Coastal upwelling systems, such as the Peru and the Canary Current Systems, are among the most productive ecosystems in the global ocean. Several observations and numerical models suggest that primary productivity declined over the past a few decades, and will to continue during the next century due to the impact of global warming. However, if at a global scale, a warming of the ocean surface is observed since the beginning of the industrial era, at a regional scale, the surface waters in the upwelling zones indicate, on the contrary, a cooling. To explain this occurrence, Bakun (1990) hypothesized that global warming increases the land-sea temperature gradients which in turn increase the alongshore winds favorable to upwelling. Although the driving mechanism is still the subject of debate, most experimental and numerical models studies indicate that coastal upwelling has intensified in recent years and the trend is toward further increases. However, there are very few studies that go beyond the 20th century. To make up for this lack of data, here, we present new data from high-resolution sediment cores covering the last two centuries and cored off the Peru and Senegal coasts. We used alkenones to reconstruct Sea Surface Temperature (SST); Total Organic Carbon (TOC) as proxy of surface productivity and granulometric parameters as a proxy of transport of terrigenous material. Our results indicate a systematic cooling developing over the last century ranging from 0.5 to 1°C off Peru. In contrast, sea surface waters withn the Canary current system exhibit warming at latitude lower than 20°N and a strong cooling at 31°N, with a marked steepening in the second part of the 20th century. There was also a close relationship between SST and TOC. In Peru and at 20°N off Senegal, cooling is coupled with an increase in TOC, suggesting an increase in surface productivity, while warming of the sea surface observed off the coast of Senegal at 31°N is parallel to a decrease in TOC. Preliminary data from the Pacific Ocean upwelling indicate complex covariance between SSTs and the silts content fraction.