Land Use Policy

(1)

Land Use Policy 111 (2021) 105771

Available online 28 September 2021

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

The role of spatial planning in land change: An assessment of urban planning and nature conservation efficiency at the southeastern coast of Brazil

Ana Beatriz Pierri Daunt

^a^,^*

, Luis Inostroza

^b^,^c

, Anna M. Hersperger

^a

aSwiss Federal Institute for Forest, Snow and Landscape Research WSL, Zürcherstrasse 111, CH-8903 Birmensdorf, Switzerland

bRuhr-University Bochum, Institute of Geography, Universit¨atsstraße 150, D-44801 Bochum, North Rhine-Westphalia, Germany

cUniversidad Aut´onoma de Chile, Av. Pedro de Valdivia 425, Providencia, Regi´on Metropolitan, Chile

A R T I C L E I N F O Keywords:

Nature conservation Planning outcome Spatial modelling Urban compactness

A B S T R A C T

Urban expansion is expected to continue at a fast rate, precisely in peri-urban areas of developing countries surrounded by biodiversity hotspots. The need to assess and potentially restructure urban and environmental planning instruments becomes apparent in scenarios where urban expansion is difficult to manage. Indicators based on spatially explicit datasets have been suggested as effective tools to evaluate spatial planning outcomes because they can shed light on the efficiency of planning measures and the fulfilment of claimed goals. In this work, we evaluated the conformance of stated spatial planning goals and the outcomes in terms of urban compactness, basic services and housing provision, and nature conservation for different land-use strategies. We evaluate the 2005 Ecological-Economic Zoning (EEZ) and two municipal master plans from 2006 in a coastal region in S˜ao Paulo State, Brazil. We used Partial Least Squares Path Modelling (PLS-PM) to explain the relationship between the plan strategies and land-use change ten years after implementation in terms of urban compactness, basic services and housing increase, and nature conservation. Our findings suggest that the evaluated plans were influenced by the land-use pattern at the time when the plan was approved (2005). For all evaluated plans, the Urban Use strategy was important to explain the Urban Compactness, but most of the new urban isolated areas occurred outside of the zones where the Urban Use strategy was applied. Two out of three of the evaluated plans were considered efficient in terms of nature conservation. In general, the Urban Use strategy can be considered successful in promoting more compact patterns of new build-up areas (axial and infill growth), but not in containing the emergence of new isolated areas outside the zones with Urban Use strategy. Our findings are in line with those from similar studies showing that areas outside of urban cores are often deprived of efficient spatial planning. The increase in Basic Services and Housing was not sufficient to attend the regional demand, and the inadequacy of these services remains a problem in the region. Future policies for land-use management in NCSP need to address the increasing demand for basic services and housing and to enable urban development inside urban core areas.

1. Introduction

Latin America and the Caribbean (LAC) is the second most urbanized region in the world, with 80.87% of the population living in urban areas (World Bank, 2020), a large proportion of informal housing, and spatial segregation with the most vulnerable populations heavily concentrated in peri-urban areas (Inostroza, 2017). The urban areas in LAC grow very rapidly, mostly at the periphery of existing urban areas as informal

settlements (Inostroza et al., 2013), which usually exist outside the planning systems and land-use governance (Fernandes, 2007).

The rate of urban expansion is expected to continue to increase rapidly, precisely in peri-urban areas and in coastal zones surrounded by biodiversity hotspots (Ellis and Ramankutty, 2008; Elmqvist et al., 2013;

Verburg et al., 2015). The fast and intense urbanization process in coastal zones is usually linked with vulnerability to extreme climate events, pollution, deforestation, habitat and biodiversity loss, violence

* Corresponding author.

E-mail addresses: beatriz.daunt@wsl.ch, beatrizpd@gmail.com (A.B.P. Daunt), luis.inostroza@ruhr-uni-bochum.de (L. Inostroza), anna.hersperger@wsl.ch (A.M. Hersperger).

Contents lists available at ScienceDirect

Land Use Policy

journal homepage: www.elsevier.com/locate/landusepol

https://doi.org/10.1016/j.landusepol.2021.105771

Received 5 March 2021; Received in revised form 9 August 2021; Accepted 14 September 2021

(2)

and crime (Elmqvist et al., 2013; Maricato, 2015; World Bank, 2006).

There is a great need for effective tools, policy instruments and programmes to manage these rapidly changing environments (World Bank, 2006). Land-use and environmental planning are the key activities through which planners can promote more efficient urban expansion and conservation activities. There is a great variety of planning instruments and purposes, including project planning, master planning, land-use planning and strategic planning, which influence patterns of land use and cover (Couclelis, 2005; Lyles et al., 2016; Rudolf and Gr˘adinaru, 2017).

A number of land-use and environmental policies, as well as spatial strategies, were developed in Brazil at the end of the 20th and early 21st century to achieve more efficient urban growth. The 1988 Brazilian Federal Constitution established principles for the housing property function and the democratic management of cities (Brasil, 1988).

Ecological-Economic Zoning (EEZ) was developed in Brazil during the 1980s as one of the principal instruments of national environmental policy, by coupling concepts of ecological protection with economic development. The 2001 City Statute requires a municipal master plan for cities with more than 20,000 inhabitants (Cohen et al., 2019; World Bank, 2006), and the creation of the City Ministry (2003) and the Pro- gram for Development Acceleration (PAC) were very important to increase investments needed to improve housing programmes (Rocco et al., 2019). The Brazilian policies for urban management, developed during the early 21st century, are recognized worldwide as a progressive reform and an important means to influence and discuss the right to the city and social justice throughout a new urban agenda. Despite being considered remarkable in the history of Brazilian urban policy-making, the principles provided by the City Statute have not been sufficient to solve land-use problems that originated from historical social in- equalities (Maricato, 2010). Around 87% of the Brazilian population lives in urban areas (World Bank, 2020), and the majority of the population still lives in informal housing in large cities, and rarely receive the benefits of legal regulations (Maricato, 2015; Rocco et al., 2019). For this reason, the suburbanization phenomenon is often deprived of efficient governance and strategic planning (Fernandes, 2007). The neces- sity to better understand the effectiveness of urban and environmental planning instruments in Brazil and throughout Latin America becomes apparent in scenarios where urban expansion and sprawl are difficult to manage (Inostroza et al., 2010; World Bank, 2006), and where social and environmental conflicts and vulnerability tend to increase (Castells, 1977; Elmqvist et al., 2013; Maricato, 2015).

