Phosphorus (P) deficiency is a global issue which can severely impact the yield of crops, including the P demanding and important food crop potato. Diagnosis of P status directly in the field can be used to adapt P fertilization strategies to the needs of the evolving crop during the growing season and is often estimated by analyzing P concentrations in leaf tissue. In this study, we investigate how diagnosis of P status in field grown potato plants is affected by leaf position and time of measurement in a randomized block experiment. The concentrations of many essential plant nutrients are highly dynamic, and large differences in nutrient concentrations were found in potato leaves depending on leaf age and time of sampling. During tuber initiation, P concentrations decreased in a steep gradient from the youngest leaves (0.8%) towards the oldest leaves (0.2%). The P concentrations in the youngest fully expanded leaf decreased by 25-33% within just 7 days, due to a high remobilization of P from source to sink tissue during crop development. 40 days later P concentrations in all leaves were near or below the established critical P concentration of 0.22%. The P concentration in leaf tissue thus depends on sampling time and leaf position on the plant, which in a practical setting might prevent a meaningful interpretation in terms of fertilizer recommendation. The chlorophyll a fluorescence parameter "P-predict", derived from the fluorescence transients, is an alternative to the classical chemical analysis of nutrient concentrations in leaf tissue. P-predict values serve as a proxy for the bioavailable P pool in the leaf and can be measured directly in the field using handheld technology. However, in conditions of high solar irradiation, the P-predict values of the most light-exposed leaf positions, i.e. the younger leaves, were found to be severely impacted by photoinhibition, preventing accurate characterization of the P status in potatoes. Shading the plants can reverse or prevent photoinhibition and restore the diagnostic capabilities of the P-predict approach.