Turkish Journal of Zoology Turk J Zool (2018) 42: 330-336 http://journals.tubitak.gov.tr/zoology/ © TÜBİTAK Research Article doi:10.3906/zoo-1709-21

Potential distribution of the guava psyllid Triozoida limbata (, Psylloidea), today and in global climate change scenarios

1 1 1 Dalva Luiz de QUEIROZ , Marcos Silveira WREGE , Talita Benedcta Santos KÜNAST , 1 2, Marilice Cordeiro GARRASTAZU , Daniel BURCKHARDT * 1 Embrapa Florestas, Colombo, Paraná, Brazil 2 Natural History Museum, Basel, Switzerland

Received: 19.09.2017 Accepted/Published Online: 28.03.2018 Final Version: 04.05.2018

Abstract: The jumping plant-louseTriozoida limbata is monophagous on guava (Psidium guajava), inducing leaf galls. Guava is planted today in warmer regions around the world but T. limbata is restricted to Central and South America. Recently, the latter has become a key pest in commercial plantations in Brazil (Pernambuco and São Paulo). Modeling of potential distributions is a good tool for developing control strategies of pests. Based on distributional data from the Americas, we created models for the psyllids for the present and future using different scenarios of climate change. The OpenModeller eco-niche modeling program was used with the Environmental Distance and Envelope Score algorithms. The potential distribution of T. limbata covers Central America, tropical and subtropical South America, sub-Saharan Africa except for the Kalahari and Cape regions, the southwest of the Indian subcontinent, Southeast Asia, and the north of Australia. In India, the largest guava producer of the world, the conditions for T. limbata are most favorable in the west and south. The effects of global climate change will be most felt in Brazil, where the decrease of suitable areas for T. limbata will concentrate the pest in the east and northeast and thus put more pressure on guava, increasing the potential for psyllid-induced damage.

Key words: Jumping plant-lice, , Psidium guajava, Myrtaceae, Environmental Distance, Envelope Score, leaf roll galls, Neotropics, Brazil

1. Introduction html). In Brazil, the annual harvest approximates 350,000 Guava, Psidium guajava (Myrtaceae), is a fruit with an t, of which about 240,000 t are produced in the states of exquisite flavor and high vitamin C content. It originates São Paulo and Pernambuco (Agrianual, 2016). Although from tropical America, probably between southern guava easily adapts to a variety of climatic conditions, it is Mexico and northern South America, and is widely attacked by a variety of pests. Over 100 insect species have distributed and well established today throughout the been reported from guava plantations in Brazil (Maricone tropical and subtropical regions of the world (Medina et and Soubihe Sobrinho, 1961). Key pests are Tephritidae al., 1991; Gonzaga Neto, 2007; Singh, 2007). According to (Diptera), which attack the fruits (Gould and Raga, 2002). Singh (2007), guava trees produce heavy crops every year Among the secondary pests are many Hemiptera and provide good economic returns as very little input is species, including the jumping plant-louse Triozoida required. limbata (Enderlein) (Psylloidea: Triozidae) [reported The main guava producer in the world is India as Triozoida sp. by Gould and Raga (2002)]. The species (Pommer et al., 2006), which produced 1,710,000 t of is monophagous on Psidium guajava; the record of guava in 2004. Other important guava producing countries P. cattleianum as host by Semeão et al. (2012), citing are China, Thailand, Pakistan, Mexico, Indonesia, Brazil, Colombi and Galli (2009), is erroneous. T. limbata occurs and Bangladesh (http://www.worldatlas.com/articles/top- today in Argentina, Bolivia, Brazil, Colombia, Costa Rica, guava-producing-countries-in-the-world.html). In recent Ecuador, El Salvador, French Guiana, Mexico, Panama, years, the cultivation of guava has increased in terms Peru, and Trinidad and Tobago (Brown and Hodkinson, of the surface planted as well as in the intensification 1988; Burckhardt, 1988). The record from Chile: Punta of cultures (Singh, 2007; http://www.worldatlas.com/ Arenas (Burckhardt, 1988) is erroneous and concerns articles/top-guava-producing-countries-in-the-world. Costa Rica: Puntarenas. In Brazil, the psyllid was reported * Correspondence: [email protected] 330 QUEIROZ et al. / Turk J Zool from the states of Amazonas, Bahia, Maranhão, Minas curling of the leaf margins, mainly on the shoots, where Gerais, Pernambuco, Paraná, Rio de Janeiro, and São the immature can be observed. Initially, the leaf Paulo (Burckhardt and Queiroz, 2012). The adults are rolls are yellowish or reddish, but they later become about 2 mm long and are characterized by a black stripe necrotic, reducing the photosynthetic active leaf area. This, on the fore wings, by the head shape, and by the male and in turn, reduces the growth and vigor of the plants, leading female terminalia (Figures 1–4). Immature specimens to diminished fruit production and consequently to poor are elongate, dorsoventrally flattened with short, curved economic performance of the crop (Barbosa et al., 2001, antennae and bear conspicuous spines on the fore margin 2003; Gallo et al., 2002). of the head (Figure 5) (Brown and Hodkinson, 1988; For a long time, T. limbata was of minor economic Burckhardt, 1988; Burckhardt and Brown, 1992). importance, but it has recently become a key pest in Immature T. limbata live in roll galls on the leaves commercial guava plantations in Brazil (states of São of their host (Figure 6) and are covered in whitish wax Paulo and Pernambuco) because of the change in the (Figure 7). The symptoms of attack ofT. limbata are the production system and regular pruning throughout the

