Journal of Science: Vol. 13 | Article 102 Zhu et al.

Geographic distribution and niche divergence of two stink- bugs, japonensis and Parastrachia nagaensis

Gengping Zhu1a*, Guoqing Liu2b, Wenjun Bu2c, Jerzy A. Lis3d

1Tianjin Key Laboratory of and Plant Resistance, College of Life Sciences, Tianjin Normal University, Tianjin 300387, China 2Institute of Entomology, College of Life Sciences, Nankai University, Tianjin 300071, China 3Department of Biosystematics, Opole University, Oleska 22, 45-052 Opole, Poland

Abstract is a small stinkbug family containing only one and two species, Parastra- chia japonensis (Scott) (: : ) and Parastrachia nagaensis Distant. The geographic distribution of the genus has been poorly studied. Niche conservatism refers to that idea that closely related species are more ecologically similar than would be ex- pected, whereas niche divergence predicts they occupy distinct niche spaces. The existence of only two species within one genus suggests niche conservatism or differentiation might exist among them. Herein, the distribution of the genus was mapped, potential distributions were pre- dicted using ecological niche modeling, and climate spaces occupied by the two species were identified and compared. Our outlined map supports the general spreading route proposed by Schaefer et al. The potential distributions suggest that the genus’ range could extend beyond its presently known distribution, and further investigation into this area could aid in their conserva- tion, particularly P. nagaensis. The niche space inferred by ecological niche modeling suggests the two species do not occupy identical habitat, but the differences between their models could simply be due to the differential availability of habitat in the different regions that they occupy.

Keywords: ecological niche, ecological niche modeling, potential distribution, principle component analysis Abbreviations: AUC, area under the curve; ENM, ecological niche modeling, PCA, principal component analysis Correspondence: a [email protected], b [email protected], c [email protected], d [email protected], *Corresponding author Editor: Robert Jetton was editor of this paper. Received: 3 February 2012 Accepted: 5 October 2013 Published: 5 October 2013 Copyright: This is an open access paper. We use the Creative Commons Attribution 3.0 license that permits unrestricted use, provided that the paper is properly attributed. ISSN: 1536-2442 | Vol. 13, Number 102 Cite this paper as: Zhu G, Liu G, Bu W, Lis JA. 2013. Geographic distribution and niche divergence of two stinkbugs, and Parastrachia nagaensis. Journal of Insect Science 13:102. Available online: http://www.insectscience.org/13.102

Journal of Insect Science | http://www.insectscience.org 1

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. Introduction Sigwalt and Lis 2008; Lis and Ziaja 2010). However, the geographic distribution of the It is widely accepted that ecology plays an genus throughout the world, and especially in important role in speciation (Rice et al. 2003; China, has been poorly studied (Schaefer et al. Rissler and Apodaca 2007; Pyron and Bur- 1988; Schaefer and Kikuhara 2007). The fact brink 2009). Generally, the speciation event that there are only two species within one ge- takes place in geographic dimensions and may nus suggests niche conservatism or not be accompanied by ecological innovation differentiation might exist among them. (Peterson et al. 1999; Pyron and Burbrink 2009; McCormack et al. 2010; Peterson Principle component analysis (PCA) and eco- 2011). When populations enter into a new en- logical niche modeling (ENM) are two vironment, however, the ecological niche of a approaches that have been widely used to species might diverge to adapt to the novel study niche conservatism and differentiation environment, and subsequent natural selection (Broennimann et al. 2007; Fitzpatrick et al. might promote this process and facilitate spe- 2007; Warren et al. 2008; Rödder and Lötters ciation (Graham et al. 2004; Wiens and 2009; McCormack et al. 2010; Medley 2010; Graham 2005; Sánchez-Fernández et al. Zhu et al. 2012). PCA uses pooled environ- 2011). Ecological niche differences among mental variables associated with species different species can be visualized and ana- occurrence to reveal reduced significant com- lyzed to assess the likely ecological and ponents that account for the observed evolutionary forces that shape the species’ distribution. The defining niche space of re- geographical distributions and habitat prefer- duced dimension allows investigation of niche ences (Graham et al. 2004; Wiens and conservatism and differentiation. ENM seeks Graham 2005; Raxworthy et al. 2007; to characterize environmental conditions that Sánchez-Fernández et al. 2011). are suitable for the species, and then to identi- fy where suitable environments are distributed Parastrachiidae is a small stinkbug family spatially (Pearson 2007). ENM has been wide- containing only one genus and two species, ly used in biological responses to climate Parastrachia japonensis (Scott) (Hemiptera: change, setting conservation priorities, and the Heteroptera: Pentatomoidea) and Parastra- study of evolutionary biology (Guisan and chia nagaensis Distant. The genus has Thuiller 2005; Raxworthy et al. 2007; Rissler attracted the attention of some heteropterists and Apodaca 2007; Martínez-Gordillo et al. due to its unstable position in Pentatomoidea 2010). (i.e., in different families or subfamilies) (Goel and Schaefer 1970; Schaefer et al. In this study, the global distribution of the ge- 1988; Gapud 1991; Hasan and Kitching 1993; nus was mapped with many records, Hasan and Nasreen 1994; Schuh and Slater especially in China, potential distributions 1995), and the maternal care behavior of P. were predicted using ENM, and niche spaces japonensis (Tsukamoto et al. 1994; Filippi et occupied by the two species were identified al. 1995a, b, 2000a, b, 2001; Nomakuchi et al. and compared using PCA and ENM. The eco- 1998, 2001; Hironaka et al. 2003a, b, 2005, logical niche of a species here can be defined 2007). The genus was raised to the family lev- as “a set of environmental conditions under el (Sweet and Schaefer 2002), which was which it is able to maintain populations with- widely accepted (Grazia et al. 2008; Pluot- out immigrational subsidy” (Grinnell 1917,

Journal of Insect Science | http://www.insectscience.org 2

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al.

