Asia (Fig. S5). Conversely, the northern parts of Latin America Our models demonstrated good results based on a bioclimatic
will become more suitable in climate change conditions. P10 analysis approach. However, a range of factors determine species
models show that the bioclimatic suitability of the parasitoids distributions and distribution change dynamics, including biotic
wileyonlinelibrary.com/journal/ps © 2021 The Authors. Pest Manag Sci 2021
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
6
Southern armyworm bioclimatic horizon scanning www.soci.org
Figure 4. Subtracted P10 binary suitability map of Spodoptera eridania from that of its parasitoids Telenomus remus, Trichogramma pretiosum, Chelonus
insularis and Cotesia marginiventris: (a) current suitable habitats, (b) and (c) future suitable habitats; 1, global; 2, Africa. Red gridcells show bioclimatic
potential for the establishment of the pest and in the event of parasitoids absence.
interactions (such as host plant or host/prey availability), evolu- studies on the species ecology which demonstrated the occur-
tionary change and dispersal ability.38 Future climates together rence of the pest in the American tropics.43,44 In southern Europe,
with landscape management may also influence the regulation subtropical dry forests can be particularly suitable for S. eridania
of pests by natural enemies.39 Another pitfall of the method used with a changing climate condition (RCP6.0 and 8.5). The total
in this study is the integration in the models of factors such as irri- P10 unsuitability of the Benin Republic (Fig. 1) currently could
gation. A significant effect of all these predictors might lead to a help to explain why extensive field sampling and pheromone
mismatch between host plants, pests and natural enemies in trapping efforts conducted in the country the past 2 years did
space and time, therefore decreasing the establishment likeli- not discover the pest again. This leads to an assumption that ini-
hood of biocontrol agents.40 Nevertheless, it is widely agreed that tial accidental introduction of the pest to the country did not sur-
bioclimatic analysis can provide useful first estimates and guide vive long with the difficult weather conditions. Conversely, the
decision making for medium- and long-term pest country is suitable for MTP but further outbreaks can only be
management.41,42 explained by migration from Central Africa (Fig. 2). Most of west
Africa would become unsuitable in the event of climate change
4.2 SAW habitat suitability (RCP6.0 and 8.5; Fig. 1), suggesting that the increase in tempera-
ture will be deadly to SAW in contrast to the cool climates of
For the moment S. eridania has been reported in two countries in
southern andwestern Europe and of Asia and Australia. This could
West Africa (Benin and Nigeria) and two in Central Africa
explain why SAW is listed as an A1 quarantine pest by the
(Cameroon and Gabon).8 Our models demonstrate that the pest
European Food Safety Authority (EFSA).11 SAW is already con-
can establish in tropical moist and rain forests in Madagascar,
firmed in west (Nigeria) and central Africa (Cameroon and Gabon).
and west and central Africa. We anticipate that the dispersal of
the pest east and south from Cameroon and Gabon to suitable
areas is likely to happen particularly in cool ecologies. The tropical 4.3 Parasitoid habitat suitability
mountain system in eastern Africa and subtropical humid forests Tropical dry, moist and humid forests are suitable environments
in the Americas, eastern Asia and Australia are also shown to be for T. remus, except large parts of rainforests in the Congo basin
suitable for SAW. These findings are supported by previous under current climates. Tropical shrubland can be suitable for
Pest Manag Sci 2021 © 2021 The Authors. wileyonlinelibrary.com/journal/ps
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
7
www.soci.org GT Tepa-Yotto et al.
