Climate as a Driver of Aboveground Biomass Density Variation: A Study of Ten Pine Species in Mexico
Abstract
1. Introduction
2. Materials and Methods
2.1. Characteristics of the Study Area and Species Selection
2.2. Database Acquisition
2.3. Aboveground Biomass Estimation and Data Cleaning
Species | Equation | R2 | n | Author |
---|---|---|---|---|
P. arizónica | 0.97 | 66 | [47] | |
P. ayacahuite | 0.97 | 58 | [47] | |
P. cembroides | 0.98 | 30 | [48] | |
P. devoniana | 0.98 | 20 | [49] | |
P. leiophylla | 0.93 | 27 | [47] | |
P. montezumae | 0.99 | 16 | [50] | |
P. oocarpa | 0.96 | 33 | [51] | |
P. patula | 0.99 | 25 | [52] | |
P. pseudostrobus | 0.99 | 20 | [49] | |
P. teocote | 0.99 | 56 | [47] |
2.4. Statistical Analysis
3. Results
3.1. Basic Comparative Analysis
3.2. Distribution Patterns and Climatic Tolerances
3.3. Influence of Bioclimatic Variables on Aboveground Biomass Density
3.4. Quantile-Based Analysis of AGBd Response to Climate Variables
4. Discussion
Biomass Distribution
5. Conclusions
Supplementary Materials
Author Contributions
Funding
Data Availability Statement
Acknowledgments
Conflicts of Interest
Appendix A
Variable Code | Description |
BIO1 | Annual Mean Temperature (°C) |
BIO2 | Mean Diurnal Range (Mean of Monthly (°C) |
BIO3 | Isothermality (BIO2/BIO7) (×100) |
BIO4 | Temperature Seasonality (SD ×100) |
BIO5 | Max Temperature of Warmest Month (°C) |
BIO6 | Min Temperature of Coldest Month (°C) |
BIO7 | Temperature Annual Range (BIO5-BIO6) |
BIO8 | Mean Temperature of Wettest Quarter (°C) |
BIO9 | Mean Temperature of Driest Quarter (°C) |
BIO10 | Mean Temperature of Warmest Quarter (°C) |
BIO11 | Mean Temperature of Coldest Quarter (°C) |
BIO12 | Annual Precipitation (mm) |
BIO13 | Precipitation of Wettest Month (mm) |
BIO14 | Precipitation of Driest Month (mm) |
BIO15 | Precipitation Seasonality (Coefficient of Variation) |
BIO16 | Precipitation of Wettest Quarter (mm) |
BIO17 | Precipitation of Driest Quarter (mm) |
BIO18 | Precipitation of Warmest Quarter (mm) |
BIO19 | Precipitation of Coldest Quarter (mm) |
Appendix B
Specie | Bioclimatic Variable | Rho | Specie | Bioclimatic Variable | Rho |
P. arizonica | P. oocarpa | BIO1 | 0.11 * | ||
BIO1 | −0.3 *** | BIO2 | −0.22 *** | ||
BIO5 | −0.21 *** | BIO3 | 0.16 *** | ||
BIO6 | −0.27 *** | BIO4 | −0.19 *** | ||
BIO8 | −0.21 *** | BIO6 | 0.16 *** | ||
BIO9 | −0.23 *** | BIO7 | −0.22 *** | ||
BIO10 | −0.21 *** | BIO9 | 0.11 * | ||
BIO11 | −0.25 *** | BIO11 | 0.14 *** | ||
BIO12 | 0.22 *** | ||||
BIO13 | 0.18 *** | ||||
BIO16 | 0.21 *** | ||||
P. cembroides | BIO2 | −0.15 *** | P. pseudostrobus | ||
BIO3 | 0.19 *** | BIO2 | −0.27 *** | ||
BIO4 | −0.23 *** | BIO3 | 0.19 ** | ||
BIO5 | −0.16 *** | BIO4 | −0.24 *** | ||
BIO6 | 0.18 *** | BIO5 | −0.17 * | ||
BIO7 | −0.22 *** | BIO7 | −0.3 *** | ||
BIO8 | −0.12 *** | BIO12 | 0.22 *** | ||
BIO10 | −0.13 *** | BIO16 | 0.2 *** | ||
BIO11 | 0.18 *** | ||||
BIO13 | −0.15 *** | ||||
BIO16 | −0.13 *** | ||||
P. leiophylla | BIO18 | −0.15 *** | P. teocote | ||
BIO19 | −0.14 *** | ||||
BIO3 | 0.22 *** | ||||
BIO4 | −0.20 *** | BIO2 | −0.12 * | ||
BIO5 | −0.18 *** | BIO3 | 0.21 *** | ||
BIO7 | −0.18 *** | BIO4 | −0.22 *** | ||
BIO8 | −0.15 *** | BIO7 | −0.19 *** | ||
BIO10 | −0.16 *** | BIO11 | 0.13 ** | ||
BIO12 | 0.24 *** | BIO12 | 0.14 ** | ||
BIO13 | 0.18 *** | ||||
BIO16 | 0.21 *** | ||||
Note: Significance of Spearman’s correlation; * p < 0.05, ** p < 0.01, *** p < 0.001. |
Appendix C
QUANTILE 1 | ||||||
P. arizonica | P. cembroides | P. leiophylla | P. oocarpa | P. pseudostrubus | P. teocote | |
Bioclimatic variable | BIO1 | BIO4 | BIO12 | BIO12 | BIO7 | BIO4 |
Min | 9.10 | 168.80 | 431.00 | 600.00 | 13.30 | 101.19 |
Max | 18.60 | 678.78 | 1230.00 | 2216.00 | 32.00 | 564.99 |
Mean | 12.86 | 458.46 | 778.01 | 1121.35 | 20.22 | 347.51 |
Mediana | 12.67 | 490.94 | 774.00 | 1050.50 | 19.20 | 358.27 |
SD | 1.67 | 120.52 | 171.03 | 317.93 | 4.10 | 99.50 |
QUANTILE 2 | ||||||
P. arizonica | P. cembroides | P. leiophylla | P. oocarpa | P. pseudostrubus | P. teocote | |
Bioclimatic variable | BIO1 | BIO4 | BIO12 | BIO12 | BIO7 | BIO4 |
Min | 9.80 | 216.53 | 422.00 | 494.00 | 13.50 | 90.44 |
