Wei 2019

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Research Article

Received: 30 April 2019 Revised: 18 October 2019 Accepted article published: 13 November 2019 Published online in Wiley Online Library: 3 December 2019

(wileyonlinelibrary.com) DOI 10.1002/ps.5684

Potential distribution of two invasive


pineapple pests under climate change
Jiufeng Wei,a Lingfei Peng,b Zhiqiang He,c Yunyun Lua and Fang Wangd*

Abstract
BACKGROUND: The number of global invasive species has significantly increased during the past two centuries due to
globalization. The understanding of species invasion under climate change is crucial for sustainable biodiversity conservation,
community dynamics, ecosystem function, and resource distribution. Two invasive species, Dysmicoccus brevipes (Cockerell)
and D. neobrevipes (Beardsley) have greatly expanded their ranges during recent years. These insects are now considered as
extremely serious pests for various plants, especially pineapple. In addition, they are the primary vectors for pineapple wilt
associated virus. However, the potential distribution range and management strategies for these pests are unclear.

RESULTS: In this study, potential risk maps were developed for these pests with MaxEnt (maximum entropy) based on occurrence
data under different environmental variables. The potential distributions of these pests were projected for 2050s and 2070s
under three climate change scenarios as described in the Special Report on Emissions Scenarios of the Intergovernmental Panel
on Climate Change. Results showed that both pests have similar potential distributions, with high environmental suitability
in South America, Africa and South Asia. In addition, potential range expansions or reductions were predicted under different
climate change scenarios. The annual mean temperature was the most important factor, accounting for 43.4% of D. brevipes
distribution. The minimum temperature of coldest month and mean temperature of coldest quarter was found to be responsible
for 90.3% of D. neobrevipes distribution.

CONCLUSION: This research provided a theoretical reference framework to develop policies in the management and control of
these invasive pests.
© 2019 Society of Chemical Industry

Supporting information may be found in the online version of this article.

Keywords: climate change; invasive pests; MaxEnt; scale insect; species distribution model

1 INTRODUCTION The pink pineapple mealybug (PPM) was first described by


Various invasive insect species have been recently introduced to Cockerell in Jamaica on pineapple plant (Ananascomosus (L.) Merr.)
new areas through trade and transport of goods, and have success- in 1893.10 PPM not only feed on ornamental plants, fruits, and
fully thrived outside of their native regions.1–3 Most of these insects vegetable crops, but also can act as a vector for transmitting
are herbivores, including more than 450 alien insects that have cacao mottle leaf virus (CMLV), cacao swollen shoot virus (CSDV)
colonized forests and urban trees and 14% of these insects have and cacao Trinidad virus (CTV).11,12 Gray pineapple mealybug
caused considerable damage to the trees.4 In addition, invasive (GPM) was first reported by Beardsley in Hawaii (1959) and was
species pose considerable threats to global agriculture. Among later detected in other areas, including Mexico, Sicily, and other
invasive insects, scale insects are major agricultural pests that Neotropical countries.13,14 Tanka et al. discovered this species on
cause serious problems when introduced into new regions with- Japan’s Ishigaki Island.15 In China, GPM was first reported on sisal
out natural predators and more than 23% of the species of all
scale insects were introduced in the United States.5 The pineapple

mealybug (Hemiptera: Pseudococcidae) Dysmicoccus spp. mainly Correspondence to: F Wang, Hebei Key Laboratory of Animal Physiology,
includes two species: pink pineapple mealybug (PPM) D. brevipes Biochemistry and Molecular Biology, College of Life Sciences, Hebei Normal
University, Shijiazhuang, Hebei 050024, China. E-mail: wangf0110@163.com
(Cockerell), and gray pineapple mealybug (GPM) D. neobrevipes
Beardsley, which pose serious threats to commercial pineapple a Department of Entomology, Shanxi Agricultural University, Taigu, Shanxi,
production.6,7 These two species are considered invasive in many China
countries and are native to tropical areas in Central and South b State Key Laboratory of Ecological Pest Control for Fujian and Taiwan Crops,
America.8 Moreover, they exhibit quite similar external morpho- Biological Control Research Institute, Fujian Agriculture and Forestry University,
logical characteristics, including discoidal pores near the eyes, ven- Fuzhou, China
tral multilocular pores restricted to segments VI, VII and VIII, and
c College of Plant Science, Tarim University, Alar, China
absence of oral rim tubular ducts.9 Due to these similarities, the
two species have been considered as biparental and partheno- d Hebei Key Laboratory of Animal Physiology, Biochemistry and Molecular
1652

genetic forms of the same species.9 Biology, College of Life Sciences, Hebei Normal University, Shijiazhuang, China

Pest Manag Sci 2020; 76: 1652–1663 www.soci.org © 2019 Society of Chemical Industry
Potential distribution of Dysmicoccus brevipes and D. neobrevipes www.soci.org

