Research Article
Research Article
Probability of a Central American locust Schistocerca piceifrons piceifrons upsurge in the Yucatan Peninsula, Mexico
expand article infoMario A. Poot-Pech§
‡ Yucatan State Plant Health, Yucatan, Mexico
§ Grasshopper-Locust Management in Central America, Yucatan, Mexico
Open Access


From ancient times to the present, infestations of the Central American locust (CAL) [Schistocerca piceifrons piceifrons (Walker, 1870)] have occurred periodically and with varying intensities in the Yucatan Peninsula (YP), Mexico. Despite efforts to survey the recession zone, an upsurge is still difficult to predict and prevent, and high economic costs are incurred in controlling this pest. For this study, two models were developed to determine the probability of an upsurge in the YP. The first was the Markov chain (MC) with transition probability matrix, which estimates probability by determining the proportion of times that the system moved from one state to another (n2) over 71, 33, and 24 years in Yucatan, Campeche, and the Quintana Roo States, respectively, divided into different periods; a correlation of the matrix and probability (n2) of the next period was performed to evaluate the accuracy of the estimation. The other method is the classic probabilistic (CP) model, which uses the number of times the upsurge could happen and the number of possible events. In the MC model, great variation was found in CAL upsurge probabilities between periods, with a similar number of upsurges from the past to the present but with varying intensity. In recent years, the treated area with insecticides has been less than that of the past. The CP model revealed that the locust population reached its maximum peak every four years, with the migration of swarms to neighboring states at the end/start of the year. Validation of the MC and CP models was performed considering information on areas treated in 2019 and 2020, and good accuracy was obtained. Both models provide information on the probability of an upsurge in the YP. This information can be incorporated into economic models to improve management decisions, such as when to announce early warnings, and to implement preventive control strategies.


early warning, forecast, preventive management, recession period


Locusts are among the most devastating pests of human agriculture (Lockwood 2015), making them a major threat to global food security and the livelihoods of farmers in numerous countries (Simpson and Sword 2008). Schistocerca piceifrons (Walker, 1870) (Orthoptera: Acrididae) is the true locust of tropical America (Harvey 1983). This species has two subspecies: the Central American locust (CAL; S. piceifrons piceifrons), known in Mexico and Central America, and the Peruvian locust (S. piceifrons peruviana), distributed in Peru and Ecuador (Harvey 1983, Barrientos et al. 1992).

CAL damage has been documented from the Mayan culture and colonial period, which reported drought and hunger as results of this pest (Flores 2011, García 2012), to the present (SENASICA-DGSV 2016). According to Trujillo (1975), a CAL swarm can feed on 30 tons of vegetation per day and, based on its food preference and distribution in 10 Mexican states, the species could affect 5.9 million ha (SENASICA-DGSV 2016).

An important CAL breeding zone is located in the Yucatan Peninsula (YP), Mexico (SENASICA-DGSV 2016), which possesses ecological conditions–such as mainly grass vegetation–for high-density locust development (Poot-Pech et al. 2018). According to SENASICA (2019), the YP receives 50% of the national budget for locust control because of the importance of breeding zones. The YP is located in the southeast of Mexico. It is made up of three states–Yucatan, Campeche, and Quintana Roo–that comprise 181,000 km2 (9.2% of the national territory). The overall climate is tropical, with strong seasonality and dry (November–May) and rainy (June–October) seasons (Barber et al. 2001). Tropical forest is the dominant vegetation, but grassland is increasing (Durán and García 2010). The soil is heterogeneous and mostly stony (Bautista et al. 2011).

Currently, CAL in Mexico are controlled through government regulations and advice from the International Regional Organization for Plant and Animal Health (OIRSA), an intergovernmental organization founded in 1953. OIRSA was created as a cooperative effort of locust control between Central American countries that are part of the insect’s migration path (Barrientos et al. 1992).

Locust plagues occur after a series of events that increase the number of locusts. The series normally begins with a calm period of recession followed by localized outbreaks and upsurges from which a plague may develop and eventually decline (Symmons and Cressman 2001). Locusts survive the dry season as adults and oviposit in bare areas. Young nymphs hatch and feed on adjacent short ephemeral grasses, while older nymphs and young adults survive on taller vegetation (Hunter and Spurgin 1999). An outbreak occurs when multiplication causes a marked increase in locust numbers and densities so that individuals become gregarious and, unless checked, hopper bands and/or swarms are formed (Van Huis et al. 2007). The gregarization process takes several months (Symmons and Cressman 2001). Upsurges are periods after an outbreak that are followed by two or more successive generations of transient-to-gregarious breeding, with populations occupying an expanding breeding area (Van Huis et al. 2007). Periods of one or more years during which upsurges are widespread are considered plagued years (Symmons and Cressman 2001).

The CAL is an old pest in Mexico (Contreras and Galindo 2014) that is still present in breeding areas today. To analyze and forecast the distribution of CAL, Hernández-Zul et al. (2013) built a dynamic simulation to identify suitable conditions for high CAL reproduction rates. Galindo et al. (2014) proposed an early warning system that used multicriteria models and satellite images. Aldama et al. (2014) suggested that outbreak characterization of endemic channels could help in forecasting short-term locust thresholds in zones where outbreaks can occur. The climate, El-Niño Southern Oscillation (ENSO) (Retana 2000, Contreras and Galindo 2014), vegetation, locust density, and predators are additional indicators that could be valuable in predicting a locust outbreak (Henschel 2015).