Max | 16.95 | 669.43 | 1269.00 | 2490.00 | 29.70 | 575.05 |
Mean | 12.55 | 472.08 | 784.34 | 1130.05 | 20.55 | 345.69 |
Mediana | 12.20 | 496.24 | 771.00 | 1062.00 | 20.40 | 354.40 |
SD | 1.66 | 114.29 | 176.30 | 331.83 | 3.85 | 104.14 |
QUANTILE 3 | ||||||
P. arizonica | P. cembroides | P. leiophylla | P. oocarpa | P. pseudostrubus | P. teocote | |
Bioclimatic variable | BIO1 | BIO4 | BIO12 | BIO12 | BIO7 | BIO4 |
Min | 9.60 | 170.86 | 394.00 | 544.00 | 12.80 | 89.06 |
Max | 17.05 | 658.56 | 1206.00 | 2195.00 | 28.70 | 634.82 |
Mean | 12.50 | 468.18 | 814.92 | 1195.24 | 19.86 | 323.20 |
Mediana | 12.17 | 494.95 | 829.00 | 1165.00 | 19.10 | 347.05 |
SD | 1.72 | 108.17 | 178.56 | 329.96 | 3.37 | 106.57 |
QUANTILE 4 | ||||||
P. arizonica | P. cembroides | P. leiophylla | P. oocarpa | P. pseudostrubus | P. teocote | |
Bioclimatic variable | BIO1 | BIO4 | BIO12 | BIO12 | BIO7 | BIO4 |
Min | 8.99 | 222.35 | 394.00 | 538.00 | 13.90 | 94.44 |
Max | 18.84 | 672.84 | 1219.00 | 2265.00 | 28.70 | 579.92 |
Mean | 11.93 | 454.05 | 830.28 | 1300.00 | 19.06 | 311.09 |
Mediana | 11.65 | 476.88 | 828.00 | 1273.00 | 18.70 | 349.60 |
SD | 1.84 | 113.42 | 171.85 | 344.91 | 3.35 | 106.51 |
QUANTILE 5 | ||||||
P. arizonica | P. cembroides | P. leiophylla | P. oocarpa | P. pseudostrubus | P. teocote | |
Bioclimatic variable | BIO1 | BIO4 | BIO12 | BIO12 | BIO7 | BIO4 |
Min | 8.74 | 195.01 | 542.00 | 692.00 | 13.00 | 74.84 |
Max | 15.82 | 664.21 | 1940.00 | 2462.00 | 24.10 | 532.58 |
Mean | 11.57 | 387.17 | 898.22 | 1288.27 | 16.97 | 272.41 |
Mediana | 11.23 | 350.06 | 904.50 | 1235.00 | 16.90 | 303.20 |
SD | 1.47 | 124.36 | 162.16 | 338.90 | 2.36 | 110.99 |
References
- SEMARNAT. Anuario Estadístico de la Producción Forestal 2017–2018; SEMARNAT: Mexico City, Mexico, 2021. [Google Scholar]
- Galicia, L.; Chávez-Vergara, B.M.; Kolb, M.; Jasso-Flores, R.I.; Rodríguez-Bustos, L.A.; Solís, L.E.; de la Cruz, V.G.; Pérez-Campuzano, E.; Villanueva, Y.A. Perspectives of the socioecological approach in the preservation, utilization and the payment of environmental services of the temperate forests of Mexico. Madera Bosques 2018, 24, e2421443. [Google Scholar] [CrossRef]
- Perry, J. The Pines of Mexico and Central America; Timber Press: Portland, OR, USA, 1991. [Google Scholar]
- Sánchez-González, A. Una visión actual de la diversidad y distribución actual de los pinos de México. Madera Bosques 2008, 14, 107–120. [Google Scholar] [CrossRef]
- Pan, Y.; Birdsey, R.A.; Phillips, O.L.; Jackson, R.B. The structure, Distribution, and Biomass of the World’s forests. Annu. Rev. Ecol. Syst. 2013, 44, 593–622. [Google Scholar] [CrossRef]
- Martin, A.R.; Doraisami, M.; Thomas, S.C. Global patterns in wood carbon concentration across the world’s trees and forests. Nat. Geosci. 2018, 11, 915–920. [Google Scholar] [CrossRef]
- IPCC. Uso de la tierra, cambio de uso de la tierra y silvicultura. In Resumen para Responsables de Políticas; IPCC: Valencia, Spain, 2000; Volume 24. [Google Scholar]
- Ullah, F.; Gilani, H.; Sanaei, A.; Hussain, K.; Ali, A. Stand structure determines aboveground biomass across temperate forest types and species mixture along a local-scale elevational gradient. For. Ecol. Manag. 2021, 486, 118984. [Google Scholar] [CrossRef]
- Pan, Y.; Birdsey, R.A.; Fang, J.; Houghton, R.; Kauppi, P.E.; Kurz, W.A.; Phillips, O.L.; Shvidenko, A.; Lewis, S.L.; Canadell, J.G.; et al. A large and persistent carbon sink in the world’s forests. Science 2011, 333, 988–993. [Google Scholar] [CrossRef] [PubMed]
- Rodríguez-Veiga, P.; Saatchi, S.; Tansey, K.; Balzter, H. Magnitude, spatial distribution and uncertainty of forest biomass stocks in Mexico. Remote Sens. Environ. 2016, 183, 265–281. [Google Scholar] [CrossRef]
- Coomes, D.; Simonson, W.; Burslem, D. Forests and Global Change; Cambridge University Press: Cambridge, UK, 2014; ISBN 9781107041851. [Google Scholar]
- Ali, A.; Lin, S.L.; He, J.K.; Kong, F.M.; Yu, J.H.; Jiang, H.S. Climatic water availability is the main limiting factor of biotic attributes across large-scale elevational gradients in tropical forests. Sci. Total Environ. 2019, 647, 1211–1221. [Google Scholar] [CrossRef] [PubMed]
