Characterization, variability and long‐term trends on local climate in a Mexican tropical dry forest

Based on 40 years of meteorological data, we characterized and analysed long‐term climatic trends in the Chamela‐Cuixmala Biosphere Reserve on the Pacific coast of central Mexico. The region is still covered by a large proportion of well‐preserved tropical dry forest, characterized by a short period of rain and a dry season of around 8 months. We found a sustained temperature increase which is likely driven by global climate change. An increasing trend in rainfall was also found, but the trend was detected only during the wet season. Correspondingly, an increase in runoff during the rainfall season was also detected; this seems to be linked with an increasing number of intense rainfall events, rather than an increase in the total number of rainy days. El Niño years are likely to present below average precipitation during the wet season and above average precipitation during the dry season. The opposite is expected during La Niña years, when tropical cyclones are likely to come closer to the coast and cause intense rainfall events during the hurricane season. As many of the studied variables are likely to change under climate change scenarios, our results highlight the need to understand the expected impacts of global climate change on tropical dry forests. In particular, it is necessary to monitor changes in water availability to anticipate its consequences for the forest and the human communities that depend on it.


| INTRODUCTION
Tropical dry forests (TDFs) worldwide are threatened by climate change (Curry, 2021;Miles et al., 2006;Siyum, 2020). Global increases in temperature and changes in rainfall patterns are expected to have an impact on TDF nutrient dynamics (Baccini et al., 2019;Jaramillo & Murray-Tortarolo, 2019), biodiversity (Stan & Sanchez-Azofeifa, 2019) and ecosystem services (Ryan et al., 2016), which will have an effect over people who inhabit and depends on these ecosystems (Siyum, 2020). However, climate change is not expected to equally alter all TDFs, with a higher proportion of forest at risk expected for the forests located in the Americas (Miles et al., 2006). Despite the previous, it is only recently that long-term climatic shifts have been studied in this ecosystem.
TDFs occur in tropical regions and are characterized by its rainfall seasonality and interannual variability, which results in a dry season that lasts for several months (Bastin et al., 2017;García-Oliva et al., 2002;Hasnat & Hossain, 2020;Miles et al., 2006). While the beginning of the rainy season has little variation, the duration, intensity and amount of rain is highly variable from one year to another (Ceccon et al., 2006;Maass et al., 2018); likewise, the onset and duration of the dry season has a high interannual variability (Ceccon et al., 2006;Maass et al., 2018). This distinctive climate has determined the structure, function and dynamics of the forest (Maass et al., 2005;Portillo-Quintero et al., 2015;Siyum, 2020), as well as the physiology and metabolism of the biological species within it (Portillo-Quintero et al., 2015;Stan & Sanchez-Azofeifa, 2019).
Increases in temperature, changes in precipitation patterns and the growing incidence of droughts, floodings, hurricanes and fires due to climate change can importantly alter TDFs functioning and threaten its stability (Curry, 2021;Siyum, 2020). For instance, although TDFs are physiologically adapted to cope with lack of water during several months per year, the incidence of frequent severe drought events or a lengthening of such events could reduce the forest capacity to recover by a reduction in its functionality and loss of biomass (Curry, 2021;Gonz alez et al., 2021).
Similarly, the incidence of hurricanes impacts all levels of TDFs from populations to landscapes . Several authors have described extensive canopy removal and reduction of canopy height after hurricane impact (Gao & Yu, 2021;Jaramillo et al., 2022;Parker et al., 2018;Stan & Sanchez-Azofeifa, 2019). Also, significant modifications in vegetation structure and phenology have been observed to impact trophic relationships and forest function (Williams-Linera et al., 2021;Suazo-Ortuño et al., 2018) as well as litter composition, production (Jaramillo et al., 2022;Martínez-Yrízar et al., 2018) and decomposition levels (Stan & Sanchez-Azofeifa, 2019). Under natural conditions (i.e., well preserved forest prior to hurricane impact), native tree species are well adapted to intense hydrometeorological events  resulting in a rapid re-growth and almost complete leaf-flush within a few weeks after disturbance . However, under harsh management conditions, recycling processes are pushed to a threshold where they do not necessarily recover as rapidly . If such extreme events increase in frequency or intensity, forests might not be able to recover at all (Curry, 2021;Siyum, 2020;Stan & Sanchez-Azofeifa, 2019).
