**1. Introduction**

A reader cannot safely assume to understand what the word 'forest' means in any new context, as the ecological (vegetation-based) and social (institutional, rule-based) meanings of the word 'forest' only partially overlap [1,2]. There can be 'trees outside a forest' [3] and 'forests without trees' [4]. Conversion of natural forest to plantation forestry is not considered to be deforestation, an issue that sparked debate when the UNFCCC climate change convention considered forest-specific policies [5]. What is excluded from the 'tree' concept, at the heart of common forest definitions, also matters: rubber (*Hevea brasiliensis*) can be considered to be a forest tree when grown for timber, or an agricultural commodity when tapped for its latex [6]. A number of global studies, however, have included the conversion of rubber monoculture (regardless of its primary purpose) to an oil palm monoculture under global 'deforestation' statistics [7]. The debate whether or not palms are included under the 'tree' concept has direct implications for whether or not the conversion of natural forest to an oil palm

monoculture can be called 'deforestation' [8] and for how much a 'deforestation-free' commitment of the palm oil industry means [9]. National definitions of the tree and forest concepts can differ from the global ones [10,11]. At the social-ecological system scale of a landscape, the concept of 'agroforest' describes tree-based vegetation managed by farmers, who often see labeling this as a forest as a threat [12,13]. Institutionally such land use may be legalized under 'community-based forest management' rules [14], but these arrangements may maintain forest authorities in the 'landlord' role, expecting a share in any harvestable goods or sellable services that the land may generate [15]. The absence, at least until recently, of formal recognition for agroforestry as a valid form of land use intermediate to 'forest' and 'agriculture' has not prevented the existence and spread of such land uses that defy the rules [16] (Figure 1).

**Figure 1.** Forest-Agriculture interface with the current segregated policy interpretation distinct from the more fluid 'integrated' reality in the landscape, with consequences for the meaning of 'deforestation' and 'loss of forest functions' [17].

In the international debate over consequences of tropical commodity production, 'forest' is primarily seen as a vegetation type or ecosystem, observable through remote sensing [18], with a rather arbitrary tree cover threshold used in a dichotomous classification. Global data sets based on a continuum interpretation with tree cover as metric are available and have been used to monitor changes in tree cover in agricultural lands [19]. Concerns about deforestation, and the concomitant loss of biodiversity and net C emissions, have led to institutional support in the climate change convention for government action to support 'reforestation' [20] and 'reducing emission from deforestation and forest degradation' [21,22]. In parallel, citizens in importing countries expressing concern about the footprint of their consumption have induced corporate actors to dissociate themselves from deforestation-linked supply chains [23]. Palm oil has been a primary target for the emergence of voluntary private standards [24,25]. The Roundtable on Sustainable Palm Oil (RSPO) and related institutions have triggered responses in Malaysia and Indonesia (jointly around 85% of global palm oil production) that saw their national sovereignty at risk in international markets [26]. A study of RSPO impacts in Kalimantan suggested that it increased forest conversion in agricultural lands but decreased such within the forest estate [27].

Recent estimates, discussed below, show that, based on remote sensing, 15–20% of Indonesian oil palms are located within the 'forest zone' (Kawasan hutan), a primarily institutional indication of land status. We indicate these as Forest Zone Oil Palm (FZ-OP). The FZ-OP percentage can be interpreted in multiple ways. To an international audience that has come to believe that most, if not all, of Indonesian palm oil is linked to deforestation, the number may seem to be small. To the Indonesian palm oil sector, it may give the impression of some bad apples that need to be dealt with, as they spoil the good name of the vast majority of those falsely accused of deforestation. To the forest authorities, the numbers are a clear sign of illegality, as current rules do not allow oil palm to be planted within the forest zone as FZ-OP. Actually, oil palms are the most readily recognizable of the four major tree crops (coffee, cocoa, rubber and oil palm) that occur both outside and inside the forest zone [28], but part of the smallholder oil palm may be part of a finer-grained mosaic and not as easily noticeable. Key questions regarding FZ-OP are: (1) What came first: local land use or designation of the land as part of the forest zone?, (2) How does the land use with tree crops relate to the primary designated forest function in (A) production forest (income generation), (B) protection forest (watershed protection) and (C) Conservation areas (Biodiversity), with recent additional interests in restricting drainage of peat soils in the production forest zone, (3) How do the various stakeholders along the palm-oil value chain respond to the lack of legality of forest-based oil palm production?

Issues related to oil palm (and its product palm oil) require an understanding of the whole value chain, as concerns of end-users, often based on limited or biased information they have access to, inform the product manufacturers, who want to be able to claim a 'clean' product, and their suppliers, the partly refined products derived from Crude Palm Oil (CPO), processed from Fresh Fruit Bunches (FFB) transported from farmgate to mill, and grown by smallholders or large-scale plantations. Current leverage on 'illegal' palm oil may primarily come from making mills responsible for their source areas. However, it is not clear whether these efforts mostly lead to 'shifting blame' [29,30] or whether they actually contribute to changing land-use and effectively protecting forests.

While uncertainty over the exact numbers and locations on FZ-OP continues, we set out to formulate lessons from earlier interactions between forest authorities and the other tree crops (coffee, cocoa, rubber) and to contextualize and analyze policy options and how these might be applied to the various contexts that contribute to the overall issue. We will review the available data and literature on:


We conclude by formulating more specific policy research questions as a follow-up.