- Li, Q.; Liu, Z.; Jin, G. Impacts of stand density on tree crown structure and biomass: A global meta-analysis. Agric. For. Meteorol. 2022, 326, 109181. [Google Scholar] [CrossRef]
- Fotis, A.T.; Murphy, S.J.; Ricart, R.D.; Krishnadas, M.; Whitacre, J.; Wenzel, J.W.; Queenborough, S.A.; Comita, L.S. Above-ground biomass is driven by mass-ratio effects and stand structural attributes in a temperate deciduous forest. J. Ecol. 2018, 106, 561–570. [Google Scholar] [CrossRef]
- Balima, L.H.; Kouamé, F.N.G.; Bayen, P.; Ganamé, M.; Nacoulma, B.M.I.; Thiombiano, A.; Soro, D. Influence of climate and forest attributes on aboveground carbon storage in Burkina Faso, West Africa. Environ. Chall. 2021, 4, 100123. [Google Scholar] [CrossRef]
- Arasa-Gisbert, R.; Vayreda, J.; Román-Cuesta, R.M.; Villela, S.A.; Mayorga, R.; Retana, J. Forest diversity plays a key role in determining the stand carbon stocks of Mexican forests. For. Ecol. Manag. 2018, 415–416, 160–171. [Google Scholar] [CrossRef]
- Álvarez-Dávila, E.; Cayuela, L.; González-Caro, S.; Aldana, A.M.; Stevenson, P.R.; Phillips, O.; Cogollo, Á.; Peñuela, M.C.; Von Hildebrand, P.; Jiménez, E.; et al. Forest biomass density across large climate gradients in northern South America is related to water availability but not with temperature. PLoS ONE 2017, 12, e0171072. [Google Scholar] [CrossRef]
- Jia, B.; Guo, W.; He, J.; Sun, M.; Chai, L.; Liu, J.; Wang, X. Topography, Diversity, and Forest Structure Attributes Drive Aboveground Carbon Storage in Different Forest Types in Northeast China. Forests 2022, 13, 455. [Google Scholar] [CrossRef]
- Usuga, J.C.L.; Toro, J.A.R.; Alzate, M.V.R.; de Jesús Lema Tapias, Á. Estimation of biomass and carbon stocks in plants, soil and forest floor in different tropical forests. For. Ecol. Manag. 2010, 260, 1906–1913. [Google Scholar] [CrossRef]
- Chen, X.; Luo, M.; Larjavaara, M. Effects of climate and plant functional types on forest above-ground biomass accumulation. Carbon Balance Manag. 2023, 18, 5. [Google Scholar] [CrossRef] [PubMed]
- Reich, P.B.; Luo, Y.; Bradford, J.B.; Poorter, H.; Perry, C.H.; Oleksyn, J. Temperature drives global patterns in forest biomass distribution in leaves, stems, and roots. Proc. Natl. Acad. Sci. USA 2014, 111, 13721–13726. [Google Scholar] [CrossRef] [PubMed]
- Bennett, A.C.; Penman, T.D.; Arndt, S.K.; Roxburgh, S.H.; Bennett, L.T. Climate more important than soils for predicting forest biomass at the continental scale. Ecography 2020, 43, 1692–1705. [Google Scholar] [CrossRef]
- Ma, Y.; Eziz, A.; Halik, Ü.; Abliz, A.; Kurban, A. Precipitation and Temperature Influence the Relationship between Stand Structural Characteristics and Aboveground Biomass of Forests—A Meta-Analysis. Forests 2023, 14, 896. [Google Scholar] [CrossRef]
- IPCC. Summary for Policy Markers:Climate Change 2023: Synthesis Report; IPCC: Geneva, Switzerland, 2023. [Google Scholar]
- Harvey, J.E.; Smiljanić, M.; Scharnweber, T.; Buras, A.; Cedro, A.; Cruz-García, R.; Drobyshev, I.; Janecka, K.; Jansons, Ā.; Kaczka, R.; et al. Tree growth influenced by warming winter climate and summer moisture availability in northern temperate forests. Glob. Change Biol. 2020, 26, 2505–2518. [Google Scholar] [CrossRef]
- Alfaro-Ramírez, U.; Ramírez-Albores, E.; Vargas-Hernández, J.; Franco-Maass, S.; Pérez-Suárez, M. Potential reduction of Hartweg´s Pine (Pinus hartwegii Lindl.) geographic distribution. PLoS ONE 2020, 15, e0229178. [Google Scholar] [CrossRef] [PubMed]
- Villers-Ruiz, L.; Trejo-Vázquez, I. Assessment of the vulnerability of forest ecosystems to climate change in Mexico. Clim. Res. 1997, 9, 87–93. [Google Scholar] [CrossRef]
- Rodríguez-Zúñiga, J.; Flores-Ortiz, C.M.; González-Guillén, M.D.J.; Lira-Saade, R.; Rodríguez-Arévalo, N.I.; Dávila-Aranda, P.D.; Ulian, T. Cost Analysis of Seed Conservation of Commercial Pine Species Vulnerable to Climate Change in Mexico. Forests 2022, 13, 539. [Google Scholar] [CrossRef]
- INEGI. México en Cifras. Available online: https://www.inegi.org.mx/app/areasgeograficas/#collapse-Res (accessed on 13 March 2024).
