Ecological Niche Modeling for the Prediction of Suitable Habitat for Chrysodeixis chalcites (Noctuidae) in the Contiguous United States

doi:10.21203/rs.3.rs-3239726/v1

Download PDF

Research Article

Ecological Niche Modeling for the Prediction of Suitable Habitat for Chrysodeixis chalcites (Noctuidae) in the Contiguous United States

https://doi.org/10.21203/rs.3.rs-3239726/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The golden twin-spot moth, Chrysodeixis chalcites Esper (Lepidoptera: Noctuidae), is a polyphagous, polyvoltine crop pest occurring natively from northern Europe to Mediterranean Africa and the Canary Islands. Larvae feed on a wide variety of naturally occurring plants as well as soybean and other legume crops, short staple cotton, tomato, potato, peppers, tobacco, and banana. Chrysodeixis chalcites has been recorded in agricultural lands in the Ontario peninsula in eastern Canada and in northern counties of Indiana, USA. Given the strong potential for C. chalcites to invade USA crop lands, it is important to identify habitats most likely to sustain growing populations of this pest. Using occurrence data from its homerange, and environmental predictors including bioclimatic conditions, elevation, and human disturbance, we trained three ecological niche models, and used these models to estimate an ensemble prediction of habitat suitability in the contiguous US. Because human impact is potentially a confounding predictor, models were trained both with and without it. High habitat suitability was projected for the Atlantic coast from New England to Florida, the Gulf coast, the lower Midwest, and the Pacific coast and Central Valley of California. Though model predictions were robust, we recommend caution in their interpretation. First, agricultural lands are bioclimatically altered landscapes, and these alterations not reflected in bioclimatic data gathered from weather stations. Second, though the inclusion of human impact did not alter predictions on a large scale, it produced predictions favoring major metropolitan areas as suitable habitat, which we interpret as an artifact.

Ecological Modeling

Chrysodeixis chalcites

ecological niche modeling

invasive species

pest

tomato looper

Pest and pathogen management represents an enormous cost to global agriculture. The United States exports more agricultural products than any other country, accounting for US$ 177 billion in 2021 (USDA-FAS 2022). Pesticide alone costs US farmers up to $15 billion per year (Fernandez-Cornejo et al. 2014), a figure that fails to capture the cost of other methods of pest damage mitigation, such as tillage, pest/pathogen monitoring, post-harvest storage, and development of pest and pathogen resistant crops. Non-native species are often transported into habitats lacking the predators, pathogens, and competitors of their native range, leading to biological invasions that exacerbate threats to the environment, human health, and agriculture (Clark et al. 1997; Mack et al. 2000; Albins and Hixon 2013; Castorani and Hovel 2015). Invasive crop pests and pathogens in the US cause up to $40 billion in damage to crops and timber resources yearly (Paini et al. 2016).

The golden twin-spot moth, Chrysodeixis chalcites Esper (Lepidoptera: Noctuidae), is a polyphagous, polyvoltine crop pest occurring natively from northern Europe to Mediterranean Africa (CABI 2022) and the Canary Islands (del Pino et al. 2011) but has become an invasive pest in southern Africa (Vermeulen and Catling 1980). Larvae feed from 16 different plant families with the most economically important crop hosts including bananas, soy and other legume crops, short staple cotton, tomato, potato, peppers, and tobacco. (Sullivan and Molet 2007; del Pino et al. 2011; Roméo et al 2015; Nouri-Ganbalani et al. 2015; Alami et al. 2014; Cakmak et al. 2019; CABI 2022). First instar larvae of Chrysodeixis chalcites graze from the lower epidermis of host leaves, creating a series of translucent patches in the leaf (Goody 1991). Later instar larvae tie leaves with silk and feed from within this shelter. They consume holes completely through the leaves between the leaf veins (Rashid et al 1971). There are also reports of larval damage to fruits including tomatoes (Romeo et al. 2015) and crop legumes (Sullivan and Jones 2007). Chrysodeixis chalcites is the pest of highest concern for Canary Island banana farmers resulting in yield loss due to both leaf and fruit damage (del Pino et al. 2011; Polaszek et al. 2012; Fuentes et al. 2018).

Female adults lay eggs in masses varying in number from 14–189 (Harakly and Farag 1975) with females able to lay as many as 630 eggs in a lifetime (Goodey 1991). Rate of development of all life stages, egg, larvae, pupae, and adults depends strongly on temperature and relative humidity, with warmer, wetter conditions promoting faster growth (Rashid et al. 1971; Harakly and Farag 1975; del Pino et al. 2020). The number of generations that occur each year is therefore strongly dependent on temperature with as many as nine generations occurring per year in Egypt (Harakly and Farag 1975), while populations in northern Europe may develop slowly enough to be univoltine (del Pino et al. 2020). Chrysodeixis chalcites cannot reproduce at temperatures above 35°C. Further, eggs do not enter diapause, and so it has been surmised that C. chalcites may not overwinter in sites with temperatures dropping below 15°C for long periods (del Pino et al. 2020). Though we have found no study indicating the flight range of adults, there is evidence that on at least one occasion an adult female was able to cross the English Channel (33km wide at its narrowest point) to lay eggs in southeast England (Sparks et al 2007). Migratory noctuid moths have been shown capable of long-distance dispersal by flying in airstreams that exceed self-powered flight speed (Alerstam et al. 2011).

