Spatiotemporal Variability in Disturbance Frequency and Severity across Mexico: Evidence from Conifer Tree Rings

Abstract

It is critical to understand the ecological processes of forest dynamics by identifying past forest disturbances to take appropriate management actions. Tree-rings are commonly used for this purpose due to their reliability and accuracy. Here, we used a network of ring-width data distributed along a broad ecological gradient for the spatiotemporal identification of forest disturbances in 15 species of 13 sites of young Mexican conifers. We found different spatial and temporal patterns in the occurrence of forest disturbance. The species with the highest disturbance rates were Pinus patula (Ppa; El Jacalón), Pinus montezumae (Pmo; La Cueva), Pinus hartwegii (Phc; Cerro El Moro), and Pinus teocote (Pts; Rancho Joyas del Durazno) of the Gulf of Mexico, and Pinus ayacahuite (Pay; Corral de los Borregos) of the Mexico Center. The highest number of disturbances was found in 1953, 1956, 1976, and 1980; however, we noted that the number of disturbances tended to increase from 1975 to 1990 and particularly from 2000 onwards. The species close to the Gulf of Mexico presented a higher frequency and severity of disturbances than those located close to the Pacific Ocean. Thus, our results demonstrate a spatiotemporal pattern of disturbance events, which deserves further analysis, including regarding the drivers of disturbance. This knowledge is crucial for a better understanding of the dynamics of contemporary species.

1. Introduction

Understanding forest ecosystem dynamics has essential implications for the demands of sustainable forest management in the face of climate change scenarios [1].

It is recognized that forest ecosystems undergo constant change due to disturbances, but also because of the direct and indirect effects of changing temperatures and CO₂ emissions [2,3]. Such disturbance regimes lead to changes in the forest structure and composition, especially the transformation of old stands into young stands [4]. Understanding the role of temporal and spatial disturbances can provide critical insights into the biogeography of forests [5]. Forest ecology research must therefore focus on accurate spatial and temporal information regarding past forest disturbances [6]. In this regard, the retrospective study of releases (i.e., abrupt and sustained increases in radial tree growth) using tree-ring analysis is a widely used approach to reconstruct historical forest disturbances [7,8]. The tree-ring approach provides a robust estimate of past disturbance events [9].

Tree-rings have been used worldwide for the growth release analysis and reconstruction of historical disturbance [10], disturbance dynamics [11], the identification of disturbance agents and temporal and spatial differences in disturbance severity, and the proportion of the stand disturbed [12], comparisons of disturbance reconstruction among radial increment cores [13], tree canopy disturbance in order to refine dendroclimatic reconstructions [14], and the integration of paleo-ecological approaches [15], etc.

Globally, most tree-ring-based disturbance reconstructions have been conducted in Europe and North America [16], while few studies are available for other parts of the world, including Mexico [17]. In this country, despite the tree diversity found in various ecological niches, there have been only limited applications of dendroecology in the study of disturbances such as hurricane damage [18] or fires [19], or with BAI modelling used as a proxy for disturbances [20]. Consequently, the study of disturbance must be improved, using a spatiotemporal perspective as an ideal strategy to improve the scientific knowledge of forest dynamics in these ecosystems. These forests are characterized by diverse and mostly young conifers, and represent the current structure [21], and their analysis is gaining importance given their high productivity [22]. The region is home to a great diversity of taxa, comprising 10 genera including Pinus, Abies, and Juniperus, with a high degree of endemism [23].

Using a network of ring datasets distributed along a broad ecological gradient, we identified disturbances and analyzed the spatial and temporality variability of past disturbances along geographical gradients in contemporary Mexican conifer forests. We hypothesized that a network of tree-ring datasets will identify spatial and temporal patterns in disturbance events across contemporary Mexican coniferous forests.

This cross-species approach is expected to increase knowledge of the spatiotemporal patterns of forest disturbances in contemporary Mexican conifer forests for subsequent research.

2. Materials and Methods

2.1. Study Area

Thirteen species were selected that occur along a bioclimatic gradient of Mexico, which is characterized as a highly diverse region of the planet [24] (

Table 1. Geographic location, statistical description, and climate conditions of sampled sites.