The strategies for locust control have has recently changed, with attempts focusing on prevention with the use of forecasting tools. Methods for producing Short- and medium-term forecasts have been made indicating potential locust migration and breeding areas (Cressman 1996). Zhang et al. (2019) proposed including monitoring and forecasting as part of current preventive locust management. Presently, biomodels and geographic information systems (GIS) are used to support the monitoring and forecasting of CAL outbreaks (Barrientos et al. 2021), but efficient statistical tools, with or without historical information, are important for the early detection of rapid changes in the breeding and migration of pest species that lead to outbreaks (Naumova and MacNeill 2005). Bayesian statistics and Markov models offer an effective method for long-term population forecasting of pests and can provide locust-control agencies and organizations with the information necessary to implement appropriate management strategies (Xiang-Hui et al. 2010). Knowledge of indirect effects is also a valuable tool for predicting the probability of CAL upsurge, as it is important to be aware of the relationship between areas treated with any pesticide and locust evolution (outbreak, upsurge, or pest) (Prospatin 2012). Van Huis et al. (2007) described the treated areas with any control agent as an answer to the function of locust area and time.

The objective of this research was to develop and compare two models that can produce probability estimates for future upsurges in the YP by analyzing non-weather-related historical CAL data.

Materials and methods

Data acquisition.—To document the occurrence of CAL upsurges, the following sources of historical information on locust were used: Márquez (1963), Trujillo (1975), Pereyra (1991), Chi (2000), newspapers, interviews with locust field officers, and information provided by the National Service of Health, Safety and Agrifood Quality (SENASICA) on the locust control campaign in the YP, Mexico, through its informatics system called the Phytosanitary Campaign System (SICAFI;

For Yucatan state, CAL data for 71 years (1948–2018) were available (every year had a value). Data for 33 years (1986–2018) were available for Campeche, and 24 years of data (1995–2018) were available for Quintana Roo.

Locust upsurge definition.— Upsurges were used instead of outbreaks because there were two high-density generations–transients-to-gregarious or gregarious-to-gregarious–and migration to other regions (Van Huis et al. 2007); when swarms move beyond breeding areas to other regions, the overall situation exceeds a hazily demarcated threshold from “outbreak” to “upsurge” (Showler 2019). The CAL was present in Yucatan State in a gregarious form in almost years which is reflected in larger treated surface with insecticides (Fig. 1).Upsurges, therefore, allow consideration of the large number of migrant swarms that formed and reached the state of Campeche. As a result of such upsurges, the treated area required for CAL control was high in one year as in 2004, 2006, 2009, 2010, 2014, and 2018 or consecutive years 1952–1955, 1979–1981 and 1986–1989 on gregarious populations (Fig. 1) that reached Campeche during migration. The reaction in this state was the development of highly controlled surface in 2007, 2011, 2015, and 2019 because subsequent swarms (451 in 2006–2007, 245 in 2010–2011, 243 in 2014–2015, and 120 in 2018–2019) migrated from Yucatan (Fig. 2). In contrast, in years without an upsurge, the number of swarms detected in Yucatan was at a minimum (0–60 swarms) and failed to reach Campeche.

Fig. 1. 

Treated area in Yucatan State from 2003–2018 (above) and 1952–1955/1977–1989 (below).

Fig. 2. 

Swarms detected and treated in the Yucatan Peninsula from November–December (2006, 2010, 2014, and 2018) and January–April (2007, 2011, 2015, and 2019).

Markov chain probabilistic model.—The data were organized by assigning “0” to years with none or low intensity of CAL swarms and “1” to those in which an upsurge occurred. Second, the data were organized into periods of upsurge based on number of years, resulting in 5 periods of 12 years and one of 11 years in Yucatan state; 3 periods of 12 years, except for the first period of 9 years, in Campeche; and 2 periods of 12 years in Quintana Roo.

Classic probabilistic model.—The maximum peak years were obtained using the treated area (ha) in Yucatan state from 2003–2018. Subsequently, a consecutive number was assigned to each of the following years, finalizing in the next peak. A probabilistic analysis of occurrence was performed for each consecutive year. The probability of occurrence in the most recent years (2003–2018) was compared with that of previous periods [1977–1989 (Pereyra 1991) and 1952–1955 (Márquez 1963)].

Model validation.—The probabilistic results of the two models were compared to the results of the treated area in the YP in 2019 and 2020.

Data analysis

Markov chain probabilistic model.—The probability of matrix transition for every period was estimated by determining the proportion of times that the CAL upsurge situation, 0 and 1, moved from one state to another (Kemp 1987). In this context, there were four possible transitions: no upsurge → no upsurge (0→0), no upsurge → upsurge (0→1), upsurge → no upsurge (1→0), and upsurge → upsurge (1→1) (modified from Zimmerman et al. 2004). These estimates were represented by a p matrix developed for each year, where


Table 1.

Matrix (one-step transition) and n2, for CAL upsurges in a 71-year period.

Number Years Yucatan Number of years Upsurge years Upsurge average Probability n2
Matrix (One-step transition)
0→0 0→1 1→0 1→1 0→0 0→1 1→0 1→1
1 1948–1959 0.5 0.5 0.43 0.57 12 7 1.71 0.46 0.54 0.46 0.53
2 1960–1971 0.5 0.5 0.6 0.4 12 5 2.40 0.55 0.45 0.54 0.46
3 1972–1983 0.71 0.29 0.75 0.25 12 4 3.00 0.72 0.28 0.72 0.28
4 1984–1995 0.2 0.8 0.67 0.33 12 6 2.00 0.57 0.43 0.35 0.65
5 1996–2007 0 1 0.72 0.28 12 7 1.71 0.72 0.28 0.2 0.8
6 2008–2018 0.58 0.42 0.66 0.34 11 4 2.75 0.61 0.39 0.60 0.4
71 (total) 33 (total) 2.15 (mean)

The elements P00 and P01 refer to the probability that in X year (present), the following year’s CAL population will be classified as low (P00) or high (P01) if X year has low densities. Alternately, if X year has a high locust density, P10 is the probability that the following year’s populations will be classified as having a low density, and P11 is the probability that the state will continue to be classified as having a high density of CAL the next year (Kemp 1987).

