Study area

The natural distribution of pine forests in Mexico was considered as a reference (Figure 1), present in almost all of the country, except for the Yucatán Peninsula (^{Rzedowski, 2006}; ^{INEGI, 2016}). Physiographically, they predominate in the Western Sierra Madre, the Eastern Sierra Madre, the Neovolcanic Axis and the Southern Sierra (^{INEGI, 2001}). The climates where they grow are temperate sub-humid [C(w)], semi-warm sub-humid [(A)C], semi-arid and arid (BS), semi-cold sub-humid (Cb), warm sub-humid (Aw) (^{García and Conabio, 1998}). The altitude at which they thrive varies from 200 to 4 200 masl (^{Inegi, 2018}). The prevailing temperatures are 0 to 17 °C (^{Guevara and Arroyo, 2016}; ^{INEGI, 2016}).

Source: ^{INEGI (2016)}.

Figure 1 Location of coniferous forests in the Mexican republic  

Data analysis

Geographic and general data on forest notifications. The geographic coordinates of the presence of the Mexican bark beetle were obtained from the databases provided by the Forest Health Analysis and Reference Department of the Forest Health and Genetic Resources Conservation Directorate of the General Directorate of Forest and Soil Management (DGGFS) of the Subsecretaría de Gestión para la Protección Ambiental of the Secretaría de Medio Ambiente y Recursos Naturales (Semarnat) (Undersecretariat of Management for Environmental Protection of the Ministry of the Environment and Natural Resources).

The information corresponded to the years 2009 to 2018 and comprised two types:

1) Pest information. Structured by the number of notifications of the presence of the pest, termed "Indicator", during the 2009 to 2013 and 2014 to 2018 periods.

2) Annual information. Referred to as "Indicator - geographic", from 2009 to 2018, on polygons of properties affected by Dendroctonus mexicanus.

The databases were first analyzed at the DGGF's Forest Health Analysis and Reference Laboratory and then debugged by specialists from the Forest Health area of the Centro Nacional de Investigación Disciplinaria en Conservación y Mejoramiento de Ecosistemas Forestales (Cenid Comef) (National Center for Disciplinary Research in Conservation and Improvement of Forest Ecosystems).

The criteria used for screening the information were that all notifications coincided with the natural distribution and host of the insect, as cited by different authors (^{Cibrián et al., 1995}; ^{Salinas-Moreno et al., 2010}; ^{Armendáriz-Toledano et al., 2018}). In addition, the scientific name recorded in the notifications was validated by entomologists from the Forest Health area of Cenid Comef.

For spatial data management purposes, the geographic coordinates in the notifications were converted to decimal degrees. The information was imported into the geographic information system (SIG) ArcGis^TM 10.1 in order to create a layer of points in shapefile (SHP) format with the data corresponding to each record.

Subsequently, the geo-referenced data were evaluated, and the vertices of the polygons sampled for the pest were checked to ensure that they coincided with the logbook number and the indicated state. Notifications that did not meet this criterion were discarded. The analysis was performed by state of the republic in addition to screening the table of attributes of each spatial layer by logbook number, in order to avoid duplicity. Through geoprocessing in ArcGis^TM, information corresponding to the type of vegetation of each record was added, according to INEGI's Land Use and Vegetation Chart Series 6 (2016).

Statistical analysis

Based on the affected forest area in each state, the variable estimated in this study was generated: total affected surface area. From their geographic distribution, the state areas were grouped into one of the following zones: North, Central and South (Table 1). The zones were assigned for the purpose of data analysis and to keep the original arrangement of the data source.

The classification by large zones was due to the fact that abiotic factors, such as extended periods of drought, can trigger outbreaks of Dendroctonus sp.; in addition, the distribution of bark beetles is associated with altitude and latitude, among other factors (^{González-Hernández et al., 2020}).

The following fixed effects model was used for statistical analysis (^{Montgomery, 2013}):

Where:

yij = Total affected area in federal entity i of the zone j i = 1, 2, …,17

j = 1,2,3

μ = Overall average

τi = Effect of zone j εij = Random error

Hypothesis: H0:μj=μj´ ; j=1,2; j´=2,3 ; H1:μj≠e μj´

Least squares mean differences were calculated as a measure of comparison between the different zones. In the case of the normality tests of the residuals, the tests of Shaphiro-Wilk, Kolmogorov-Smirnov, Cramer-von Misses, and Anderson-Darling were used.

All statistical analyses, fixed effects tests and residual analysis of the model were carried out within the statistical analysis system environment SAS^® v9.3 (^{SAS, 2017}).