Table 1. Principal components analysis of 13 environmental vari- climatic variables were chosen. Among the ables associated with occurrences of the genus Parastrachia, variables, 6 (i.e., BIO 1, 5, 6, 12, 13, 14) were eigenvalues for significant variables (> 0.8) are in bold. summarizing aspects of temperature and pre- cipitation and were obtained from WorldClim

(Hijmans et al. 2005), and 3 (i.e., BIO 20, 21, 22) were summarizing aspects of radiation

from CliMond (Kriticos et al. 2011). Topog- raphy variables represented by elevation,

slope, aspect, and compound topographic in- dex were derived from the US Geological

Survey’s HYDRO1k Elevation Derivative Da- tabase (USGS 2001). Variables at a resolution

of 2.5 min were used for model calibration.

Principle component analysis

1924). This study also highlighted the correla- PCA was used to explore significant variables tive approach for biodiversity conservation, that related to the genus’distribution and to

especially for the localized endemic species. compare niche spaces occupied by the 2 spe- cies. After superimposing the occurrence data

Materials and Methods on bioclimate and topography grids, values for each record were extracted in ArcGIS 10. A

Occurrence data correlation matrix of 40 × 13 was prepared for Occurrence records were assembled from ex- the PCA performed in SPSS 19 (IBM 2009).

isting literature (Tsukamoto and Tojo 1992; To facilitate data visualization, the occurrence Hironaka et al. 2003a; Schaefer and Kikuhara records were pooled for P. nagaensis and P.

2007; Hosokawa et al. 2010) and our speci- japonensis respectively. mens (see Supplementary Figure 1). Localities

were georeferenced in Google Maps Ecological niche modeling (http://maps.google.com), Gazetteer of China A wide range of methods have been explored

(1997), or BioGeomancer for ENM. Among them, the maximum entro- (http://bg.berkeley.edu/latest/), then mapped py algorithm implemented in the Maxent

in ArcGIS 10 (ESRI 2006). A total of 34 and software (Phillips et al. 2004, 2006; Elith et al. 6 occurrence records were prepared for P. ja- 2011) generally performs better than other

ponensis and P. nagaensis respectively. algorithms (Elith et al. 2006; Phillips et al. 2006; Ortega-Huerta and Peterson 2008).

Environmental variables Maximum entropy is a machine-learning Environmental variables were selected by technique that predicts species distributions by

considering climate and topography factors using detailed environmental variables associ- that might affect the genus’ distribution (Tsu- ated with species occurrence. It follows the

kamoto et al. 1994; Filippi et al. 1995a, b, principle of maximum entropy and spreads 2000a, b, 2001; Nomakuchi et al. 1998, 2001; out probability as uniformly as possible, but

Hironaka et al. 2003a, b, 2005, 2007) (Table subject to the caveat that they must match 1). Annual trends and extreme or limiting bio- empirical information such as known presence

Journal of Insect Science | http://www.insectscience.org 3

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. (Phillips et al. 2004, 2006). Maxent is less was set as the minimum square area that cov- sensitive to sample size (Wisz et al. 2007) and ered all the occurrences of P. japonensis. The can be applied to sample sizes as small as five omission rate at the 10th percentile training (Pearson et al. 2007). The default convergence presence threshold was also adopted for mod- threshold (10-5), maximum number of itera- el evaluation; the 10th percentile threshold is tions (500), and the logistic output with highly conservative in estimating a species suitability values ranging from 0 (unsuitable tolerance for each predictor, which has been habitat) to 1 (optimal habitat) were adopted. A more commonly used (Liu et al. 2011). Half jack-knife procedure was used to evaluate the of the records were used for model calibra- relative importance of each predictor variable tion, and the other half were used for AUC and the ability to correctly predict new occur- and omission rate test. In the end, the overall rences in the model (Pearson et al. 2007). occurrence was used to calibrate the model for exhibiting P. japonensis. To visualize niche in ecological dimensions, environmental grids (including annual mean For P. nagaensis, the 6 occurrence records temperature and annual precipitation) and the were inappropriate for a typical model evalua- final predictions were extracted from a mask tion approach involving partitioning the data built using minimum convex polygon for the into training and testing subsets. A modified genus in Hawth’s Tools (Beyer 2004) and jackknife approach specifically designed for then combined for each species in ArcGIS 10. small sample size was used (Pearson et al. The associating attribute tables were imported 2007). In this method, independent Maxent into SPSS 19, where data density was reduced models were generated iteratively, excluding by selecting a random 10% of the records. one locality in each turn. The lowest suitabil- These reduced tables were then exported to ity score of a presence point, or lowest Microsoft Excel, where scatter plots were pre- presence threshold, for each model was then pared for visualization (e.g., Kambhampati used to determine areas of predicted presence. and Peterson 2007; Donalisio and Peterson The proportion of the training area predicted 2011). as present and the failure or success of the model to predict jackknifed points were then Model evaluation used to calculate the probability of the ob- For P. japonensis, the area under the curve served degree of coincidence between (AUC) of the receiver operating characteristic independent test data and predicted areas of plot and omission rate were adopted for model suitability for P. nagaensis (Pearson et al. evaluation. AUC values range from 0 to 1, 2007). The overall occurrence was used to where 1 is a perfect fit. Useful models pro- calibrate the model for exhibiting in the end. duce AUC values of 0.7–0.9, and models with “good discriminating ability” produce AUC Niche identity and background test values above 0.9 (Swets 1988). The AUC of Niche overlaps were measured by testing the the receiver operating characteristic plot is a similarities between habitat suitability predic- threshold-independent measure of model ac- tions of the 2 species using ENM tools curacy, which juxtaposes correct and incorrect (Warren et al. 2008, 2010). Although only 6 predictions over a range of thresholds. Since occurrence records were available for P. na- AUC is sensitive to background size for sam- gaensis, which might lead to low significance, pling pseudoabsence data, the background the 6 records were widely distributed and