- DOF. NORMA Oficial Mexicana de Emergencia NOM, EM 001-SARH3-1994; Diario Oficial de la Federación: Mexico City, Mexico, 1994; pp. 5–7. [Google Scholar]
- SEMARNAT; CONAFOR. La Producción de Resina de Pino en México; Comisión Nacional Forestal: Zapopan, Mexico, 2012. [Google Scholar]
- CONAFOR; CONABIO. Pinus montezumae Lamb. Available online: http://www.conafor.gob.mx:8080/documentos/docs/13/971Pinus%20montezumae.pdf (accessed on 21 June 2024).
- CONAFOR; CONABIO. Pinus oocarpa Shiede. Available online: http://www.conafor.gob.mx:8080/documentos/docs/13/974Pinus%20oocarpa.pdf (accessed on 21 June 2024).
- Sáenz-Romero, C.; Guzmán-Reyna, R.R.; Rehfeldt, G.E. Altitudinal genetic variation among Pinus oocarpa populations in Michoacán, Mexico. Implications for seed zoning, conservation, tree breeding and global warming. For. Ecol. Manag. 2006, 229, 340–350. [Google Scholar] [CrossRef]
- CONAFOR; CONABIO. Pinus devoniana Lindley. Available online: http://www.conafor.gob.mx:8080/documentos/docs/13/957Pinus%20devoniana.pdf (accessed on 21 June 2024).
- Gómez-Romero, M.; Soto–Correa, J.; Blanco-García, J.A.; Sáenz–Romero, C.; Villegas, J.; Linding-Cisneros, R. Estudio de especies de pino para restauración de sitios degradados. Agrociencia 2012, 46, 795–807. [Google Scholar]
- CONAFOR; CONABIO. Pinus cembroides Zucc. Available online: http://www.conafor.gob.mx:8080/documentos/docs/13/955Pinus%20cembroides.pdf (accessed on 21 June 2024).
- SEMARNAT; CONAFOR; CONACYT. Catálogo de Recursos Forestales Maderables y no Maderables Árido, Tropical y Templado; CONAFOR: Zapopan, Mexico, 2008. [Google Scholar]
- Constante García, V.; Villanueva Díaz, J.; Cerano Paredes, J.; Cornejo Oviedo, E.H.; Valencia Manzo, S. Dendrocronología de Pinus cembroides Zucc. y reconstrucción de precipitación estacional para el Sureste de Coahuila. Cienc. For. México 2009, 34, 17–39. [Google Scholar]
- CONAFOR; CONABIO. Pinus patula Schl. et Cham. Available online: http://www.conafor.gob.mx:8080/documentos/docs/13/975Pinus%20patula.pdf (accessed on 21 June 2024).
- CONAFOR; CONABIO. Pinus ayacahuite Ehren. Available online: http://www.conafor.gob.mx:8080/documentos/docs/13/953Pinus%20ayacahuite.pdf (accessed on 21 June 2024).
- CONAFOR; CONABIO. Pinus arizonica Engelm. Available online: http://www.conafor.gob.mx:8080/documentos/docs/13/952Pinus%20arizonica.pdf (accessed on 21 June 2024).