Site Name	Spp	masl	Long.	Lat.	MAT (°C)	TAP (mm)	SA	AHMI	TH (m)	Dq (cm)	AB (m2)
Pacific Ocean (PO)
La Laguna = LAG	Pce	1897	−109.99	23.54	22	420	66 ± 5	7.62	12.4	38	2.1
Potrero de Mulas = MUL	Poj	818	−104.98	20.74	22.5	1637	43 ± 1	1.99	14.2	26.3	1.1
Alto de la Lagunita = ALT	Phe	2716	−106.48	25.19	24.5	810	51 ± 4	4.26	16.82	30.9	1.5
Papanton = PAP	Jdp	3078	−103.78	23.67	15.1	459	35 ± 2	5.47	5.8	26	0.5
Cerro del Bravo = BRA	Plu	2391	−101.72	20.54	18.1	554	74 ± 5	5.07	24.8	45.8	3.3
Alto de la Lagunita = ALT	Pdu	2716	−106.48	25.19	24.5	810	52 ± 2	4.26	18.3	29	1.3
Gulf of Mexico (GM)
La Mesa = MES	Ptn	1900	−100.13	25.19	19.3	540	40 ± 1	5.43	24.4	51.7	4.2
Cerro El Morro = MOR	Phc	3401	−100.36	25.2	16.1	648	73 ± 7	4.03	15	39.4	1.8
Orillas de Ocotal Chico = OCO	Pov	713	−94.85	18.27	23.7	2100	28 ± 0.5	1.6	15.2	36	2
El Jacalón = JAC	Ppa	2602	−99.45	23.88	16.6	875	67 ± 2	3.04	30	47.6	3.5
La Cueva = CUE	Pmo	2579	−99.44	23.87	22.7	836	72 ± 3	3.91	22.4	49.1	3.7
Rancho Joyas del Durazno = RAN	Pts	2030	−100.35	21.89	17.8	477	44 ± 2	5.83	12	29	1.3
La Mesa = MES	Pse	1898	−100.13	25.19	19.3	540	32 ± 1	5.43	20.6	41.9	2.7
Mexico Center (MC)
Corral de los Borregos = COR	Phm	3500	−98.75	19.41	14.7	600	65 ± 5	4.12	24	55.8	4.9
Pozuelos = POZ	Pay	2985	−96.44	17.38	17.2	968	56 ± 3	2.81	29.8	53	4.4

where Spp = species, masl = meters above sea level, Long = longitude, Lat. = latitude, MAT = mean annual temperature, TAP = total annual precipitation, SA = stand age, AHMI = annual heat–moisture index, TH = top height, Dq = quadratic mean diameter, and AB = basal area.

). The gradient ranged in elevation from 713 to 3500 m asl. All of the sites corresponded to young and unevenly aged forest stands in which the predominant species in each site comprised at least 70% of its basal area.

At each site, a central tree was randomly selected for each species, and a further nineteen of the closest individual trees were then sampled. From these, at least two radial cores were taken using a Pressler increment borer (Ø = 5 mm) at a height of 1.3 m above ground level, a standard known in forestry as diameter at breast height (DBH, cm).

2.2. Ring Width Data

Radial increment cores were mounted on wooden racks and polished with increasingly fine sandpaper. Tree-ring width (TRW) was measured using a VELMEX ultra-precision (0.01 mm) measuring system. Cross-dating of samples from the same species and location was performed. Visual dating was verified by statistical validation using the program COFECHA [25], which compares the ring series of each individual tree with a master chronology for each species. We built species tree-ring width chronologies using standard dendrochronological procedures [25]. We calculated the mean values of inter-tree correlation (Rbt), subsample signal strength (SSS), and expressed population signal (EPS), where a minimum threshold of 0.85 is normally considered indicative of well-replicated chronologies [26] (see

Table A1. Dendrochronological statistics of the sampled species.

Site Name	Spp	n (nc)	TRW ± sd	Years ± sd	Rbt	EPS	Period
Pacific Ocean (PO)
La Laguna = LAG	Pce	15 (30)	1.28 ± 0.07	66 ± 5	0.32	0.87	1970–2019
Potrero de Mulas = MUL	Poj	20 (40)	1.57 ± 0.05	43 ± 1	0.41	0.96	1987–2021
Alto de la Lagunita = ALT	Phe	20 (40)	2.36 ± 0.12	51 ± 4	0.43	0.97	1982–2021
Papanton = PAP	Jdp	16 (33)	2.16 ± 0.19	35 ± 2	0.54	0.96	1986–2020
Cerro del Bravo = BRA	Plu	18 (36)	1.38 ± 0.09	74 ± 5	0.45	0.94	1917–2020
Alto de la Lagunita = ALT	Pdu	20 (40)	1.68 ± 0.08	52 ± 2	0.3	0.94	1992–2021
Gulf of Mexico (GM)
La Mesa = MES	Ptn	20 (40)	4.16 ± 0.13	40 ± 1	0.52	0.97	1977–2020
Cerro El Morro = MOR	Phc	15 (31)	1.61 ± 0.18	73 ± 7	0.48	0.97	1987–2021
Orillas de Ocotal Chico = OCO	Pov	16 (33)	3.16 ± 0.20	28 ± 0.5	0.26	0.9	1986–2020
El Jacalón = JAC	Ppa	20 (41)	2.46 ± 0.09	67 ± 2	0.49	0.97	1952–2020
La Cueva = CUE	Pmo	20 (43)	2.01 ± 0.11	72 ± 3	0.38	0.95	1937–2020
Rancho Joyas del Durazno = RAN	Pts	19 (38)	2.50 ± 0.15	44 ± 2	0.59	0.97	1970–2020
La Mesa = MES	Pse	20 (40)	4.34 ± 0.19	32 ± 1	0.42	0.95	1983–2020
Mexico Center (MC)
Corral de los Borregos= COR	Pay	20 (40)	2.92 ± 0.11	56 ± 3	0.33	0.93	1959–2020
Pozuelos = POZ	Phm	13 (27)	1.64 ± 0.11	65 ± 5	0.26	0.86	1975–2019