Figures 1–7. Triozoida limbata: 1- Habitus of adult; 2- head; 3- male terminalia, in profile; 4- female terminalia, in profile; 5- habitus of immature; 6- open leaf roll gall on guava with immature specimens; 7- leaf roll galls on guava.

331 QUEIROZ et al. / Turk J Zool year (Marcelino, 2013). Though currently not known Museum, Basel, Switzerland; United States National from outside of the Americas, there is a possibility of an Museum, psyllid collection of the USDA, Beltsville, MD, accidental introduction of T. limbata to other continents. United States. Hodkinson (2009), in his seminal review of factors Completeness and consistency of the data set (points of influencing the life cycles of psyllids, concluded that the occurrence) were checked and doubtful or incomplete main factors for the adaptation of life history are the records were eliminated. ambient temperature and the availability of water, acting 2.2. Current climate maps and future climate scenarios directly on the psyllids or mediated through their host Current climate maps and future climate scenarios were plants. created using multiple linear regression, correlating Modeling of population dynamics is an essential part climatic variables with the surface model of the landscape of research on and management of forest pest insects. Its (altitude) and the geographical latitude and longitude. success depends on the selection of an appropriate level The maps were made for the world, at a scale of 1:250,000, of complexity that corresponds to the modeling objectives with spatial resolution equivalent to 90 m at the equator. and to available data (Sharov, 1996). Ecological niche The digital elevation model (DEM) used was GTOPO30, models are often used to predict the potential distribution developed by the United States Geological Survey (https:// of insect species. lta.cr.usgs.gov/GTOPO30), based on the Shuttle Radar Species distributions are dynamic entities independent Topography Mission (SRTM) project (Farr and Kobrick, of the scale of analysis, and so are the factors affecting 2000). them. Integrating how the elements of niche, distribution, For the modeling, we used variables available in and species–habitat association change through time Worldclim, such as average monthly values of air will further enhance our understanding of the dynamics temperature and rainfall. We selected 5 from 19 bioclimatic variables (Table 1). The variables were selected based on of spatial distributions across spatial scales (Hortal their relative importance, defined by a jackknife test, using et al., 2010). The Environmental Distance model is a the software MaxEnt. They were Bio7 (annual temperature mathematical model of prediction of species distribution variation), Bio4 (temperature seasonality), Bio11 (average known as ‘Domain’, which combines the occurrence data temperature of the coldest quarter), Bio14 (rain in the of the species with the environmental variables favorable driest month), and Bio9 (average temperature in the driest to its survival (Rodrigues, 2012). The objective of our study quarter). The relative contribution of each variable was is to help decision making in the management of the pest 33.8%, 29.6%, 9.1%, 6.3%, and 5.8%, respectively. T. limbata in predicting its potential distribution, using For future climate scenarios, the less (A2) and the Environmental Distance model, under present-day the more (B1) pessimistic global models (GCMs) of conditions and for future climatic scenarios outlined in HadGEM2-ES were chosen for the periods 2041–2060 the 4th report of the Intergovernmental Panel on Climate and 2061–2080, which accord with the fourth report of Change (IPCC, 2007). the Intergovernmental Panel on Climate Change (IPCC, 2007). For the future scenarios, the same variables were 2. Materials and methods selected as for the present. 2.1. Occurrence data of Triozoida limbata 2.3. Modeling the potential distribution of Triozoida For the modeling, we used 116 points of occurrence of limbata Triozoida limbata from the Americas (numbers of points The potential distribution of T. limbata was calculated per country or state: Bolivia 1. ‒ Brazil: AM 6; BA 2; CE on the basis of the known distribution in the Americas 3; ES 1; MA 1; MG 5; MS 7; MT 3; PA 1; PB 1; PE 2; PR expressed by the geographical coordinates for the present 4; RJ 1; RR 1; SC 5; SP 42. ‒ Colombia 8. ‒ Costa Rica 2. and for the future using scenarios of climate change. ‒ Ecuador 1. ‒ El Salvador 1. ‒ French Guiana 1. ‒ Mexico We chose the OpenModeller eco-niche modeling 3. ‒ Panama 5. ‒ Peru 4. ‒ Trinidad and Tobago 5; Figure software (Muñoz et al., 2009). The program works with 8) as the species is presently known only from there. The geographic distributions of species (latitudinal and record from Argentina (Burckhardt, 1988) is not included longitudinal coordinates) and environmental variables as it does not specify where the material was collected. (climate, soil, and relief), calculating a mathematical The data are from personal information provided by model for the geographic distribution of species (Muñoz researchers and producers (see Acknowledgments), from a et al., 2009). Here we used the Environmental Distance critical literature survey, and from specimens deposited in and Envelope Score algorithms. The selection of the best following collections: Entomology Laboratory, Embrapa model was made through an area under the curve (AUC) Florestas, Colombo, PR, Brazil; Muséum d’histoire analysis. The contribution of each variable was analyzed naturelle, Geneva, Switzerland; Naturhistorisches by jackknife estimations. Each data set was submitted to