One of the most studied TDFs in the Americas is in the Chamela-Cuixmala Biosphere Reserve (CCBR) in the Southern Pacific Coast of Mexico where the Chamela Biological Station from the National Autonomous University of Mexico (UNAM) is located. This forest has been under protection for over 50 years, being an important remnant of pristine primary vegetation. In addition, research in the site has resulted in more than 1300 scientific products (Castillo et al., 2021), with long-term ecological research (LTER) standing out for their uniqueness in the understanding of the forest structure and dynamics Campo et al., 2001;Jaramillo & Murray-Tortarolo, 2019;Maass et al., 1988Maass et al., , 2005Martínez-Yrizar et al., 1996;Mason-Romo et al., 2017;Sarukh an & Maass, 1990;Vihervaara et al., 2013).
Recent studies have recognized that climate change is altering this forest functioning. Yamanaka (2012) described variations in the ecosystem productivity and the length of the growing season due to erratic rainfall patterns. Prieto-Torres et al. (2016) stated that climate models suggest changes in the distribution of Mexican TDFs outside its current geographical range. Esperon-Rodriguez et al. (2019) described an expected shift in the climatic conditions for several natural protected areas in Mexico, including the CCBR.
Not unlike other TDFs, Chamela-Cuixmala's rainfall regime is markedly seasonal. Bullock (1986) and García-Oliva et al. (2002) characterized Chamela-Chixmala's climate as warm, subhumid with summer rainfall that last for 5 months, from June to October, and is importantly influenced by tropical cyclones (TCs). Accordingly, the greatest amount of precipitation is recorded during September (Bullock, 1986;Maass et al., 2018), when TCs are most frequent (García-Oliva et al., 2002). However, El Niño-Southern Oscillation (ENOS) also has an important role in the rainfall pattern. There is a significant decrease in precipitation during the wet season and an increase in winter rainfall during El Niño years, with the opposite effect during La Niña years (García-Oliva et al., 2002;Maass et al., 2018).
Although there are some characterizations on Chamela-Cuixmala's climate (Bullock, 1986;García-Oliva et al., 2002) and a very comprehensive description on its hydrological dynamic , most long-term changes in seasonal and annual weather patterns remain to be investigated. A description of the climatic shifts in this forest could offer valuable information for climatic, ecological and socioecological research looking forward to understanding the impact of climate change in TDFs and responding accordingly.
In order to contribute to the above-mentioned gap, we present an analysis of climate change during the last four decades in the Chamela-Cuixmala Biosphere Reserve and the surrounding area. Our specific research questions were (a) Which are the long-term shifts in annual climatic means (temperature and precipitation)?; (b) What is the influence of tropical cyclones and El Niño-Southern Oscillation on precipitation variability? and (c) What insights can we get on seasonal dynamics based on simple water budgets (e.g. precipitation minus potential evapotranspiration, or precipitation and runoff)?
Maximum and minimum temperature, and precipitation at a daily resolution from 1982 to 2021 were provided by the EBCH, located in the CCBR (19 29 0 N, 105 02 0 W). Daily mean temperature was calculated based on the arithmetic average of minimum and maximum temperatures. Runoff data from 1983 to 2015 was obtained from Maass et al. (2018) and corresponds to five small watersheds (12-28 ha) under natural conditions within the Reserve.
Data from cyclonic events including tropical depressions, tropical storms and hurricanes were obtained from the revised Northeast and North Central Pacific hurricane database (HURDAT2; https://www.aoml.noaa.gov/ hrd/hurdat/hurdat2-nepac.html) from the National Oceanic and Atmospheric Administration (NOAA). This database is the result from a revision of the original HURDAT in 2016 and includes the name, date and cyclone number, as well as location (latitude and longitude) and maximum sustained wind (in knots) recorded every 6 h. For the present work, only cyclonic events from the Northeast Pacific Basin (NEP; cyclones formed in the North Pacific east of 140 W) were considered. Landfalling events were confirmed with reports emitted by either the NOAA or the National Water Commission (Conagua) in Mexico.