Journal of Insect Science | http://www.insectscience.org 4

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al.

Table 2. Six occurrence records for Parastrachia nagaensis, together with the model’s success in predicting the excluded point in ques- tion, and the suitability score of each point in Maxent models trained using all P. nagaensis occurrence points.

could represent the geographic range of the minimum squares that covered all the occur- species’ distribution. The Schoener’s D rence of P. japonensis or P. nagaensis were

(Schoener 1968) and Warren’s I statistic set as the backgrounds. This disjunct disposal (Warren et al. 2008) were used because they fit well the two species’ general dispersal pat-

were directly based on suitability scores and terns (Schaefer et al. 1988, 1991), and could have been widely used for niche overlap represent their geographic distribution range.

measurements (McCormack et al. 2010; 500 replicates were used for both the identity Hawlitschek et al. 2011; Peterson 2011). The and background test in ENM tools (Warren et

metrics D and I were calculated by taking the al. 2010). difference between species in suitability score

at each grid cell. The two metrics ranged from Results 0 (species have completely discordant ENM)

to 1 (species have identical ENM) (Warren et Geographic distribution al. 2010). In 2007, Schaefer and Kikuhara mapped the

distribution of the genus, and reported a new Niche identity and background tests were then country record of P. nagaensis in Laos, which

performed to determine whether the ENM extended the known species’ range about 300 generated for the 2 species were identical or km to south. Some records appeared on the

exhibited significant difference, and whether map with unidentified specimens from China. these differences were caused by the envi- Our results suggest that some of them were P.

ronmental feature spaces. The niche identity nagaensis, and the others were P. japonensis test works by pooling actual occurrence points (Figure 1). P. japonensis showed a continuous

and randomizing their identities to produce 2 distribution from Hengduan region to southern new samples with the same numbers of obser- Japan with the north extended to western Mt.

vations as empirical data. Niche overlap Qinling, while P. nagaensis exhibited a cryp- values generated by the actual occurrence data tic habitat with sporadic distributional records

were then compared with those generated by (Figure 1). the empirical data. The background test was

used to ask whether the 2 species were more Principle component analysis or less similar than expected based on the dif- PCA of pooled environmental variables re-

ferences in the available environmental vealed reduced significant components, backgrounds. It works by comparing actual defining a realized niche space occupied by P. niche overlap with those generated using japonensis and P. nagaensis. The first 3 com- points drawn at random from the region de- ponents of the PCA were significant, and fined as environmental background for one of together explained 72.94% of the overall vari- the species (Warren et al. 2008, 2010). The ance. The first component (PC-1) was related

Journal of Insect Science | http://www.insectscience.org 5

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. to temperature and precipitation, mainly con- portant variable. In the ecological dimensions, tributed to by annual mean temperature, the 2 species occupied 2 distinct climate spac- minimum temperature in the coldest month, es (Figure 4), P. nagaensis occupied climate annual precipitation, and precipitation in the spaces of lower precipitation and higher tem- driest month. The second component (PC-2) peratures compared to P. japonensis, was associated with the lowest weekly radia- suggesting it is more tolerant to dry and high tion. The third component (PC-3) was less temperature conditions. clearly associated with a single dimension (Table 1). Much niche space of P. nagaensis Niche identity and background test was differentiated from P. japonensis along The Schoener’s D and Warren’s I were 0.375 the second and third components in the 3- and 0.619 respectively for the actual niche dimensional plot, suggesting niche space overlap. Values of D were generally lower might be diverged between the 2 species (Fig- than those of I. In the identity test, the niche ure 2). overlaps of D and I were 0.500 ± 0.056 and 0.750 ± 0.041 for the random resample occur- Ecological niche modeling rence data. The actual niche overlaps (D and The model output of P. japonensis showed I) were quite different from those of the ran- good performance compared to random ex- dom data (Figure 5), were outside the 95% pectation (AUC = 0.82); the omission rate at confidence intervals, and thus were signifi- the conservative threshold of the tenth percen- cant. In the background test, the focal species tile training presence was 33.3%. The model P. nagaensis with the background of P. ja- of P. nagaensis as measured by the Pearson ponensis showed that the D and I were 0.438 jackknife-based test procedure was signifi- ± 0.037 and 0.708 ± 0.036, and the focal spe- cantly better than random expectations (p > cies P. japonensis with the background of P. 0.01) (Table 2). The suitability score at each nagaensis showed that the D and I were 0.297 occurrence point for P. nagaensis ranged from ± 0.040 and 0.554 ± 0.055. Although actual 0.26 to 0.83, while suitability for P. japonen- niche overlap values appeared at the lower sis ranged from 0.06 to 0.95. Areas in central end of the distribution of overlaps from ran- southern China and southern Japan showed domly drawn points (Figure 5), the actual high suitability for P. japonensis, while the niche overlap (D and I) was inside the 99% eastern Taiwan, mainland of South Korea, and confidence intervals of the background test northern Japan were also suitable (Figure 3). results and therefore were not significant. Highly suitable areas identified by the model of P. nagaensis, including the southeastern Discussion Qinghai-Tibet Plateau, western Mt. Qinling, Guizhou and Guangxi Provinces, eastern In- Model interpretation dochina, and the disjunct areas in Limitations on the material and methodology Heilongjiang and Iner-Monglia were also employed in this study need to be addressed identified as suitable (Figure 3). Significant here. Due to cryptic habits, locally restricted variables identified by the jackknife test for P. distributions, or low sampling effort, the oc- japonensis included precipitation in the driest currence data of P. nagaensis was limited to month and the lowest weekly radiation. The only 6 available records. Ecological dimen- jackknife test for P. nagaensis showed that sions could not be directly compared and highest weekly radiation was the most im- statically tested using scarce occurrence data,