Several aspects of the lifecycle of Chrysodeixis chalcites contribute to its potential invasiveness. 1) As a generalist, C. chalcites can make use of many different host plants allowing it to thrive in natural and anthropogenic habitats alike, 2) high fecundity and the potential for multiple generations per year increase the potential for population growth and dispersal, 3) C. chalcites can migrate long distances by human transportation of fruit, and 4) C. chalcites is likely able to fly long distances. In 2008, C. chalcites was discovered in southern Ontario, feeding on tomatoes and legumes less than 50km from the US border (Murillo et al. 2013). Because C. chalcites exhibits seasonal migration in Europe, it is possible C. chalcites populations in Canada overwinter in the United States (Murillo et al. 2013). This is supported by survey efforts that report C. chalcites having been found in northern counties of Indiana in 2021 without evidence of population establishment (CERIS 2022).

In this study, we developed ecological niche models (ENMs) for Chrysodeixis chalcites. ENMs quantify current distributions of species and derive response functions that maximize correlation between species occurrence and environmental data, such as temperature, precipitation, seasonality, or anthropogenic conditions (Pearson et al. 2002; Elith et al. 2006; Pearson 2007; Kozak et al. 2008). These models can then be used to predict the most probable distribution of a species on a different landscape, or under different environmental conditions. Additionally, with a suitable selection of environmental conditions, ENMs are useful in identifying aspects of the environment most important for determining habitat suitability. In this study, we developed three ENMs incorporating a range of statistical and philosophical considerations. Generalized Additive Models (GAM) fit multiple limited functions to response variables using splines, producing functions that are less prone to extrapolation error than General Linear Models (Guisan et al. 2002). Maximum Entropy (ME) models are used to select the least informative predictive distributions while maximizing fit to the data, compatible with both the Bayesian statistical paradigm and machine learning methods (Phillips et al. 2006). Boosted Regression Tree (BRT) models are constructed by making use of regression trees and boosting, a machine-learning algorithm that tracks the explanatory power of populations of regression trees (Elith et al. 2006, 2008). We predicted habitat suitability for Chrysodeixis chalcites in the contiguous US, as well as environmental conditions driving that potential distribution.

Unless otherwise indicated, all analyses took place in the R coding language (R Core Team 2021).

Collection and Sanitization of Occurrence Data

Occurrence records for Chrysodeixis chalcites were obtained through the Global Biodiversity Information Facility (GBIF 2022). All occurrence records in the dataset without latitude or longitude were removed. Chrysodeixis chalcites most commonly occurs in Europe, the Middle East, and North Africa (Sullivan and Molet 2007, CABI 2022), and so models were trained to occurrence points in this range. Occurrence points in sub-Saharan Africa represent invasive populations that may not have fully occupied the entire landscape available to them violating the assumption of species-environment equilibrium (Guisan and Thuiller 2005), and so were not included. Occurrence points from Asia and the Pacific may be C. eriosoma, sometimes considered to be a separate species from C. chalcites (Murillo et al. 2013), and so these were also excluded from the analysis. Removal outside of the home range left 4312 occurrence points. Duplicate points were then removed, resulting in 2156 records. After this, we stratified the data set by removing all points with duplicate latitudes and longitudes when rounded to a tenth of a degree. This was done to mitigate the effects of collector bias (Phillips et al. 2009). After this cleaning process, the dataset contained 679 occurrence records. 20% of these records (137) were randomly removed from the data set and set aside to assess model accuracy. The remaining 80% (542) were used to train the models (Fig. 1)

Generation of Background/Absence Data

All models used in this study require background/absence points. To mitigate the effects of sampling bias inherent in occurrence data, background data were sampled from within the range of presence data (Phillips et al. 2009). First, k-means spatial clusters were identified (k = 1:15), and the smallest number of clusters was chosen that simultaneously minimized the sum of squared average geographical distance. Circles were generated around each occurrence point in the training data set, with the radius being equal to the average distance among occurrence points within each cluster. Circles were merged into a buffer representing the range of occurrence points, and absence and background points were drawn randomly from within this buffer.

Predictor Variables

Thirty-five environmental variables were obtained from the CliMond Bioclim dataset (Kriticos et al. 2012). GTOPO30 elevation data were also included (USGS-EROS 2018) along with a layer summarizing human impact (HI), as a composite of nine global data layers covering human population pressure (population density), human land use and infrastructure (built-up areas, nighttime lights, land use/land cover), and human access in the form of coastlines, roads, railroads, navigable rivers (Wildlife Conservation Society, 2005). The use of such a human impact layer may be confounding. On one hand, many invasive pests are well adapted to disturbance. However, such a pattern could also be generated by collection practices that are biased toward locations close to human-made structures such as roads and settlements (Kadmon et al. 2004; Moerman and Estabrook 2006; Williams and Lutterschmidt 2006, Phillips et al. 2009; Daru et al. 2018). Therefore, models were trained both with and without HI predictor data.