| Data analysis
Statistical analyses were performed to identify the annual mean, standard deviation and linear trend for maximum and minimum temperature, precipitation, potential evapotranspiration, runoff, and tropical cyclone (TC) frequency and intensity. For the annual trends a simple linear regression was fitted between the variable and the year. A Sen's slope estimator with a Mann-Kendall trend test was used to identify the statistical significance of the trends when the data did not fit a normal distribution. The same analyses were performed at the seasonal scale with the time series divided into the wet (June-October) and dry (November-May) seasons.
Monthly water balance was established using precipitation minus potential evapotranspiration (P-PET) or precipitation minus actual evapotranspiration (P-AET). To assess the change in seasonal water balance, we sum the P-PET or P-AET values of each season's months and evaluated the long-term trend with a simple linear regression and a Sen's slope estimator.
Correlation analyses were made between oceanatmosphere oscillations and both mean annual and seasonal precipitation and temperature, as well as tropical cyclone incidence.
To assess the intensity of tropical cyclones, the annual accumulated cyclonic energy (ACE) was calculated through the equation: ACE = P v 2 max , where v max is the maximum sustained winds with values ≥35 knots at 6-h intervals derived from HURDAT2.
To calculate the precipitation associated with tropical cyclones, mean location (latitude and longitude) of TCs was calculated for each day. Then, precipitation registered on days when a TC was present at 500 km or less from the Chamela Biological Station was summed, avoiding data duplication on days when more than one TC was present (Dominguez & Magaña, 2018). In accordance with Jiang and Zipser (2010) and Dominguez and Magaña (2018), the 500 km radius is considered adequate to estimate TC-related rainfall.

| Data limitations
Daily meteorological record from the Chamela Biological Station is nearly complete for the last four decades, with only some days missing record for temperature or precipitation, which makes it a very valuable database to study climate. However, it should be noticed that topography in the study area is very heterogeneous (Bullock, 1986;Maass et al., 2018) and while tropical dry forest is the predominant vegetation, riparian forests, mangroves and tropical semideciduous forests occur near the riverbeds S anchez-Azofeifa et al., 2008;Trejo, 1998).
Furthermore, human settlements near the Chamela-Cuixmala Biosphere Reserve have transformed the forest through deforestation for agriculture and cattle farming use, resulting in a highly fragmented and, sometimes, degraded landscape (Flores-Casas & Ortega-Huerta, 2019; Maass, 1995;S anchez-Azofeifa et al., 2008). Hence, while our study offers an overview of climate and climate trends for the CCBR and adjacent areas, it is likely that local microclimate variations are found due to the abovementioned features.

| Mean annual climate
Over these 40 years (1982-2021) the mean annual temperature (MAT) was 25.8 ± 0.7 C; the annual minimum mean temperature (Tmin) was 20 ± 1.0 C, with the mean lowest temperature recorded during the month of March (average 16.6 ± 1.3 C) and the annual maximum mean temperature (Tmax) was 31.3 ± 1.0 C, with the highest temperature recorded during the month of July (average 32.9 ± 1.2 C).
Mean annual precipitation (MAP) was 832 ± 277 mm with a noteworthy high interannual variability (CV 33.3%), there is a fourfold variation between extreme years: the wettest year (2015) had 1394 mm of rainfall, while the driest (2005) only received 340 mm. There is also a high intra-annual variability, with the highest precipitation records in September (average 243 ± 165 mm) while the lowest corresponded to April (average 0 ± 0.0 mm); correspondingly, mean annual runoff was 99 ± 119 mm of which 47.6% occurred during September (average 47 ± 90 mm), while no runoff was ever recorded during April (0 ± 0 mm).
Mean annual PET was 1553 ± 140 mm, the highest values were recorded during July (183 ± 24 mm), and the lowest during February (average 81 ± 13 mm), which correspond to the months with the hottest and the second coldest (16.7 ± 1.4 C) average temperature record, respectively. However, for AET, the highest values were found in September and the lowest in April, which are the wettest and driest months, respectively (Table 1). This difference is likely due to the differential weight of either temperature or precipitation when estimating PET and AET, respectively.