where Spp = species, n = no. of trees, nc = no. of cores, TRW = tree-ring width, Years = no. of years, Rbt = correlation between trees, and EPS = expressed population signal.

; Figure A1 in Appendix A).

2.3. Disturbance Identification

The boundary line method proposed by Black and Abrams [27] was used to identify disturbances. This methodology included prior growth, in which releases within 20%–49.9% and 50%–100% of the limit line were considered. The effect of prior growth on the release response was evaluated by plotting prior growth against percentage change in growth. The percent change in growth was calculated using the following formula (Equation (1)):

$$Percent growth change=\left(M_2-M_1\right)/M_1$$

(1)

where:

M₁ = average growth over the previous 10 years;

M₂ = average growth over the next 10 years.

Prior growth was defined as the average gross growth in the 10 years preceding each annual ring. Prior growth and percentage change in growth were both calculated for each year of each series using the RWL files.

The TRADER package [28] was used to determine the frequency and intensity of disturbances using R software [29]. We analyzed only the common period from 1950 to 2010, and all detected releases above a 20% threshold were considered disturbance events according to the suggested threshold by the original methodology [27]. We did not divide releases into categories as we are interested in the general pattern, and we could not connect these with the recorded severity of disturbances as these are not known [6]. Detected releases were grouped into 5-year intervals. This strategy was based on our previous knowledge of the territory and experience in dendrosciences [17].

2.4. Spatial Characterization of Disturbances

For spatial characterization of disturbances, growth releases were identified according to their vicinity with respect to the Gulf of Mexico, Mexico Center, and the Pacific Ocean. In addition, kernel density estimation (KDE) was used for two different time periods: 1950–1980 and 1981–2010. This is a nonparametric statistical method for estimating spatial density that has been used previously in Mexico for predicting forest disturbances, especially wildfires, e.g., [30]. Although disturbances drivers were beyond the scope of our study, we complementarily explored drought as a potential factor in the spatial variation in disturbance distribution. We calculated the Standardized Precipitation Evapotranspiration Index (SPEI) (see https://spei.csic.es/map/maps.html#months=7#month=8#year=2021, accessed on 23 May 2022). This multi-scalar drought index is an ideal strategy to address geographic disturbances–hydroclimatic relationships, given that it utilizes data pertaining to spatio-temporally accumulated droughts. This evidences a close correlation between evaporative demand and its influence on growth fluctuations. Thus, negative values signify a water deficit, indicating drought, and vice versa. We obtained SPEI data at a 0.5° spatial resolution from the SPEI Global Drought Monitor webpage (https://spei.csic.es/map/maps.html, accessed on 30 May 2022).

3. Results

3.1. Identification of Temporal Disturbances

Disturbances were recorded for at least one species in 36 of the 60 years analyzed during the common period from 1950 to 2010 (Figure 1).

The periods with the highest number of disturbances per species (Pay, Ppa, Pmo, Phc, and Ptn) were 1953, 1956, 1976, and 1980. In general, a higher occurrence of disturbances was observed from 2000 onward (Figure 2). The species with the highest number of disturbances neighboring the Pacific Ocean was Pce during 1967, 1978, and 2009. At the easterm sites close to the Gulf of Mexico, Ptn, Pmo, Ppa, Phc, and Pse presented the highest percentage of cores with disturbances for 1975, 1951, 2007, 1980, and 2010. Finally, those sites located in central Mexico, in which Pts and Phm were found, had 50% of trees evidencing disturbance for the period 1953–1957 and 1960–1963, respectively (see Appendix B).

