Human activities have greatly intensified since the Industrial Revolution, resulting in worldwide environmental degradation and biodiversity loss ^[1][2][1,2]. Many countries have established a series of protected areas to reduce human pressure on the ecosystem and protect biodiversity. Protected areas cover roughly 15.4% of Earth’s surface [3]; however, they have shortfalls in effectiveness [4]. There are 20 “Aichi Biodiversity Targets” set by the United Nations Conference on Biological Diversity in 2010, including “Address the underlying causes of biodiversity loss by mainstreaming biodiversity across government and society” and “Reduce the direct pressures on biodiversity and promote sustainable use” (https://www.cbd.int/sp/targets accessed on 15 December 2021). By 2020, however, none of them have been fully achieved, and only six goals have been partially achieved [5].

It has become increasingly important to assess whether protected areas are effective. At the global scale, the effectiveness of protected areas in implementing biodiversity measures [6], reducing forest loss [7], and protecting natural land cover [8] was assessed. At the national scale, some scholars evaluated the conservation effectiveness for amphibian biodiversity [9] and carnivores [10]. Moreover, some evaluation studies are also available at the regional scale ^[11][12][13][11,12,13]. These studies primarily focus on changes of species or ecosystem services, which has provided a good foundation for further studies. However, assessing changes in biodiversity and ecosystem services is not only time- and cost-consuming. It also requires integrating a great number of biotic and abiotic variables that are not fully comprehended and differ from period to period and area to area. As the targets of protected areas were achieved by reducing human activity, evaluating the change in human activity intensity should also be feasible. It is accurate and efficient because the human pressure data, including population density and land use intensity, is easier to obtain than biodiversity and ecosystem services data. Some studies have attempted to assess the effectiveness of protected areas from the perspective of human activity, including land use ^[14][15][14,15], roads [16], grazing ^[17][18][17,18], and energy consumption such as nighttime lights ^[19][20][19,20]. Nevertheless, most of them have investigated only one or two human impact factors, thus resulting in insufficient assessments.

Some researchers ^[21][22][21,22] mapped human footprint (HF), measuring human activities from multiple perspectives. It is a map that shows the impacts of human activity on the planet’s surface [21]. It measures the cumulative disturbance intensity of some categories of human pressures on ecosystems, including built environments, crop and pasture lands, population density, nighttime lights, roads and railways, and navigable waterways [23]. By contrast with collecting biodiversity and ecosystem services data, it is easier to obtain these satellite-based human pressure data. So HF has been widely adopted in studies at various scales. Specifically, at the regional scale, it has been used to evaluate the effectiveness of nature reserves in Tibet [24], the Hengduan Mountains region [25], and south Ecuador [26]. It has been mapped for Iran [27] and Argentina [28] at the national scale. At the global scale, it has been used to evaluate the effectiveness of global terrestrial nature reserves for different years ^[22][29][30][22,29,30]. These studies have shown that HF is increasingly being used to evaluate the effectiveness of nature reserves [27].

As the Yellow River, Yangtze River, and Mekong River source, the Sanjiangyuan region is rich in biodiversity and ecosystem services. China and the Chinese government have designated a nature reserve in this area known as Sanjiangyuan nature reserve (SNR), one of the largest in China situated in the Qinghai-Tibet plateau. In terms of biodiversity and ecosystem services, evaluating the SNR’s effectiveness is challenging because it is nearly impossible to collect data covering a long time on this remote plateau with a harsh natural environment. A more flexible way to assess its effectiveness is measuring HF and its changes after SNR is established.

The intensity of human activities in the Sanjiangyuan region was generally low from 1995 to 2015, with a fluctuating increasing trend. The average HF value in the Sanjiangyuan region was only 1.0307 in 20 years (Figure 1). The HF value was considerable in the eastern parts of the Sanjiangyuan region and low in its western parts.

The average HF value increased from 0.9885 in 1995 to 1.0161 in 2005, indicating human activity intensity was increasing for the Sanjiangyuan region. For 2005–2010, the average HF value decreased from 1.0161 to 0.9697. After that, it increased to 1.1857 in 2015. Spatially, the HF value is high in the east and south of the Sanjiangyuan region and low in the west of the region. Collectively, the spatial pattern of HF remained stable for the whole study period (Figure 1).

There was an increasing trend for population density, land use intensity, nighttime lights, and roads from 1995 to 2015. In particular, the pressure values of roads and population density increased by 0.0973 and 0.1612, respectively, over 20 years (Table 1). Nevertheless, grazing intensity showed a decreasing trend.