The objective of systematic conservation planning is to select areas to protect or rehabilitate ecological assets in the most efficient way. After setting targets for ecological assets, heuristic algorithms or optimization techniques are employed to meet these targets. This technique – traditionally only used in terrestrial and marine settings – has recently been adapted to river management, acknowledging spatial constraints arising from the connected nature of rivers. However, terrestrial heuristics and optimization techniques employed to solve the minimum-set or maximum-coverage problem in conservation planning scenarios have been designed to deal with non-connected systems. Therefore, different algorithms will perform better or worse in a riverine setting. In this study, we compare the performance of two different techniques to identify important cells for meeting ecological targets in terms of efficiency, congruence and computational effort. The first technique is a heuristic algorithm, often used in classic conservation planning problem. Heuristics operate in a stepwise manner, selecting for the most taxa rich or the rarest feature first, then recalculate the selection matrix and run until all conservation targets are covered. To ensure connectivity of planning units is preserved, we modified the rules of the heuristic: Isolated planning units in the middle of a river system cannot be selected. Instead the entire catchment area upstream will have to be protected The second method is an extension of the conservation software package MARXAN. After allocating a random initial reserve, planning units are randomly added to and taken out. Each step is evaluated against an objective function that considers the achieved conservation targets, as well as cost and compactness of the reserve system. The last measure – compactness of the reserve system – is used to accommodate MARXAN to lotic systems. Instead of penalising for all boundaries of a planning unit, only the planning units that are crossed by a river are counted. We found that while the heuristic assigned a higher range of irreplaceability values, the areas of high conservation value were similar in both algorithms. When comparing the best solutions (also termed near-minimum sets), we found that an increasing boundary penalty in MARXAN also increases the reserve network. While a run without penalty only needs 27 out of 1854 planning units, this increases to 174 units at penalty 10 and 696 at penalty hundred. At boundary penalty 100, not all features were captured, as the penalty for compactness exceeded the penalty for not meeting targets. The 174 units at penalty 10 take up a slightly smaller area than the best solution of the heuristic algorithm. Boundary penalty 10 seems to be the optimal penalty in the current dataset. While it is not as strict in the upstream protection as the modified heuristic, it still creates a network of compact reserves at a configuration that is easier to achieve. However, because of the lack in the strict upstream protection, the reserve network might not be adequate for some of the targets, depending on the strength of upstream disturbance. While this is a great step forward to advance river conservation planning, more research into the tradeoffs between whole catchment protection and practicality will have to be conducted. Currently, the node-based approach in the heuristic ensures to find a near-optimal set under the constraints that whole-catchment protection is needed. With a medium boundary penalty setting, MARXAN can deliver more efficient reserve designs, but this could lead to inadequate protection. As a future research direction, we recommend to include information about the downstream extent of disturbance to ensure adequacy.