- CONAFOR. Procedimientos de Muestreo Version 19.0; SEMARNAT-CONAFOR: Guadalajara, Mexico, 2017. [Google Scholar]
- Fick, S.E.; Hijmans, R.J. WorldClim 2: New 1-km spatial resolution climate surfaces for global land areas. Int. J. Climatol. 2017, 37, 4302–4315. [Google Scholar] [CrossRef]
- Lê, S.; Josse, J.; Husson, F. FactoMineR: An R Package for Multivariate Analysis. J. Stat. Softw. 2008, 25, 1–18. [Google Scholar] [CrossRef]
- Rzedowski, J. Vegetación de México; Edición Digital; Comisión Nacional para el Conocimiento y Uso de la Biodiversidad: Mexico City, Mexico, 2006. [Google Scholar]
- Návar, J. Allometric equations for tree species and carbon stocks for forests of northwestern Mexico. For. Ecol. Manag. 2009, 257, 427–434. [Google Scholar] [CrossRef]
- Guerrero, L. Cuantificación de Biomasa, Carbono y Producción de Oxígeno de Pinus cembroides Zucc. Bachelor’s Thesis, Universidad Autónoma Agraria Antonio Narro, En Mazapil, Zacatecas, Mexico, 2013. [Google Scholar]
- Méndez-González, J.; Luckie-Navarrete, S.L.; Capó-Arteaga, M.A.; Nájera-Luna, J.A. Ecuaciones alométricas y estimación de incrementos en biomasa aérea y carbono en una plantación mixta de Pinus devoniana Lindl. y Pinus pseudostrobus Lindl., En Guanajuato, México. Agrociencia 2011, 45, 479–491. [Google Scholar]
- Carrillo-Anzúres, F.; Acosta-Mireles, M.; Flores-Ayala, E.; Juárez-Bravo, J.E.; Bonilla-Padilla, E. Estimación de biomasa y carbono en dos especies arboreas en La Sierra Nevada, México. Rev. Mex. Cienc. Agrícolas 2018, 5, 779–793. [Google Scholar] [CrossRef]
- Navarro-Martínez, J.; Godínez-Jaimes, F.; López-López, M.Á.; Rosas-Acevedo, J.L.; Juárez-López, A.L.; Reyes-Umaña, M. Ajuste de ecuaciones alométricas para estimar biomasa aérea en Pinus oocarpa y Quercus resinosa en Guerrero, México. Madera Bosques 2020, 26, e2611964. [Google Scholar] [CrossRef]
- Díaz-Franco, R.; Acosta-Mireles, M.; Carrillo-Anzures, F.; Buendía-Rodríguez, E.; Flores-Ayala, E.; Etchevers-Barra, J.D. Determinación de ecuaciones alométricas para estimar biomasa y carbono en Pinus patula Schl. et Cham. Madera Bosques 2007, 13, 25–34. [Google Scholar] [CrossRef]
- Moyroud, N.; Portet, F. Introduction to QGIS. In QGIS and Generic Tools; John Wiley & Sons Ltd.: Hoboken, NJ, USA, 2018; Chapter 1; pp. 1–17. [Google Scholar]
- Revelle, W. Psych: Procedures for Psychological, Psychometric, and Personality Research; R Core Team: Vienna, Austria, 2023. [Google Scholar]
- De Mendiburo, F. Una Herramienta de Análisis Estadístico para la Investigación Agrícola; Universidad Nacional de Ingenieria: Rímac, Peru, 2009. [Google Scholar]
- Kassambara, A. rstatix:Pipe-Friendly Framework for Basic Statical Tests; R Core Team: Vienna, Austria, 2023. [Google Scholar]
- Makowski, D.; Ben-Shachar, M.S.; Lüdecke, D. bayestestR: Describing Effects and their Uncertainty, Existence and Significance within the Bayesian Framework. J. Open Source Softw. 2019, 4, 1541. [Google Scholar] [CrossRef]
- Kassambara, A. ggpubr: “ggplot2” Based Publication Ready Plots; R Core Team: Vienna, Austria, 2023. [Google Scholar]
- Wang, L.Q.; Ali, A. Climate regulates the functional traits—Aboveground biomass relationships at a community-level in forests: A global meta-analysis. Sci. Total Environ. 2021, 761, 143238. [Google Scholar] [CrossRef]
- Wang, L.-Q.; Ali, A.; Wang, L.-Q.; Ali, A. Functional identity regulates aboveground biomass better than trait diversity along abiotic conditions in global forest metacommunities. Ecography 2021, 2022, e05854. [Google Scholar] [CrossRef]
- Cañas, R.A.; Canales, J.; Muñoz-Hernández, C.; Granados, J.M.; Ávila, C.; García-Martín, M.L.; Cánovas, F.M. Understanding developmental and adaptive cues in pine through metabolite profiling and co-expression network analysis. J. Exp. Bot. 2015, 66, 3113. [Google Scholar] [CrossRef]
