Introduction

Tree species of the Gondwanan Nothofagus genus are one of the most important floristic components of the temperate rainforests in countries of Oceania, and in Chile and Argentina in South America (36°–55°S)1,2. These species inhabit a large variety of environments, changing their abundance as a function of local climate, elevation, soil type, topography and disturbance regimes2,3. Several Nothofagus species from the temperate forests of South America have been used in tree-ring research including climate reconstructions4,5, forest dynamics2,6,7,8,9, forest dieback10,11, climate effects on tree-growth and physiological patterns12,13,14,15,16,17, insect outbreaks18,19, and disturbances such as volcanic eruptions, landslides and fires20,21,22. Concerning climate reconstructions, only three species have been used for this purpose, N. pumilio for temperature4,5, streamflow23 and precipitation reconstructions24, N. betuloides for temperature25 and the southern annular mode (SAM) reconstructions, and recently N. macrocarpa for soil moisture reconstruction26. In this regard, recent findings suggest that the growth of N. obliqua trees is sensitive to precipitation and temperature in both Chile and Argentina16,27. However, most N. obliqua tree-ring chronologies in Chile span no more than 150 years16 and 300 years in Argentina27, although evidence indicates that the species’ life-span can reach approximately 500 years28. This represents a significant potential for developing long-term growth series, which could be valuable for climate reconstructions, forest ecology research, as well as archeological and historical studies.

To date, most climate reconstructions in South America have relied on Andean high-altitude species, such as Araucaria araucana, Austrocedrus chilensis, Fitzroya cupressoides and N. pumilio4,24,29,30,31,32. Additionally, for historical dating, tree-rings of Pilgerodendron uviferum has been used in southern Chile33, Prosopis spp. in the Atacama Desert34 and in the Central Andes35, among with Polylepis tarapacana36. However, the potential of Nothofagus species remains largely unexplored for these challenges so far. In this scenario, Nothofagus species present an untapped opportunity, particularly through the study of historical wood, which could provide valuable records for climate reconstructions and historical dating. This would allow the identification and dating of various historical structures, complementing existing datasets and offering new insights into the climatic and human history of lower-elevation regions, along with other environments where this wood may have been utilized by human communities.

Despite the extensive use of Nothofagus tree species for dendrochronological studies, only a few species have precisely dated and well replicated tree-ring width chronologies extending prior to 1700: N. betuloides (1489–2008)37, N. pumilio (1546–1990)5, N. dombeyi (1612–2019) and N. alpina (1634––2019)38. For N. obliqua, old-growth forests in the east slope of the Andes have allowed to build a tree-ring width chronology from 1706 to 201327. Regarding the use of relict deadwood to extend Nothofagus chronologies in temperate forests, only one attempt has been made with N. betuloides yielding very promising results37. This fact offers a new avenue to explore and extend current and future Nothofagus chronologies taking advantage of the preservation capacity and wood quality of many species from this genus. In this context, N. obliqua has exceptional wood mechanical properties and rot resistance, especially when the wood originates from old trees, reason why it has been intensively logged in most of its geographical distribution28,39,40,41,42. This wood has been widely used in the construction of houses, barns, bridges, beams, poles, and for firewood43,44,45, and it was the main species used for the elaboration of railroad ties in the early 1800s46,47. To this day, its logging is still legally allowed and it continues to be used for several uses, such as firewood and in construction41,48. The valuable and appreciated wood of N. obliqua has been used by human communities for a long time. Anthracological studies have found carbonized wood of N. obliqua in archaeological sites in southern Chile, showing its use as a raw material in fireplaces, residential structures and funeral rituals49. At the same time, ethnographic accounts show the continuity of its use in various structures of the Mapuche people, such as their dwellings50,51, suggesting an ancient and constant knowledge of the properties of N. obliqua wood. All these uses result in diverse wood pieces which can contain old growth patterns and provide an opportunity to extend growth chronologies in ausency of old growth forest in the present.

One of the main uses of N. obliqua in the last two centuries was the construction of the railroad across the country. Railroads have been a cornerstone of economic and social development in Chile since their introduction in the ninteenth century52. With nearly 9000 kilometers of tracks47,53, the railway supported the growth of agriculture, mining, and other industries, with the mining and nitrate industry standing out in the north54, and the forestry industry standing out in the south of the country52. By the late Nineteenth Century century, the main user of railroad ties in Chile was the National Railway Company (Empresa de Ferrocarriles del Estado, EFE)55, which, for example, during the 1960s consumed approximately 1,593,880 units per year of creosote-treated railroad ties, mainly from N. obliqua wood, for the line construction46, (Supplementary material, Fig. S1). However, the use of wood railroad ties has gradually declined and been replaced by concrete structures in the 2000s. Although still utilized in rural areas for various purposes, the production of wooden railroad ties has decreased since the 1970s, mainly due to the reduction of the N. obliqua forests near the exploitation sites46. This transition from wood to concrete ties has left behind a significant repository of historical N. obliqua ties, which not only reflect the forestry practices of the past, but also offer a preserved archive for dendrochronological studies. According to former railroad ties manufacturers’ stories in the Panguipulli area (~39°S), within the temperate rainforest region, smooth, knotless and healthy N. obliqua trees over 20 meters tall with large diameters (1–2.5 meters) were selected for tie production (Interviews with two former lumberjacks, Panguipulli, May, 2021). These tall and mature trees, referred to as “Pellín” by the Mapuche indigenous people56, had a significant portion of heartwood which is far more durable than sapwood in N. obliqua wood42. In contrast, younger N. obliqua trees, also called “Hualles”56, are composed mainly of less durable sapwood. For this reason, railroad construction was one of the main drivers of the extensive exploitation of old-growth forests of this species, contributing significantly to their decline, and also helps explain the current scarcity of old trees alive in the forest.

