Host manipulation by parasites and parasitoids is a fascinating phenomenon within evolutionary ecology, representing an example of extended phenotypes. To elucidate the mechanism of host manipulation, revealing the origin and function of the invoked actions is essential. Our study focused on the ichneumonid spider ectoparasitoid Reclinervellus nielseni, which turns its host spider (Cyclosa argenteoalba) into a drugged navvy, to modify the web structure into a more persistent cocoon web so that the wasp can pupate safely on this web after the spider's death. We focused on whether the cocoon web originated from the resting web that an unparasitized spider builds before moulting, by comparing web structures, building behaviour and silk spectral/tensile properties. We found that both resting and cocoon webs have reduced numbers of radii decorated by numerous fibrous threads and specific decorating behaviour was identical, suggesting that the cocoon web in this system has roots in the innate resting web and ecdysteroid-related components may be responsible for the manipulation. We also show that these decorations reflect UV light, possibly to prevent damage by flying web-destroyers such as birds or large insects. Furthermore, the tensile test revealed that the spider is induced to repeat certain behavioural steps in addition to resting web construction so that many more threads are laid down for web reinforcement.

#### Appendix 1 radii in normal orb webs

Raw data of tensile test of radii in normal orb webs with graphs regarding force-strain curve. In all graphs, X-axis is strain and Y-axis is force [mN]. First graph (above) is raw force-strain curve, second (middle or below) is thirty points' moving-averaged force-strain curve, and third (below) is extract of initial slope used for calculation of Young's modulus, if any. All sample names described in each sheet in this category are designated as OR (meaning orb web radius) with the number behind hyphen representing each sample (individual web) and the lowercase alphabet representing replicates.

#### Appendix 2 frames in normal orb webs

Raw data of tensile test of frame threads in normal orb webs with graphs regarding force-strain curve. In all graphs, X-axis is strain and Y-axis is force [mN]. First graph (above) is raw force-strain curve, second (middle) is thirty points' moving-averaged force-strain curve, and third (below) is extract of initial slope used for calculation of Young's modulus, if any. All sample names described in each sheet in this category are designated as OF (meaning orb web frame thread) with the number behind hyphen representing each sample (individual web). In this category, there is no replicates (a thread from one web).

#### Appendix 3 radii in resting webs

Raw data of tensile test of radii in resting webs with graphs regarding force-strain curve. In all graphs, X-axis is strain and Y-axis is force [mN]. First graph (above) is raw force-strain curve, second (middle) is thirty points' moving-averaged force-strain curve, and third (below) is extract of initial slope used for calculation of Young's modulus. All sample names described in each sheet in this category are designated as RR (meaning resting web radius) with the number behind hyphen representing each sample (individual web) and the lowercase alphabet representing replicates.

#### Appendix 4 frames in resting webs

Raw data of tensile test of frame threads in resting webs with graphs regarding force-strain curve. In all graphs, X-axis is strain and Y-axis is force [mN]. First graph (above) is raw force-strain curve, second (middle) is thirty points' moving-averaged force-strain curve, and third (below) is extract of initial slope used for calculation of Young's modulus. All sample names described in each sheet in this category are designated as RF (meaning resting web frame thread) with the number behind hyphen representing each sample (individual web). In this category, there is no replicates (a thread from one web).

#### Appendix 5 radii in cocoon webs

Raw data of tensile test of radii in cocoon webs with graphs regarding force-strain curve. In all graphs, X-axis is strain and Y-axis is force [mN]. First graph (above) is raw force-strain curve, second (middle or below) is thirty points' moving-averaged force-strain curve, and third (below) is extract of initial slope used for calculation of Young's modulus, if any. All sample names described in each sheet in this category are designated as CR (meaning cocoon web radius) with the number behind hyphen representing each sample (individual web) and the lowercase alphabet representing replicates.

#### Appendix 6 frames in cocoon webs

Raw data of tensile test of frame threads in cocoon webs with graphs regarding force-strain curve. In all graphs, X-axis is strain and Y-axis is force [mN]. First graph (above) is raw force-strain curve, second (middle) is thirty points' moving-averaged force-strain curve, and third (below) is extract of initial slope used for calculation of Young's modulus. All sample names described in each sheet in this category are designated as CF (meaning cocoon web frame thread) with the number behind hyphen representing each sample (individual web) and the lowercase alphabet representing replicates.

Raw data used for statistical analysis of tensile properties. Three sheets are presented; "Thread diameter (μm)", "Tensile properties", and "Analysis".

The first sheet shows raw data on thread diameters in column D, the average values of intra-samples in column E, and the log-transformed averaged values in column F. Note that there are several errors and omitted data at the time of publication, so corresponding data are coloured and annotated in columns I-M. Correct values and their log-transformed values are shown in columns G and H.

The second sheet shows breaking strain (column D), breaking force (column F), raw Young's modulus (column J), breaking stress (column L), and Young's modulus (column N). Column I shows representative diameters used to calculate breaking stress and Young's modulus. The values in columns L and N, which are partly incorrect because of incorrect thread diameters, were used for statistical analysis at the time of publication. Columns M and O show correct breaking stress and Young's modulus recalculated using correct thread diameters.

The third sheet shows tidy data (linked with the previous two sheets) used for statistical analysis and a table of the results of ANOVA and the Tukey-Kramer test for multiple comparisons as a post hoc test. The tidy data show both datasets used for published results and those with corrected values. Changed values since publication are cloured. Regarding the statistical results of thread diameter, breaking stress, and Young's modulus, both published results with partly wrong dataset and reanalyzed results with correct dataset exhibit no difference among statistical relationships.