hemp in the Hainan Province of China in 1998 and Guangdong and sources: the Centre of Agriculture and Bioscience International
Guangxi Province in 2008.15–17 In addition, GPM was considered (CABI, https://www.cabi.org/), the Global Biodiversity Information
as a vector of wilt-associated viruses, which severely reduced Facility (GBIF, https://www.gbif.org/) and previously published
pineapple yields.18 Both PPM and GPM can survive on pineapple sources (see supporting information, Table S1). Accurate geo-
and are primary vectors of pineapple mealybug wilt associated graphical coordinates for each distribution location were either
virus.19 PPM is mostly found on the root, crown and lower stem of obtained from the literature or by using Google Earth coordinates.
pineapple plants, and GPM is usually observed on the upper parts Duplicate occurrences and potential errors in distribution data
of plants, such as leaf whorls and developing fruits.7,20 (such as outside of baseline climate or baseline map) were cau-
Thailand, the Philippines, Brazil, China, and India are major tiously checked and excluded. Data indicating country locations
pineapple producers in the world. The global pineapple produc- were removed, and the data referring to county, city and town
tion was 13 444 000 metric tons in 2000.21 The largest pineapple were used. One hundred and nineteen georeferenced occurrences
production is located in Hawaii, accounting for more than half of for PPM (42 native range and 77 invasive range points), and 40
the global production and has an estimated value of 100 million occurrences for GPM (seven native and 33 invasive range points)
dollars (US).8,22 However, pineapple production is seriously threat- were assembled in this study.
ened by PPM and GPM. PPM created a serious economic prob- A distribution site is usually biased to the regions that are easily
lem in pineapple farming, resulting in 40% reduction in the yield accessible for humans, or nearby cities and other areas of human
due to transmission of PMWaV-3 (Pineapple Mealybug Wilt asso- settlement.30,31 Thus, distribution site data can significantly influ-
ciated Virus-3) disease in Cuba.23 In Central Uganda, the incidence ence the outcome of a model due to spatial autocorrelation.32,33
of pineapple mealybug wilt disease ranged from 15% to 100% in In order to reduce spatial autocorrelation and sample bias among
the production fields, which was observed in 2013.24 Moreover, occurrence data, all occurrences for these two pests were filtered
PPM has expanded to 126 countries outside of its native region according to workflow based on the previously described method
and has been recorded on more than 140 genera in 58 families of of Li et al.34,35 A coarse resolution (a grid cell size of 1 × 1 km, simi-
plants.14 GPM afflicted serious damage to sisal hemp by inhibit- lar to the resolution of environment variables) was generated and
ing its growth, leading to death. Almost 6700 ha of sisal hemp a single point was randomly selected from each cell that included
was damaged by GMP, causing 30% production loss in the Guang- one or more sampling points. One hundred and nineteen locations
dong province of China in 2008.17 Since 2006, GPM distribution for PPM and 35 for GPM remained after the filtering process. The
was increased from 2670 ha to 6700 ha in China.17 To date, GPM has distribution list and maps of the locations for PPM and GPM used
been reportedly distributed in 41 countries and feeds on 64 gen- in this study are presented in Figure 1 and Table S2 (see support-
era in 39 families of plants, including banana, pineapple, sweet- ing information). The workflow was conducted in ArcGIS 10.1 (ESRI,
sop, oranges, grape.14 Thus, there is an urgent need to control and Redlands, CA, USA) (http://www.esri.com/).
manage these pests due to their ongoing global spread and wide
host range. 2.2 Environmental data
Currently, no effective control treatments have been developed Climate and topographical information are the main factors used
for these pests. Therefore, an alternative approach is to implement to reflect ecological niches in environmental variables. At large
strict quarantine measures in countries where PPM and GPM are spatial scales, climatic variables are primary factors that deter-
yet to be detected. In addition, increasing evidence suggested that mine species niches and have been frequently used for insect
climate change will further aggravate the impacts of naturalization niche modeling.28,36,37 The species distribution is usually corre-
and subsequent invasion of alien species into new ecosystems and lated with two principal climate factors, such as precipitation and
communities.25,26 Climate change can influence the distribution temperature at a large scale.38 The 19 climatic datasets, includ-
and abundance of invasive insects directly (e.g., by altering the ing a combination of means, extremes and seasonal differences,
occurrence of species and hosts) and indirectly (e.g., via changes were obtained from WorldClim Global Climate Database (http://
in population growth rates, propagule pressure, and spread).27 www.worldclim.org) for current climate.31 Current climate condi-
Therefore, modeling how climate change impacts invasive pests, tions were represented by minima, maxima and average values
can provide vital information for controlling and managing the of monthly, quarterly and annual ambient temperature, and pre-
spread of these pests. cipitation values were recorded from 1950 to 2000. Topographic
Several studies have used species distribution models (SDMs) variables represented by altitude (ALT) were derived from National
to predict the potential impact of invasive species in order to Geophysical Data Center (http://www.ngdc.noaa.gov/mgg/topo/
facilitate planning and mitigation of future impacts.28,29 In this globe.html). All data were represented by 1 km2 (30 arc-seconds)
study, current and future potential distributions of PPM and GPM spatial resolution in the present study.
were estimated based on occurrence data using MaxEnt software. Multicollinearity among climate variables could hinder the anal-
The objectives of this study were: (i) to identify the major climatic ysis of species-environment relationships.39 Thus, Pearson’s cor-
variable that restricts the potential distribution of these pests; (ii) relation coefficient was used for each pairwise comparison of all
to predict the trends of suitable habitat ranges under different 19 environmental variables to minimize multicollinearity. Highly
climate change scenarios; (iii) to provide a theoretical reference correlated variables (r2 ≥ |0.8|) for PPM and GPM were removed
framework for policy-making to manage and control these two from original environmental variables, which filtered those below
invasive pests. the threshold values for final analysis (see supporting informa-
tion, Table S3-1 and S3-2). Multicollinearity was examined using
ENMTools version 1.0.40 Seven climate variables remained for
2 MATERIALS AND METHODS PPM: Bio1 (annual mean temperature), Bio2 (mean diurnal range),
2.1 Occurrence data and sample bias Bio7 (temperature annual range), Bio14 (precipitation of driest
The distribution data of D. brevipes (Cockerell) and D. neobrevipes month), Bio17 (precipitation of driest quarter), Bio18 (precipitation
1653

Beardsley used in this study were obtained from the following of warmest quarter) and Bio19 (precipitation of coldest quarter).

Pest Manag Sci 2020; 76: 1652–1663 © 2019 Society of Chemical Industry wileyonlinelibrary.com/journal/ps
www.soci.org J Wei et al.

Figure 1. The localities of Dysmicoccus brevipes and D. neobrevipes used in current modeling. The base map were created with Natural Earth Dataset (http://
www.naturalearthdata.com/). Gray, unsuitable habitat area; Yellow, low habitat suitability area; Blue, moderate habitat suitability area; Red, highly habitat
suitability area. (A) D. brevipes; (B) D. neobrevipes.

Similarly, six climatic variables remained for GPM: Bio1 (annual Dynamics Laboratory Coupled Model V3 (GFDL-CM3) and the
mean temperature), Bio6 (min temperature of coldest month), Model for Interdisciplinary Research on Climate (MIROC5) for
Bio11 (mean temperature of coldest quarter), Bio17 (precipitation the years 2050 (average for 2041–2060) and 2070 (average for
of driest quarter), Bio18 (precipitation of warmest quarter) and 2061–2080) from the Coupled Model Inter-comparison Project
Bio19 (precipitation of coldest quarter). In addition, the variable Phase 5 (CMIP5) of the fifth assessment of the International Panel
altitude (ALT) was not included (note that Alt was not analyzed for on Climate Change (IPCC).44
multicollinearity but was included in final analysis due to it is not a In 2013, the Fifth Assessment Report was published by the UN’s
climatic variable. Intergovernmental Panel on Climate Change (IPCC), with four rep-
To predict the future potential distribution of these two pests
resentative concentration pathways (RCPs, such as RCP 2.6, RCP
on a global scale, climatic variables for future projections (same
4.5, RCP 6.0 and RCP 8.5).45 Four possible greenhouse gas emis-
as current climate variables) were downloaded from WorldClim
sion trajectories, ranging from lowest (RCP 2.6) to highest (RCP
database. Considering the uncertainty of future climate scenarios,
assessments of the impact should incorporate data from a range 8.5), equivalent to the increments in global radiative forcing val-
of climate models that represent different climate sensitivities to ues in the year 2100, relative to preindustrial values (2.6, 4.5, 6.0
various possible future climate change projections.41,42 Thus, to and 8.5 w/m2 , respectively). The global mean temperature increase
reduce this impact on global circulation models (GCMs) on which ranges from 0.3 to 4.8 ∘ C across all RCPs.46 The high emission sce-
the future climate datasets are based, an ensemble forecasting narios will likely have severe impacts on the future geographic
procedure was used to estimate future distribution range.43 In this distribution of both PPM and GPM. Therefore, RCP 2.6 (the mini-
study, three random GCMs were used: Hadley Global Environment mum greenhouse gas emission scenario) and RCP 8.5 (the maxi-
1654

Model 2-Atmosphere Ocean (HADGEM2-AO), Geophysical Fluid mum greenhouse gas emission scenario) for the years 2050 and

wileyonlinelibrary.com/journal/ps © 2019 Society of Chemical Industry Pest Manag Sci 2020; 76: 1652–1663
Potential distribution of Dysmicoccus brevipes and D. neobrevipes www.soci.org