the egg parasitoid on the Indian subcontinent and parts of east- crop yields in specific hotspots in the coming years. We anticipate
ern Africa. T. remus can also survive climates of the subtropical that the spread of the pest east and south from Cameroon and
dry forests of Southern Europe in the event of global warming Gabon may happen any time, provided there are suitable path-
(RCP6.0 and 8.5). Almost all rainforests in Latin America will ways. We demonstrated that all the considered parasitoids might
become suitable with increased temperature regimes as the cli- be generally good candidates for biological control of SAW glob-
mate changes (RCP6.0 and 8.5).45,46 The habitat suitability of ally, except they will not establish in specific habitats. The egg par-
T. pretiosum almost mirrors that of the first egg parasitoid asitoids can be excellent candidates for inundative biological
T. remus, except that southern parts of North America are suitable control against SAW. The current work is another demonstration
and southeastern Asia is unsuitable for the former. This parasitoid that the guild of parasitoids shared between SAW and FAW53,54
is a ubiquitous insect present almost everywhere.47 In contrast to represents a perfect opportunity to pursue further work for bio-
T. remus, T. pretiosum can establish almost all over the Congo control of both pests by the studied parasitoids. The interpreta-
basin in current climates. The bioclimate envelop used in the cur- tion of the models in this study is based only on bioclimatic
rent models shows least suitability of T. pretiosumwith indications suitability thresholding, and output data are only as good as the
that the egg parasitoid will not survive in tropical dry and moist input data available. Hence, more work is needed to validate
forests in southern Africa. The egg-larval parasitoid C. insularis these findings in local contexts, taking microclimatic conditions
has the most limited geographic range compared to all other par- into account. This would contribute to calibration of the model
asitoids considered in this study. All of Europe and large parts of inputs for more precise predictions and aid better interpretations
Middle Africa are unsuitable for this egg-larval parasitoid. How- of outputs. In addition, this paper does not consider the efficacy of
ever, C. insularis offers a good opportunity for the biocontrol of individual parasitoids or intraguild competition between and
SAW on Madagascar (in the event of the introduction of this pest among these and other natural enemies present in a given envi-
to the island). The C. marginiventrismodels predict unsuitable por- ronment. The present work is one important step towards devel-
tions of the Indian subcontinent and of coastal West African ecol- oping biocontrol of these two important pests, newly
ogies close to Central Africa.48 Likewise, Australia is only suitable introduced and established on the African continent.
to the larval parasitoid in its eastern parts. High temperatures
(RCP8.5) will tangibly decrease the establishment capabilities of
C. margiventris in the tropical moist and rain forests of Africa but ACKNOWLEDGEMENTS
not in the subtropical humid forests of southern Europe (RCP6.0 The authors thankfully acknowledge the financial support pro-
and 8.5).49,50 vided by the World Bank to projects aimed at Accelerating Impact
Our models show that the climate suitability of the parasitoids of CGIAR Climate Research in Africa (P173398, AICCRA-Ghana).
match well that of SAW (Figs 1, 3, 4 and S1–S5). Most native and Similarly, the authors are grateful to the International Institute of
current invaded regions suitable for SAW are also suitable for Tropical Agriculture (IITA) for strategic funds allocation to the Bior-
the four selected parasitoids combined (Figs 3 and S5). Only the isk Management Facility (BIMAF) partly covering the first author
climates of the southernmost parts of the Sahel and small por- time (GTY). The efforts are also part of the portfolio of the Global
tions of southern Europe are suitable for the SAW parasitoids. Integrating CGIAR Research Program on Climate Change, Agricul-
The ensemble suitability for the selected parasitoids will decrease ture and Food Security (CCAFS), which is carried out with support
under the RCP6.0 and 8.5 climate change scenarios (Fig. S5) partic- from the CGIAR Trust Fund and through bilateral funding agree-
ularly in the Congo basin and on the Indian subcontinent. How- ments. The views expressed in this paper cannot be taken to
ever, most climates of East Asia (east and south-central China) reflect the official opinions of these organizations.
and of the insular regions of southeast Asia will remain suitable
to the parasitoids despite climate change. On the other hand,
the P10 climate suitability range shift of the four parasitoids CONFLICT OF INTEREST
(Fig. 4) suggests potential for biological control-based long term The authors declare that they have no conflict of interest.
management of the pest, as opposed to the worst-case scenario
of the MTP map (Fig. S6) with maximal distribution for the pest
and limited suitability for biocontrol. SUPPORTING INFORMATION
Overall, the egg parasitoids might be excellent candidates for Supporting informationmay be found in the online version of this
inundative biological control, the releases of large numbers of article.
parasitoids (e.g. Trichogramma spp.), as opposed to inoculative
biological control.51 The present study demonstrates that all the
considered parasitoids are generally good candidate biological REFERENCES
control agents of SAWworldwide, except they will not be efficient 1 Kenis M, Auger-Rozenberg M-A, Roques A, Timms L, Péré C, Cock MJW
et al., Ecological effects of invasive alien insects. Biol Invasions 11:21–
in specific habitats of northern Latin America, West Africa and the 45 (2009).
Congo basin under current climate conditions (Fig. 4). However, 2 Nagoshi RN, Brambila J and Meagher RL, Use of DNA barcodes to iden-
we anticipate that additional management methods should com- tify invasive armyworm Spodoptera species in Florida. J Insect Sci
plement biological control, particularly in changing climate condi- 11:1–11 (2011).
tions (RCP6.0 and 8.5).15,52
3 Avtzis DN, Coyle DR, Christopoulos V and Roques A, Biological inva-
sions, national borders, and the current state of non-native insect
species in Greece and the neighbouring Balkan countries. Bull Insec-
tol 70:161–169 (2017).