The historical information that we know about the railroad industry in Chile today, mainly comes from the work of chroniclers, naturalists and historians who documented timber exploitation from the late 16th century onward39,52. However, accurate dating techniques have not been applied to provide precise information about the ties construction, provenance and age of the wood used for this purpose in the country. The availability of N. obliqua old railroad ties constitute a unique opportunity to extend back in time the short and scarce tree-ring width chronologies of this species. Considering that the precise dating offered by tree-ring crossdating methods ensures that each ring can be attributed to a specific year with unmatched accuracy, it can be used for reconstructing detailed past environmental conditions and human activities57,58. Moreover, these ties themselves, become a new historical record that offers the possibility of dating and studying from different perspectives the historical dynamics of the construction, maintenance and repairs of the Chilean railroad network. Although there are no specific studies concerning the use of Nothofagus historical wood, research with species of the same order Fagales from the Northern Hemisphere, such as Fagus, Quercus, and Castanea, have demonstrated their usefulness in dating historical pieces and ancient structures in different parts of the world, including medieval buildings, shipwrecks, and historical artifacts59,60,61,62,63. These studies have highlighted not only the potential of dendrochronology for precise dating but also its ability to reconstruct historical climate conditions and human-forest interactions64,65,66. This global perspective underscores the potential of Nothofagus railroad ties to provide valuable insights into historical forestry practices and environmental changes in South America and the world.

Considering the dry conditions of arid and semi-arid regions in northern Chile (19–32°S), which help preserve wood and provide a unique opportunity to recover well-preserved N. obliqua samples from the ties of abandoned railroad tracks in the southern Atacama Desert, our study seeks to bridge the gap between historical forestry practices and modern climate research. The main goals of this study are: (i) to crossdate these historical wood samples with available N. obliqua tree-ring width chronologies, (ii) to extend back in time the length of the current N. obliqua tree-ring network, and (iii) to assess the climate signal of the new regional N. obliqua composite chronology and its potential to be used in climate reconstructions in Chile. Given the scarcity of old N. obliqua trees in the temperate rainforests of Chile due to historical overexploitation, the integration of these samples could significantly contribute to strengthening existing tree-ring chronologies, expanding the opportunity of rescue historical information of its geographical distribution, and enhancing climate reconstructions. The use of tree-ring series from railroad ties, particularly from lowland areas of southern Chile, presents a promising new avenue for historical, environmental and climate research, and could serve as a novel example for replication, offering valuable insights in other countries and ecosystems worldwide.

Methods

Study region: the Northern Longitudinal Railway (FCLN)

The study area includes removed, abandoned and/or ruined railway tracks from the Northern Longitudinal Railway Project (Ferrocarril Longitudinal Norte, FCLN) (Fig. 1A,B and 2H), located on the semi-arid southern edge of the Atacama Desert in Chile (Norte Chico, 25–32°S). This region experiences a mean annual precipitation of 150 mm and a mean annual temperature of 15.5 °C (Fig. 1C,D). The semi-arid and desert environment of the transverse valleys in Norte Chico (Fig. 2A) has facilitated the preservation of wooden railroad ties. The FCLN railway line originally spanned approximately from 20° to 32° S, according to the former layout of the tracks (Fig. 1A). The first tracks were privately built in an east-to-west direction by mining companies to extract material from mines and transport them to coastal ports for export. The Copiapó-Caldera line, which began operations in 1851, is an example of these early tracks54. These initial tracks were built in an isolated and disconnected manner. The government’s FCLN project aimed to connect northern Chile with the capital, Santiago, by integrating privately owned railway branches into a national network52,54. Construction began around 1888 but faced 20 years of delays due to financial limitations and political resistance from influential landowners in the Parliament43,52. By the early Twentieth century, the FCLN was entirely managed by the National Railway Company (EFE)54, reaching the northern part of the country where the nitrate and mining industries thrived, particularly in the Tarapacá and Antofagasta regions (19–25°S). This railway system facilitated the transport of both freight and passengers. However, since the mid to late Twentieth century, the railway network gradually declined as tracks were systematically suspended, primarily due to political, technical and economic challenges. Major reasons included decreased mining output, the end of the nitrate industry, annual deficits caused by high operating costs in rugged terrain, reduced goods and products traffic, the low-grade minerals being transported, inadequate infrastructure, and obsolete equipment, but mostly, lack of political-technical will54. The competition from the trucking industry, boosted by continued government investment in road infrastructure since 1945, the end of state subsidies during Augusto Pinochet’s dictatorship in the 1970s, and the reprivatization of the northern railway system in 1997, among other factors exacerbated the cease of operation of many railway lines53,54. This culminated in a widespread crisis for EFE47, leaving behind a legacy of abandoned infrastructure, including numerous tracks and N. obliqua railroad ties. Today, only a few branches of the railway remain active for freight transportation.

Fig. 1
figure 1

(Source Esri, GEBCO, Garmin, NaturalVue | Esri, GEBCO, IHO-IOC GEBCO, Garmin, NGS), both publicly available for their use.

(A) Map of Chile indicating the distribution of N. obliqua forests across south-central Chile, the locations of existing N. obliqua tree-ring width (TRW) chronologies (red triangles), and the railroad network with the sampling sites located in southern Atacama Desert (yellow triangles), (B) map of the Northern Longitudinal Railway Project (FCLN) and the three sampling sites. Red-colored lines represent disused railway tracks, while black lines represent abandoned tracks, (C) and (D) Precipitation and temperature descriptions of the study regions, (E), (F) and (G) zoom-in of the railroad ties sampling sites. All maps were generated using open-source software QGIS version 3.28.15 (https://qgis.org). Base maps were obtained from Google Satellite (Maps data ©2025 Google, Terrametrics) and ESRI Ocean

Full size image
Fig. 2
figure 2

(A) The semiarid landscape of Norte Chico region along the railway at the southern Atacama Desert in Chile where railroad ties sampling was carried out, abandoned railroad tracks in (B) Ovalle (OVA) and (C) Alcaparrosa (ALC), (D) illustration of the different orientations of ties manufacturing in a cross-section of a N. obliqua trunk with its annual growth rings, (E) detail of railroad ties sampling with a chainsaw, (F) visualization of N. obliqua annual growth rings from a railroad tie wood section, (G) Temperate rainforest biome in southern Chile, with Riñihue, Panguipulli and Neltume lakes from left to right, a region which was an epicenter for the manufacture of N. obliqua railroad ties, and (H) part of the cultural heritage consisting of the abandoned Alcaparrosa tunnel and station infrastructure (notice the water crane manufactured in Bath UK 1906). Pictures A and H from Cristian Campos (https://www.campografia.cl/alcaparrosa/).