332 QUEIROZ et al. / Turk J Zool five replications, and cross-validation was performed for are no records from Chile and Paraguay and it is unknown each model, where the data were divided into two subsets, whether the species occurs there. one with 80% and the other with 20% of the data, used to Although T. limbata is known only from the Americas, calibrate and validate the final model (http://ncep.amnh. guava, its host, is widely cultivated in tropical and org/linc). The result of the final mean of the models was subtropical regions around the globe. The accidental analyzed by the AUC (Table 2) and by the test set omission introduction of T. limbata to other continents therefore rate (http://ncep.amnh.org/linc). constitutes a risk. The optimal potential distribution The resulting maps of the models were transformed of T. limbata today (Figure 8) covers, in the Americas, into values ranging from 0 to 1, with 0 = no probability an area that is slightly larger than that from which it is of insect occurrence and 1 = maximum probability of reported, stretching further south with high probabilities insect occurrence. The maps created by OpenModeller of occurrence in Paraguay and areas in Rio Grande do in text format, containing these values, were imported Sul (Brazil). Outside the Americas, the probability of into ArcGIS and transformed first into the ‘raster’ format. occurrence is high in most of sub-Saharan Africa (except Classes were then created with a gradient from 0 to 1, for the Kalahari Basin and the Cape Region), Madagascar, representing the gradient of areas with no to those with the extreme south of the Arabian Peninsula, the south of maximum probability of insect occurrence. the Indian subcontinent including Sri Lanka, Southeast Asia, New Guinea, and a narrow strip in the north of 3. Results Australia. In Brazil, the potential occurrence of T. limbata The known distribution of Triozoida limbata stretches is high everywhere, except for regions in the south at between Mexico in the north and Brazil (Santa Catarina) higher altitudes with temperate climate. in the south (Figure 8), covering a variety of different In the most pessimistic global change scenario for biomes. The species occurs also in Argentina but no the period of 2061–2080 (Figure 9), the optimal areas of details are known from there (Burckhardt, 1988). There occurrence of T. limbata will decrease globally and, in

Figures 8 and 9. Predicted distribution of Triozoida limbata using the Environmental Distance algorithm; black circles indicate the most important guava-producing countries or states: 8- for the present, including points of occurrence of T. limbata (gray circles); 9) for the period of 2061–2080, worst-case scenario.

333 QUEIROZ et al. / Turk J Zool

Table 1. Selected climatic variables for the preparation of maps Table 2. AUC results for the five replicated runs used for model predicting the potential occurrence of guava and T. limbata. evaluation.