2070 were selected for potential distribution based on future cli-


Table 1. Relative contribution of each environmental variables to
mate variables. Overall, four climate change scenario/year combi- Maxent model for two invasive species
nations were selected in the current study: RCP 2.6–2050 (average
for the years 2041–2060 under RCP 2.6), RCP 8.5–2050 (average for Variable PPM GPM
the years 2041–2080 under RCP 8.5), RCP 2.6–2070 (average for
Annual mean temperature (Bio1) 43.4 1.0
the years 2041–2080 under RCP 8.5), and RCP 8.5–2070 (average
Mean diurnal range (Bio2) 6.4 –
for the years 2061–2080 under RCP 8.5). All future climate pro-
Min temperature of coldest month (Bio6) – 80.9
jection data were downloaded from the World Climate Database
Temperature annual range (Bio7) 39 –
(Http://www.worldclim.org/). All data used for the ENM had a spa-
Mean temperature of coldest quarter (Bio11) – 9.4
tial resolution of 1 km2 (30 arc-second).
Precipitation of driest month (Bio14) 1.0 –
Precipitation of driest quarter (Bio17) 2.5 2.5
2.3 Modeling approach
Precipitation of warmest quarter (Bio18) 1.4 3.1
Various software packages, such as CLIMEX, BIOCLIM, DOMAIN, Precipitation of coldest quarter (Bio19) 3.8 2.6
GARP, GLMs and GAMs have been used to simulate the poten- ALT 2.5 0.5
tial distribution for invasive species.47 However, MaxEnt (maximum
entropy) was one of the mostly frequently used for present-only The first two most contributing environmental variables are shown in
SDMs for modeling and projecting current and future distribu- bold
tions of invasive species.48,49 Compared with other software (GARP,
Mahalanobis typicalities, and random forests), MaxEnt has out-
future climate scenarios in order to reduce variations among gen-
performed other methods for estimating potential species dis-
eral climate circulation models (GCMs).
tributions, regardless of the number or geographical range of
species records.31,50 Thus, MaxEnt (version 3.4.1) was used for final
analyses.51 2.4 Model evaluation
Feature types and a regularization multiplier were included in The area under the receiver operating characteristic curve (AUC)
MaxEnt to optimize the models and control overparameteriza- has been extensively applied to measure the performance of
tion. These feature types represented different transformations MaxEnt.28,29,33 However, there are disadvantages to this approach
of the covariates and included linear, product, hinge, threshold, due to equal weighting of omission and commission errors, and
and quadratic features, allowing the software to be optimized even AUC cannot provide information on the spatial distribution
for selected species to prevent oversimplified or overcomplicated of model errors.56–58 Therefore, an alternative partial receiver oper-
models.52 The R package ‘ENMeval’, an automated method to exe- ating characteristic (ROC) metric method was developed to evalu-
cute models using an user-specified range of regularization mul- ate the model performance,59 which was implemented using the
tiplier (RM) value, and feature combinations (FCs) was used to NicheToolbox site with 1000 replicates and E = 0.05 (http://shiny
improve the performance of MaxEnt and avoid overfitting.53 The .conabio.gob.mx:3838/nichetoob2/).
RM range from 0.5 to 4, with the increments of 0.5 and 8 FCs,
were examined for two invasive species: (i) Linear (L); (ii) Linear
(L) and Quadratic (Q); (iii) Linear (L), Quadratic (Q) and Hinge (H); 3 RESULTS
(iv) Linear (L), Quadratic (Q), Hinge (H), and Product (P); (v) Lin- 3.1 Model performance for two invasive species
ear (L), Quadratic (Q), Hinge (H), Product (P), and Threshold (T); The model performance for both invasive species demonstrated a
(vi) Quadratic (Q), Hinge (H) and Product (P); (vii) Quadratic (Q), significant predictive ability of partial ROC tests (see supporting
Hinge (H), Product (P), and Threshold (T); (viii) Hinge (H), Prod- information, Fig. S2). The mean value for partial AUC at 0.05 is
uct (P) and Threshold (T). The ENMeval package was executed in 0.951461 for PPM (P < 0.001) and 0.949562 for GPM (P < 0.001).
R (v 3.1.3).54 The ‘checkerboard2’ approach was applied by cal- A ‘10th percentile training presence logistic threshold’ value of
culating the standardized Akaike information criterion coefficient 0.2637 and 0.2386 were obtained for D. brevipes and D. neobrevipes,
(AICc), and lowest delta AICc scores were selected to run the final respectively. These values indicated unsuitable habitat for both
MaxEnt models. The results are shown in Fig. S1 and Tables S4 PPM and GPM.
(see supporting information). Thus, RM = 4, and FC = LQH (Linear,
Quadratic, Hinge) were selected for D. brevipes in final software 3.2 Environments that constrain the model
configuration. Similarly, for D. neobrevipes, RM = 2 and FC = LQH The analysis of single variable contribution showed that annual
(Linear, Quadratic,Hinge) were selected. The logistic output of mean temperature (Bio1) (43.4%) and temperature annual range
MaxEnt was used for all analyses. The suitable and unsuitable habi- (Bio7) (39%) were major factors, influencing the model perfor-
tats or binary presence/absence maps for the two invasive species mance for PPM (Table 1). Other contributing factors were mean
were defined by a 10th percentile training presence logistic thresh- diurnal range (Bio2, 6.4%), precipitation of coldest quarter (Bio19,
old. This threshold has been widely applied in species distribu- 3.8%) and precipitation of driest quarter (Bio17, 2.5%). Minimum
tion modeling, especially when data were collected by different temperature of the coldest month and mean temperature of cold-
collectors.55 In order to explain the reasonable simulated results est quarter showed a total contribution of 90.3% for GPM. Similarly,
of two invasive species, the potential distribution map was reclas- precipitations of warmest quarter (Bio18), driest quarter (Bio17),
sified into four levels: < threshold, unsuitable habitat (no risk); and coldest quarter (Bio19) contributed to the model by 3.1%,
threshold-0.4, low habitat suitability (low risk); 0.4–0.6, moderate 2.5% and 2.6%, respectively. Interestingly, altitude is less affected
habitat suitability (medium risk); and 0.6–1, high habitat suitability in the process of forming a potential distribution map for two
(high risk). Finally, 10-fold cross-validation was used to run MaxEnt pests, 2.5% for PPM and 0.5% for GPM. These results showed that
to prevent random errors from predicted samples. In addition, a thermal conditions influenced the modeled distribution for both
1655

final suitability map was generated by averaging the maps from all PPM and GPM much more than precipitation.

Pest Manag Sci 2020; 76: 1652–1663 © 2019 Society of Chemical Industry wileyonlinelibrary.com/journal/ps
www.soci.org J Wei et al.

Table 2. Area with suitability different climate scenarios (km2 )

Range Level Current RCP2.6–2050 RCP8.5–2050 RCP2.6–2070 RCP8.5–2070

PPM 0.2637–1 ∼4.50 × 107 ∼4.42 × 107 (−1.8%) ∼4.38 × 107 (−2.66%) ∼4.41 × 107 (−2.0%) ∼3.92 × 107 (−12.8%)
0.2637–0.4 ∼1.55 × 107 ∼1.53 × 107 ∼1.58 × 107 ∼1.54 × 107 ∼1.71 × 107
0.4–0.6 ∼2.31 × 107 ∼2.26 × 107 ∼2.21 × 107 ∼2.25 × 107 ∼2.14 × 107
0.6–1 ∼6.40 × 106 ∼6.27 × 106 (−2.0%) ∼5.8 × 106 (−10.3%) ∼6.13 × 106 (−4.2%) ∼3.92 × 106 (−38.75%)
GPM 0.2386–1 ∼2.20 × 107 ∼2.36 × 107 (7.2%) ∼2.55 × 107 (15.9%) ∼2.38 × 107 (8.1%) ∼2.99 × 107 (35.9%)
0.043–0.4 ∼9.82 × 106 ∼1.17 × 107 ∼1.36 × 107 ∼1.18 × 107 ∼1.47 × 107
0.4–0.6 ∼8.69 × 106 ∼8.94 × 106 ∼9.22 × 106 ∼8.99 × 106 ∼1.22 × 107
0.6–1 ∼3.53 × 106 ∼2.92 × 106 (−17.2%) ∼2.72 × 106 (−22.9%) ∼2.96 × 106 (−16.1%) ∼3.06 × 106 (−13.3%)