| Interannual precipitation variability and the influence of the oceanatmosphere oscillations
We found no statistically significant correlation between PDO, NAO or MEI with MAT or MAP. Nonetheless, when the relationships were evaluated seasonally, we found a positive and significant correlation between MEI and rainfall during the dry season (Spearman's rho 0.37) and a negative correlation between MEI and rainfall during the wet season (Spearman's rho −0.39), with similar values for El Niño3.4 index ( Figure 1 and Table S1, Supporting Information). In other words, during El Niño years (hot phase of ENSO), precipitation tends to be below average during the wet season and above in the dry season, with the opposite effect being observed during La T A B L E 1 Mean values of annual temperature and values for the hottest and coldest months with their corresponding standard deviation (SD)

Temperature
Mean ± SD Hottest month (Jul), mean ± SD Coldest month (Mar), mean ± SD Niña years. This balancing effect might explain why ENSO influence is undetectable at the annual scale. Illustrative of ENSO effect is 1992 which was a strong El Niño year and, as expected, precipitation was above average during the dry season (+645 mm anomaly) and below average during the rainy season (−294 mm anomaly); the opposite pattern was observed during 2011, a strong La Niña year (−110 mm and +488 mm anomaly during the dry and wet season, respectively). This highlights the influence of the ENSO phenomenon in the rainfall regime during both seasons.
Interestingly, we found a positive and significant correlation between the number of hurricanes and major hurricanes in the NEP and MEI, Niño3.4, PDO and NAO; and a negative and significant correlation with AMO. We also found such correlations between these ocean-atmosphere oscillations and ACE (Table 2).

| Long-term trends
Over the analysed period, we found a significant increment in most climatic variables. Mean annual temperature increased by almost 2 C (Figure 2) while Tmax increased by 1.93 C and Tmin by 2.54 C (Figure 2). All increasing trends in temperature were statistically significant ( p < 0.001) for both simple linear regression and Man-Kendall trend test. As a result of the temperature increment, we also found a rise of 283 mm in mean annual PET (Figure 3).
A positive increment in MAP is also detectable despite its large interannual variability (Figure 3). The total increase during the studied period was 444 mm or closely 50% of the MAP for the last four decades. With increasing rainfall there is also an increase in runoff, which reaches a total of 145 mm from 1983 to 2015 (Figure 3).

| Shifts in seasonal water availability
Precipitation in the region has two clearly distinctive seasons: the wet season, which ranges from June to October and has an average rainfall of 726 ± 241 mm (87% of MAP) and the dry season, which last from November to May and has an average rainfall of 110 ± 164 mm. Mean seasonal precipitation increased significantly during the wet season at a rate of 8 mmÁyear −1 , which accounts for a total increment of 337.2 mm (or 75% of the annual trend) (Figure 4 and Table 3) and an increase in AET during the wet season was also found (Table 3). However, no detectable trend was found over the dry season for either rainfall or AET. Correspondingly, during the dry months only 9% of total annual runoff was recorded (average 8 ± 28 mm) while a contrasting 91% was recorded during the wet season (average 91 ± 117 mm). We did not find an increase in the number of rainy days (those with more than 0.1 mm of rain), but there was an increase in the number of large storms (>50 mm) throughout the year ( Figure 5).
While the wet season is getting wetter, changes in the dry season are not so clear. We found that the increase in temperature (mean, maximum and minimum) is higher during the dry season (Table 2), which in turn leads to higher PET increases. Such an increase is not significant for AET, likely because this variable is highly tied to precipitation, which is also nonsignificant during the dry season.
When analysing water balance, no statistically significant trend was detected in either season when evaluated with PET, but a decreasing trend during the first three decades of the study was identified during the dry season ( Figure 6). However, an increasing trend was found during the wet season when using P-AET, which is likely a response to the increase in rainfall (Table S2).