Full size image

Sampling sites for railroad ties were pre-selected using the railroad database from the website www.amigosdeltren.cl to identify abandoned sections of the line. These locations were later verified in the field prior to the sampling. Samples were collected from three locations within the Norte Chico region: Ovalle (OVA), Alcaparrosa (ALC) and Chinchillas (CHI) (Fig. 1).

Sampling, processing and measuring of railroad ties samples

At each site, a minimum of 15 ties were selected from abandoned rail tracks in a well-preserved state that could contain a reasonable number of growth rings (Fig. 2B,C). Depending on the primary cutting elaboration technique the railroad could include more or less annual rings (Fig. 2D). Transverse sections of railroad ties approximately 10–15 cm width were cut using a chainsaw (Fig. 2E). These sections were stored in individual plastic bags, labeled and transported to the laboratory. Broken or shredded samples were glued with soluble wood glue to avoid deterioration during handling. The exposed surface of the samples (transverse plane) was sanded with progressively finer sandpaper (50–800 grit) following standard dendrochronological protocols67, to maximize the visibility of the rings under 10–50× magnification (Fig. 2F). Prior to sanding, samples with high creosote content—a toxic substance used for wood preservation52—were discarded to prevent handling and mitigate potential health risks associated with this toxic compound68. Annual growth rings were first identified as N. obliqua wood by comparing wood anatomy cuts from railroad ties with reference samples and their description69, and then visually dated under a high-resolution stereomicroscope, assuming the first ring as zero in each sample (undated). The ring-width was measured using a microscope coupled to a Velmex Inc. stage measuring system with an accuracy of 0.001 mm. Wherever possible and depending on the difficulty and definition of the growth rings, multiple radii were measured in cross-sections, visually matched for accuracy, and then averaged to create an overall mean series for each individual sample. None of the samples presented bark, but some presented a waney edge indicating the curvature of the last growth ring.

N. obliqua composite reference tree-ring width chronology

Considering the unknown provenance and temporality of the railroad ties, and in order to ensure a robust basis for cross-dating these series, we developed a composite reference N. obliqua tree-ring width chronology. This reference chronology combined eight ring-width chronologies previously developed by Urrutia-Jalabert et al.16 and two new ring-width chronologies, all from the temperate rainforest region of Chile (35°7’S–39°6’S, Fig. 2G). The two new chronologies were developed by sampling living trees in remnant forests from the Nahuelbuta National Park in the Coastal Mountain Range (37°47’S–72°59’O, Fig. 1A). At each site (NBT and NBR), tree-cores from 17 and 20 trees, respectively, were collected between 2021 and 2022 using increment borers following standard dendrochronological techniques67. In the laboratory, samples were dried, sanded, and dated using the Schulman’s convention for the Southern Hemisphere70, which assigns each ring the calendar year corresponding to the start of radial growth (austral spring). Ring-widths were visually cross-dated, measured, and annually verified using the COFECHA software71. Ring-width chronologies were developed using the ARSTAN_41d software, with standardization performed through the application of negative exponential or linear regression curves72,73. To check if there was a clear regional tree-growth signal for the complete N. obliqua distribution, we assessed the relationship between the 10 chronologies using the Pearson correlation coefficients (see Supplementary material, Fig. S2). Once this was validated, we were able to build a composite N. obliqua ring-width chronology, which was used as the reference series to date the railroad ties (Table 1, Fig. 1). This reference chronology was built using the measurements of the 326 samples from the 10 study sites (Table 1). Tree-ring width measurements were detrended using a spline curve with a 50% cutoff, and finally averaged to obtain a regional reference tree-ring width chronology of N. obliqua from living trees.

Table 1 Location and characteristics of the N. obliqua tree-ring width chronologies from living trees growing across the species distribution in the temperate rainforest region and the composite reference chronology.
Full size table

Dating tree-ring series from railroad ties

Once the rings of the railroad tie samples were measured, the exact annual crossdating of the series with the reference chronology was verified at a high-frequency level using COFECHA. We used lagged correlations between the measured tree-ring series after detrending with a 32-years spline and removing its temporal persistence by autoregressive modeling71. The statistical validation of the comparison between the ring-width series from the railroad ties and the standard reference chronology was conducted using the correlation coefficient and Student’s t-value between the high-frequency standardized tree-ring time series. The “t-value” assesses the strength of the association between two tree-ring width series, adjusting the correlation coefficient by sample size74. A t-value over 3.5 suggests some degree of match with a small margin of statistical error74,75, while a t-value exceeding 6.0 indicates a robust and conclusive agreement76. Once the floating series were assigned a chronological starting year, a regional N. obliqua tree-ring width chronology was developed by combining the series from both living trees with dated railroad ties, thereby strengthening and extending the chronology further back in time. The regional composite chronology was built using the ARSTAN software72, following the same procedure previously described for the reference chronology.