To avoid model inaccuracy due to collinearity of predictor variables (De Marco et al. 2018) contribution of each variable to overall collinearity was estimated as the Variance Inflation Factor (VIF) (Guisan et al. 2002, 2017). A stepwise algorithm for reducing collinearity was performed by first estimating VIF for all predictor values associated with occurrences, removing the layer contributing most to collinearity, and recalculating VIF, until remaining layers had VIF values less than 10 (Chatterjee and Yilmaz 1992); this was performed using the vifstep function in the ‘usdm’ package (Naimi et al. 2014).

For GAM, the relative effect of each predictor was measured as a z-score, or the number of standard deviations from the mean under the null expectation of no-effect. On the other hand, predictor importance for the machine learning models ME and BRT was measured by relative contribution. These evaluations are clearly meant to establish relative importance of predictors within rather than among trained models. However, for uniformity of presentation, z-scores for the GAM model were converted to relative contribution scores by calculating their contribution to overall variance.

Ecological Niche Models

We employed three models that differed markedly in underlying statistical properties, the Generalized Additive Model (GAM), Maximum Entropy (ME), and Boosted Regression Trees (BRT). The Generalized Additive Model (GAM) was trained using the ‘mgcv’ package (Wood 2022). Maximum Entropy models (ME) were constructed using MaxEnt software (Phillips et al. 2006, 2023), run through the ‘dismo’ package (Hijmans et al. 2023). Booted Regression Tree (BRT) models were constructed using the ‘dismo’ package as well using tree complexity of 5 with a learning rate of 0.001. To train the Generalized Additive Model (GAM), we sampled absence points equal to the number of occurrences used to train the model. For Maximum Entropy (ME) and Boosted Regression Tree (BRT) models, we used 10,000 background/pseudoabsence points. Models were then used to predict probability of presence, called ‘habitat suitability’ here, for the contiguous US.

Model Evaluation and Ensemble Prediction

Accuracy of the models in predicting withheld occurrence data was evaluated using three metrics: Area Under the receiver operating characteristic Curve (AUC), Pearson’s correlation coefficient (COR), and Cohen’s kappa (kappa). Because these three modelling approaches are philosophically and mathematically different, agreement in prediction among them is likely due to correct discrimination between signal and noise. Therefore, we incorporated an ensemble approach (Araujo and New 2007; Stohlgren et al. 2010) to summarize models, by calculating the AUC-weighted average of GAM, ME, and BRT predictions of habitat suitability. User accuracy metrics were also calculated for ensemble predictions.

Environmental variables

When HI was included, the VIF algorithm retained 14 predictors (Table 1). When HI was not included, 13 predictors were retained. For the most part, the same predictors were obtained for both predictor sets. When HI was included, Maximum Temperature in the Warmest Week (bio05) and Radiation in the Coldest Quarter (bio27) were not retained. When HI was not included, Temperature Seasonality (bio04) was not retained.

Table 1

Environmental predictors retained after the VIF algorithm when HI is included and excluded among the predictors.
	With HI	Without HI
Isothermality (bio03)	+	+
Temperature Seasonality (bio04)	+
Maximum Temperature in the Warmest Quarter (bio05)		+
Mean Temperature in the Wettest Quarter (bio08)	+
Annual Precipitation (bio12)	+	+
Precipitation in the Wettest Week (bio13)	+	+
Precipitation in the Driest Week (bio14)	+	+
Precipitation Seasonality (bio15)	+	+
Highest Weekly Radiation (bio21)	+	+
Radiation Seasonality (bio23)	+
Radiation in the Wettest Quarter (bio24)	+	+
Radiation in the Driest Quarter (bio25)	+	+
Radiation in the Coldest Quarter (bio27)		+
Mean Moisture in the Coldest Quarter (bio35)	+	+
Elevation	+	+
Human Impact	+

Model evaluation

When HI was included, the resulting predictor set produced models with marginally higher ability to discriminate true presence from true absence (AUC = 0.866–0.954) than when HI was not included (AUC = 0.805–0.934). This pattern was also true for model calibration (COR) and predictive accuracy (kappa; Table 2).

Table 2

Evaluation metrics Area Under the receiver operating characteristic Curve (AUC), Pearson’s correlation coefficient (COR) and Cohen’s kappa (Kappa) for three models, the Generalized Additive Model (GAM), Maximum Entropy (ME), Boosted Regression Trees (BRT), and their AUC weighted Ensemble.
With HI
	AUC	COR	Kappa
GAM	0.866	0.638	0.583
ME	0.916	0.725	0.693
BRT	0.954	0.669	0.781
Ensemble	0.922	0.725	0.702
Without HI
	AUC	COR	Kappa
GAM	0.805	0.525	0.517
ME	0.877	0.660	0.610
BRT	0.934	0.638	0.750
Ensemble	0.887	0.669	0.621

Contribution of environmental variables

When HI was included as a predictor, it was positively correlated with habitat suitability. HI had the greatest contribution for GAM (19.6%) and the second greatest contribution for ME (22.2%) and BRT (14.6%; Table 3). Radiation Seasonality was negatively correlated with habitat suitability and also played an important role in all three models, having the second highest contribution for GAM (13.6%), the highest for BRT (16.2%) and the third highest for ME (18%). Elevation was positively correlated with habitat suitability and was the most important predictor for ME (24.0%) and the third highest for BRT (11.6%).