Journal of Insect Science | http://www.insectscience.org 6

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. and model extrapolation based on such data Schaefer et al. (1988, 1991) suggested that P. should be performed with caution (Stockwell japonensis might be the more plesiomorphic and Peterson 2002; Pearson et al. 2007; Wisz of the 2 species, and it might more closely re- et al. 2008; Peterson 2011). In our study, the semble the 2 species’ ancestor. They model output of P. nagaensis could not be in- hypothesized that P. nagaensis was the west- terpreted as predicting actual limits of the ern descendant of a more widely ranging species’ range, but could identify regions that ancestor, while P. japonensis was the more had similar environmental conditions to areas eastern descendant. The mapped distribution where the species was known to occur (Pear- supported the above hypothesis and further son et al. 2007), while the model output of P. filled in the blank region between southwest- japonensis could be interpreted as the suitabil- ern China and southern Japan (Figure 1). The ity index in the geographic space. current distribution of P. japonensis might be the result of population expansion from the Geographic distribution Hengduan region, as the Hengduan region has Southwestern and southern China is one of the acted as a refuge for hosting many ancient most species-rich regions in the world (Dai et species (Huang et al. 2006). Recent studies al. 2011). Potential distributional areas in- showed close a relationship of the genus ferred by ENM for the genus Parastrachia Parastrachia with the Ethiopian genus Dis- might be useful for future field surveys in the- megistus Amyot and Serville (Grazia et al. se areas. ENM seeks to characterize the 2008; Pluot-Sigwalt and Lis 2008), however, realized niche in a way that approaches the the genus Dismegistus was poorly studied, and fundamental niche without consideration of the divergence of these 2 lineages might be the species’ dispersal ability or biotic interac- related to historical geographic events. tions (Soberón and Peterson 2005). In some areas identified as suitable, like eastern Tai- Ecological niche wan, the mainland of South Korea, and Significant variables identified by the PCA northern Japan for P. japonensis, or the Hei- and ENM were in accordance with the general longjiang and Iner-Monglia for P. nagaensis, biology of P. japonensis. The habitat of P. the species were absent, potentially due to the japonensis was restricted to early-stage sec- dispersal limits or lack of availability of suita- ondary forests in foothill areas (Filippi et al. ble host plants. The distribution of P. 2001). This insect is a specialist feeder that nagaensis also occurred in the indo-Burma utilizes drupes of the Olacaceous tree, Schoep- and South Central China hotspot (Mayers et fia jasminodora, as its sole food resource al. 2000). Species in hotspots tend to be scarce (Tachikawa and Schaefer 1985). It requires within their range, which increases their prob- good shade and usually forms aggregations on ability of extinction (Brown 1984; Gaston the underside of hard waxy leaves for summer 1994). The potential distribution suggests this estivation (Filippi et al. 2001). In winter, the insect’s range could extend beyond its pres- consolidated, massive aggregations enter into ently known distribution to further south. the large soil holes under leaf litter for hiber- Further investigation could aid in forming a nation (Hironaka et al. 2005). In the PCA, the more complete picture of the genus’ distribu- first component was associated with tempera- tion, which in turn could aid in their ture and precipitation, whereas the second was conservation, particularly for P. nagaensis. related to sunshine. In the ENM, 2 significant variables were identified for P. japonensis,