| Influence of tropical cyclones on the rainfall regime
TCs, which includes tropical depressions, tropical storms and hurricanes, play an important role in the precipitation regime in our study area. Officially, hurricane season in the NEP last from May 15 to November 30. However, there are records of storms as early as April 26 in 1983 and as late as December 6 in 2020.
TCs are common in the Pacific coast of Mexico with an average 18.62 eventsÁyear −1 . From 1982 to 2021, there were 745 tropical cyclones in the NEP of which 13% were tropical depressions (maximum sustained surface winds of <34 knots), 39% were tropical storms (34-63 knots) and 48% were hurricanes (>64 knots), nearly half of these (51%) were major hurricanes (category 3 or higher in the Saffir-Simpson scale). Out of the 745 tropical cyclones, 95 made landfall, mostly in the  (Table 4).
According to our calculations, TCs contribute an average 31% of the mean annual precipitation in the Southern Coast of Jalisco, but this contribution was as low as 4% in 1985 and as high as 59% in 1995. Although there is a positive correlation between MAP and the annual TC-related rainfall (Spearman's rho 0.644, p < 0.01), TC-related rainfall alone does not explain the occurrence of extremely wet or dry years. However, we found a positive correlation between the number of days when TCs were 500 km or less from Chamela Biological Station and the precipitation recorded during the entire rainfall season (Spearman's rho = 0.377, p < 0.05). We evaluated the occurrence of TCs and annual ACE from 1982 to 2021 to determine if there was an increase in the frequency or intensity of these events but no significant trend was found. Interestingly, there was an increase in the number of landfalling TCs and we found an increasing trend in precipitation associated to TCs but no trend was found for non-TC-related rainfall during the wet season (Table 4).

| Extreme climate events
We identified two different extreme events in relation to temperature and hurricanes. In terms of the first, we found that the five hottest years on record occurred in the second half of the analysed period; these years were 2015 (+1.09 C anomaly), 2009 (+0.99 C), 2014 (+0.86 C), 2006 (+0.86 C) and 2004 (+0.79 C). During these years, Tmax was equal or above 35 C for at least 66 days (in 2006) and for as much as 94 days (in 2009 and 2014). For comparison, the average number of days with this condition was 35 ± 1 days during the 1982-2021 period.
In terms of the second type of extreme, TCs worth mentioning are hurricanes Jova 2011 and Patricia 2015 which made landfall near Chamela Biological Station as category 2 and 4 systems, respectively. During 2011, there were 13 TCs in the NEP and an annual ACE of 124.9; TCs that year contributed 30% of the annual precipitation. Contrastingly, during 1992, which is the most active hurricane season on record in the NEP (27 TCs of which 9 were major hurricanes), the ACE was 282.7 and TCs contributed with only 5% of the annual rainfall.
Hurricane Patricia in 2015 is the only hurricane in the NEP to make landfall as a category 4 system in the last four decades. This year corresponds to the 2nd most active hurricane season in the NEP, with an ACE of 260.3 and a recorded 22 TCs of which 10 were major hurricanes. This year, TCs contributed 23% of annual rainfall.

| Mean climate and temporal trends
Values for mean annual temperature and precipitation described in this study are similar to those reported previously by Bullock (1986), García-Oliva et al. (2002) and Maass et al. (2018). Likewise, increasing trends in temperature had been reported by García-Oliva et al. (2002) who identified a 1 C increase in MAT from 1982 to 2000. The previous rising pattern (0.052 CÁyear −1 ) has maintained similarly from 1982 to 2021 (0.047 CÁyear −1 ). Gavito et al. (2014) stated that the temperature increases in the Chamela region from 1977 to 2013 were slight and gradual, and they detected no significant trend in their data. However, our results suggest a clear and worrisome increasing temperature trend, which is like that reported on a national scale (Murray-Tortarolo, 2021). The rate of increase is comparable to that observed globally under human-caused global warming due to greenhouse gas emissions (IPCC, 2021), which suggests that increasing local temperature trends are likely related to this global process.
High interannual variability in precipitation and runoff at the study area makes it difficult to detect trends in these variables; however, an increasing trend in MAP and mean annual runoff were recently noted by Maass et al. (2018). Precipitation increases are consistent with the IPCC report (IPCC, 2021), where an increasing trend in heavy precipitation and flooding is expected at 1.5-2 C global warming for the North, Central and South American regions. Particularly, our results suggest that precipitation in the Chamela-Cuixmala forest could be driven by more frequent intense events of rainfall, likely related to TCs influence in the region, than by an increment in total rainfall events.