To verify the tree-ring dating of the N. obliqua railroad ties chronology we used annually resolved radiocarbon series following a “Miyake-like” wiggle matching approach77,78, but using the “bomb-peak” period as the reference event when atmospheric radiocarbon levels almost doubled due to the large number of atmospheric thermonuclear bomb tests detonated in the Northern Hemisphere79. For this purpose, we choose the previously ring-width dated ALC002 N. obliqua railroad tie sample (which covered the period 1640–1981) and selected six consecutive annual rings corresponding to years around the bomb-peak period (1963–1968; with Schulman’s convention70 annual rings 1962–1967). Then, the 14C content of each of the 6 years were determined and the sequence wiggle-matched against the single-year-resolution reference curve of the 14C content in tree-rings from the Southern Hemisphere zone 1-2 developed by Hua et al.78,80,81. To determine its exact calendar year, we used a classical Chi-squared (χ2) fit of the 14C 6 years railroad tie sequence to the 14C reference curve in conjunction with 10,000 Monte-Carlo simulations to obtain a goodness of fit and a realistic range of possible solutions77. Besides being the largest tree-ring sample, the ALC002 presents the most recent annual ring of all the tie series and presents in one of its corners a classical waney edge indicating a bark edge78. The ALC002 sample preparation for 14C content determination was done by cutting each of the six selected dated annual rings under a binocular microscope using sterile ceramic knives over an acrylic table, polished with diamond tools to remove the surface that was in contact with extraction and sanding tools, stored in falcon tubes and sent to DirectAMS, Washington, USA. All 14C determinations were carried out over the α-cellulose extracted from wood samples to avoid translocation between growth rings82 and following standard procedures83.

Climatic signal in the regional composite N. obliqua chronology

To assess the spatial and temporal relationship between the regional composite N. obliqua tree-ring width chronology (comprising both living trees and historical railroad ties) and the climate variability, we analyzed its correlation with various climatic datasets and paleoclimatic reconstructions. This assessment was conducted using both internationally validated climate products and long-term hydroclimate reconstructions (all published in scientific journals), allowing us to explore N. obliqua sensitive to past and present climate variability.

To assess the climatic signal in the tree-ring width chronology and its spatial representation, we generated correlation maps for 3, 6 and 12-months windows between the standard chronology and the high resolution 0.25° × 0.25° latitude-longitude gridded precipitation and maximum temperature data from ERA584 for the period 1950–2020. These time windows were chosen based on the observation that N. obliqua exhibited significant correlations with climate variability across multiple seasonal combinations. For the period prior to 1950, and to validate the climate signal of the chronology including railroad ties, we compared our chronology with longer climate datasets extending back to 1901, such as the CRU TS4.08 precipitation dataset (0.5° × 0.5° latitude-longitude)85. Correlation analyses were performed using this dataset for the period 1901–1950 to verify the consistency of N. obliqua’s response to precipitation across different climate products. To further ensure coherence, we conducted correlation analyses for the period 1950–2020, allowing for direct comparison with ERA5 data, and for the entire common period covered by CRU TS4.08 and the N. obliqua chronology (1901–2020).

To further extend the assessment, we used the 0.5° × 0.5° gridded South American Drought Atlas (SADA)31, which spans South America from 12°S to south, covering a period from 1400 to the present. The SADA is a tree-ring based field reconstruction (0.5° × 0.5°) of austral summer (December–February) soil moisture, using the self-calibrated Palmer Drought Severity Index (scPDSI) as a target metric, developed using more than 300 tree-ring width chronologies from over 10 tree species as predictors, not including N. obliqua. Therefore, our regional composite chronology was completely independent from this reconstruction. However, given that our chronology reaches an EPS > 0.8 only from 1740 onward, we restricted the use of the SADA dataset and further analyses to the period 1740–2020, ensuring that all analyses were conducted with robust population signals. The Expressed Population Signal (EPS) statistic measures the common variability within a chronology which is dependent upon sample depth86,87. When the EPS value drops below a predetermined level (often 0.85), the chronology begins to reflect individual tree-level signals rather than a coherent stand-level signal87. Nevertheless, it can still be reliably dated and used in archaeological studies87. Also, some studies argue that an EPS >0.8 is sufficient to represent a strong population signal88,89,90,91. Based on this criterion, we truncated the composite chronology at 1740, ensuring that all subsequent analyses were based on data with EPS > 0.8.

Following this approach, we performed correlation analyses between our chronology and the precipitation reconstruction for Northern Patagonia92 for the periods 1900–1988, 1800–1988 and 1740–1988. Additionally, we examined its relationship with streamflow reconstructions located in different parts of the geographical distribution of N. obliqua, including the northern area represented by the Maule river93, south-central area using the Biobío94 and Imperial rivers95, and the Puelo river representing the southern area of the species distribution96. The streamflow reconstructions from these rivers were correlated with the N. obliqua chronology for the periods 1900–2000, 1800–2000 and 1740–2000.

Finally, we assessed the correspondence between historical extreme climate events documented in historical records and our chronology. For this purpose, we followed the methods and techniques of historical climatology, which involve the search and compilation of information from periods preceding or parallel to instrumental records97,98,99. We consulted various historical sources, including chronicles, archival documents, newspapers and magazines, obtained from libraries, archives and museums in major cities of central-southern Chile (Santiago, Temuco, Valdivia, Puerto Montt), covering the period from the arrival of European colonizers to the present. This information was systematized and classified into years of drought (dry years) and high-rainfall (wet years). The search covered central-southern Chile from 1619 onward; however, for this study, we only considered climatic events between 1740 and 1920, given the EPS constraints and the availability of precise instrumental data in more recent years.

Results

Composite reference tree-ring width chronology

The composite reference N. obliqua ring-width chronology from living trees had a total of 326 series with an intercorrelation of r = 0.407 (p < 0.01) (Table 1). The span of this chronology ranged from 1627 to 2020, though the low number of old trees allowed a well replicated chronology since 1759 (n > 5 tree-ring width series), but reaching an EPS > 0.85 since 1780, with a marginal decline in the period 1805–1830. Moreover, the intercorrelation values between the five ring-width series that covered the period 1760–1809 was r = 0.51 (p < 0.05) (with series from NBR and MEL site), demonstrating the coherence of the growth patterns in the composite chronology even with a low number of ring-width series. The two new chronologies developed in the Nahuelbuta Mountain Range (codes NBT and NBR) included the oldest living trees of N. obliqua reported to date (Supplementary material, Fig. S3), playing a crucial role for strengthening the composite chronology of the species.

