Response of Mediterranean grassland species to changing rainfall. A reply to Figueroa and Davy

Orsis, 8: 121-126 (1993) Response of Mediterranean grassland species to changing rainfall. A reply to Figueroa and Davy R. Fernández Alés Departamento de Biologia Vegetal y Ecologia. Universidad de Sevilla. Apdo. 1095. 41080 Sevilla Key words: Mediterranean annuals, temporal changes, dehesa. Resumen. Respuesta de especies pratenses mediterráneas a cambios en la precipitaci6n. Una réplica a Figueroa y Davy. Se han revisado 10s resultados publicados por Figueroa & Davy en el Journal of Ecology (vol 79: 925-941, 1991) sobre la relación entre las variaciones en frecuencia de las espe- 'cies de pastos mediterráneos del W de España y la precipitación, debido a que sus resultados contradicen 10s de otros autores que también han estudiado estas relaciones en otros pastos mediterráneos. La causa de esta contradicción estriba en que 10s datos de precipitación utilizados por Figueroa & Davy contienen muchos errores. Cuando se utilizan 10s datos correctos 10s resultados son concordantes con 10s de 10s otros autores. Abstract. The results of a paper of Figueroa & Davy in the Journal of Ecology (vol 79: 925-941, 1991) on the relationships between changes in species frequency and rainfall in a Mediterranean grassland of W Spain have been revised, because they do not agree with the results of other authors that have studied this relationships in other Mediterranean grasslands. The lack of agreement is the result of the inaccuracies contained in the rainfall data used by Figueroa & Davy. When the correct rainfall data are used, the results agree with those of the other authors. Introduction Recently, M.E. Figueroa and A.J. Davy have published a paper in the Journal of Ecology (1991) on the relationships between changes in species abundance and rainfall in a Mediterranean grassland in W Spain. Using a long series of data (1 1 years) on species frequency in a dehesa grassland in Alia (Cáceres), and rainfall data from the meteorological station of Guadalupe (Cáceres), the authors show that winter rainfall and the rainfall over the whole growing season are good predictors of the abundance of a considerable number of species (25%), better than autumn and spring rainfall. These results do not fit with other studies on Mediterranean grasslands of California (Duncan & Woodmansee, 1975; Pitt & Heady, 1978), Australia (Rossiter, 1966), Israel (Naveh, 1982) and Spain (Ortega & Fernández Alés, 1987). All these authors find better relationships between species abundance and seasonal rainfall (specially with autumn and spring rainfall) than with annual rainfall. Hobbs & Mooney (1985), Marañón (1985) and Fernández Alés (1991) have shown that, in Mediterranean annual grasslands, only a small part of the seed bank emerge as seedlings, and almost all of them emerge during the first month after the first autumn rains, with many of them dying in this time of the year. Mortality is very low during winter and spring, and almost all survivors die at the end of the spring, after producing seeds. This means that the botanical composition and species abundance in the grassland are determined during the germination and seedling emergence period, remaining constant during the rest of the growing season (Heady, 1958; Ortega & Fernández Alés, 1987). Thus, the distribution and abundance of rainfall during this critica1 period (autumn) will be a paramount factor in determining the abundance of the species in the grassland (Bartolome, 1979; Ortega & Fernández Alés, 1987). Since autumn rainfall is not correlated with winter or total rainfall (Figueroa & Davy, 1991), the set of processes described cannot explain the relationships found by Figueroa & Davy. The objective of this study is to revise the data of Figueroa & Davy, in order to investigate the causes of the lack of agreement between the results of these authors and the results of the other studies cited above. Methods and results The first matter investigated is the effect of data transformation on the