| Seasonal trends
A defining characteristic of TDFs is its rainfall seasonality. Rainfall seasonality patterns found in our study area are similar to those previously described by Bullock (1986), Maass and Burgos (2011) and Maass et al. (2018). Interestingly, the increment in wet season rainfall has also been observed at a national scale (Murray-Tortarolo, 2021).
Late rainfall events can extend the rainy season period, sometimes up to 8 months, as happened in 1991, 1992. On the other hand, the steady increase in temperature drives an increase in PET, which in turn leads to an increased water loss via evapotranspiration. Maass and Burgos (2011) pointed out that transpiration is the main mechanism by which this forest loses water and warned that such a trend could increase with continued global warming. We found that P-PET during the dry season decreases significantly during the first three decades of study and increases in the last decade. This is likely due to the overall increase in mean annual precipitation, as also suggested by the observed increase in P-AET during the wet season and, particularly, due to the fact that several years in the last decade exhibit above average precipitation during the dry season (Figure 4, top panel), which might be compensating water loss via transpiration.
Overall, it will be necessary to closely monitor rainfall, PET and ET patterns in the upcoming years to recognize changes in water availability, particularly as global drylands are expected to expand during this century (Feng & Fu, 2013) and because water is recognized as a limiting resource in the region . However, if the seasonal increase in precipitation continues to accentuate, it is likely that water availability during the wet season increases. This is consistent with the observation by  that rainfall magnitude, duration and timing have increased in several seasonally dry tropical regions.
Altogether, our results suggest that the already marked seasonality in the Chamela-Cuixmala region has T A B L E 4 Tropical cyclones climatology in the NEP  # (% of total TCs) Average ± SD become more contrasting, a pattern likely to intensify in the future with drier dry seasons and wetter wet seasons (Murray-Tortarolo et al., 2016). This trend will be importantly influenced by the expected increase in both La Niña and El Niño extreme events (Cai et al., 2014(Cai et al., , 2015a(Cai et al., , 2015b, and other ocean-atmosphere oscillations could also influence seasonal rainfall patterns. Interestingly, we found a positive and significant correlation between AMO and precipitation during the wet season, and between PDO and precipitation during the dry season (Table S1). To the authors' knowledge, such correlations had not been previously described for our study region, but Montero-Martínez et al. (2022) reported an increase in precipitation and temperature during AMO positive phase in central Mexico. Also, Elder et al. (2014) and Park et al. (2017) described an influence of AMO and PDO in drought events in Mexico, which stress the importance of keeping track of these ocean-atmosphere oscillations in future assessments of local climate trends.
It should be noted that while seasonality in our study area seems to be accentuating, this trend is not uniform in other regions of the tropics  and worldwide (Feng & Zhang, 2015;Greve et al., 2014;Murray-Tortarolo et al., 2017), which is why seasonal studies of local climate conditions should be continuously assessed as more data is available.