Code Variable Run number Values Bio04 Temperature Seasonality (standard deviation × 100) 1 0.9985 Bio07 Temperature Annual Range (Bio5 – Bio6) 2 0.9905 Bio09 Mean Temperature of Driest Quarter 3 0.9955 Bio11 Mean Temperature of Coldest Quarter 4 0.9965 Bio14 Precipitation of Driest Month 5 0.9915 particular, will become much more restricted in Central diversity of host species will therefore be important for Brazil. Similar trends can be seen in the less pessimistic their adaptation to the new environmental conditions, scenarios and for less distant periods in the future. as a larger genetic diversity will help to respond more efficiently to climatic changes. 4. Discussion Gutierrez and Ponti (2013) developed a model to Because of the intensification of production with regular evaluate the potential geographic distribution and relative pruning of the plants throughout the year, Triozoida abundance of Diaphorina citri on citrus, its parasitoid, limbata has recently become a key pest in commercial guava and citrus greening disease (huanglongbing, HLB) in plantations in Brazil (states of São Paulo and Pernambuco) North America and the Mediterranean Basin. The authors (Marcelino, 2013). Modeling of the potential distribution observed that the areas with potential for development of this pest is part of the strategy to control T. limbata in of citrus, the psyllid, and HLB overlap in some regions, guava cultures in Brazil (and the world) by identifying representing a large risk for Southeast Spain, Sicily, and the the regions with the most favorable conditions for the eastern Mediterranean region. The authors discussed the occurrence of the pest and therefore where strategically efficiency of biological control, using the joint prediction sanitary control measures should be prioritized. of climatic favorability for the development of T. radiata. Through the AUC analysis, the Environmental Lashkari et al. (2013), using GIS and MaxEnt-related Distance model was selected, being the best with the software, showed that the main areas in Iran suitable highest AUC values for T. limbata. In this work, current for D. citri are restricted to South Kerman, Southeast and future scenarios presented AUC values of 0.981 with Fars, Central and Eastern Hormozgan, and Central 0% omission error. Similar values were also observed in Sistan-Baluchestan. Based on a jackknife test, the annual distributional models of Diaphorina citri in Iran using GIS temperature range has the largest effect in modeling the and the software Maximum Entropy (MaxEnt) (Lashkari distribution of D. citri. The AUC was 0.988, indicating a et al., 2013). Ecological modeling was used by Queiroz et perfect prediction. López-Collado et al. (2013), in a study al. (2013) to estimate the dispersal potential of Glycaspis of habitat distribution of D. citri in Mexico, showed that brimblecombei, based on environmental variables of 502 the most suitable regions for D. citri are the states at the occurrence points of G. brimblecombei worldwide. The Gulf of Mexico, Yucatan, and areas scattered throughout authors tested several models and concluded that the the states along the Pacific coast. Northern and central Environmental Distance model was the most adequate to Mexico, on the other hand, are less suitable for D. citri. predict the potential distribution of G. brimblecombei in This indicates that the citrus-producing states are among new regions. the most suitable regions for the occurrence, development, Our model (Figure 8) predicts for T. limbata favorable and population growth of D. citri, hence increasing the conditions everywhere in Brazil. T. limbata has a short life risk of HLB dispersion. cycle and good migration capacity (Dalberto et al., 2004), According to Christ et al. (2013), the Brazilian allowing for a rapid adaptation to new environmental peppertree, Schinus terebinthifolia, a perennial plant conditions after colonization (introduction) of new native to Brazil, Argentina, Uruguay, and Paraguay, has areas or, in the future, coping with climate change. For become one of the most aggressive invasive weeds in host species that have a long life cycle, the adaptation Florida. A psyllid, terebinthifolii Burckhardt & process should be slower than the evolution of climate Basset (Hemiptera: ), has been identified as change (Hamrick, 2004), and most of the negative a potential biological control agent for this plant. Using effects caused by the loss of genetic diversity will be felt MaxEnt for ecological niche modeling, the authors after a few generations. The maintenance of the genetic confirmed that there was a climatic overlap between

