With the emergence of the Internet of Things, environmental sensing has been gaining interest, promising to improve agricultural practices by facilitating decision-making based on gathered environmental data (i.e., weather forecasting, crop monitoring, soil moisture sensing). Environmental sensing, and by extension what is referred to as precision or smart agriculture, pose new challenges, especially regarding the collection of environmental data in presence of connectivity disruptions, their gathering, and their exploitation by end-users or by systems that must perform actions according to the values of those collected data. In this paper, we present a middleware platform for the Internet of Things that implements disruption tolerant opportunistic networking and computing techniques, and that makes it possible to expose and to manage physical objects through Web-based protocols, standards and technologies, thus providing interoperability between objects and creating a Web of Things (WoT). This WoT-based opportunistic computing approach is backed up by a practical experiment whose outcomes are analyzed in this article.