Pressure and friction.

Lava upwellings under the hotspots. Other than that, we don't know.

1) "Characteristics: J. Tuzo Wilson came up with the idea in 1963 that volcanic chains like the Hawaiian Islands result from the slow movement of a tectonic plate across a "fixed" hot spot deep beneath the surface of the planet. Hotspots are thought to be caused by a narrow stream of hot mantle convecting up from the Earth's core-mantle boundary called a mantle plume, although some geologists prefer upper-mantle convection as a cause. This in turn has re-raised the antipodal pair impact hypothesis, the idea that pairs of opposite hotspots may result from the impact of a large meteor. Geologists have identified some 40–50 such hotspots around the globe, with Hawaii, Réunion, Yellowstone, Galápagos, and Iceland overlying the most currently active. Most hotspot volcanoes are basaltic because they erupt through oceanic lithosphere (e.g., Hawaii, Tahiti). As a result, they are less explosive than subduction zone volcanoes, in which water is trapped under the overriding plate. Where hotspots occur under continental crust, basaltic magma is trapped in the less dense continental crust, which is heated and melts to form rhyolites. These rhyolites can be quite hot and form violent eruptions, despite their low water content. For example, the Yellowstone Caldera was formed by some of the most powerful volcanic explosions in geologic history. However, when rhyolitic magma is completely erupted, it may eventually turn into basaltic magma because it is no longer trapped in the less dense continental crust. An example of this activity is the Ilgachuz Range in British Columbia, which was created by an early complex series of trachyte and rhyolite eruptions, and late extrusion of a sequence of basaltic lava flows. - Trail: As the continents and seafloor drift across the mantle plume, "hotspot" volcanoes generally leave unmistakable evidence of their passage through seafloor or continental crust. In the case of the Hawaiian hotspot, the islands themselves are the remnant evidence of the movement of the seafloor over the hotspot in the Earth's mantle. The Yellowstone hotspot emerged in the Columbia Plateau of the US Pacific Northwest. The Deccan Traps of India are thought to be the result of the emergence of the hotspot currently under Réunion Island, off the coast of eastern Africa. Geologists use hotspots to help track the movement of the Earth's plates. Such hotspots are so active that they often record step-by-step changes in the direction of the Earth's magnetic poles. Thanks to lava flows from a series of eruptions in the Columbia Plateau, scientists now know that the reversal of magnetic poles takes about 5,000 years, fading until there is no detectable magnetism, then reforming in near-opposite directions." Source and further information: http://en.wikipedia.org/wiki/Hotspot_(geology%29 2) "On Dec. 15 in San Francisco's Moscone Center, four scientists attending the 2004 meeting of the American Geophysical Union debated the cause of hotspots at a press conference titled "Plumes or Not?" Professors Norman H. Sleep of Stanford and Donald J. DePaolo of the University of California-Berkeley posited that plumes of hot material rising from the deep mantle, possibly as deep as the Earth's core, are the best explanation for hotspots. Professors Gillian R. Foulger of the University of Durham (in the United Kingdom) and James H. Natland of the University of Miami argued that models based on shallow processes better fit the data for many hotspots. "Hotspots reveal that we don't know what's happening in the deep Earth," said Sleep during an interview before the press conference. Sleep has received funding from the National Science Foundation to study the base of the lithosphere—the rocky rind encircling the Earth's mantle like the tough skin of an onion. "Plumes are probably what's causing the majority of these hotspots, but they're not well understood." Plume advocates painted a picture of searing material rising from the planet's deep interior to reach the base of the lithosphere. "If the plate is moving, it's like moving your hand over a candle," Sleep said. "Slowly you get a series of burns." Plume skeptics, by contrast, advocated that small cracks produced by stress in a plate let magma out. "In the crack and magma theory, you have partial melt everywhere beneath the lithosphere—very widespread," Sleep explained. Cracks propagate in regions such as the Basin and Range Province, where land from Idaho to Mexico and from California to Utah has been stretched out. Here the crust thins. Cracks form and magma pushes through, creating hotspots. While a crack and magma are required to make a volcano, Sleep said, "the [question] is whether the crack is a secondary feature or the primary one." Canadian geophysicist J. Tuzo Wilson came up with the hotspot theory in 1963 to explain long-lived volcanic activity in certain regions of the world, such as the Hawaiian Islands. Magma forced through cracks creates volcanoes. Volcanism could only sustain itself over long periods if hotspots existed below the plates. In 1971, W. Jason Morgan of Princeton built on the hotspot theory by proposing that ridges and volcanoes could form when an oceanic plate passes over a hot mantle plume. At Hawaii, a stationary spot deep in the mantle persistently produces magma, lighter than surrounding rock, that erupts onto the surface and forms new land. Driven by plate movement, eventually the newly formed land moves on, becoming cut off from its hotspot. The magma then gets forced through different cracks to form even newer land. That process is recorded in the rocks of Hawaii's Big Island, which sports active volcanoes, and islands further away, such as Oahu, now inactive and eroding, and the Midway Islands, which have subsided almost to sea level. Ironically, in the 1970s Sleep had been a major proponent of the crack hypothesis, which had been proposed as early as the 1800s. But he changed his mind around 1985 when he was trying to explain the swell of the Hawaiian Islands—a region of uplifted seafloor that extends hundreds of kilometers roughly east of the Big Island. He decided the crack hypothesis required contrivances that made it "too complicated to be true." With the plume hypothesis, as the plate moves over the plume, it drags plume material with it "just like smoke out of a chimney gets dragged in the wind," said Sleep, who had intended to show that it was a geometric impossibility for plumes to form the uplifted region. "My 'proof of impossibility' explained its shape—literally on the first plot I did," Sleep laughed. "I didn't have an absurdity at all. I had evidence for plumes." The other major line of evidence supporting the plume hypothesis is from the four or five spots on Earth where hotspots cross ridges. Plumes can cross ridges without any problem, but it doesn't make a lot of sense for a propagating crack to cross a ridge axis, Sleep said. Still, he admitted, there's a place for cracks in the big picture. "Cracks are a mechanism that produces minor hotspots and modulates the hotspots that we actually have. … We also may have small hotspots that are cracks that aren't associated with plumes. And we definitely have hotspots that are secondary ones that are produced from the plume material flowing toward the ridge axis."" Source and further information: http://news.stanford.edu/news/2005/january12/agu_sleepsr-011205.html