Table 3

Percent contribution of predictors in training three models trained with predictor sets including human impact.
Predictor	GAM	ME	BRT
Human Impact	19.6	22.2	14.6
Radiation Seasonality	13.6	18.0	16.2
Elevation	0.4	24.0	11.6
Annual Precipitation	8.5	11.5	8.1
Temperature Seasonality	7.6	4.1	10.7
Radiation in the Wettest Quarter	7.5	9.6	3.9
Isothermality	8.4	6.3	4.9
Mean Temperature in the Wettest Quarter	8.4	0.3	4.8
Precipitation in the Driest Week	7.6	0.5	4.2
Mean Moisture in the Coldest Quarter	5.4	0.5	5.1
Highest Weekly Radiation	3.7	1.1	5.4
Radiation in the Driest Quarter	4.3	1.0	3.6
Precipitation in the Wettest Week	4.4	0.7	3.3
Precipitation Seasonality	0.5	0.3	3.5

When HI was not included as a predictor, elevation was positively correlated with habitat suitability and had the highest contribution for ME (45.5%) and BRT (19.0%; Table 4). Radiation in the coldest quarter had the highest contribution for GAM (14.1%), the second highest contribution for BRT (16.3%), and the third highest for ME (13.6%). Moisture in the Coldest Quarter was the second most important contributor for GAM (10.4%) but did not play a strong role for ME or BRT. Similarly, Radiation in the Wettest Quarter was the second strongest contributor to ME (15.7%) but was not a strong predictor of habitat suitability for either GAM or BRT.

Table 4

Percent contribution of predictors in training three models trained with predictor sets *not* including human impact.
Predictor	GAM	ME	BRT
Elevation	6.2	45.5	19.0
Radiation in the Coldest Quarter	14.1	13.6	16.3
Radiation in the Wettest Quarter	9.0	15.7	5.1
Annual Precipitation	9.4	7.2	6.9
Precipitation in the Driest Week	8.4	5.5	6.0
Mean Moisture in the Coldest Quarter	10.4	1.4	6.1
Mean Temperature in the Wettest Quarter	9.3	0.9	6.9
Precipitation in the Wettest Week	9.5	1.3	4.5
Isothermality	6.5	1.7	7.0
Radiation in the Driest Quarter	6.1	1.3	6.8
Highest Weekly Radiation	1.5	5.2	5.7
Maximum Temperature in the Warmest Quarter	6.4	0.7	4.7
Precipitation Seasonality	3.4	0.0	5.1

Predicted Distribution of Chrysodeixis chalcites

Regardless of whether HI was included as a predictor, ensemble predictions were as accurate as at least two of the three models evaluated by AUC, COR, or kappa (Table 2), and will be the focus of interpretation (Fig. 2). Predictions for individual models are available as supplementary data (Supplement Figs. 1 and 2). Chrysodeixis chalcites is predicted to find suitable habitat throughout the eastern half of the contiguous USA and along portions of the west coast, with areas of high habitat suitability along the Atlantic and Gulf coasts and through the lower Great Lakes and midwestern states—Ohio, Indiana, Michigan, and Illinois (Fig. 2). Nearly all of Florida is predicted to be highly suitable. Both sets of models indicate moderate to high suitability for the Ontario peninsula, where C. chalcites is known to occur (Murillo et al. 2013). Predictions from maps trained with or without HI also predict moderate suitability in the Central Valley of California, especially in the Sacramento River Drainage in the north.

Though predictions were similar for models trained with and without Human Impact, there were also some differences. Human Impact was a strong enough contributor that predictions from models trained with HI are somewhat granular, with areas of high habitat suitability clustered in US metropolitan areas; Pittsburgh, Cleveland, Columbus, Cincinnati, Louisville, Indianapolis, Chicago, Saint Louis, Kansas City, Des Moines, and Minneapolis are all clearly visible (Fig. 2a) as high habitat suitability areas. Additionally, models trained with HI predict a more limited distribution of habitat suitability in New England, eastern Ontario and Quebec and in the Pacific Northwest, than models trained without HI.

Though models trained to predictor sets including and excluding HI differed somewhat in the heterogeneity of suitable habitat, both agreed that suitable habitat for Chrysodeixis chalcites occurs along the mid- to south Atlantic coasts, in the Great Lakes and lower Midwest, along the Gulf Coast, and along the Pacific coast into the central valley of California. Under either analysis, Florida is predicted to be highly suitable in its entirety. Florida ranks second among the states in tomato and bell pepper production, after only California, and 13th in potato production (USDA-ERS 2021). Similarly, North Carolina and Virginia constitute highly suitable habitat, and rank first and fourth respectively in tobacco production (USDA-ERS 2021). Perhaps more alarming, five of the top-ranking states in soybean production constitute highly suitable habitat in the lower Midwest: Illinois (1st ), Iowa (2nd ), Indiana (4th ), Missouri (6th ), and Ohio (7th ). California’s central valley leads the US states in production of many vegetable crops, with tomatoes and peppers among them (USDA-ERS 2021). Clearly, crops susceptible to damage from C. chalcites grow in areas of the US predicted to be suitable for populations of C. chalcites to flourish.