1.1. Evaluation of spatial planning outcomes

The evaluation of land policy requires parameters that allow a long- term perspective on economic, ecological and social implications (Inostroza et al., 2010). Indeed, the evaluation of the quality and effectiveness of spatial plans plays an important role in the planning process (Hersperger et al., 2017). There is an increasing number of studies evaluating planning outcomes, traditionally conducted based on conformance and performance evaluation of plan implementation (Alfasi et al., 2012; Gr˘adinaru et al., 2017; Lyles et al., 2016) or based on comparative research (Cortinovis et al., 2019; Jehling and Hecht, 2021).

The quantitative comparison of planning goals and outcomes in terms of an effective land-use change remains a challenge, although advances have been made to evaluate plans outcomes, systematic methods for operationalizing information from planning documents into spatial data are still underdeveloped (Hersperger et al., 2018).

There is still a great need for robust instruments to evaluate the causal relationship between planning goals and outcomes in terms of land change, nature conservation and socioeconomic transformation (Cortinovis et al., 2019; Dembski et al., 2019; Jehling and Hecht, 2021;

Menzori et al., 2021; Rodrigues and Cazalis, 2020; World Bank, 2006).

Specifically regarding urban compact development, quantitative research has shown that urban planning instruments did not effectively

control urban sprawl in China (Wu et al., 2017), Brazil (Menzori et al., 2021; Pierri-Daunt et al., 2021) and Israel (Alfasi et al., 2012), with a particular concern regarding the role of these policies in managing urban growth in the south globe (Horn, 2020). Environmental policies and protected areas have been extensively discussed as important mechanisms for nature conservation (Rodrigues and Cazalis, 2020;

Steiner, 2008), and quantitative research has shown that these policies can be successfully in promoting forest persistence and afforestation in Brazilian Atlantic Forest (Pierri-Daunt et al., 2021; Silva et al., 2016), and in enhancing biodiversity (Topor et al., 2019).

The aim of this study was to develop a framework to evaluate the outcomes of spatial plans, and to apply it in a study area in a coastal region in S˜ao Paulo State, Brazil. Our study contributes to much-needed research on evaluating spatial planning outcomes, with a framework based on a quantitative and causal relationship, that can be applied to different scales and contexts, and thus could be useful for a wide range of evaluations. This method helped us to describe, in a spatially explicit manner, the efficiency of the plans in fostering compact cities, providing basic services and supplying housing, and ensuring nature conservation.

Specifically, the following questions were addressed: Can the observed land-use changes be attributed to plans or rather to other drivers? Which strategies in spatial plans have been effective for nature conservation?

Which strategies in spatial plans have been successful in promoting urban compactness? Which strategies in spatial plans have been successful in promoting basic services and housing increase?

2. Analytical framework for evaluating plan outcomes: from plan to land change

Our analytical framework consisted of a conceptual model (Section 2.1) and concepts for measuring plan outcomes (Section 2.2).

2.1. Conceptual model

Fig. 1 shows the conceptual model used to evaluate the plan outcomes, which involves mechanisms from plan-making to plan- implementation stages, based on Hersperger et al. (2019), LeGates and Stout (2011) and Steiner (2008). Spatial planning provides strategic actions for several goals related to more efficient urban growth, landscape multifunctionality and nature conservation, such as new residential areas, improvement of the transportation network and basic services, expansion of green and rural areas, and preservation of native forests (Hersperger et al., 2019; Steiner, 2008) (Fig. 1, Plan Goals). Plan implementation is a complex process in itself that includes various strategies and procedures to realize the goals established by a spatial plan, and it involves many actors regarding the territorial governance, such as politicians, planners, lawyers and public administrators (Albrechts, 2010; Steiner, 2008), funding mechanisms (Hersperger et al., 2019; Oliveira and Hersperger, 2018) and external forces (Hersperger et al., 2019; Palka et al., 2020) (Fig. 1, Implementation Process). Imple- mentation leads to outcomes regarding urban growth, nature conservation, and housing and basic services. (Fig. 1, Outcomes).

Fig. 1. Conceptual model to evaluate spatial planning outcomes in terms of patterns of urban expansion, forest cover and quality, and basic services and housing provision.

Adapted from Hersperger et al. (2019), LeGates and Stout (2011) and Steiner (2008).

(3)

To assess the efficiency of the spatial plans in fostering urban development and nature conservation, the proposed framework focuses on measurable outcomes in terms of land change, i.e. patterns of urban expansion and changes in forest cover and quality, and on the assessment of basic services and housing provision.

Although spatial planning can be important to drive land-use change, plans are not the only cause of change. Patterns of urban expansion and changes in forest cover are usually a result of a complex interaction of different actors and driving forces (Geist and Lambin, 2002; Plieninger et al., 2016). To identify these other drivers and their relative contribution to land-use change processes, the framework recognizes possible drivers, such as changes in population density and socioeconomic con- ditions, and topographic information (Fig. 1, Other drivers).

This framework can be applied at different scales and in different geographic contexts, and it can include different objectives and thus be useful for a wide range of spatial planning evaluations. Indicators based on spatially explicit data and respective change across time have been suggested as an effective tool to evaluate the potential divergence between planning objectives and outcomes in terms of observed land changes on the ground, and the effectiveness of the implementation process (Cortinovis et al., 2019; Gr˘adinaru et al., 2017).

In this study, we evaluated the deviation between stated spatial planning goals and the outcome for different land-use strategies, from regional to municipal scales. We evaluated the conformance of planning outcomes in terms of Urban Compactness, Basic Services and Housing provision, and Nature Conservation, whereas we had no intention to assess or explain the implementation process. In our study, we focused on understanding the relationship between three types of strategies (Urban Use, Multifunctional Use and Native Forest Maintenance (NFM)). (Sup- plementary Material I) and plan outcomes (Section 2.2).

2.2. Measuring plan outcomes: from goals to outcomes

For this study, we define efficiency as the ability of a plan or strategy to successfully foster a desired or intended goal after ten years of plan implementation. To evaluate the planning outcomes, we proposed three concepts: Urban Compactness, Basic Service and Housing (BSH), and Nature Conservation, as described below.

2.2.1. Urban compactness

The compact city model has been discussed as a way to promote efficient urban development, because compact cities are associated with higher resource efficiency and reduced travel costs, citizen well-being, and social and cultural cohesion (Bibri et al., 2020; Dieleman and Wegener, 2004; LeGates and Stout, 2011; United Nations, 2015).

Regarding urban expansion, three patterns have been discussed in the literature for evaluating urban form and its changes in terms of frag- mentation and compactness: axial, infill and isolated (leap-frog) expansion (Aguilera et al., 2011; Inostroza et al., 2013). Axial and infill are considered to conform to the goal of efficient urban growth, while isolated expansion does not.

