1

Selecting ecotonal landscape units

1 2 3 4 5 6 7 9 10 11 12
*

Selecting ecotonal landscape units on Meridional Plateau, Southern Brazil

a

Rosemeri Segecin Moro, bIngrid Aparecida Gomes, bTiaro Katu Pereira

Corresponding author: aUEPG/DEBIO. Universidade Estadual de Ponta Grossa. Programa de Mestrado em Geografia PhD. Conservao de Ecossistemas, Ponta Grossa, PR Brazil, moro.uepg@gmail.com Territrio ABSTRACT

8 b Universidade Estadual de Ponta Grossa. Programa de Mestrado em Geografia UEPG, MSc. Gesto do

13The Escarpa Devoniana Environmental Protection Area, located in the Meridional Plateau of South 14Brazil, supports a high diversity mosaic of native Araucaria forests and grasslands patches. In order to 15conduct studies of its dynamics, the Biodiversity of grassland-forest ecotones in South Brazil Project 16team has been developing a methodological approach for sorting ecotonal areas for sampling. This 17essay validates this methodology to the major grasslands occurrence area in the Brazilian Atlantic 18Forest Biome, the Campos Gerais region. The entire area was divided into a 1 x 1 km grid that was 19overlayed with vegetation typology files. The criteria for quadrat selecting were: (a) 25% or more of 20forest remnants; (b) 25% or more of grasslands remnants; (c) at least 20% of contact between them. 21The non-supervisioned classification pPotential ecotonal areas were validated in a non-supervisioned 22classification through visual checking in a high resolution qualitative way, always by the same 23operator. The quadrats could be achieved by the researchers in a *.kmz file, which allows someone 24selecting them in the Google Earth plataform. 25Keywords: Landscape Ecology, Campos Gerais, GIS, sampling methods 26 27 RESUMEN

28El Area de Preservacin Ambiental de la Escarpa Devoniana, situado en la meseta meridional del 29Brasil admite un mosaico de gran diversidad de bosques de Araucaria y pastizales nativos. Para llevar a 30cabo estudios de su dinmica, el equipo de lo proyecto Biodiversidad de ecotonos de bosques y 31pastizales en el sur de Brasil han evolucionado un enfoque metodolgico para ordenar reas ecotonales 32de muestreo. Este ensayo valida la metodologa para la zona de ocurrencia de grandes pastizales en el
3 4

Selecting ecotonal landscape units

33bioma de bosque Atlntico, la regin de Campos Gerais. Toda el rea fue dividido en cuadrculas de 1 34x 1 Km que fue enchapada con archivos de tipologa de vegetacin. Los criterios para la seleccin de 35quadrculas fueron: (a) 25% o ms de los restos de bosque; (b) 25% o ms de los restos de pastizales; 36(c) al menos el 20% de contacto entre ellos. La clasificacin no supervisionada de reas potencialmente 37ecotonales fue validada a travs de comprobacin cualitativa visual de alta resolucin, siempre por lo 38mismo operador. Las cuadrculas podran lograrse mediante el equipo de investigacin en un archivo 39*.kmz, que permita que alguien seleccionndolos en la plataforma de Google Earth. 40INTRODUCTION 41 Grasslands in southern Brazil are natural ecosystems present in this region prior to the forest

42expansion in the Late Quaternary (Behling and Pillar 2007, Behling and Lichte 1997). Grasslands form 43ecotones with Araucaria Forest isolated rounded patches, linear riparian forests, or larger forests over 44slopes (Moro and Carmo 2007). 45 The natural grasslands dynamic are associated with water and soil conditions (Curcio 2006) and 46with disturbances such as fire and pasture (Overbeck et al. 2007). Investigations on the Araucaria forest 47expansion over the grasslands in the Campos Gerais region, in the Paran State, are being conducted 48with a proposed, not published yet, methodology for sorting areas developed to the Pampas Biome, in 49Rio Grande do Sul (Hasenack et al. 2011, pers. com.). In that region, forests spread mainly from their 50edge over the open areas (Oliveira and Pillar 2004), or from little woody nucleous not far from the 51forest (Duarte et al. 2007). 52 This essay validates this GIS methodology to select and sort ecotonal areas on the major 53grasslands occurrence in the Brazilian Atlantic Forest Biome, the Campos Gerais region. 54METHODS 55 The systematic and standardized approach to select suitable sampling units combined