Tree-ring dating of railroad ties

In total, 26 out of the 45 railroad ties series were dated: 12 from the ALC site and 14 from the OVA site. The high creosote impregnation of the CHI site samples hindered the accurate identification of the growth rings, preventing the dating of these samples. The temporal extension of the wooden pieces ranged from 23 to 345 years (Table 2, Fig. 3). Specifically, six samples were under 50 years old, nine were between 50 and 100 years, five ranged between 100 and 150 years, four were between 150 and 250 years, and only two were more than 300 years (ALC002 and ALC020) (Fig. 3). The samples covered different time periods, with the oldest starting in 1534 (ALC020) and the most recent ending in 1981 (ALC002) (Table 2, Fig. 3). Railroad ties series’ correlations with the reference chronology ranged from 0.175 to 0.532, and t-values ranged from 2.09 to 6.38 (Table 2). Only 7 out of 26 tree-ring series were cross-dated with the reference chronology before 1780, and still obtained good statistics in several samples (Table 2, Fig. 4). Each tree-ring series from railroad ties was cross-dated and plotted against the reference chronology (Fig. 4).

Table 2 Characteristics of the N. obliqua tree-ring series from railroad ties and corresponding dating statistics.
Full size table
Fig. 3
figure 3

Temporal extension of railroad ties tree-ring width series arranged chronologically. Horizontal bars represent the extension of each series. Vertical lines show the year since when the composite reference N. obliqua chronology from living trees has > 5 and > 10 series, respectively.

Full size image
Fig. 4
figure 4

Residual version of the living trees composite chronology (gray) and the railroad ties superimposed on it (each one named with individual codes). Colors in the time series represent the start period of the first ring for each railroad tie series (dark green = from 1533 to 1700, brown = from 1701 to 1800, and orange = from 1801 to 1915). r = correlations coefficients, and t = t-value statistics for the common period of railroad series and the reference chronology.

Full size image

The samples from railroad ties came from wood pieces of similar size, however, they exhibited a wide variation in the number of rings, resulting in growth series of different time lengths. This variability is attributed to differences in tree growth rates and variations in the wood-cutting orientation during the railroad tie manufacturing process (Fig. 2D). In most cases, this variation can be explained by the young age effect, where younger trees tend to exhibit wider rings mainly due to higher growth rates100, as observed in samples where the railroad tie was crafted from the youngest sections of the wood. This age-related growth pattern is well documented for N. obliqua, a pioneer species characterized by fast growth during its early establishment stages48,101. Thus, samples with a low number of years (e.g. OVA004 with 23 years) displayed higher growth rates (i.e. wider rings) compared to samples of similar size but with a significantly higher number of rings (i.e. narrower rings), indicating slower growth rates (e.g. ALC002 with 345 years), likely obtained from old trees. Only one rotten sample was found (ALC017) in the total sampling. Furthermore, no particular growth patterns were found with respect to each site, probably due to the mixed origin of the wood used in the manufacturing process. Future studies comparing growth patterns from different sites could help determine whether local growth patterns can be used to identify the provenance of the historical N. obliqua wood pieces.

Verification of dating by 14C wiggle matching

A total of six radiocarbon measurements were carried out on the ALC002 series, which was the largest tree-ring series analyzed in this study (period 1640–1981). The wiggle-match of the 6-year radiocarbon series from ALC002 against the annually resolved tree-ring 14C reference curve from the SH 1–2 region80, corroborated the exactitude of the ALC002 cross-dating by classical ring-width patterns. The lowest χ2 value of the wiggle-matching indicated that the 1962–1967 calendar period assigned by cross-dating was perfectly dated and corresponded to the same 14C period in the calendar years of the Hua et al.80 reference 14C curve (Fig. 5).

Fig. 5
figure 5

14C wiggle matching of the N. obliqua ALC002 tree-ring sample using as a reference curve the series of the 14C content in annual rings from the Southern Hemisphere zone 1-2 developed by Hua et al.80 (curve in red). Six consecutive years (blue) of the N. obliqua ALC002 series wiggle-matched to the exact calendar year during the bomb peak period. Inset plot: output of the Chi-square (χ2) Monte-Carlo test at 95% probability (horizontal orange line), t-statistic= 11.07; df = 6, the red dot indicates the lowest χ2 value (χ2 = 1.7; maximum probability) indicating the calendar year of the most recent year of the 14C N. obliqua series.

Full size image

N. obliqua regional chronology and its climate signal

After verifying the ties dating through cross-dating with the composite reference chronology and radiocarbon analysis, a regional ring-width chronology was constructed using all the growth series. By cross-dating both the undated series and the living tree-rings chronologies, it was possible to extend and strengthen the N. obliqua regional chronology, particularly in the oldest period (1600s) and especially since the Eighteenth Century. This improvement enhanced the correlation and increased the number of samples during this period, reaching between 10 and 18 series (Fig. 6). The final regional chronology spans from 1534 to 2020, comprising 353 ring-width series with an intercorrelation of 0.404 (p < 0.01). The EPS dropped below 0.85 in the earlier centuries of the chronology (prior to 1865) due to a decreasing sample size. However, since 1740 it remained above 0.80 indicating that these sections of the chronologies still contained a strong population signal.

Fig. 6
figure 6

(A) Comparison between the residual versions of the living-trees (gray) and railroad-ties (blue) N. obliqua chronologies, (B) sample depth of each chronology, (C) standard version of the N. obliqua regional chronology composed by tree-ring series from living trees growing in the temperate forests of southern Chile and railroad ties series collected in the southern Atacama Desert. The chronology was plotted from the year when EPS > 0.80. The red line represents a 20-year Gaussian filter.