2.2.2. Basic Services and Housing (BSH)

The urban basic services, such as water, sanitation, drainage and energy, are vital for the development of more resilient and sustainable cities (United Nations, 2015). In Latin America, urban sprawl is frequently linked with an increase in areas deprived of these services (Cohen et al., 2019; Maricato, 2015). In Brazil, the national urban agenda has established that the State is the main party responsible for ensuring the provision of basic services and sufficient housing for the resident population (Brasil, 2001, 1988). For this reason, basic services and housing provision represent an important indicator of efficient urban growth in our study area.

2.2.3. Nature conservation

The discussion about promoting more sustainable and green cities

has increased both in research and in policy initiatives (LeGates and Stout, 2011). After many years of rapid deforestation, the Brazilian environmental policies are based on principles of deforestation decrease and ecosystem restoration. For this reason, most of the Brazilian spatial plans regulate, per zone or strategy, the percentage of native forest that must be conserved or recovered. In this way, changes in forest cover and quality are key aspects for evaluating efficiency in promoting nature conservation.

3. Methods

3.1. Study area and planning instruments

The study area lies in a coastal region in S˜ao Paulo State, Brazil, named here as the Northern Coast of S˜ao Paulo State (NCSP) (Fig. 2).

The region is an administrative division for regional planning that hosts four municipalities, covering an area of 1948 km²: Caraguatatuba, Ilhabela, Sao Sebasti˜ ˜ao and Ubatuba, with a total GDP of R$201,994,048 ($40,034,493). The NCSP faces a relevant planning challenge caused by considerable population growth during the last decades, which went from 87,777 inhabitants in 1980 to 281,800 in 2010 (Instituto Brasileiro de Geografia e Estatística, 2010; Instituto Brasileiro de Geografia e Estatística, 1980) and by rapid urban growth (167%) during the last 30 years (Pierri Daunt and Silva, 2019), mostly resulting from policies aimed at developing the tourism and transportation sectors (Ab’S´aber, 1986; Pierri-Daunt et al., 2021; Teixeira, 2013). At the same time, around 80% of NCSP is covered by native Atlantic Forest, and most of this forest area is located inside the limits of three protected State Parks (Pierri Daunt and Silva, 2019). The Brazilian Protected Areas Act of 2000 prohibits any human settlement within parks’ boundaries and it limits their use to nature conservation, research and educational purposes. Another important challenge for planners in NCSP region is the outline of the already started construction of large scales logistic en- terprises, such as roads and port expansion and increases in oil explo- ration (Teixeira, 2013). Considering that the region shows high rates of population growth (1.6% per year) (FSEADE, 2016) that might demand the construction of new residences (Rosemback et al., 2017), the region may be fast approaching a land use planning and management crisis. In order to better guide future planning activities, NCSP is an ideal study area to apply the proposed framework.

Under the national and regional legal systems, the municipalities are usually in charge of the urban spatial planning in Brazil. The national and state governments also play an important role in environmental and regional planning, such as the demarcation of protected areas, watershed management plans and Ecological-Economic Zoning (Rocco et al., 2019). To apply the framework at both the regional and local scale, we selected the Ecological-Economic Zoning (S˜ao Paulo, 2005) and two municipal master plans. Ecological-Economic Zoning (EEZ) was intro- duced by the state government in 2005, coupling ecological and economic dimensions, and has since become one of the main instruments to regulate land-use change in NCSP. It is a mixed policy that regulates different land-use types. Within the area covered by the EEZ plan, we chose two municipal plans in order to understand the role of both regional and municipal spatial planning. The 2006 Ilhabela (Ilhabela, 2006) and Ubatuba (Ubatuba, 2006) master plans have established the respective municipal land-use zoning with mixed policies that regulate various land-use types. The municipal plans should conform to the EEZ strategies, and are allowed to be even more restrictive. Although the Ilhabela plan has excluded the protected areas and the State Parks from its planning perimeter, we included these areas in our modelling to provide comparable results with the other plans. The two municipal plans are in line with many of the priority concepts from the contem- porary urban agenda, described in Section 2, and are grounded in the national policy for urban development, the City Statute. The three documents should be updated every 10 years (Fig. 2), and we evaluated the changes after 10 years of plan implementation (from

(4)

2005/06–2015).

We analysed the content of the three spatial plans, focusing on zoning regulations and descriptions. We assigned each zone to a land-use strategy: Urban Use, Multifunctional Uses and Protected Areas and Native Forest Maintenance (NFM) (Table 1, see detailed description of the dominant land-use strategies in Supplementary Material I).

3.2. Modelling approach

To understand the relationship between the three land-use strategies and the plan outcomes, we applied Partial Least Squares Path Modelling (PLS-PM) using R software (“plspm” package), of the R programming language, version 3.6 (R. Core Team, 2020), following Sanchez (2013) (Fig. 4). PLS-PM is frequently applied to estimate complex cause–effect relationships, and it is a powerful multivariate method for analysing multiple relationships between a set of blocks of variables, designated as

“latent variables” (LVs) (Fan et al., 2016; Latan and Noonan, 2017).

PLS-PM has been traditionally applied in social science and business studies (Latan and Noonan, 2017), and more recently in land-change science (Fan et al., 2016). Since PLS-PM is less restrictive in the combination and use of different variables (Latan and Noonan, 2017; Nitzl and Chin, 2017), it is expected to be especially suited to quantify the causal relationship between stated spatial planning goals and other driving forces to observed land-change outcomes. Therefore, it is thus a promising method for the outlined research questions.

We create a total of nine LVs (Fig. 3), organized into three categories:

land-use strategy, plan outcomes, and other drivers. The land-use strategy LVs, as described in Section 2.1, refers to Urban Use, Multifunctional Use and NFM. To represent the spatial plan outcomes, we organized the data into Urban Compactness, BSH and Nature Conservation concepts (see Section 2.2). Each LV was composed by a conjunct of measured variables (Fig. 3, Table 2) and the contribution of each variable was weighted with loading values from 0 (no contribution) to 1 (highest contribution).

The contribution of each land-use strategy LV to each plan outcome LV, as well the contribution of each driver LV to each plan outcome LV, were analysed using the coefficient of contribution (C). The expected and modelled relationship between all LVs can be visualized in the theoretical framework (Fig. 4). For the purpose of this study, we understand efficiency when a land-use strategy LV is contributing positively (C>0.2) with a specific plan outcome LV, and, at the same time, presented a contribution higher than the other drivers LVs. Specifically, the NFM strategy is considered efficient when it contributes to the Nature Conservation LV; the Urban Use strategy is considered efficient when it positively contributes (C>0.2) to the increase of BSH and Urban Compactness LVs; and the Multifunctional Use strategy is considered efficient if it contributes positively (C>0.2) to all plan outcomes LVs.