3.3 Current invasive pattern temperature (Bio1) of 27 to 32 ∘ C, minimum temperature of coldest


Based on distribution records and existing environmental vari- month (Bio6) of 20 to 24 ∘ C, mean temperature of coldest quar-
ables, current potential distribution map of PPM and GPM are illus- ter (Bio11) of 25 to 27 ∘ C, precipitation of driest quarter (Bio17)
trated in Figure 1(A) and (B). between 180 to 650 mL, precipitation of warmest quarter (Bio18)
MaxEnt predictions indicated that the potential distribution between 400 to 1700 mL, precipitation of coldest quarter between
range of PPM was almost globally distributed, with the main 480 to 2200 mL(Fig. 3). However, Alt showed no strong relation to
potential distribution areas located in South America, Africa and high habitat suitability in the model for GPM.
Asia (Fig. 1(A)). In about approximately 4.50 × 107 km2 of the
world’s land mass, approximately 6.40 × 106 km2 (∼14.2% of total 3.4 Future invasive area
potential invasive area) exhibited high habitat suitability (high The MaxEnt models for potential distribution of PPM and GPM
risk) and 2.31 × 107 km2 (∼51.3% of total potential invasive area) under future climate change scenarios RCP 2.6 and RCP 8.5 for the
showed moderate habitat suitability. In Asia, largest high habi- years 2050 and 2070 are illustrated in Figures 4 and 5.
tat suitability areas under current climate scenarios were located
in Southern China, Vietnam, Laos, Thailand, Burma, Malaysia, the
Philippines, a small part of India, and Indonesia. Northeastern 3.4.1 PPM
Australia showed high habitat suitability in the model’s projec- The total potential distribution of PPM showed that areas of suit-
tions for PPM potential distribution under current climate. In addi- ability would decrease in 2050 under emission scenario RCP 2.6 by
tion, Nicaragua, Costa Rica, Panama, Columbia, Uruguay, and a approximately 4.42 × 107 km2 of the world’s land mass, account-
small part of Brazil showed high habitat suitability within cen- ing for approximately 1.8% of current predicted area (Fig. 4(A) and
tral and South America. Further, Africa, Madagascar, Liberia, Ivory Table 2). Further, the total predicted range would also decrease
Coast, Ghana, Nigeria, Cameroon, Congo, Angola, and south of by approximately 3.26%, which is approximately 4.38 × 107 km2
the Republic of South Africa also exhibited high habitat suitability under RCP 8.5–2050 (Fig. 4(C) and Table 2). Meanwhile, high suit-
(Fig. 1(A), Table 2). ability areas would decrease to approximately 6.27 × 106 km2
Similarly, the potential distribution map showed that all habi- and approximately 5.80 × 107 km2 , which appropriates to approx-
tat suitability for GPM were located from near north latitude of imately 2.0% and 10.3% reduction of current predicted area under
40∘ to near south latitude of 33∘ , which is mainly concentrated the RCP 2.6–2050 and RCP 8.5–2050, respectively.
within Asia, Africa, and Central and South America (Fig. 1(B)). Similarly, the potential distribution range of all suitable habi-
Within approximately 2.20 × 107 km2 of this area, approximately tats would expand to approximately 4.41 × 107 km2 under RCP
3.53 × 106 km2 (∼16.0% of total potential invasive area) exhibited 2.6–2070 (a ∼2.0% decrease compared to the current suitable
high habitat suitability (high risk), and 8.69 × 106 km2 (∼39.5% of habitat). In addition, the area of high suitability habitats declined
total potential invasive area) showed moderate habitat suitabil- in some regions to approximately 6.13 × 106 km2 (a decrease of
ity (Table 2). Southern and Central America, Southeast of Mex- ∼4.2% compared to the current range) (Fig. 4(B) and Table 2). In
ico, Guatemala, Nicaragua, Costa Rica, Panama, Cuba, Columbia, 2070, the potential distribution area would decrease to approx-
Venezuela, and northeast Brazil exhibited high habitat suitabil- imately 3.92 × 107 km8 under RCP 8.5–2070 (an decrease of
ity under current climatic conditions. Further, Liberia, Ivory Coast, ∼12.8% compared to the current suitable habitat), meanwhile,
Ghana, Nigeria, Cameroon, Gabon, and Congo also showed high highly suitable habitat would also reduce to approximately
habitat suitability within Africa. Another high habitat suitability 3.92 × 107 km6 under RCP 8.5–2070 (decrease of ∼38.75%)
region was found in Southern Asia, including the Southern India, (Fig. 4(D) and Table 2).
the Southern coastal zone of China, Vietnam, Malaysia, Indonesia,
and the Philippines (Fig. 1(B),Table 2). 3.4.2 GPM
The response curves (Fig. 2) of distribution model of PPM indi- Under RCP 2.6, it was estimated that the total area of habitat suit-
cated high probability of occurrence of this species in regions ability would increase in 2050 and would expand to approximately
with the temperature annual range (Bio7) of 12 to 18 ∘ C, pre- 2.36 × 107 km2 (an increase of ∼7.2% over the current suitable
cipitation of warmest quarter (Bio18) of 400 to 1800 mL, precip- area) (Fig. 5(A) and Table 2). However, the potential distribution
itation of coldest quarter (Bio19) of 300 to 2700 mL,altitude of area with highly suitable habitat decreased under RCP 2.6–2050,
below 300 m. However, Bio1, Bio2, Bio14 and Bio17 showed no with a decrease of approximately 17.2% compared to current
relation to high habitat suitability in the model for PPM. Similarly, highly suitable habitat. MaxEnt predicted that the area of suitable
1656

there was a high probability of GPM in areas with annual mean habitats under RCP 2.6 would be approximately 2.38 × 107 km7 in

wileyonlinelibrary.com/journal/ps © 2019 Society of Chemical Industry Pest Manag Sci 2020; 76: 1652–1663
Potential distribution of Dysmicoccus brevipes and D. neobrevipes www.soci.org

Figure 2. Response curves showing the relationships between the probability of presence of Dysmicoccus brevipes and eight bioclimatic variables. Values
shown are average over 10 replicate runs: blue margins show ± SD calculated over 10 replicates.

2070, which is an increase of approximately 8.1% compared to worldwide, especially for Asia, Australia and Africa. The unsuitable
the current area. However, highly suitable areas would decrease areas for the current distribution of these two pests showed that
to 2.96 × 106 km6 (∼16.1% decrease) (Fig. 5(B) and Table 2). they were inhibited by extreme temperatures or rainfall and these
Similarly, under RCP 8.5–2050, the potential distribution area areas are expected to either increase or decrease as the climate
would increase to approximately 2.55 × 107 km2 , which is an changes.
increase of approximately 15.9% of the current suitable area. In this study, the results indicated that thermal conditions were
Yet, the potential area of distribution with highly suitable habi- the most important environmental variables limiting the distri-
tat decreased in this scenario (a decline of ∼22.9% over the cur- bution map of PPM and GPM. Temperature is the driving force
rent area) (Fig. 5(C) and Table 2). Under RCP 8.5–2070, suitable for mealybug development and survival, although developmen-
habitat now covered approximately 2.99 × 107 km7 , showing an tal times and thresholds might differ among species.60,61 Annual
increase of approximately 35.9% over the current suitable habi- mean temperature (Bio1) is the most important climatic vari-
tat area. However, highly suitable area would decrease to approxi- able that defines the current global distribution of PPM. Previous
mately 3.06 × 106 km7 (a decrease of ∼13.3% over the current area) study on the temperature threshold for PPM demonstrated that
(Fig. 5(D) and Table 2). 28.6 ∘ C was the optimal temperature for the survival of its nymphal
In summary, the results obtained in this study suggested that stage.62 However, the most suitable temperature for PPM develop-
the potential distribution of PPM would decrease and GPM would ment was 30 ∘ C.62 Further, the mean temperature range of 8–35∘ C
increase under future climate change scenarios. In addition, highly were estimated for the life cycle of PPM, which suggested that
suitable habitats would decrease in most regions for both PPM and this species can thrive in different table grape-producing areas in
GPM under future climate change. Brazil. The simulated results showed that PPM population would
increase within the temperature range of 12–18∘ C of a year. Labo-
ratory temperature results were different from field temperatures,
4 DISCUSSION which can be attributed to the complexity of natural conditions in
D. brevipes (PPM) and D. neobrevipes (GPM) have been regarded field environments. The temperature values for the most suitable
as the pests of economically important plants in some parts of climatic conditions for PPM growth and some current unsuitable
the world.6,60 These species are native to South America and are habitats (especially Yangtze River banks) may become highly suit-
considered potential pests for agricultural crops and forests.61 able in the future because the temperature of current unsuitable
Therefore, the potential distribution maps of their expansion areas are increasingly hot due to global warming. On the contrary,
1657

are urgently needed for economic and quarantine purposes few areas, such as Southern China and Northern Australia may

Pest Manag Sci 2020; 76: 1652–1663 © 2019 Society of Chemical Industry wileyonlinelibrary.com/journal/ps
www.soci.org J Wei et al.