| Influence of El Niño-Southern Oscillation and tropical cyclones on the rainfall regime
Explaining the precipitation regime during the rainfall season necessarily implies accounting for ENSO and tropical cyclones' influence. On the one hand, we found a negative correlation between the hot phase of ENSO and the amount of rainfall during the wet season, a pattern that had already been described by García-Oliva et al. (2002) and Maass et al. (2018), and which is consistent with patterns described for Mexico south of latitude 22 (Bravo-Cabrera et al., 2017;Caso et al., 2007). On the other hand, we confirmed the positive correlation between El Niño and TCs in the NEP described by previous authors (Martínez-S anchez & Cavazos, 2014;Romero-Vadillo et al., 2007). Such correlations may seem contradictory, since one would expect an increasing number of tropical cyclones to be accompanied by an increase in rainfall during the wet season. However, Fu et al. (2017) found that TCs activity during El Niño years undergoes a westward shift, and both genesis and track density increases in the west portion of the Northeast Pacific basin (west of 116 W), while decreasing in the eastern portion (see also Lin et al., 2020). This could explain why increasing TC activity does not necessarily result in an increasing rainfall in the Mexican Pacific coast and overall explain our findings of decreasing precipitation during the rainfall season in El Niño years (as depicted in Figure 1).
Evidence from Fu et al. (2017) also indicates that during La Niña years, the opposite pattern from El Niño years is observed, which results in TCs activity switching eastward of the NEP, that is, closer to the coastline. Moreover, Dominguez et al. (2020) found that during La Niña years there is an increase in TCs related rainfall and in the number of days with extreme precipitation related to TCs on the Mexican coast of the central Pacific.
Furthermore, we identified an increasing trend in the TC-related rainfall and an increase in the number of landfalling TC events. Overall, these results point to TCs as the main cause for the increasing rainfall pattern during the wet season. Interestingly, and in line with the results of other authors (Dominguez et al., 2020;Fu et al., 2017;Lin et al., 2020;Magaña & Dominguez, 2017), it is not the number of TC events that determines the local rainfall regime, as exemplified by the cases described in Table 5, but their trajectories and closeness to the coastline.
Damages from landfalling hurricanes were extensively described for the Chamela-Cuixmala TDF after hurricanes Jova in 2011 and Patricia in 2015 . While the TDF was highly capable of recovery after these disturbance events (Bhaskar et al., 2018;Holm et al., 2017;Martínez-Yrízar et al., 2018), several authors have also recognized that if such events become more frequent, and with other stressors in play, the forest capacity to recover might be hindered (Curry, 2021;Siyum, 2020). For example, after hurricane Patricia, which caused a massive deposition of woody debris unprecedent in long-term studies , two fires were recorded inside the Chamela-Cuixmala Biosphere Reserve in (Del Castillo, 2016Renton et al., 2018). This is a cause of concern, since TDFs rarely experience natural fires and forest species are not adapted to such events (Dexter et al., 2018). Besides ENSO, other ocean-atmosphere oscillations also play an important role in TC behaviour. PDO, in particular, has been found to have an influence on the periodicity of TCs activity (Pazos & Mendoza, 2013) and landfalling events (Raga et al., 2013). Importantly, we found a positive correlation between MEI and PDO (Spearman's rho 0.678; p < 0.05) and other authors reported that a warm phase of PDO can strengthen an El Niño event (Huang et al., 2021;Wang & Liu, 2016). Moreover, interbasin teleconnections with the North Atlantic also modulates TC activity in the NEP. Zhang et al. (2022) found that a NAO positive phase promotes TC genesis frequency in the North Pacific. Conversely, several authors described a decrease in TC genesis in the North Pacific when AMO is in a positive phase (Gong et al., 2021;Patricola et al., 2016;Wang et al., 2022).
It will be important to further analyse the relation between ocean-atmosphere oscillations and TCs in the NEP, since timescales and dominant phases of several of them are expected to change in a warmer climate (Michel et al., 2020;Rind et al., 2005;Zhang & Delworth, 2016). Changes in the dynamic of these variables could have important implications for major hurricanes and landfalling TCs in the Mexican Pacific coast.

| CONCLUSIONS
Several climatic shifts are taking place in the Chamela-Cuixmala TDF, which are identifiable in long-term analysis of climate trends. Mainly, we recognized an annual increase in temperature and an increase in precipitation that is limited to the wet season; such an increase seems to be associated with an increasing of the number of large storms. As a result, there is also an increase in runoff. It is important to point out that greater precipitation does not necessarily imply greater water availability to local people or ecosystems. Intense storms generally exceed infiltration capacity of the soil and cause runoff and flooding events. Conversely, there is a decrease in water availability during the dry season, particularly during the first three decades of the study, but further data is needed to confirm this trend.
Our results suggest that it is necessary to monitor changes in the behaviour of ENSO and TCs due to global warming and climate change to further understand their relation with changes in local climate and water availability. This knowledge can be very useful to develop, along with the local communities, stakeholders, and government authorities, strategies and programs to better respond and adapt to changes. Moreover, the uncertainty associated with our rapidly changing world requires an adaptive management strategy coupled with long-term climate monitoring as necessary tools to navigate and understand several local processes in a more comprehensive way.