Full size image

Among the spatial climate analysis, the new regional tree-ring width chronology showed significant positive correlations with late spring - early summer (November–January) precipitation (r values from 0.23 to 0.53) and significant negative correlations with maximum temperature (r values from − 0.23 to − 0.46) during the period 1950–2020 (p < 0.05, Fig. 7A,B). Similarly, significant correlations were found for 6 months (southern hemisphere spring-summer) and 12 months windows (winter–winter and spring–spring) for both variables (Fig. S4). The highest correlations were found in the temperate rainforest region from northern Patagonia, encompassing the Andes and the lowlands to the west in Chile, and extending toward the eastern Patagonia steppe in Argentina following the Arid Diagonal climatic pattern102. The spatial correlation patterns between the regional chronology and the SADA from 1740 to 2020, when EPS > 0.80, displayed the highest significant positive correlations (r values from − 0.49 to 0.42) across the N. obliqua distribution in the temperate forest biome in Chile, and also extending to the desert steppe at the leeward of the Andes in Argentina (Fig. 7C).

Fig. 7
figure 7

Spatial correlation maps between the regional N. obliqua tree-ring width chronology composed by tree-ring series from living trees and railroad ties, and: (A) late spring - early summer (November–January) ERA5 precipitation during the 1950–2020 period, (B) mean late spring - early summer (November–January) ERA5 maximum temperature during the 1950–2020 period, and (C) Austral summer self-calibrated Palmer Drought Severity Index (scPDSI) field reconstruction indicated by the South American Drought Atlas (SADA) for the 1740–2020 period. The black dashed lines in panel C indicate the approximate N. obliqua forest distribution in Chile. Only significant correlation values (p < 0.05) are shown.

Full size image

Regarding the relationship between the regional chronology, historical precipitation records and climate reconstructions, our results indicate that the new regional chronology of N. obliqua successfully records some historical drought events from central-south Chile (Fig. 8A). For example, documented droughts such as the megadrought of 1770–1782103,104,105 coincide with a decline in tree-ring width (Fig. 8A). Drought years that also coincided with a decline in tree-ring width were 1791–1792 and 1800–1803106, as well as of 1869–1876 (a period of relative dryness interspersed with normal years)104,106,107,108, and of 1907–1917 (except for 1914, a wet year)106,108. Apparently, wet years were not particularly reflected in the tree-ring chronology (Fig. 8A).

Fig. 8
figure 8

Comparison of the regional composite chronology with historical precipitation records and climate reconstructions. (A) Standard composite chronology (black line) smoothed with a 20-years Gaussian filter (red line), alongside historically identified dry years (yellow bars) and wet years (blue bars), according to Ortlieb98, Steiger et al.103, Vicuña Mackenna 104, Taulis106, Camus & Jaksic107, “El Diario Austral de Temuco” newspaper108, Carvallo Goyeneche109. Comparison of the composite chronology with: (B) precipitation from the CRU TS4.08 dataset (1901–2020) (blue line) and the precipitation reconstruction for Northern Patagonia (1740–1988) (magenta line) by Villalba et al.92; (C) Maule River streamflow reconstruction (green line) from Urrutia et al.93. In both panels, significant correlation values (r (p < 0.05)) are shown for three different periods.

Full size image

The correlation analysis between the regional chronology and precipitation data from the CRU TS4.08 dataset for the period 1901–2020 yielded an r value of 0.45 (p < 0.05), being stronger in the second half of the century (1950–2020, r = 0.46, p < 0.05), compared to the early Twentieth century (1901–1950, r = 0.42, p < 0.05) (Fig. 8B). Regarding the Northern Patagonia precipitation reconstruction from Villalba et al.92, which overlaps with our regional composite chronology from 1740 to 1988, it showed significant correlation values across periods: r = 0.25 (1740-1988, p < 0.05), r = 0.31 (1800–1988, p < 0.05), and r = 0.51 (1900–1988, p < 0.05) (Fig. 8B).

Finally, correlation analyses with the available streamflow reconstructions within the N. obliqua distribution range—encompassing the Maule93, Biobío94, Imperial95, and Puelo96 rivers—, yielded varying results. The Maule river streamflow reconstruction displayed the highest correlation values (Fig. 8C), with r = 0.22 (p < 0.05) for the entire period (1740–2000), r = 0.28 (p < 0.05) for 1800-2000, and r = 0.46 (p < 0.05) for 1900–2000 (Fig. 8C). In contrast, correlations with the Biobío and Puelo streamflow reconstructions were more moderate, with r = 0.16 (p < 0.05) and r = 0.15 (p < 0.05), respectively, for 1740–2000, and r = 0.35 (p < 0.05) and r = 0.31 (p < 0.05), respectively for 1900–2000. Correlations with the Imperial River streamflow reconstruction were not significant for the evaluated periods.

Discussion

Dendroarchaeological potential of historical N. obliqua wood

To the best of our knowledge, this study represents the first initiative dating tree-rings from railroad ties. The historical expansion of railways across Chile contributed significantly to the country’s connectivity and economic development in the Nineteenth Century and Twentieth century, while also led to the degradation of N. obliqua forests in Northern Patagonia39,52,54. This exploitation impacted regional biodiversity with some effects persisting to this day in an ecosystem type that remains underrepresented in Chilean protected areas110,111. Additionally, it has reduced the availability of old-growth N. obliqua forests and ancient trees, hindering the development of multi-century N. obliqua tree-ring width chronologies, with implications for long-term ecological, environmental and paleoclimatological studies in the region28,112. By integrating traditional crossdating ring-width methods with wiggle matching of radiocarbon series from railroad ties, this study successfully dated wood samples from these historical artifacts, significantly extending the regional N. obliqua tree-ring chronology in Northern Patagonia back to AD 1534. This newly extended chronology, comprising 353 growth series, enhances previously underrepresented periods, providing a more robust temporal framework. Furthermore, this chronology now constitutes the strongest reference series for dating N. obliqua wood, widely used in construction of houses, bridges, and various artifacts and infrastructure44,47,49, being a valuable new resource for studying past human wood-use technologies and historical environmental conditions.