56stratification and randomization from knowledge about the current distribution and spatialization of 57grasslands and forests ecotones. For this, we used the database related to the Escarpa Devoniana 58Environmental Protetion Area (APA in Portuguese), a conservation unit of extensive use with 59392,363 ha (SEMA/IAP 20049). Images from 2004 were a fusion of Landsat 7 satellite bands. 60Analyses were performed by the software ArcGIS 9.x using the Vector-based Landscape Analysis 61Tools Extension (ESRI 2006).
7 8

Selecting ecotonal landscape units

62

We adopted in the first step, as reference, a standardized spatial grid of 10 x 10 km recommended

63by the Brazilian Government (Servio Florestal Brasileiro/ Ministrio do Meio Ambiente) for sampling 64to the National Forestry Inventory. According to them, this resolution allowed ecotone delimitations in 65several different regional landscapes no matter the predominant matrix. Thiessen polygons resulted 66from the equidistant lines interacting between points and the APA limits were overlayed with 67vegetation typology files of the APA. The attribute table containeds patches information, as patch 68identification number, grid code, use class (Araucaria forest, grasslands, silviculture, agriculture, and 69water bodies), total area (m2 and hectares), and perimeter. 70 Using the class attribute, the criteria for quadrat selection were: (a) 25% or more of forest

71remnants; (b) 25% or more of grasslands remnants. These arbitrary criteria were considered enough to 72provide a good minimum probability of grasslands and forests werebe potentially in contact. As the 73quadrats selected this way could not be real ecotonal areas, once they could be apart, validation of the 74non-supervisioned classification was conducted in a qualitative way, always by the same operator, 75through visual checking of the contact between grasslands and forests patches. Every quadrat was 76analyzed by means of the more recent images of Google Earth and only the contact of at least 20% of 77the two vegetation were considered ecotones for our purposes. This criterion 79settlement. 80RESULTS and Discussion 81 The 85 10 x 10 Km polygons have covered 850,000 ha, so we perceived that this grid size was arbitrary too, 78considering the finest scale of the mapping interpretations and the field necessities for replica plotting

82covered much more than the APA area (fig. 1-A). Besides, the initial proposed resolution, of 30 meters, 83was unable to discriminate some small ecotones in this region and, in spite of the Brazilian government 84technical recommendations, a new delimitation was necessary. So each polygon was divided again, this 85time in a 1 x 1 km grid with 4,647 polygons covering almost 465,000 ha (fig. 1-B), including just 15% 86of marginal areas outside the APA. 87 This APA grid generated one attribute table containing 77,003 vegetation patches, where 42,083

88were classified as agriculture and 8,956 as silviculture. So, aAlmost 66% of the vegetation patches 89were anthropogenic. There were 12,528 Araucaria forest patches, and 14,930 patches identified as 90grasslands and/or savannas, and/or wetlands. After this screening, circa 1,950 quadrats (195,000 ha)

11 12

13

Selecting ecotonal landscape units

91potentially could have ecotones, i.e., quadrats that presented grasslands and Araucaria forests, 92corresponding to 22.9% of the total area. 93

94 95Figure 1. Natural vegetation on Escarpa Devoniana APA under 85 10 x 10 Km Thiessen polygons (A), 96and under 4,647 1 x 1 Km grid (B). 97 98 It was necessary a visual checking on the higher resolution images of Google Earth to

99discriminate the grasslands areas from savanna and wetlands. In this procedure, we have found also 100several Araucaria Forest classification errors related to actual silviculture areas. Forests deeply located 101into canyons were desconsidered as well, due to their hypotetical difficulty to contact grasslands above. 102After the visual validation, only 1,544 quadrats (18.2% of the total area) achieved the third proposed 103criterion for real ecotone presence in the studied scale - at least 20% of contact between grasslands and 104Araucaria forests patches (fig. 2).
15 16

17

Selecting ecotonal landscape units

105 106 107 108

109 110Figura 2.Vegetation remnants on a 1 x 1 Km grid on the Escarpa Devoniana APA: A - grasslands, B 111Araucaria forests, and C - ecotonal areas. 112 113 The quadrats were then transposed by the research team into a *.kmz file for the project site, alloweds someone selecting them in the Google Earth platform. In the site

114which

115(https://sites.google.com/site/sisbiotaparana), acessing the *.kmz files, one could see a georeferenced 116index file with a 1 x1 km grid. Sorting any one, the noted UTM coordinates allowed to access shape 117files of the systematic mapping where one could see the grid sequencial numbers and its related 118attribute tables. The former 10 X 10 Km proposed grid was not suitable for ecotones identification in 119this region due to the finer scale required. Also, the shape and extension of the APA contributed for the 120selection of larger areas without interest outside the APA limits. 121DISCUSSION AND CONCLUSIONS
19 20