3.3. Data acquisition and organization

We acquired the data for the explanatory variables from different sources, as exemplified in Table 2 and detailed in Supplementary Ma- terial I. Since the model is spatially explicit, all input data were Fig. 2. (A) Map of Brazil showing the location of S˜ao Paulo State (SP). (B) Northern Coast of S˜ao Paulo State (NCSP) showing the Ecological-Economic Zoning (2005). (C) Ilhabela Spatial Plan. (D) Ubatuba Spatial Plan. The Ecological-Economic Zoning and the municipal strategic plans are detailed in Table 1.

(5)

transformed to a 30 m resolution raster. To deal with differences in measurement units, including indices, proportions and densities per unit area, all explanatory variables were normalized from − 2 to +2 using the norm function in the package “scales” in R (Wickham and Seidel, 2019).

To avoid multicollinearity, for each model we tested the correlations between the pairs of independent variables described in Table 2, using

the corr function in the “stats” package in R, which returns a simple correlation matrix, and selected only variables with a correlation coefficient of less than 0.7. To evaluate spatial autocorrelation, we used bootstrap validation with 1000 samples for each model, which is part of the “plspm” package in R. The major steps are presented in the workflow diagram (Fig. 3).

Table 1

Spatial planning instruments in the Northern Coast of S˜ao Paulo State (NCSP). Zone descriptions and assigned land-use strategies, i.e. Urban Use, Multifunctional Uses, and Protected Areas and Native Forest Maintenance (NFM). APP=Permanent Preservation Areas, defined by the Brazilian Forest Code as areas with the environmental function of preserving water resources, the landscape, geological stability and biodiversity.

Spatial plan Scale Land-use strategy

EEZ Ecological-Economic Zoning

(2005) Regional: Caraguatuba, Ilhabela, S˜ao Sebasti˜ao, Ubatuba

Zone Zone regulation /description Area

(km²) Dominant land-use strategy

Protected areas (PA) Forest and ecosystem conservation, under Federal Law (SNUC) and SP State Law that creates the

state parks and other protected areas 1409.8 NFM

Z1 Forest and ecosystem conservation, ecosystem restoration 195.3 NFM

Z2 Natural resource and ecosystem conservation and water supply, landscape heritage conservation, increase in basic services. High quantity of APPs and slopes from 30% to 47%, flood and slipping risk. Allow rural use and a few new built-up areas (<20% of use/cover);

aquiculture, mining and sustainable nature management

107.2 Multifunctional Use

Z3 Agricultural use, rural villages, and multiple land-uses are allowed, sanitation service

improvement is a premise/assumption (allowed on <30% of non-native forest use/cover) 86.6 Multifunctional Use Z4 Dense and efficient urban development. Urban area maintenance to avoid urban sprawl. Increase

in basic urban services and mobility, complete urban services and infrastructure (basic services for all residents), social housing programmes, APP restoration, increase in green infrastructure, prioritization of the previous occupied areas (compact cities)

89.0 Urban Use

Z4OS Lower the impact of urban development, maintain urban areas, avoid urban sprawl, complete urban services and infrastructure (basic services for all residents), APP restoration and increase in green infrastructure

34.1 Urban Use

Z5 Industrial use, dense and populated urban areas, complete urban services and infrastructure

(basic services for all residents), increase in urban green areas 23.6 Urban Use Municipal Strategic Plan (2006) –

Ilhabela Local: Ilhabela Area

Zone Zone regulation /description

PA Forest and ecosystem conservation, under Federal Law (SNUC) and SP-State Law that created the

Ilhabela State Park. 283.1 NFM

ZRT Terrestrial Restricted Zone Forest and ecosystem conservation. Any land-use is forbidden as result of ecological and/or

geotechnical characteristics; no changes are allowed. 2.2 NFM

ZR1 restricted Very restricted zone due to geotechnical characteristics. A small amount (maximum 10% of use/

cover) of built-up area is allowed for residential and commercial purposes. Services are the land owner’s responsibility. No deforestation of native forest is allowed.

ZR2 restricted Very restricted urban use for residential purposes, and a small (maximum 15% for residential use and 20% for commercial use) increase in built-up area is allowed. The sanitation project and water cistern colocation are mandatory for approval and are the land owner’s responsibility. No deforestation for any land-use is allowed.

ZU 1 Urban Restricted Zone Urban use with restricted increase in urban use (maximum of 20% for residential use and, 30%

for commercial use). The sanitation project and water cistern colocation are mandatory for approval and are the land owner’s responsibility. Deforestation is regulated through state and federal laws.

1.9 Urban Use

ZU 2 Urban Restricted Zone Urban use with restricted increase in urban use (maximum of 30% for residential use and 40% for

commercial use). Basic services are mandatory and are provided by the municipality. 6.4 Urban Use ZIE Specific Interest – traditional

communities and archaeological sites Landscape as a common heritage, archaeological sites and traditional communities, agriculture and extraction, touristic uses. Basic services are the municipality’s responsibility and all residents should receive them. APP restoration, sustainable uses, and sparse build up is allowed.

Improvements to quality of life, security, health and education are mandatory (provided by govern)

Municipal Strategic Plan (2006) Local: Ubatuba Area

Zones Zone regulation /description

PA Forest and ecosystem conservation, under Federal Law (SNUC) and SP-State Law that created the

state parks. 513.7 NFM

Traditional communities Traditional territory and community use, shared uses, landscape as a common heritage, nature conservation and touristic uses. Build up is not encouraged, but communities need infrastructure under the federal law for traditional communities and for sustainable protected areas (SNUC).

Basic services should be developed.

Public lands Shared and public uses, landscape as a common heritage, nature conservation and touristic uses. 5.1 NFM Islands and coastal steep areas Landscape as a common heritage, biodiversity conservation and touristic uses. 109.0 NFM

Coastal areas Touristic sector development, tourist houses and accommodation. 30.7 Urban Use

Backlands Agricultural uses, multifunctional landscapes, tourism development, residential housing development, a few built-up areas, improved connectivity and housing increase near the transportation network, increase in public services and other urban services such as commerce, accommodation and restaurants.

Dense urban Efficient urban development, accessibility, urban services – priority for City Statute development 21.7 Urban Use

(6)

We created a presence-absence raster (binary) for each zone in each plan. The EEZ and Ubatuba plan include the protected areas that were previously created in the zoning, while the Ilhabela plan excludes them.

We organized the zones into three strategic groups of latent variables, as described above: Nature Conservation, Multifunctional Use, and Urban Use (Fig. 3 and Table 1).

To infer the value of Urban Compactness, we used the built-up cover changes from 2005 and 2015. We acquired two Landsat Collection 1 Higher-Level Surface Reflectance images distributed by the U.S.