The probability of correctly determining the locust density in the next period was obtained using the recursive properties of a two-state Markov chain (Bhat 1972). The matrix P (n–1) was multiplied by the matrix P to generate the matrix p(n). The element-by-element calculations for P(n) were as follows:

P 00 ( n ) = P00(n-1)P00 + P01(n-1)P10;

P 01 (n) = P00(n-1)P01+ P01(n-1)P11;

P 10 (n)= P10(n-1)P00 + P11(n-1)P10;

P 11 (n) = P10(n-1)P01 + P11(n-1)P11

The correlation between the probability n2 of the two-state Markov chain and the probability of the next period (one-step transition)–for example, the probability n2 of period 1 with the matrix of period 2–was determined using Pearson correlation (p < 0.05) in R software version 3.6.0 (R Core Team 2019).

Classic probabilistic model.—We obtained the classic probability using P(1) = #1/M, where P(1) is the probability of an upsurge, #1 is the number of times the upsurge can happen, and #M is the number of possible events (Infante and Zárate 1990). Upsurge years were designated as 1, with years of no upsurge designated as 0.


Markov chain probabilistic model

Yucatan state.—A great deal of variation was found in CAL upsurge probabilities between periods, except in transitions 0→0 and 0→1 of periods 1 and 2. In the first period, transition 1→1 (0.57) was highest, indicating years with contiguous upsurges. The second period was characterized by a reduction in upsurge frequency of 1→1 (0.4). In the third period, this transition had its biggest reduction (0.25), indicating an increase in “recession years,” and the transition 0→0 was increased to 0.5 to 0.71 . During periods 4 and 5, the values of the transition matrix were very similar (0→0: 0–0.2, 0→1: 0.8–1, 1→0: 0.67–0.72, and 1→1: 0.28–0.33). Thus, an upsurge appeared in one year, and in the next, it was reduced. The last period was very similar to the second period, where the recession years were increasing 0→0 (0.58) and 1→0 (0.66) and remained high. From 1948 to 2018, there were 33 upsurges at an average of 2.15 per period and with a range of 4 to 7. Periods 1 and 5 had a major upsurge, and periods 3 and 6 had a minor locust presence.

In 4 of 5 cases, the correlation between the matrix and probability n2 in the next period was negative, and the P-value was not statistically significant (P > 0.05). In the period 1972–1983, the correlation was high, positive, and statistically significant (P ≤ 0.01).

Table 2.

Pearson correlation of the matrix and probability n2 in Yucatan state (P < 0.05).

Period Correlation P-value
1960–1971 -0.65 0.34
1972–1983 0.98 0.01
1984–1995 -0.26 0.73
1996–2007 -0.75 0.24
2008–2018 -0.64 0.35

Campeche State.—In the matrix, the three periods of CAL upsurge in Campeche State were different. In period 1, which was the shortest, 0→1 (1) and 1→1 (0.72) stood out. Periods 2 and 3 showed identical values–1→0 (1) and 1→1 (0)–and had similar 0→0 (0.57–0.75) and 0→1 values (0.43–0.25).

The number of upsurges per year decreased; the first period had the highest number of upsurges (7), followed by periods 2 (4) and 3 (3).

Table 3.

Matrix (one-step transition) and n2, for CAL upsurges in a 33-year period.

Number Period Campeche Years Upsurge years Upsurge overage Probability n2
Matrix (one-step transition)
0→0 0→1 1→0 1→1 0→0 0→1 1→0 1→1
1 1986–1994 0 1 0.28 0.72 9 7 1.29 0.28 0.72 0.2 0.8
2 1995–2006 0.57 0.43 1 0 12 4 3 0.75 0.25 0.57 0.43
3 2007–2018 0.75 0.25 1 0 12 3 4 0.81 0.19 0.75 0.25
33 (total) 14 (total) 2.3 (mean)
Table 4.

Pearson correlation of the matrix and probability n2 in Campeche (P > 0.05).

Period Correlation P-value
1986–1994 -0.88 0.11
1995–2006 0.67 0.32

The Pearson correlation was different in the two periods. In 1986–1994, it was negative, and in 1995–2006, there was a positive association. However, the differences were not statistically significant (P > 0.05).

Quintana Roo State.—In the 24 years of data (1995–2018), only one year had a CAL upsurge: 2006. Therefore, the probability for 1→0 and 1→1, equal to 1 and n2 (Table 5), could not be obtained. According to data from the Quintana Roo Plant Health locust control from Yucatan migrated and caused damage to corn and bean crops and cultivated grasslands and caused defoliation in a nature reserve and on urban trees from August–September. In 2006, locust control operations were undertaken in 1,381 ha.

These swarms were able to oviposit in Quintana Roo and complete the second generation. They also returned to Yucatan at the end of 2006 as swarms. In that generation, 173 ha required control operations. In subsequent years, the locust population was present at a low density in the solitarious phase.

Table 5.

Matrix (one-step transition) for CAL upsurges over a 24-year period.

Period Quintana Roo Years Upsurge years Upsurge average
Matrix (one-step transition)
0→0 0→1 1→0 1→1
1995–2006 0.9 0.1 0 0 12 1 1
2007–2018 1 0 0 0 12 0 0
Total 24 1 1

Classic probabilistic model

Fig. 1 shows a tendency for upsurges to peak every four years (2006, 2010, 2014, and 2018), except for an additional two peaks in 2004 and 2009, from 2010 to 2018, this tendency was clearer than from 2003 to 2009.Therefore and consequently it was assigned values from years 1 to 4, starting with the maximum peak, e.g: year 2006 (year value 1), 2007 (2), 2008 (3), 2009 (4), 2010 (1), 2011 (2) so on (Table 6); from 1952–1955 and 1977–1989, the tendency of the maximum peaks was less clear, and the treated area was larger than in recent years.

This information was used to identify the upsurge years (Table 6). Then, the probability of an upsurge in the past and current periods was determined (Table 7).

Table 6.

Upsurge value for 2003–2018 in the YP and 1952–1955/1977–1989 in Yucatan state.

States Upsurge value per year
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Yucatan 0 1 0 1 0 0 1 1 0 0 0 1 0 0 0 1
Campeche 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0
Q. Roo 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0
Yucatan 1952 1953 1954 1955 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988
1 1 1 1 0 0 1 1 1 0 0 0 0 1 1 1
Year value 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4
Table 7.