Though spread of crop pests is most often the result of accidental transportation by people (Crespo-Pérez et al 2011; Bebber et al 2014; Cordeiro et al. 2019), natural hosts may provide a home for source populations in regions of high habitat suitability. Once Chrysodeixis chalcites has established populations on natural hosts, these populations may provide a source for continuing dispersal into agricultural habitats. Chrysodeixis calcites is migratory in its home range and if migratory behavior is maintained in new ranges, it could increase the chances of finding new suitable habitats and hosts (Sullivan and Molet 2007). Natural hosts for C. chalcites include Echium vulgare (viper's-bugloss), Marrubium vulgare (horehound), and Urtica dioica (stinging nettle) (Sullivan and Molet 2007). These three species occur broadly through the USA, though much more densely along the mid-Atlantic coast and lower Midwest, with Marrubium vulgare and Urtica dioica also commonly occurring throughout California into the Pacific Northwest (GBIF 2022). Through much of the suitable range, therefore, C. chalcites should be able to find suitable natural hosts, even when agricultural hosts are unavailable, or are inaccessible due to pesticide use.

Interpretation of habitat suitability predictions must account for bioclimatic differences in microhabitat between agricultural and natural landscapes. Bioclimatic data used here are the result of interpolation of measurements taken at weather stations (Kriticos et al. 2012; Fick and Hijmans 2017). As such, they do not necessarily reflect conditions of cultivated agricultural land. For example, irrigation alters soil moisture and replaces water normally obtained through precipitation. Tree clearing likely alters available radiation and metrics associated with temperature. These common agricultural practices are not reflected in available climate data, but could alter habitat suitability for Chrysodeixis chalcites. Pesticide use may also alter habitat suitability in ways not captured by ecological niche models. C. chalcites infestations have shown to be mitigated by the use of pesticides like indoxacarb or Bt crops (Fuentes et al. 2018). Despite high habitat suitability predicted by our models, agricultural lands that are dominated by pesticide use may provide poor habitat for C. chalcites. Further, highly modified environments such as greenhouses may provide excellent resources for C. chalcites establishment far away from suitable habitat predicted by these models.

Biological invasions are often correlated with anthropogenic disturbance in which the invader is either well-adapted to anthropogenically altered ecosystems, dispersed by humans, or both (González-Moreno et al. 2015; Paudel and Battaglia 2015; Jauni and Ramuli 2017). Human disturbance may also create an advantage for pests by eliminating competitors from habitats available for dispersal (Bauer 2012). Such an ecological correlation would argue for the use of human impact as a predictor variable as we have done here. On the other hand, occurrence data are inherently biased by collection practices that favor habitats close to roadways, human habitation, and collection facilities (Kadmon et al. 2004; Moerman and Estabrook 2006; Williams and Lutterschmidt 2006, Phillips et al. 2009; Daru et al. 2018). Occurrences with this sampling bias will certainly favor models including human disturbance as a factor, regardless of any biological reality. Without controlled experiments to measure the effect of human disturbance on population fitness, predictors like human impact are best used with caution.

Acknowledgments

We are grateful to Steven Passoa (USDA/APHIS/PPQ) for guidance on the taxonomy and biology of Chrysodeixis chalcites. Funding for this work came from the United States Department of Agriculture Farmbill 7221-1a awards AP20PPQS&T00C153 and AP21PPQS & T00C019 to JKW and CPR

Funding:

This work was supported by USDA-APHIS Farmbill 7221-1a awards AP20PPQS&T00C153 and AP21PPQS & T00C019 to JKW and CPR

Competing Interests:

The authors have no relevant financial or non-financial interests to disclose

Author contributions :

Nicholas A. Galle, Justin K. Williams and Christopher P. Randle designed the study; analysis was carried out by Nicholas A. Galle and Christopher P. Randle; All authors participated in manuscript preparation.

Data availability

Supplementary figures and raw data are available at (insert Dyad URL on acceptance)

Code availability

Scripts are available at (https://github.com/randle-cp/Chrysodeixis-chalcites.git) and all scripts and data have been archived at

Conflict of interest The authors have no conflicts of interest to declare that are relevant to the content of this article.

Ethical approval Research did not involve experiments or invasive sampling of vertebrate animals.

Consent to participate Research did not involve human participants.

Consent for publication All authors approved the final version of the manuscript.