Geological Survey (USGS), covering the entire study area (paths 76 and 77, row 220, WRS-2 reference system, https://earthexplorer.usgs.gov/).

We classified one image acquired by the Landsat 5 Thematic Mapper (TM) sensor on 2005–05–150, and one image from the Landsat 8 Operational Land Imager (OLI) sensor from 2015 to 08–15, using the Support Vector Machine (SVM) supervised algorithm (Hsu et al., 2003), in ENVI 5.0 software (see Supplementary Material I for details regarding the classification accuracy). To describe the spatial configuration of the urban expansion, we classified the new built-up areas into axial, infill and isolated, following Inostroza et al. (2013) (See Supplementary Ma- terial I).

To infer the value of Nature Conservation, we used the changes in forest cover (acquired as described above) and changes in Normalized Difference Vegetation Index (NDVI). We selected the areas with persistent forest (pixels with forest cover in 2005 that remained as forest cover in 2015) and afforestation (pixels without forest cover in 2005 that were converted to forest cover in 2015). The NDVI was obtained from the same source used for the LULC data. With the Google Engine platform, we used an annual average for the best pixels (without clouds) for 2005 and 2015, and we calculated the changes between dates. We used increases of >0.2 NDVI to represent an improvement in forest quality.

The data used to infer the values of BSH and the Socioeconomic and Population drivers was derived from the Brazilian Federal Census data (Instituto Brasileiro de Geografia e Estatística, 2010; Instituto Brasileiro de Geografia e Estatística, 2000). Population density, permanent housing density, mean income, basic education, and the percentage of houses receiving waste collection, sanitation and water provision services, called basic services in the context of this study, were calculated per 30 m pixel. The Human Development Index is only available at the municipality level. We attributed the HDI for the vector file with the municipality border, and we rasterized (30 m resolution) this file in QGIS. Annual rates of change were then calculated to allow compara- bility between LULC periods. To infer the BSH, we used only areas with an increase in permanent housing density and basic services provision (See Supplementary Material I).

Topographic measures are frequently modelled as a driver of urban growth (Silva et al., 2016). To infer the values of the topographic driver, we used the slope data and the Topographic Index Position (TPI) based on the digital elevation model from SRTM (30 m resolution) produced by ALOS (freely available at eorc.jaxa.jp/ALOS/en/about/about_index.

htm), and both variables were considered constant from 2005 to 2015 (See Supplementary Material I).

4. Results

4.1. Measuring plan outcomes

In general, the evaluated plans were similar regarding the variable’s loadings (see loadings in Table 3). Persistent forest covered 84.75% of the study area (Fig. 5), and it was the most important variable in inferring the value of Nature Conservation LV in all models. All the others LULC change covered 15.25% of the study area. Although 68.9%

of the study area showed an increase of NDVI over time, this variable had a low contribution to this LV. Urban persistence was the most important variable in inferring the value of Urban Compactness LV in all models (Table 3 and Fig. 6). These findings suggest that the evaluated plans are influenced by the land-use condition at the time the plan was

approved (land-use in 2005).

Regarding the classification of new urban areas, the percentage of urban isolated areas was higher than that of axial or infill areas. Despite this finding, the loadings for urban isolated areas were lowest, which suggests that the Urban Compactness LV illustrates the development of more compact cities (Table 3). For the area of the EEZ and the Ilhabela plan, urban isolated areas occurred mainly inside zones that are classified as Multifunctional Use strategy; for the area of the Ubatuba plan, the urban isolated areas occurred mainly inside zones that are classified as NFM strategy (Supplementary Material II).

Our results show different loadings for the variables used to infer the value of BSH LV (Fig. 7): For the EEZ, water and waste services presented the highest loadings, whereas water service and housing density had the highest loadings for the Ilhabela plan and sanitation and water services for the Ubatuba plan.

4.2. Dominant land-use strategies: regional and municipal plans

The regional and municipal plans were not always consistent. For the territory of Ilhabela, the municipal plan delineates smaller areas for Urban Use (8.3 km²) and NFM (285.3 km²) than specified in the EEZ (20.1 km²and 306.1 km², respectively). In contrast, the Ilhabela municipal plan dedicates a larger area to the Multifunctional Use strategy (38.3 km²) than specified in the EEZ (9.2 km²). In Ubatuba territory, Urban Use (52.4 km²) and NFM (627.8 km²) strategies amount to a larger area than foreseen in the EEZ (40.4 km²and 598.5 km², respectively), while the municipal plan dedicates a smaller area to the Multifunctional Use strategy (30.5 km²) than indicated in the EEZ (65.1 km²) (Supplementary Material II Figure B).

4.3. Ecological-Economic Zoning

Goodness-of-Fit (GoF) measures the overall prediction performance of the model. The model generated to evaluate the EEZ’s efficiency per- formed well regarding the GoF (0.43) and regarding all R²values (Fig. 8). All the evaluated relationships were significant (p-value

<0.001), except for the relationship between Socioeconomic drivers and the Urban Compactness LVs (Supplementary Material II Table D1). The NFM strategy explained the Nature Conservation (C=0.39). The Multifunctional Use strategy explained only the BSH (C=0.29). The Urban Use strategy explained the BSH (C=0.39) and the Urban Compactness (C=0.65). Socioeconomic drivers explained only the BSH (C=0.39). Population and Topography did not influence the plan outcomes, which suggests that EEZ is an important policy in fostering nature conservation inside the limits of the NFM strategy, and for the development of more compact cities with an increase of basic services and housing inside the zones for Urban Use strategy. However, we must highlight that the urban isolated areas grew mainly inside the zones for Multifunctional Use and NFM strategies, which suggests that the plan has not been efficient in containing the emergence of new isolated areas outside of zones with the Urban Use strategy.

4.4. Ilhabela Municipal Strategic Plan

The model generated to evaluate the Ilhabela plan’s efficiency per- formed best regarding GoF (0.46). All latent variables were considered important enough for use in evaluating plan efficiency (R²>0.3), and all evaluated relationships were significant (p-value <0.001) (Supple- mentary Material II Table D2). The Urban Use strategy explained the Urban Compactness (C=0.43) and the BSH (C=0.27) LVs. The Multifunctional Use strategy also explained the BSH (C =0.39). The NFM strategy was less efficient in promoting nature conservation (C=0.22) compared with the other plans. The inconsistencies between the regional and local plans might have influenced the lower coefficient value for the NFM strategy when compared with the EEZ model. We quantified 0.33 km²of urban isolated areas inside the zones that

(7)

regulate the NFM strategy (Supplementary Material II Table E). Popu- lation explained the Urban Compactness (C=0.22) and Nature Con- servation (C= − 0.22) LVs, and Socioeconomic drivers explained BSH (C= −0.43), but in general, the contribution of the other drivers was smaller than that of the three land-use strategies (Fig. 9).