Probability of an upsurge, P(1), from 1952–1955/1977–1988 and 2003–2018.

Year value 1952–1955 and 1977–1988 2003–2018
Yucatan Campeche Q. Roo
Upsurges P(1) Upsurges P(1) Upsurges P(1) Upsurges P(1)
1 1,0,1,0 0.5 0,0,0,0 0 1,1,1,1 1 0,0,0,0 0
2 1,0,0,1 0.5 1,0,0,0 0.25 0,0,0,0 0 0,0,0,0 0
3 1,1,0,1 0.75 0,1,0,0 0.25 0,0,0,0 0 0,0,0,0 0
4 1,1,0,1 0.75 1,1,1,1 1 0,0,0,0 0 1,0,0,0 0.25

For the period 2003–2018 in Yucatan, the highest probability of upsurge was in year 4 (P: 1). There was no probability of an upsurge (P: 0) in year 1 and a minimal probability (P: 0.25) in years 2 and 3. These results were different from the data from the past (1952–1955 and 1977–1988), where years 1 and 2 were similar (P: 0.5) and years 3 and 4 were similar (P: 0.75). For Campeche, the P(1) was 1 in year 1, and years 2–4 had no values. In Quintana Roo, there was only a remote probability of an upsurge, but the probability increased if Yucatan had an upsurge so intense (similar to 2006) that it resulted in locust invasion.

Validation.—According to SENASICA (Table 8), the area treated for locust control was 2,224 ha in 2019 and 3,714 ha in 2020. This information was used to validate the model.

Table 8.

Treated area (ha) in the Yucatan Peninsula in 2019 and 2020.

State Years
2019 2020
Yucatan 813 3,420
Campeche 1411 294
Quintana Roo 0 0
Total 2224 3714

Validation of Markov chain probabilistic model

In Yucatan state starting in 2018, which had a value of 1, the transition 1→0 (one step: 0.66, n2:0.6) had the highest value. Therefore, the following year, 2019, an upsurge value of 0 would be expected and starting in 2019 as upsurge value 0, the transition (0→0: 0.58, 0.61) was the highest value, so it would also be expected an upsurge value 0 in 2020, both results obtained of the probabilistic model correspond to the results of the low treated area (Table 8). In Campeche in 2018, value = 0, the transition 0→0 (0.75; 0.81) had the highest value, a year with value 0 was expected, but it was not fulfilled, in 2019 according to the treated area the value was 1. The value 1→0 (1, 0.75) was highest for 2020 and it was the upsurge value, through the treated surface (Table 8). In Quintana Roo, despite good results in the probability of 0→0 for 2019 and 2020, no application was made in transition n2 with an origin of 1 (1→1 or 1→0) because it is an invasion area and not a locust breeding zone.

Validation of Classic probabilistic model

The comparison of the probability of an upsurge P(1), with previous information 2003–2018, with values 2019–2020 indicate that in Yucatan it was according to the model in 2019 and very close to 2020 (P(1)=0.25, value=0), in Campeche and Quintana Roo as predicted.


Markov Chain Model

Yucatan State.—The CAL upsurge from 1948 to 2018 was a sign that the frequency of occurrence had changed, as there were no equal periods. The transition 1→1 had the highest value in the first period (1948–1959); therefore, it was called a “plague period” in which “favorable breeding conditions are present and control operations fail… two or more regions are affected simultaneously” (Symmons and Cressman 2001). Upsurge had a higher value in the past than in recent years because there was no organization in charge of a permanent locust control program. Control measures were more recently undertaken over gregarious populations, mainly swarms (Marquez 1963). From 1941–1942, CAL swarms invaded Mexico and Central America,originating from some area of this same region. In 1947, the International Committee for Locust Control (CICLA) was formed; in 1955, the name was changed to OIRSA to include locust surveillance (Marquez 1963, Trujillo 1975). In 1948, the Mexican General Direction of Agricultural Protection was created, an institution that consolidated technical cooperation with CICLA. The first period (1948–1959) in Mexico was characterized by the creation of national and international institutions for locust management (Ortíz and Zuleta 2020).

In the second period (1960–1971), the transition 1→1 decreased, perhaps as an effect of more organized locust control and unfavorable weather conditions, and the number of upsurges per period was reduced from 7 to 5. In period 3, 1972–1983, the transition 0→0 was the highest (0.71). Consequently, because of the fewer number of upsurges (4), the locust was declared to be in its recession period, i.e., a period of several years when the locust population is low (Symmons and Cressman 2001).

In periods 4 (1984–1995) and 5 (1996–2007), the values of the transition matrix were very similar, the transition 0→1 and 1→0 were high, that is the upsurges occurred approximately every 2 years. When starting with a solitarious population, CAL needs three generations to reach the gregarious phase (Barrientos et al. 1992). The CAL presents two generations per year (Astacio 1966, Trujillo 1975); therefore, a third generation is achieved within two years, which concurs with the calculations of Marquez (1963), who indicated that “the locust does not form suddenly; it takes two or more years to develop.” In periods 4 and 5, transition 1→1 was reduced because locust management was strengthened with the creation of state-run locust control programs. For example, in 1988, the Ministry of Agriculture and Water Resources of Mexico granted greater autonomy to phytosanitary programs in each state, including the locust control program. The Federal Plant Health Law (Ley Federal de Sanidad Vegetal 1994) created plant health committees to develop multiple crop protection programs with government funding. These programs encompassed the locust control campaign. In Yucatan, such a committee was created in 1997 and is currently responsible for the permanent monitoring and control of this pest.

Although Yucatan periods 1 and 5 had similar upsurge years (7), the severity differed. The overall area treated with insecticides in period 1 (1952–1955) was 56,000 ha (Márquez 1963), while it was 26,363 ha in period 5 (2003–2006), which was 47% less than in the first period.