- Gómez-Mendoza, L.G.; Arriaga, L. Modeling the Effect of Climate Change on the Distribution of Oak and Pine Species of Mexico. Conserv. Biol. 2007, 21, 1545–1555. [Google Scholar] [CrossRef]
- Aceves-Rangel, L.; Méndez-González, J.; García-Aranda, M.A.; Nájera-Luna, J.A. Distribución potencial de 20 especies de pinos en México. Agrociencia 2018, 52, 1043–1057. [Google Scholar]
- Alfaro-Reyna, T.; Retana, J.; Arasa-Gisbert, R.; Vayreda, J.; Martínez-Vilalta, J. Recent dynamics of pine and oak forests in Mexico. Eur. J. For. Res. 2020, 139, 179–187. [Google Scholar] [CrossRef]
- Slik, J.W.F.; Aiba, S.I.; Brearley, F.Q.; Cannon, C.H.; Forshed, O.; Kitayama, K.; Nagamasu, H.; Nilus, R.; Payne, J.; Paoli, G.; et al. Environmental correlates of tree biomass, basal area, wood specific gravity and stem density gradients in Borneo’s tropical forests. Glob. Ecol. Biogeogr. 2010, 19, 50–60. [Google Scholar] [CrossRef]
- Liu, Y.; Yu, G.; Wang, Q.; Zhang, Y. How temperature, precipitation and stand age control the biomass carbon density of global mature forests. Glob. Ecol. Biogeogr. 2014, 23, 323–333. [Google Scholar] [CrossRef]
- Escobar-Alonso, S.; Rodríguez-Trejo, D.A.; Escobar-Alonso, S.; Rodríguez Trejo, D.A. Estado del arte en la investigación sobre calidad de planta del género Pinus en México. Rev. Mex. Cienc. For. 2019, 10, 4–38. [Google Scholar] [CrossRef]
- De Frenne, P.; Lenoir, J.; Luoto, M.; Scheffers, B.R.; Zellweger, F.; Aalto, J.; Ashcroft, M.B.; Christiansen, D.M.; Decocq, G.; De Pauw, K.; et al. Forest microclimates and climate change: Importance, drivers and future research agenda. Glob. Change Biol. 2021, 27, 2279–2297. [Google Scholar] [CrossRef]
- Ali, A.; Sanaei, A.; Li, M.; Nalivan, O.A.; Ahmadaali, K.; Pour, M.J.; Valipour, A.; Karami, J.; Aminpour, M.; Kaboli, H.; et al. Impacts of climatic and edaphic factors on the diversity, structure and biomass of species-poor and structurally-complex forests. Sci. Total Environ. 2020, 706, 135719. [Google Scholar] [CrossRef] [PubMed]
- Bar-On, Y.M.; Phillips, R.; Milo, R. The biomass distribution on Earth. Proc. Natl. Acad. Sci. USA 2018, 115, 6506–6511. [Google Scholar] [CrossRef]
- Ni, Y.; Jian, Z.; Zeng, L.; Liu, J.; Lei, L.; Zhu, J.; Xu, J.; Xiao, W. Climate, soil nutrients, and stand characteristics jointly determine large-scale patterns of biomass growth rates and allocation in Pinus massoniana plantations. For. Ecol. Manag. 2022, 504, 119839. [Google Scholar] [CrossRef]
- González-Cásares, M.; Acosta-Hernández, A.C.; Zúñiga-Vazquez, J.M.; Herrera-Soto, G.; Camarero, J.J. Responses of Growth to Climate and Drought in Two Sympatric Mexican Pine Species Marcos. In Latin American Dendroecology; Pompa-García, M., Camarero, J.J., Eds.; Springer Nature: Cham, Switzerland, 2020; pp. 61–77. ISBN 9783030369293. [Google Scholar]
- Pompa-García, M.; Hadad, M.A. Sensitivity of pines in Mexico to temperature varies with age. Atmósfera 2016, 29, 209–219. [Google Scholar] [CrossRef]
- Vivar-Vivar, E.D.; Pompa-García, M.; Camarero, J.J. Tree Rings Elucidate Differential Drought Responses in Stands of Three Mexican Pines. Forests 2024, 15, 994. [Google Scholar] [CrossRef]
- Chave, J.; Réjou-Méchain, M.; Búrquez, A.; Chidumayo, E.; Colgan, M.S.; Delitti, W.B.C.; Duque, A.; Eid, T.; Fearnside, P.M.; Goodman, R.C.; et al. Improved allometric models to estimate the aboveground biomass of tropical trees. Glob. Change Biol. 2014, 20, 3177–3190. [Google Scholar] [CrossRef]
- Vieilledent, G.; Gardi, O.; Grinand, C.; Burren, C.; Andriamanjato, M.; Camara, C.; Gardner, C.J.; Glass, L.; Rasolohery, A.; Rakoto Ratsimba, H.; et al. Bioclimatic envelope models predict a decrease in tropical forest carbon stocks with climate change in Madagascar. J. Ecol. 2016, 104, 703–715. [Google Scholar] [CrossRef]
- Rosas-Chavoya, M.; López-Serrano, P.M.; Vega-Nieva, D.J.; Hernández-Díaz, J.C.; Wehenkel, C.; Corral-Rivas, J.J. Estimating Above-Ground Biomass from Land Surface Temperature and Evapotranspiration Data at the Temperate Forests of Durango, Mexico. Forests 2023, 14, 299. [Google Scholar] [CrossRef]