results of the correlation between plant frequency and rainfall, since the correlation coefficient r is not invariant to logarithmic transformations. In order to test for the effect of the transformation used, the non transformed frequency data for the most abundant species (the only plant data available from Figure 2 of Figueroa & Davy, 1991) have been correlated with the total rainfall over the whole growing season (Figure 1 in Figueroa & Davy, 1991), and the results compared with the ones published by Figueroa & Davy using log-transformed plant frequency data (tables 1, 2, 3, 4 and 5 in Figueroa & Davy, 1991). The results (Table 1) show that the number of species showing significant correlation increases from 7 to 10 when using non transformed data. Thus, it can be concluded that data transformations are not the responsible for the discrepancies detected. The second matter investigated is the existence of errors in the data. The revision of the rainfall data used by Figueroa & Davy has shown that there are some inaccuracies in them. Thus, the data on annual precipitation in Guadalupe meteorological station used in figures 1 and 3 by Figueroa & Davy (Figure la), do not fit with the data provided by the Spanish Meteorological Institute for Guadalupe station (Figure lb), nor with the data from Alia station (figure lc,), that lies closer to the study area. In view of such discrepancies, the non transformed plant frequency data has been correlated with the rainfall data from Alia Meteorological station. This station has been selected because it has no rnissing data, as it happens with Guadalupe station (Figure 1). Product-moment correlation coefficients (n=ll) were calcula- REPLY TO FIGUEROA & DAVY 1 O00 - 500 E E - - s .- 2 o Years 1000 - 500 E E - - - s .- 2 o Years 1 O00 - 500 E E - - s .- 2 o Years Figure 1. Total rainfall during the growing season (October-May) at two meteorological stations close to the study area of Figueroa & Davy, for the years in which plant frequency was recorded. Separate seasonal components: ( . ) Autumn (October- Novernber), ( O ) winter (December-Febmary), ( . ) Spring (March-May). A. Data from Figueroa & Davy (1991) for Guadalupe station. B. Data provided by the Spanish Meterological Institute for Guadalupe Station. This station has missing data (months in which rainfall was not recorded) in 1983,1984, 1985 and 1986. The nurnber of rnonths with missing data each year are marked at the top of the bar, in brackets. C. Data provided by the Spanish Meteorological institute for Alfa Station, closer to the study area than Guadalupe and with no rnissing data. ted between the total frequency of each species and total rainfall during the whole growing season and over auturnn (October, November), winter (December, January, February) and spring (March, April, May). The results (Table 2) show that, when considering the rainfall data of Alia instead of the ones provided by Figueroa & Davy, the number of species exhibiting significant correlation with rainfall over the whole growing season falls from 10 to 6. The most striking differences are found when comparing the correlations with the seasonal rainfall (Table 2). Thus, while Figueroa & Davy find that the largest number of species are significantly correlated with winter rainfall, the correlations with Alia rainfall data show that much more species are related with the autumn precipitation (7) than with the winter (1) and spring (0) one. Thus, it can be concluded that the grasslands studied by Figueroa & Davy behave as the other Mediterranean grasslands, and that the differences found by these authors were related with the inaccuracies in the rainfall data. Table 1. Correlations of transformed and non transformed plant frequency with the accumulated rainfall over the growing season (October - May). Data of transformed plant frequency and rainfall from Figueroa & Davy (1991). Coefficients with an asterisk are significant for p< 0.05. Species Transformed Not transformed Vulpia bromoides 0.78" O 81" Anthoxanthum aristatum 0.59 0.62" Aira