Progress and spatial pattern of huanglongbing in Persian lime in Nayarit, Mexico

F. J. Márquez-Pérez1; J. L. Flores-Sánchez2; L. Rodríguez Mejía1; J. Márquez Gómez3; S. J. Michereff4; V. Ancona5; A. Robles-Bermúdez6; S. Domínguez-Monge2*

1. Universidad Autónoma Chapingo, Departamento de Parasitología Agrícola, 56230 Texcoco, Estado de México, México., Universidad Autónoma Chapingo, Universidad Autónoma Chapingo, Departamento de Parasitología Agrícola,

<state>Estado de México</state>
, Mexico , 2. Colegio de Postgraduados, Programa de Fitosanidad, 56230 Texcoco, Estado de México, México., Colegio de Postgraduados,
<state>Estado de México</state>
, Mexico ,
3. Comité Estatal de Sanidad Vegetal de Nayarit, 63000 Tepic, Nayarit, México., Comité Estatal de Sanidad Vegetal de Nayarit,
, Mexico ,
4. Universidade Federal Rural de Pernambuco, Departamento de Agronomia, 52171-900 Recife, Pernambuco, Brasil., Universidade Federal Rural de Pernambuco, Universidade Federal Rural de Pernambuco, Departamento de Agronomia,
, Brazil ,
5. Texas A&M University-Kingsville Citrus Center, 78599 Weslaco, TX, USA., Texas A&M University-Kingsville,
, USA ,
6. Universidad Autónoma de Nayarit, Unidad Académica de Agricultura, 63155 Xalisco, Nayarit, México., Universidad Autónoma de Nayarit, Universidad Autónoma de Nayarit, Unidad Académica de Agricultura,
, Mexico

Correspondence: * . Corresponding Author: Domínguez Monge, Santiago, Colegio de Postgraduados, Programa de Fitosanidad, Carretera México-Texcoco Km. 36.5, C.P.56230 Texcoco, Estado de México, México. Phone: (595) 1101 657. E-mail: E-mail:


Huanglongbing (HLB), caused by Candidatus Liberibacter asiaticus (CLas), causes severe losses to citrus (Citrus spp.) growers in Mexico since it was first detected in 2009. This study aimed at analyzing the temporal and spatial progression of the disease at four commercial orchards with HLB presence cultivated with Persian lime (Citrus latifolia Tanaka) located in Xalisco, Nayarit, Mexico. The evaluation of the disease incidence was done by visual inspection of all trees for six months. The epidemics were fitted through the flexible model Weibull (y= 1-[t/b]c), with respect to initial incidence (y0), final incidence (yf), rate of disease progression (b-1 ), and area under disease progress curve (AUDPC). The spatial pattern of the disease was analyzed by geostatistical maps, “ordinary runs” and fitting the beta-binomial distribution. The analysis of the progress curves showed that the epidemic was faster in orchard ED-2, associated with a higher initial incidence proportion, relatively young trees (<3 years) and probably with increased exposure to asian psyllid. Disease incidence increased in all orchards ranging from 1.9 to 32.4 % during the evaluation period. The within rows aggregation increased along with the time and the increase in the disease incidence in EM-1, ED-2 and HM-3 orchards. In orchards IP-4 and HM-3, there was a slight predominance of aggregation within the rows while, in orchard EM-1 and ED-2 the across rows aggregation of diseased trees was prevalent.

Received: 2017 September 18; Accepted: 2018 May 5

revbio. 2020 Mar 23; 5(spe1): e351
doi: 10.15741/revbio.05.nesp.e351

Keywords: Key words: Epidemiology, Candidatus Liberibacter asiaticus, Citrus latifolia, Sampling.


Huanglongbing (HLB) is currently the most devastating disease of the citrus worldwide (Bové, 2006; Bassanezi et al., 2013; Robles-González et al., 2013; Flores-Sánchez et al., 2017). Currently, three species of Candidatus Liberibacter associated with HLB are known: Ca. L. asiaticus, Ca. L. africanus, and Ca. L. americanus, (Lópes, et al., 2009; Gottwald, 2010; Santivañez, et al., 2014). In Mexico, HLB is caused by Candidatus Liberibacter asiaticus (CLas) (Mora-Aguilera et al., 2014a) and transmitted by Asian citrus psyllid (ACP) Diaphorina citri Kuwayama (Hemiptera: Libiidae) (Hall et al., 2013; Torres-Pacheco et al., 2013). The infected trees show Mottling on the leaves, deformed fruits, premature fruit fall and premature mortality in plants (Esquivel-Chávez et al., 2012; Robles-González et al., 2013). The losses in production are affected by 18 % in Persian lime and between 60 to 80 % in Mexican lime (Bassanezi et al., 2011; Robles-González et al., 2013; Flores-Sánchez et al., 2015), which represents a true threat for the Mexican citrus industry.