- Toledo, M.; Peña-Claros, M.; Bongers, F.; Alarcón, A.; Balcázar, J.; Chuviña, J.; Leaño, C.; Licona, J.C.; Poorter, L. Distribution patterns of tropical woody species in response to climatic and edaphic gradients. J. Ecol. 2012, 100, 253–263. [Google Scholar] [CrossRef]
- Zhang, X.; Zhou, Y.; Ji, Y.; Yu, M.; Li, X.; Duan, J.; Wang, Y.; Gao, J.; Guo, X. Climate Factors Affect Above–Belowground Biomass Allocation in Broad-Leaved and Coniferous Forests by Regulating Soil Nutrients. Plants 2023, 12, 3926. [Google Scholar] [CrossRef] [PubMed]
- Stegen, J.C.; Swenson, N.G.; Enquist, B.J.; White, E.P.; Phillips, O.L.; Jørgensen, P.M.; Weiser, M.D.; Monteagudo Mendoza, A.; Núñez Vargas, P. Variation in above-ground forest biomass across broad climatic gradients. Glob. Ecol. Biogeogr. 2011, 20, 744–754. [Google Scholar] [CrossRef]
- Lie, Z.; Xue, L.; Jacobs, D.F. Allocation of forest biomass across broad precipitation gradients in China’s forests. Sci. Rep. 2018, 8, 10536. [Google Scholar] [CrossRef] [PubMed]
- Sandel, B.; Goldstein, L.J.; Kraft, N.J.B.; Okie, J.G.; Shuldman, M.I.; Ackerly, D.D.; Cleland, E.E.; Suding, K.N. Contrasting trait responses in plant communities to experimental and geographic variation in precipitation. New Phytol. 2010, 188, 565–575. [Google Scholar] [CrossRef]
- Lin, D.; Xia, J.; Wan, S. Climate warming and biomass accumulation of terrestrial plants: A meta-analysis. New Phytol. 2010, 188, 187–198. [Google Scholar] [CrossRef]
- Saatchi, S.; Houghton, R.A.; Dos Santos Alvalá, R.C.; Soares, J.V.; Yu, Y. Distribution of aboveground live biomass in the Amazon basin. Glob. Change Biol. 2007, 13, 816–837. [Google Scholar] [CrossRef]
- IPCC. Synthesis Report, Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. In Climate Change 2023; Lee, H., Romero, J., Eds.; IPCC: Geneva, Switzerland, 2023; pp. 35–115. [Google Scholar]
- McKenney, D.W.; Pedlar, J.H.; Lawrence, K.; Campbell, K.; Hutchinson, M.F. Potential Impacts of Climate Change on the Distribution of North American Trees. Bioscience 2007, 57, 939–948. [Google Scholar] [CrossRef]
- Hammond, W.M.; Williams, A.P.; Abatzoglou, J.T.; Adams, H.D.; Klein, T.; López, R.; Sáenz-Romero, C.; Hartmann, H.; Breshears, D.D.; Allen, C.D. Global field observations of tree die-off reveal hotter-drought fingerprint for Earth’s forests. Nat. Commun. 2022, 13, 1761. [Google Scholar] [CrossRef] [PubMed]
- Cuervo-Robayo, A.P.; Ureta, C.; Gómez-Albores, M.A.; Meneses-Mosquera, A.K.; Téllez-Valdés, O.; Martínez-Meyer, E. One hundred years of climate change in Mexico. PLoS ONE 2020, 15, e0209808. [Google Scholar] [CrossRef] [PubMed]
- Larjavaara, M.; Lu, X.; Chen, X.; Vastaranta, M. Impact of rising temperatures on the biomass of humid old-growth forests of the world. Carbon Balance Manag. 2021, 16, 31. [Google Scholar] [CrossRef] [PubMed]
- Gutiérrez, E.; Trejo, I. Efecto del cambio climático en la distribución potencial de cinco especies arbóreas de bosque templado en México. Rev. Mex. Biodivers. 2014, 85, 179–188. [Google Scholar] [CrossRef]
Species | H (m) | DBH (cm) | AGBd (t ha−1) | ||||||
---|---|---|---|---|---|---|---|---|---|
Min | Max | Mean | Min | Max | Mean | Min | Max | Mean | |
P. arizonica | 0.20 | 29.50 | 8.57 e | 7.50 | 61.50 | 16.62 f | 0.013 | 19.07 | 3.11 c |
P. ayacahuite | 0.20 | 30.00 | 8.89 d | 7.50 | 62.40 | 15.77 f | 0.022 | 12.32 | 1.14 f |
P. cembroides | 0.20 | 16.20 | 5.52 f | 7.50 | 60.50 | 14.82 g | 0.002 | 14.30 | 2.33 d |
P. devoniana | 0.20 | 41.50 | 12.31 ab | 7.50 | 85.10 | 24.55 a | 0.010 | 12.40 | 1.16 f |
P. leiophylla | 0.20 | 35.00 | 9.60 d | 7.50 | 67.80 | 18.39 e | 0.020 | 26.45 | 1.92 e |
P. montezumae | 0.20 | 35.70 | 12.32 ab | 7.50 | 97.00 | 24.76 ab | 0.020 | 82.33 | 8.08 a |
P. oocarpa | 0.20 | 36.70 | 11.67 b | 7.50 | 89.00 | 22.61 b | 0.014 | 58.23 | 5.07 a |