cariophyllea 0.63" 0.68" Poa bulbosa 0.58 O 61" Chaetopogon fasciculatus 0.23 O. 16 Ornithopus compressus 0.90" 0.92" Trifolium glomeratum 0.78" 0.81" Trifolium campestre 0.70" 0.70" Trifolium arvense 0.61 * 0.62" Trifolium gemellum 0.46 0.5 1 Tolpis barbata 0.72" 0.83" Leontodon taraxacoides 0.16 0.28 Hipochaeris glabra -0.20 -0.08 Chamaernelum rnixtum 0.10 -0.04 Logfia gallica 0.41 0.42 Tuberaria guttata 0.52 0.69" Plantago bellardi 0.57 0.58 Euphorbia exigua -0.05 0.28 REPLY TO HGUEROA & DAVY 125 Conclusions The lack of agreement between the results of Figueroa & Davy and the results of other studies made in other Mediterranean grassland is related with the inaccuracies in the rainfall data used by the former authors. When the correct rainfall data are used, this lack of agreement disappear. In any case, it should be noted that the number of species correlated with rainfall over the whole growing season is almost as high as when considering only autumn rains. Since many of the studies cited above have established the relationships between species abundance and rainfall grouping the species into families, or have used short series of data, it would be very interesting that these authors reconsider their results with the proper rainfall data, in order to check the relati- ve importance of total and autumn rainfall in determining species abundance. Table 2. Correlations of non transformed plant frequency with accumulated rainfall of Alia meteorological station. Growing season, October-May; Autumn, October-November; Winter, December-February; Spring, March-May. Coefficients with an asterisk are significant for p< 0.05. Accumulated rainfall Growing Species season Autumn Winter Spring Vulpia bromoides 0.82" 0.64" 0.59 -0.08 Anthoxanthum aristatum 0.63" 0.62* 0.34 0.11 Aira cariophyllea 0.7 1 * 0.21 0.68" -0.13 Poa bulbosa 0.48 0.66" 0.23 -0.16 Chaetopogon fasciculatus 0.26 -0.13 0.35 -0.10 Ornithopus compressus 0.63* 0.55 0.35 0.22 Trifolium glomeratum 0.50 0.72" 0.14 0.15 Trifolium campestre 0.81* 0.60" 0.50 0.25 Trifolium arvense 0.78" 0.75" 0.4 1 0.20 Trifolium gemellum 0.50 0.56 0.17 0.35 Tolpis barbata 0.35 0.5 1 0.14 -0.07 Leontodon taraxacoides 0.34 -0.10 0.39 0.07 Hipochaeris glabra -0.11 -0.39 0.1 1 -0.19 Chamaemelum mixtum 0.10 -0.16 0.09 0.37 Logfia gallica 0.07 0.61" 0.26 0.17 Tubera ria guttata 0.29 -0.05 0.33 0.05 Plantago bellardi 0.58 0.05 0.52 0.30 Euphorbia exigua 0.47 0.12 0.50 -0.22 References Bartolome, J.W. 1979. Gerrnination and seedling emergence in California annual grassland. J. Ecol 67: 273-28 1. Duncan, D.A. & Woodmansee, R.C. 1975. Forecasting forage yield from precipitation in California's annual rangeland. J. Range Manag. 28(4): 327-329. Fernández Alés, R. 1991. Population dynarnics in annual species of Mediterranean grasslands. IV International Rangeland Congress. Montpellier, 22-26 of April of 1991. Figueroa, M.E. & Davy, (1991). Response of Mediterranean grassland species to changing rainfall. J. Ecol. 79: 925-941. Heady, H.F. 1958. Vegetational changes in the California annual type. Ecology 39: 402- 416. Hobbs, R. & Mooney, H.A. 1985. Community and population dynamics of serpentine grassland in relation to gopher disturbance. Oecologia 67: 342-35 1. Marañón, T. 1985. Reserva de semillas en el suelo en una dehesa de Sierra Morena: relación con la vegetación. Anal. Edaf. y Agrobiol XLIV: 1805-1816. Naveh, Z. 1982. The dependence of the productivity of an semiarid Mediterranean hil1 pasture ecosystem on climatic fluctuations. Agriculture and Environment 7: 47-61. Ortega, F. & Fernández Alés, R. 1987. Trends in floristic change in time in Mediterranean annual grasslands of South Western Spain. in Di Castri, F.; Floret, Ch.; Rambal, S. & Roy, J. (eds.). Time scales and Water Stress. IUBS. Paris. pp. 451-456. Pitt, M.D. & Heady, H.F. (1978). Responses of annual vegetation to temperature and rainfall patterns in Northern California. Ecology 59: 336-350. Rossiter, R.C. (1966). Ecology of Mediterranean annual-type pasture. Adv. Agron. 18: 1-56. Manuscript received on December 1992