Photovoltaic (PV) hosting capacity studies usually consider uniform probability distribution to install PV by consumers. The history of PV installation in urban areas of several cities shows that the adhesion of PV in low-voltage (LV) networks has a spatial influence, i.e., the inhabitants' decision to install a PV system is influenced by their neighbors. This paper presents a novel methodology to estimate PV hosting capacity (HC) by distribution transformers considering the spatial dependence on the adhesion of consumers. In the proposed method, the first law of geography is used to obtain the probabilities of installing PV by consumers considering their spatial influence. These probabilities provide the generation of diverse PV integration scenarios. The evaluation is performed using an LV feeder of a distribution company from São Paulo city in Brazil. A comparison with a uniform probability approach is also presented. Results indicate that the HC occurs at lower levels of PV penetration when the spatial dependence is considered. Therefore, the proposed method can aid distribution companies in better characterizing PV penetration scenarios in HC studies