The dating of railroad ties tree-ring series from southern Atacama Desert, using reference chronologies located further south within the species’ distribution in Northern Patagonia, resulted in the longest N. obliqua tree-ring record to date, and one of the longest for the genus worldwide. This underscores the regional environmental signal contained in N. obliqua tree-ring data, as evidenced by correlations with climate variables (Figs. 7 and 8). Although the obtained intercorrelation was slightly lower than those reported for other regional tree-ring chronologies of angiosperms in southern South America, such as N. pumilio113,114, this outcome may be partially explained by the heterogeneous spatial distribution of individual chronologies into the wide environmental amplitude were N. obliqua grows44, and the likely low-elevation origin of the railroad ties, were higher interspecific competition, combined with greater accessibility for human communities, has historically led to more intensive logging44. These factors can introduce greater growth variability, potentially weakening the climate-growth relationship compared to tree-ring chronologies from less disturbed or higher-elevation sites, such as N. pumilio113,114,115. Individually, the longest railroad ties tree-ring series showed relatively low correlations with the reference chronology, however we confirmed the dates by identifying the globally recognizable 14C bomb peak, a signal produced by the massive detonation of nuclear tests during the 1960s79.

Despite the cessation of passenger transport on the railway lines in the study region between 1975 and 1977, freight transport continued until 1987–199254. The last year recorded in our railroad tie chronology was 1981, indicating that the line was actively maintained and repaired at least until then. This finding underscores that the southern Atacama railway network may contain historical wood suitable for tree-ring studies, spanning from its construction in the late Nineteenth Century century through the early 1980s, offering a valuable resource for historical dendroarchaeology and environmental research (Fig. 2H). While historical dendroarchaeology has been widely applied in various regions of the world—such as North America, Europe and Asia—to date buildings, wooden artifacts, and cultural heritage sites61,116,117,118,119, examples from South America remain scarce. In historical archeology, tree-ring research has been used to date a whaling shipwreck from North America discovered on the Atlantic Patagonia coast and to determine the provenance of its wood materials76. Similarly, dendrochronology using Pilgerodendron uviferum has been employed to corroborate the years of the construction and, similarly to our results with railroad ties, allowed to identify periods of maintenance not recorded in historical documents associated to heritage churches on Chiloé Island in southern Chile33. Additionally, studies have examined tree-growth patterns in archaeological wood from Prosopis tamarugo and Prosopis alba from pre-Hispanic periods in the Atacama Desert34,35, and dated archaeological sites in the Central Andes back to the 13th century using Polylepis tarapacana tree-rings 36. Nevertheless, the use of historical wood from railroad ties represents an innovative approach in dendroarchaeology studies in South America and beyond, especially since old trees were historically selected for the manufacture of ties due to their wood properties42. It should be noted that Domínguez-Delmás et al.120 proposed the potential use of various wooden structures, including railroad ties, for dendroarchaeological studies in Spain’s Iberian Peninsula. This suggests that abandoned railway infrastructure in other regions may serve as valuable sources for similar research, and our study could help foster international research synergies in this field. Previous tree-ring research related to railways has primarily focused on reconstructing disturbance histories and tree ages in a railroad logging camp in Oregon121, or tracing the history and provnance of timbers used in constructing the Terminal Warehouse connected to the railway lines of New York118. Given that the Nothofagus genus is widely distributed across Chile, Argentina, Australia, New Zealand, New Guinea and New Caledonia122, and that N. obliqua railroad ties from Chile were exported to several countries including Germany and South Africa123,124 (Fig. S1), this study opens new avenues for tree-ring research using this historical railway wood in the Global South and beyond. Furthermore, additional species such as oak (Quercus spp.), which was widely used for railroad ties125,126,127 and has been extensively studied in dendrochronology64,65,66, could provide valuable new tree-ring records for this field. By extending tree-ring chronologies from railroad ties, future studies could focus on a variety of research topics, including paleoclimatic records, international historical timber markets, wood properties and species selection for railway ties, the extent of forest exploitation driven by this demand, the current status of remaining old-growth forest, and growth rates comparisons between historical and contemporary individuals of these species.

Dendroclimatic potential of N. obliqua chronologies

Utilizing historical wood samples can significantly extend and enhance existing tree-ring chronologies for environmental and climate studies128,129. This is especially important in regions where old-growth forests and ancient trees are scarce due to severe historical human disturbance, as is the case with N. obliqua forests. Our resulting composite chronology spans multiple centuries and reveals a clear hydroclimate signal across Northern Patagonia (Figs. 7 and 8). Also, significant correlations between the chronology and hydroclimatic reconstructions remark this attribute (Fig. 8).

Spatial correlations between the new regional composite N. obliqua chronology and gridded climate data (ERA5) showed significant positive patterns with precipitation and negative patterns with maximum temperature. These correlations were stronger than those reported by Urrutia-Jalabert et al.16 for specific site chronologies and weather stations, potentially due to a large number of samples that the new chronology has involved. The precipitation correlation map revealed a strong climate signal across the core of N. obliqua’s distribution, as well as in other Nothofagus species known for their sensitivity to hydroclimate variability3,12,13,15. The correlation map with maximum temperatures highlighted the detrimental effect of warmer summers on N. obliqua radial growth, consistent with previous findings for specific site chronologies16. The negative effect of spring-summer temperatures on growth has also been reported for other Nothofagus species in the area12,13,15,113. The strong and significant correlation patterns between our new regional N. obliqua chronology and a completely independent paleoclimate reconstruction of soil moisture (scPDSI) represented by the SADA31, spanning the last 280 years, confirm the consistent hydroclimate signal contained in this tree-ring record. All the spatial correlation patterns between our regional N. obliqua chronology and ERA5 precipitation and temperature (period 1950–2020), along with the SADA (period 1740–2020), consistently capture climate variability within the N. obliqua distribution area and extend across Northern Patagonia (Fig. 7).