334 QUEIROZ et al. / Turk J Zool

Florida and the native area of the psyllid in South America, The effects of global warming on guava production in and a map was created providing the appropriate sites for Brazil may be negative as the climate will become drier the establishment of the psyllid to be used as an agent of and less suitable for guava. Global warming may thus biological control. increase the vulnerability of guava by attacks of T. limbata. Our results concerning the potential distributions of T. As the current scenario will change with the increase limbata and guava differ from the studies on D. citri and C. of global temperature, the monitoring of populations terebinthifolii in that the area where the crop is planted and of this pest will be important over the next years. The the optimal area for its pest overlap only in part. Whereas predicted potential distributions presented here should be in tropical and subtropical South America conditions for considered when planning new guava plantations in Brazil T. limbata and guava are generally good, Mexico, Florida, (and the world), for preventing T. limbata from turning India, and continental Southeast Asia provide good into a widely distributed major pest. conditions for guava but are less suited for T. limbata. In conclusion, although Triozoida limbata occurs at Acknowledgments present in relatively restricted areas in South and Central We thank José Emílio Bettiol, Valmir Antonio Costa, America, with the expansion of guava cultivation it has a Fernanda Dalberto, Kling Dantas, Pedro José Ferreira large potential for spread in South America, sub-Saharan Filho, Guilherme Melo, and Wilson Carlos Pazini for Africa, southern India, Sri Lanka, and large areas in providing distributional data; Debra D Creel and Cheryle Southeast Asia, and also in future global climate change A O’Donnell (USNM) for access to the collections under scenarios with temperature increases. In India, the major their care; the Instituto Nacional de Pesquisas Espaciais guava producer in the world, the conditions for T. limbata (INPE) for climatic data for future scenarios; and Ibama/ are mostly suitable in the southern part of the peninsula Sisbio (permits 11832-2) for granting collection permits were its establishment and spread are most likely in case for Brazil. The research was partly funded by the Embrapa of an introduction. project number 02.12.01.028.00.00.

References

Agrianual (2016). Agrianual 2015, Goiaba – Produção Brasileira. São Christ LR, Cuda JP, Overholt WA, Vitorino MD, Mukherjee A. Paulo, Brazil: Informa Economics, FNP (in Portuguese). (2013). Biology, host preferences, and potential distribution of Calophya terebinthifolii (Hemiptera: Calophyidae), a candidate Barbosa FR, Ferreira RG, Kiill LHP, Souza EA, Moreira WA, Alencar for biological control of Brazilian peppertree, Schinus JA, Haji FNP (2003). Studies of damage level, weed host plants, terebinthifolius, in Florida. Fla Entomol 96: 137-147. natural enemies and control of Triozoida sp. in guava plants at the São Francisco river valley. Rev Bras Frutic 25: 425-428 Colombi CA, Galli JC (2009). Population dynamic and evolution (article in Portuguese with an abstract in English). of damage of Triozoida limbata (Hem.: Psillydae) in guava orchard in Jaboticabal city, State of São Paulo. Cienc Agrotec Barbosa FR, Souza EA de, Siqueira KMM de, Moreira WA, Alencar 33: 412-416 (article in Portuguese with an abstract in English). JA de, Haji FNP (2001). Efficiency and selectivity of insecticides in the control of Triozoida sp. in guava plants. Pesticidas, Dalberto FMS, Menezes Junior AO, Simões HC, Benito NP, Pitwak J Revista de Ecotoxicologia e Meio Ambiente 11: 45-52 (article (2004). Population dynamic of guava psyllid Triozoida limbata in Portuguese with an abstract in English). (Hemiptera: ) in Londrina Parana, Brazil. Semin Cienc Agrar 25: 87-92 (article in Portuguese with an abstract Brown RG, Hodkinson ID (1988). and Ecology of the in English). Jumping Plant-Lice of Panama (Homoptera: Psylloidea). 1st ed. Leiden, the Netherlands: Brill. Farr TG, Kobrick M (2000). Shuttle radar topography mission produces a wealth of data. Eos Trans AGU 81: 583-585. Burckhardt D (1988). Jumping plant lice (Homoptera: Psylloidea) of the temperate neotropical region. Part 3: Calophyidae and Gallo D, Nakano O, Carvalho RPL, Baptista GC, Berte Filho EB, Triozidae. Zool J Linn Soc-Lond 92: 115-191. Parra JR, Zucchi RA, Alves SB, Vendramin JD, Marchini LC et al. (2002). Entomologia agrícola. 3rd ed. Piracicaba, Brazil: Burckhardt B, Brown RG (1992). Larvae of Panamanian jumping Fundação de Estudos Agrários Luiz de Queiroz – FEALQ (in plant-lice (Homoptera: Psylloidea). In: Qintero D Aiello A, Portuguese). editors. Insects of Panama and Mesoamerica. 1st ed. Oxford, UK: Oxford University Press, pp. 290-301. Gonzaga Neto L (2007). Produção de goiaba. 1st ed. Fortaleza, Brazil: Instituto Frutal (in Portuguese). Burckhardt D, Queiroz DL (2012). Commented checklist of the jumping plant-lice (Hemiptera: Psylloidea) from Brazil. Gould WP, Raga A (2002). Pest of guava. In: Peña JE, Sharp JL, Zootaxa 3571: 26-48. Wysoki M, editors. Tropical Fruit Pests and Pollinators: Biology, Economic Importance, Natural Enemies and Control. New York, NY, USA: CAB International, pp. 295-313.