4.5. Ubatuba Municipal Master Plan

The overall performance of the Ubatuba master plan model was

worse than that of the EEZ and Ilhabela plan (GoF=0.38). BSH and Urban Compactness LVs were considered important (R²>0.3) enough for use in evaluating plan efficiency, but the R²for the Nature Conser- vation LV was rather low (R²=0.2). All the evaluated relationships were significant (p-value <0.001 or <0.01), except for the relationship between Socioeconomic drivers and Urban Compactness. The Urban Use strategy explained the Urban Compactness (C=0.51) and the BSH LVs (C=0.30). The BSH was also explained by the Socioeconomic drivers (C=0.38). Population explained the observed increase in BSH (C=0.21), Urban Compactness (C=0.16) and Nature Conservation (− 0.18). The Multifunctional Use strategy did not explain any evaluated any plan outcome LV. The inconsistencies between the regional and local plans might have influenced the lower coefficient value for the Multifunctionality strategy, when compared with the EEZ model. The Nature Conservation LV was therefore explained by the NFM strategy (C=0.30) and by the Topography drivers (C=0.19) (Fig. 10), but the urban isolated areas mainly occurred under the NFM strategy (Supple- mentary Material II Table E).

5. Discussion

We developed a method to evaluate plan outcomes with spatial data, which made it possible to evaluate the efficiency of the plans in fostering the observed patterns of built-up and forest cover, and the increase in BSH. In this analysis, the evaluated plan outcomes were able to illustrate the investigated changes after 10 years of plan implementation at both the regional and local scale. Differences in model performance and in the contribution of the evaluated relationships were found and are discussed in the following sections.

5.1. Can the observed land-use changes be attributed to plans or rather to other drivers?

Urban Compactness and Nature Conservation were mainly explained Fig. 3. Study workflow Data organization, pre-processing and Partial least Squares – Path Modelling main steps. Abbreviation: Multi=multifunctional.

Fig. 4. Partial Least Squares Path Modelling theoretical framework with three categories of latent variables: land-use strategies (blue), plan outcomes (red), and other drivers (yellow). Abbreviations: NFM=Native Forest Maintenance, Multi Use=Multifunctional Use; BSH=Basic Services and Housing. Aiming for reproducibility, the three models have the same structure: the NFM strategy is always evaluated with the Nature Conservation; the Multifunctional Use strategy is always evaluated with the three LVs for plan outcomes and the Urban Use strategy is evaluated with BSH and Urban Compactness. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(8)

by the Urban Use and NFM strategies, respectively. These findings suggest that the observed changes regarding urban form and nature conservation can be attributed to the evaluated plans, although some particularities were found and are discussed below.

Under the national urban agenda (Brasil, 2001, 1988), the zones with the Urban Use strategy recommend the provision of BSH for the entire resident population. We found that the BSH has been more influenced (higher coefficient) by the changes in socioeconomic data than by the Urban Use strategy, which suggests that the spatial plans might not be efficient enough to foster the observed BSH increase. A report produced by the NCSP Watershed Management Committee has shown that the increase in BSH was not sufficient to meet the regional demand (Comite de Bacias Hidrogr´aficas do Litoral Norte CBHLN, 2016). The same committee estimated that around 40% of NCSP houses receive public sanitation services and 80% of the houses receive clean water and waste collection (Comite de Bacias Hidrogr´aficas do Litoral Norte CBHLN, 2018). Future policies in NCSP need to address the increasing demand for basic services and housing inside the zones with Urban and Multi- functional Use strategies.

The BSH in Ubatuba and the Urban Compactness in Ilhabela have also been influenced by population dynamics, although with smaller coefficients than the Urban Use strategy. In contrast, Population did not contribute to any EEZ plan outcome, which may reflect differences between regional and local findings, and municipal planners should therefore be urged to better address resident-population dynamics.

These findings are interesting when contrasted with results from earlier studies that the importance of population density in driving land changes has been decreasing worldwide (Lambin et al., 2001) and in NCSP in particular (Pierri-Daunt et al., 2021).

The Nature Conservation was best explained by the NFM strategy, although Topography also contributed to explaining Nature Conserva- tion in the Ubatuba and NCSP models. The region has many steep areas that are unstable for human settlement, and the protected areas and the zones that regulate nature conservation in NCSP are frequently located in steep areas, which might explain our findings.

5.2. Has the NFM strategy been effective for nature conservation?

Our findings suggest that the NFM strategy has been efficient in promoting nature conservation in terms of forest persistence, especially for the EEZ and Ubatuba, but we could not confirm its efficiency in improving NDVI in NCSP and in the two municipalities. The NFM strategy mainly regulates restricted and protected areas, especially in three state parks, which have already been suggested as an important driver for forest persistence in NCSP (Pierri-Daunt et al., 2021). Envi- ronmental policies and planning activities have been discussed as an important tool for nature conservation (Rodrigues and Cazalis, 2020;

Steiner, 2008), and especially for landscape stability (Plieninger et al., 2016), forest persistence and afforestation (Pierri-Daunt et al., 2021;

Table 2

Dependent and independent variables used to infer the values of the latent variables to model the relationship between the three types of strategies and the plan outcomes.

Variable Information/

description Source/data origin Time range Spatial plans (independent variables)

Ecological- Economic Zoning (EEZ)

EE zones CPLA – SP 2005

Ilhabela master

plan (zoning) Ilhabela master plan

zones Ilhabela

municipality 2006 Ubatuba master

plan (zoning) Ubatuba master plan

zones Ubatuba

municipality 2006 Measuring plan outcomes (depended variables)

Urban Compactness Urban persistence Built-up

2005=Built-up 2015

Landsat Collection 1 Higher-Level Surface Reflectance 30 m

2005–2015

Urban infill New built-up pixels

classified as infill Landsat Collection 1 Higher-Level Surface Reflectance 30 m

2005–2015

Urban axial New built-up pixels

classified as axial Landsat Collection 1 Higher-Level Surface Reflectance 30 m

2005–2015

Urban isolated New built-up pixels

classified as isolated Landsat Collection 1 Higher-Level Surface Reflectance 30 m

2005–2015

Nature Conservation Forest cover

persistence Native forest cover persistence from 2005 to 2015

2005–2015

Forest cover gain Native forest cover gains from 2005 to 2015

NDVI Changes in

Normalized Difference Vegetation Index

2005–2015

Basic Services and Housing Permanent

housing density increase

Increase in the density of permanent housing units

Federal census

(BIGS)a 2000–2010

Waste service

increase Increase in waste collection service (%)