3.2. Spatial Characterization of Disturbances

It has generally been found that sites located close to the Gulf of Mexico are more intensely disturbed, with CUE, JAC, and MOR exhibiting frequent releases (Appendix B, Figure A2, Figure A3 and Figure A4).

CUE and JAC at the same site (Güemes, Tamaulipas) showed the same pattern of release, unlike Pov and Ptn. MES showed more frequent major releases (>50%) (Appendix B, Figure A5, Figure A6 and Figure A7). MOR experienced disturbances continuously during 1976–2003, with the highest proportion of trees showing strong releases in 1980 and 1993, while at POZ there were moderate releases (Appendix B, Figure A8).

In the sites of central Mexico, a greater number of trees responded moderately to disturbance. Pts, Phm, and Plu evidenced disturbance only in 1957, 1960, 1961, and 1962 (Appendix B, Figure A9, Figure A10 and Figure A11).

At sites neighboring the Pacific Ocean, Pce showed medium intensity disturbance (Appendix B, Figure A12). Jdp showed a strong release in 1976, but a lower incidence in number of trees (Appendix B, Figure A13). Pdu recorded a low rate of forest disturbance from 1973 to 2007 (Appendix B, Figure A14). Phe presented moderate disturbances from 1978 to 2011 (Appendix B, Figure A15). Poj had a lower intensity from 2005 to 2011 (Appendix B, Figure A16).

Our results indicate that disturbance frequency increased during the second study period (1981–2010) (Figure 3). Nevertheless, some sites had similar disturbance frequencies over both periods (e.g., LAG, RAN, or COR; see Appendix B). In particular, sites in northeastern Mexico presented a greater intensity in releases. In contrast, the sites of central Mexico showed a moderate change in disturbances between the two periods studied.

The SPEI dynamic indicates differences in drought frequency and intensity between the Pacific Ocean and the Gulf of Mexico, especially during the last two decades (Figure 4). Thus, the trees growing near the Gulf of Mexico differed in release events, which may reflect their differential responsiveness to precipitation episodes (e.g., hurricanes), particularly from 1996 onwards.

4. Discussion

To our knowledge, this study is the first of its kind in Mexico to identify disturbances. We identified past disturbances in 13 species of young Mexican conifers along a diverse spatial ecological gradient. We analyzed the release of ring width growth in each species over a range of years from 1950 to 2010. We did not examine the effects of stand characteristics (Table 1) on apparent disturbances, a theme that merits further research.

4.1. Identification of Temporal Disturbance

Variability was found in the reconstruction of disturbances among the 13 species studied. The ring-width data showed widespread disturbances for the period 1950 to 2010, suggesting complex disturbance patterns within and among the studied species, and supporting the previous reports of other authors [31]. Species exhibited different temporal patterns of disturbance represented by the growth failures observed.

Given that these forests are unevenly aged, this indicates that they have not been disturbed by a single catastrophic event that completely killed the stand. The frequency and intensity of the releases detected here indicates that there have only been episodes of gap forming disturbances. However, the short temporal scale of the study of these species is inconclusive in terms of disturbances, since the previous occurrence of these events (prior to 1950) remains unknown. According to our results (Figure 4), we hypothesize that disturbances in Mexico will be associated with the combination of drought and other climatological phenomena, such as hurricanes or landslides caused by heavy rains, as well as pathogens and diseases associated with drought and fire regimes. Further research on these phenomena and their relationships to the disturbances is therefore vital at the site level as it is impossible to capture such complexity over a large area without long-term observation and detailed records.

4.2. Spatial Characterization of Disturbances

Species found in the eastern sites presented a higher frequency of disturbance compared to those in the west and center. This was most evident during the period from 1950 to 1980, and was consistent with a global study indicating that the increased frequency and severity of droughts due to climate change has become the main factor affecting tree growth. Another similar result was the change in response as a function of the period of analysis, since differences in growth response to climatic conditions and drought were found in the periods 1950–1980 and 1981–2010. This suggests that climate variability is an important factor to consider when reconstructing disturbance regimes. Our results suggest specific interactions between disturbance trends and seasonal hydroclimate variability.

Temporally, disturbances were recurrent, but only severe in certain years. This reconstruction suggested that these disturbances were infrequent but more severe, a finding that was consistent with that reported in studies from other countries [10]. This is attributed to the response of these species to large-scale natural disturbances such as hurricanes [18] and circulatory phenomena such as ENSO (El Niño Southern Oscillation) [32,33]. There is consistent evidence of spatially extended perturbations that synchronously affect different biome types. Several of the reconstructions of regional-scale disturbances are associated with intense regional drought events triggered by the variability of ENSO in the Pacific Ocean [31]. Across the species, we found that some explain the disturbance through geographic proximity, while others respond differently to disturbances. This has ecological implications for forest successional processes and structural characteristics [34]. In other words, forest composition is determined by resistance or resilience to disturbance, which provides a tool for managers to shape their silvicultural regimes [12].