General data of the notifications

A total of 4 231 records of the Mexican pine beetle were obtained for the 2009-2018 period. It must be noted that Dendroctonus mexicanus is sometimes confused with D. frontalis and D. vitei Wood (1974). However, both species it can be reliably differentiated by the shape of the seminal rod and the anchor; in particular, it differs from D. vitei by the absence of circular concavities on the anterior face of the antennal club (^{Cibrián et al., 1995}; ^{Armendáriz-Toledano et al., 2018}). As this differentiation is impossible to be made in the field, it was assumed that most of the records are correct after the quality control carried out by expert entomologists and indicated in the methods section.

The pest occurred in 25 of Mexico's 32 states, including Mexico City, Sonora, (^{Atkinson, 2019}) and Guanajuato; this contrasts with the assertion by ^{Salinas-Moreno et al. (2010)} that the bark beetle is distributed in 23 states. Michoacán had the highest number of registrations, followed by the states of Méx, Dgo, Qro, NL, and Chih. On the other hand, Ags, Sin, Tam, Ver, followed by Son, Coah, Col, and Chis (Acronyms in Spanish), had the lowest number of records (Figure 2).

Figure 2 Logbooks with records of Dendroctonus mexicanus Hopkins (1905) by state (2009 to 2018). 

Regarding the number and distribution of records of the Mexican bark beetle, the largest number coincided with the states with the highest timber production, while the lowest number corresponded to states with an intermediate to low level of timber production (^{Semarnat, 2016b}).

As for its spatial distribution, a predominant tendency was observed in the Transverse Neovolcanic Axis, followed by the Western Sierra Madre, the Eastern Sierra Madre and the Southern Sierra Madre (Figure 3).

Simbología = Symbology; Registros de descortezador mexicano = Records of Mexican bark beetle; Bosque de coníferas = Coniferous forest.

Figure 3 Distribution in Mexico of the Dendroctonus mexicanus Hopkins (1905) records in coniferous vegetation. 

Pest records showed a decline in the 2009-2011period, and an increase from 2012 to 2014, from which it declined to its lowest point (203 cases) in 2016, and then ascended (Figure 4).

Figure 4 Number of Dendroctonus mexicanus Hopkins (1905) records from 2009 to 2018. 

In general, the percentages of bark beetle records by vegetation type are described as follows: 72.66 % are found in natural coniferous forests, 47.2 % in mixed (pine-oak) forests, and 27.3 %, in disturbed (secondary) vegetation, according to the current nomenclature of the Instituto Nacional de Estadística y Geografía, Inegi (National Institute of Statistics and Geography). The data indicated that the pest was predominant mainly in primary forests (72.66 %).

The states with the highest presence of D. mexicanus in the period under analysis were, Nuevo León, Chihuahua, Durango and Zacatecas, adding up to 59.6 % of the total in the country. These states accounted for 39 895.7 ha of damaged pine forest (Table 2).

Durango, with 25 % of its forests, and Chihuahua, with 18 %, were the states with the largest forest area and timber production affected by bark beetles. Those with the lowest percentages were Puebla (0.2 %), Mexico City and Coahuila (0.2 %), and Aguascalientes, Colima, Sonora, Tamaulipas and Veracruz, which together totaled 0.05 %. 12 635 ha (equivalent to 19 % of the total surface area) of the states located in the central part of the country ―Jalisco, Michoacán, State of México, Morelos, Mexico City, Hidalgo, Puebla and Veracruz―were affected, this being a region in which, according to certain authors, Dendroctonus mexicanus is commonly found (^{Salinas-Moreno et al., 2010}; ^{González-Hernández et al., 2020}). In the southern zone ―Guerrero, Chiapas and Oaxaca― a total infested area of 6 513.1 ha (9.72 %) was recorded. It should be noted that the surface area affected by the Mexican bark beetle in Oaxaca has increased in the last decade (^{Conafor, 2020}).

In regard to the seasonality of bark beetle damage during the period under study, the area with damage was variable. However, it was observed that in 2014, the highest percentage was 34.2 % of the total for the period, while in 2013 it represented 18.6 %, and 16.2 % in 2015, corresponding to 69.1 % of the total of the evaluated years.

The volume of damaged wood by Dendroctonus mexicanus was higher in Michoacán, Oaxaca, Durango and State of Mexico, which together totaled 1,014 455.6 m³, equivalent to 71.1 % of the country's total. In contrast, those with the least damage were Veracruz, Tamaulipas, Aguascalientes, Coahuila, Colima, Sonora and Sinaloa, which accounted for 0.22 % of the total (Table 3). It is important to note that these entities do not have a considerable extension of conifers (^{INEGI, 2016}), except for Sinaloa, where, we assume, due to its small forest area, there were fewer records than in the other states.