Figure 3. Response curves showing the relationships between the probability of presence of Dysmicoccus neobrevipes and seven bioclimatic variables.
Values shown are average over 10 replicate runs: blue margins show ± SD calculated over 10 replicates.

become unsuitable as a result of global warming. The reduction in suitable habitat and moderate suitable would increase in all four
habitat suitability in these potential regions might limit the spread scenarios relative to the current area. The model developed in this
and further development of PPM outside its current suitable range. study to predict the changes in PPM and GPM distribution patterns
In addition, our model showed that several environmental vari- on a global scale would provide a platform to develop monitoring
ables could explain the current distribution of GPM worldwide. The strategies to detect future infestations in currently non-infested
results of potential distribution for GPM suggested that their popu- regions. The early warnings for invasive species spread and mon-
lation would increase at minimum temperature of 20–24 ∘ C in the itoring the direction of spread are critical processes. Distribution
coldest month, and the mean temperature for highly suitable areas modeling of invasive species is a cost-effective method to iden-
would be 25–27 ∘ C in the coldest quarter. A lab study in China tify the potential distribution areas of pests, which can help in the
showed that 29 ∘ C was the most suitable temperature for GPM sur- development of strategies to control them within these areas.66
vival from first instar nymph to adult stage. Similarly, a previous Our models suggested that Africa and Australia have suitable habi-
laboratory study indicated that most suitable temperature range tats for the survival of GPM and Europe for PPM. However, various
for GPM was 23-29 ∘ C, which was consistent with our results.6 data and information or hazard records have not been reported in
Another study indicated that the adult female exhibited highest these regions at present. To date, GPM has not been reported in
fecundity at 29 ∘ C.63 Further, Chen and his colleagues reported that Africa, but there was an outbreak of pineapple mealybug wilt dis-
GPM population increased at 20 to 36 ∘ C, and the suitable temper- ease in Central Uganda.67 Therefore, these areas should formulate
ature for GPM development was 24 to 28 ∘ C in China.64 The map strict quarantine measures as in Australia and Europe, especially
also showed that the potential distribution model did not reveal within vulnerable pineapple productions that have suitable habi-
altitude as crucial factor for two pests (2.5% for PPM and 0.5% tat for PPM and GPM. However, some countries in Sub-Saharan
for GPM). Africa, such as Liberia, Ivory Coast, Ghana, Nigeria, Cameroon, and
Various studies have reported that global warming would greatly Gabon, strict quarantine policies are highly unlikely to succeed
influence species distributions by causing expansion, shifts, or due to limited resources.68 Early detection of new invasions is an
contractions in the species ranges.28,65 Our model suggested that important step in any prevention strategy. Incomplete data on
the proportion of both highly suitable habitat and the suitable the distribution of invasive species often hamper early detection
habitat for PPM in all four future climatic scenarios would decrease and other efforts to minimize their impacts.69 Citizen science is
relative to the current area (Table 2). It was observed that, for an alternative method, which encourages the public to help col-
PPM, low habitat suitability would decrease by 2070, and moderate lect large volumes of information by conducting surveys for target
1658

suitable habitat would decline in all four scenarios. For GPM, low invasive pest species.70 This method was effective in monitoring

wileyonlinelibrary.com/journal/ps © 2019 Society of Chemical Industry Pest Manag Sci 2020; 76: 1652–1663
Potential distribution of Dysmicoccus brevipes and D. neobrevipes www.soci.org

Figure 4. Future species distribution models of Dysmicoccus brevipes on global scale under different climate scenarios predicted by MaxEnt. Gray,
unsuitable habitat area; Yellow, low habitat suitability area; Blue, moderate habitat suitability area; Red, highly habitat suitability area. The base map
was created with Natural Earth Dataset (http://www.naturalearthdata.com/). (A) RCP 2050-2.6; (B) RCP 2070-2.6; (C) RCP 2050-8.5; (D) RCP 2070-8.5.

new invasions and determining new areas where treatments were Pineapple is a tropical fruit with an abundant source of vitamins
required, such as South Africa,71 USA70 and Australia.72 In addi- (especially vitamin C), minerals, substantial calcium, potassium
tion, the results showed that PPM and GPM have many overlapping and fiber contents.73 The major pineapple growing countries in the
areas of potential distribution, so they can be dealt together when world are Brazil, Thailand, the Philippines, Costa Rica, China, India,
1659

developing control and quarantine measures. and Indonesia.73 Our model indicated that east of Brazil, south

Pest Manag Sci 2020; 76: 1652–1663 © 2019 Society of Chemical Industry wileyonlinelibrary.com/journal/ps
www.soci.org J Wei et al.

Figure 5. Future species distribution models of Dysmicoccus neobrevipes on global scale under different climate scenarios predicted by MaxEnt. Gray,
unsuitable habitat area; Yellow, low habitat suitability area; Blue, moderate habitat suitability area; Red, highly habitat suitability area. The base map was
created with Natural Earth Dataset (http://www.naturalearthdata.com/). (A) RCP 2050-2.6; (B) RCP 2070-2.6; (C) RCP 2050-8.5; (D) RCP 2070-8.5.

of Thailand, the northern portion of the Philippines, Costa Rica, and Brazil from 2050 to 2070 under the RCP-2.6. Therefore, it
South China and South India showed high potential distribution is necessary to strengthen the quarantine measures for these
ranges for PPM and GPM. It was further observed that the suitable two pests in these countries with large pineapple production
area for PPM in these countries would increase under climate industries.
change scenarios. Both suitable area and highly suitable area Interestingly, results showed that these two morphologically
would increase in Brazil from 2050 to 2070 under the RCP-2.6. similar pests could occupy similar potential distribution ranges
1660

For GPM, the suitable area would also increase in south China with great overlap in many regions, such as South America

wileyonlinelibrary.com/journal/ps © 2019 Society of Chemical Industry Pest Manag Sci 2020; 76: 1652–1663
Potential distribution of Dysmicoccus brevipes and D. neobrevipes www.soci.org