The spatial perspective shows certain geographic patterns are attributed to the occurrence of site-level disturbances. However, different species at the same site showed differences in the occurrence of disturbances. This could be explained by the fact that frequent and less severe natural forest disturbances generally occur at smaller spatial scales [35]. From a temporal perspective, Figure 2 shows an increase in disturbance. This indicates a warning signal in the vegetation dynamics that deserves further research in order to elucidate the ecological or anthropogenic mechanisms underlying these disturbances. In general, species from sites closer to the Gulf of Mexico (between 95° and 100° W) were disturbed more frequently than those closer to the Pacific Ocean (between 100° and 105° W). Although we did not investigate the causes of disturbance in depth, our dendroecological approach showed a clear differentiation in responses to disturbances between the Gulf of Mexico and Pacific Ocean regions.

We found that in Gulf of Mexico there are, in general, fewer droughts and drought and wet periods alternate. However, there have been two relatively long and intense drought periods (one in 1960 and one before 2000), which did not occur in the Pacific. This intense drought could induce tree mortality. Thus, we suggest that the higher intensity of detected releases in the Gulf of Mexico may be related to drought [36]. In addition, the Gulf of Mexico is significantly more affected by hurricane activity, and thus we suggest that the combination of drought and hurricanes is mainly responsible for forest disturbances in this area (see Appendix C to analyze relationships between geographic space and environmental influences on the disturbances).

We can speculate that the climate–growth relationships are acting to regulate the disturbances in each zone. Thus, the west is controlled by the North American summer monsoon related to El Niño [37]. The positive correlations between growth and SPEI confirm that El Niño events are associated with radial growth. On the other hand, the forests of the east appear to present promising opportunities under the occurrence of hurricanes. Moreover, both regions offer different microsite conditions, including topography, genetic load, seasonal growth, etc., that merit further investigation. Previous studies have documented that the presence of disturbances (hurricanes) has an impact on Earlywood (EW) and Latewood (LW) anatomical characteristics [38]. We therefore recommend further investigation with higher temporal resolution proxies (EW, LW) for future studies. It is known that the seasonal responses of trees can convey a fingerprint of disturbance that is not reflected in the total ring width [39].

Another limitation of our study could be the short duration of the analysis period, and data available in the International Tree-Ring Data Bank (ITRDB) could be utilized to complement the temporal perspective [6]. Expanding the temporal perspective could greatly benefit the scope of our study, as could explaining the causes of such disturbances. Understanding their dynamics and examining their causes in the future will allow predictions to be made with respect to migration, decline, and mortality episodes, including tree mortality.

Other factors for consideration should include species resilience, recovery, and plasticity since, as seen here, there are differences in the frequency and severity of disturbances found among coexisting species. Forest resilience has been used to measure the severity of disturbances such as drought and is a good indicator of forest composition [40]. Studies of intraspecific genetic variation and phenotypic plasticity are also essential in order to estimate the vulnerability of forest species to drought [41,42]. Further research is recommended in order to test these arguments.

The histogram of detected growth releases provided insights into the disturbance regime. Thus, the disturbances have consequences for forest development due to the fact that they interrupt continuous succession and alter growth rates. This has direct implications in terms of redefining the concept of natural succession, which in turn impacts the composition and structure of species. This tool should be integrated into the portfolio of the forest manager, as a complement to inventory data. For example, the estimation of disturbance frequency produced in this study is crucial for decision-making because the studied species are subjected to forest management and logging [21,39]; it has been documented that the age of rotation within forest management should not exceed the frequency of the disturbances [7].

5. Conclusions

Here, we investigated the recording of spatial and temporal disturbances by contemporary conifer species representative of forest ecosystems in Mexico.

The results suggest that forest disturbances follow a spatial and temporal trend. It is difficult to reach a definitive conclusion and to generalize regarding the multifactorial causes of forest disturbances across a large study area. Although the influence of anthropogenic disturbances has already been demonstrated, in this study it is evident that climate and large-scale circulatory patterns influence the release rates. Overall, disturbances appear to follow an increasing rate from 2000 onwards. The proxies used in this study produced parameters that improve our understanding of forest dynamics and should be considered in vegetation dynamics models to reduce uncertainty in the face of climate vulnerability. Our approach constitutes a suitable route for identifying disturbances as parameters of great importance that can act to configure the structure, composition, and function of forest ecosystems.