The temporal and spatial dynamics of HLB have been studied in different countries. For the HLB progress, models that describe the temporal increase of the incidence of HLB at orchard level have been fitted. These models indicate that indidence can reach more than 95 % in a period from three to thirteen years after the first appearing of the symptoms (Bassanezi et al., 2006; Gatineu et al., 2006; Gottwald et al., 2007; Gottwald et al., 2010). Also, it has been observed that the areas in which the disease is endemic and there is an abundant source of inoculum, the disease’s progress rates are faster in young plantations, even with the use of insecticides (Bassanezi et al., 2013). About spatial patterns of HLB, in previous studies performed in China, Brasil and Florida, USA, it has been found that the adding of initial foci of infected tres with HB is associated with secondary focus to a distance from 25 to 50 meters (Gottwald et al., 1991; Bassanezi et al., 2005; Gottwald et al., 2010).

Since CLas was detected for the firts time in Yucatan in 2009 (Salcedo et al., 2010; Trujillo-Arriaga, 2010), 24 citrus growing states have been confirmed with presence of infected trees by HLB (SENASICA, 2018). Despite Despite widespread HLB, in most states of central Mexico and gulf, where approximately 80 % of citrus is produced, HLB is regionally restricted to foci (Flores-Sánchez et al., 2017). Therefore, epidemiologic studies should be reinforced in order to reduce the risk of intruduction, establishment, and spread of CLas. To this effect, the study of te progress curve and the spatial pattern constitutes an important tool for the understanding of epidemic dynamics, revealing the way the disease disseminates and helping with the development of control programs (Madden et al., 2007). Thus, this study aimed to analize spatial and temporally the progress of Huanglongbing at orchard level, as well as determining the rates of incidence per orchard. This way, it is intended to understand the behavior of this disease and which factors contribute to its dispersion.

Materials and Methods

The study was carried out during the period from November, 2010, to February, 2011 in the municipality of Xalisco, Nayarit, Mexico. Four commercial orchards of Persian lime were randomly chosen with presence of HLB and little efficient control of psyllids (Table 1). The collecting of samplings of vegetal tissue for the molecular identification of CLas, as well as the recording of HLB data, were monthly done in all of the sites. In each orchard, a plot of 10 x 10 trees was chosen, consisting of N1 = 10 rows and N2 = 10 trees per row, to have a total population of 100 trees per orchard. The criteria to consider for the selection of the plots were: low number of missing trees, high number of visually healthy trees, and absence of defined foci of HLB (Pérez-Hernandez) et al., 2004). On each plot, the trees were inspected through typical symptoms of HLB (Esquivel-Chávez et al., 2012). Leaf samples were taken from the symptomatic trees in order to confirm CLas. Each sample was composed of for leaves collected from each cardinal point of the tree canopy. For the molecular analysis and diagnosis, the samples composed of vegetal tissue were analyzed in the lab of Centro Nacional de Referencia Fitosanitaria del Servicio Nacional de Sanidad, Inocuidad y Calidad Agroalimentaria (SENASICA). DNA of vegetal tissue was isolated through the CTAB (cetyl tri-methyl-ammonium bromide) method (Dellaporta et al., 1983). CLas detection was done through qPCR following the protocol described by Li et al., (2006).

Table 1.

Characteristics of the 4 orchards evaluated to determine the spatial and temporal progress in Xalisco, Nayarit, Mexico between 2010 and 2011.

Field Trees
Latitude Longitude
EM-1 267 1.5 5x5 7 10 3.75 SO 902 21.317 104.925
ED-2 105 0.5 5x5 2 8 7.62 Volka 1,032 21.331 104.916
HM-3 185 1.2 5x5 5 7 3.79 SO 1,053 21.243 104.920
IP-4 144 1 5x5 6 2 1.38 Volka 1,013 21.356 104.889

TFN11 Number of initial positive trees. 2 Percentage of initial incidence. 3 Rootstock type: SO-Sour Orange; Volka-Volkameriana.