(Colombia, Venezuela, Ecuador, Eastern Brazil), Asia (Southern Thirdly, PPM and GPM have been associated with various ant
India, Vietnam, Malaysia, Indonesia and the Philippines), and species. This mutualistic relationship between ants and pineapple
Northern Australia. Further, the potential distribution area of PPM mealybugs helps in increasing their population in the pineapple
and GPM showed high suitability in Africa. Thus, based on the fields because ants protect them from their natural enemies.85
results of this study, it can be inferred that species with similar mor- The ant genera Pheidole and Solenopsis are mainly associated with
phological or phenotypic characteristics might have similar poten- PPM,85 and Pheidole and Tetramorium are associated with GPM
tial distributions, especially for high suitability areas. in some areas.15 Thus, the control of ants associated with GPM
All models in this study showed an expansion in the poten- and PPM could be another strategy for their management and
tial distribution range for PPM and GPM under climate change. treatment.86 Previous studies suggested that the control of ants
Results were found to be consistent with other studies, espe- in the pineapple or crop fields can significantly reduce mealybug
cially for invertebrates.74 Several studies suggested that some populations, especially PPM.85,87
invasive species are likely to experience range reductions due
to climate change.28,45,75 The range contractions as predicted in
this study should provide important information for the man- 5 CONCLUSION
agement and eradication of these two pests. Actually, reduced In brief, regions that have been invaded by PPM and GPM should
climatic suitability on currently invaded areas may reduce the manage and control their spread through different pest manage-
competitiveness of these invasive species, which may ultimately ment and surveillance strategies. This study provided the first
lead to their decline.74,76 In addition, some other abiotic and/or potential distribution map of PPM and GPM based on current and
biotic factors, which were not included in the models, might future climate scenarios. The range of suitable habitats will con-
have rapidly changed over time elucidating significant range con- tinue to increase under future climate conditions compared to the
traction of the species. There are many factors that can influ- current climate conditions. The present study provided a reference
ence and limit the potential distribution of invasive species. In to manage the pests and develop policies for their control. More-
this study, only the constraints caused by climate variables were over, strict quarantine measures should be used in areas where
considered, but other biotic factors, including host-plant avail- PPM and GPM have not yet been reported based on the results of
ability were not considered.77 Pineapple is one of the hosts for current research. Different variables, such as host-plant availabil-
both GPM and PPM, thus, their distribution range is affected by ity and species’ dispersal capacity should be considered in future
the distribution of pineapple. Human-mediated transport or trade studies to develop a concrete strategy for the control of these inva-
might be another factor, which increases the invasive risk of alien sive pests.
species.78 Another possible factor is a species’ dispersal capacity,
which also influenced the distribution range of these two inva-
sive species. 72 The process of invasion itself may induce strong
selection pressure on the dispersive abilities of species, resulting
ACKNOWLEDGEMENTS
This study was supported by the National Natural Science Foun-
in increased dispersal during the colonization of novel regions.79,80
dation of China (31802001) and Shanxi Agricultural University of
Interspecific interactions can also influence the accuracy of fore-
Science and Technology Innovation fund projects (2015YJ03).
casted models by affecting the species range expansion rates.81
In addition, other variables also influence the result of simulated
models, such as geographic barriers, land use and cultivation
practices of host plants. These factors should be considered in SUPPORTING INFORMATION
future research. Supporting information may be found in the online version of this
article.
Control methods and reliable management strategies are key
approaches for preventing the spread of invasive species. Various
factors should be considered, including climate change, land use, REFERENCES
and human activity.76 Along with strict quarantine measures, other 1 Hulme PE, Bacher S, Klotz S, Kuhn I, Minchin D, Nentwig W et al.,
approaches and management strategies for controlling PPM and Grasping ate the routes of biological invasions: a framework
GPM have been proposed as follows, based on the results of for integrating pathways into policy. J Appl Ecol 45:402–414
(2008).
current and future potential distribution studies.
2 Seebens H, Blackburn TM, Dyer EE, Genovesi P, Hulme PE, Jeschke
Firstly, insecticides can be an alternative method to control PPM JM et al., Global rise in emerging alienspecies results from
and GPM. Insecticides, such as soaps, tobacco extract, mercury, increased accessibility of new source pools. Proc Natl Acad Sci
sulfur, and oils have been recommended in field applications.8 It U S A 115:E2264–E2273 (2018).
has been reported that Sicilian lemon oil was highly effective to 3 Liebhold AM, Yamanaka T, Roques A, Augustin S, Chown SL, Brocker-
hoff EG et al., Plant diversity drives global patterns of insect inva-
control PPM.82 sions. Sci Rep 8:12095 (2018).
Secondly, the discovery of natural predators of PPM and 4 Aukema JE, McGUllough DG, Holle BV, Liebhold AM, Britton K and
GPM can be an effective approach to reducing their spread. Frankel SJ, Historical accumulation of nonindigenous forest pests in
Entomopathogenic nematodes (Heterorhabdtis indica Poliner, the continental united states. Bio Science 60:886–897 (2010).
5 Miller DR, Miller GL, Hodges GS and Davidson JA, Introduced scale
Karunakar & David) have great potential to control PPM due
insects (Hemiptera: Coccoidea) of the United States and their
to their high mortality rates upon PPM infection.83 One natural impact on US agriculture. P Entomol Soc Wash 107:123–158 (2005).
predator of GPM was evaluated for its effectiveness in controlling 6 Qin ZQ, Qiu BL, Wu JH, Cuthbertson AGS and Ren SX, Effect of tem-
GPM. On the other hand, there are 24 natural enemies belonging perature on the life history of Dysmicoccus neobrevipes (Hemiptera:
to 13 genera in 7 families for PPM, including Rhinoleucophenga Pseudoccocidae): an invasive species of gray pineapple mealybug
in south China. Crop Prot 45:141–146 (2013).
capixabensi, which is a potential predator of PPM.84 However, no 7 Suma P, Mazzeo G, Pergola AL, Nucifora S and Russo A, Establishment
biological control field studies have been performed within the of the pineapple mealybug Dysmicoccus brevipes (Hemiptera: Pseu-
1661

PPM infected areas. dococcidae) in Italy. EPPO Bull 45:218–220 (2015).

Pest Manag Sci 2020; 76: 1652–1663 © 2019 Society of Chemical Industry wileyonlinelibrary.com/journal/ps
www.soci.org J Wei et al.

