Borderlands are critical security zones but remain poorly understood. In regions plagued by drug violence and conflict, violent groups compete for territorial control, cooperate in illegal cross-border activities, and substitute for the functions of the state in these areas. Despite undermining physical security, fuelling fear, and challenging the state’s sovereignty, the exact modi operandi of these groups are little known.

Against this backdrop, this thesis explores how different interactions among violent non-state actors (VNSAs) in the Colombian-Ecuadorian and Colombian-Venezuelan borderlands impact on citizen security. These border areas attract rebels, paramilitaries and criminal organisations alike: they constitute geo-strategic corridors for the global cocaine industry and are sites of supply and operation for the major actors involved in Colombia’s decades-long armed internal conflict.

Adopting an interdisciplinary approach, this thesis consolidates the literature on conflict, security and organised crime, borders and borderlands, and anthropological approaches to fear and violence. It integrates theories of cooperation among social actors with original empirical research. It is based on a comparative, multi-sited case-study design, using ethnographic methods complemented by quantitative data. The research involved over twelve months of fieldwork with 433 interviews and participant observation on both sides of the crisis-affected Colombia-Ecuador and Colombia-Venezuela borders, and in Bogotá, Caracas and Quito.

Developing a typology of VNSA interactions, I argue that these create not only physical violence but also less visible types of insecurity: when VNSAs fight each other, citizens are exposed to violence but follow the rules imposed by the opposing parties. Fragile alliances produce uncertainty among communities and erode the social fabric by fuelling interpersonal mistrust. Where VNSAs provide security and are socially recognised, "shadow citizen security" arises: security based on undemocratic means. I show that the geography of borderlands reinforces the distinct impacts of VNSA arrangements on citizen security yet renders them less visible.