It must be noted that the volume of affected timber fluctuated during the period under study; however, the trend was upward. In particular, from 2012 to 2014 there was a cumulative damage to timber of almost 50 % (Figure 5).

Figure 5 Percentage of damaged wood volume by Dendroctonus mexicanus Hopkins (1905) during the 2009-2018 period. 

According to ^{Safranyik et al. (2010)} and ^{Rubín-Aguirre et al. (2015)}, the growth of the population of bark beetles is limited by environmental variables such as temperature and humidity, which would explain the annual behavior for each particular area. For example, in 2012, the months in which most of the records were concentrated were August to December, January to May in 2013, and February to July in 2014 (Figure 6); for this reason, they must be analyzed according to the climatic conditions of each state.

Figure 6 Area affected (ha) by month in Mexico by Dendroctonus mexicanus Hopkins (1905) in the 2009-2018 period. 

Two main flight peaks were detected: from April to June and from September to December, depending on the geographical region of the country (^{Vázquez-Collazo et al., 2007}; ^{Avilés-Carrillo et al., 2016}; ^{Morales-Rangel et al., 2016}; ^{del-Val and Sáenz-Romero, 2017}) (Figure 6).

Bark beetles are part of the forest dynamics, but lately, as the climate changes, especially with the anomalous temperatures (usually high), conditions are more favorable for the development of these insects as a pest of Mexican temperate coniferous forests (^{IPCC, 2014}). Dendroctonus mexicanus has shown a preferential distribution towards temperate, humid and sub-humid environments, with a mean annual temperature ranging between 12 °C and 18 °C (^{Cuéllar et al., 2012}; ^{González-Hernández et al., 2020}). In terms of precipitation, D. mexicanus has been observed in areas with 600 to 1 200 mm of rainfall (^{Armendáriz-Toledano et al., 2018}); however, previous modeling has indicated its potential distribution in an optimal range of approximately 800 mm (^{González-Hernández et al., 2020}).

The rate of development of certain species of bark beetles of the genus Dendroctonus, including Dendroctonus mexicanus, depends mainly on temperature (^{Mitton and Sturgeon 1982}). Therefore, the monthly behavior would be associated with the particular temperature conditions for each year and state evaluated. Although there is a lack of data, it has been shown that the lack of rainfall is related to the increase in the number of bark beetle reports (^{Esparza, 2014}); which would explain a sudden increase in the number of pest records from 2009 to 2011 in northern Mexico.

Correlation of the effect of geographic area on total affected area

The value of the overall test of F _2,14 ─ fixed effects─ was 17.99, and its corresponding p-value was 0.0001. Therefore, there is sufficient evidence to conclude that there are highly significant differences between the three zones under study (North, Center and South) with respect to the response variable: total area affected by bark beetles. However, the least squares means were different from zero for the Central and North zones, with a significance level of 0.05; whereas, the South zone had a level of 0.10 (Table 4).

The estimated average for the North zone (near to 10,000 ha) was more than five times higher than the average for the Central zone, and three times higher than the mean for the South zone; likewise, the highest variability was characteristic of the North zone, while in the South zone the values were more similar (Figure 7).

Superficie total afectada = Total area affected; Centro = Center; Norte = North; Sur= South.

Figure 7 Mean values and variability of the total area affected by bark beetles by zone. 

It was observed in particular, that the differences between the Central and North zones, as well as North and South were highly significant for the variable total area affected by bark beetles. This behavior was not detected in the Central or South zones (Table 5).

Normality tests indicated that the normal distribution of the studentized residuals was normal. The four statistical tests applied had p-values above 0.05 (Table 6), which is congruent with the symmetrical distribution of the residuals, whose extremes are lower than 3.0. In any case, any potential violation of other model assumptions is not critical, given that the p-values corresponding to the overall fixed effects test and the differences in means are robust, i.e., they are far from the benchmark value of 0.05.

Differences between geographic areas could be due to the levels of forest fragmentation in each region (^{Moreno-Sanchez et al., 2012}), regional anomalies of climatic variables (^{Romero-Sanchez et al., 2018}), and, possibly, the institutional response of forest health agencies in the states. Finally, high-density populations of Mexican bark beetles have been responsible for economic, environmental and social losses in forest ecosystems (^{Moser et al. 2005}; ^{Manzo-Delgado et al., 2014}). Therefore, it is important to identify the areas with the highest historical incidence of this pest in order to make appropriate decisions for both forest management and the management of this particular debarking pest.