5 CONCLUSIONS 4 Botha AM, Kunert KJ, Maling'a J and Foyer CH, Defining biotechnolog-
It is likely that SAW detection in west and central Africa occurred ical solutions for insect control in sub-Saharan Africa. Food Energy
Secur 9:1–21 (2020).
shortly after its invasion of the region. It may not be a severe prob- 5 Armstrong KF and Ball SL, DNA barcodes for biosecurity: invasive spe-
lem outside its native range for themoment but may compromise cies identification. Philos Trans R Soc B 360:1813–1823 (2005).
wileyonlinelibrary.com/journal/ps © 2021 The Authors. Pest Manag Sci 2021
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
8
Southern armyworm bioclimatic horizon scanning www.soci.org
6 Hurley BP, Garnas J, Wingfield MJ, Richardson DM, Branco M and ed. by Stocker TF, Qin D, Plattner G-K, Tignor MMB, Allen SK,
Slippers B, Increasing numbers and intercontinental spread of inva- Boschung J et al. Cambridge University Press, Cambridge, United
sive insects on eucalypts. Biol Invasions 18:921–933 (2016). Kingdom and New York, NY, p. 1535 (2013).
7 Goergen G, Kumar PL, Sankung SB, Togola A and Tamò M, First report 27 Ruane AC and McDermid SP, Selection of a representative subset of
of outbreaks of the fall armyworm Spodoptera frugiperda (J E smith) global climate models that captures the profile of regional changes
(Lepidoptera, Noctuidae), a new alien invasive pest in west and Cen- for integrated climate impacts assessment. Earth Perspect 4:1–20
tral Africa. PLoS One 11:1–9 (2016). (2017).
8 GoergenG, Southern armyworm, a new alien invasive pest identified in 28 McDermid SS, Mearns LO and Ruane AC, Representing agriculture in
west and Central Africa. Crop Prot 112:371–373 (2018). earth system models: approaches and priorities for development.
9 Pogue, M. A World Revision of the Genus Spodoptera Guenée:(Lepi- J Adv Model Earth Syst 9:2230–2265 (2017).
doptera: Noctuidae). (2002) Available: https://www.ars.usda.gov/ 29 Alsamadisi AG, Tran LT and Papeş M, Employing inferences across
research/publications/publication/?seqNo115=110657 scales: integrating spatial data with different resolutions to enhance
10 Manuwoto S and Scriber JM, Consumption and utilization of experi- Maxent models. Ecol Model 415:108857 (2020).
mentally altered corn by southern armyworm: iron, nitrogen, and 30 Phillips SB, Aneja VP, Kang D and Arya SP, Modelling and analysis of the
cyclic hydroxamates. J Chem Ecol 11:1469–1483 (1985). atmospheric nitrogen deposition in North Carolina. Int J Glob Environ
11 Bragard C, Dehnen-Schmutz K, Di Serio F, Gonthier P, Jacques MA, Issues 6:231–252 (2006).
Jaques Miret JA et al., Pest categorisation of Spodoptera eridania. 31 Hastie T and Fithian W, Inference from presence-only data; the ongo-
EFSA J 18:1–29 (2020). ing controversy. Ecography 36:231–252 (2013). https://doi.org/10.
12 Montezano DG, Specht A, Sosa-Gómez DR, Roque-Specht VF and De 1111/j.1600-0587.2013.00321.x.
Barros NM, Immature stages of Spodoptera eridania (Lepidoptera: 32 Elith J, Kearney M and Phillips S, The art of modelling range-shifting
Noctuidae): developmental parameters and host plants. J Insect Sci species. Methods Ecol Evol 1:330–342 (2010).