Alami S, Naseri B, Golizadeh A, Ramjou J (2014) Age-stage, two-sex life table of the tomato looper, Chrysodeixis chalcites (Lepidoptera: Noctuidae), on different bean cultivars. Arthopod-Plant Interact 8:475–484. https://doi.org/10.1007/s11829-014-9330-3
Albins MA, Hixon MA (2013) Worst case scenario: potential long-term effects of invasive predatory lionfish (Pterois voltans) on Atlantic and Caribbean coral-reef communities. Environ Biol Fishes 96:1151–1157. https://doi.org/10.1007/s10641-011-9795-1
Alerstam T, Chapman JW, Bäckman J, Smith AD, Karlsson H, Nilsson C, Reynolds DR, Klaassen RHG, Hill JK (2011) Convergent patterns of long-distance nocturnal migration in noctuid moths and passerine birds. Proc Royal Society, Ser B: Biol Sci 278(1721):3074–3080. https://doi.org/10.1098/rspb.2011.0058
Araujo MB, New M (2007) Ensemble forecasting of species distributions. Trends Ecol Evol 22(1):42–46. https://doi.org/10.1016/j.tree.2006.09.010
Bauer J (2012) Invasive species: “back-seat drivers” of ecosystem change. Biol Invasions 14:1295–1304. https://doi.org/10.1007/s10530-011-0165-x
Bebber DP, Holmes T, Garr SJ (2014) The global spread of crop pests and pathogens. Glob Ecol Biogeogr 23:1398–1407. https://doi.org/10.1111/geb.12214
CABI (2022) Center for Agriculture and Bioscience International Compendium of Invasive Species. https://www.cabi.org/isc/datasheet/13243#todistribution. Accessed 14 December 2022.
Cakmak T, Diaz AP-B, Suarez EH, Acosta CA (2019) Chrysodeixis chalcites (Easper)(Lepidoptera: Noctuidae) oviposition preferences on different growing stages of banana. Phytoparasitica 47:485–498. https://doi.org/10.1007/s12600-019-00755-w
Castorani MC, Hovel KA (2015) Invasive prey indirectly increase predation on their native competitors. Ecol 97(7):1741–2033. https://doi.org/10.1890/14-1538.1
CERIS (2022) Survey Status of Golden twin spot moth - Chrysodeixis chalcites. Center for Environmental and Research Information Systems. Purdue University. https://pesttracker.org/map/ITBCQGA/2021. Accessed 16 December 2022.
Chatterjee S, Yilmaz M (1992) A review of regression diagnostics for behavioral research. Appl Psychol Measurement 16(3):209–227. https://doi.org/10.1177/014662169201600301
Clark JM, Heath DD, Stankiewicz M (1997) The ectoparasites of brushtail possum Trichosurus vulpecula in New Zealand. New Zealand J Zool 24:199–204. https://doi.org/10.1080/03014223.1997.9518114
Cordeiro EM, Campbell JF, Phillips, E. Akhunov (2019) Isolation by distance, source-sink population dynamics and dispersal facilitation by trade routes: Impact on population genetic structure of a stored grain pest. Genes Genomes Genetics 9:1457–1468. https://doi.org/10.1534/g3.118.200892
Crespo-Pérez V, Rebaudo F, Silvain J-F, Dangles O (2011) Modeling invasive species spread in complex landscapes: the case of the potato moth in Ecuador. Landsc Ecol 26:1447–1461. https://doi.org/10.1007/s10980-011-9649-4
Daru BH, Park DS, Primack RB, Willis, CG, Barrington DS, Whitfield TJS, Seidler TG, Sweeney PW, Foster DR, Ellison AM, Davis CC (2018) Widespread sampling biases in herbaria revealed from large-scale digitization. New Phytol 217:939–955. https://doi.org/10.1111/nph.14855
De Marco Jr. P, Nóbrega CC (2018) Evaluating collinearity effects on species distribution models: An approach based on virtual species simulation. PLoS ONE 13(9):e0202403. https://doi.org/10.1371/journal.pone.0202403del.
del Pino M, Cabello T, Hernández-Suárez E (2020) Age-stage, two-sex life table of Chrysodeixis chalcites (Lepidoptera: Noctuidae) at constant temperatures on semi-synthetic diet. Environ Entomol 49(4):777–788. https://doi.org/10.1093/ee/nvaa050
del Pino M, Carnero A, Cabello T, Hernández E (2011) La lagarta o bicho camello, Chrysodeixis chalcites (Esper, 1789), una plaga emergente en los cultivos de platanera de Canarias. Phytoma Esp 225:21–24.
Elith J, Graham CH, Anderson RP, Dudík M, Ferrier S, Guisan A, Hijmans RA, Huettmann F, Leathwick JR, Lehmann A, Li J, Lohmann LG, Loiselle BA, Manion G, Moritz C, Nakamura M, Nakazawa Y, Overton JMM, Peterson AT, Phillipis SJ, Richardson KA, Scachetti-Pereira R, Schapire R, Soberón J, Williams S, Wisz MS, Zimmerman NE (2006) Novel methods improve prediction of species' distributions from occurrence data. Ecography 29:129–151. https://doi.org/10.1111/j.2006.0906-7590.04596.x
Elith J, Leathwick JR, Hastie T (2008) A working guide to boosted regression trees. J Anim Ecol 77:802–813. https://doi.org/10.1111/j.1365-2656.2008.01390.x