The disease epidemics were characterized using incidence per plot with the Weibull distribution model (Pennypacker et al., 1980) modified at two parameters (Mora-Aguilera et al., 1996). This model (y = 1 - [t/b]c ) was chosen due to its flexibility; where “y” represents the proportion of the disease incidence, “t” is time in months, “b” is the epidemic rate parameter estimated to its inverse form (b-1 ) and “c” is the curve shape parameter (Mora-Aguilera et al., 1996). The estimations of b and c were done using the nonderivative method DUD of PROC NLIN of SAS 9.0 (Jesus et al., 2004). In each evaluation, the proportion of diseased plants was considered as the dependent variable, while as for the time passed in months after the first evaluation was considered as the independent variable. The model fitted for each empidemic was determined through values of the coefficient of determination (R2) between observed and expected values, and the existence or lack of patterns in the graphic of residuals versus expected values (Campbell and Madden, 1990). Additionally, AUDPC was calculated through original data of incidence (Madden et al., 2007).

The analysis of the spatial adjustment was monthly performed in each plot through interpolative geostatistic maps, odinary runs, and fitting the beta-binomial distribution (Madden et al., 2007). To examine the spatial pattern using the frequencies to the beta-binomial distribution, the plots were divided in 25 quadrants with four trees in each one (2 furrows for every two trees).

The geostatistic maps were done considering the position of each tree (presence or absence of disease) in the plot with the program Surfer® 10 (Golden Software Inc., Golden, Colorado, USA, 2011).

The analysis of ordinary runs, allowed to define the existence of aggregation among adjacent symptomatic trees inside each row and between rows. A run (U) is described as a succession of one or more symptomatic or non-symptomatic trees. The expected number of runs E(U) under the null hypothesis of randomness, is given for E(U) = 1 + [2m (N-m)/N], where “m” is the number of trees with symptoms and “N” is the total numbers of trees per row or between rows. The standard deviation of U, under the null hypothesis, is given for S(U) = {[(2m(N-m)) (2m(N-m)-N)]/(N2(N-1)}. For determining the significance of the symptomatic trees aggregated, a normal test of Z was used, where Z(U) = {[(U+0.5) - E(U)]/S(U)}, when the values of Z are lower than -1.67 (p = 0.05), indicate a rejection of the null hypothesis (random arrangement) and the alternate hypothesis is accepted (aggregated arrangement) (Kranz, 1993). With the results obtained, the percentage of rows that showed aggregation between symptomatic trees inside each plot, was analyzed.

To examine the presence of aggregation in the quadrants, the data of the symptomatic trees frequency in the quadrants were used to adjust the beta-binomial distribution with the program BBD (Madden and Hughes, 1994). The analysis provides the standard normal value (Z) and the respective probability (P) for the test C(α), associated to the beta-binomial distribution, where the values Z ≥ 1.64 o P(Z) ≤ 0.05, indicate that the arrangement of symptomatic trees follows the beta-binomial distribution (aggregated), while values of Z < 1.64 o P(Z) > 0.05, indicate that the arrangement of the symptomatic trees follows the bbinomial distribution (random).

Results and Discussion

The analysis of the progress curves, the incidence of HLB in the four orchards varied from 19 to 32.4 % in the months (Figure 1). In general terms, the dissemination of the disease has been fast and presents a pattern of dispersion similar to the one reported in Brazil (Gottwald et al., 2007) and Florida, USA (Gottwald et al., 2010), where the incidence of HLB increased from 6 to 27 % in 9 to 10 months and from 2 to 39 % in 10 months, respectively.

[Figure ID: f1] Figure 1.

Progress curves of the incidence of huanglongbing in four Persian lime orchards in Xalisco, Nayarit, Mexico.

The progress of the disease was adequately described by the Weibull model with a coefficient of determination (R2 ) from 0.84 to 0.95 % (Table 2). On the analysis of epidemics, it was verified that there was a marked increase of the rate in orchard ED-2, compared to the other orchards (Table 2). That difference can be attributed to the fact that a proportion of initial major incidence was found in that orchard. This hypothesis is backed up because the source of the inoculum plays a major role on the progress rate of the disease (Mora-Aguilera et al., 2014a). Also, in this orchard the trees were relatively young (younger than three years old) and had constant vegetal growth, which created favorable conditions for the presence of high population of adults and nymphs of the ACP (Gottwald et al., 2007).

Table 2.