8 Rohrbach KG, Beardsley JW, German TL, Reimer NJ and Sanford tick, Haemaphysalis longicornis in North America. Sci Rep 9:498
WG, Mealybug wilt, mealybugs, and ants of pineapple. Plant Dis (2019).
72:558–565 (1988). 34 Li GQ, Du S and Guo K, Evaluation of limiting climatic factors and sim-
9 Lim WH, Studies on the bisexual race of Dysmicoccus brevipes (Ckll): ulation of climatically suitable habitat for Chinese sea buckthorn.
its bionomics and economic importance. Malays Agr J 49:254–267 PLoS One 10:e0131659 (2015).
(1973). 35 Rodríguez-Castañeda G, Hof RA, Jansson R and Harding LE, Predicting
10 Cockerell TDA, The west Indian species of Dactylopius. The Entomol the fate of biodiversity using species’ distribution models: enhanc-
26:177–179 (1893). ing model comparability and repeatability. PLoS One 7:e44402
11 CABI. Dysmicoccus brevipes (pineapple mealybug). Invasive species (2012).
compendium. Datasheets, maps, images, abstracts and full text on 36 Luoto M, Virkkala R and Heikkinen RK, The role of land cover in bio-
invasive species of the world. Available: http://www.cabi.org/isc/ climatic models depends on spatial resolution. Glob Ecol Biogeogr
datasheet/20248 [1 January 2019]. 16:34–42 (2007).
12 Harris KF, Arthropod and nematode vectors of plant viruses. Annu Rev 37 Medley KA, Niche shifts during the global invasion of the Asian tiger
Phytopathol 19:391–426 (1981). mosquito, Aedes albopictus Skuse (Culicidae), revealed by reciprocal
13 Beardsley JW, On the taxonomy of pineapple mealybugs in Hawaii, distribution models. Glob Ecol Biogeogr 19:122–133 (2010).
with a description of a previously unnamed species (Homoptera: 38 Bell DM, Bradford JB and Lauenroth WK, Early indicators of change:
Pseudococcidae). P Hawaiian Entomol Soc 17:29–37 (1959). divergent climate envelopes between tree life stages imply range
14 García Morales M, Denno BD, Miller DR, Miller GL, Ben-dov Y and Hardy shifts in the western United States. Glob Ecol Biogeogr 23:168–180
NB, ScaleNet: a literature-based model of scale insect biology and (2014).
systematics. Database Available: http://scalenet.info [12 November 39 Heikkinen RK, Araújo MB, Virkkala R, Thuiller W and Sykes MT, Methods
2018]. and uncertainties in bioclimatic envelop modeling under climate
15 Tanka H and Uesato T, New records of some potential pest mealybugs change. Prog Phys Geog 30:1–27 (2006).
(Hemiptera: Coccoidea: Pesudococcidae) in Japan. Appl Entomol 40 Warren DL, Glor RE and Turelli M, ENMTools: a toolbox for comparative
Zool 47:413–419 (2012). studies of environmental niche models. Ecography 33:607–611
16 Qin ZQ, Wu JH, Ren SX and Wan FH, Risk analysis of the alien invasive (2010).
gray pineapple mealybug (Dysmicoccus neobrevipes Beardsley) in 41 Guisan A, Tingley R, Baumgartner JB, Naujokaitis-Lewis I, Sutcliffe PR,
China. Sci Agr Sin 43:626–633 (2010). Tulloch AIT et al., Predicting species distributions for conservation
17 Wu JH, Lin L, Ren SX and Qin ZQ, New pest on sisal in Leizhou decisions. Ecol Lett 16:1424–1435 (2013).
Peninsula-A briefing about gray pineapple mealybug, Dysmicoccus 42 Sultana S, Baumgartner JB, Dominiak BC, Royer JE and Beaumont LJ,
neobrevipes Beardsley. Guangdong Agr Sci 4:47–48 (2008). Potential impacts of climate change on habitat suitability for the
18 Sether DM, Ullman DE and Hu JS, Transmission of pineapple mealybug Queensland fruit fly. Sci Rep 7:13025 (2017).
wilt-associated virus by two species of mealybug (Dymicoccus spp.). 43 Araujo MB and New M, Ensemble forecasting of species distributions.
Phytopathology 88:1224–1230 (1998). Trends Ecol Evol 22:42–47 (2007).
19 Sether DM, Melzer MJ, Busto J, Zee F and Hu JS, Diversity and 44 Moss RH, Edmond JA, Hibbard KA, Manning MR, Rose SK, van Vuuren
mealybug transmissibility of Ampeloviruses in pineapple. Plant Dis DP et al., The next generation of scenarios for climate change
89:450–456 (2005).
research and assessment. Nature 463:747–756 (2010).
20 Ito K, Studies on the life history of the pineapple mealybug, Pseudo-
45 Petersen MJ, Evidence of a climatic niche shift following North Ameri-
coccus brevipes (Ckll.). J Econ Entomol 31:291–298 (1938) (1938).
can introductions of two crane flies (Diptera: genus Tipula). Biol Inva-
21 FAO, Production Yearbook. Food and Agriculture Organization of the
sions 4:885–897 (2013).
United Nations, Rome (2001).
46 van Vuuren DP, Edmonds J, Kainuma M, Riahi K, Thomson A, Hibbard K
22 German TL, Ullman DE and Gunasinghe UB, Mealybug wilt of pineap-
et al., The representative concentration pathways: an overview. Clim
ple. Adv Dis Vector Res 9:241–259 (1992).
Change 109:5–31 (2011).
23 Hernandez L, Ramos PL, Pena M and Perez JM, First report of pineapple
47 Byeon DH, Jung JM, Jung SH and Lee WH, Prediction of global geo-
mealybug wilt associated virus-3 infesting pineapple in Cuba. New
Dis Rep 22:18 (2001). graphic distribution of Metcalfa pruinosa using CLIMEX. Entomol Res
24 Bua B, Karungi J and Kawube G, Occurrence and effects of pineapple 48:99–107 (2017).
mealybug wilt disease in Central Uganda. J Agr Sci Tech 3:410–416 48 Cooper DM, Dugmore AJ, Scharf AK, Wilting A and Kitchener AC,
(2013). Predicted Pleistocene-Holocene range shifts of the tiger (Panthera
25 Ekesi S, Meyer MD, Mohamed SA, Virgitio M and Borgemeister C, tigris). Divers distrib 22:1199–1211 (2016).
Taxonomy, ecology, and management of native and exotic fruit fly 49 Marchioro CA and Krechemer FS, Potential global distribution of
species in Africa. Annu Rev Entomol 61:219–238 (2016). Diabrotica species and the risks for agricultural production. Pest
26 Paini DR, Sheppard AW, Cook DC, De Barro PJ, Worner SP and Thomas Manag Sci 74:2100–2109 (2018).
MB, Global threat to agriculture from invasive species. Proc Natl Acad 50 Elith J, Graham CH, Anderson RP, Dudik M, Ferrier S, Guisan A et al.,
Sci U S A 113:7575–7579 (2016). Novel methods improve prediction of species’ distributions from
27 Lantschner MV, Villacide JM, Garnas JR, Croft P, Carnegie AJ, Liebhold occurrence data. Ecography 29:129–151 (2006).
AM et al., Temperature explains variable spread rates of the invasive 51 Phillips SJ, Anderson RP, Dudik M, Schapire RE and Blair ME, Open-
woodwasp Sirex noctilio in the Southern Hemisphere. Biol Invasions ing the black bos: an open-source release of Maxent. Ecography
16:329–339 (2014). 40:887–893 (2017).
28 Wei JF, Zhao Q, Zhao WQ and Zhang HF, Predicting the poten- 52 Morales NS, Fernandez IC and Baca-Gonzalez V, MaxEnt’s parameter
tial distributions of the invasive cycad scale Aulacaspis yasumatsui configuration and small sample: are we paying attention to recom-
(Hemiptera: Diaspididae) under different climate change scenarios mendations? A systematic review. Peer J 5:e3093 (2017).
and the implications for management. Peer J 6:e64832 (2018). 53 Muscarella R, Galante PJ, Soley-Guardia M, Boria RA, Kass JM, Uriarte M
29 Chen YT, Vasseur L and You MS, Potential distribution of the inva- et al., ENMeval: an R package for conducting spatially independent
sive loblolly pine mealybug, Oracella acuta (Hemiptera: Pseudococ- evaluations and estimating optimal model complexity for MaxEnt
cidae) in Asia under future climate change scenarios. Clim Change ecological niche models. Methods Ecol Evol 5:1198–1205 (2014).
141:719–732 (2017). 54 R Core Team, R: a language and environment for statistical computing.
30 Kadmon R, Farber O and Danin A, Effect of roadside bias on the R Foundation for Statistical Computing. Available at http://www.
accuracy of predictive maps produced by bioclimatic models. Ecol R-project.org/, Vienna (2015).
Appl 14:401–413 (2004). 55 Bosso L, Febbraro MD, Cristinzio G, Zoina A and Russo D, Shedding
31 Hijmans RJ, Cameron SE, Parra JL, Jones PG and Jarvis A, Very high light on the effects of climate change on the potential distribu-
resolution interpolated climate surfaces for global land areas. Int tion of Xylela fastidiousa in the Mediterranean basin. Biol Invasions
J Climatol 25:195–204 (2005). 18:1759–1768 (2016).
32 Elith J, Phillips SJ, Hastie T, Dudik M, Chee YE and Yates CJ, A statistical 56 Zarzo-Arias A, Penteriani V, Delgado MM, Torre PP, García-González R,
explanation of MaxEnt for ecologists. Divers Distrib 17:43–57 (2011). Mateo-Sánchez MC et al., Identifying potential areas of expansion
33 Raghavan RK, Barker SC, Cobos ME, Barker D, Teo EJM, Foley DH et al., for the endangered brown bear (Ursus arctos) population in the
1662