Fernandez-Cornejo J, Nehring R, Osteen C, Wechsler S, Martin A, Vialou A (2014) Pesticide Use in U.S. Agriculture: 21 Selected Crops, 1960-2008, EIB-124, U.S. Department of Agriculture, Economic Research Service. https://www.ers.usda.gov/webdocs/publications/43854/46734_eib124.pdf. Accessed 30 July 2023
Fick SE, Hijmans RJ (2017) WorldClim 2: new 1km spatial resolution climate surfaces for global land areas. Int J Climatol 37(12):4302–4315. https://doi.org/10.1002/joc.5086
Fuentes EG, Hernández-Suárez E, Simone O, Williams T, Caballero P (2018) Chrysodeixis chalcites, a pest of banana crops on the Canary Islands: Incidence, economic losses, and current control measures. Crop Prot 108:137–145. https://doi.org/10.1016/j.cropro.2018.02.020
GBIF (2022) Global Biodiversity Information Facility. https://www.gbif.org. Accessed 15 December 2022.
González-Moreno P, Diez JP, Richardson DM, Vilà M (2015). Beyond climate: disturbance niche shifts in invasive species. Glob Ecol Biogeogr 24:360–370. https://doi.org/10.1111/geb.12271
Goodey B (1991) Chrysodeixis chalcites—observations on the life-cycle in captivity. Entomol Rec J Var 103:111–118.
Guisan A, Edwards J, Thomas C, Hastie, T (2002) Generalized linear and generalized additive models in studies of species distributions: setting the scene. Ecol Model, 157:89–100. https://doi.org/10.1016/S0304-3800(02)00204-1
Guisan A, Thuiller W (2005) Predicting species distribution: offering more than simple habitat models. Ecol Lett 8:993–1009. https://doi.org/10.1111/j.1461-0248.2005.00792.x
Guisan A, Thuiller W, Zimmerman NE (2017) Habitat suitability and distribution models. Cambridge Press, Cambridge
Harakly FA, Farag SS (1975). Biological studies on the tomato looper Chrysodeixis chalcytes (Esper) in Egypt. Bull Soc Entomol Egypte 59:295–299
Hijmans RJ, Phillips SJ, Leathwick J, Elith J (2023). Dismo: Species distribution modeling. https://CRAN.R-project.org/package=dismo. Accessed 13 June 2023.
Jauni M, Ramula S (2017) Demographic mechanisms of disturbance and plant diversity promoting the establishment of the invasive Lupinus polyphyllus. J Plant Ecol 10(3):510–517. https://doi.org/10.1093/jpe/rtw049
Kadmon R, Farber O, Danin A (2004) Effect of roadside bias on the accuracy of predictive maps. Ecol Appl 14(2):401–413. https://doi.org/10.1890/02-5364
Kozak KH, Graham CH, and Wiens JJ. (2008) Integrating GIS–based environmental data into evolutionary biology. Trends Ecol Evol 23:141–148. https://doi.org/10.1016/j.tree.2008.02.001
Kriticos, DJ, Webber BL., Leriche A, Ota N, Macadam I, Bathols J, Scott JK (2012) CliMond: global high resolution historical and future scenario climate surfaces for bioclimatic modelling. Methods Ecol Evol 3:53–64. https://doi.org/10.1111/j.2041-210X.2011.00134.x
Mack RN, Simberloff D, Lonsdale WM, Evans H, Clout M, Bazzaz FA (2000) Biological invasions: Causes, epidemiology, global consequences and control. Ecol Appl 10(3):689–710. https://doi.org/10.1890/1051-0761(2000)010[0689:BICEGC]2.0.CO;2
Moerman DE, Estabrook GF (2006). The botanist effect: counties with maximal species richness tend to be home to universities and botanists. J Biogeogr 33:1969–1974. https://doi.org/10.1111/j.1365-2699.2006.01549.x
Murillo H, Hunt DWA, VanLaerhoven SL (2013) First records of Chrysodeixis chalcites (Lepioptera: Noctuidae: Plusiinae) for east-central Canada. Can Entomol 145:338–342. https://doi.org/10.4039/tce.2012.106
Naimi B, Hamm NAS, Groen TA, Skidmore AK, Toxopeus AG (2014) Where is positional uncertainty a problem for species distribution modelling? Ecography 37(2):191–203. https://doi.org/10.1111/j.1600-0587.2013.00205.x
Nouri-Ganbalani G, Mardani-Talaee M, Haji-Ramezani MR (2015) Age-stage, two-sex life history of the golden twin spot moth, Chrysodeixis chalcites (Lepidoptera: Noctuidae), on six commercial tomato cultivars. Can Entomol 148:92–101. https://doi.org/10.4039/tce.2015.16
Paini DR, Sheppard AW, Cook DC, De Barro PJ, Wornerf SP, Thomas MB (2016) Global threat to agriculture from invasive species. Proc Natl Acad Sci USA 113(27):7575–7579. https://doi.org/10.1073/pnas.1602205113
Paudel S, Battaglia LL (2015) The role of light, soil, and human factors on the probability of occurrence of an invasive and the native plant species in coastal transitions of coastal Mississippi, USA. J Plant Ecol 8(5):491–500. https://doi.org/10.1093/jpe/rtu045