Only 1 out of 5 correlations between the probability n2 and the matrix was positive and significant. This may be because the outcome of the Markov chain depends on the outcome of previous events, meaning that the next state of the system depends on the present state, and locust outbreaks are erratic events (Symmons 1992). Showler (2002) outlined broad strategic approaches to locust outbreaks based on intervention timing (prevention or reaction) and tactics (survey, control, economic effects, insecure areas, and contingency planning). Management, weather, climate change [World Meteorological Organization-Food and Agriculture Organization (WMO-FAO) 2016], and political decisions differ in different periods, making it difficult to obtain reliable probability.

Markov models have limitations, are problematic with short time intervals, cannot be derived rigorously from deterministic, dynamic models, and rarely provide the range of time for which the model is appropriate (DelSole 2000). However, they permit knowledge of the outbreak population in each period without the necessity of searching for causation through correlative procedures (Kemp 1987). Weather conditions that could have an influence on outbreaks include the amount of rain in the year (Steinbauer 2011), temperature, relative humidity (Al-Ajlan 2007), wind for migration (WMO-FAO 2016), ENSO (Contreras and Galindo 2014), and precipitation in previous years (Skinner and Child 2000, Chiconela et al. 2003).

In some species of Orthoptera, there has been a decrease in outbreaks; for example, outbreaks of rangeland grasshoppers in Wyoming are highly erratic events, with instances of infestations persisting for multiple years being quite low, so there is little basis for prorating the benefits of control beyond the year of treatment (Zimmerman et al. 2004).

Campeche state.—Historically, the Campeche state has been invaded by swarms from Yucatan (Márquez 1963). The entry and migration routes are found in the north, but these swarms can quickly move to the south of the state (Chi 2000). This movement is made by the second generation and at the beginning of the year (Cullen et al. 2017). In the second period (1995–2006), there was a reduction in the number of upsurges per year, probably because the locust campaign of Yucatan Plant Health established control operations in 1997, reducing the possibility of migration to Campeche. Before this period, the largest number of upsurges occurred, with the transition 1→1 indicating several continuous years with an upsurge or a reduced recession period (0→0). In the third period, there was an increase in the recession period 0→0 because the upsurge was dependent on outbreak and migration from Yucatan every 4 years (2006, 2010, and 2014). The probability of upsurge in 2019 was expected because of the migration of swarms from Yucatan in 2018.

With the information obtained on the control and migration of swarms, it was possible to construct the migration route of CAL to Campeche. At the beginning of the year, Campeche is an invasion zone (Fig. 2) for swarms originating from places in Yucatan, where it is difficult to undertake survey and control operations. Therefore, it is important to carry out locust management tasks in Yucatan state with new methods to reduce migration to Campeche. Despite the large number of swarms formed in Yucatan in the upsurge year, the damage was minimal for two reasons: a) in the maximum period of migration that occurs at the end-beginning of the year, there is less cultivated area, as most of the agriculture occurs in June–September when there is a greater amount of precipitation to help rainfed crops, and b) the size of the swarms, according to the classification of Cressman (2001), corresponds to very small swarms (<1 km2) due to preventive management of first stage nymphs.

Quintana Roo State.—Vegetation is very important for locust development (Sword et al. 2010). Its distribution (Despland et al. 2000), coverage, and status (green or dry) influence locust gregarization (Cisse et al. 2013). Models could be enhanced by using associations between plant communities and insects to predict risk areas (Van Der Werf et al. 2005). The CAL is highly associated with the grass Panicum maximum (Poot-Pech et al. 2018); however, in Quintana Roo, the vegetation is 90% tropical forest, 3.3% agriculture, 3.2% grasses, and 3.5% other uses (INEGI 2011). According to survey information from the locust campaign, P. maximum is not the dominant grass in Quintana Roo.

Classic probabilistic model

In Yucatan in 1952–1955 and 1977–1988, there were 10 upsurges and 6 recession years, while in 2003–2018, there were 6 upsurges and 10 recession years. This may be because the structure of the locust program was modernized during the latter period, with greater autonomy and economic resources for developing the program and prevention strategies (Rodríguez 2000). The locust program in Mexico receives technical support from OIRSA (Ortíz and Zuleta 2020), and in general, biological and ecological knowledge about locusts has increased, which has helped create better management practices for this pest (Lecoq 2021).

Periods of 4 years (Table 7) were formed, and a linear regression of areas treated for CAL was obtained (Figs 3, 4), finding a positive relationship. Generally, the first year had the least amount of treated area, and the fourth year had the most amount of treated area. In the fourth year, there was a high migration of swarms to Campeche (Fig. 2), so the gregarious area remained at low infestation for three generations. Two years later, the population became high once again (Márquez 1963, Barrientos et al. 1992). However, this pattern occurs when the swarms are controlled along their migration route; if the swarms are not controlled, then year one would have a high infestation when the swarms oviposit and return in the next generation, flying in the opposite direction (Cullen et al. 2017) and repeating the pattern presented in previous periods (1952–1955 and 1977–1988; Tables 6, 7) with at least two years of high infestation. This has been seen recently, such as in the upsurge in 2019 and 2020 of the desert locust in Pakistan (Sultana et al. 2021) and Africa (Peng et al. 2020, Salih et al. 2020).

Fig. 3. 

Linear regression of the treated area in periods of 4 years from 2003–2018 in Yucatán (SENASICA information).

Fig. 4. 

Linear regression of the treated area in periods of 4 years from 1952–1955 (Marquez 1963) and 1997–1988 (Pereyra 1991) in Yucatán.

There were intermediate years–year 2 in 2004 and year 3 in 2009–which were likely the result of suitable weather conditions such as precipitation in the breeding zone. A marked increase in locust numbers on a local scale due to concentration, multiplication, and gregarization can lead to the formation of hopper bands and swarms (Roffey and Popov 1968).

In 1952–1955, there was a “plague” of CAL, requiring 4 years of intense control; gradually, the size of the controlled area was reduced. The opposite situation occurred in 1977–1980 and 2007–2010, with an intermediate rebound in 1979. This situation lasted until 1981, and the treated surface area was reduced until 1984. There was another plague from 1985–1989, with 1986 having the largest controlled area.