Journal of Insect Science | http://www.insectscience.org 7

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. including precipitation in the driest month and the lowest weekly radiation. The jackknife test Soberón and Peterson (2011) proposed to for P. nagaensis may not be reliable due to its adopt the fundamental niche, potential niche, small sample size. The host plant and habitat and realized niche (or existing fundamental for P. nagaensis are currently unknown. P. niche) to elaborate the ENM based niche dif- nagaensis is usually distributed in mountain ferentiation issue. They insist that the ENM areas of high altitude (500–3500 m a.s.l.). characterizes somewhere between the realized Schaefer et al. (1991) outlined the general dis- and potential niche. The above observed niche tribution of Olacaceous trees and Schoepfia difference can be considered as the realized spp. in China, Japan, and India, and suggested niche difference, but may not indicate that the that S. jasminodora may be the host of P. ja- 2 species are actually biologically different at ponensis in China, and some other Schoepfia all (i.e., they may not have any significant dif- species may be the host(s) of P. nagaensis. ferences in their fundamental niche), since the Further field investigations may reveal the fundamental niche is generally and not fully real host plant and cryptic habits of P. na- manifested in a certain environment (Soberón gaensis. and Peterson 2011). The observed niche dif- ference might also be due to high Niche divergence dimensionality of the environmental variables Reduced niche dimensions identified by the and the limited distributional records (Peter- PCA occupied by the 2 species allows investi- son 2011). Although we tried to reduce such gation of niche conservatism and negative effects in niche modeling, the niche differentiation. In 3 dimensions, P. nagaensis divergence analysis here is tentative, based on showed a shifting niche space from P. ja- the limited records available for P. nagaensis. ponensis along the second and third The results suggest that the 2 species are oc- components, and much of the niche space of cupying different habitats, but that may be P. nagaensis was differentiated (Figure 2). driven primarily by availability (Warren et al. The second component was associated with 2008). radiation while the third was less clearly asso- ciated with a single dimension (Table 1). The ENM offers useful information for the identi- visualization of modeled niche space in eco- fication of poorly known species. Evidence of logical dimensions also suggests the 2 species niche divergence had been explored for spe- occupied 2 climate spaces, with P. nagaensis cies delimitation (Raxworthy et al. 2007; more tolerant of dry and high temperature Rissler and Apodaca 2007; Leaché et al. 2009; condition (Figure 4). In the identity and back- Martínez-Gordillo et al. 2010), and many ground test, a significant niche identity test studies have shown its coincidence with DNA indicates significant differences in niches be- sequence data (Fitzpatrick et al. 2007; Waltari tween the 2 species; however, the possibility et al. 2007; Flanders et al. 2010; Hawlitschek that this differentiation is primarily driven by et al. 2011). Somatic characters for discrimi- allopatry cannot be ruled out. Although the 2 nating between the 2 species sometimes species do not occupy identical habitats, the appear unstable (Schaefer and Kikuhara background test suggests the differences be- 2007); however, genital characters support tween their ENMs could simply be due to the their distinct existence (Schaefer et al. 1988, differential availability of habitat in the differ- 1991; Schaefer and Kikuhara 2007). ENM for ent regions that they occupy. species delimitation is quite useful for taxo-

Journal of Insect Science | http://www.insectscience.org 8

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. nomic groups with low vagility, localized en- Subjects (Insect Systematics, No. J0630963), demism, and poorly known distribution and a talent introduction program award to (Raxworthy et al. 2007), which might be use- Gengping Zhu from Tianjin Normal Universi- ful for the genus Parastrachia. ty.

Conclusion References Parastrachiidae is a small stinkbug family containing one genus and two species, P. ja- Beyer HL. 2004. Hawth's Analysis Tools for ponensis and P. nagaensis. The former ArcGIS. Available online: showed a continuous distribution from the http://www.spatialecology.com/htools Hengduan region to southern Japan, with the north extended to western Mt. Qinling, Broennimann O, Treier UA, Müller-Schärer whereas the latter exhibited a cryptic habitat H, Thuiller W, Peterson AT, Guisan A. 2007. with sporadic distributional records. The po- Evidence of climatic niche shift during tential distributions of the 2 species were biological invasion. Ecology Letters 10: 701– anticipated using ENM, which may prove use- 709. ful in searches for new populations and in conservation of this genus. The ecological di- Brown JH. 1984. On the relationship between mensions occupied by the 2 species were abundance and distribution of species. The identified and compared, and the comparison American Naturalist 124: 255–279. suggested the 2 species do not occupy identi- cal habitats. But, the differences could simply Dai C, Zhao N, Wang W, Lin C, Gao B, Yang be due to the differential availability of habitat XJ, Zhang Z, Lei F. 2011. Profound climatic in the different regions they occupy. effects on two East Asian black-throated Tits (Ave: Aegithalidae), revealed by ecological Acknowledgements niche models and phylogeographic analysis. PLOS ONE 6: e29329. The authors thank Dr. Dávid Rédei (Hungari- an Natural History Museum, Hungary) for Donalisio MR, Peterson AT. 2011. verifying identification of Parastrachia in Environmental factors affecting transmission Nankai University, Dr. Schaefer (University risk for hantaviruses in forested portions of of Connecticut, USA) for encouragement to southern Brazil. Acta Tropica 119: 125–130. do this work, Dr. Peterson (University of Kan- sas, USA) for valuable discussions on Elith J, Graham CH, Anderson RP. 2006. ecological niche, and Dr. Mao Chen (Monsan- Novel methods improve prediction of species’ to Co., USA) and Prof. Shuxia Wang (Nankai distributions from occurrence data. Ecography University, China) for reviewing earlier ver- 29: 129–151. sions of the manuscript. Special thanks to Dr. Warren (University of Texas at Austin, USA) Elith J, Phillips SJ, Hastie T, Dudík M, Chee for comments on model protocols and linguis- YE, Yates CJ. 2011. A statistical explanation tic writing. This research was supported by of MaxEnt for ecologists. Diversity and the National Natural Science Foundation of Distributions 17: 43–57. China (No. 30870328), the National Educa- tion Project in Basic Science for Special