Pearson RG (2007) Species’ distribution modeling for conservation educators and practitioners. Lessons Conserv 3:54–89.
Pearson RG, Dawson TP, Berry PM, Harrison PA (2002) SPECIES: A spatial evaluation of climate impact on the envelope of species. Ecol Model 154:289–300. https://doi.org/10.1016/S0304-3800(02)00056-X
Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy modeling of species geographic distributions. Ecol Model 90:231–259. https://doi.org/10.1016/j.ecolmodel.2005.03.026
Phillips SJ, Dudík M, Elith J, Graham CH, Lehmann A, Leathwick J, Ferrier S (2009) Sample selection bias and presence-only distribution models: implication for background and pseudo-absence data. Ecol Appl 19(1):181–197. https://doi.org/10.1890/07-2153.1
Phillips SJ, Dudík M, Schapire RE (2023). Maxent software for modeling species niches and distributions. https://biodiversityinformatics.amnh.org/open_source/maxent. Accessed 13 June 2023.
Polaszek, A Rugman-Jones PF, Stouthammer R, Hernández-Suárez E, Cabello T, del Pino Pérez (2012) Molecular and morphological diagnosis of fives species of Trichogramma: biological control agents of Chrysodeixis chalcites (Lepidoptera: Noctuidae) and Tuta absoluta (Lepidoptera: Gelechiidae) in the Canary Islands. BioControl 57:21–35. https://doi.org/10.1007/s10526-011-9361-y
R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Accessed 12 June 2023.
Rashid, FF, Hammad AM, Hassan SM (1971) The biology of Autographa chalcites L. in Alexandria region (Lepidoptera: Noctuidae). Bull Soc Entomol Egypte, 55:419–426.
Roméo HC, Champlain DL, Chantal AD, Stephan EAP (2015) Diversity and agronomic status of tomato and pepper fruit pests in two agro-ecological zones of southern Cameroon: Western Highland and the southern Plateau of Cameroon. Afr J Agric Res 10:1224–1232.
Sparks TH, Dennis RLH, Croxton PJ, Cade M (2007) Increased migration of Lepidoptera linked to climate change. Eur J Entomol 104:139–143.
Stohlgren, TJ, Ma P, Kumar S, Rocca M, Morisette JT, Jarnevich CS, Benson, N (2010). Ensemble habitat mapping of invasive plant species. Risk Analysis 30(2):224-235. https://doi.org/10.1111/j.1539-6924.2009.01343.x
Sullivan M, Jones E (2007). Soybean commodity-based survey guideline. United States Department of Agriculture, Animal and Plant Health Inspection Service, Center for Plant Science Health and Technology. http://download.ceris.purdue.edu/file/392. Accessed 15 December 2022.
Sullivan M, Molet T (2007). CPHST Pest Datasheet for Chrysodeixis chalcites. USDA-APHIS-PPQ-CPHST. http://download.ceris.purdue.edu/file/3027. Accessed 14 December 2022.
USDA-FAS (2022). United States Department of Agriculture, Foreign Agricultural Service. https://www.fas.usda.gov/2021-country-overview. Accessed 14 December 2022.
USDA-ERS (2021). United States Department of Agriculture, Economic Research Service. Cash receipts by commodity state ranking. https://data.ers.usda.gov/reports.aspx?ID=17844#Paef9fc54c0c042988b142a1245abc26b_10_xA. Accessed 3 July 2023.
USGS-EROS (2018). United States Geographical Survey Earth Resources and Observation Science Center. https://www.usgs.gov/centers/eros/science/usgs-eros-archive-digital-elevation-global-30-arc-second-elevation-gtopo30. Accessed 12 June 2023
Vermeulen JB, Catling HD (1980) Establishment of a grid of light traps in South Africa and some preliminary results for the Noctuidae. Phytophylactica 12:31–43.
Wildlife Conservation Society (2005) Last of the wild project, V2, (LWP-2): Global Human Influence Index (HII) dataset (geographic). NASA Socioeconomic Data and Applications Center (SEDAC). https://sedac.ciesin.columbia.edu/data/set/wildareas-v2-human-influence-index-geographic. Accessed 13 June 2022.
Williams JK, Lutterschmidt WI (2006) Species-area relationships indicate large-scale data gaps in herbarium collections. Lundellia 2006(9):41–50. https://doi.org/10.25224/1097-993X-9.1.41
Wood S (2022) Package ‘mgcv’. https://cran.r-project.org/web/packages/mgcv/index.html. Accessed 7 April 2022.

Supplement.docx
Supplement

Download PDF

Version 1

posted

You are reading this latest preprint version

Ecological Niche Modeling for the Prediction of Suitable Habitat for Chrysodeixis chalcites (Noctuidae) in the Contiguous United States

Status:

Version 1

Abstract

Figures

Introduction

Methods

Results

Discussion

Declarations

References

Supplementary Files

Status:

Version 1