From 1952–1955, as shown in Fig. 4, there was a negative regression, indicating a plague of CAL (Symmons and Cressman 2001). Efforts were made to reduce the population each year, but control was aimed at the gregarious phase and mostly swarms (Márquez 1963). International cooperation among countries was still in development (Yam and Zuleta 2020). In 1977–1980, there was a positive regression; an increase in the population occurred in 1979–1980, but in 1981–1984, there was again a negative regression, with a rebound in 1981. Finally, in 1985–1989, the regression value was low because in 1986, there was a higher peak; however, in the period 1986–1987, there was a plague.

In his book, An Account of the Things of Yucatán, written in 1566, Diego de Landa (2003), bishop of Yucatan, wrote about the consequences of locusts and noted the 4-year recession period, saying, “the locust developed for 5 years, which left no green thing, it was so hungry that people fell dead on the roads ... with 4 good years after the locust population had improved somewhat.”

Table 9.

Results of Markov matrix in 2018 compared to results from 2019 and 2020 upsurge values.

State (value year 2018) Matrix Values Upsurge value
0→0 0→1 1→0 1→1 2019 2020
Yucatán One-step transition (2018=1) 0.58 0.42 0.66 0.34 0 0
n2 0.61 0.39 0.6 0.4
Campeche One-step transition (2018=0) 0.75 0.25 1 0 1 0
n2 0.81 0.19 0.75 0.25
Q. Roo One-step transition (2018=0) 1 0 0 0 0 0
Table 10.

Probability of an upsurge P (1) and results of upsurge values 2019 and 2020 in the YP.

Year value Years Yucatan Campeche Q. Roo
P (1) Value P (1) Value P (1) Value
1 2019 0 0 1 1 0 0
2 2020 0.25 0 0 0 0 0
3 0.25 0 0
4 1 0 0.25

Infestation of the desert locust, S. gregaria, in Africa occurred in four out of five years between 1860 and 1963, and subsequently, in one year out of six (Symmons 1992). In India, from 1863–1962, there were 10 locust infestations, with 5–6 consecutive years of widespread reproduction, swarm production, and damage to crops, followed by 1–8 years of low activity; from 1963 to 2012, there were 18 upsurges (Sharma 2014).

The WMO-FAO (2016) listed the effects of the following factors on the development of a locust outbreak or upsurge: management; the failure to implement a preventive control strategy; inexperienced field survey teams and campaign organizers; insufficient or inappropriate resources; lack of training of the field officers; inaccessibility to important breeding areas; the weather (precipitation, temperature, and wind); and climate change. These factors are likely to have influenced the high occurrence of CAL in the intermediate years (years 2 or 3).

Additionally, limited financial capacity and ongoing armed conflicts could have rendered some of the locust breeding areas inaccessible, and the coronavirus pandemic lockdown has further hampered control efforts (Salih et al. 2020). Political and socio-economic conditions (Meynard et al. 2020) can also greatly influence the likelihood of locust outbreak by affecting management decisions (Gay et al. 2018). In period 2003–2018 there was an important change of government in 2007 in Yucatan State ( that included a change in locust-control decision-makers. A clear difference between the periods 2003- 2010 and 2011–2018 can be seen: Four upsurges were present in the first period, with only two in the second. In the second period, upsurges only occurred every four years as a result of the preventive actions and the incorporation of new technologies in the locust-control program (Barrientos et al. 2021).

Plagues arise when locusts breed frequently and successfully over a period of one or more years, with repeated and widespread rains in successive, often widely separated, seasonal breeding areas. This allows swarms to form and invade the agricultural zones surrounding the recession area. Pest control, drought, and migration to unsuitable areas can have an effect on ending plagues, although their relative importance is not always clear (Pedgley 1981). The uncertainty of pest outbreaks and associated damage (risk) can be reduced by preventive practices and by the selective use of pesticides based on monitoring and forecasting (Daamen and Rabbinge 1991). It is very important to predict and identify locust recession periods because these populations begin the invasion when ecological conditions become favorable (Lazar et al. 2016), and early prevention strategies have been proven to prevent damage to major agricultural zones in the invasion area (Sharma 2014). However, the response to an alert must be quick; otherwise, swarms will invade several regions, making the forecast pointless (Ceccato 2007).

Curiously, from 2003–2018, the largest treated area against CAL (Fig. 3) coincided with the years in which a soccer championship was held (2006 Germany, 2010 South Africa, 2014 Brazil and 2018 Russia), so in Yucatan these CAL upsurges are known as “soccer world cup locusts”.

Both probabilistic models are functional as long as the permanent monitoring system is sustained and allocated resources are maintained or increased. A reduction in budget would risk the development of the pest, as has occurred in the past. The results of the models discussed here provide insights into the probabilities of CAL upsurges in the YP, and this information can be incorporated into ecological models to improve CAL monitoring and aid management decisions.


I thank the field locust officers of the YP, SENASICA, and OIRSA. Special thanks to Mario Marin-Correa and Eudaldo Pereyra-Cuevas, retired YP locust officers who helped rebuild history. I also thank CONACyT (Consejo Nacional de Ciencia y Tecnología) and ITC (Instituto Tecnológico de Conkal) for their support of my PhD studies. Special thanks to the reviewer of this document for his insightful comments and corrections.