Journal of Insect Science | http://www.insectscience.org 9

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. ESRI. 2006. ArcGIS 10. Environmental Filippi L, Hironaka M, Nomakuchi S. 2001. A Systems Research Institute. review of the ecological parameters and implications of subsociality in Parastrachia Fitzpatrick MC, Weltzin JF, Sanders NJ, japonensis (Hemiptera: Cydnidae), a Dunn RR. 2007. The biogeography of semelparous species that specializes on a poor prediction error: why does the introduced resource. Population Ecology 43: 41–50. range of the fire ant over-predict its native range? Global Ecology and Biogeography 16: Gapud V. 1991. A generic revision of the 24–33. subfamily Asopinae with consideration of its phylogenetic position in the family Flanders J, Wei L, Rossiter SJ, Zhang S. 2011. Pentatomidae and superfamily Pentatomoidea Identifying the effects of the Pleistocene on (Hemiptera: Heteroptera). Philippine the greater horseshoe bat, Rhinolophus Entomology 8: 865–961. ferrumequinum, in East Asia using ecological niche modelling and phylogenetic analyses. Gaston KJ. 1994. Rarity. Chapman and Hall. Journal of Biogeography 38: 439–452. Gazetteer of China. 1997. Sinomaps Press. Filippi L, Nomakuchi S, Kuki K, Tojo S. 1995a. Adaptiveness of parental care in Goel SC, Schaefer CW. 1970. The structure of Parastrachia japonensis (Hemiptera: the pulvillus and its taxonomic value in the Cydnidae). Annals of the Entomological land Heteroptera (Hemiptera). Annals of the Society of America 88: 374–383. Entomological Society of America 63: 307– 313. Filippi L, Nomakuchi S, Tojo S. 1995b. Habitat selection, distribution, and abundance Graham CH, Ron SR, Santos JC, Schneider of Parastrachia japonensis (Hemiptera: CJ, Moritz C. 2004. Integrating phylogenetics Cydnidae) and its host tree. Annals of the and environmental niche models to explore Entomological Society of America 88: 456– speciation mechanisms in dendrobatid frogs. 464. Evolution 58: 1781–1793.

Filippi L, Hironaka M, Nomakuchi S, Tojo S. Grazia J, Schuh RT, Wheeler WC. 2008. 2000a. Provisioned Parastrachia japonensis Phylogenetic relationships of family groups in (Hemiptera: Cydnidae) nymphs gain access to Pentatomoidea based on morphology and food and protection from predators. Animal DNA sequences (Insecta: Heteroptera). Behaviour 60: 757–763. Cladistics 24: 1–45.

Filippi L, Nomakuchi S, Hironaka M, Tojo S. Grinnell J. 1917. The niche-relationships of 2000b. Insemination success discrepancy the California thrasher. Auk 34: 427–433. between long-term and short-term copulations in the provisioning shield bug, Parastrachia Grinnell J. 1924. Geography and evolution. japonensis (Hemiptera: Cydnidae). Journal of Ecology 5: 225–229. Ethology 18: 29–36. Guisan A, Thuiller W. 2005. Predicting species distribution: Offering more than

Journal of Insect Science | http://www.insectscience.org 10

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. simple habitat models. Ecology Letter 8: 993– (Heteroptera: Cydnidae), under different 1009. photic conditions. Zoological Science 20: 423–428. Hasan SA, Kitching IJ. 1993. A cladistic analysis of the tribes of the Pentatomidae Hironaka M, Nomakuchi S, Iwakuma S, (Heteroptera). Japanese Journal of Filippi L. 2005. Trophic egg production in a Entomology 61: 651–669. subsocial shield bug, Parastrachia japonensis Scott (Heteroptera: Parastrachiidae), and its Hasan SA, Nasreen S. 1994. Structure and functional value. Ethology 111: 1089–1102. phylogeny of pretarsus in pentatomoid bugs (Heteroptera: Pentatomoidea). Proceedings of Hosokawa T, Kikuchi Y, Nikoh N, Meng XY, Pakistan Congress of Zoology 14: 183–190. Hironaka M, Fukatsu T. 2010. Phylogenetic position and peculiar genetic traits of a midgut Hawlitschek O, Porch N, Hendrich L, Balke bacterial symbiont of the stinkbug M. 2011. Ecological niche modelling and Parastrachia japonensis. Applied and nDNA sequencing support a new, Environmental Microbiology 76: 4130–4135. morphologically cryptic beetle species unveiled by DNA barcoding. PLOS ONE 6: Huang XL, Qiao GX, Lei FM. 2006. Diversity e16662. and distribution of aphids in the Qinghai- Tibetan Plateau-Himalayas. Ecological Hijmans RJ, Cameron SE, Parra JL, Jones PG, Entomology 31: 608–615. Jarvis A. 2005. Very high resolution interpolated climate surfaces for global land IBM. 2009. SPSS 19. IBM SPSS Statistics. areas. International Journal of Climatology 25: 1965–1978. Kambhampati S, Peterson AT. 2007. Ecological niche conservation and Hironaka M, Tojo S, Nomakuchi S, Filippi L, differentiation in the wood-feeding Hariyama T. 2007. Round-the-clock homing cockroaches, Cryptocercus, in the United behavior of a subsocial shield bug, States. Biological Journal of the Linnean Parastrachia japonensis (Heteroptera: Society 90: 457–466. Parastrachiidae), using path integration. Zoological Science 24: 535–541. Kriticos DJ, Webber BL, Leriche A, Ota N, Macadam I, Bathols J, Scott JK. 2011. Hironaka M, Horiguchi H, Filippi L, CliMond: global high resolution historical and Nomakuchi S, Tojo S, Hariyama T. 2003a. future scenario climate surfaces for Progressive change of homing navigation in bioclimatic modeling. Methods in Ecology the subsocial bug, Parastrachia japonensis and Evolution 3: 53–64. (Heteroptera: Cydnidae). Japanese Journal of Entomology (New Series) 6: 1–8. Leaché AD, Koo MS, Spencer CL, Papenfuss TJ, Fisher RN, McGuire JA. 2009. Hironaka M, Nomakuchi S, Filippi L, Tojo S, Quantifying ecological, morphological, and Horiguchi H, Hariyama T. 2003b. The genetic variation to delimit species in the directional homing behaviour of the subsocial coast horned lizard species complex shield bug, Parastrachia japonensis

Journal of Insect Science | http://www.insectscience.org 11

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. (Phrynosoma). Proceedings of the National preferred food. Journal of Insect Behavior 11: Academy of Sciences USA 106: 12418–12423. 605–619.

