1：Experimental Design and Method Selection:

The PNMS system was built on top of an inverted optical microscope using a high-numerical-aperture oil-immersion objective. A 680 nm, 15 mW superluminescent light-emitting diode (SLED) laser was used as the light source. BK-7 glass coverslips coated with a 2 nm Cr adhesion layer and 48 nm of Au were used as sensing chips. A polydimethylsiloxane (PDMS) sample cell was placed on this chip. The chip was modified with hydrophilic self-assembled monolayers to control the adhesion strength of bacteria.

2：Sample Selection and Data Sources:

Escherichia coli (E.coli, CICC 10003) and Bacillus thuringiensis (B. thuringiensis, CICC 10322) were purchased from China Center of Industrial Culture Collection (CICC). Vibrio qinghaiensis sp. Q67 was from the culture collection of our group.

3：List of Experimental Equipment and Materials:

Inverted optical microscope, oil-immersion objective (60×) with a high numerical aperture (NA =

4：49), 680 nm, 15 mW superluminescent light-emitting diode (SLED) laser (SUPERLUM, Ireland), BK-7 glass coverslips coated with a 2 nm Cr adhesion layer and 48 nm of Au, polydimethylsiloxane (PDMS) sample cell. Experimental Procedures and Operational Workflow:

Bacteria solution (OD600 =

5：2) was added into the sensing cell before each batch of toxic tests. After 5 min deposition, some bacteria from the bulk solution adhered to the sensing surface, and the vibrations of these bacteria were recorded by the PNMS as control. Then, toxicant was also added in sensing cell. For the joint toxicity assessment, the addition was a mixture of two toxicants (Cu2+ and phenol). Data Analysis Methods:

The plasmonic intensity of each ROI were extracted by the image-processing software, ImageJ. The vibration amplitudes were determined as the standard deviation of the vertical displacement in time series (typically within 2000 frames). The inhibition ratios of treated bacteria cells were fitted with the toxicological models.