To confirm the accuracy of the new N. obliqua chronology as a potential predictor for climate reconstruction, we examined the coincidence of extreme events identified on historical documents and validated climate reconstructions. Our historical analysis, combined with the comparison between the regional chronology and various precipitation and streamflow reconstructions (Fig. 8), reveal significant insights into temporal climate variability. On the one hand, and consistent with the positive relationship between N. obliqua tree growth and precipitation16 [and this study], our composite chronology effectively captured all the documented dry periods through a decline in ring-width (1764–1784, 1791–1792, 1800–1803, 1869–1876, 1907–1917, Fig. 8A). The latter period has been described as a prolonged dry decade, with 1914 being an exception as a wet year106,108, a pattern that is distinctly reflected in our tree-ring chronology (Fig. 8A). Unlike droughts, wet years were not clearly recorded in tree-rings, due to the decreasing sensitivity of tree growth to rainfall above a certain threshold, as documented by Villalba et al.92.

On the other hand, our analyses relating the regional composite chronology with hydroclimate reconstructions, further support these findings. A stronger relationship was observed between our tree-ring chronology and the northernmost studied river (Maule, Fig. 8C). The Maule River watershed is located within the north Mediterranean-temperate transition zone, where N. obliqua thrives at its northern distribution limit93. This streamflow reconstruction better reflects periods of water scarcity93, aligning with the drought signals in our chronology. The correlation decreases for southern rivers, possibly due to increased precipitation and reduced dependence on groundwater storage, which may buffer tree growth against short-term droughts130,131.

These findings highlight the potential of combining tree-ring width chronologies from living N. obliqua trees with historical wood from railroad ties to enhance the temporal coverage and improve data resolution for future climate reconstructions. This approach could significantly improve dendroclimatic studies using a species that has not been employed for this purpose before. Future research integrating this extended dataset could refine our understanding of long-term climate variability in the region and improve hydroclimatic reconstructions for southern South America, and at the same time could provide a new paleoclimatic time series to compare the climate variability between Andean and low-elevation areas.

Implications and future directions

The lack of precise documentation regarding the origin of the trees, felling dates, and the initial installation and maintenance of railway lines poses challenges that must be addressed. Our findings emphasize the importance of continuing research on these historical structures, as they represent a valuable data source for dendrochronological and environmental studies. N. obliqua from other historical infrastructures such as beams, bridges, tunnels and warehouses, could potentially provide even longer environmental records and historical information about other human activities in the past. Also, developing new N. obliqua chronologies from remaining relict old-growth forests in protected areas, would greatly improve the cross-dating of historical wood samples. The absence of the inner trunk arc in most analyzed wood samples—likely due to rooting—suggests that the original trees were significantly older than the rings we were able to date. Given the lack of information regarding the provenance of the railroad ties, future studies are encouraged to incorporate genetic analyses and advanced techniques such as isotopic signatures and chemical fingerprints132 to better determine the origin of the wood, facilitating a more precise identification of the source forests. Loader et al.133 suggest that the use of oxygen isotope chronologies could improve the dating quality of wood samples by refining the accuracy of complacent ring sequences in master chronologies. However, the contamination of the railroad ties with creosote and other compounds46 may present challenges for this approach. Creosote—an oily preservative derived from tar, petroleum or wood tar— was widely used in railroad manufacturing to increase durability and to protect the wood from rot, fungi and insect infestations46. Additionally, compounds soluble in organic solvents such as pentachlorophenol, and soluble salts such as zinc chloride or fluoride, copper sulfate, among others, were also extensively used in Chile during the 60s and 70s for wood preservation46. Once installed, many railroad ties may have been exposed to heavy mining load in northern Chile52, which could have introduced additional sources of contamination. The living trees used for railroad ties were likely not exposed to contaminants before processing46.

The railroad ties, previously central to the country’s railway infrastructure, now offer an important resource for environmental and earth science research, and can contribute valuable information for archeological, historical and socio-ecological studies. This places the railway’s significance as an essential part of Chile’s cultural heritage. The remaining past railway infrastructure, including abandoned railroad tracks in the Atacama Desert, are key for future tree-rings studies across south-central Chile, and constitute a unique archive for the study and the understanding of the historical relationship between humans and the forest.

Final remarks

The results obtained through traditional crossdating methods, strongly developed in dendrochronology field, corroborated by radiocarbon wiggle matching using the bomb-peak, and the presence of well-preserved tree-ring samples over 300 years old, underscore the scientific importance of this tree-ring heritage. The strengthened N. obliqua chronology developed here is not only valuable for climate reconstructions but also for socio-environmental studies on the forests of southern Chile and the human impacts on them, being one of the only vestiges to contain long-standing information about the growth of the trees in the past. However, the preservation of Chile’s railway heritage is currently at risk, threatened by the theft of railroad ties and the looting and disappearance of stations. This underscores the urgent need for protection measures to safeguard this historical legacy.

The abandoned railway network in parts of the Atacama Desert not only tells the history of Chilean industrialization processes and economic development, but also contributes significantly to socio-environmental research in ecology, anthropology, archaeology, history, heritage studies and geography. Preserving Chilean railway infrastructure contributes to safeguarding national cultural heritage and documenting the historical role of British engineering and mining companies in the region. Integrating railway history with tree-ring research offers an opportunity to connect Chile’s industrial past with contemporary efforts toward sustainability. The abandoned railroad ties in the Atacama Desert emerge as a witness of the human impact on earth at regional and global scales, reflecting the exploitation of native forests in Patagonia and the massive detonation of nuclear bombs in the Northern Hemisphere.