Lis JA, Ziaja DJ. 2010. Pretarsal structures in Nomakuchi S, Filippi L, Hironaka M. 2001. the family Cydnidae sensu lato (Hemiptera: Nymphal occurrence pattern and predation Heteroptera: Pentatomoidea). Zootaxa 2545: risk in the subsocial shield bug, Parastrachia 23–32. japonensis (Heteroptera: Cydnidae). Applied Entomology and Zoology 36: 209–212. Liu X, Guo Z, Ke Z, Wang S, Li Y. 2011. Increasing potential risk of a global aquatic Ortega-Huerta MA, Peterson AT. 2008. invader in Europe in contrast to other Modeling ecological niches and predicting continents under future climate change. PLOS geographic distributions: a test of six ONE 6: e18429. presence-only methods. Revista Mexicana de Biodiversidad 79: 205–216. Martínez-Gordillo D, Rojas-Soto O, de los Monteros AE. 2010. Ecological niche Pearson RG. 2007. Species’ distribution modelling as an exploratory tool for modeling for conservation educators and identifying species limits: an example based practitioners. Synthesis. American Museum of on Mexican muroid rodents. Journal of Natural History. Available online: Evolutionary Biology 23: 259–270. http://ncep.amnh.org

McCormack JE, Zellmer AJ, Knowles LL. Pearson RG, Raxworthy CJ, Nakamura M, 2010. Does niche divergence accompany Peterson AT. 2007. Predicting species allopatric divergence in Aphelocoma jays as distributions from small numbers of predicted under ecological speciation? occurrence records: a test case using cryptic Insights from tests with niche models. geckos in Madagascar. Journal of Evolution 64: 1231–1244. Biogeography 34: 102–117.

Medley KA. 2010. Niche shifts during the Peterson AT. 2011. Ecological niche global invasion of the Asian tiger mosquito, conservatism: a time-structured review of Aedes albopictus Skuse (Culicidae), revealed evidence. Journal of Biogeography 38: 817– by reciprocal distribution models. Global 827. Ecology and Biogeography 19: 122–133. Peterson AT, Soberón J, Sánchez-Cordero V. Myers N, Mittermeier RA, Mittermeier CG, 1999. Conservatism of ecological niches in da Fonseca GAB, Kent J. 2000. Biodiversity evolutionary time. Science 285: 1265–1267. hotspots for conservation priorities. Nature 403: 853–858. Phillips SJ, Anderson RP, Schapire RE. 2006. Maximum entropy modeling of species Nomakuchi S, Filippi L, Tojo S. 1998. geographic distributions. Ecological Selective foraging behavior in nest- Modelling 190: 231–259. provisioning females of Parastrachia japonensis (Hemiptera: Cydnidae): Cues for Phillips SJ, Dudik M, Schapire RE. 2004. A maximum entropy approach to species

Journal of Insect Science | http://www.insectscience.org 12

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. distribution modeling. Proceedings of the Sánchez-Fernández D, Lobo JM, Abellán P, Twenty-First International Conference on Millán A. 2011. Environmental niche Machine Learning 655–662. divergence between genetically distant lineages of an endangered water beetle. Pluot-Sigwalt D, Lis JA. 2008. Morphology of Biological Journal of the Linnean Society the spermatheca in the Cydnidae (Hemiptera: 103: 891–903. Heteroptera): Bearing of its diversity on classification and phylogeny. European Schaefer CW, Dolling WR, Tachikawa S. Journal of Entomology 105: 279–312. 1988. The shieldbug genus Parastrachia and its position within the Pentatomoidea (Insecta: Pyron RA, Burbrink FT. 2009. Lineage Hemiptera). Zoological Journal of the diversification in a widespread species: roles Linnaean Society 93: 283–311. for niche divergence and conservatism in the common kingsnake, Lampropeltis getula. Schaefer CW, Kikuhara Y. 2007. Molecular Ecology 18: 3443–3457. Parastrachia nagaensis (Distant) (Hemiptera: Parastrachiidae) from Laos. Oriental Raxworthy CJ, Ingram CM, Rabibisoa N, 41: 459–462. Pearson RG. 2007. Applications of ecological niche modeling for species delimitation: a Schaefer CW, Zheng LY, Tachikawa S. 1991. review and empirical evaluation using day A review of Parastrachia (Hemiptera: geckos (Phelsuma) from Madagascar. Cydnidae: Parastrachiinae). Oriental Insects Systematic Biology 56: 907–923. 25: 131–144.