Parameters estimated by fitting incidence of HLB with Weibull distribution model, in four citrus orchards in Xalisco, Nayarit, Mexico from September 2010 to February 2011.

Orchard (epidemic) Weibull modela AUDPC y0 yf
y=1-(t/b)c c b1 R2
EM-1 y=1-(t/962.8)0.53 0.53 0.001 0.84 10.92 0.04 0.12
ED-2 y=1-(t/90.93)1.0S 1.08 0.011 0.89 27.05 0.08 0.32
HM-3 y=1-(t/305.9)0.62 0.62 0.003 0.95 16.41 0.04 0.17
IP-4 y=1-(t/13439.4)0.45 0.45 0.00007 0.84 4.36 0.01 0.04

TFN2a The model parameters were estimated by nonlinear regression with model Weibull equation y=1-(t/b)c , where c is the curve shape and b is the epidemic rate parameter, y is the disease measured as the incidence of diseased trees, y t is the time in months. y0 is the proportion of initial incidence y y f is the proportion of final incidence of HLB in Persian lime orchards (Citrus latifolia) evaluated in Xalisco, Nayarit, Mexico.

While mapping the contour graphics it was possible to distinguish the possible patterns of spatial arrangement of the disease only in the first evaluations until the fourth month (Figure 2). In the orchards IP-4 and HM-3 the tree foci with symptoms were more evident inside the rows (among the trees), while as for the orchards EM-1 and ED-2 between rows (Figure 2). The mapping of the orchards provided a quick visualization of the arrangement of the tress with and without symptoms of HLB, being considered by this as the first element of an analysis.

[Figure ID: f2] Figure 2.

Geoestatistical maps of four orchards of Persian lime showing trees with symptoms of HLB, evaluated during September 2010 to February 2011, in Xalisco, Nayarit, Mexico. Dark areas indicate diseased trees.

Through the analysis of ordinary runs, in the orchards EM-1, ED-2, and HM-3 the percentage of rows showing aggregation symptomatic trees inside the rows was elevated overtime, seeing that in the last evaluation (sixth month), 13, 32 and 13 % of rows with aggregation were registered, respectively (Table 3). In the orchards IP-4, the percentage was maintained from the third month (7.7 %).

Table 3.

Spatial arrangement of HLB in four plots of Persian lime, in Xalisco, Nayarit, Mexico, analyzed by the techniques of “ordinary runs” and fitting the beta-binomial distribution [test C(α)].

Orchard Evaluation Ordinary runsb
Aggregated lines
Test C(α)c
Z P(Z)
EM-1 Sep 7.7 nad na
Oct 7.7 na na
Nov 7.7 na na
Dec 11.3 2.320 0.029
Jan 13.0 2.620 0.014
Feb 13.0 2.620 0.014
ED-2 Sep 7.7 1.67 0.106
Oct 9.5 1.65 0.112
Nov 18.0 2.69 0.012
Dec 21.1 2.96 0.006
Jan 26.9 3.65 0.001
Feb 32.0 3.84 0.0007
HM-3 Sep 3.9 0.920 0.368
Oct 5.8 1.320 0.198
Nov 9.5 2.000 0.056
Dec 9.5 2.000 0.056
Jan 11.3 2.320 0.029
Feb 13.0 2.620 0.014
IP-4 Sep 3.9 1.430 0.166
Oct 3.9 1.430 0.166
Nov 7.7 na na
Dec 7.7 na na
Jan 7.7 na na
Feb 7.7 na na

TFN3a Evaluation period. bCalculated according Kranz (1993), considering 10 trees/line and 10 lines per orchard. Percent of symptomatic trees lines with aggregation. cNormal distribution pattern (Z) of the test C(α) and associated probability [P(Z)]. Values Z>1.64 or P(Z)<0.05 indicate rejection of H0; arrangement follows the binomial distribution (random) and H1; arrangement follows the beta-binomial distribution (aggregate). Estimated values with the help of the BBD program (Madden and Hughes 1994). dNot applicable.

By the fitting to the beta-binomial distribution, and considering quadrants with four trees, in the orchards IP-4 all the evaluations evidenced a random arrangement of trees with HLB symptoms, while in the other orchards a pattern in the form of aggregated was evident from the third one (ED-2 and HM-3) and fourth one (EM-1) evaluation (Table 3). On the other hand, it was verified that, on low levels of incidence of the disease, as reported in Shantou, China (Gottwald et al., 1991), a random arrangement of diseased trees predominated, because the results obtained from Z were not significant (p>0.05) for the test C(α), giving that there was no fit on the beta-binomial (Madden and Hughes, 1994). According to Bassanezi et al., (2005), the spatial arrangement of trees with HLB are influenced by the interaction of several factors, mainly by the source of available inoculum and the vector populations.

