This paper raises the prospect of citizen science being used to record species interactions, and that this has benefit for citizen scientists and for ecological knowledge, with two examples being specifically highlighted: invasive species and urban habitats.

I agree with many of the suggestions and arguments proposed by the authors and there is some value in re-stating these and framing them in the context of advances in citizen science. I realise that this is a short forum article, but I did not find that the arguments were sufficiently novel or coherent. The problems identified are common to many aspects of citizen science or species interaction data, while the solutions are widely-stated (albeit important) solutions regarding biodiversity data. I believe that the paper would have been better if it had had a stronger framing (it was not clear whether the invasive species angle was core or peripheral to the paper) and a clearer sense of novelty and vision for this topic, especially regarding the challenges, opportunities and solutions to citizen science species interaction data.

L150 The problem of 'free form' responses could have been expanded, e.g. a user recording "bumblebee on dandelion" could be pollen transfer, nectar feeding, sitting on a petal, or sitting on a leaf! Some people have explored hierarchical ontologies of interactions and this could have been introduced and the authors could have given their opinion of the value of this (for ecological knowledge and for engagement)

L159 and elsewhere. I felt that the user engagement aspect of networks was not explored very much. I agree with the argument of 'emotional connection' (L122), but I would have liked to have known where the authors think that this could go. Does it add to recorder biases? Could it reduce recorder bias? (I can think of examples in both cases.) They state that visualisation is likely to be important but beyond drawing networks for users, what is the value of this, and what is the evidence for the value? GLoBI does this already - is there evidence that it is valuable?

L171-187 The question of bias is important. The authors consider how positive bias could be beneficial (e.g. for species strongly associated with humans, e.g. some pests and garden plants), but do not stated how this might be problematic.

L212 Biases are raised, but the reader is "left to speculate on what interaction biases may exist". I would rather consider the authors' thoughts on this, the importance of different types of bias, and how this might limit our scientific inference.

L239 The questions of standards for interaction data are vital, but this wasn't really explored here and the "Relational Ontology" was not explained. How could this be done? How far could it develop in a way that works for science and citizen engagement? Combining and visualising data from multiple sources "will be challenging" but in what way, and how can this be addressed?

L277 Leveraging image analysis and machine learning is interesting and should have been explored in the text, rather than raised in the conclusions. How do these interact or support citizen science?

L286 Concluding with the fact that citizens will "enjoy and learn from" the recording of interactions seems at odds with the motivations raised in the paper about disease transmission via interactions, or the imperative of recording invasive species interactions.

Minor points:
L241 I do not think that Isaac et al 2020 refers to joint species distribution models.
L269 co-occurrence in the context of these references is not directly relevant to directly observed species interactions, which appears to be the focus of the paper.