Rice NH, Martínez-Meyer E, Peterson AT. Schoener TW. 1968. The Anolis lizards of 2003. Ecological niche differentiation in the Bimini: resource partitioning in a complex Aphelocoma jays: a phylogenetic perspective. fauna. Ecology 49: 704–726. Biological Journal of the Linnean Society 80: 369–383. Schuh RT, Slater JA. 1995. True Bugs of the World (Hemiptera: Heteroptera). Rissler LJ, Apodaca JJ. 2007. Adding more Classification and Natural History. Cornell ecology into species delimitation: ecological University Press. niche models and phylogeography help define cryptic species in the black salamander Soberón J, Peterson AT. 2005. Interpretation (Aneides flavipunctatus). Systematic Biology of models of fundamental ecological niches 56: 924–942. and species’ distributional areas. Biodiversity Informatics 2: 1–10. Rödder D, Lötters S. 2009. Niche shift versus niche conservatism? Climatic characteristics SPSS. 2009. SigmaPlot for Windows. Version of the native and invasive ranges of the 11.2. SPSS, Inc. Mediterranean house gecko (Hemidactylus turcicus). Global Ecology and Biogeography Stockwell DRB, Peterson AT. 2002. Effects 18: 674–687. of sample size on accuracy of species distribution models. Ecological Modelling 148: 1–13.

Journal of Insect Science | http://www.insectscience.org 13

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al. Waltari E, Hijmans RJ, Peterson AT, Nyári Sweet MH, Schaefer CW. 2002. ÁS, Perkins SL, Guralnick RP. 2007. Parastrachiinae (Hemiptera: Cydnidae) raised Locating Pleistocene Refugia: Comparing to family level. Annals of the Entomological Phylogeographic and Ecological Niche Model Society of America 95: 441–448. Predictions. PLOS ONE 2: e563.

Swets JA. 1988. Measuring the accuracy of Wiens JJ, Graham CH. 2005. Niche diagnostic systems. Science 240: 1285–1293. conservatism: integrating evolution, ecology, and conservation biology. Annual Review of Tachikawa S, Schaefer CW. 1985. Biology of Ecology, Evolution, and Systematics 36: 519– Parastrachia japonensis (Hemiptera: 539. Pentatomoidea: ?-idae). Annals of the Entomological Society America 78: 387–397. Wisz MS, Hijmans RJ, Li J, Peterson AT, Graham CH, Guisan A, NCEAS Predicting Tsukamoto L, Kuki K, Tojo S. 1994. Mating Species Distributions Working Group. 2008. tactics and constraints in the gregarious insect Effects of sample size on the performance of Parastrachia japonensis (Hemiptera: species distribution models. Diversity and Cydnidae). Annals of the Entomological Distributions 14: 763–773. Society of America 87: 962–971. Zhu GP, Bu WJ, Gao YB, Liu GQ. 2012. Tsukamoto L, Tojo S. 1992. A report of Potential geographic distribution of Brown progressive provisioning in a stink bug, Marmorated Stink Bug invasion Parastrachia japonensis (Hemiptera: (Halyomorpha halys). PLOS ONE 7: e31246. Cydnidae). Journal of Ethology 10: 21–29.

USGS. 2001. HYDRO1k elevation derivative database. US Geological Survey.

Warren DL, Glor RE, Turelli M. 2008. Environmental niche equivalency versus conservatism: quantitative approaches to niche evolution. Evolution 62: 2868–2883.

Warren DL, Glor RE, Turelli M. 2010. ENMTools: a toolbox for comparative studies of environmental niche models. Ecography 33: 607–611.

Journal of Insect Science | http://www.insectscience.org 14

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al.

Figure 1. Geographic distribution of Parastrachia. Low to high

altitudes are shown as a ramp from white to black. The Chinese

province names in the text were simplified on the map (NM:

Inner Mongolia, HLJ: Heilongjiang, SC: Sichuan, GZ: Guizhou,

GX: Guangxi, YN: Yunnan). Numbers beside black dots indicate

the P. nagaensis records (1: Vietnam, New country record, 2:

India, 3: Ya’an, China, 4: Yunnan, China, 5: Mt. Qingcheng, Chi- Figure 3. Potential distributions of Parastrachia japonensis and na, 6: Laos). High quality figures are available online. P. nagaensis based on Maxent. Model predictions from low to

high suitability are shown as a ramp from white to black. White

and black dots indicate the occurrence of P. japonensis and P.

nagaensis. High quality figures are available online.

Figure 2. Principal component analysis of 13 variables associ-

ated with occurrences of Parastrachia japonensis and P. nagaensis.

Symbols represent the insects’ occurrence in reduced 3 dimen-

sions. High quality figures are available online. Figure 4. Two-dimensional visualization of the ecological

niche of Parastrachia japonensis and P. nagaensis. Model predicted

presence of P. japonensis (square) and P. nagaensis (triangle) according to the combinations of climate variables and the final

predictions. High quality figures are available online.

Journal of Insect Science | http://www.insectscience.org 15

Journal of Insect Science: Vol. 13 | Article 102 Zhu et al.

Figure 5. Results of niche identity (upper two panels) and background tests (lower two panels). Gray and black columns represent niche overlap values (X-axis) of D and I created in replicates of identity and background tests. Arrows indicate actual niche overlap distributed in the frequency (Y-axis) of the replicates. For the background tests, results are given both for Parastrachia japonensis (compared to the background of P. na- gaensis) and P. nagaensis (compared to the background of P. japonensis). High quality figures are available online.

Supplementary Figure 1. Specimens examined in the Institute of Entomology at Nankai University

Journal of Insect Science | http://www.insectscience.org 16