335 QUEIROZ et al. / Turk J Zool

Gutierrez AP, Ponti L (2013). Prospective analysis of the geographic Medina JC, Castro JV, Sigrist JMM, Martin ZJ, Kato K, Maia ML, distribution and relative abundance of Asian citrus psyllid Garcia AEB, Leite RSSF (1991) Goiaba: cultura, matéria-prima, (Hemiptera: ) and citrus greening disease in North processamento e aspectos econômicos. 2nd ed. Campinas, America and the Mediterranean Basin. Fla Entomol 96: 1375- Brazil: Intituto de Tecnologia de Alimentos (in Portuguese). 1391. Muñoz MES, Giovanni R, Siqueira MF, Sutton T, Brewer P, Pereira Hamrick JL (2004). Response of forest trees to global environmental RS, Canhos DAL, Canhos VP (2009). OpenModeller: a changes. Forest Ecol Manag 197: 323-335. generic approach to species’ potential distribution modelling. GeoInformatica 15: 111-135. Hodkinson ID (2009). Life cycle variation and adaptation in jumping plant lice (Insecta: Hemiptera: Psylloidea): a global synthesis. J Pommer CV, Murakami KRN, Watlington F (2006). Goiaba no Nat Hist 43: 65-179. mundo. O Agronômico, Campinas 58: 22-26 (in Portuguese). Hortal J, Roura-Pascual N, Sanders NJ, Rahbek C (2010). Queiroz DL, Majer J, Burckhardt D, Zanetti R, Fernandez JR, Queiroz Understanding (insect) species distributions across spatial EC, Garrastazu M, Fernandes BV, Anjos N (2013) Predicting scales. Ecography 33: 51-53. the geographical distribution of Glycaspis brimblecombei (Hemiptera: Psylloidea) in Brazil. Aust J Entomol 52: 20-30. IPCC (2007). Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Rodrigues ESC (2012). Teoria da Informação e adaptatividade na IPCC. Cambridge, UK: Cambridge University Press. modelagem de distribuição de espécies. PhD, Polytechnic School of the University of São Paulo, São Paulo, Brazil (in Lashkari M, Sahragard A, Manzari S, Hosseini R, Erfanfar D (2013). Portuguese). Niche modeling of Asian citrus psyllid, Diaphorina citri Kuwayama (Hem.: Psyllidae), in Iran. Plant Pests Res 3: 45-58 Semeão AA, Martins JC, Picanço MC, Chediak M, Silva EM da, Silva (article in Farsi with an abstract in English). GA (2012). Seasonal variation of natural mortality factors of the guava psyllid Triozoida limbata. B Entomol Res 102: 719- Lopez-Collado J, López-Arroyo JI, Robles-García PL, Márquez- 729. Santos M (2013). Geographic distribution of habitat, development, and population growth rates of the Asian citrus Sharov AA (1996). Modeling insect dynamics. In: Korpilahti E, psyllid, Diaphorina citri, in Mexico. J Insect Sci 13: 1-17. Mukkela H, Salonen T editors. Caring for the Forest: Research in a Changing World. Congress Report, Vol. 2 of the IUFRO Marcelino MCS (2013). Distribuição espacial e amostragem (International Union of Forestry Research Organizations) sequencial de Triozoida limbata (Hemiptera: Triozidae) em XX World Congress; 6–12 August 1995; Tampere, Finland. goiabeira. PhD, São Paulo State University, São Paulo, Brazil Jyvaskyla, Finland: Gummerus Printing, pp. 293-303. (in Portuguese). Singh G (2007). Recent development in production of guava. Acta Maricone FAM, Soubihe Sobrinho J (1961). Contribuição para o Hortic 735: 161-176. conhecimento de alguns insetos que depredam a goiabeira (P. guajava L.). Piracicaba, Brazil: University of São Paulo Luiz de Queiroz College of Agriculture (in Portuguese).

336