Biological control protects tropical forests

Biological control of invasive species can restore crop yields, and thus ease land pressure and contribute to forest conservation. In this study, we show how biological control against the mealybug Phenacoccus manihoti (Hemiptera) slowed deforestation across Southeast Asia. In Thailand, the newly-arrived mealybug caused an 18% decline in cassava yields over 2009-2010, a shortfall in national production and an escalation in the price of cassava products. This spurred an expansion of cassava cropping in neighboring countries from 713,000 ha in 2009 to >1 million ha by 2011: satellite imagery reveal 388%, 330%, 185% and 608% increases in peak deforestation rates in Cambodia, Lao PDR, Myanmar and Viet Nam focused in cassava crop expansion areas. Following release of the host-specific natural enemy Anagyrus lopezi (Hymenoptera) in 2010, mealybug outbreaks were reduced, cropping area contracted and associated deforestation slowed by 31-94% in individual countries. When used with due caution and according to current guidelines, biological control offers broad benefits for people and the environment.

must identify and promote all appropriate options including those, such as arthropod biological 48 control, that are often disregarded. Here, for the first time, we show how biological control can 49 reduce pressure on land and thus spare forests. 50 Invasive species, including many agricultural pests, constrain the production of food and other 51 commodities 7 , and often impose additional costs such as the disruption of ecosystem services 52 (e.g., nutrient cycling), damage to infrastructure or increased disease in humans 8 . Since the late 53 1800s, more than 200 invasive insect pests and over 50 weeds across the globe have been 54 completely or partially suppressed through biological control, often with favorable benefit:cost 55 ratios (ranging from 5:1 to >1,000:1) 9, 10 . Modern biological control, centered on a careful 56 selection and subsequent introduction of a specialized natural enemy (from the pest species" 57 region of origin), thus offers an effective solution for invasive species problems 11 . This approach 58 is particularly useful in smallholder farming systems, as biological control is self-propagating 59 and requires little involvement from local stakeholders 12 . Nonetheless there are risks as 60 exemplified by few (poorly-selected) control agents that have subsequently become major 61 problems themselves, such as the cane toad Buffo marinus or the weevil Rhinocyllus conicus 13, 62 14 . A consequence is that, despite significant improvements in risk assessment and management 63 over the past three decades, concerns often obscure the potential benefits and result in biological 64 control being avoided when it could be valuable 13 . While the failures of the last century appear 65 4 well known, the more modern success stories require wider recognition. Our goal here is to 66 present one such story. 67 In 2010, biological control was implemented against the invasive mealybug, Phenacoccus

85
In this study, we characterized how the cassava mealybug invasion and ensuing biological 86 control are associated with agricultural expansion and forest loss in mainland Southeast Asia.

87
These forests include the most species-rich and biologically-valuable habitats in the region 19,20 . Our surveys, conducted across mainland Southeast Asia between 2014 to 2017 (i.e., 6-9 years 99 and 5-8 years following the initial P. manihoti detection and A. lopezi introduction, respectively), 100 showed that P. manihoti was present in 37.0% of the fields (n= 549) and comprised 20.8% 101 abundance within a speciose mealybug complex (Fig. 1). Among sites, P. manihoti reached 102 field-level incidence of 7.6 ± 15.9% (mean ± SD; i.e., proportion mealybug-affected tips) and semi-perennial crop, routinely harvested at 8-10 months of age), and reduced total cassava 118 production. Over the ensuing 2009-10 cropping season, province-level yields dropped by 12.59 ± 119 9.78% (area-weighted mean: -18.2%) and country-wide aggregate yields declined from 22.67 120 t/ha to 18.57 t/ha (Fig. 2). Regional production followed similar trends: total production in Viet  Thailand"s cassava production to recover, helped stabilize cassava trade, averted the need for 181 insecticides in neighboring countries and reduced cassava-driven forest loss in the region.

182
Demand for cassava is an important driver of land-use pressure and forest loss in the Greater 183 Mekong sub-region, yet it is not the only one, and the A. lopezi introduction alone thus will not 184 avert future deforestation. Other drivers of forest loss are the establishment of oil palm, pulp and 185 paper plantations, rubber and food crops 24 ; crops that are cultivated across tropical Asia through 186 10 significant engagement from agro-enterprises 24,25 , with their actions regularly affected by "place-187 less" incentives (e.g., varietal improvements) 26,27 , (foreign-based) consumer demand 15 , or 188 chronic soil fertility loss e.g., for cassava in fragile upland settings 28 . Yet, during 2010-2012, our 189 analyses revealed the marked role of cassava area growth in deforestation at a multi-country 190 level. To stabilize the forest margin, a multi-functional "landscape approach" and a systematic 191 analysis of the various drivers of land-use change will be necessary 29 . Also, in order to enhance 192 the capacity of cropping systems to absorb (or recover from) perturbances such as the P.  Though few cases have justifiably blemished the reputation of arthropod biological control, 208 current practices and safeguards minimize such risks 13,39 . Our study equally helps put those 209 11 risks into perspective, as the rapid A. lopezi introduction proved essential to alleviate the 210 disruptive impacts of P. manihoti attack 34,40 . The benefits gained through the A. lopezi release 211 thus need to be viewed in light of the multi-faceted ecosystem impacts of invasive species 41,42 , 212 and the environmentally-disruptive actions that are regularly employed for their mitigation 43,44 .

213
Our study illustrates how an invasive pest can lead to substantial loss of forest 45 and accelerating 214 species loss and extinctions 24,46 , while biological control offered a powerful, environmentally-215 benign tool to permanently resolve those impacts 11 . Now, by concurrently highlighting the 216 harmful and beneficial impacts of P. manihoti and A. lopezi respectively, our work can enable a 217 concerted search for "win-win" solutions that address invasive species mitigation, biodiversity 218 conservation and profitable farming.

219
Biological control requires a reassessment by all those responsible for achieving a better world             Upon acceptance of the manuscript, all data will be made available in an appropriate public 528 structured data depository, and the accession number(s) provided in the manuscript.  In Thailand, farmer surveys were conducted in a total of 33 cassava-growing provinces over 575 the course of 2014 (i.e., 6 years after the initial detection of P. manihoti). In each province, a 576 variable number of farmers was interviewed by DoAE personnel, ranging from n= 20 577 Payao) to n= 348 (Karnchanaburi), attaining a grand total of 2,505 cassava farmers in the 578 national territory. Sample size was determined by local authorities, and is only partially 579 reflective of the number of cassava growers in a given province. In Cambodia and Vietnam, 580 nation-wide surveys were carried out in a more systematic fashion in at least 15 districts per 581 country during late 2016 (i.e., 8 years following the initial detection of P. manihoti in Thailand).

582
In these two countries, interviews were focused in districts with the largest area of cassava 583 production. Within each district, a total of 15 cassava growers were randomly selected and 584 interviewed, attaining the following respective sample sizes for Cambodia and Vietnam: n= 240 585 (16 districts) and n= 206 (15 districts). District-level adoption rates were pooled per province for 586 either country, and mapped at a regional scale.