Federal census

(BIGS)a 2000–2010

Sanitation service

increase Increase in sanitation service provision (%)

Federal census

(BIGS)a 2000–2010

Water service

increase Increase in clean water provision service (%)

Federal census

(BIGS)a 2000–2010

Other drivers (independent variables) Socioeconomic

Basic education Basic education in % per census sector (change per year)

Federal census

(BIGS)a 2000–2010

Human Development Index (HDI)

HDI per municipality (change per year)

Federal census

(BIGS)a 2000–2010

Mean income Mean income in Reais (R$) per census sector (change per year)

Federal census

(BIGS)a 2000–2010

Population

Population density Population density per pixel (change per year)

Federal census

(BIGS)a 2000–2010

Topography

Slope Slope ^◦ ALOS 30 m

Table 2 (continued)

Variable Information/

description Source/data origin Time range Data from 2005 and 2015^b Topographic Index

Position (TPI) Topographic Index

Position (TPI) ALOS 30 m Data from

2005 and 2015^b a All data from Federal Census (Instituto Brasileiro de Geografia e Estatística) refers to permanent inhabintant and permanent housing unit. The Federal Census data was transformed into denity values foa a pixel size of 30 m (see Supplementary Material I).

b The information is the same for 2005 and 2015.

Source:Sources: Brazilian Institute of Geography and Statistics (2000; 2010); Sao ˜ Paulo State Environmental Planning Division (CPLA-SP).

(9)

Table 3

Variables areas and percentages in the North Coast of S˜ao Paulo State (NCSP; 1948 km²) and in the Ilhabela (335.57 km²) and Ubatuba (710.6 km²) municipalities. The loading values quantify the contribution of the variable to each LV of plan outcome, and are measure from 0 (no contribution) to 1 (highest contribution).

Latent Variable Variable NCSP Ilhabela Ubatuba

Area km² % Loading Area km² % Loading Area km² % Loading

Nature Conservation Forest persistence 1651.01 84.75 0.99 305.46 91.03 0.98 639.84 90.04 0.98

Forest gain 20.38 1.05 0.36 3.32 0.99 0.47 6.19 0.87 0.40

NDVI increase 1342.73 68.93 0.15 72.57 21.63 -0.08 324.72 45.70 0.07

Urban Compactness Urban persistence 86.48 4.44 0.92 7.87 2.34 0.82 24.76 3.48 0.92

Urban infill 11.5 0.59 0.28 1.87 0.56 0.4 1.77 0.25 0.29

Urban axial 23.5 1.21 0.23 2.46 0.73 0.33 8.14 1.15 0.23

Urban isolated 25.53 1.31 0.11 3.18 0.95 0.14 8.5 1.20 0.01

Basic Services and Housing Permanent housing density increase 682.93 35.06 0.34 44.50 13.26 0.90 267.05 37.58 0.56 Waste service increase 497.94 25.56 0.75 133.95 39.92 0.38 119.53 16.82 0.73 Sanitation service increase 153.17 7.86 0.65 20.08 5.98 0.73 37.67 5.30 0.74

Water service increase 280.61 14.41 0.88 19.57 5.83 0.74 83.05 11.69 0.82

Fig. 5. Nature Conservation. Forest cover change (forest persistence and forest gain; A–C) and NDVI increase (D–F) from 2005 to 2015. (A, D) North Coast of Sao Paulo State (NCSP); (B, E) Ubatuba municipality; (C, F) Ilhabela municipality.

(10)

Silva et al., 2017).

We identified urban isolated areas subject to the NFM strategy in Ilhabela, which might explain the lower coefficient of NFM in that municipality than in EEZ and Ubatuba. If urban growth continues to grow inside these areas, the efficiency in terms of forest protection might decrease in the near future.

5.3. Which strategies in spatial plans have been successful in promoting efficient urban development?

For all three evaluated plans, the Urban Use strategy was important to explain the Urban Compactness and BSH LVs. The values for Urban Compactness were mainly influenced by urban persistency, axial and infill build-up, with a very low contribution from the new isolated areas, Fig. 6. Urban Compactness. Spatial characterization of increases in built-up area from 2005 to 2015. (A) North Coast of Sao Paulo State (NCSP); (B) Ubatuba municipality; (C) Ilhabela municipality.

Fig. 7. Basic Services and Housing LV (BSH). Areas in the North Coast of Sao Paulo State (NCSP) with an increase in the density of permanent housing units and basic services provision (sanitation, water and waste services) from 2000 to 2010.

(11)

Fig. 8.PLS-PM for the Ecological-Economic Zoning: Model for the relationship between the three types of strategies and the EEZ outcomes. Nature Conservation R²=0.34;

BSH R²=0.53; Urban Compactness R²=0.5. Values refer to the coefficient of contribution (C). Abbreviations:

EEZNFM=Ecological-Economic Zones (EEZ) regulating native forest maintenance; EEZmult=EEZ with multifunctional use strategies; EEZurb=EEZ for the Urban Use strategy; Socioeco=Socioeconomic; Urban Com- pac=Urban Compactness; BSH=Basic Services and Housing; Nature=Nature Conservation. Only significant relationships are shown.

Fig. 9.PLS-PM for the Ilhabela (IB) Municipal Strategic Plan: Model for the relationship between the three types of strategies and the Ilhabela plan outcomes. Nature Conser- vation R²=0.30; BSH R²=0.69; Urban Compactness R²

=0.43. Values refer to the coefficient of contribution (C).

Abbreviations: IBNFM=IB zones regulating native forest maintenance; IBmult=IB zones with multifunctional use strategies; IBurb=IB zones for the Urban Use strategy;

Socioeco=Socioeconomic; Urban Compac=Urban Compactness; BSH=Basic Services and Housing; Nature-

=Nature Conservation. All evaluated relationships were significant.

Fig. 10.PLS-PM for the Ubatuba (UBA) Municipal Master Plan: Model for the relationship between the three types of strategies and the Ubatuba plan outcomes. Nature Con- servation R²=0.20; BSH R²=0.38; Urban Compactness R²=0.38. Values refer to the coefficient of contribution (C). Abbreviations: UBANFM=UBA zones regulating native forest maintenance; UBAmult=UBA zones with multifunctional use strategies; UBAurb=UBA zones for the Urban Use strategy; Socioeco=Socioeconomic; Urban Compac=Urban Compactness; BSH=Basic Services and Housing; Nature=Nature Conservation. Only significant relationships are shown.

(12)

which suggests that new build-up land classified as axial and infill emerged inside the zones designated for urban use. Nonetheless, BSH was explained better or equally by Socioeconomic drivers (Section 5.1) and by the Multifunctional Use strategy in Ilhabela.