21

Selecting ecotonal landscape units

122

This is a realiable method for assessing cartographic representative areas for sampling. The

123previous quadrats selection helps the research team on randomly sorting ecotonal areas before their 124field works, especially for those that do not have much a local vegetation knowledge. The former 10 X 12510 Km proposed grid is not suitable for ecotones identification in Campos Gerais region due to the 126finer scale required. Also, the extension and convex shape of the APA lead to the selection of larger 127areas without interest outside the APA limits. To divide the area in a smaller grid, as 1 x 1 km, is an 128option, but now we have to achieve another resolution patch. If finer images were not available, one 129could use the Google Earth plataform. In that case, one could be aware that, before mapping, quadrat 130images must be georeferenced by means of at least 5 GPS field points every sampling. 131 Nevertheless, even bearing ecotones, some of them quadrats may not be suitable for sampling.

132Each sorted quadrat should be then evaluated according to its distance and access conditions, relief, 133special permission entrance needs, and so on. A landscape context analysis should be performed as 134well, particularly according to spatial redundancy, connectivity or contiguity with another ecotonal 135area, barriers existeance, or any other parameter that could possibly influence the quality of the ecotone 136at each focused biocoenosis. 137ACKNOWLEDGMENTS 138 We are grateful to our coleague Atila Santana of the project Biodiversidade dos campos e dos

139ectonos campo-floresta no sul do Brasil: bases ecolgicas para sua conservao e uso sustentvel , 140supported by the MCT/CNPq/MEC/CAPES/FNDCT N 47/2010 SISBIOTA BRASIL. 141REFERENCES 142Behling H, M Lichte. 1997. Evidence of dry and cold climatic conditions at glacial times in tropical 143 145 146 147 149 150
23 24

southeastern Brazil. Quaternary Research (48): 348-358. Brazilian highland and their implication for conservation and management of modern Araucaria forest and grassland ecosystems. Philosophical Transactions Royal Society B 362: 243251. fluviais do rio Iguau, Paran, Brasil. Tesis Ingeniero Florestal. Curitiba, Brasil. Universidade Federal do Paran. 510 p.
6

144Behling H, VD Pillar. 2007. Late Quaternary vegetation, biodiversity and fire dynamics on the southern

148Curcio GR 2006. Relaes entre geologia, geomorfologia, pedologia e fitossociologia nas plancies

25

Selecting ecotonal landscape units

151Duarte LS, MB Carlucci, SM Hartz, VD Pillar. 2007. Plant dispersal strategies and the colonization of 152 153 155 157 158 160 161 162 164 166 167 Araucaria forest patches in a grassland-forest mosaic. Journal of Vegetation Science18: p. 847-858. for Everyone. Amostrais de Paisagem (UAPs) em campos e em ectonos campo/floresta no Rio Grande do Sul Projeto SISBIOTA. Guimares eds. Patrimnio natural dos Campos Gerais do Paran. Ponta Grossa, Brasil. Editora UEPG. p 93-98. Acessed at 10 fev. 2012. Available at: http://ri.uepg.br:8080/riuepg//handle/123456789/452 between 1974 and 1999 in Southern Brazil. Community Ecology 5: p. 197-202. Forneck. 2007. Brazils neglected biome: The South Brazilian Campos. Perspectives in Plant Ecology,

154ESRI. Environmental Systems Research Institute.2006. ArcGIS. The Geographic Information System 156Hasenack H, E Weber E, E Vlez E, R Cardoso. 2011. Procedimento para seleo de Unidades

159Moro RS, MRB Carmo. 2007. A vegetao campestre nos Campos Gerais. In MS Melo, RS Moro, GB

163Oliveira JM, VD Pillar. 2004. Vegetation dynamics on mosaics of Campos and Araucaria forest 165Overbeck GE, SC Mller, A Fidelis, J Pfadenhauer, VD Pillar, CC Blanco, II Boldrini, R Both, ED

168Evolution and Systematics 9: 101-116. 169SEMA/IAP. 2004. Plano de Manejo: Zoneamento Ecolgico-Econmico, Plano de Manejo e 170 171 172 regulamentao legal da rea de Proteo Ambiental da Escarpa Devoniana. Curitiba, Brasil. IAP/MRS. 350p. Acessed at 22 jan. 2012. Available at: www.pr.gov.br/meioambiente/iap/pdf/devoniana_pl_manejo.pdf

27 28