P. patula | 0.20 | 36.70 | 13.37 a | 7.50 | 84.90 | 20.88 c | 0.010 | 53.59 | 6.66 a |
P. pseudostrobus | 0.20 | 44.40 | 13.27 a | 7.50 | 101.0 | 25.54 a | 0.010 | 99.78 | 9.01 a |
P. teocote | 0.20 | 33.00 | 10.28 c | 7.50 | 71.10 | 19.21 d | 0.022 | 30.14 | 3.48 b |
Parameter | Pari | Paya | Pcem | Pdev | Plei | Pmon | Pooc | Ppat | Ppse | Pteo |
---|---|---|---|---|---|---|---|---|---|---|
BIO1 | 2.11 e × 103 *** | 0.094 | 1.46 e × 109 *** | 0.857 | 2.17 e × 107 *** | 1.89 | 1.15 | 6.75 * | 2.84 | 50.51 *** |
BIO2 | 0.353 | 0.608 | 2.04 e × 1018 *** | 0.313 | 0.696 | 2.05 | 1.46 e × 109 *** | 0.338 | 1.09 e × 106 *** | 2.97 e × 109 *** |
BIO3 | 0.153 | 0.147 | 9.59 e × 1022 *** | 0.198 | 72.00 *** | 0.637 | 2.69 e × 104 *** | 0.456 | 239.75 *** | 3.32 e × 108 *** |
BIO4 | 0.183 | 0.307 | 5.84 e × 1028 *** | 0.178 | 11.35 ** | 0.536 | 7.69 e × 107 *** | 0.742 | 1.13 e × 104 *** | 8.87 e × 1013 *** |
BIO5 | 19.88 ** | 0.278 | 8.23 e × 1012 *** | 2.03 | 4.46 e × 1011 *** | 11.61 ** | 1.12 | 10.94 ** | 1.12 e × 105 *** | 0.91 |
BIO6 | 115.62 *** | 0.142 | 8.54 e × 1031 *** | 0.31 | 0.922 | 0.424 | 1.47 e × 103 *** | 1.39 | 0.15 | 2.27 e × 107 *** |
BIO7 | 0.106 | 0.542 | 1.17 e × 1035 *** | 0.19 | 42.36 *** | 2.38 | 9.90 e × 108 *** | 0.812 | 2.19 e × 107 *** | 2.72 e × 1013 *** |
BIO8 | 10.05 ** | 0.186 | 4.15 e × 103 *** | 3.34 * | 4.11 e × 1011 *** | 2.94 | 0.074 | 7.46 * | 62.22 *** | 0.089 |
BIO9 | 66.21 *** | 0.089 | 0.405 | 0.44 | 1.20 e × 107 *** | 1.33 | 0.737 | 3.12 * | 0.648 | 22.90 ** |
BIO10 | 12.52 ** | 0.18 | 3.50 e × 105 *** | 1.64 | 4.39 e × 1011 *** | 3.29 * | 0.075 | 7.65 * | 141.23 *** | 0.082 |
BIO11 | 4.99 e × 103 *** | 0.105 | 3.74 e × 1025 *** | 0.465 | 10.14 ** | 0.948 | 99.08 *** | 4.50 * | 0.194 | 1.46 e × 106 *** |
BIO12 | 0.117 | 0.185 | 76.19 *** | 0.158 | 0.108 | 0.241 | 7.30 e × 104 *** | 0.36 | 156.21 *** | 5.38 e × 103 *** |
BIO13 | 0.128 | 0.09 | 7.41 e × 1016 *** | 0.154 | 0.09 | 0.359 | 393.53 *** | 0.565 | 8.92* | 0.749 |
BIO14 | 0.113 | 0.093 | 2.82 e × 104 *** | 0.276 | 0.104 | 0.212 | 0.138 | 0.362 | 0.13 | 0.083 |
BIO15 | 0.173 | 0.381 | 2.29 e × 1025 *** | 0.19 | 0.084 | 0.215 | 0.077 | 0.205 | 0.749 | 0.138 |
BIO16 | 0.11 | 0.099 | 3.14 e × 1013 *** | 0.179 | 0.092 | 0.272 | 2.87 e × 103 *** | 0.352 | 102.24 *** | 10.02 ** |
BIO17 | 0.125 | 0.262 | 3.77 e × 107 *** | 0.287 | 0.182 | 0.215 | 0.079 | 0.422 | 0.136 | 0.094 |
BIO18 | 0.103 | 0.091 | 3.75 e × 1016 *** | 0.719 | 0.125 | 0.216 | 0.074 | 0.361 | 1.26 | 90.19 *** |
BIO19 | 0.112 | 0.138 | 977.44 *** | 0.228 | 0.37 | 0.313 | 9.29 * | 0.262 | 0.202 | 1.15 e × 104 *** |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Girón-Gutiérrez, D.; Méndez-González, J.; Osorno-Sánchez, T.G.; Cerano-Paredes, J.; Soto-Correa, J.C.; Cambrón-Sandoval, V.H. Climate as a Driver of Aboveground Biomass Density Variation: A Study of Ten Pine Species in Mexico. Forests 2024, 15, 1160. https://doi.org/10.3390/f15071160
Girón-Gutiérrez D, Méndez-González J, Osorno-Sánchez TG, Cerano-Paredes J, Soto-Correa JC, Cambrón-Sandoval VH. Climate as a Driver of Aboveground Biomass Density Variation: A Study of Ten Pine Species in Mexico. Forests. 2024; 15(7):1160. https://doi.org/10.3390/f15071160
Chicago/Turabian StyleGirón-Gutiérrez, Dioseline, Jorge Méndez-González, Tamara G. Osorno-Sánchez, Julián Cerano-Paredes, José C. Soto-Correa, and Víctor H. Cambrón-Sandoval. 2024. "Climate as a Driver of Aboveground Biomass Density Variation: A Study of Ten Pine Species in Mexico" Forests 15, no. 7: 1160. https://doi.org/10.3390/f15071160
APA StyleGirón-Gutiérrez, D., Méndez-González, J., Osorno-Sánchez, T. G., Cerano-Paredes, J., Soto-Correa, J. C., & Cambrón-Sandoval, V. H. (2024). Climate as a Driver of Aboveground Biomass Density Variation: A Study of Ten Pine Species in Mexico. Forests, 15(7), 1160. https://doi.org/10.3390/f15071160