On the other hand, the new built-up areas classified as urban isolated areas occurred mainly inside the zones with Multifunctional Use and NFM strategies. Although the Multifunctional Use strategy allows a few new built-up areas, we suggest that the evaluated plans have not been efficient in containing the emergence of new isolated areas. Our finding is in agreement with the literature about the weak role of the national urban agenda and spatial planning in driving urban development in Brazil (Cohen et al., 2019; De Souza, 2001) and the lack of efficient administration and spatial planning in the sprawling urban areas outside the city core (Fernandes, 2007). Worldwide, the inefficiency in regulating urban expansion, specifically regarding urban sprawling, has been documented as a result of multiple factors (Wu et al., 2017), such as territorial governance and political decision during the implementation process (Alfasi et al., 2012; Menzori et al., 2021; Horn, 2020), economic interests, such as neoliberal urban agenda and real estate market in- fluences (Wu et al., 2017; Horn, 2020; Pierri-Daunt et al., 2021), and top-down approaches without the involvement of stakeholders (Stein- berg, 2005).

Steinberg (2005) summarized the relationship between spatial planning and urban growth in Latin America as two approaches: a top-down approach, which focuses on the strategic demand of a city, and a bottom-up approach, which represents the social articulation of the citizens. In our case study, the top-down approach can be illustrated by the densification of the cities inside the limits of the urban zones. The bottom-up approach can be illustrated by the new urban isolated areas mapped inside the zones with Multifunctional Use and NFM strategies.

Further, our findings show that the evaluated plans were influenced by the land-use condition at the time when the plan was approved (land- use in 2005). These findings suggest that NCSP pursues a defensive planning strategy (Ahern, 2006), by reacting to the effects of historical processes that caused the observed spatial patterns, instead of proac- tively preventing urban growth. In Europe, Cortinovis et al. (2019) and (Shaw et al., 2020) have suggested that changes in peri-urban areas have changed or influenced the planning process in most of the evaluated case studies. In Israel, the limit of zones with urban strategies was updated aiming to include new urban areas not predicted by previous versions (Alfasi et al., 2012). Although these initiatives can be realistic in terms of planning success, authorities should develop more long-term planning tools and allocate efforts to successfully implement policies to manage urban growth. On the other hand, a literature review has shown that it is unlikely that urban growth management will work in developing countries, and that there is a need to move from traditional mechanisms of controlling sprawl to policies encouraging urban densification and enabling urban core development (Horn, 2020).

We found no clear pattern regarding the Multifunctional Use strategy outcomes: this strategy contributed to the BSH increase in Ilhabela and EEZ and had no influence on any LV for plan outcome in Ubatuba. Our models did not point to any LV that can be used to evaluate the Multi- functional Use strategy specifically, which might explain the low contribution from this strategy to the evaluated LVs. On the other hand, the zones with the Multifunctional Use strategy regulate the percentage of allowed agricultural and built-up areas, and these areas are frequently planned as buffer areas to mitigate the impact from the urban envi- ronment on protected areas (S˜ao Paulo, 2005). In general, we can suggest that the Multifunctional Use strategy is rather restrictive in terms of land change, and therefore was not efficient in promoting multifunctionality. Policies to develop combined urban and rural sustain- ability might be more efficient in promoting multifunctionality.

The dichotomy between forest reserves and urban growth has been documented for the study area and explained as a result of a complex combination of economic and political drivers for developing dense urban areas for tourism and transportation sectors, and environmental

policies for nature conservation (Ab’S´aber, 1986; Pierri-Daunt et al., 2021; Teixeira, 2013). As a consequence, other land-uses, such as agricultural uses, and the interests of the local residents have received less attention from the authorities and planners (Pierri-Daunt et al., 2021;

Pierri Daunt and Silva, 2019), and the presence of new isolated built-up areas has increased outside the zones with the Urban Use strategy.

In Brazil and Latin America in general, urban sprawl and spatial segregation have frequently been documented as the result of historical inequality of public policies and economic interest (Villaça, 2012);

consequently, they have been considered to have no direct link with planning activities. Future policies for land-use management in NCSP need to address the increasing demand for basic services and housing, and to contain urban sprawl, not only inside the zones with the Urban Use strategy. Improvements in Multifunctional Use strategies and policies can help to mitigate the impact of urban sprawl on protected areas and, further, to conciliate urban development and nature conservation.

Better communication between the S˜ao Paulo state government and the municipalities could help to decrease contradictions between plans (i.e.

between the EEZ and the municipal plans) and therefore help to drive more efficient urban development.

5.4. Limitations of the method and data set

The selection of variables for model the relationship between the three land-use strategies, drivers and plan outcomes should depend on the theoretical and behavioural assumptions (Verburg et al., 2004), but we were limited to the available data. The inclusion of additional variables would likely improve the model results and the assessment of efficiency in promoting urban development. For example, the transportation network and other accessibility assessments could be important measures of urban infrastructure improvement (Geurs and Ritsema van Eck, 2003), but this data is not available for the study area.

Other variables could also better represent the Socioeconomic drivers, if available; in particular, variables related to land prices would have improved our results. The approach to monitoring the forest cover dynamics can be considered a simplification of Nature Conservation assessment, and other variables might better describe the success in protecting the Atlantic Forest ecosystems, such as landscape metrics, diversity indexes, species distribution information, and other ecological indicators (Ribeiro et al., 2009).

The variables used to infer the values of the latent variables were acquired from different sources and differed in the unit and scale of measurement. Errors are inherent to LULC and spatial modelling, and the combination of these variables and might generate uncertainties in the model results. The fact that the Federal Census years (2000 and 2010) differ from the LULC years (2005 and 2015) might have intro- duced additional uncertainties. We partly adjusted for the mismatch by calculating the annual rates of change. However, the extrapolation beyond the common period from 2010 to 2015 remains a potential issue, especially because annual population growth rates were smaller from 2010 to 2020 (1.48) than from 2000 to 2010 (2.31) for the region (Fundaç˜ao SEADE, 2020). Unfortunately, this information is only available for the municipality level. For this reason, the effect of population on the evaluated LVs might have been overestimated in our models.

On the other hand, the variables acquired from the same source might be more correlated with themselves than with the other variables, and this might also have an effect on the model results. Although we tested the correlation and spatial autocorrelation with the bootstrapping resampling test, the quantified coefficients of Socioeconomic and Pop- ulation drivers on the BSH may have been influenced by the fact that this data was acquired from the Federal Census, and as a consequence, these relationships might also have been overestimated.

Interviews with planners and stakeholders could provide a more detailed comprehension of the complex process regarding plan- implementation activities and could help to strengthen the link