  • Al-Ajlan AM (2007) Relationship between desert locust Schistocerca gregaria (Forskal), infestation, environmental factors and control measures in Gazan and Makkah Regions, Saudi Arabia. Pakistan Journal Biological Sciences 10: 3507–3515.
  • Aldama AC, Hipólito CG, García MFJ, Mata CF (2014) Caracterización de los brotes de langosta Centroamericana (Schistocerca piceifrons piceifrons (Walker) en el Sureste de México. In: Galindo MMG, Contreras SC, Ibarra ZE (Eds) La plaga de la langosta, Schistocerca piceifrons piceifrons (Walker) una visión multidisciplinaria desde la perspectiva del desastre fitosanitario en México. Colección Sanidad Vegetal Tomo 1. Vigilancia Epidemiológica Fitosanitaria. Universidad Autónoma de San Luís Potosí, México, 160–172.
  • Astacio CO (1966) Observaciones Sobre la Biología, Ecología y Control del Chapulín (Schistocerca paranensis Burn) en Nicaragua. OIRSA, Nicaragua, 26 pp.
  • Barber A, Tun J, Crespo MB (2001) A new approach on the bioclimatology and potential vegetation of the Yucatan Peninsula (Mexico). Phytocoenología 31: 1–31.
  • Barrientos LL, Astacio CO, Álvarez BF, Poot MO (1992) Manual Técnico Sobre la Langosta Voladora (Schistocerca piceifrons piceifrons Walker, 1870) y Otros Acridoideos de Centroamérica y Sureste de México. FAO-OIRSA, El Salvador, 162 pp.
  • Barrientos LL, Song H, Rocha SAY, Torres CJA (2021) State of the art management of the Central American Locust Schistocerca piceifrons piceifrons (Walker, 1870). Agronomy 11: 1024.
  • Bautista F, Palacio-Aponte G, Quintana P, Zinck JA (2011) Spatial distribution and development of soils in tropical karst areas from the Peninsula of Yucatán, Mexico. Geomorphology 135: 308–321.
  • Bhat UN (1972) Elements for Applied Stochastic Process. Wiley, New York, 488 pp.
  • Ceccato P, Cressman K, Giannini A (2007) The desert locust upsurge in West Africa (2003–2005): Information on the desert locust early warning system and the prospects for season climate forecasting. International Journal of Pest Management 53: 7–13.
  • Chi QG (2000) Problemática Agrícola de la Langosta Schistocerca piceifrons (Walker) en el Estado de Campeche, Periodo 1995–1998. Memoria de Experiencia Profesional, Universidad Autónoma Chapingo, México, 76 pp.
  • Chiconela T, Chongo D, D´Uamba P, Ngazero A, Santos L (2003) Predicting the occurrence of red locust outbreak in Mozambique. African Crop Science Conference Proceedings 6: 224–230.
  • Cisse S, Ghaout S, Mazih A, Ould BEMA, Benahi AS, Piou C (2013) Effect of vegetation on density thresholds of adult desert locust gregarization from survey data in Mauritania. Entomologia Experimentalis et Applicata 149: 159–165.
  • Contreras SC, Galindo MMG (2014) La plaga de la Langosta Centroamericana Schistocerca piceifrons (Walker) en México: Reconstrucción ambiental 1592–2000. In: Galindo MMG, Contreras SC, Ibarra ZE (Eds) La plaga de la langosta, Schistocerca piceifrons piceifrons (Walker) una visión multidisciplinaria desde la perspectiva del desastre fitosanitario en México. Colección Sanidad Vegetal Tomo 1. Vigilancia Epidemiológica Fitosanitaria. Universidad Autónoma de San Luís Potosí, México, 88–110.
  • Cullen DA, Cease AJ, Latchininsky AV, Ayali A, Berry K, Buhl J, De Keyser R, Foquet B, Hadrich JC, Matheson T, Ott SR, Poot-Pech MA, Robinson BE, Smith JM, Song H, Sword GA, Vanden Broeck J, Verdonck R, Verlinden H, Rogers SM (2017) From molecules to management: Mechanisms and consequences of locust phase polyphenism. Advances of Insect Physiology. 53: 169–285.
  • Cressman K (2001) Desert locust Guidelines. 2 Survey. FAO, Rome, 33 pp.
  • de Landa D (2003) An account of the things of Yucatan: written by the bishop of Yucatan, based on the oral traditions of the ancient Mayas. Reprint. Monclem Ediciones. Mexico, 174 pp.
  • Durán GR, García CG (2010) Distribución espacial de la vegetación. In: Durán R, Méndez M (Eds) Biodiversidad y Desarrollo humano en Yucatán. CICY, PPD-FMAM, CONABIO, SEDUMA, México, 131–135.
  • Flores GF (2011) Las plagas de langosta en el área maya: Ambiente e historia de una calamidad en la época prehispánica. Península 2: 27–46.
  • Galindo MMG, Ibarra ZE, Hipólito CG, Aldama AC (2014) Sistema de alerta temprana para el control de la Langosta Centroamericana (Schistocerca piceifrons piceifrons (Walker) en la Península de Yucatán aplicando modelos multicriterio e imágenes de satélite NOAA_AVHRR. In: Galindo MMG, Contreras SC, Ibarra ZE (Eds) La plaga de la langosta, Schistocerca piceifrons piceifrons (Walker) una visión multidisciplinaria desde la perspectiva del desastre fitosanitario en México. Colección Sanidad Vegetal Tomo 1. Vigilancia Epidemiológica Fitosanitaria. Universidad Autónoma de San Luís Potosí, México, 173–215.
  • García QA (2012) La langosta, los mayas y el colonialismo en Yucatán, México, 1883. Relaciones 129: 215–249.
  • Gay P-E, Lecoq M, Piou C (2018) Improving preventive locust management: insights from a multi-agent model. Pest Management Science 74: 46–58.
  • Harvey AW (1983) Schistocerca piceifrons (Walker) (Orthoptera: Acrididae), the swarming locust of tropical America: a review. Bulletin of Entomological Research 73: 171–184.
  • Henschel JR (2015) Locust times – monitoring populations and outbreak controls in relation to Karoo natural capital. Transactions of the Royal Society of South Africa 2: 135–143.