14:1–10 (2014). 33 Fitzpatrick MC and Hargrove WW, The projection of species distribu-
13 Specht A, Montezano DG, Sosa-Gómez DR, Paula-Moraes SV, Roque- tion models and the problem of non-analog climate. Biodiversity
Specht VF and Barros NM, Reproductive potential of Spodoptera eri- Conserv 18:2255–2261 (2009).
dania (Stoll) (Lepidoptera: Noctuidae) in the laboratory: effect of 34 Moilanen A, Anderson BJ, Eigenbrod F, Heinemeyer A, Roy DB,
multiple couples and the size. Braz J Biol 76:526–530 (2016). Gillings S et al., Balancing alternative land uses in conservation prior-
14 Ramirez-Cabral NYZ, Kumar L and Shabani F, Future climate scenarios itization. Ecol Appl 21:1419–1426 (2011).
project a decrease in the risk of fall armyworm outbreaks. J Agric Sci 35 Allouche O, Tsoar A and Kadmon R, Assessing the accuracy of species
155:1219–1238 (2017). distribution models: prevalence, kappa and the true skill statistic
15 Tougeron K, Brodeur J, Le Lann C and van Baaren J, How climate (TSS). J Appl Ecol 43:1223–1232 (2006).
change affects the seasonal ecology of insect parasitoids. Ecol Ento- 36 Moilanen A, Landscape zonation, benefit functions and target-based
mol 45:167–181 (2020). planning: unifying reserve selection strategies. Biol Conserv 134:
16 Li XJ, Wu MFM, Jian G, Bo Y, Wu CL, Chen AD et al., Prediction of 571–579 (2007).
migratory routes of the invasive fall armyworm in eastern China 37 Rabosky DL, Grundler M, Anderson C, Title P, Shi JJ, Brown JW et al.,
using a trajectory analytical approach. Pest Manage Sci 76:454–463 BAMMtools: an R package for the analysis of evolutionary dynamics
(2020). on phylogenetic trees. Methods Ecol Evol 5:701–707 (2014).
17 Torres JA, Lepidoptera outbreaks in response to successional changes 38 Pearson RG and Dawson TP, Predicting the impacts of climate change
after the passage of hurricane Hugo in Puerto Rico. J Trop Ecol 8:285– on the distribution of species: are bioclimate envelope models use-
298 (1992). ful? Global Ecol Biogeogr 12:361–371 (2003).
18 da Silva RS, Kumar L, Shabani F, Ribeiro AV and Picanço MC, Dry stress 39 Stireman JO, Dyer LA, Janzen DH, Singer MS, Lill JT, Marquis RJ et al.,
decreases areas suitable for Neoleucinodes elegantalis (Lepidoptera: Climatic unpredictability and parasitism of caterpillars:
Crambidae) and affects its survival under climate predictions in implications of global warming. Proc Natl Acad Sci USA 102:17384–
South America. Ecol Inf 46:103–113 (2018). 17387 (2005).
19 Musolin DL, Insects in a warmer world: ecological, physiological and 40 Thomson LJ, Macfadyen S and Hoffmann AA, Predicting the effects of
life-history responses of true bugs (Heteroptera) to climate change. climate change on natural enemies of agricultural pests. Biol Control
Global Change Biol 13:1565–1585 (2007). 52:296–306 (2010).
20 Mutamiswa R and Nyamukondiwa C, Superior basal and plastic ther- 41 Biber-Freudenberger L, Ziemacki J, Tonnang HEZ and Borgemeister C,
mal responses to environmental heterogeneity in invasive exotic Future risks of pest species under changing climatic conditions. PLoS
stemborer Chilo partellus Swinhoe over indigenous Busseola fusca One 11:1–17 (2016).
(fuller) and Sesamia food preference and foraging activity of ants: 42 Hannah L, Midgley GF andMillar D, Climate change-integrated conser-
recommendations for field applications of low-toxicity baits view vation strategies. Glob Ecol Biogeogr 11:485–495 (2002).
project climate change and integrated Pest control view project. 43 Sampaio F, Krechemer FS and Marchioro CA, Temperature-dependent
Artic Physiol Entomol 43:108–119 (2018). development models describing the effects of temperature on the
21 Warren DL, Glor RE and Turelli M, ENMTools: a toolbox for comparative development of Spodoptera eridania. Pest Manage Sci 77:919–929
studies of environmental niche models. Ecography 33:607–611 (2010). (2020). https://doi.org/10.1002/ps.6101.