Potential spatial distribution of the newly introduced long-horned Cantabrian Mountains (NW Spain). PLoS One 14:e0209972 (2019).

wileyonlinelibrary.com/journal/ps © 2019 Society of Chemical Industry Pest Manag Sci 2020; 76: 1652–1663
Potential distribution of Dysmicoccus brevipes and D. neobrevipes www.soci.org

57 Gilfillan D, Joyner TA and Scheuerman P, Maxent estimation of aquatic 73 Kannojiya R, Kumar G, Ravi R, Tiyer NK and Pandey KM, Extraction
Escherichia coli stream impairment. Peer J 6:e5610 (2018). of Pineapple fibers for making commercial products. J Environ Res
58 Lobo JM, Jiménez-Valverde A and Real R, AUC: a misleading measure Develop 7:1385–1390 (2012).
of the performance of predictive distribution models. Glob Ecol 74 Bellard C, Thuiller W, Leroy B, Genovesi P, Bakkenes M and Cour-
Biogeogr 17:145–151 (2008). champ F, Will climate change promote future invasions? Glob Chang
59 Peterson AT, Papes M and Soberón J, Rethinking receiver operating Biol 19:3740–3748 (2014).
characteristic analysis applications in ecological niche modeling. 75 Bertelseier C, Luque GM, Hoffmann BD and Courchamp F, Worldwide
Ecol Model 213:63–72 (2008). ant invasions under climate change. Biodivers Conserv 24:117–128
60 He YB, Zhan RL, Sun GM, Wu JB and Zhao YL, Phylogeography of pink (2014).
pineapple mealybugs Dysmicoccus brevipes (Cockerell) reveals the 76 Pyke CR, Thomas R, Porter RD, Hellmann JJ, Dukes JS, Lodge DM et al.,
history of pineapple introduction and cultivation in China. Genet Current practices and future opportunities for policy on climate
Mol Res 14:9890–9897 (2015). change and invasive species. Conserv Biol 22:585–592 (2008).
61 He YB, Zhan RL, Liu YH, Sun GM and Zhao YL, Effect of ant trap- 77 Lemoine NP, Climate change may alter breeding ground distributions
ping on population of Dysmicoccus brevipes (Cockerell) and occur- of eastern migratory monarchs (Danaus plexippus) via range expan-
rence of mealybug wilt of pineapple. J South Agr 45:1814–1817 sion of Asclepias host plants. PLoS One 10:e01186114 (2015).
(2014). 78 Banks NC, Paini DR, Bayliss KL and Hodda M, The role of global trade
62 Bertin A, Lerin S, Botton M and Parra J, Temperature thresholds and and transport network topology in the human-mediated dispersal
thermal requirements for development and survival of Dysmicoccus of alien species. Ecol Letters 18:188–199 (2015).
brevipes (Hemiptera: Pseudococcidae) on Table Grapes. Neotrop 79 Phillips BL, Chipperfield JD and Kearney MR, The toad ahead: chal-
Entomol 48:71–77 (2019). lenges of modelling the range and spread of an invasive species.
63 Hu ZY, Shao WD, He YJ, Zhang JD and Xu ZH, Effects of temperature Wildlife Res 35:222–234 (2008).
on the growth, development and reproduction of Dysmicoccus 80 Gallien L, Münkemüller T, Albert CH, Boulangeat I and Thuiller W,
neobrevipes Beardsley (Hemiptera: Pseudococcidae). Chin J Applied Predicting potential distributions of invasive species: where to go
Ecol 28:651–657 (2017). from here? Divers Distrib 16:331–342 (2010).
64 Chen ZT, Zhang XD, Zhang N, Yan Z and Ren MY, Life table of 81 Svenning JC, Gravel D, Holt RD, Schurr FM, Thuiller W, Münkemüller T
the laboratory population of Dysmicoccus neobrevipes (Bardsley) at et al., The influence of interspecific interactions on species range
different temperature. Chin J Topi Crops 31:464–468 (2010). expansion rates. Ecography 37:1198–1209 (2014).
65 Biber-Freudenberger L, Ziemacki J, Tonnang HEZ and Borgemeister B, 82 Martins GDSO, Zago HB, Costa AV, Junior LMDA and Carvalho JRD,
Future risks of pest species under changing climatic conditions. PLoS Chemical composition and toxicity of citrus essential oils on
ONE 11:e0153237 (2016). Dysmicoccus brevies (Hemiptera: Pseudococcidae). Rev Caatinga
66 Warren DL, In denfense of “niche modeling”. Trends Ecol Evol 30:811–817 (2017).
27:497–500 (2012). 83 Ferreira KDS, Ferreira TF, Souza RM and Dolinski C, Potential of
67 Kabi S, Karungi J, Sigsgaard L and Ssebuliba JM, Dysmicoccus brevipes entomopathogenic nematodes (Rhabditida) for control of
(Cockerell) occurrence and infestation behavior as influenced by pink pineapple mealybug adult females, Dysmicoccus brevipes
farm type, cropping systems and soil management practices. Agr (Hemiptera: Pseudococcidae), under laboratory conditions. Nema-
Ecosyst Environ 222:23–29 (2016). toda 2:e092015 (2015).
68 Pluess T, Cannon R, Jarošík V, Pergl J, Pyšek P and Bacher S, When are 84 Culik MP and Ventura JA, New species of Rhinoleuco phenga, a
eradication campaigns successful? A test of common assumptions. potential predator of pineapple mealybugs. Pesqui Agropecu Bras
Biol Invasions 14:1365–1378 (2012). 44:417–420 (2009).
69 Lodge DM, Williams S, Macisaac HJ, Hayes KR, Leung B, Reichard S 85 Win AT, Kinoshita T and Tsuji K, The presence of an alternative food
et al., Biological invasions: recommendations for US policy and source changes the tending behavior of the big-headed ant, Phei-
management. Ecol Appl 16:2035–2054 (2006). dole megacephala (Hymenoptera: Formicidae) on Dysmicoccus bre-
70 Crall AW, Jarnevich CS, Young NE, Panke BJ, Renz M and Stohigren vipes (Homoptera: Pseudococcidae). Appl Entomol Zool 53:253–258
TJ, Citizen science contributes to our knowledge of invasive plant (2018).
species distributions. Biol Invasions 17:2415–2427 (2015). 86 Jahn GC, Beardsley JW and Gonzálz-hernandez H, A review of the asso-
71 Gaertner M, Larson BMH, Irlich UM, Holmes PM, Stafford L, van Wilgen ciation of ants with mealybug wilt disease of pineapple. P Hawaiian
BW et al., Managing invasive species in cities: A framework from Entomol Soc 36:9–28 (2003).
Cape Town, South Africa. Landsc Urban Plan 151:1–9 (2016). 87 Carabali-Banguero DJ, Wyckhuys KAG, Montoya-Lerma J, Kondo T
72 Vall-llosera M, Woolnough AP, Anderson D and Cassey P, Improved and Lundgren JG, Do additional sugar sources affect the degree
surveillance for early detection of a potential invasive species: the of attendance of Dysmicoccus brevipes by the fire ant Solenopsis
alien Rose-ringed parakeet Psittacula krameria in Australia. Biol geminate? Entomol Exp Appl 148:65–73 (2013).
Invasions 19:1273–1284 (2017).

1663

Pest Manag Sci 2020; 76: 1652–1663 © 2019 Society of Chemical Industry wileyonlinelibrary.com/journal/ps

You might also like