The predominance of the aggregated arrangement indicates that the infection of the trees with HLB could be originated from a close source of inoculum or from one inside the orchard. This most probable hypothesis is that the infections mainly occurred through the primary inoculum transmitted by the Asian psyllids of the citrus, coming from the near infected orchards with HLB. On the other hand, the secondary transmission from tree to tree by the ACP inside the orchards had an influence on the epidemic, mainly on the orchard ED-2, where the aggregation of diseased trees was detected in great proportions. Studies performed in China, Brazil, and Florida (USA), also corroborated that the ACP showed an aggregated distribution on the fields (Gottwald et al., 1991; Bassanezi et al., 2005; Gottwald et al., 2010). This pattern of dispersion is associated with diseases caused by systemic pathogens in which the vectors are the main way of mode of dispersion (Mora-Aguilera et al., 2014b).

The spatial relationship between diseased trees, when is combined with biological knowledge about the pathosystem, helps with the comprehension of the temporal process of the diseas and the factors that in-fluence the dispersion on the field (Mora-Aguilera et al., 2014b). The spatial mapping indicated that the occurrence of secondary dispersion of HLB in the orchards, with the participation of the ACP as vector, up to 40 meters long, a behavior similar to the one found in Brazil (Bassanezi et al., 2005).

The establishment of area-wide management of ACP (ARCO), a current strategy applied in Mexico (SENASICA, 2012, 2016), is a viable alternative that can drastically reduce the source of inoculum and the vector insect population (Flores-Sánchez et al., 2017). The use of the ARCOs can be a way to minimize the incidence of HLB and to modify the pattern of dispersion of the insect, decreasing the aggregation of the pathogen by its transmission.


The rate of incidence of the trees with HLB symptoms in the orchards evaluated in the municipality of Xalisco, Nayarit, Mexico, increased up to 32.4 % in a six-month period (September, 2010 - February, 2011). The trees between two and five years old were the most susceptible. The spatial studies of HLB in Mexico at parcel level has been restricted to the use of distribution maps of diseased trees, with no quantitative analysis. This research confirms this aspect and demonstrates that the spatial arrangement of trees of Persian lime with HLB symptoms is aggregated predominant.

fn1Cite this paper: Márquez-Pérez, F. J., Flores-Sánchez, J. L., Rodríguez Mejía, L., Márquez Gómez, J., Michereff, S. J., Ancona, V., Robles-Bermúdez, A., Domínguez-Monge, S. (2018). Progress and spatial pattern of huanglongbing in Persian lime in Nayarit, Mexico. Revista Bio Ciencias 5(nesp), e351. doi:


Enlaces refback

  • No hay ningún enlace refback.

Revista Bio Ciencias, Año 13, vol. 9,  Enero 2022. Sistema de Publicación Continua editada por la Universidad Autónoma de Nayarit. Ciudad de la Cultura “Amado Nervo”,  Col. Centro,  C.P.: 63000, Tepic, Nayarit, México. Teléfono: (01) 311 211 8800, ext. 8922. E-mail:,, Editor responsable: Dra. Karina J. G. Díaz Resendiz. No. de Reserva de derechos al uso exclusivo 04-2010-101509412600-203, ISSN 2007-3380, ambos otorgados por el Instituto Nacional de Derechos de Autor. Responsable de la última actualización de este número Dra. Karina J. G. Díaz Resendiz Secretaria de Investigación y Posgrado, edificio Centro Multidisciplinario de Investigación Científica (CEMIC) 03 de la Universidad Autónoma de Nayarit. La opinión expresada en los artículos firmados es responsabilidad del autor. Se autoriza la reproducción total o parcial de los contenidos e imágenes, siempre y cuando se cite la fuente y no sea con fines de lucro.

Licencia Creative Commons
Revista Bio Ciencias por Universidad Autónoma de Nayarit se encuentra bajo una licencia de Creative Commons Reconocimiento-NoComercial-SinObraDerivada 4.0 Internacional

Fecha de última actualización 9 de Junio de 2022


licencia de Creative Commons Reconocimiento-NoComercial-SinObraDerivada 4.0 Internacional


Dra. Karina J. G. Díaz Resendiz