22 de Freitas JG, Takahashi TA, Figueiredo LL, Fernandes PM, Camargo LF, 44 Scriber JM, The behavior and nutritional physiology of southern army-
Watanabe IM et al., First record of Cotesia scotti (Valerio and Whit- worm larvae as a function of plant species consumed in earlier
field, 2009) (hymenoptera: Braconidae: Microgastrinae) comb. nov. instars. Entomol Exp Appl 31:359–369 (1982).
parasitising Spodoptera cosmioides (walk, 1858) and Spodoptera eri- 45 Pomari-Fernandes A, de Freitas Bueno A and De Bortoli SA, Size and
dania (Stoll, 1782) (Lepidoptera: Noctuidae) in Brazil. Rev Bras Ento- flight ability of Telenomus remus parasitoids reared on eggs of the
mol 63:238–244 (2019). factitious host Corcyra cephalonica. Rev Bras Entomol 60:177–181
23 Santos N, Almeida R, Padilha I, Araújo D and Creão-Duarte A, Molecular (2016).
identification of Trichogramma species from regions in Brazil using 46 Pomari AF, Bueno ADF, Bueno RCODF and Menezes ADO, Biological
the sequencing of the ITS2 region of ribosomal DNA. Braz J Biol characteristics and thermal requirements of the biological control
75:391–395 (2015). agent Telenomus remus (hymenoptera: Platygastridae) reared on
24 Pomari AF, Bueno AF, Bueno RCOF and Menezes AO, Telenomus remus eggs of different species of the genus Spodoptera (Lepidoptera:
Nixon egg Parasitization of three species of Spodoptera under dif- Noctuidae). Ann Entomol Soc Am 105:73–81 (2012).
ferent temperatures. Neotrop Entomol 42:399–406 (2013). 47 Cave RD, Biology, ecology and use in pest management of Telenomus
25 Shimbori EM, Onody HC, Fernandes DRR, Silvestre R, Tavares MT and remus. Biocontrol News Inf 21:21–26 (2000).
Penteado-Dias AM, Hymenoptera ‘Parasitica’ in the state of Mato 48 López MA, Martínez-Castillo AM, García-Gutiérrez C, Cortez-Mondaca E
Grosso do Sul, Brazil. Iheringia Sér Zool 107:e2017121 (2017). and Escobedo-Bonilla CM, Parasitoids and entomopathogens asso-
https://doi.org/10.1590/1678-4766e2017121. ciated with fall armyworm, Spodoptera frugiperda, in northern Sina-
26 IPCC, Climate change 2013. The physical science basis, in Working loa. Southwest Entomol 43:867–881 (2018).
Group I Contribution to the Fifth Assessment Report of the Intergovern- 49 Chidawanyika F, Mudavanhu P and Nyamukondiwa C, Global climate
mental Panel on Climate Change. Abstract for Decision-Makers, change as a driver of bottom-up and top-down factors in
Pest Manag Sci 2021 © 2021 The Authors. wileyonlinelibrary.com/journal/ps
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
9
www.soci.org GT Tepa-Yotto et al.
agricultural landscapes and the fate of host-parasitoid interactions. Front Ecol Evol 7:485 (2019). https://doi.org/10.3389/fevo.2019.
Front Ecol Evol 7:1–13 (2019). 00485.
50 Fiaboe KKM, Fernández-Triana J, Nyamu FW and Agbodzavu KM, Cote- 53 Tepa-Yotto GT, Tonnang HEZ, Goergen G, Subramanian S, Kimathi E,
sia icipe Sp. N., a new Microgastrinae wasp (hymenoptera, Braconi- Abdel-Rahman EM et al., Global habitat suitability of Spodoptera fru-
dae) of importance in the biological control of lepidopteran pests giperda (JE smith) (Lepidoptera, Noctuidae): key parasitoids consid-
in Africa. J Hymenopt Res 61:49–64 (2017). ered for its biological control. Insects 12:273 (2021). https://doi.org/
51 Kogan M, Gerling, D, and Maddox, JV, Enhancement of Biological Con- 10.3390/insects12040273.
trol in Transient Agricultural Environments Fisher, Bellows, T. S., Jr. & 54 Bortolotto OC, Silva GV, De Freitas Bueno A, Pomaria AF, Martinelli S,
T. W. Fisher, (eds) 1999.Handbook of Biological Control: Principles and Head GP et al., Development and reproduction of Spodoptera erida-
Applications. Academic Press, San Diego, CA. (1999). nia (Lepidoptera: Noctuidae) and its egg parasitoid Telenomus remus
52 Clarke CW, Calatayud P-A, Sforza RFH, Ndemah RN and (hymenoptera: Platygastridae) on the genetically modified soybean
Nyamukondiwa C, Editorial: Parasitoids' ecology and evolution. (Bt) MON 87701×MON 89788. Bull Entomol Res 104:259–260 (2015).
wileyonlinelibrary.com/journal/ps © 2021 The Authors. Pest Manag Sci 2021
Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
10