  • Hernández-Zul MI, Quijano-Carranza JA, Yañez-López R, Ocampo-Velázquez RV, Tórres-Pacheco I, Guevara-González RG, Castro-Ramírez AE (2013) Dynamic simulation model of the Central American Locust Schistocerca piceifrons (Orthoptera: Acrididae). Florida Entomologist 4: 1274–1283.
  • Hunter DM, Spurgin PA (1999) Origin of outbreaks of Austracris guttulosa (Walker) in Australia and Nomadacris septemfasciata (Serville) in Southern Africa and its implications for their management. International Journal Tropical Insect Science 19: 307–311.
  • INEGI (2011) Uso de suelo y vegetación. Conjunto de datos vectoriales de la carta de uso de suelos y vegetación. Escala 1: 250, 000. Serie V. Quintana Roo, México, 1–200.
  • Infante GZ, Zárate LGP (1990) Métodos estadísticos. Un enfoque interdisciplinario. Editorial Trillas, México, 643 pp.
  • Kemp WP (1987) Probability of outbreak for rangeland grasshoppers (Orthoptera: Acrididaes) in Montana: Application of Markovian principles. Journal of Economic Entomology 6: 1100–1105.
  • Lazar M, Piou C, Doumandji-Mitiche B, Lecoq M (2016) Importance of solitarious desert locust population dynamics: lessons from historical survey data in Algeria. Entomologia Experimentalis et Applicata 3: 168–180.
  • Márquez DA (1963) La Lucha Contra la Langosta en México. Colegio de Ingenieros Agrónomos de México A. C, México, 220 pp.
  • Meynard C, Lecoq M, Chapuis M-P, Piou C (2020) On relative role of climate change and management in the current desert locust outbreak in East Africa. Global Change Biology 26(7): 3753–3755.
  • Ortíz YI, Zuleta MC (2020) Asuntos de vecinos: Langosta, defensa agrícola y la construcción de la sanidad vegetal en México y Centroamérica, siglo XX. Historia Mexicana 70(1): 313–373.
  • Pedgley DE (1981) Desert Locust Forecasting Manual. Vol. 1. Centre for Overseas Pest Research, London, UK, 268 pp.
  • Peng W, Ling MN, Zhang D, Zhou Q, Yue X, Ching KS, Yang H, Guan R, Chen H, Zhang X, Wang Y, Wei Z, Suo C, Peng Y, Yang Y, Shiung LS, Sonne C (2020) A review of historical and recent locust outbreaks: Links to global warming, food security and mitigation strategies. Environmental Research 191: 110046.
  • Pereyra CE (1991) Biología, Ecología y Control de la Langosta Schistocerca piceifrons (Walker) (Orthoptera: Acrididae) en el Estado de Yucatán. Tesis, Universidad Autónoma de Chapingo. México, 75 pp.
  • Poot-Pech MA, Ruiz-Sánchez E, Gamboa-Angulo M, Ballina-Gómez H, Reyes-Ramírez A (2018) Population fluctuation of Schistocerca piceifrons piceifrons (Orthoptera: Acrididae) in the Yucatán Península and its relation with the environmental conditions. Biología Tropical 66: 403–414.
  • Prospatin P (2012) Multisensor monitoring system for assessment of locust hazard risk in the Lake Balkhash Drainage Basin. Environmental Management 50: 1234–1246.
  • R Core Team (2019) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
  • Retana BJ (2000) Relación entre algunos aspectos climáticos y el desarrollo de la langosta centroamericana Schistocerca piceifrons piceifrons en el Pacífico Norte de Costa Rica durante la fase cálida del fenómeno El Niño-Oscilación Sur (ENOS). Tópicos Meteorológicos y Oceanográficos 2: 73–87.
  • Rodríguez VJ (2000) Historia de la Fitosanidad en México siglo XXI. SAGAR, México, 180 pp.
  • SENASICA-DGSV (2016) Ficha Técnica Langosta Centroamericana Schistocerca piceifrons piceifrons (Walker, 1870) (Orthoptera: Acrididae). Servicio Nacional de Sanidad, Inocuidad y Calidad Agroalimentaria – Dirección General de Sanidad Vegetal – Centro Nacional de Referencia Fitosanitaria – Grupo Especialista Fitosanitario. México, 18 pp.
  • Sharma A (2014) Locust control management: Moving from traditional to new technologies – An empirical analysis. Entomology, Ornithology & Herpetology: Current Research 4: 1.
  • Showler AT (2019) Desert locust control: The effectiveness of proactive interventions and goal of outbreak prevention. American Entomologist 65: 180–191
  • Skinner KM, Child RD (2000) Multivariate analysis of the factors influencing changes in Colorado grasshopper abundances. Journal of Orthoptera Research 9: 103–109.
  • Sultana R, Kumar S, Samejo AA, Soomro S, Lecoq M (2021) The 2019–2020 upsurge of the desert locust and its impact in Pakistan. Journal of Orthoptera Research 30: 145–154.
  • Symmons PM, Cressman K (2001) 1. Biology and Behavior. Desert Locust Guidelines. FAO, Rome, 25 pp.
  • Trujillo GP (1975) El problema de la langosta Schistocerca paranensis Burm. Sociedad de Geografía y Estadística de Baja California, Tijuana, México, 77 pp.
  • Van Der Werf W, Woldewahid G, Van Huis A, Butrous M, Sykora K (2005) Plant communities can predict the distribution of solitarious desert locust Schistocerca gregaria. Journal of Applied Ecology 42: 989–997.
  • WMO-FAO (2016) Weather and Desert Locust. WMO-No. 175, Geneva, Switzerland, 38 pp.
  • Xiang-Hui Y, Huai L, Jin-Jun W, Zhi-Mo Z, Fangneng H, Deng-Fa C (2010) Population forecasting model of Nilaparvata lugens and Sogotella furcifera (Homoptera: Delphacidae) based on Markov Chain Theory. Environmental Entomology 6: 1737–1743.
  • Yam IO, Zuleta MC (2020) Asuntos de vecinos: langosta, defensa agrícola y la construcción de la sanidad vegetal en México y Centroamérica, Siglo XX. Historia Mexicana 70: 313–373.
  • Zimmerman K, Lockwood JA, Latchininsky AV (2004) A spatial, Markovian model of rangeland grasshopper (Orthoptera: Acrididae) population dynamics: Do long-term benefits justify suppression of infestations. Environmental Entomology 2: 257–266.
login to comment